Endothelial deletion of SHP2 suppresses tumor angiogenesis and promotes vascular normalization

Abstract

SHP2 mediates the activities of multiple receptor tyrosine kinase signaling and its function in endothelial processes has been explored extensively. However, genetic studies on the role of SHP2 in tumor angiogenesis have not been conducted. Here, we show that SHP2 is activated in tumor endothelia. Shp2 deletion and pharmacological inhibition reduce tumor growth and microvascular density in multiple mouse tumor models. Shp2 deletion also leads to tumor vascular normalization, indicated by increased pericyte coverage and vessel perfusion. SHP2 inefficiency impairs endothelial cell proliferation, migration, and tubulogenesis through downregulating the expression of proangiogenic SRY-Box transcription factor 7 (SOX7), whose re-expression restores endothelial function in SHP2-knockdown cells and tumor growth, angiogenesis, and vascular abnormalization in Shp2-deleted mice. SHP2 stabilizes apoptosis signal-regulating kinase 1 (ASK1), which regulates SOX7 expression mediated by c-Jun. Our studies suggest SHP2 in tumor associated endothelial cells is a promising anti-angiogenic target for cancer therapy.

Fig. 1. Hyper-activated SHP2 in tumor endothelial cells is necessary for tumor growth.

a Western blot for p-SHP2^Y542 and SHP2 in HUVECs treated with various NSCLC cancer cell-conditioned media for 10 min (p-SHP2, n = 3) or 24 h (SHP2, n = 5). β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM. p-values were shown and generated by one-way ANOVA with multi-comparisons. b, c Representative images showing p-SHP2^Y542 (b, green) and SHP2 expression (c, green) in CD31⁺ vessels (red) in NSCLC tumor tissues and paired adjacent normal tissues (n = 7 for p-SHP2 and n = 4 for SHP2). Quantitative data were shown as mean ± SEM. p-values were shown and generated by using the two-tailed Student’s t-test. Scale bar: 50 μm; inset 100 μm. d Shp2 mRNA expression in mouse normal and tumor endothelial cells was extracted from the GEO dataset (GSE118904, n = 1000 cells). p-value was shown and generated by using the two-tailed Student’s t-test. e Images of explant LLC tumors from Shp2^f/f and Shp2^iECKO mice. Scale bar: 10 mm. f, g The volumes and weights of LLC tumors in Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. The weights were recorded 16 days after cancer cell injection. Quantitative data were shown as mean ± SEM. p-values were shown and generated by two-way ANOVA with Tukey’s post hoc test (f) or by using the two-tailed Student’s t-test (g). h Images for orthotopic E0771 tumors in Shp2^f/f (n = 4) and Shp2^iECKO (n = 5) mice. Scale bar: 10 mm. i, j The volumes (i) and weights (j) for orthotopic E0771 tumors in Shp2^f/f (n = 4) and Shp2^iECKO (n = 5) mice. Tumor weights were measured 15 days after cancer cell injection. Quantitative data were shown as mean ± SEM. p-values were shown and generated by two-way ANOVA with Tukey’s post hoc test or by using the two-tailed Student’s t-test. k, l Representative images of LLC tumor necrosis (k) and apoptosis (l). H&E staining was conducted and necrosis area was labeled by dot lines. Scale bar: 5 mm. Apoptosis was measured by TUNEL staining. Scale bar: 5 mm. DAPI was used to label nuclei. Quantitative data were shown as mean ± SEM. n = 5 for each group. p-values were shown and generated by using the two-tailed Student’s t-test. m Representative images of hypoxyprobe-1-labeled areas in LLC tumors. Quantitative data were shown as mean ± SEM. n = 3 for each group. p-value was shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. Source data are provided as a Source data file.

Fig. 2. Shp2 deletion in endothelial cells impairs tumor angiogenesis and promotes vascular normalization.

a Immunofluorescence images of CD31 in LLC tumors in Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Quantitative data were shown as mean ± SEM. p-value was shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. b Images of Matrigel plugs from Shp2^f/f and Shp2^iECKO mice. c Immunofluorescence images of CD31 in plugs from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Images were analyzed by the Image J software and quantitative data were shown as mean ± SEM. p-value was shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. d Sprouting assay of aortic rings from Shp2^f/f (n = 3) and Shp2^iECKO (n = 3) mice. Quantitative sprouting area were shown as mean ± SEM. p-value was shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. e Tube formation in vitro of MLECs isolated from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Junction numbers, tube lengths, and branching lengths were measured by the Image J software and were shown as mean ± SEM. p-values were shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. f Western blot for SHP2 expression in SHP2-knockdown HUVECs with SHP2^WT re-expression. β-Actin was used as a loading control. Results were repeated for three independent experiments. g Tube formation in vitro in SHP2-knockdown HUVECs with SHP2 re-expression. Junction numbers, tube lengths, and branching lengths were measured by using the Image J software and shown as mean ± SEM. p-values were shown and generated by using one-way ANOVA with multi-comparisons. n = 5 for each group. Scale bar: 100 μm. h–j Immunofluorescence images of CD31 (h), αSMA (i), Collagen IV (j, Col4) in LLC tumors from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Branching index, pericyte coverage, basement membrane coverage were measured by using Image J software and quantitative data were shown as mean ± SEM. p-values were shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. k, l Representative images for vessels perfused with lectin (k) and dextran (l) in LLC tumors from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Images were analyzed by the Image J software and Quantitative data were shown as mean ± SEM. p-values were shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. m Immunofluorescence images of CD11b⁺ myeloid cells in LLC tumors from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Images were analyzed by the Image J software and Quantitative data were shown as mean ± SEM. p-values was shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. Source data are provided as a Source data file.

Fig. 3. SHP2 inhibitor reduces tumor angiogenesis and growth.

a Tumors were established in BALB/c nude mice by subcutaneous injection of LM3 cells expressing SHP2^WT or SHP099-resistant mutant (SHP2^T253M/Q257L, SHP2_TM/QL) and treated with vehicle, SHP099 (75/38.7/18.75 mg/kg body weight, daily) or cediranib (1.5 mg/kg body weight, daily) from day 3. b Images for LM3 tumors. Scale bar: 10 mm. c, d The volumes (c) and weights (d) for LM3 tumors. Tumor weights were measured 13 days after cancer cell injection. Quantitative data were shown as mean ± SEM. p-values were shown and generated by two-way ANOVA with Tukey’s post hoc test or by one-way ANOVA with multi-comparisons. n = 5 for SHP2^WT + control group, n = 7 for the other each group. e–h Immunofluorescence staining for CD31 (e, f), αSMA (g), Collagen IV (h, Col4) in LM3 tumors. Scale bar: 100 μm. n = 5 for SHP2^WT + control group, n = 7 for the other each group. i–l Vessel density (i), branching index (j), pericyte coverage (k), basement membrane coverage (l) were measured by using Image J software and quantitative data were shown as mean ± SEM. p-values were shown and generated by one-way ANOVA with multi-comparisons. n = 5 for SHP2^WT + control group, n = 7 for the other each group. Source data are provided as a Source data file.

Fig. 4. SHP2 regulates SOX7 expression.

a Heatmap showing qPCR results of angiogenesis-related genes in SHP2 knockdown (shSHP2) and control (shScr) HUVECs. b Western blot for SOX7 expression in SHP2-knockdown and control HUVECs. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-value was shown and generated by using the two-tailed Student’s t-test. c Western blot for SOX7 in overexpression of SHP2 or its A72G active mutation and control HUVECs. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by one-way ANOVA with multi-comparisons. d Western blot for SOX7 in HUVECs treated with various NSCLC cancer cell-conditioned media for 24 h. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for five independent experiments. p-values were shown and generated by one-way ANOVA with multi-comparisons. e–h Western blot for SOX7 expression in SHP2 knockdown and control HUVECs treated with VEGF-A (10 ng/ml) or FGF2 (10 ng/ml) for indicated time. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by two-way ANOVA with Bonferroni’s multiple comparisons test. i Representative images showing SOX7 expression (green) in CD31⁺ vessels (red) in LLC tumors from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Endothelial SOX7 was analyzed by the Image J software and quantitative data were shown as mean ± SEM. p-value was shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. j Representative images showing SOX7 expression (green) in CD31⁺ vessels (red) in NSCLC tumor tissues and paired adjacent normal tissues (n = 5). SOX7 expression in endothelial cells was analyzed by the Image J software and quantitative data were shown as mean ± SEM. p-value was shown and generated by using the two-tailed Student’s t-test. Scale bar: 100 μm. Source data are provided as a Source data file.

Fig. 5. SOX7 is required for SHP2 to promote tumor angiogenesis and vessel abnormalization.

a Images of LLC tumors from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice treated with AAV-Sox7. b, c Tumor volumes (b) and tumor weights (c) of LLC tumors from Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice treated with AAV-Sox7. Arrows indicated the time points for AAV and tamoxifen injection. Tumor weights were measured 18 days after cancer cell injection. Quantitative data were shown as mean ± SEM. p-values were shown and generated by two-way ANOVA with Tukey’s post hoc test (b) or one-way ANOVA with multi-comparisons (c). d Representative images of LLC tumor necrosis. Necrosis areas were labeled by dot lines. Images were analyzed by the Image J software and quantitative data were shown as mean ± SEM. n = 5 in each group. p-values were shown and generated by one-way ANOVA with multi-comparisons. Scale bar: 10 mm. e Representative images of hypoxyprobe-1 labeled areas in LLC tumors. Images were analyzed by the Image J software and quantitative data were shown as mean ± SEM. n = 5 in each group. p-values were shown and generated by one-way ANOVA with multi-comparisons. Scale bar: 100 μm. f–m Immunofluorescence images of CD31 (f, g), αSMA (h), and collagen IV (i, Col4) in LLC tumors. Vessel density (j), branching index (k), pericyte coverage (l), basement membrane coverage (m) were measured by using the Image J software and quantitative data were shown as mean ± SEM. n = 5 in each groups. p-values were shown and generated by one-way ANOVA with multi-comparisons. Scale bar: 100 μm. Source data are provided as a Source data file.

Fig. 6. SHP2 regulates SOX7 expression through c-Jun signaling.

a, b Western blot for c-Jun, ERK, and P38 and associated phosphorylated forms in SHP2-knockdown HUVECs and MLECs isolated from Shp2^f/f and Shp2^iECKO mice. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by one-way ANOVA with multi-comparisons or the two-tailed Student’s t-test. c Western blot for c-Jun and associated phosphorylated forms in SHP2 knockdown and control HUVECs treated with VEGF-A (10 ng/ml) or FGF2 (10 ng/ml) for 10 min. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by two-way ANOVA with Bonferroni’s multiple comparisons test. d Representative images showing p-c-Jun S63 level (green) in CD31⁺ vessels (red) in LLC tumors in Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Scale bar: 50 μm; inset 100 μm. e Luciferase reporter assay in HEK293 cells with c-Jun overexpression. The data were normalized to the Renilla luciferase activity. Quantitative data were shown as mean ± SEM for three independent experiments. p-value was shown and generated by the two-tailed Student’s t-test. f, g Western blot for SOX7, c-Jun, and its phospho-forms in HEK293 cells with c-Jun overexpression (f) and SOX7, c-Jun, phosphor-c-Jun, ERK, phosphor-ERK, P38, and p-P38 in HUVECs treated with various concentrations of JNK inhibitor SP600125 (24 h) (g). β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by using the two-tailed Student’s t-test or by one-way ANOVA with multi-comparisons. h Western blot for c-Jun and associated phosphorylated forms in HUVECs treated with various NSCLC cancer cell-conditioned media for 24 h. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by one-way ANOVA with multi-comparisons. Source data are provided as a Source data file.

Fig. 7. SHP2 positively regulates c-Jun/SOX7 signaling by inhibiting ASK1 degradation.

a qPCR for ASK1 in SHP2-knockdown HUVECs. Quantitative data were shown as mean ± SEM for three independent experiments. p-value was shown and generated by using the two-tailed Student’s t-test. b Western blot for ASK1 in SHP2-knockdown HUVECs with re-expression of SHP2WT or its activated (A72G) and catalytic-dead (C459S) mutants. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by one-way ANOVA with multi-comparisons. c, d Western blot for ASK1 in HUVECs treated with VEGF-A (d; 10 ng/ml) or FGF2 (e; 10 ng/ml). β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by two-way ANOVA with Tukey’s post hoc test. e Western blot showing co-immunoprecipitation assay for HA-tagged ASK1 and Myc-tagged SHP2 and its truncated mutations in HEK293 cells. Results were repeated for three independent experiments. f Ubiquitin conjugation assay for ASK1 and its Y718E mutation in HEK293 cells. Results were repeated for three independent experiments. g Representative images showing ASK1 expression (green) in CD31⁺ vessels (red) in LLC tumors in Shp2^f/f (n = 5) and Shp2^iECKO (n = 5) mice. Endothelial ASK1 were analyzed by the Image J software and quantitative data were shown as mean ± SEM. Scale bar: Scale bars: 50 μm; inset 100 μm. p-value was shown and generated by using the two-tailed Student’s t-test. h Western blot for ASK1 in HUVECs treated with various NSCLC cancer cell-conditioned media for 24 h. β-Actin was used as a loading control. Quantitative data were shown as mean ± SEM for three independent experiments. p-values were shown and generated by one-way ANOVA with multi-comparisons. i Representative images showing ASK1 expression (green) in CD31⁺ vessels (red) in NSCLC tissues (n = 7) and paired adjacent normal tissues (n = 7). Quantitative data were shown as mean ± SEM. p-value was shown and generated by using the two-tailed Student’s t-test. Scale bar: 10 μm. Source data are provided as a Source data file.

Fig. 8. Targeting endothelial SHP2 inhibits tumor growth.

a Schematic representation of the experimental model to delete Shp2 after tumor growth (late deletion). For late Shp2 deletion, tumors were grown in Shp2^f/f and Shp2^iECKO mice before tamoxifen was administered when the tumors reached a diameter of 5 mm. b Images of explant LLC tumors from Shp2^f/f and Shp2^iECKO mice (early deletion or late deletion). Scale bar: 10 mm. c, d The volumes (c) and weights (d) of LLC tumors from Shp2^f/f (n = 5) and Shp2^iECKO (early deletion, n = 3; late deletion, n = 5) mice. The weights were recorded 11 days after cancer cell injection. Quantitative data were shown as mean ± SEM. p-values were shown and generated by two-way ANOVA with Tukey’s post hoc test (c) or by using the two-tailed Student’s t-test (d). e Schematic illustration showing the role and mechanism of SHP2 in regulating tumor angiogenesis and vessel abnormality. In tumor-associated endothelial cells, expression of protein tyrosine phosphatase SHP2 is upregulated in response to signals from cancer cells. SHP2 dephosphorylates and stabilizes ASK1, which promotes c-Jun signaling to induce SOX7 expression. SOX7 is a proangiogenic factor and promotes tumor angiogenesis to produce tumor vessels with inefficiency in pericyte coverage, barrier integrity, and vessel perfusion. When SHP2 is genetically deleted or pharmacologically inhibited, through ASK1-c-Jun signaling, SOX7 expression is reduced. Reduced SOX7 expression results in a decrease in tumor angiogenesis while produces tumor vessels with well pericyte coverage, barrier integrity, and vessel perfusion. SHP2-ASK1-c-Jun-SOX7 is an important signaling axis in regulation tumor angiogenesis and vessel normalization, thus rendering the signaling pathway, especially SHP2, whose allosteric inhibitor is available, a valuable target for the development of anti-angiogenic therapy in cancers. Source data are provided as a Source data file.