(A) Blood vessels were detected by immunohistochemical analysis employing a CD31 Ab. (B) SEMA3A was present in some blood vessels in normal cervix (N) (arrowheads) and in the basal squamous epithelium (arrows); SEMA3A was expressed at high levels in vessels proximal to CIN-3 lesions (arrowheads), in squamous epithelium (arrows), and in scattered stromal cells (asterisks). Notably SEMA3A was barely detectable in SCC. SEMA3A expression was evaluated by immunostaining utilizing anti-SEMA3A Ab in parallel to CD31 analysis in consecutive sections. These images are representative of experiments repeated 3 times; 12 different samples per stage and 5 fields per sample were evaluated. Scale bars: 50 μm.

(A) Fluorescence confocal microscopy of RipTag2 mice using an anti-Sema3A Ab (red) revealed Sema3A expression in vessels (arrows) and epithelium of normal islets; Sema3A was highly expressed in angiogenic islets (arrows) and absent in tumors. Sema3A was mainly expressed by ECs, as revealed by colocalization of Meca-32 (green) with Sema3A (red) (arrows); Sema3A was present to a lesser extent in dysplastic epithelium (arrowheads). (B) In HPV/E₂ mice, Sema3A (red) was expressed in Meca-32–positive (green) blood vessels (arrows) and in the squamous epithelium (arrowheads) of E₂-treated normal cervix and CIN-3 lesions. Sema3A was also detected in scattered cells in the stroma underlying CIN-3 lesions (asterisks). Sema3A was barely detectable in SCC. Confocal analysis was performed on tissue sections from 10 mice per stage of both models and images are representative of 5 fields observed per stages. E, epithelium; S, stroma; Scale bars: 50 μm.

(A) Real-time RT-PCR analysis of integrin expression during RipTag2 tumorigenesis showed an increase in Itga5 and Itgb1 mRNA in both angiogenic islets and tumors, compared with normal islets. RQ values are mean ± SD of 4 experiments. Total RNA derived from a pool of islets of 10 mice per stage. (B and D) Confocal analysis revealed an increase in total β₁ integrins in ECs of angiogenic islets and tumors, compared with normal islets, as detected by colocalization of CD31 (green) with total β₁ (red) (D); graph shows the percentage of total β₁ integrin colocalized with ECs (B) (67% increase in angiogenic islets, 69% increase in tumors vs. normal islets, ***P < 0.0001). (C and D) Colocalization of CD31 (green) with active β₁ integrins (red) showed an increase in β₁ integrin activation in tumors compared with angiogenic islets and normal islets (D); graph shows the percentage of active β₁ integrin present on ECs (C) (51% increase in tumors vs. angiogenic islets, **P < 0.001). (E) Four-week-long AAV8-Sema3A treatment induced a decrease in active β₁ integrins (red) in tumor CD31-positive (green) ECs compared with controls. (F and G) Percentage of total and active β₁ integrins present on tumor ECs. β₁ integrin activation decreased by 53% in AAV8-Sema3A–treated mice compared with controls (G, **P < 0.001). Arrows and arrowheads, respectively, indicate total and active β₁ integrin colocalization with ECs and pericytes. Values are mean ± SD (n = 10 animals per group). Dotted lines indicate normal islets and lesions surrounded by the exocrine pancreas. Scale bars: 50 μm.

(A and B) Real-time RT-PCR revealed that Sema3a, Sema3f, and, to a lesser extent, Sema3e transcripts were strongly upregulated both in angiogenic islets (A) of RipTag2 mice (A) and in CIN-3 of HPV/E₂ mice (B) and downregulated in tumors (T) and SCC, as compared with normal islets (N) and E₂-treated normal cervix (N/E2), respectively. (C and D) Compared with that in normal stages, Nrp1 and Nrp2 expression increased in both angiogenic islets and tumors (C) as well as in CIN-3 and SCC (D). (E and F) Compared with that in normal islets and E₂-treated normal cervix, Plxna1, Plxnd1, and Plxnb1 mRNA was upregulated in both angiogenic islets and tumors (E) in addition to CIN-3 and SCC (F). In contrast, Plxna2 transcript increased in angiogenic islets and CIN-3 and decreased in tumors and SCC. Normalized relative quantification (RQ) values are compared with normal stages and are mean ± SD of 4 experiments. Sema3A transcript was present in sizeable amounts (pg per 100 ng of total RNA) in normal islets (Sema3A, 46 pg) and in normal cervix (Sema3A, 53 pg; see Supplemental Methods). Values were normalized to the endogenous Gus housekeeping gene. Total RNA derived from pools of islets or tissue lesions of 10 mice per stage (20 mice for normal islets). H, hyperplastic islets; CIN-1, -2, and -3, low, moderate, and high CIN, respectively.

(A) Compared with 14-week-old AAV8-LacZ controls, AAV8-Sema3A treatment of 12-week-old RipTag2 mice promoted apoptosis in ECs (arrows) and tumor cells (arrowheads) 2 (short trial) and 4 (long trial) weeks after AAV gene delivery, respectively. EC apoptotic rate was detected by colocalization of Meca-32 (green) with activated caspase-3 (red). Images are representative of 5 fields per mouse from a total of 10 mice per 2- and 4-week-long treatment. (B and C) Percentage of cleaved caspase-3⁺ cells on total cells in the short (B) and long (C) trial, compared with controls (**P < 0.001). (D) Upon Sema3A treatment, no apoptotic ECs were detected in blood vessels of normal pancreatic tissue in either short or long trial. (E) In long trials, anti-Ki67 immunostaining revealed no difference in proliferation rate in AAV8-Sema3A–treated compared with control animals. (F) Islet tumor hypoxia as detected by the formation of pimonidazole adducts (green). In short trials, Sema3A transiently enhanced tumor hypoxia compared with control, while in long trials, Sema3A-treated insulinomas were normoxic. Values are mean ± SD (n = 10 mice per 2- and 4 week-long treatment). Scale bars: 50 μm.

(A) Fluorescence confocal microscopy brought to light an increase in pericyte coverage of tumor blood vessels after 4 weeks of AVV8-Sema3A treatment compared with controls. Pericyte coverage was evaluated by analysis of colocalization (arrows) of Meca-32 with NG2, α-SMA, or PDGFR-β. Arrowheads indicate pericytes in the immediate vicinity of ECs. Images are representative of 5 fields per mouse from a total of 10 mice per treatment group. Scale bars: 50 μm. (B–D) Percentage of colocalization of pericyte markers on tumor ECs; quantification analysis revealed an increase in pericyte coverage of 44% for NG2⁺ cells (B), 45% for SMA⁺ cells (C), and 40% for PDGFR-β⁺ cells (D) in AAV8-Sema3A–treated versus untreated tumors (**P < 0.001). Values are mean ± SD (n = 10 animals per treatment group). Pericyte colocalization was measured as fluorescence intensity ratio between red (NG2, SMA, and PDGFR-β) and green (Meca-32) channels (see Methods). (E) While the NG2 gene was upregulated, the Rgs5 gene, a marker of activated pericytes, was strongly downregulated in AAV8-Sema3A–treated tumors (4 weeks) compared with controls, as detected by real-time RT-PCR. (F) Human recombinant Sema3A inhibited EC motility and enhanced human SMC migration in chemotaxis assays. Relative cell migration is expressed as fold increase compared with control unstimulated cells. Values are mean ± SD of 4 independent experiments.

Sema3A inhibition increased the angiogenic activity of islet purified from 10.5-week-old RipTag2 mice in a collagen gel bioassay. (A) Angiogenic islets in the absence of ECs and (B) ECs without islets in a collagen gel. (C) Angiogenic islets in the absence (untreated control) (D) or presence of SM-216289 alone (E) or VEGF-blocking Ab alone (F) or both SM-216289 and VEGF-blocking Ab, were embedded into a 3D collagen matrix containing ECs. SM-216289 neutralized the inhibitory effect induced by an anti-VEGF Ab (arrows). Scale bars: 50 μm (A–F). (G) EC migration in the absence or presence of SM-216289 and Sema3A, Sema3E, or Sema3F. SM-216289 did not interfere with either Sema3E or Sema3F inhibitory activity. (H) SM-216289 or saline were locally administrated by osmotic minipumps in 8.5-week-old RipTag2 mice for 2 weeks. Gross pathology images of pancreatic islets and tumors from 10.5-week-old animals after 2 weeks of saline treatment showing angiogenic islets and small tumors (left panel) compared with a significantly enhanced tumor volume due to continuous inhibition of Sema3A (right panel). Scale bars: 800 μm (H). (I) Increased tumor incidence in SM-216289– versus saline-treated animals (by 50%). (J) Increased tumor volume in SM-216289–treated animals compared with controls (84%, ***P < 0.0001). (K) SM-216289 treatment enhanced the number of angiogenic islets compared with controls (36%, **P < 0.001). (L) In a PT, Sema3A overexpression delayed the angiogenic switch, as indicated by a 60% reduction in the number of angiogenic islets in AAV8-Sema3A–treated mice compared with controls (n = 12, ***P < 0.0001).

(A) rAAV8 vector expresses myc-tagged Sema3A or LacZ under the control of the constitutive CMV promoter. (B) Tumor volume of AAV8-Sema3A–treated mice was reduced by 65% compared with that of controls (n = 10 and n = 12 for 12-week and 14-week RipTag2 controls, respectively; n = 12 for AAV8-Sema3A–treated 16-week-old mice; **P < 0.001). (C) Kaplan-Meier survival curve generated by injecting 12-week-old RipTag2 mice with either AAV8-Sema3A (n = 20) or AAV8-lacZ (n = 20) and monitoring their survival; Sema3A reexpression in tumors significantly extended median survival to 8.7 weeks compared with controls (log-rank [Mantel-Cox] test, P < 0.0001). Values are mean ± SD. (D and G) Vessel density, as assessed by Meca-32 immunostaining, was significantly reduced in AAV8-Sema3A tumors compared with controls (41% reduction in vessel area, *P < 0.01). (E) High-resolution confocal image stacks of vessels stained with anti–Meca-32 Ab were reconstructed by isosurface rendering using Imaris software. Tumor vascular morphology of Sema3A-treated mice was more linear, less branched, and with a smaller diameter compared with controls. (F) Higher magnification of the boxed areas in E of the 3D reconstruction in treated and untreated tumors. Analysis of confocal images revealed a reduction in vessel branching (53%, ***P < 0.0001) (H) and vessel diameter (46%, **P < 0.001) (I). Results are mean ± SD of 5 fields per mouse from a total of 10 mice per treatment group. (J) AAV8-Sema3A treatment did not affect normal vessels of exocrine pancreas. Scale bars: 50 μm.