Conceived and designed the experiments: SGU KHK BI. Performed the experiments: SGU SJ CD SC OB CS FH PP. Analyzed the data: SGU. Contributed reagents/materials/analysis tools: CD SC OB CS FH PP XM LM JA CR KHK. Wrote the paper: SGU KHK BI.
A patent is pending on GKT136901 with PP as inventor and GenKyoTex as owner of the patent. PP, CS and K-HK own shares of GenKyoTex. A patent is pending on fulvene-5, with JLA as inventor and Emory University as owner of the patent. This does not alter the authors' adherence to all the PLoS ONE policies on sharing data and materials.
Reactive oxygen species, ROS, are regulators of endothelial cell migration, proliferation and survival, events critically involved in angiogenesis. Different isoforms of ROS-generating NOX enzymes are expressed in the vasculature and provide distinct signaling cues through differential localization and activation. We show that mice deficient in NOX1, but not NOX2 or NOX4, have impaired angiogenesis. NOX1 expression and activity is increased in primary mouse and human endothelial cells upon angiogenic stimulation. NOX1 silencing decreases endothelial cell migration and tube-like structure formation, through the inhibition of PPARα, a regulator of NF-κB. Administration of a novel NOX-specific inhibitor reduced angiogenesis and tumor growth in vivo in a PPARα dependent manner. In conclusion, vascular NOX1 is a critical mediator of angiogenesis and an attractive target for anti-angiogenic therapies.
Angiogenesis is a complex process occurring in physiological situations such as embryogenesis and wound repair, and contributes to pathological conditions such as diabetes, psoriasis, arthritis and cancer. Angiogenesis is a critical determinant of cancer progression. In its absence, tumors are unable to grow beyond the size of microscopic lesions and persist as dormant, non-expanding nodules
NADPH oxidases are enzymes that produce reactive oxygen species (ROS). Depending on concentration and sub-cellular localization, ROS can mediate a variety of cellular functions, including pathogen killing, cell migration, proliferation and differentiation (for review
The nuclear hormone receptors peroxisome proliferator-activated receptors (PPAR) dimerize with the retinoid X receptor. Upon activation by lipids, this complex regulates gene transcription by binding to peroxisome proliferator-responsive elements. PPARα, a member of the family, was shown to mediate anti-inflammatory activity through inhibition of the transcription factor NF-κB. In the vascular system, PPARα inhibits NF-κB transactivation either by direct interaction with the p65 subunit or by up-regulation of I-κB, the NF-κB inhibitory subunit
In this study, we analyzed the role of NOX1 in human and mouse angiogenesis and observed an increased expression and activity of NOX1 during the angiogenic switch. Furthermore, blood vessel formation in NOX1-deficient mice was dramatically reduced in response to angiogenic factors and in tumors. NOX1 deficiency also lead to reduced endothelial cell migration and reduced formation of tube-like structures. We analyzed the mechanism by which NOX1 regulates angiogenesis and showed that NOX1 down-regulates expression and activity of the anti-inflammatory and anti-angiogenic nuclear receptor PPARα.
In order to test whether NOX-dependent ROS production participates in blood vessel formation, we performed
Matrigel was loaded with 500 ng/ml bFGF and implanted subcutaneously into NOX deficient mice. After one week, iodinated liposomes were injected i.v. and the plugs analyzed by X-ray tomography. (a) Quantification of matrigel plug vascularization. Graph shows mean of grey density ± s.e.m. For WT, NOX1 KO and NOX1/2 KO n = 8 and for NOX2 KO and NOX4 KO n = 6. (b) Photographs of excised plugs, scale bars represent 1 cm. c. Blood vessel density in plugs from the experiment in (a). Graph shows percentage of PECAM-1 positive area ± s.e.m. (d). Photographs of PECAM-1 immunostaining, PECAM-1 in green, nuclei in blue, scale bars represent 20 µm. Images were acquired with a 20x/0.5 numeric aperture lens and analyzed using LSM510 Meta confocal microscope (Carl Zeiss). e. Vessel size analysis; vessel with lumen under 20 µm are considered as small (black), from 20 to 50 medium (dark grey) and over 50 µm as large (light grey) ± s.e.m. Anova p<0.01; * p<0.05, ** p<0.01, *** p<0.001.
From these results, we conclude that NOX1 is essential for bFGF-induced angiogenesis.
To determine whether the aberrant angiogenesis observed in NOX1-deficient mice was due to impaired endothelial cell function; we studied the effect of NOX1 inhibition on mouse lung endothelial cells (MLEC)
NOX1 expression was measured by quantitative real-time PCR in mouse primary lung endothelial cells (MLEC, a), HUVEC (b), mouse endothelioma cell line (c) after stimulation with VEGF or b-FGF at 20 ng/ml for 3 hours. The quantity of NOX1 mRNA was normalized to the quantity of a housekeeping gene, tubulin for mouse cells and β2-microglobulin for human cells, ± s.e.m, n = 3. Activity of NOX1 was evaluated by ROS measurement using DHE substrate on MLEC (d,g), HUVEC (e, h) and mouse endothelioma (f, i). ROS production is up-regulated by VEGF and b-FGF stimulation in a NOX1-dependent manner. g, h, i. Graphs show quantification of ROS level in endothelial cells stimulated in presence or absence of VEGF or b-FGF. g. MLEC WT (black bar), MLEC WT + GKT136901 (grey bar), MLEC NOX1KO (white bar). h. HUVEC untreated (black bar) or treated with GKT136901 at 10 µM (grey bar). i. Mouse endothelioma cell line untreated (black bar), treated with GKT136901 at 10 µM (grey bar), treated with NOX1 siRNA (white bar). Results are expressed in fold increase ± s.e.m, n = 3. *p<0.05, ***<0.001 using Student's t-test.
Next, we analyzed ROS production in murine and human endothelial cells in response to VEGF and bFGF stimulation. As expected, these factors increased intracellular ROS levels (
(a) Concentration-response curve of GKT136901 (◊) and DPI (x) on membranes prepared from NOX1 over-expressing cells. Results represent one out of four experiments performed in triplicate. Values are presented as means ± s.e.m. (b) Affinities of GKT136901 and DPI on NOX1, NOX2, NOX4 and Xanthine Oxidase. GKT136901 is a NADPH-oxidase specific inhibitor, with selectivity on NOX1 and NOX4 over NOX2, whereas DPI inhibits all the NADPH oxidases tested with the same potency.
Taken together, the results obtained with NOX1 deficient cells and the inhibitor GKT136901 demonstrate that NOX1 is responsible for ROS production in endothelial cells stimulated with the pro-angiogenic factors VEGF or bFGF.
Angiogenesis requires endothelial cell proliferation, sprouting and migration
In vitro migration was analyzed by wound-healing assay on mouse primary endothelial cells (a), human primary endothelial cells (b) and endothelioma cell lines (c). Tubular structure formation was measured by 3D culture of mouse primary endothelial cells (d), human primary endothelial cells (e) or endothelioma cell lines (f). Results are expressed in % of control ± s.e.m, n = 3. *p<0.05, **p<0.01, ***<0.001 using Student's t-test. GKT136901 was used at 10 µM.
Moreover, using a NOX1 expressing vector we observed that NOX1 overexpression (thirteen-fold increase) is sufficient to increase endothelioma cell migration and tube-like structure formation (
These experiments demonstrate that NOX1 is an important protein involved in migration and tube-like structure formation of endothelial cells with no detectable role in cell proliferation.
PPARα is a nuclear hormone receptor with anti-inflammatory functions able to block the angiogenic activity of VEGF
NOX1 deficiency induces an up-regulation of PPAR
To test this hypothesis, we treated NOX1-deficient MLEC with the PPARα antagonist GW6471 and found that the compound restored cell migration and tube formation compared to untreated NOX1-deficient cells (
Taken together, these results indicate that NOX1 promotes endothelial cell migration and sprouting by suppressing PPARα expression and activity.
We then set out to further identify the vascular NOX1 signaling pathways implicated in angiogenesis. Activation of both Akt and ERK1/2 were observed following 10 minutes of stimulation with the angiogenic factor bFGF. In NOX1-deficient MLEC, Akt activation was reduced compared to WT cells, while ERK1/2 activation was unaffected (
Since PPARα is known to inhibit NF-κB activation, we monitored NF-κB activation in NOX1-deficient MLEC in response to VEGF or bFGF stimulation (30 minutes). NOX1-deficient MLEC showed reduced nuclear translocation of the NF-κB p65 subunit compared to WT cells in response to VEGF or bFGF as demonstrated by Western blotting on nuclear extracts and immunofluorescence of endothelial cells (
LMEC NOX1 KO | tEnd siRNA NOX1 | tEnd siRNA NOX1 + GW 6471 | |
Catalase | |||
GPX3 | |||
VE-cadherin | |||
MMP-2 | |||
MMP-9 | |||
uPAR | |||
VEGF | |||
bFGF | |||
VCAM-1 |
Level of expression of target genes in NOX1 deficient primary endothelial cells (MLEC), or silenced cells endothelial cell lines compared to control cells, analyzed by quantitative real-time PCR. Results are expressed in fold increase or decrease ± s.e.m. n = 3.
The above findings led us to investigate whether NOX1 may contribute to tumor progression by promoting tumor vascularization. To this end, we implanted tumorigenic B16F0 melanoma cells or Lewis Lung Carcinoma (LLC1) cells subcutaneously into WT and NOX1-deficient mice. LLC1 cells expressed high level of NOX1 in contrast to B16F0 cells. We observed reduced tumor vascularization in NOX1-deficient animals with B16F0 melanoma but not with LLC1 tumors (
B16F0 (a) or LLC1 (b) tumor cells were injected subcutaneously into WT or NOX1 KO mice. Tumors were allowed to develop for 10 days and the density of tumor vasculature was identified using PECAM-1 staining by measuring the positive (vascular) and total (tumor) area ± s.e.m expressed as percentage. n = 8; WT mice injected with LLC1 tumor cells were then treated with the NOX inhibitor GKT136901 at 40 mg/kg/day by oral administration for 8 days; (c). Graph represents tumor weight in mg ± s.e.m n = 8 per group. (d). Tumor vessel density in tumors from the experiment in (c). Graph shows tumor vascularization expressed in percentage of PECAM-1 positive area ± s.e.m.; (e). Changes in tumor volume in mm3 after therapeutic treatment starting 8 days post tumor cell injection with GKT136901 (black arrows) or anti-VEGFR2 (DC101) (pointed arrows). Tumor volume is measured using a caliper and the formula V = 4/3π(L/2*l/2*w/2). n = 8; (f). Blood vessel density in tumors from the experiment in (e). at day 12. Graph shows tumor vascularization expressed in percentage of PECAM-1 positive area ± s.e.m. WT or PPAR
To assess whether NOX1 may be a valuable target for cancer therapy, we used the inhibitor GKT136901. This inhibitor did not interfere with tumor cell proliferation and apoptosis
To further analyze the link between NOX1 and PPARα
From these observations, we conclude that NOX1 promotes tumor angiogenesis by inhibiting the anti-angiogenic factor PPARα.
Reactive oxygen species are important players in cancer biology
(a) Tumor cells and tumor-infiltrating leukocytes during tumor development produce proangiogenic factors such as VEGF and bFGF. These factors activate the preexisting vessels to form neovessels by sprouting, migration and proliferation. (b) Endothelial cells receive the angiogenic stimulus by fixation of the angiogenic factor to the surface receptor. This interaction initiates a signaling cascade, which leads to NOX1 activation through Rac1. NOX1 dependent ROS inhibits the nuclear hormone receptor PPAR
Inhibition of NOX1 decreases endothelial cell migration. A link between NOX1, ROS and cell migration has been previously described with non-endothelial cells such as colon carcinoma cell lines
Importantly, we do not find evidence for the involvement of the other tested NOX isoforms 2 and 4 in bFGF-induced angiogenesis in vivo. This is in contrast to other reports suggesting an angiogenic role for these NOX isoforms. Angiopoietin-1, VEGF, or thrombin-induced neovascularization in vitro and in vivo have been suggested to be dependent on NOX2
Our results reveal a remarkable specificity of NOX isoforms. Clearly, endothelial cells express NOX1, NOX2, NOX4, and – at least in humans - also NOX5. Yet, their functions appear to be non-redundant. The specificity of NOX isoforms relies on several elements, including different subcellular localization (NOX4 is predominantly intracellular, while NOX1 rather localizes to caveolae
With respect to tumor growth, we observed differences between NOX1-deficient mice and wildtype mice treated with the NOX inhibitor. Indeed, while plug- and tumor-induced angiogenesis was efficiently decreased in the NOX1-deficient mouse and by the NOX inhibitor, the latter was markedly more efficient in decreasing tumor growth. The most likely explanation for this difference is the possibility that NOX enzymes within the tumor cells are contributing to tumor growth and that the NOX inhibitor targets these enzymes. Indeed, there is ample evidence in the literature for a role of NOX enzymes in enhancing growth of tumor cells
Under homeostatic conditions, ROS levels are balanced by scavenger and antioxidant enzymes. As in certain pathologies this balance becomes deregulated and much effort has been put into development of inhibitors of ROS production
We observed an increase of PPARα expression when NOX1 was not expressed; suggesting that NOX1 activity constitutively represses PPARα expression. As previously mentioned, in the absence of NOX1 expression, endothelial cells were less able to migrate and to form tubular structures, we showed that this effect was reversed using an antagonist of PPARα. Moreover, in NOX1-deficient endothelial cells, NFκB was not activated after VEGF or bFGF stimulation. This effect was reversed by treatment with a PPARα antagonist. In addition, anti-oxidants and anti-migratory genes were up-regulated and pro-angiogenic genes down-regulated by NOX-1 deficiency. These differences in gene expression depended on PPARα transactivation and they explain the reduced migratory phenotype and tube formation of NOX1 deficient endothelial cells. PPARα activity, which itself controls PPARα expression, could be directly regulated by the catalytic activity of NOX1. Indeed, PPARα activity is regulated by several post-transcriptional modifications such as phosphorylation or SUMOylation
Currently, the anti-VEGF antibody bevacizumab (Avastin), and several small molecular VEGFR tyrosine kinase inhibitors, are used as anti-angiogenic drugs to treat patients with advanced cancers
In conclusion, we have identified NOX1 as a novel mediator of angiogenesis and as a candidate therapeutic target for anti-angiogenic therapies in cancer.
NOX1, NOX2, NOX1/2, NOX4 deficient mice were inbred on the C57BL/6J background for more than 6 generations. The PPARα null animals were originally described in
Endothelial cell line (thymus derived endothelioma
7 to 10 week old females were injected subcutaneously with 400 ml of growth factor reduced Matrigel supplemented with 500 ng/ml of bFGF. One week later, mice were scanned using Micro-CT (Skyscan-1076). Mice were scanned before and after retro-orbital injection of 400 ml iodinated liposomes (BR22, BracoResearch, Plan-les-Ouates) to visualize the vessel density in the plug as described previously
Murine lung endothelial cells were isolated as described previously
Total RNA from treated cells was extracted using RNeasy minikit (Qiagen). Total RNA was reverse-transcribed with the Superscript III first strand RT-PCR kit (Invitrogen). Quantitative real-time PCR was performed using SybrGreen master mix (Applied Biosystems) on Step one plus Real-time PCR machine (Applied Biosystems). Primer sequences are listed in the
Endothelial cells were seeded on glass slides and stained with dihydroethidium (DHE). Images were captured with an inverted microscope and analyzed with Metafluor imaging software (MDS Analytical Technologies). Quantification was performed by measuring the fluorescence intensity of over minimum 50 endothelial cells per slide.
Membranes from NOX2 expressing PMN cells or from cells overexpressing NOX1 or NOX4 were prepared as previously described
For mouse cells, siRNAs were nucleofected using the Amaxa technology (Lonza). Gene silencing was assessed 48 hours after nucleofection by quantitative RT-PCR. For HUVEC, shRNA vectors (SABiosciences) were nucleofected using Amaxa. Gene silencing was assessed 48 hours after nucleofection by quantitative RT-PCR.
Wound healing assay was performed as described previously
Sprouting assay was performed as described previously
After stimulation, nuclear and cytoplasmic proteins of endothelial cells were extracted according to Tauzin et al.
Cells were stimulated for the indicated time and then lysed in TNT buffer (50 mM Tris, 150 mM NaCl, 0.5% Triton X-100) complemented with protease inhibitor cocktail CLAP and phosphatase inhibitor cocktail [25 mM NaF, 20 mM b-Glycerophosphate, 5 mM HEPES, 2.5 mM EDTA, 0.5 mM Orthovanadate]. Membranes were blocked in PBS containing 0.5% BSA and hybridized with different antibodies. Blots were revealed with peroxidase coupled secondary reagent (Jackson Immunoresearch) followed by ECL and quantified by densitometry using Image J software.
5. 105 of LLC1 or B16F0 were injected subcutaneously on the back of mice. Mice were treated with NOX inhibitor, GKT136901 or vehicle (Carboxymethyl-cellulose, CMC) at 40 mg/kg everyday per os or i.p with anti-VEGFR2 antibody (DC101) at 0.8 mg every 2 days. When the control tumor reached approximately 1 cm in length, mice were sacrificed and the tumor excised, weighed and frozen. Frozen sections of tumors were stained with anti-PECAM-1 antibody (rat monoclonal,
All statistical analysis was performed using Anova on multiple variable analyses and Student's t-test on paired analyses. *(p = 0,05), **(p = 0,01), ***(p = 0,001).
Inhibitory effect of the inhibitor GKT 136901 on ROS producing enzymes, redox-sensitive enzymes and others proteins.
(0.03 MB DOC)
MLEC isolation from WT and NOX deficient mice. a. Flow cytometry analysis of endothelial surface molecules on isolated MLEC. PECAM-1, VE-Cadherin, ICAM-2 and Meca-32 expression level in MLEC. b. PECAM-1 immunofluorescence staining of WT, NOX1 KO and NOX4 KO MLEC. Nuclei in blue (DAPI), and PECAM-1 in purple (Cy5). Images were acquired with a 40x/1.3 numeric aperture lens and analyzed using LSM510 Meta microscope (Carl Zeiss).
(6.76 MB TIF)
Inhibition of NOX-dependent ROS production by GKT136901 and DPI. a. Concentration-response curves of GKT136901 on NOX1 (x), NOX2 ▴), NOX4 (◊) and Xanthine Oxidase (XO) (□). b. Concentration-response curve of DPI on NOX1 (x), NOX2 (▴), NOX4 (◊) and Xanthine Oxidase (XO) (□) Results are from one experiment performed in triplicate, representative of four performed. Values are presented as means ± s.e.m.
(2.02 MB TIF)
NOX dependant ROS blocking agents efficiently block endothelial cell migration and branching capacities. a. Migration of endothelial cells was analyzed by a wound-healing assay in presence of different inhibitors that block NADPH dependant ROS production. b. Tubular structure formation was measured by 3D culture using the mouse endothelial cell line in presence of different inhibitors that block NADPH dependant ROS production. Results are expressed in % of control ± s.e.m, n = 3.
(5.89 MB TIF)
NOX1 over-expression enhances endothelial cell migration and tube-like structure formation. a. In vitro migration was analyzed by wound-healing assay using endothelioma cell lines transfected with NOX1 expressing vector. b. Tubular structure formation was measured by 3D culture of endothelioma cell lines transfected with NOX1 expressing vector. Results are expressed in % of control ± s.e.m. *p< 0.05 using Student's t-test.
(5.88 MB TIF)
AKT but ERK 1/2 activation is affected by NOX1 deficiency. NOX1-deficient MLEC does not activate Akt after bFGF stimulation but present no difference in ERK1/2 activation. a. Western blot analysis of Akt phosphorylation in WT and NOX1-deficient MLEC after 10 min stimulation with 20 ng/ml of bFGF. The graph shows the abundance of phosphorylated Akt relative to total Akt ± s.e.m as determined by densitometry. n = 3. b. Western blot analysis of ERK1/2 phosphorylation in WT and NOX1-deficient MLEC stimulated for 10 min with 20 ng/ml of bFGF. The graph shows the abundance of phosphorylated ERK1/2 relative to total ERK1/2 ± s.e.m as determined by densitometry. n = 3.
(6.13 MB TIF)
NF-κB nuclear translocation is inhibited in the absence of NOX1 and dependent on PPARα activation. VEGF or b-FGF stimulation of endothelial cells induced p65 NF-κB translocation into the nucleus. This nuclear translocation is not observed in NOX1-deficient cells but restored by PPARα antagonist treatment (GW6471). Immunofluorescence, anti-p65 NF-κB of MLEC (a) and endothelioma cell lines (b) stimulated with VEGF or bFGF in presence or absence of GW6471 (10mM). NF-κB in green (Alexa 488), nuclei in blue (DAPI). Scale bar represent 20 mm. Images were acquired with a 40x/1.3 numeric aperture and analyzed using LSM510 confocal microscope (Carl Zeiss).
(7.09 MB TIF)
Effect of GKT 136901 on tumor cells. a. LLC1 and B16F0 cell proliferation was measured by EdU incorporation and propidium iodide staining of DNA content, 24h after incubation with 10 μM of GKT136901. b. LLC1 apoptosis was measured by AnnexinV/PI staining after 24h of incubation with 10 μM of GKT136901. c. ROS levels produced by LLC1 are inhibited by GKT136901. ROS production was quantified by DHE substrate 1h after incubation with 10 μM of the inhibitor. *** p<0.001 (student t-test).
(6.11 MB TIF)
Non toxic effect GKT 136901 on mice organs. Heart (a), Kidney (b), Liver (c) and Lung (d) of mice treated orally with vehicle or with vehicle plus GKT136901 inhibitor at 40 mg/kg per day during 8 days, stained by Hematoxilin/Eosin. Scale bars represent 100 μm on the full picture and 20 μm on the zoom. Images were acquired with a 20x/0.8 numeric aperture and analyzed using Mirax (Carl Zeiss).
(8.38 MB TIF)
Real-time PCR primer sequence list.
(0.05 MB DOC)
We would like to thank C. Chaponnier (Centre medical Universitaire, Geneva) for the gift of the anti-Actin antibody; N. Deblon (Centre medical Universitaire, Geneva) for advice concerning PPARa; N. Imaizumi (CePO and University of Lausanne, Lausanne) for her assistance in Matrigel plug assay; B. Lee (Centre medical Universitaire, Geneva) for his assistance in the manuscript editing.