H-Ras regulates angiogenesis and vascular permeability by activation of distinct downstream effectors

Abstract

Angiogenesis and vascular permeability occur following endothelium activation by vascular endothelial growth factor (VEGF). Downstream mechanisms that define these vascular responses remain unknown. H-Ras activation has been associated with the angiogenic response. However, active H-Ras initiates a wide spectrum of other biological responses through multiple downstream effectors. To identify vascular signaling by H-Ras and the immediate effectors we activated the Erk/MAPK or PI3K pathways in chicken and mouse endothelial tissues by ectopic expression of the Ras effector mutants H-RasV12S35 or H-RasV12C40, respectively. Constitutive activation of the Erk/MAPK pathway by H-RasV12S35 was sufficient to induce angiogenesis and not vascular permeability, whereas activation of the PI3K pathway by H-RasV12C40 was required for both angiogenesis and vascular permeability. Pharmacological inhibition of PI3K (α/β) suppressed both Ras- or VEGF-mediated vascular response in vivo, and survival of primary human endothelial cells in vitro. However, inhibition of PI3K (γ/δ) suppressed Ras- or VEGF-mediated vascular permeability in vivo, with no effect on survival of primary endothelial cells. This was supported by genetic studies since PI3K p110γ knock-out mice showed impaired vascular permeability response to VEGF or H-RasV12C40 treatment yet produced a wild type angiogenic response to H-RasV12S35. We conclude that downstream of VEGF, H-Ras serves as cellular switch that controls neovascularization and vascular permeability by activation of distinct effectors.

Introduction

Angiogenesis and vascular permeability occur in response to VEGF activation of endothelial cells.¹ Described first as a vascular permeability factor by Dvorak et al., ² VEGF is the only angiogenic growth factor that induces both vascular phenotypes. Other growth factors like bFGF or PDGF are only angiogenic.^{3, 4} Genetic deficiency of VEGF results in embryonic lethality due to failed development of vasculature.⁵ Angiogenesis is also associated with pathological processes such as tumor growth and metastasis, proliferative retinopathies, age-related macular degeneration, and rheumatoid arthritis.^{6, 7}

The GTPase Ras, proximally positioned upstream of a number of important signal transduction networks, becomes activated in the proangiogenic response to VEGF in adult tissues,^{8, 9} and in developmental angiogenesis since mice deficient in p120-ras GAP or NF-1, which facilitate Ras inactivation, fail to form organized vascular networks,^{10, 11} deletion of Sos1, a positive regulator of Ras activation, leads to cardiovascular and yolk salk defects, and embryonic lethality,¹² and disruption of the Ras effector B-Raf, results in vascular defects in mice and mid-gestational death.¹³ However, the coordinated circuitry and effectors that participate in angiogenic signaling downstream of Ras to manifest distinct vascular responses to specific growth factors have not been established. Additionally, Ras contribution to vascular permeability has not been documented to date.

Ras regulates cell growth, survival, and proliferation in all eukaryotic cells through signaling pathways that respond to peptide growth factors, cytokines, and hormones. These factors activate many downstream effectors through Ras, including Raf, p120 Ras GAP, RalGDS, phosphatidylinositol 3-kinase (PI3K), etc.^{14, 15} Deciphering the contribution of each specific Ras effector is challenging. Ras activity is controlled by GTP/GDP cycle, with residues corresponding to switch I (30-37) and II (59-76) regions defining conformational differences between the inactive GDP- and active GTP-Ras. An intact Ras effector domain (residues 32-40) is essential for all effectors interactions.¹⁶ Mutation of spanning residues 25-45 results in differential impairment of effector interactions and provides thus useful elegant tools to isolate contribution of specific effectors to Ras function.^{16, 17} Using such effector mutants, studies have revealed a bifurcation of the signaling pathways downstream of Ras leading to remodeling of the actin cytoskeleton and DNA synthesis.¹⁸

We have used here two such Ras effector mutants to identify selective contributions of Ras effectors of the ERK/MAPK- or PI3K- pathway to vascular phenotypes in vivo. RasV12C40 [G12→V12, T40→C40], binds to and selectively activates PI3K, while RasV12S35 [G12→V12, Y35→S35], binds to Raf1, and selectively activates the ERK/MAPK pathway.¹⁸ Ectopic expression of these mutations in chick or mouse endothelial tissues show that Ras-induced selective activation of the ERK/MAPK pathway or the PI3K pathway is sufficient to induce differential vascular phenotypes in vivo. Although VEGF-induced angiogenesis is accompanied by a vascular permeability response, VEGF-induced vascular permeability is not required for angiogenesis.¹⁹ Thus angiogenesis and vascular permeability are regulated independently downstream of VEGF. Here, we provide the first evidence that Ras may function as a cellular switch that controls angiogenesis and vascular permeability by activation of distinct downstream effectors.

Materials and Methods

Endothelial Cell Morphogenesis, Permeability, and Survival/Proliferation Assays were performed according to established protocols detailed in Figure Legends and supplemental Methods.

Angiogenesis Assays

10d fertilized chick embryos (standard pathogen free grade; SPAFAS, Preston, CT) were incubated at 37°C, 70% humidity. The chorioallantoic membrane was exposed and treated as detailed previously.^{8, 20} Sterile cortisone acetate–treated filter disks were soaked with 20μl of VEGF [200ng] in PBS, PBS alone, or adenovirus (10⁸ pfu in PBS) and added directly to the CAM, N=24 for each treatment. Blocking inhibitors (at concentrations indicated in the Figure Legends) were added in a volume of 10μl 1 h before VEGF treatment, and daily at 24h intervals for 3d. VEGF concentration and doses of PD98059, PI3K inhibitors, and adenoviruses were based on previously published results.^{8, 20} 5 days after treatment CAMs were explanted examined for new vessel branch points (capillary-sized). Quantification and photographs were obtained at 4× magnification through an Olympus SZH10 microscope using a Spot Camera and a Spot Diagnostic Detection System.^{8, 20}

Gene delivery of the human RasV12, RasV12S35 or RasV12C40 cDNA in adenovirus vectors in mice was completed by intradermal injection of the vectors in the ears of Nu/Nu, p110γ ^+/+ or ^-/- mice (3×10⁸ pfu per injection). Human GFP, RasN17 or VEGF cDNA in same vectors were used as internal controls and injected in the opposite ears as shown. Neovascularization, first observed 36h post-injection, was quantified by counting blood vessel branch points 8d post-infection at 4× magnification through an Olympus SZH10 microscope. Photographs were obtained at the same magnification.

Miles assay

All animal studies followed current “NIH Guidelines for the Use of Laboratory Animals” and IACUC-approved protocols. Nu/Nu mice, p110γ ^+/+ or ^-/- mice (18-21 g) were dosed i.p. with inhibitors or vehicle, when inhibitors were used. 1 h after inhibitor treatment, at time 0 with no treatment, or 5 days post-infection with adenovirus, 100 μl of 1% Evans blue dye (Sigma, St. Louis, MO) was administered i.v. For VEGF or cytokine treatment, animals were injected intradermally on each flank with 100 μL of either saline, VEGF (400 ng), or Pertussis Toxin (600ng). 30min following dye delivery mice were perfused, injection sites were photographed, and circular regions including the injection sites (8mm diameter) were excised. Permeability was quantified by elution of the Evan's blue in these sections in 400 μl of formamide at 56°C for 24 hr followed by absorbance measurements at 600 nm.

An expanded Materials and Methods section that includes resources and detailed Procedures on Antibodies and Reagents, Cell Culture, Adenovirus Preparation, Western Blotting, RNA isolation and cDNA synthesis, Quantitative Real-Time RT-PCR, Immunohistochemistry and Microscopy, and Statistics is available in the online data supplement.

Results

Selective Activation of MAPK Pathway by RasV12S35 is sufficient to induce angiogenesis

Tube formation by endothelial cells is a critical step in angiogenesis.⁹ An in vitro endothelial cell morphogenesis assay using human umbilical vein endothelial cells (HUVEC) expressing RasV12, RasV12S35 and RasV12C40 was performed. Representative photographs are shown in Fig. 1A, top, and the total tube length (mm) in three separate (10×) fields is shown in Fig. 1A, bottom. Compared with VEGF, RasV12 and RasV12S35 induced a significant increase in formation of capillary-like tubular structures that are sustained for up to 120h vs. 72h for VEGF. This is sustained by constitutive activation of the Erk/PI3K by the selective Ras mutations. In the VEGF treatment, an immediate activation of ERK/PI3K is induced by VEGF followed by depletion/inactivation of the VEGF from the serum at 37°C with time (Supplemental Fig. S1, A). Significantly, HUVEC expressing RasV12 and RasV12S35 induced similar levels of branching morphogenesis, while RasV12C40 failed to induce tube formation. Further treatment of HUVEC expressing RasV12S35 or RasV12C40 with VEGF produces little increase in morphogenesis without a synergistic effect (Supplemental Fig. S1, B). These findings reveal that Ras-induced activation of the ERK/MAPK pathway in cultured HUVEC is sufficient to induce tube formation in vitro while activation of PI3K is not.

Figure 1. Selective activation of the ERK/MAPK pathway by AdRasV12S35 is sufficient to produce angiogenesis in vitro and in vivo.

(A) Branching morphogenesis of HUVEC expressing RasV12, RasV12S35, RasV12C40, RasN17 and GFP. HUVEC were infected with AdRasV12, AdRasV12S35, AdRasV12C40, AdRasN17 and AdGFP at a MOI of 1, or treated with 2.5ng/ml VEGF. Scale bar represents 100μ (top). Total tube lengths (mm) in response to AdRasV12 (black triangle), AdRasV12S35 (red triangle), AdRasV12C40 (blue triangle), AdRasN17 (purple triangle), AdGFP (grey triangle), VEGF (green triangle) and PBS treatments are shown. (B-D) Ras-induced angiogenesis in CAM: (B) 10-d-old chick CAMs were treated with AdRasV12, AdRasV12S35, AdRasV12C40, AdRasN17, AdGFP [10⁸ pfu], VEGF (200ng), or PBS. Photomicrographs taken 5d post-infection are shown. Scale bars represent 0.5 mm. (C) CAMs lysed 36 h post-infection were analyzed by Western blot (100 μg/lane) to detect total Ras, P-Akt, P-ERK1/2, total Akt and ERK1,2. (D) Angiogenesis in treated CAMs was quantified by counting blood vessel branch-points double blinded as previously described ⁸. Each bar represents the mean ± SEM of three replicates (N=24). *, P<0.05 relative to control; **, P<0.05 relative to treatment.

Activated Ras and Ras effector mutants were transduced into HUVEC using adenoviral vectors. Equivalent expression by the adenoviral constructs carrying RasV12, RasV12S35, or RasV12C40 mutations, and GFP was observed by immunoblotting and immunofluorescence analysis (Supplemental Figure S2, Figure S3). The selective effectors activated by the Ras mutations were identified by monitoring kinase activities (phosphorylation status of specific substrates) as shown in Fig. S3. Constitutively active Ras (RasV12) induced ERK/MAPK activation as monitored by increased P-Erk levels (Fig. S3), and increased P-Akt, a measure of its capacity to activate PI3K.¹⁸ RasV12S35 produced increased P-Erk, but failed to induce phosphorylation of Akt, while RasV12C40 produced increased P-Akt compared to controls, yet had no impact on Erk.

The outcome of selective activation of the ERK/MAPK pathway by H-Ras in endothelial tissue in vivo was assessed by ectopic expression of Ras mutations in the chick chorioallantoic membrane (CAM). Filter disks saturated with AdRasV12, AdRasV12S35, or AdRasV12C40, were placed on the CAM of 10-day-old chick embryos (N=24 for each treatment), and the angiogenic response was assessed 5 days post-infection (Materials and Methods). Representative images of the angiogenic response to treatments are shown in Fig. 1B. Lysates of the transduced CAMs were evaluated for Ras expression, ERK- and PI3K-activity by immunoblotting specific antibodies to Ras, P-Erk and P-Akt [Ser473] (Fig 1C). A marked angiogenic response associated with activated Erk was detected in the CAMs treated with VEGF or those expressing RasV12 and RasV12S35 compared with controls (Fig. 1C, D). CAMs expressing RasV12C40 showed no angiogenic response or Erk activation (Fig. 1C, D) even though phosphorylation of Akt in these tissues is observed (Fig. 1C). Ectopic expression of RasN17, a dominant negative Ras [S17→N17], disrupted the angiogenic response to VEGF in CAMs (Fig. 1D), indicating that Ras activation is required for the angiogenic response downstream of VEGF. Detergent lysates of these CAMs (15 min after VEGF treatment) were evaluated for Ras expression, ERK and PI3K-activity as above (Fig 1C). Our findings indicate that Ras-induced selective activation of the ERK/MAPK pathway is sufficient for neovascularization both in vitro and in vivo.

Intradermal Expression of RasV12S35 and RasV12C40 in Mice Leads to Angiogenesis or Vascular Permeability

We used a murine model ²⁰ to observe the outcome of ectopic expression of Ras mutations on vascular phenotypes. Intradermal injections of the AdRasV12S35 or AdRasV12C40 were performed in the right ear, while the internal controls, AdVEGF and AdGFP, were injected in the left ear of each mouse (N=18 per each treatment). VEGF over-expression produced a robust neovascular response and extensive vascular permeability (Fig. 2A, B).²⁰ Representative images are shown (Fig. 2B). While both RasV12S35, and RasV12-mutations induced new blood vessel growth (Fig. 2A, B), RasV12 produced a combined angiogenic and vascular permeability response, though the response was not synergistic (fold compared with VEGF and each other) (Fig. 2A, B). Comparatively, RasV12C40 showed no angiogenic response (Fig. 2A, B). Consistent with the results in the CAM model, these findings indicate that selective activation of the ERK/MAPK-pathway by the RasV12S35 mutant is sufficient to induce an angiogenic phenotype in vivo. Expression of RasN17 prior to VEGF stimulation disrupted the vascular permeability response to VEGF in a Miles assay (Fig. 2C) indicating that active Ras is required for the vascular permeability response downstream of VEGF.

Figure 2. Angiogenesis and vascular permeability are mediated by Ras-induced-activation of distinct downstream effectors.

(A, B) Ectopic expression of human RasV12, RasV12S35 or RasV12C40 cDNA was completed by intradermal injection of adenoviral vectors in the ears of Nu/Nu mice (3×10⁸ pfu per injection). Adenoviruses carrying human GFP, RasN17 or VEGF cDNA were injected as internal controls in the opposite ears. Neovascularization was quantified by counting new vessel branch points 8d post-infection (Methods). (C) Human GFP and RasN17 cDNA was delivered in ears of Nu/Nu mice as above. 5d-post infection VEGF (400ng) was injected at the AdRasN17- and the AdGFP-transduced sites. Vascular permeability induced by VEGF was determined using Miles assay (Methods). (D, E) Human RasV12, RasV12S35, RasV12C40, RasN17, GFP, and VEGF cDNA was delivered in the ears of Nu/Nu mice in adenoviral vectors as above. 5d-post infection the vascular permeability was quantified using Miles assay. Each bar (A, C, D) represents the mean ± SEM of three replicates (N=18). *, P<0.05 relative to control; **, P<0.05 relative to treatment.

Significantly, RasV12C40 mutation produced no angiogenic response (Fig. 2A, B), but induced considerable vascular permeability (Fig. 2D, E) as measured in a Miles assay (Materials and Methods). RasV12C40 induced ∼ 4.5-fold increase in vascular permeability compared with the GFP controls, and only ∼ 1.5-fold less response than the corresponding VEGF adenovirus (Fig. 2D). Representative images are shown in Fig. 2E. In contrast, RasV12S35 mutation produced no vascular permeability relative to RasV12C40 or VEGF (Fig. 2D, E). RasV12 induced a permeability response comparable with that of VEGF, consistent with its activation of both the ERK/MAPK- and PI3K-pathway. These results demonstrate that selective activation of the PI3K-pathway by RasV12C40 mutant is sufficient to induce vascular permeability.

To associate the vascular effects observed were with gene expression in endothelial cells we analyzed the adenoviral-transduced murine tissues. Tissue sections were examined for the endothelial marker CD31 and GFP expression driven from the same CMV promoter as the Ras cDNA in the adenoviral vectors (Fig. S4). We observed co-localization of GFP and CD31 (Fig. S4A, arrowheads) consistent with the endothelium expressing the adenoviral genes. To evaluate additionally for Ras-mediated ERK/MAPK and PI3K activation, tissue sections were stained for co-localization of P-ERK and/or P-Akt and CD31 (Fig. S4B). Consistent with the results in vitro (Fig. S3), we identify co-localization of increased P-Erk and CD31 in AdRasV12S35 treated sections (Fig. S4B), and co-localization of increased P-Akt and CD31 in AdRasV12C40 treatment (Fig. S4B, b) relative to control treatment (Fig. S4B, c). Control AdGFP-treated sections were stained for CD31 (Fig. S4B, c) or treated with secondary antibodies alone prior to staining for P-Erk and P-Akt (Fig. S4B, d). To determine if ectopic expression of RasV12, RasV12S35 and RasV12C40 leads to altered VEGF expression, we isolated the total RNA form these tissues and performed reverse transcription followed by Real-Time Quantitative PCR analysis of VEGF-A expression relative to the endogenous gene cyclophilin (CPH) (Methods). We found no evidence of increased VEGF-A expression with RasV12, RasV12S35 and RasV12C40 over-expression in the mouse ears (Table S1). Additionally, treated tissues did not show altered VEGF levels by western blotting (data not shown), indicating that VEGF half-life has not been altered by post-translational stabilization upon adenoviral treatment. To exclude other potential paracrine effects induced by the Ras mutations we have evaluated additionally effects of various autacoid inhibitors and the PI3K δ/γ inhibitor TG100-115 on the vascular permeabilitity induced by RasV12C40 (Supplemental Materials and Methods, Figure S5, A-F). TG100-115 blocked the transendothelial flux of FITC-fluorescent beads associated with RasV12C40 (Figure S5, B), while the NO inhibitor Nω-Nitro-L-Arginine, the serotonin inhibitor 4-chloro-L-phenylalanine, the inhibitor of Cox-1 and Cox-2 indomethacin, and the histamine inhibitor cyproheptadine hydrochloride have no specific effect on the RasV12C40-induced transendothelial flux of FITC-beads (Figure S5, C-F). These findings support our conclusion that the vascular permeability associated with ectopic expression of RasV12C40 in vitro and in vivo is specific to the activation of PI3K δ/γ pathway.

RasV12C40 or VEGF - induced vascular permeability but not angiogenesis is blocked by pharmacological or genetic disruption of PI3K γ/δ

ERK and PI3K activation has been linked to cell survival and angiogenesis.²¹ To identify further Ras effectors in endothelial phenotypes, mice transduced with AdRasV12C40, AdRasV12S35 or treated with VEGF were exposed to inhibitors of PI3K or MEK (Fig. 3A, B), and angiogenesis or vascular permeability were measured as described above. Inhibitors of PI3K α and PI3K β ²² disrupted both angiogenesis and vascular permeability in vivo regardless of the stimulus (Fig. 3A, B). In vitro, we observed a dose-dependent inhibition of HUVEC survival by these inhibitors (Fig. 4A). These results imply a general survival function for these two isoforms in vascular cells as observed previously in other tissues.²³ Significantly, an inhibitor of PI3K δ/γ ²⁴, TG100-115, disrupted vascular permeability induced by RasV12C40 or VEGF in vivo (Fig. 3A), yet had no effect on angiogenesis (Fig. 3B) or in vitro survival of HUVEC (Fig. 4A). Blockade of MEK completely disrupted angiogenesis (Fig. 3B) without influencing vascular permeability induced by RasV12C40 or VEGF (Fig. 3A). To further evaluate to role of PI3K γ in the vascular permeability response we performed the Miles assay in mice deficient in p110γ^−/− p110γ^−/− mice showed impaired vascular permeability induced by RasV12C40 or VEGF (Fig. 4C), and normal/wild type response to VEGF- or RasV12S35- induced angiogenesis (data not shown). As seen in Figure 2, we find here that Ras activation of MEK/ERK is sufficient for angiogenesis. Significantly, Ras activation of δ/γ PI3K isoforms is sufficient for vascular permeability, and Ras activation of the α and/or β PI3K isoforms is required for survival of endothelial cells and thus indirectly participates to the angiogenic response.

Figure 3. RasV12C40 or VEGF - induced vascular permeability but not angiogenesis is disrupted with pharmacological inhibition of PI3K γ/δ.

(A, top) PI3K isoform specific inhibitors were PIK75 for PI3Kα, TGX115 for PI3Kβ, and TG100-115 for PI3Kδ/γ ²². MEK-1 inhibitor was PD98059. The vascular permeability induced by RasV12C40 in the ear of Nu/Nu mice treated with PD98059 (2.5mg/kg), PIK75 (.5mg/kg), TGX115 (2.5mg/kg), and TG100-115 (2.5mg/kg) was assessed (Methods). Each inhibitor was administered i.p. at 12h interval, daily, for 5d ²², ²⁴. (A, bottom) VEGF-induced vascular permeability in the skin of Nu/Nu mice treated with PI3K inhibitors (concentrations as above) for 1h, followed by VEGF (400 ng) or PBS treatment. Each bar represents the mean ± SEM of three replicates (N=18). *, P<0.05 relative to control; **, P<0.05 relative to treatment. (B, top) 10-d-old chick CAMs were exposed to filter paper discs saturated with AdRasV12S35 or AdGFP [10⁸ pfu per disc]. Inhibitors were added daily at 24h interval for 4 consecutive days. Neovascularization was quantified day 5 post-treatment by counting vessel branch points double-blinded ³⁸. (B, bottom) 10-d-old chick CAMs were treated with VEGF (200ng) or PBS. Inhibitors were added 1h prior to VEGF treatment and daily for 4 consecutive days after treatment. Neovascularization was quantified as above. Each bar represents the mean ± SEM of three replicates (N=24). *, P<0.05 relative to control; **, P<0.05 relative to treatment.

Figure 4. VEGF or RasV12C40 – induced vascular survival and permeability are selectively regulated by PI3K isoforms.

(A) Survival of HUVECs treated with VEGF and isoform specific PI3K inhibitors using XTT assay (Online Supplemental Methods). Background contribution from non-treated cells has been subtracted. Values represent averages of duplicates in two independent experiments. (B) Western Blot analyses of tissue lysates of mice ears treated with VEGF and PI3K or MEK-1 inhibitors (100 μg total protein/lane) were processed to detect P-Akt, P-ERK1/2, total Akt and ERK-1,2 using specific antibodies. (C) VEGF-, AdRasV12C40, and Pertussis Toxin (positive control for gene knock-out) were injected subcutaneously or in the ear of p110γ ^+/+ and ^-/- mice, and vascular permeability was assessed as detailed in Methods and above. Each bar represents the mean ± SEM of two replicates (N=12). *, P<0.05 relative to treatment in wild type. (D) Western Blot analyses of tissue lysates of mice ears treated with VEGF and AdRasC40 (100 μg total protein/lane) were processed to detect P-VE cadherin (Y-731-left panel, Y-658-right panel), VE cadherin, P-ERK1/2, and ERK-1,2 using specific antibodies.

To correlate these in vivo responses with downstream effectors, we monitored Akt or Erk phosphorylation. VEGF treatment induced Akt and Erk phosphorylation readily detectable in tissue lysates and, pretreatment with TG100-115 partially blocked Akt phosphorylation with no effect on Erk (Fig. 4B). Similar Akt inhibition was observed for tissues expressing RasV12C40 and treated with PI3K inhibitors (data not shown). We observed inhibition of both Akt and Erk phosphorylation upon treatment with VEGF and PI3K α and β inhibitors (Fig. 4B). We note that Akt activation by phosphorylation at Ser 473 recognized here by a pan-antibody does not differentiate substrate specficity. Our data thus indicate that Ras/ERK/MEK activation is sufficient to induce neovascularization, while Ras/PI3K δ/γ activation appears sufficient to disrupt vascular barrier function.

Discussion

Under pathological conditions such us ischemic injury or cancer, blood vessels undergo proliferation as well as loss of barrier function. While growth factors and inflammatory mediators can influence the growth and integrity of blood vessels, it is unclear how angiogenesis and vascular permeability are differentiated intracellular. Many growth factors and cytokines activate Ras, which in turn stimulates a wide range of signaling pathways. Here we asked if selective Ras mutations distinguish angiogenesis and vascular permeability via activation of downstream effectors. We show that Ras-induced selective activation of MEK/ERK pathway is sufficient to mediate angiogenesis while Ras-induced selective activation of the PI3Kδ/γ/Akt pathway promotes vascular permeability. This is depicted schematically in Fig. 5.

Figure 5. Ras regulates angiogenesis and vascular permeability by differential activation of downstream effectors.

VEGF stimulation of new blood vessel growth is critical for embryonic development ⁵. However, the ability of VEGF to promote vascular permeability has profound pathological consequences.²⁵ For example, following stroke or myocardial infarction, VEGF-mediated vascular permeability causes a significant increase in the level of infarcted tissue immediately following injury.^{24, 26} Given that the blood vessel growth promoting activity of VEGF would benefit ischemic tissues it is important to understand how vascular permeability and neovascularization are regulated at the molecular levels. Previous studies suggest distinct vascular responses mediated by differential regulation of signaling pathways downstream of VEGF or other growth factors: inhibition of Protein Kinase C, a VEGF downstream effector, disrupts angiogenesis while enhancing VEGF-induced vascular permeability;²⁷ while both VEGF and bFGF are angiogenic, only VEGF mediates vascular permeability;³ mice deficient in individual Src family kinases, pp60c-src or pp62c-yes show angiogenesis but no vascular permeability in response to VEGF.^{19, 20} Our data provide first evidence that Ras can serve as a cellular switch for angiogenesis or vascular permeability downstream of VEGF, by activation of distinct effectors of the MEK/ERK- or PI3K/δ/γ/Akt pathway, respectively.

We used both in vitro and in vivo approaches to assess the role of Ras signaling in endothelial cell signaling. We found that ectopic expression of RasV12S35, and RasV12 on basal RasV12 background promoted branching morphogenesis and tube formation by HUVEC in vitro, while expression of RasV12C40 did not. Thus Ras to Erk signaling appears sufficient for endothelial cell tube formation. In vivo we observed similarly that only RasV12S35 and RasV12 induced angiogenesis in the embryonic chick CAM, or mouse ears. As RasV12 and RasV12S35 promoted Ras to Erk signaling and RasV12C40 activated PI3K signaling to Akt, the lack of an angiogenic response to RasV12C40 was not due to its incapacity to initiate a signaling response. Rather, selective Ras-derived signals propagate distinct vascular responses. Activation of effectors of the Ras/ERK pathway, Raf or MEK1, has been shown to promote proliferative responses and growth, associated with the angiogenic response observed here.⁸ Consistently, we observed inhibition of MEK-1 disrupted the angiogenic response induced by RasV12S35 in vivo. We asked if these vascular phenotypes may be indirect autocrine/paracrine effects mediated by increased VEGF expression in the cells expressing the Ras mutations. We find that ectopic expression of RasV12, RasV12S35, or RasV12C40 does not alter VEGF expression, or increase the half-life of VEGF by post-translational stabilization in vivo, consistent with previous in vitro analyses.¹³ Comparatively, the vascular phenotypes observed here are clearly distinct as RasV12S35 induces angiogenesis whereas RasV12C40 induces vascular leak, while VEGF is concurrently angiogenic and a vascular permeability factor.

While the RasV12C40-induced activation of the PI3K pathway is sufficient to induce vascular permeability, not all PI3K isoforms function in this regard. PI3K α and β isoforms have been linked to general cell survival in other tissues,²⁸ while PI3K γ and δ have been associated with inflammation.²⁴ Accordingly, inhibitors of PI3K α and β isoforms disrupted both vascular growth and permeability, due to their ability to induce dose-dependent inhibition of endothelial cell survival (in vitro). In agreement, p110 α- and β-deficient mice are embryonic lethal due to defects in DNA synthesis and cell survival.^{28, 29} In contrast, inhibition of the PI3K δ/γ isoforms selectively blocked vascular permeability yet had no effect on new blood vessel growth or HUVEC survival in vitro. Moreover, the vascular permeability response to both RasV12C40 and VEGF was significantly diminished in p110γ -/- mice untreated and treated further with a PI3K δ inhibitor (data not shown). These findings suggest that PI3K δ and γ are be both necessary and sufficient for vascular permeability. Consistent with this proposal, p110 δ – and γ – deficient mice are viable, but show reduced inflammatory and immune response which may in part be associated with reduced edema.²⁴ In contrast to the role that PI3K plays in vascular permeability, inhibition of MEK-1 has no effect on vascular permeability (Fig. 3A). However, it clearly prevents the angiogenic response induced by RasV12S35 or VEGF (Fig. 3B), which is consistent previous studies.^{4, 8, 23}

Vascular barrier function depends on the integrity of VE-cadherin-mediated cell-cell junctions and the phosphorylation state of VE-cadherin and associated proteins.^{30, 31} Therefore, we probed for phosphorylation of VE-cadherin in the RasV12C40 and VEGF-treated tissues. We observed phosphorylation of VE-cadherin in both treatments (Figs. 4D, S6) consistent with disassembly of the adherens junctions.³⁰ We have also considered the involvement of Src in RasV12C40 mediated vascular permeability and found that a Src inhibitor that blocks VEGF-mediated leak had no effect on RasC40-induced vascular permeability (data not shown). This suggests that Src mediates vascular permeability independently or upstream of Ras. In fact, integrin-ligation leading to Src activation may involve Ras downstream in this process.^{8, 20}

VEGF-induced vascular permeability is known to precede calcium/solute flux, and tissue perfusion in vasodilatation,³² regulate the female menstrual cycle,³³ and fibrin deposition in wound repair.³⁴ VEGF or other growth factor-mediated angiogenesis is observed in early development,²¹ and in the adult associated with implantation and placentation in the ovary and uterus.³³ We establish here that differential Ras activation is sufficient to mediate vascular permeability or angiogenesis. When the mechanisms governing these phenotypes become deregulated pathologies occur. As such, mutational activation of Ras is not well tolerated, as seen in the Costello syndrome, the Noonan syndrome and the cardio-facio-cutaneous syndrome.^{15, 35} Vascular edema and VEGF expression also occur during ischemic injury and cancer.³⁶ Therefore it is imperative to understand how Ras and its immediate effectors differentially regulate angiogenesis and the vascular barrier function downstream of VEGF.

A number of clinical trials have been aimed at disrupting Ras signal transduction. The fact that Ras controls a broad spectrum of biological responses may explain why this approach has not met success. For example inhibitors of farnesylation that disrupt H-Ras activation were inefficient due to transfer of farnesyl/geranyl specificity, and combination therapies were found highly toxic.¹⁵ Similarly, the first approved VEGF inhibitors in cancer have been found to prolong survival in cancer patients by months. However, VEGF neutralization caused a large increase in the circulating red blood cells,³⁷ which though acceptable to terminal patients, is of concern to individuals with non-life-threatening conditions, such us blindness or arthritis. Therefore, efforts that have been more focused on therapies targeting the ERK/MAPK and PI3K pathways independently may provide a greater degree of safety, and ultimately improve the treatment options for arthritis, blindness, ischemic disease, defective wound repair, endometriosis and lastly cancer.^{33, 36}