| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 278, Issue 38, 36830-36840, September 19, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the University of Texas Southwestern and The Dallas Veterans Medical Center, Dallas, Texas 75216
Received for publication, March 4, 2003 , and in revised form, June 20, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Reactive oxidants have been linked to both proliferative pathways and cytoskeletal dynamics and thus may present a common mediator of these two facets of growth factor signaling. PDGF, epidermal growth factor, and insulin initiate a small burst of oxidants critical for protein tyrosine phosphorylation, MAPK activation, and DNA synthesis, whereas oxidant scavengers suppress these events (1-3). Rac1, through its involvement with the NADPH oxidase, also mediates H-RasV12-induced mitogenesis and PDGF-dependent cyclin D1 expression (4, 5). The use of oxidants is not limited to pathways linked to tyrosine kinase receptors, because smooth muscle NADPH oxidase is activated during the proliferative signaling of both angiotensin II and the ganglioside GD3 (6, 7).
Oxidants have also been linked to cytoskeletal reorganization. High levels of exogenous oxidants increase recruitment of actin to the cytoskeleton, and conversely actin cytoskeletal disruption precipitates oxidant-dependent NF-
activation (8, 9). Oxidant production is also most prominent in actively migrating endothelial cells at the margin of a monolayer wound and, in particular, is concentrated in membrane ruffles of these highly motile cells (10). This localization of oxidant production is comparable to that found in phorbol-ester-stimulated neutrophils whose oxidative burst is concentrated within ruffling membranes (11). Consistent with these observations, we recently found constitutive localization of the NADPH oxidase subunit p47phox to the cytoskeleton of endothelial cells with translocation of p47-GFP to membrane ruffles (12).
In the present study, we identify the ruffle catalyst WAVE1/suppressor of cAMP receptors as a binding partner of p47phox. Upon stimulation of endothelial cells with VEGF, p47phox associates with WAVE1 as well as with Rac1 and PAK1, proteins involved in both ruffle formation and oxidase activation. p47phox-dependent oxidants appear to mediate VEGF-induced ruffle formation, tyrosine phosphorylation, and JNK activation.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Yeast Two-hybrid Screening and Mating—AH109 yeast (Clontech) were stably transformed with pGBKT7-p47-(1-298) and tested negative for autonomous transactivation. A HUVEC library previously dropped out in pGAL4-AD (13) was used to secondarily transform bait-containing yeast, which were selected under high stringency for bait and library vectors (reversal of Trp and Leu auxotrophy) and interaction (reversal of His and Ade auxotrophy and lacZ expression). Surviving colonies were recloned and retested, and library plasmids were extracted, passaged through Escherichia coli, characterized, and transformed into AH109 yeast. Library plasmid-containing AH109 yeast were mated with bait-containing Y187 yeast, and diploids were plated and assessed for lacZ expression with a filter lift assay. Truncated forms of p47phox and WAVE1 were tested for interaction using a similar yeast mating strategy. Strongly positive interactions caused blue color development within 2 h. Negative interactions remained white for >24 h. Weak positive interactions caused distinct blue coloration between 2 and 24 h.
GST Pull Down Assay—BL21-RP E. coli (Stratagene) harboring pGEX-2TK, pGEX-p47, or pGEX-p47-(1-289) were induced with isopropyl-1-thio--D-galactopyranoside for 4 h, and the soluble proteins were captured on glutathione-Sepharose 4B (Amersham Biosciences). Full-length or truncated WAVE1 proteins were transcribed and translated in vitro from pCIN-WAVE1 in the presence of [35S]methionine using the TNT Quick Coupled system (Promega). 5 µg of the fusion proteins were incubated with 10 µl of translation mixture for 2 h at 4 °C prior to extensive washing.
Adenoviral Construction—Adenoviruses harboring lacZ and p47phox were described previously (12, 13). Sequential site-directed mutations were introduced with PCR to create p47-(S303D,S304D,S328D), which was then subcloned into pShuttle and then pAdX (Clontech). Ad-p47-(W193R) and Ad-p47-(S303D,S304D,S328D) were generated, cloned, and characterized in human embryonic kidney 293 cells. Efficiency of transgene expression in HUVEC assessed by lacZ expression was >95%. Expression of p47-wild type, p47-(W193R), and p47-(S303D,S304D,S328D) proteins were equivalent by immunoblot at equal m.o.i.
Cell Culture and Microscopy—HUVEC (Clonetics) used in this study were propagated in EGM-2 medium containing 2% fetal calf serum, epidermal growth factor, fibroblast growth factor, VEGF, insulin growth factor, ascorbic acid, hydrocortisone, heparin, and gentamicin. Transfection was performed after cell synchronization 6-8 h after thymidine release using electroporation with a constant amount of total plasmid DNA. For adenoviral transduction, cells were infected for 1 h at an m.o.i. of 200:1, and harvested 24 h later. For microscopy experiments, cells were plated on fibronectin-coated chambered glass cover-slips and examined live. Prior to VEGF stimulation, cells were serum-starved in medium containing 0.1% fetal calf serum without growth factors for 24 h. For some experiments, cells were treated with 100 µM SOD mimetic MnTBAP (Calbiochem) for 1 h at 37 °C. The number of ruffling cells was counted in all of the actin-GFP-expressing cells in at least ten random fields in at least three separate slides between 15 and 20 min after the addition of VEGF (30 ng/ml, PeproTech). Confocal images were obtained with a Zeiss Axiovert S100TV and LSM 410 laser-scanning system. In some experiments, cells were fixed, permeablized, and stained with 5 units of rhodamine phalloidin (Molecular Probes) prior to examination (12). Fluorescence ratio imaging was performed using Zeiss LSM software, version 3.98.
Immunoprecipitation and Western Immunoblot—FLAG-p47-transfected HUVEC were rocked in ice-cold lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM Na2 EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM -glycerophosphate, 1 mM Na3VO4, 1 µg/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride) for 20 min, harvested, and sonicated once for 3 s. Insoluble material was removed by centrifugation at 4 °C for 20 min at 10,000 x g prior to immunoprecipitation with anti-FLAG (M2, Sigma). Endogenous and adenovirally transduced p47phox were immunoprecipitated using mouse anti-p47phox (Upstate Biotechnology). Immunoblots were performed with antibodies against WAVE1 (Upstate Biotechnology), p47phox (Upstate Biotechnology and a gift of Dr. Bernard Babior), PAK1 (Santa Cruz Biotechnology), Rac1 (Transduction Laboratories), and phosphotyrosine (Santa Cruz Biotechology).
p47phox Phosphorylation and Oxidant Production—HUVEC were transfected with pCINF-p47, recovered for 24 h, and then serum-starved for an additional 20 h. Culture medium was replaced with phosphate-free Dulbecco's modified Eagle's medium (Sigma) containing 0.5 mCi/ml [32P]orthophosphate for 4 h. Cells were then stimulated with VEGF (30 ng/ml) for 15 min, and cell lysates were subjected to immunoprecipitation with anti-FLAG, PAGE separation, and autoradiography. A third of the immunoprecipitate was immunoblotted for p47phox to assess capture and loading. Oxidant production was assessed as oxidation of 2',7'-dichlorofluorescin diacetate (DCF, Molecular Probes). HUVEC were washed and loaded with DCF (10 µM) in Hanks' balanced salt solution for 20 min at 37 °C. They were then washed and exposed to VEGF (30 ng/ml for 15 min), and DCF fluorescence was measured as described previously (12).
JNK and PAK Kinase Assays—Following stimulation, cells were washed with ice-cold phosphate-buffered saline, scraped, pelleted, and lysed in lysis buffer. HA-JNK or PAK was immunoprecipitated with anti-HA or anti-PAK1 (Santa Cruz Biotechnology), and the immunocomplexes were washed twice with lysis buffer and once with kinase buffer. The immunocomplex kinase assay contained either 2 µg of GST-Jun (Santa Cruz Biotechnology) for JNK or 5 µg of myelin basic protein (Upstate Biotechnology) for PAK as a substrate in the presence of [
-32P]ATP (PerkinElmer Life Sciences) at 30 °C for 30 min (13).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
Using the yeast-mating technique, general binding domains were identified on both proteins. The WAVE1-binding region of p47phox was contained within the central tandem SH3 domains with the N-proximal domain (SH3a) in isolation binding more robustly than the isolated distal (SH3b) domain (Fig. 1c). Both p47-(1-205), containing the N-terminal Phox (PX) and SH3a domains, and p47-(205-390), containing the SH3b domain and C-terminal region, failed to bind WAVE1, presumably because of SH3 surface masking by proline-rich sequences within the adjacent PX and C-terminal regions, respectively. Interestingly, p47-(153-219,Arg-193), containing a Trp
Arg mutation disrupting the SH3a binding surface (12, 15) also bound WAVE1, suggesting a nonconventional mechanism of interaction perhaps unrelated to the SH3a surface.
p47phox in turn appeared to bind to the mid-region of WAVE1 (Fig. 1c). Again, the binding mechanism was not easily deduced. Neither residues 1-269 nor 270 -599, which together span the entire protein, bound p47-(1-298). Similarly, neither 1-362 nor 363-559 bound p47-(1-298). An intact proline-rich region (residues 278-442) appeared to be necessary for binding as truncations not containing the full proline-rich region did not bind p47. GST pull down experiments also confirmed the ability of WAVE1-(278-442) and WAVE1-(100-442) to bind p47-(1-298) (Fig. 1d). Curiously, WAVE1-(270-559) bound p47-(1-298) in vitro but not in the yeast two-hybrid system, possibly due to interference with p47 binding by the N-terminal fused Gal4-AD in the WAVE1 truncation of the latter system.
VEGF Stimulates Interaction between WAVE1 and p47phox within Membrane Ruffles—As with its myeloid paralog WASP, WAVE1 functionally catalyzes Arp2/3-dependent actin nucleation and polymerization (16) and is thus found concentrated within membrane ruffles (17). In live HUVEC, we found that a DsRed-p47 fusion consistently colocalized with actin-GFP within membrane ruffles following VEGF stimulation, whereas DsRed-p47 was diffusely localized within the cytosol in unstimulated cells (Fig. 2). Specific localization to ruffles was demonstrated by fluorescence ratio imaging of cells cotransfected with DsRed-p47 and free EGFP (Fig. 3, a-c). In addition, the reversal of chromophores using p47-GFP and free DsRed also demonstrated relative concentration of p47phox in membrane ruffles (Fig. 3, d-f). Thus, localization of p47phox to ruffles did not appear to be a function of DsRed itself or of the increased cytosolic volume within ruffles. Interestingly, DsRed-p47 translocated to the nucleus in 15-30% of cells following VEGF stimulation (Fig. 2, d and f). The significance of this finding is unclear at this point but may reflect cotransport of p47phox with other mitogenic or survival factors to the nucleus.
|
|
In addition, endogenous WAVE1 coprecipitated with FLAG-p47 and recovery of WAVE1 was maximal at 15 min after stimulation with VEGF (Fig. 4a). This time course coincided with the onset of ruffling activity, which peaked at 10-15 min following VEGF addition, and subsided by 30 min. Furthermore, VEGF also increased coprecipitation of endogenous WAVE1 with endogenous p47phox (Fig. 4b). As expected, spatial colocalization within ruffles of VEGF-stimulated cells was demonstrated using WAVE-GFP and DsRed-p47 (Fig. 5). Colocalization within ruffles was consistent and reproducible with all of the ruffles visualized containing both fusion proteins. Finally, ectopic expression of either the WAVE1 binding domain of p47phox (p47-(153-286)) or the p47 binding domain of WAVE1 (WAVE1-(100-442)) decreased VEGF-induced ruffling (Fig. 6), suggesting that the interaction between WAVE1 and p47phox facilitates such ruffling.
|
|
|
PAK1 Acts Upstream of VEGF-induced p47phox Phosphorylation, Oxidant Production, and Ruffle Formation—Activated Rac1 and its effector PAK1 are potent inducers of ruffles, and both proteins respond to growth factors and concentrate in linear ruffles (18, 19). Active Rac1 has also been shown to coprecipitate with WAVE1 (17), although the latter protein lacks a functional Cdc42/Rac1-interaction binding domain, suggesting an indirect interaction. We found that within 15 min of VEGF stimulation, endogenous forms of both PAK1 and Rac1 coprecipitated with FLAG-p47 (Fig. 7a). WAVE1 also appeared to be part of the PAK1-p47 complex (Fig. 7b), consistent with the formation of a ternary complex in which WAVE1 and associated protein(s) may act as a scaffold for the recruitment of Rac1, PAK1, and p47phox. Rac1 is an essential subunit of the NADPH oxidase, and PAK1 serves as an effective p47phox kinase in vitro (20), although its capacity to perform this function in intact cells has not been demonstrated. The formation of such a complex might therefore be expected to facilitate p47phox phosphorylation and consequent oxidant production. Indeed, we found that VEGF led to PAK1 activation within 5 min and caused phosphorylation of p47phox within 15 min, whereas kinase-dead dominant-negative (DN) PAK1-(K298A) blocked p47phox phosphorylation (Fig. 8, a-c). This ability of PAK1 to facilitate p47phox phosphorylation, either directly or indirectly, is consistent with recent studies of the NADPH oxidase suggesting that Rac1 activates unknown proximal signaling events leading to oxidase assembly independent from its direct involvement with the oxidase (21). In parallel, PAK1-(K298A) also blocked VEGF-induced oxidant production (Fig. 8d) and ruffle formation (Fig. 9). Furthermore, the membrane-permeant SOD mimetic MnTBAP also reduced VEGF-induced ruffling (Fig. 10), suggesting that the effects of endogenous p47phox occurred at least in part through oxidant production. Taken together, these data suggest a linear sequence of complex formation, p47phox phosphorylation, oxidant production, and membrane ruffling.
|
|
|
|
p47-(S303D,S304D,S328D) Increases Oxidant Production, Tyrosine Phosphorylation, and Ruffle Formation—To further implicate a specific role of p47-dependent oxidants in ruffle formation, we replaced three serines on p47phox known to be critically phosphorylated during activation of the phagocyte oxidase with aspartic acid residues to mimic phosphoserines. This triple mutant has been shown to cause constitutive activation of the NADPH oxidase in a cell-free reconstitution system (22). Adenoviral delivery of this mutant p47phox increased oxidant production of unstimulated HUVEC, whereas overexpression of the wild-type p47phox or the SH3a-defective p47-(W193R) had no significant effect on basal oxidant production (Fig. 11a). Overexpression of p47-(S303D,S304D,S328D) also increased total protein phosphotyrosine content, and this effect did not require the presence of VEGF and other growth factors (Fig. 11b). In contrast, overexpression of wild-type p47phox increased tyrosine phosphorylation only in the presence of growth factors, and p47-(W193R) overexpression had no appreciable effect with or without growth factors. Finally, overexpression of p47-(S303D,S304D,S328D) also resulted in the formation of circular dorsal ruffles in the absence of VEGF (Fig. 12, a-c), which were unusual in uninfected cells or cells infected with Ad-LacZ, Ad-p47, or Ad-p47-(W193R) (Fig. 12a and data not shown). WAVE1 also coprecipitated with p47-(S303D,S304D,S328D) (Fig. 12d), consistent with activation-induced association of p47phox with WAVE1 and consequent ruffle formation. We could not demonstrate association of PAK1 with this complex (Fig. 12d), possibly because the active mutant of p47phox presumably enters the pathway distal to PAK1, thus bypassing it. The reason for the appearance of circular rather than peripheral linear ruffles is also not clear; however, PAK1 concentrates in dorsal circular ruffles and active PAK1 mutants such as H83L,H86L are known to induce dorsal ruffling, whereas active Rac1 mutants acting proximal to PAK1 more typically produce linear ruffles. Again, specious activation of the oxidase through an active p47phox mutant may bypass non-oxidase-dependent functions of Rac1 or elements upstream of Rac1, potentially biasing the response toward circular rather than linear ruffles. Notably, circular ruffles are also WAVE-dependent (23) and we noted that p47-GFP concentrated in the rare circular ruffles stimulated by VEGF (Fig. 12e), further suggesting that a common WAVE- and p47-dependent complex may be responsible for both linear and circular ruffles.
|
|
VEGF Signaling Bifurcates Distal to PAK1 and Oxidant Production, Leading to JNK Activation and Ruffle Formation—We confirmed that VEGF activated JNK (Fig. 13a), a MAPK known to mediate VEGF-dependent mitogenesis (24). Furthermore, PAK1-(K298A) effectively blocked VEGF-induced HA-JNK2 activation (Fig. 13a) as did the antioxidant MnTBAP (Fig. 13b). Thus, JNK activation similar to ruffle formation appears to lie downstream of PAK1 and oxidant production. To investigate whether JNK lies within a linear activation scheme leading to ruffle formation, we cotransfected DN-JNK2 with actin-GFP and found that DN-JNK2 did not diminish VEGF-induced ruffling (Fig. 14). Similar negative results were obtained with DN-JNK1 (data not shown). Therefore, VEGF signal pathways appear to diverge beyond oxidant production, leading independently to JNK activation and ruffle formation.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Our finding that VEGF stimulates association of WAVE1 with p47phox suggests a possible linkage of WAVE1 function with oxidant-related cytoskeletal and signaling events. In support of this possibility, the reduction of oxidant levels through interventions upstream (DN-PAK1) or downstream (MnTBAP) of p47phox or ectopic expression of the isolated binding domains of the two proteins was associated with a reduction in ruffle formation. In contrast, specious activation of p47phox (p47-(S303D,S304D,S328D)) increased ruffle formation. The potential involvement of endogenous oxidants in actin dynamics is also suggested by recent studies demonstrating that NADPH oxidase inhibitors and 
scavengers abolish actin monomer incorporation at the fast-growing filament barbed ends and block migration of endothelial cells (10). More specifically, oxidants derived from the NADPH oxidase have been linked to VEGF-induced endothelial cell migration and angiogenic behavior (33). The NADPH oxidase has also been implicated in MAPK activation by PDGF and tumor necrosis factor, both factors that induce membrane ruffling (2, 12). The association of WAVE1 with p47phox may therefore reflect the existence of a molecular complex controlling both cytoskeletal and mitogenic events.
VEGF stimulates phosphorylation of endothelial p47phox, which is consistent with the unmasking of its core binding domains to allow binding to WAVE1. It is uncertain at this point, however, whether a single p47phox molecule is able to bind simultaneously to WAVE1 as well as to the NADPH oxidase. Our data do not support a conventional SH3 surface-dependent interaction between p47phox and WAVE1, allowing for the possibility that p22phox and WAVE1 may simultaneously bind different surfaces within the core p47phox region. Recent peptide walking data suggest that p22phox itself may bind p47phox through a non-SH3-dependent mechanism (34). Alternatively, the binding of p47phox to WAVE1 and p22phox may be sequential rather than simultaneous. In one scenario, p47phox may first serve its function as an NADPH oxidase adapter and then leave the complex to recruit WAVE1. Indeed, once activated, the NADPH oxidase does not require p47phox for production (35). In an alternate scenario, p47phox may first serve a novel role as a WAVE1 adapter by assisting membrane targeting of WAVE1 through its N-terminal PX domain, which binds to lipid products of phosphatidylinositol 3-kinase (36) before binding p22phox. Finally, dimerization of p47phox within a WAVE1 complex would allow presentation of two SH3a surfaces per complex. That at least one of these possibilities operates in endothelial cells is strongly suggested by the colocalization of WAVE1, p47phox, and oxidant production within membrane ruffles.
Recruitment of p47phox to a WAVE1 scaffold may reflect a general strategy of targeting oxidant production to critical subcellular sites as a means to gain signal specificity. In the context of VEGF signaling, the molecular targets of such oxidants are presently unknown. However, we found that expression of p47-(S303D,S304D,S328D) increased tyrosine phosphorylation of a number of proteins, consistent with the ability of to reversibly inactivate protein tyrosine phosphatases (37). In this regard, the association of the tyrosine kinase c-Abl with the WAVE1 complex (16) may be expected to hold significance for both mitogenic and cytoskeletal signal pathways. Indeed, c-Abl appears to mediate both oncogenic Src and PDGF effects and PDGF fails to stimulate ruffles in c-Abl-null cells (38). Potential downstream c-Abl substrates include Crk and Enabled, both linked to actin remodeling and survival signaling (39, 40).
|
FOOTNOTES |
|---|
To whom correspondence should be addressed: Dallas VAMC, MC 151, 4500 S. Lancaster Rd., Dallas, TX 75216. Fax: 214-857-0340; E-mail: Lance.Terada{at}med.va.gov.
1 The abbreviations used are: MAPK, mitogen-activated protein kinase; PDGF, platelet-derived growth factor; GFP, green fluorescent protein; JNK, c-Jun N-terminal kinase; HUVEC, human umbilical vein endothelial cell; HA, hemagglutinin; EGFP, enhanced GFP; GST, glutathione S-transferase; Ad, adenovirus; VEGF, vascular endothelial growth factor; DCF, dichlorofluorescein diacetate; SH, Src homology; PX, Phox; DN, dominant-negative; WASP, Wiskott-Aldrich syndrome protein; PAK, p21-activated kinase; BD, binding domain; DsRad, Discoma sp. Red; SOD, superoxide dismutase.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  N. Anilkumar, R. Weber, M. Zhang, A. Brewer, and A. M. Shah Nox4 and Nox2 NADPH Oxidases Mediate Distinct Cellular Redox Signaling Responses to Agonist Stimulation Arterioscler. Thromb. Vasc. Biol., July 1, 2008; 28(7): 1347 - 1354. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Md. R. Abid, K. C. Spokes, S.-C. Shih, and W. C. Aird NADPH Oxidase Activity Selectively Modulates Vascular Endothelial Growth Factor Signaling Pathways J. Biol. Chem., November 30, 2007; 282(48): 35373 - 35385. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-X. Chen, H. Zeng, M. L Lawrence, T. S. Blackwell, and B. Meyrick Angiopoietin-1-induced angiogenesis is modulated by endothelial NADPH oxidase Am J Physiol Heart Circ Physiol, October 1, 2006; 291(4): H1563 - H1572. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. S. Terada Specificity in reactive oxidant signaling: think globally, act locally J. Cell Biol., August 28, 2006; 174(5): 615 - 623. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Ushio-Fukai Localizing NADPH Oxidase-Derived ROS Sci. Signal., August 22, 2006; 2006(349): re8 - re8. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Ushio-Fukai Redox signaling in angiogenesis: Role of NADPH oxidase Cardiovasc Res, July 15, 2006; 71(2): 226 - 235. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Moldovan, K. Mythreye, P. J. Goldschmidt-Clermont, and L. L. Satterwhite Reactive oxygen species in vascular endothelial cell motility. Roles of NAD(P)H oxidase and Rac1 Cardiovasc Res, July 15, 2006; 71(2): 236 - 246. [Abstract] [Full Text] [PDF] |
