Fig 2: SOD3 upregulation alters tumor-associated vasculature. a Lectin-FITC-perfused, CD31-stained vessels in Ad-C and Ad-mSOD3-injected tumors (tumor size 0.5–1.2 cm3). Data are shown as the percentage of CD31+lectin+ vessels relative to total CD31+-stained vessels. See also Supplementary Fig. 4b for representative images. b, c Density (b) and mean area (c) of CD31+ structures in Ad-C- and Ad-mSOD3-injected tumors at end point. d CD31 staining (red) of sections from Ad-C- and Ad-mSOD3-injected LLC tumors. e–g Determination of the mean length (e), diameter (f), and branches per vessel (g) from images as in d (n = 60 vessels/condition). h Percentage of CD31+lectin+ vessels relative to total CD31+-stained vessels in Vhcl- and Lov-treated LLC tumors (1–2.5 cm3) grafted into WT and SOD3-/- mice. i Reconstruction of tumor blood vessels from Vhcl- or Lov-treated, lectin-perfused WT and SOD3-/- mice using the Imaris software (tumor sections 50 µm). j Mean vessel area in tumors from Vhcl- and Lov-treated WT and SOD3-/- mice determined from images stained with anti-CD31 antibody. k, l Determination of mean length (k) and diameter (l) of CD31-stained tumor vessels from Vhcl- or Lov-treated WT and SOD3-/- mice (n = 70 vessels/condition). m Scanning electron microscopic analysis of blood vessels in tumors from Vhcl- and Lov-treated WT and SOD3-/- mice dissected at day 20 (tumor size <1 cm3). n Vessel leakiness in LLC tumors dissected at day 16 (size <1 cm3), determined by FITC-lectin and Texas Red-dextran injection, with two-color merge (yellow). o Vessel permeability of samples as in n, determined as the ratio of fluorescence intensity of extravasated dextran to perfused vessel counts. p Determination of VEC mRNA levels, normalized to those of CD31 (EC marker), in tumors from Vhcl- and Lov-treated WT and SOD3-/- mice. q Images of VEC (red) and CD31 (green) staining in tumors from Lov-treated WT or SOD3-/- mice dissected at day 20. Bottom panels show only VEC staining in a magnified area. r VEC-stained area, determined as a percentage of VEC staining in CD31+ structures, in images as in q. In all cases, 10–20 images were analyzed from =5 mice/group; *p< 0.05, **p< 0.01, ***p< 0.001; two-tailed Student’s t-test. Bar, 50 µm

Fig 3: SOD3 stabilizes HIF-2a posttranscriptionally via NO. a Immunoblot analysis of 1G11-mock and -SOD3 cell extracts with anti-HIF-2a and -tubulin antibodies (n = 2). The ratio between bands is shown (bottom). b Immunofluorescence analysis of HIF-2a staining in mock and SOD3-expressing cells (n = 500 cells/condition). c, d Percentage of nuclear area and mean fluorescence intensity in images in b. e Relative VEC promoter activity in IOX2- or Vhcl-treated 1G11-mock and -SOD3 cells. f Scheme of the ELISA used to detect PHD activity. g PHD activity, determined by ELISA, in mock and SOD3-expressing cell extracts. SOD3 levels are shown in these cells as determined by immunoblot (actin loading control; n = 3). h Top, HIF-2a ubiquitination analysis in immunoprecipitates of MG132-treated 1G11-mock and -SOD3 cells at the indicated times. Bottom, rehybridization with anti-HIF-2a antibody. i Immunoblot of HIF-2a in extracts from 1G11-mock and -SOD3 cells cultured in hypoxia (6 h), cycloheximide treated, and cultured in normoxia for the indicated times. Blots were rehybridized for tubulin as loading control. A representative experiment is shown (n = 4). j Quantification of the HIF-2a/tubulin densitometry ratio from i, expressed as the percentage of the ratio at time 0 (hypoxic conditions, 100%). k Representative images of Vhcl- or L-NMMA-treated 1G11-SOD3 cells stained with anti-HIF-2a antibodies. l, m Relative percentage of stained nuclear area and mean fluorescence intensity in Vhcl- or L-NMMA-treated 1G11-mock and -SOD3 cells. Values were normalized to the lowest value in the Vhcl-treated group for each cell type (n = 100 cells/condition). For a, c–e, g, j, l, m, data shown as mean ± SEM from at least three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001; two-tailed Student’s t-test (c, e, g, l, m) or two-way ANOVA with Bonferroni post-hoc test (j), showing significant differences between cell lines. Bar, 30 µm

Fig 4: Coregulation of SOD3, VEC, and HIF-2a in primary CRC tumors. a, b Scatter plots showing individual data points and mean for SOD3 (a) and VEC (b) mRNA levels in non-tumor (C; n = 13) and CRC samples (CRC; n = 102). c Correlation of SOD3 and VEC mRNA levels in the CRC cohort. d, e Scatter plots as above for SOD3 (d) and VEC (e) mRNA levels in a validation cohort of 19 non-tumor and 87 stage III CRC samples. f Correlation of SOD3 and VEC mRNA levels in the CRC validation cohort. g TMA samples stained for SOD3 and HIF-2a, representative of low (#6) and high (#61) SOD3 levels. h, i Correlation of estimated H-scores for SOD3 and HIF-2a in stromal (h) and tumor (i) compartments. j Representative images from two tumor samples stained for CD34 (red), SOD3 (green), and HIF-2a (blue); the three-color merge is shown; nuclei were counterstained with DAPI (see Supplementary Fig. 11 for single-color images). White lines in the DAPI panel contour CD34-stained areas. c, f, h, i Spearman’s rank correlation test; a, b, d, e *p < 0.05, two-tailed Student’s t-test. Bar, 50 µm (g) and 20 µm (j)

Fig 5: EC-specific SOD3 expression is sufficient to increase Doxo effect. a Conditional expression allele for the SOD3-IRES-GFP gene at the ROSA26 locus. b SOD3 (green) and CD31 (red) detection in sections from tumors grown in Cre-expressing and Cre- mice after tamoxifen induction. c LLC tumor growth kinetics in SOD3EC-Tg and control (Cre-) mice. Arrows indicate tamoxifen and Doxo treatments (n = 10/group). d Doxo quantification in extracts of tumors dissected at day 16 from control and SOD3EC-Tg mice (n = 15 or 17 mice/group). e, f Images and quantification of lectin-FITC-perfused and CD31-stained vessels in control and SOD3EC-Tg tumors (0.7–1.2 cm3) dissected at day 20. g CD31 staining of control and SOD3EC-Tg tumors at end point. h–l CD31+ structure density (h), mean area covered by CD31+ structures (i), mean vessel length (j) and diameter (k), and the number of vessel branches (l), determined in images as in g. m CD31 (green) and VEC (red) staining of tumors grafted in SOD3EC-Tg and control mice. Bottom panels show red staining in a magnified area. n Quantification of the VEC area in CD31+ structures from images as in m. o Vessel permeability determined by FITC-lectin and Texas Red-dextran in tumors dissected at day 17 from control and SOD3EC-Tg mice. For b, e–o, 10–20 images were analyzed from 5 mice/group; *p < 0.05, **p < 0.01, two-tailed Student’s t-test. Bar, 50 µm