ROS-induced ROS release orchestrated by Nox4, Nox2, and mitochondria in VEGF signaling and angiogenesis

Abstract

Reactive oxygen species (ROS) derived from NADPH oxidase (NOX) and mitochondria play a critical role in growth factor-induced switch from a quiescent to an angiogenic phenotype in endothelial cells (ECs). However, how highly diffusible ROS produced from different sources can coordinate to stimulate VEGF signaling and drive the angiogenic process remains unknown. Using the cytosol- and mitochondria-targeted redox-sensitive RoGFP biosensors with real-time imaging, here we show that VEGF stimulation in human ECs rapidly increases cytosolic RoGFP oxidation within 1 min, followed by mitochondrial RoGFP oxidation within 5 min, which continues at least for 60 min. Silencing of Nox4 or Nox2 or overexpression of mitochondria-targeted catalase significantly inhibits VEGF-induced tyrosine phosphorylation of VEGF receptor type 2 (VEGFR2-pY), EC migration and proliferation at the similar extent. Exogenous hydrogen peroxide (H₂O₂) or overexpression of Nox4, which produces H₂O₂, increases mitochondrial ROS (mtROS), which is prevented by Nox2 siRNA, suggesting that Nox2 senses Nox4-derived H₂O₂ to promote mtROS production. Mechanistically, H₂O₂ increases S36 phosphorylation of p66Shc, a key mtROS regulator, which is inhibited by siNox2, but not by siNox4. Moreover, Nox2 or Nox4 knockdown or overexpression of S36 phosphorylation-defective mutant p66Shc(S36A) inhibits VEGF-induced mtROS, VEGFR2-pY, EC migration, and proliferation. In summary, Nox4-derived H₂O₂ in part activates Nox2 to increase mtROS via pSer36-p66Shc, thereby enhancing VEGFR2 signaling and angiogenesis in ECs. This may represent a novel feed-forward mechanism of ROS-induced ROS release orchestrated by the Nox4/Nox2/pSer36-p66Shc/mtROS axis, which drives sustained activation of angiogenesis signaling program.

reactive oxygen species (ROS) such as superoxide anion (${\mathrm{O}}_{\mathrm{2}}^{\mathrm{-}}$) and hydrogen peroxide (H₂O₂) at physiological level function as signaling molecules to mediate many biological responses. In endothelial cells (ECs), vascular endothelial growth factor (VEGF) induces angiogenesis by stimulating EC migration and proliferation primary through the VEGF receptor type 2 (VEGFR2). Using EC-specific catalase-overexpressing mice, we reported that endothelium-derived H₂O₂ is required for postischemic neovascularization (55). In ECs, NADPH oxidases (NOXs), especially Nox2 and Nox4, are important sources of ${\mathrm{O}}_2^{-}$ and H₂O₂ (4, 13, 18, 34, 44). Using global Nox4^−/− or Nox2^−/− mice or EC-specific Nox4 or dominant-negative Nox4-overexpressing mice, Nox2 or Nox4 or their regulators are shown to play a critical role in ROS-dependent VEGFR2 signaling as well as postnatal angiogenesis (1, 5, 7, 11, 12, 18, 27, 48, 50, 54, 58, 60, 61, 65). Nox2 interacts with regulatory subunit p22^phox, and its activation requires cytosolic components such as p47^phox and small GTPase Rac1 to produce ${\mathrm{O}}_2^{-}$ that is rapidly converted to H₂O₂. By contrast, Nox4 also interacts with p22^phox but its activation increases intracellular H₂O₂ without cytosolic organizers p47^phox or Rac1 (4, 39, 51). Both Nox2 and Nox4 are shown to exist in intracellular compartments or plasma membranes (8, 20, 56, 57). However, the functional relationship between Nox2 and Nox4 in VEGF-induced angiogenesis is poorly understood.

Mitochondria in ECs have been also implicated in ROS signaling (6, 19, 66) and are shown to be involved in H₂O₂-induced transactivation of VEGFR2 (10) or VEGF-induced cell migration in a Rac1-dependent manner in ECs (62). One of the key regulators of mitochondrial ROS (mtROS) is p66Shc, which is an adaptor protein and is phosphorylated at Ser36 in NH₂-terminal CH2 domain by oxidative stress, leading to mitochondrial H₂O₂ generation by catalyzing electron transfer to oxygen through the oxidation of cytochrome c (23, 45). We previously reported that the nonphosphorylated form of p66Shc is involved in VEGF-induced rapid Rac1 activation and subsequent H₂O₂ production which is required for VEGFR2 phosphorylation at caveolae/lipid rafts and angiogenic responses in ECs (40). We also demonstrated that VEGF stimulation increases phosphorylation of p66Shc at Ser36 through protein kinase C (PKC) or ERK/JNK pathways in ECs (40). However, the specific role of pSer36-p66Shc in VEGF-induced mtROS production and angiogenesis has not been investigated.

Fundamental questions remain how highly diffusible H₂O₂ derived from different sources NOXs and mitochondria can efficiently promote VEGF signaling and angiogenesis. Furthermore, the temporal-spatial relationship for VEGF-induced ROS production from the different subcellular compartments has not been investigated. Of note, “ROS-induced ROS release” has been proposed as a mechanism for ROS amplification and localized ROS production (14, 67). H₂O₂ is shown to activate Nox2 to produce ${\mathrm{O}}_2^{-}$ in fibroblasts and smooth muscle cells with unknown mechanism (35). However, the role and mechanism of ROS-induced ROS release in VEGF signaling in ECs are entirely unknown.

In this study, using compartment-specific ratiometric redox-sensitive green fluorescent proteins (RoGFP) (63) and other redox-sensitive probes, we demonstrate that Nox2 senses Nox4-derived H₂O₂ signal to promote mitochondrial ROS (mtROS) production via pSer36-p66Shc formation, thereby enhancing VEGFR2 activation, EC migration and proliferation. This may represent a novel ROS-induced ROS release mechanism orchestrated by the Nox4/Nox2/p-p66Shc/mtROS axis, which drives switching from a quiescent to an angiogenic phenotype.

MATERIALS AND METHODS

Materials.

Antibodies to phospho-VEGFR2 (pY1175, no. 2478) and total VEGFR2 (no. 9698) were from Cell Signaling. Anti-pSer36-p66Shc and Shc antibodies were from (Calbiochem no. 566807 and BD Bioscience no. 610879, respectively). Anti-Nox2 antibody was from BioLegend (no. 650102), which has been validated using Nox2-null ECs (31), and anti-Nox4 antibody was from Novus (NB110-58849), which has been validated using Nox4-overexpressing ECs (12) or null cells (28). Human recombinant VEGF165 was from R&D Systems (no. 293-VE). Oligofectamine and Opti-MEMI Reduced-Serum Medium were from Invitrogen. Other materials were purchased from Sigma.

Cell culture.

Human umbilical vein endothelial cells (HUVECs, Lonza no. CC-2519 from pooled donors) were grown in EndoGro (Millipore) containing 5% fetal bovine serum (FBS, Atlanta Biologicals) and supplement, and used at 90~95% confluency at passage 4–6. For serum starvation, cells were incubated with 0.5% FBS in EndoGro overnight. We used HUVECs because they were the most widely and commonly used human ECs to study the role of ROS in VEGF-induced angiogenesis, as reported previously (18, 27, 40, 41, 62, 65).

Redox biosensor RoGFP imaging.

HUVECs grown on glass-bottom 35-mm diameter culture dishes were infected with adenovirus expressing cytosol-targeted redox-sensitive RoGFP biosensor (Cyto-RoGFP) or mitochondria-targeted RoGFP biosensor (Mito-RoGFP), which were provided by Dr. Paul Shumacker (63). This RoGFP protein contains two engineered cysteine thiols and has excitation maxima at 400 and 484 nm, with emission at 525 nm. RoGFP exhibits reciprocal changes in intensity at the two excitation maxima, and its ratiometric characteristics render it insensitive to expression levels during changes in redox conditions (63). In addition, Mito-RoGFP was designed with an addition at the 5′-end of a 48-bp region encoding the mitochondrial targeting sequence from cytochrome oxidase subunit IV (63). Before experiments, medium was changed to HEPES-buffered solution. Dishes were placed in a recording chamber mounted on the stage of an inverted fluorescence microscope (Eclipse Ti-E; Nikon, Tokyo, Japan) equipped with an objective lens (S Fluor ×40 /0.90 for Cyto-RoGFP and Plan Apo VC ×100 /1.40 oil immersion for Mito-RoGFP; Nikon) and an EM-CCD camera (Evolve; Photometrics, Tucson, AZ), as previously reported (52). Cells were excited with 405 nm and 485 nm wavelengths (D405/×20 and D485/×25 filters, respectively; Chroma Technology, Bellows Falls, VT) by a xenon arc lamp (Lambda LS; Sutter Instrument, Novato, CA) and an optical filter changer (Lambda 10-B; Sutter Instrument). Emission of RoGFP was collected through a dichroic mirror (505 nm; Chroma Technology) and an emission filter (510–560 nm; Chroma Technology). Fluorescence signals in response to VEGF, dithiothreitol (DTT), or H₂O₂ were imaged with NIS Elements 3.2 software (Nikon) and measured as the ratio of fluorescence intensities (F₄₀₅/F₄₈₅) every 10 s. The recording chamber was continuously perfused with external solution at a flow rate of 2 ml/min using a mini-pump (model 3385; Control, Friendswood, TX). Fluorescence images were recorded in real-time at 37°C using an automatic temperature controller (TC-344B, Warner Instruments, Holliston, MA) (52). Of note, detection of Cyto- or Mito-RoGFP oxidation in real-time without photo-breaching or background noise was extremely difficult. However, this problem could be solved by using this fluorescence microscope with a high sensitivity and a low noise photomultiplier. We imaged the Cyto-RoGFP and Mito-RoGFP fluorescence signals from at least five individual cells in the field and repeated at least three times.

ROS measurement.

To detect intracellular ROS, HUVECs stimulated with VEGF were incubated with 20 μM CM-H₂DCFDA [5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester; Invitrogen] for 6 min at 37°C, fixed with 4% paraformaldehyde for 10 min at room temperature, and then mounted with VECTASHIELD Mounting Medium with DAPI. DCF fluorescence was measured by confocal microscopy (Zeiss) using the same exposure condition in each experiment. Relative DCF fluorescence with DAPI-positive cells were analyzed using ImageJ (National Institutes of Health, Bethesda, MD). We confirmed that DCF fluorescence was abolished by incubation with polyethylene glycol (PEG)-conjugated catalase (100 U/ml for 12 h), suggesting that DCF signal mainly detects intracellular H₂O₂, as we reported (40). Of note, whether PEG-catalase enters the cell depends on the concentration and incubation time. Our preliminary study found that VEGF-induced DCF fluorescence was not significantly inhibited by short-term incubation of PEG-catalase (100 U/ml for 1 h) or non-cell-permeable catalase (100 U/ml for 12 h). To detect mitochondrial ${\mathrm{O}}_2^{-}$, cells were incubated with 5 μM MitoSox Red (Invitrogen) for 10 min and live cell images were taken, using confocal microscopy (Zeiss), as previously described with minor modifications (47). We confirmed that MitoSox fluorescence was abolished by mitochondria-targeted SOD mimetic, Mito-TEMPO (Enzo). As an alternative method to detect mtROS, cells were incubated with MitoTracker CM-H2TMRos (Invitrogen), which was abolished by Mito-TEMPO or mitochondria-targeted catalase (Mito-catalase). Relative fluorescence intensity in cells was measured using ImageJ.

Immunoprecipitation and immunoblotting.

Growth-arrested HUVECs were stimulated with VEGF or H₂O₂, and cells were lysed in lysis buffer [50 mM HEPES (pH 7.4), 5 mM EDTA, 120 mM NaCl, 1% Triton X-100, protease inhibitors (10 μg/ml aprotinin, 1 mmol/l phenylmethylsulfonyl fluoride, 10 μg/ml leupeptin) and phosphatase inhibitors (50 mmol/l sodium fluoride, 1 mmol/l sodium orthovanadate, 10 mmol/l sodium pyrophosphate)]. Cell lysates were used for immunoprecipitation and immunoblotting, as we described (27, 40, 65).

Quantitative real-time PCR.

Total RNA from HUVECs were isolated using TRI Reagent (Molecular Research Center Inc.). Reverse transcription was carried out using high capacity cDNA reverse transcription kit (Applied Biosystems) using 2 μg of total RNA. Quantitative PCR was performed with the ABI Prism 7000, the SYBR Green PCR kit (Qiagen) and the specific primers for human genes. Nox2 primers, forward 5′-AGCTATGAGGTGGTGATGTTAGTGG-3′, reverse 5′-TGCAATATTTGTACCAGACTGACTTGAG-3′; p22^phox primers, forward 5′-GTACTTTGGTGCCTACTCCA-3′, reverse 5′-CGGCCCGAACATAGTAATTC-3′; Nox4 primers, forward 5′-CTCAGCGGAATCAATCAGCTGTG-3′ reverse 5′-AGAGGAACACGACAATCAGCCTTAG-3′. Samples were all run in triplicate to reduce variability. Gene expression was normalized and is expressed as fold change relative to 18S (internal control), as we reported (7).

siRNA and adenovirus transfection.

HUVECs were grown to 50% confluence in 100-mm dishes and transfected with 20 nM control siRNA or validated Nox2 siRNA (sc-35503) or p22^phox siRNA (sc-36149) from Santa Cruz (43) using Oligofectamine, as we reported (40, 65). Cells were used for experiments at 48 h after transfection. For adenovirus transfection, HUVECs were infected with adenovirus expressing LacZ (Ad.control); siNox4 or Nox4-wild type (WT) (from Dr. Junichi Sadoshima, Rutgers New Jersey Medical School, Newark, NJ) (2); mitochondria-targeted catalase (Mito-catalase), or p66Shc(S36A) (from Addgene); Cyto-RoGFP or Mito-RoGFP from Dr. Paul Schumacker (Northwestern University, Chicago, IL). Cells were used for experiments at 36 h after transfection, as reported (65).

Modified Boyden chamber migration assay.

HUVECs were serum starved with 0.5% FBS containing EndoGro culture medium (Millipore) overnight. Cells (60,000) were seeded on 0.1% gelatin-coated Transwell inserts with 8-μm pore size and then transferred to the lower chamber in a 24-well plate containing medium with 0.5% FBS (baseline, negative control) or 0.5% FBS + 20 ng/ml VEGF (stimulant, positive control). The wells were incubated for 6 h in 37°C CO₂ incubator and the Transwell inserts were then removed from the 24-well plate. Remaining cells that had not migrated from the top of the insert membrane were removed with wet cotton swabs and fixed with 4% paraformaldehyde for 10 min and stained with crystal violet. Migrated cells through the membranes in the bottom of Transwell inserts, not the lower chamber, were counted, as we reported (27, 40, 41, 65).

Cell proliferation assay.

HUVECs (10⁵ cells) were seeded in 5% FBS containing complete EndoGro medium on 0.1% gelatin-coated six-well plate. Upon adherence, the cells were washed twice with 1× PBS and incubated in medium containing only 0.5% FBS with or without (baseline) 20 ng/ml VEGF for 48 h. The live cells, but not dead or apoptotic cells, which were stained by Trypan blue, were analyzed by counting, as we reported (27, 40, 41, 65). To analyze the data obtained by the migration and proliferation assay, we used unstimulated cells as basal control and evaluated the VEGF-induced responses by fold increase from the basal responses. In addition, we confirmed that adherent cell numbers after plating were not different among treatment groups used in this study.

Statistical analysis.

Results are expressed as means ± SE. Statistical significance was assessed by Student's paired two-tailed t-test or analysis of variance on untransformed data, followed by comparison of group averages by contrast analysis using the Super ANOVA statistical program (Abacus Concepts, Berkeley, CA). A P value of <0.05 was considered to be statistically significant.

RESULTS

VEGF increases cytosolic RoGFP oxidation followed by mitochondrial RoGFP oxidation in ECs.

To examine the temporal and spatial changes of cytosolic and mitochondrial ROS production in ECs in response to VEGF, HUVECs were infected with adenovirus expressing Cyto-RoGFP) (Fig. 1, A–C) or Mito-RoGFP (Fig. 2, A–C). Serum-starved ECs were exposed to Hanks’ balanced salt solution (HBSS), showing the stable low level of basal redox status, and then perfused with VEGF (50 ng/ml)-containing HBSS for 20 min. This was followed by media containing reducing agent dithiothreitol (DTT, 1 mmol/l) for 10 min to fully reduce the sensor, and then switched to HBSS containing H₂O₂ (500 μM) for 20 min to fully oxidize the sensor. Ratiometric representative images (Figs. 1A and 2A), the fluorescence ratio (Figs. 1B and 2B), and the reciprocal changes in fluorescence intensity for the two excitation maxima (485 and 405 nm) (Fig. 1C) of RoGFP were monitored.

Fig. 1.

VEGF increases cytosolic RoGFP oxidation in ECs. HUVECs infected with adenovirus expressing Cyto-RoGFP (A, B, and C) were superfused with media containing VEGF (50 ng/ml) for 20 min, followed by media containing reducing agent dithiothreitol (DTT, 1 mmol/l) for 10 min to fully reduce the sensor, and then with media containing oxidizing reagent, H₂O₂ (500 μM) for 20 min to fully oxidize sensor. A: representative ratiometric fluorescence images of Cyto-RoGFP biosensors. The numbers (1–5) indicate different cells on the same field. B: representative time course of the fluorescence intensity ratio from representative individual ECs imaged in A. C: representative time course of the reciprocal changes in Cyto-GFP fluorescence intensity at the two excitation maxima (485 and 405 nm) in cell no. 1 imaged in A. The experiments were performed at least three times.

Fig. 2.

VEGF increases mitochondrial RoGFP oxidation in ECs. HUVECs infected with adenovirus expressing Mito-RoGFP (A and C) were superfused with media containing VEGF (50 ng/ml) for 20 min, followed by media containing reducing agent dithiothreitol (DTT, 1 mmol/l) for 10 min to fully reduce the sensor, and then with media containing oxidizing reagent, H₂O₂ (500 μM) for 20 min to fully oxidize sensor. A: representative ratiometric fluorescence images of Mito-RoGFP biosensors. B: representative time course of the fluorescence intensity ratio from representative individual ECs imaged in A. C: HUVECs infected with adenovirus expressing Mito-RoGFP with empty virus control (Ad.Ctrl) or Ad.Mito-catalase were superfused with media containing VEGF (50 ng/ml) for 20 min. Representative ratiometric fluorescence images of Mito-RoGFP biosensors (top) and the averaged time course of the fluorescence intensity ratio from 5 different individual ECs (bottom). The experiments were performed at least three times.

Figure 1 shows that VEGF stimulation rapidly increased Cyto-RoGFP fluorescence within 1 min with a peak at around 10 min and remained above basal at least for 60 min, which was inhibited by NOX inhibitor, apocynin (data not shown). By contrast, Fig. 2 shows that VEGF stimulation gradually increased Mito-RoGFP fluorescence within 5 min, which peaked at around 10–15 min and remained above basal for up to 60 min. We confirmed that VEGF-induced Mito-RoGFP oxidation was abolished by overexpression of mitochondria-targeted catalase (Mito-catalase, Fig. 2C), suggesting that mitochondrial H₂O₂ is involved. VEGF-induced increase in both Cyto-RoGFP and Mito-RoGFP fluorescence was reversible, because DTT exposure rapidly reduced while H₂O₂ rapidly increased RoGFP fluorescence. Of note, individual cells (marked as nos. 1–5 in Figs. 1, A and B and 2, A and B) exhibited the synchronized and almost identical time course for changes in both Cyto- or Mito-RoGFP fluorescence in response to VEGF, DTT, and H₂O₂.

In parallel, using confocal microscopy, we confirmed that VEGF stimulation rapidly increased DCF fluorescence in a time-dependent manner, which was abolished by a cell-permeable PEG-catalase (100 U/ml for 12 h) (Fig. 3A). Note that VEGF-induced DCF fluorescence was not significantly inhibited by either short-term incubation of PEG-catalase (100 U/ml for 1 h) or non-cell-permeable catalase (100 U/ml for 1 h or 12 h) (Fig. 3B). Thus, these results suggest that PEG-catalase (100 U/ml for 12 h) can enter the cells and that DCF signals mainly reflect intracellular H₂O₂ production. Of note, VEGF-induced increase in DCF fluorescence was cotemporaneous with that in Cyto-RoGFP fluorescence. Since availability of fluorescence microscopy equipped with high sensitivity and low background noise EM-CCD camera to perform real-time ratiometric RoGFP imaging was limited, we used DCF and mtROS-specific fluorescence probes, as described in materials and methods and below, to address the role of Nox4 and Nox2 in VEGF-induced cytosolic and mitochondrial ROS production in the following studies.

Fig. 3.

Knockdown of Nox2 or Nox4 inhibits VEGF-induced ROS production in ECs. A: HUVECs were stimulated with VEGF (20 ng/ml) for indicated times, or pretreated with PEG-catalase (100 U/ml) for 12 h and then stimulated with VEGF for 5 min. These cells were used to measure dichlorofluorescein (DCF) fluorescence with DAPI staining imaged by confocal microscopy. B: HUVECs were treated with PEG-catalase or catalase (100 U/ml) for 1 h or 12 h, and then stimulated with VEGF (20 ng/ml) for 5 min to measure DCF fluorescence. Graph shows the fold increase of DCF fluorescence over the basal without VEGF stimulation. C, D, and E: HUVECs were transfected with siRNAs for control (siCtrl) or Nox2 (siNox2) or Nox4 (siNox4). In C, cells were stimulated with VEGF (20 ng/ml) for 5 min and used to measure DCF fluorescence with DAPI staining. In D, cells were used to measure mRNA levels of Nox2 and Nox4 using quantitative RT-PCR (n = 3, *P < 0.05, vs. control siRNA). In E, cells were used to measure Nox2 or Nox4 protein expression.

Nox4 or Nox2- or mitochondria-derived ROS are involved in VEGF-induced VEGFR2 phosphorylation and angiogenesis in ECs.

To compare the role of Nox2 and Nox4 in VEGF-induced ROS production and angiogenic responses, we examined the effects of Nox2 or Nox4 knockdown using their specific siRNAs in HUVECs. Figure 3C shows that VEGF-induced DCF fluorescence was almost completely abolished by Nox2 or Nox4 siRNAs which specifically reduced expression of Nox2 or Nox4 mRNAs (Fig. 3D) and proteins (Fig. 3E) without affecting the other. Furthermore, Nox2 or Nox4 siRNAs almost completely inhibited VEGF-induced EC migration (Fig. 4A) and proliferation (Fig. 4B). To determine the underlying mechanisms, we next examined the role of Nox2 and Nox4 in VEGF-induced tyrosine phosphorylation of VEGFR2 (VEGFR2-pY), the most proximal signaling event activated following ligand binding to the receptor. As shown in Fig. 4C, there was very low level of basal phosphorylation while VEGF stimulation rapidly and significantly increased VEGFR2-pY1175 with a peak at 5 min, which gradually decreased to above the basal levels within 30 min. We found that Nox2 or Nox4 siRNAs significantly inhibited VEGF-induced VEGFR2-pY (Fig. 4D) without affecting expression of integrin β1 (Fig. 4E) which is involved in EC migration. Of note, VEGF-induced p-VEGFR2 normalized by total VEGFR2 protein was still inhibited by Nox2 or Nox4 siRNAs, while Nox2 siRNA, but not Nox4 siRNA, slightly reduced basal VEGFR2 protein (Fig. 4D). These findings suggest that ROS derived from both Nox2 and Nox4 are involved in VEGFR2 phosphorylation and that Nox2, but not Nox4, may have additional effect to regulate VEGFR2 expression. We then examined the role of mtROS in VEGF-induced VEGFR2 signaling and angiogenesis in ECs. Figure 5, A and B, shows that overexpression of Mito-catalase almost completely inhibited VEGF-induced EC migration (Fig. 5A) and proliferation (Fig. 5B). Moreover, VEGF-induced VEGFR2-pY, which increased with a peak at 5 min and then decreased to above basal levels within 30 min, was partially but significantly inhibited by overexpression of Mito-catalase (Fig. 5C). Taken together, these results suggest that Nox4, Nox2, or mitochondria-derived H₂O₂ is required for VEGF-induced VEGFR2 activation and angiogenic responses to a similar extent.

Fig. 4.

Nox2 and Nox4 are involved in VEGF-induced angiogenic responses as well as VEGFR2 activation in ECs. A, B, D, E: HUVECs were transfected with siCtrl, siNox2, or siNox4. In A, cell migration stimulated with VEGF (20 ng/ml) for 6 h was measured using modified Boyden chamber method. Bar graph represents averaged migrated cell number per 5 fields (×10) expressed as fold change over the basal (n = 3, *P < 0.05; ns, not significant). In B, cells were cultured in 0.5% FBS containing medium with or without VEGF (20 ng/ml) for 48 h and the cell number was counted (n = 3, *P < 0.05). C: serum-starved HUVECs were stimulated with VEGF (20 ng/ml) for indicated time, and lysates were immunoblotted with anti-VEGFR2-pY1175 or total VEGFR2 antibodies (Abs). Middle blot shows longer exposure of membrane, showing the low level of basal phosphorylation. In D, lysates from cells stimulated with VEGF for 5 min were immunoblotted with anti-VEGFR2-pY1175 or total VEGFR2 antibodies (Abs). Graph represents the averaged fold change of p-VEGFR2 per total VEGFR2 over the basal (n > 3, *P < 0.05). In E, cells were used to measure integrin β1 protein expression.

Fig. 5.

Mitochondrial ROS are involved in VEGF-induced angiogenic responses as well as VEGFR2 activation in ECs. HUVECs were infected with adenovirus expressing mitochondria-targeted catalase (Ad.Mito-catalase) or control empty virus (Ad.Ctrl). A: cell migration stimulated with VEGF (20 ng/ml) for 6 h was measured using modified Boyden chamber method. Bar graph represents averaged migrated cell number per 5 fields (×10) expressed as fold change over the basal (n = 3, *P < 0.05). B: cells were cultured in 0.5% FBS containing medium with or without VEGF (20 ng/ml) for 48 h and the cell number was counted (n = 3, *P < 0.05). C: lysates from cells stimulated with VEGF for indicated times were immunoblotted with anti-VEGFR2-pY1175 or total VEGFR2 or catalase Abs. Graph represents the averaged fold change of p-VEGFR2 per total VEGFR2 over the basal (n > 3, *P < 0.05).

Nox2 senses Nox4-derived H₂O₂ to promote mtROS production in ECs.

We next examined how Nox4- or Nox2- and mitochondria-derived ROS can coordinate to promote VEGF-induced ROS production. It is shown that Nox4 activation increases intracellular H₂O₂ (4, 39, 51) and that H₂O₂ activates Nox2 to produce ${\mathrm{O}}_2^{-}$ in nonphagocytic cells such as fibroblasts and vascular smooth muscle cells (VSMCs) (35). We found that VEGF stimulation initially increased cytosolic RoGFP, followed by Mito-RoGFP oxidation (Figs. 1 and 2). We thus examined whether Nox4-derived H₂O₂ can activate Nox2 to produce ${\mathrm{O}}_2^{-}$, which is converted to H₂O₂ and which in turn promotes mtROS production. Figure 6, A and B, shows that Nox4 overexpression increased both DCF (Fig. 6A) and MitoSox fluorescence (Fig. 6B), which were inhibited by Nox2 siRNA. Moreover, exogenous H₂O₂ at 500 μM also increased MitoSox (Fig. 6C), which was significantly inhibited by Nox2 siRNA or mitochondria-targeted SOD mimetic Mito-TEMPO, but not by Nox4 siRNA. These results suggest that H₂O₂ or Nox4 stimulation increases mitochondria-derived ${\mathrm{O}}_2^{-}$ in a Nox2-dependent manner. Preliminary studies found that exogenous H₂O₂ increased DCF or MitoSox fluorescence in a dose (100–500 μM)- and time-dependent manner and that 500 μM H₂O₂ was required to increase the same extent of mtROS production as 20 ng/ml VEGF stimulation. Furthermore, we confirmed that 500 μM H₂O₂ for 20 min did not induce apoptosis in ECs. Thus, these results suggest that Nox2 senses Nox4-derived H₂O₂ to promote mtROS production in ECs (Fig. 6D), demonstrating the ROS-induced ROS release.

Fig. 6.

Nox4 overexpression- or H₂O₂ -induced mitochondrial ROS production is inhibited by Nox2 knockdown in ECs. A and B: HUVECs were transfected with siCtrl or siNox2 for 24 h, and then infected with adenovirus expressing Nox4 (Ad.Nox4) or Ad.control (Ad.Ctrl) for 24 h. Cells were used to measure DCF fluorescence (A) or MitoSox fluorescence (B). C: HUVECs transfected with siCtrl or siNox2 or siNox4 or treated with Mito-TEMPO (50 μM) were stimulated with H₂O₂ (500 μM) for 20 min and then used to measure MitoSox fluorescence. Graph represents averaged MitoSox fluorescence intensity expressed as fold increase over the basal (n = 3, *P < 0.05). D: scheme showing that Nox4-derived H₂O₂ activates Nox2 that produce ${\mathrm{O}}_2^{-}$/H₂O₂, which in turn promotes mitochondrial ROS (mtROS) production in ECs.

H₂O₂ activates the Nox2/pS36-p66Shc axis to promote mtROS production in ECs.

We next addressed how H₂O₂ derived from the Nox4/Nox2 axis increases mtROS in ECs. It is shown that p66Shc is phosphorylated at Ser36 by oxidative stress, which in turn promotes mtROS generation in other systems (23, 45). We reported that p66Shc knockdown with siRNA in ECs inhibits VEGF-induced DCF fluorescence (H₂O₂), EC migration and proliferation and that VEGF stimulation increases pSer36-p66Shc formation (40); however, a specific role of pSer36-p66Shc in these responses was not explored. H₂O₂ stimulation rapidly increased pSer36-p66Shc within 5 min and gradually increased at least for 60 min (Fig. 7A), which was significantly inhibited by Nox2 siRNA, but not by Nox4 siRNA (Fig. 7B). Moreover, overexpression of Ser36 phosphorylation-defective mutant, p66Shc(S36A), significantly inhibited the H₂O₂-induced mtROS production (Fig. 7C) without affecting Nox2 or Nox4 protein levels (Fig. 7D). These results suggest that H₂O₂ presumably derived from Nox4 may activate the Nox2/pS36-p66Shc axis to promote mtROS production in ECs (Fig. 7E).

Fig. 7.

H₂O₂-induced pSer36-p66shc through Nox2, but not Nox4, contributes to mitochondrial ROS production in ECs. A: HUVECs were stimulated with H₂O₂ at 500 μM for indicated times (0, 5, 15, 30, 60 min) and lysates were used for immunoprecipitation (IP) with anti-Shc Ab, followed by immunoblotting (IB) with anti-pS36-p66Shc or Shc Abs. B: HUVECs transfected with siCtrl or siNox2 or siNox4 were stimulated with H₂O₂ (500 μM) for 20 min, and lysates were IP with anti-Shc Ab, followed by IB with anti-pS36-p66Shc or Shc Abs. Graph represents the averaged fold change of pS36-p66Shc/total p66Shc over the basal (n = 3, *P < 0.05). C and D: HUVECs infected with adenovirus expressing p66shc (S36A) (Ad.p66Shc-S36A) or Ad.control (Ad.Ctrl) were stimulated with H₂O₂ (500 μM) for 20 min. In C, cells were used to measure MitoSox fluorescence. Graph represents averaged MitoSox fluorescence intensity expressed as fold increase over the basal (n = 3, *P < 0.05). In D, cells were used to measure protein expression of p66Shc, Nox2, Nox4 or actin (loading control). E: scheme showing that Nox4-derived H₂O₂ activates Nox2 which produces ${\mathrm{O}}_2^{-}$/H₂O₂, thereby increasing pS36-p66Shc, which promotes mtROS production in ECs.

VEGF-induced pSer36-p66shc mediated through NOX is involved in mtROS production, p-VEGFR2 and angiogenesis in ECs.

We then examined the role of VEGF-induced pSer36-p66Shc in mtROS production and found that overexpression of p66Shc(S36A) significantly inhibited VEGF-induced MitoSox fluorescence (Fig. 8A), which was abolished by Mito-TEMPO (Fig. 8B). Furthermore, we also used Mito-Tracker Orange CM-H2RMRos, which becomes fluorescent upon its oxidation in live cells and accumulates in mitochondria (29). Figure 8C shows that VEGF-induced Mito-Tracker CM-H2RMRos fluorescence was inhibited by overexpression of p66Shc(S36A) or Mito-catalase. These results suggest that VEGF stimulation increases both ${\mathrm{O}}_2^{-}$ and H₂O₂ from mitochondria in a pS36-p66Shc-dependent manner. We confirmed that p66(S36A) overexpression almost completely inhibited VEGF-induced pSer36-p66Shc, while it partially reduced VEGFR2-pY (Fig. 9A). Of note, VEGF-induced pSer36-p66Shc was not affected by Mito-TEMPO (Fig. 9B), suggesting that VEGF-induced mtROS is not upstream, but downstream of pSer36-p66Shc formation. Furthermore, p66Shc(S36A) overexpression almost completely inhibited VEGF-induced EC migration and proliferation (Fig. 9C).

Fig. 8.

VEGF-induced pSer36-p66Shc is required for mitochondrial ROS production in ECs. A and B: HUVECs were infected with Ad.p66Shc-S36A or Ad.control (Ad.Ctrl) (A) or treated with Mito-TEMPO (50 μM) for 24 h (B). Cells were stimulated with VEGF (20 ng/ml) for 30 min, and then used to measure MitoSox fluorescence. Graph represents the averaged MitoSox fluorescence intensity. C: HUVECs were infected with Ad.Ctrl, Ad.Mito-catalase, or Ad.p66Shc-S36A for 48 h. Cells were stimulated with VEGF (20 ng/ml) for 15 min and then used to measure MitoTracker CM-H₂TMRos fluorescence imaged by confocal microscopy. Graph represents averaged fluorescence intensity expressed as fold increase over the basal. (n = 3, *P < 0.05).

Fig. 9.

VEGF-induced pSer36-p66Shc is required for VEGFR2 activation and angiogenic responses in ECs. A and B: HUVECs were infected with Ad.p66Shc-S36A or Ad.control (Ad.Ctrl) (A) or treated with Mito-TEMPO (50 μM) for 24 h (B). Cells were stimulated with VEGF for indicated times, and lysates were IB with VEGFR2-pY1175 or total VEGFR2 Abs (A); or IP with anti-Shc Ab, followed by IB with anti-pS36-p66Shc or Shc Abs (A and B), Graph represents p-VEGFR2 per total VEGFR2 expressed as fold change over the basal (A) or pS36-p66shc per total p66shc expressed as fold change over the basal (B) (n = 3, *P < 0.05). C: cell migration stimulated with VEGF for 6 h was measured using modified Boyden chamber method (left). For cell proliferation, cells were cultured in 0.5% FBS containing medium with or without VEGF (20 ng/ml) for 48 h and the cell number was counted (right) (n = 3, *P < 0.05).

We also confirmed that knockdown of p22^phox (Fig. 10D), which is a regulatory subunit for both Nox2 and Nox4, significantly inhibited VEGF-induced increase in DCF (Fig. 10A) and MitoSox (Fig. 10B) fluorescence, VEGFR2-pY (Fig. 10C), and pS36-p66Shc formation (Fig. 10E) without affecting VEGFR2 protein expression (Fig. 10C). These results further support the conclusion that the Nox/H₂O₂/pS36-p66Shc axis plays an important role in VEGF-induced mtROS production and partially in VEGFR2 activation (Fig. 10F).

Fig. 10.

p22^phox, a regulatory subunit for both Nox4 and Nox2, is involved in VEGF-induced mitochondrial ROS production and pSer36-p66shc formation in ECs. HUVECs were transfected with siRNA for control (siCtrl) or p22phox (si-p22^phox). A: cells were stimulated with VEGF (20 ng/ml) for 15 min and DCF fluorescence was measured. B: cells were stimulated with VEGF for 30 min and MitoSox fluorescence was measured. C: cells were stimulated with VEGF and were used to measure VEGFR2-pY1175 or total VEGFR2 using western analysis. D: cells were used to measure p22^phox mRNA using quantitative RT-PCR. E: lysates were IP with anti-Shc Ab, followed by IB with anti-pS36- p66Shc or Shc Abs. Graph represents the averaged fold change over the basal (n = 3, *P < 0.05). F: scheme showing that si-p22^phox prevents VEGF-induced Nox4/Nox2-derived H₂O₂, thereby inhibiting VEGF-induced pSer36-p66Shc, which promotes mtROS production and p-VEGFR2 in ECs.

DISCUSSION

Separate previous studies indicate that NADPH oxidases Nox4 or Nox2 or that mitochondria is the major source of ROS that serve as signaling molecules to promote switching from a quiescent to an angiogenic phenotype in ECs (1, 11, 13, 18, 44, 50, 56, 58, 62). Nox4 activation increases H₂O₂ while Nox2- or mitochondria generate ${\mathrm{O}}_2^{-}$ that is rapidly converted to H₂O₂ via SOD (4, 34). However, how H₂O₂ derived from NOX and mitochondria can coordinate to efficiently promote VEGF signaling and angiogenesis is poorly understood. To address this question, we applied several methods to measure intracellular redox status due to limitation of each ROS measurement (15, 24, 30, 59). We used Cyto-RoGFP and DCF-DA for cytosol redox analysis while Mito-RoGFP, MitoSox, and MitoTracker CM-H2TMRos for mitochondria redox analysis in combination with PEG-catalase (cytosolic H₂O₂ scavenger) or Mito-catalase (mitochondrial H₂O₂ scavenger) or Mito-TEMPO (mitochondria-targeted SOD mimetic) (15, 24, 59). In this study, we demonstrate that VEGF stimulation initially increases Cyto-RoGFP oxidation, which is followed by Mito-RoGFP oxidation in HUVECs. Knockdown of Nox4 or Nox2 or Mito-catalase overexpression significantly reduces VEGF-induced p-VEGFR2, EC migration and proliferation to a similar extent. Mechanistically, Nox2 senses Nox4-derived H₂O₂ to promote mtROS production via pSer36-p66Shc, thereby enhancing VEGFR2 activation and angiogenic responses. This may represent a novel positive feed-forward ROS-induced ROS release mechanism orchestrated by the Nox4/Nox2/pS36-p66Shc/mtROS axis, which promotes sustained activation of VEGFR2 angiogenesis signaling program (Fig. 11).

Fig. 11.

Scheme showing the role of ROS-induced ROS release orchestrated by Nox4, Nox2, and mitochondria in VEGF signaling and angiogenesis in ECs. Nox4-derived H₂O₂ in part activates Nox2 to promote mtROS production via pSer36-p66Shc, thereby enhancing VEGFR2 signaling and angiogenesis in ECs. This may represent a novel feed-forward mechanism of ROS-induced ROS release orchestrated by the Nox4/Nox2/pSer36-p66Shc/mtROS axis, which is required for sustained activation of angiogenesis signaling program (pathways demonstrated by present study are shown by black lines). Blue dashed lines show reported Rac1-mediated Nox2 activation independent of Nox4 activation, suggesting the possible direct regulation of Nox4- or Nox2- or mitochondria-derived ROS for activation of VEGFR2 or VEGFR2 downstream signaling pathways independent of ROS-induced ROS release.

To our knowledge, this is the first evidence for the real-time imaging of cytosol- and mitochondria redox status using compartment-specific RoGFP biosensor in response to VEGF in ECs at the single-cell level (26, 63). Of note, five individual cells showed the synchronized time course and intensity for VEGF-induced increase in both Cyto- or Mito-RoGFP oxidation, which was abolished by cytosolic catalase (63) and Mito-catalase, respectively. Thus, these results indicate that VEGF stimulation initially increases H₂O₂ from the cytosol, and then from the mitochondria in ECs. We also confirmed these findings using PEG-catalase-sensitive DCF; Mito-TEMPO-sensitive MitoSOX; and Mito-catalase-sensitive MitoTracker CMH2TMRos fluorescence. Since mitochondria-derived ${\mathrm{O}}_2^{-}$ can be rapidly converted to H₂O₂ by SOD2, it may result in the sustained production of H₂O₂ from the mitochondria.

Nox2 activation requires the assembly of cytosolic organizers such as p47^phox, p67^phox, and small GTPase Rac1 to produce ${\mathrm{O}}_2^{-}$ that is converted to H₂O₂. By contrast, Nox4, which is the most highly expressed NOX homolog in ECs, requires p22^phox but does not require Rac1 or any of the cytosolic organizers needed for Nox2 activation (16). Although Poldip2 is shown to interact with the Nox4-p22^phox complex in VSMCs (36), its expression and regulation for Nox4 activity in ECs remains unknown. The present study found that either Nox4 or Nox2 knockdown or PEG-catalase almost completely inhibits VEGF-induced DCF fluorescence, indicating that Nox4- and Nox2-derived H₂O₂ are in the same axis to increase Cyto-RoGFP oxidation in ECs. To address the relationship between Nox4 and Nox2, we found that Nox4 overexpression-induced H₂O₂ production was markedly inhibited by Nox2 siRNA. These results suggest that Nox2 senses Nox4-derived H₂O₂ signal to promote sustained cytosolic H₂O₂ production. However, role of crosstalk between Nox2 and Nox4 might be different in a cell- or agonist-specific manner, as shown in contrasting vascular effects of Nox2 vs. Nox4 in regulating blood pressure (50).

Functional linkage between NOX- and mitochondria-derived ROS (ROS-induced ROS release) has been implicated as a mechanism for amplification and compartmentalization of ROS signaling (14, 67). However, its role in VEGF signaling and angiogenesis remains unknown. The present study shows that either Nox4 or Nox2 knockdown or Mito-catalase overexpression almost completely inhibits VEGF-induced mtROS production, EC migration and proliferation but partially reduces VEGFR2-pY. Of note, siNox2, but not siNox4 or Mito-catalase, slightly but significantly reduces basal VEGFR2 protein expression. Since knockdown of p22^phox, which couples to both Nox2 and Nox4, has no effect on VEGFR2 expression, it is conceivable that Nox2 might have a noncanonical function to regulate VEGFR2 expression independent of its catalytic activity. However, VEGF-induced VEGFR2-pY normalized by total VEGFR2 is still significantly inhibited by Nox2 or Nox4 siRNAs, suggesting that Nox4 or Nox2-derived ROS are involved in VEGFR2 activation. Given that VEGF-induced cytosolic and mitochondrial ROS remained elevated for at least 60 min while p-VEGFR2 increased with a peak at 5 min and then decreased to above basal levels within 30 min, we speculate that VEGF-induced ROS may activate VEGFR2 downstream signaling pathways directly in addition to VEGFR2. Taken together, our results suggest that different ROS sources such as Nox4 and Nox2 lie in the same axis in part to promote mtROS production, which may enhance p-VEGFR2 and VEGFR2 downstream targets linked to angiogenesis (Fig. 11). Addressing molecular mechanisms of how Nox2 regulates VEGFR2 expression independent of NOX activity is beyond the scope of the present study.

In this study, we used exogenous H₂O₂ stimulation to mimic the effects of Nox4-derived H₂O₂ to demonstrate that Nox2 senses Nox4-derived H₂O₂ to promote mtROS production, which represents the novel mechanism of ROS-induced ROS release in ECs. Since VEGF-induced H₂O₂ produced via the Nox4/Nox2/mtROS axis enhances VEGFR2-pY, effects of inhibition of either Nox4 or Nox2 or mtROS on VEGF-induced responses may be secondary due to the inhibition of ligand-induced VEGFR2 activation. Thus, to bypass the effects of ROS on VEGFR2, we also used exogenous H₂O₂ in addition to VEGF stimulation to investigate the relationships among Nox4, Nox2 and mitochondria in organizing the ROS-induced ROS release, which is at least upstream for VEGFR2 activation. Consistent with our findings, Li et al. (35) reported that H₂O₂-induced ${\mathrm{O}}_2^{-}$ production is inhibited by p22^phox antisense-transfected VSMCs or in Nox2-deficient fibroblasts, indicating that H₂O₂ activates Nox2/p22^phox complex to promote ROS production. Evangelista et al. (18) reported that Nox4-derived H₂O₂ is upstream for Nox2-derived ROS, which are required for VEGF-induced S-glutathiolation of Ca²⁺ channel SERCA and EC migration. However, they did not address the linkage between the Nox4-Nox2 axis and mtROS in VEGF-induced angiogenesis. Given that SERCA is localized at the endoplasmic reticulum (ER), which structurally and functionally interacts with mitochondria (22, 37), it is possible that Nox4-derived H₂O₂ activates Nox2 to increase mtROS, which may induce S-glutathiolation of SERCA at the ER.

It remains unknown how H₂O₂ can activate Nox2. One possibility is that Nox4-derived H₂O₂ may participate in regulating Nox2 cytosolic organizers such as p47^phox phosphorylation and Rac1 activation, which are critical processes to activate Nox2 (4, 17, 62). To support this idea, it is shown that H₂O₂-sensitive cSrc kinase phosphorylates p47^phox (4) as well as Rac1 guanine nucleotide exchange factor (GEF), Vav2 (49), which is involved in VEGF-induced Rac1 activation in ECs (21). Furthermore, how Nox4 is rapidly activated by VEGF remains unclear. Recently, Xi et al. (64) reported that growth factor IGF-I stimulation induces rapid Tyr491 phosphorylation of Nox4, thereby promoting Nox4 binding to the adaptor protein Grb2, a component of plasma membrane scaffold SHPS-1 complex, which in turn induces localized ROS production and subsequent sustained Src activation. Thus, it is possible that this mechanism may also apply to VEGF-induced rapid activation of Nox4 to increase H₂O₂ that activates Nox2. Further study is warranted to investigate these possibilities. Moreover, we cannot exclude Nox4-independent activation of Nox2 mediated through VEGF-induced Rac1 activation (27, 40) (Fig. 11).

Next question is how H₂O₂ derived from the Nox4/Nox2 axis can promote mtROS production. In ECs, mitochondria functions as ROS-producing signaling organelles regulating vascular endothelial function (6, 19, 66). A previous report suggested that mtROS play an important role in VEGF-induced EC migration through Rac1 (62); however, the upstream mechanism by which VEGF increases mtROS is still poorly understood. It is reported that oxidative stress increases pSer-p66Shc via redox-sensitive PKC-β activation, and then translocates to mitochondria where it catalyzes electron transfer from cytochrome c to oxygen, thereby generating H₂O₂ from mitochondria (23, 45). Similarly, we previously demonstrated that VEGF increases pSer36-p66Shc at least through ERK/JNK or PKC in ECs (40), and that p66Shc knockdown inhibits VEGF-induced VEGFR2-pY, EC migration and proliferation (40). However, the specific role of pSer36-p66Shc in VEGFR2 signaling and angiogenesis remained unknown. In the present study, we show that VEGF-induced increase in Mito-Tracker CM-H2RMRos fluorescence was inhibited by p66Shc-S36A or Mito-catalase overexpression (Fig. 8C) as well as by ERK or PKC inhibitors or p66Shc siRNA (M. Ushio-Fukai, unpublished observations). Thus, together with other data, the current study suggests that the Nox4/Nox2 axis-derived H₂O₂ may induce pSer36-p66Shc via activation of PKC or ERK/JNK, thereby promoting mtROS, which enhances VEGFR2-pY and/or VEGFR2 downstream signaling, leading to angiogenic responses in ECs (Fig. 11). Moreover, we cannot exclude the possibility that there is direct regulation of Nox4- or Nox2-derived H₂O₂ or mtROS for VEGFR2-pY or VEGFR2 downstream signaling, independent of ROS-induced ROS release mechanism in ECs. (Fig. 11). Of note, it is likely that ${\mathrm{O}}_2^{-}$ production from mitochondria which is detected by MitoSox may be secondary due to mitochondrial H₂O₂ production induced by pS36-p66Shc. To support this notion, it is proposed that H₂O₂ induces oxidation of functional cysteines on mitochondrial proteins including electron transport chain (ETC) proteins to modulate protein activity, which may enhance ${\mathrm{O}}_2^{-}$ production via ETC (3). This mitochondria-derived ${\mathrm{O}}_2^{-}$ is rapidly converted to H₂O₂ by SOD2, thereby further increasing mitochondrial H₂O₂ production which is detected by Mito-catalase-sensitive, Mito-Tracker Orange CM-H2RMRos. Addressing this mechanism in detail is beyond the scope of present study but the subject of future study.

The mechanisms by which H₂O₂ derived from the Nox4/Nox2/pS36-p66Shc/mitochondria axis promotes ligand-induced VEGFR2-pY remain unclear. It is shown that H₂O₂ induces inactivation of protein tyrosine phosphatases (PTPs) via reversible oxidation of reactive Cys residue in the active site of the enzyme, thereby increasing growth factor signaling (46, 53). We previously reported that PTP1B negatively regulates VEGF-induced VEGFR2-pY and EC proliferation in ECs (38). Moreover, EC-specific PTP1B knockout mice show that PTP1B is an important endogenous negative regulator of VEGF signaling and angiogenesis (33). PTP1B is mainly localized at the ER where it dephosphorylates endocytosed EGFR (9, 25). In ECs, VEGF induces autophosphorylation of VEGFR2, which is followed by internalization to endosomes where VEGFR2-pY1175 is negatively regulated by PTP1B (32). Of note, Nox4 localized at the ER is shown to induce oxidative inactivation of PTP1B, which promotes epidermal growth factor (EGF) receptor-dependent signaling (9). Given that the ER and mitochondria structurally and functionally interact (22, 37), it is possible that endocytosed VEGFR2-pY, PTP1B, and mtROS may localize at close proximity in the intracellular compartment. This in turn may enhance VEGFR2 activation via Nox4/Nox2/pS36-p66Shc/mtROS-induced oxidative inactivation of PTP1B. It is possible that the Nox4- or Nox2-derived H₂O₂ prior to phosphorylating p66Shc may also activate VEGFR2 during receptor trafficking from the cell surface. In line with this, we previously reported that extracellular SOD-derived H₂O₂ enhances VEGF-induced VEGFR2-pY via oxidative inactivation of PTP1B/DEP1 in caveolae/lipid rafts, thereby promoting angiogenesis in ECs (41). Furthermore, ROS may directly activate VEGFR2 kinase activity via the redox-dependent modification of Cys residue in the catalytic sites of cytoplasmic domain, as reported for the EGF receptor kinase (42).

In conclusion, the present studies illustrate that H₂O₂ derived from Nox4 in part activates Nox2, which produces ${\mathrm{O}}_2^{-}$ that is rapidly converted to H₂O₂, thereby promoting mtROS production via pSer36-p66Shc, which in turn further enhances ROS-dependent VEGFR2 signaling in ECs. This may represent a novel positive feed-forward ROS-induced ROS release mechanism coordinated by the Nox4/Nox2/pS36-p66Shc/mtROS axis, which drives an angiogenic switch. However, we cannot exclude the possibility that there is direct regulation of Nox4- or Nox2-derived H₂O₂ or mtROS for VEGFR2 signaling and angiogenesis, independent of ROS-induced ROS release mechanism. Moreover, the contribution of other enzymatic sources of ROS including endothelial nitric oxide synthase (eNOS) that generates ${\mathrm{O}}_2^{-}$ (eNOS uncoupling) in pathological conditions should be investigated in future study. Our study should provide a novel insight into Nox2 as a sensor of Nox4-derived H₂O₂ to promote mtROS production, as well as pSer36-p66Shc as a central organizer linking the Nox4/Nox2 axis and mtROS required for sustained activation of angiogenesis signaling program mediating new vessel formation. In vivo role of Nox4, Nox2, or pSer36-p66Shc or mtROS in ECs using EC-specific Nox4^−/− or Nox2^−/−, p66Shc(S36A) knock-in mutant or Mito-catalase overexpressing transgenic mice with various angiogenesis models should be clarified in future study.

GRANTS

This research was supported by National Institutes of Health National Heart, Lung, and Blood Institute Grants NIHR01HL135584, NIHR01HL077524, NIHR01HL077524-S1, and NIHR21HL112293 (to M. Ushio-Fukai), NIHR01HL116976 (to T. Fukai and M. Ushio-Fukai), and NIHR01HL070187 (to T. Fukai); American Health Association Grant 16GRNT31390032 (to M. Ushio-Fukai); and Department of Veterans Affairs Merit Review Grant 2I01BX001232 (to T. Fukai).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Y.-M.K., S.-J.K., R.T., and H.Y. performed experiments; Y.-M.K., S.-J.K., R.T., and H.Y. analyzed data; Y.-M.K., S.-J.K., R.T., T.F., and M.U.-F. interpreted results of experiments; Y.-M.K., S.-J.K., and R.T. prepared figures; Y.-M.K. and M.U.-F. drafted manuscript; Y.-M.K., S.-J.K., R.T., H.Y., T.F., and M.U.-F. approved final version of manuscript; S.-J.K., H.Y., T.F., and M.U.-F. edited and revised manuscript; M.U.-F. conceived and designed research.