oxidant stress and endothelial cell dysfunction

8/9/2019 Oxidant Stress and Endothelial Cell Dysfunction

1/23

invited review

Oxidant stress and endothelial cell dysfunction

HAZEL LUM1 AND KENNETH A. ROEBUCK21Department of Pharmacology and 2Department of Immunology and Microbiology,

Rush Presbyterian St. Lukes Medical Center, Chicago, Illinois 60612

Lum, Hazel, and Kenneth A. Roebuck. Oxidant stress and endo-thelial cell dysfunction. Am J Physiol Cell Physiol 280: C719C741,2001.Reactive oxygen species (ROS) are generated at sites of inflam-mation and injury, and at low levels, ROS can function as signalingmolecules participating as signaling intermediates in regulation of fun-damental cell activities such as cell growth and cell adaptation re-sponses, whereas at higher concentrations, ROS can cause cellular injuryand death. The vascular endothelium, which regulates the passage ofmacromolecules and circulating cells from blood to tissues, is a majortarget of oxidant stress, playing a critical role in the pathophysiology ofseveral vascular diseases and disorders. Specifically, oxidant stress in-creases vascular endothelial permeability and promotes leukocyte adhe-sion, which are coupled with alterations in endothelial signal transduc-tion and redox-regulated transcription factors such as activator protein-1and nuclear factor-B. This review discusses recent findings on thecellular and molecular mechanisms by which ROS signal events leadingto impairment of endothelial barrier function and promotion of leukocyteadhesion. Particular emphasis is placed on the regulation of cell-cell andcell-surface adhesion molecules, the actin cytoskeleton, key protein ki-nases, and signal transduction events.

endothelial permeability; leukocyte transmigration; actin filament; cad-

herin; occludin; intercellular adhesion molecule-1; selectins; redox-sensitive signal transduction; nuclear factor-B

ACUTE AND CHRONIC OXIDANT STRESSto the vascular endo-thelium is a serious causative factor of vascular endo-thelial dysfunction and plays an important role in thepathophysiology of several vascular diseases, includingatherosclerosis, diabetes, neuronal disorders, and is-chemia-reperfusion injury. Two aspects of endothelialdysfunction may be particularly important in deter-mining the severity of a vascular disorder:1) increasedendothelial permeability and 2) increased endothelial

adhesion for leukocytes, both critical factors governingtissue edema formation and leukocyte extravasation.Reactive oxygen species (ROS) are well documented tofunction as signaling molecules, stimulating cellularactivities ranging from cytokine secretion to cell prolif-eration, and at higher concentrations, they can inducecell injury and death by oxidant modification of pro-

teins and carbohydrates, lipid peroxidation, and DNAstrand nicks. Such diverse responses are related tomultiple factors, such as the ROS prevailing at theinflammatory locus, concentration and turnover of theoxidants, and the antioxidant capacity of the locaenvironment as well as target cells. There are severaexcellent reviews regarding the signaling role of ROSand possible mechanisms of endothelial injury (132164). Therefore, this review is specifically focused onrecent findings regarding cellular and molecular activities by which ROS signal the impairment of endothelial barrier function and promotion of leukocyte adhesion and extravasation.

EFFECTS OF OXIDANT STRESS

ON VASCULAR ENDOTHELIUM

At sites of inflammation and infection, the local cellular environment is enriched with cytokines, chemokines, and ROS. During this period, the endotheliumcan be exposed to high levels of multiple species of ROSfor a prolonged period of time. In the in vivo setting

Address for reprint requests and other correspondence: H. Lum,Dept. of Pharmacology, Rush Presbyterian St. Lukes Medical Cen-ter, 2242 W. Harrison St., Suite 260, Chicago, IL 60612 (E-mail:[email protected]).

Am J Physiol Cell Physio280: C719C741, 2001

0363-6143/01 $5.00 Copyright 2001 the American Physiological Societyhttp://www.ajpcell.org C719


2/23

subpopulations of polymorphonuclear neutrophils(PMN) are found adherent to the endothelium or mi-grating through the extravascular tissue matrix. Amajor source of ROS is blood leukocytes that becomeactivated and adherent to the endothelial cell surface.In vitro studies demonstrate that suspensions of PMN(106), when activated with phorbol esters, release35360 nmol O2

/h (146, 226). Evidence indicates

that adherence to matrix proteins (i.e., laminin, fi-bronectin, vitronectin) or endothelium primes thePMN for a massive respiratory burst lasting 13 h inresponse to tumor necrosis factor- (TNF-) and thechemoattractant formylmethionyl leucylphenylalanine(fMLP) (166). Thus cytokines at the inflammatory sitemay contribute significantly to the generation of ROS,particularly by PMN.

The activation of endothelial cells also generatesROS and therefore may be a significant contributor inmaintaining the oxidant-rich environment at the in-flammatory locus. The treatment of human umbilicalvein endothelial cells with the cytokines interleukin-1(IL-1) and interferon- (IFN-) results in dose- and

time-dependent increases in O2 (153). Also, vasoactivepeptides such as bradykinin can induce the productionof O2

(40 nmol/2 106 cells) within 5 min of ligandbinding to its receptor on endothelial cells (96). Expo-sure of endothelial cells to hypoxia followed by reoxy-genation induces the production of 1050 nmolO2

h1 1.5 million cells1 (144, 257), suggesting thatendothelium-derived ROS contribute to ischemia-reperfusion injury.

With the generation of O2, other more reactive in-

termediates also are produced by the spontaneous orenzymatically catalyzed dismutation of O2

2O2 2H 3 H2O2O2 (1)

In the presence of transition metals (e.g., Fe3), O2

reduces Fe3 to Fe2 by the iron-catalyzed Haber-Weiss reaction (Eq. 2). The reduced ferrous iron reactswith H2O2 via the Fenton reaction (Eq. 3) to generatethe highly reactive hydroxyl radical OH

O2 Fe3 3 O2 Fe

2 (2)

H2O2 Fe2

3 Fe3 OHOH (3)

Furthermore, O2 may also react with nitric oxide

(NO), a reactive nitrogen species produced by endothe-lial cells (123), to form peroxynitrite, which decom-

poses to form the potent ROS OH and NO2

(21).Several experimental models of oxidant stress havedemonstrated increased leukocyte extravasation. Theperfusion of the rat pancreatic vasculature with xan-thine/xanthine oxidase resulted in increased leukocyterecruitment as measured by myeloperoxidase assay(100). Similarly, superfusion of the rat mesentery mi-crovasculature with H2O2caused increased transendo-thelial migration of leucocytes as detected by intravitalmicroscopy (200). In an in vitro model, treatment ofhuman umbilical vein endothelial cells with tert-butyl-hydroperoxide resulted in a twofold increase in trans-

endothelial migration of monocyte-like HL-60 cells(190).

There is substantive evidence that oxidant stressincreases vascular endothelial permeability. Theperfusion of ROS (e.g., H2O2) or xanthine/xanthineoxidase (which generates O2

) into isolated lungs oguinea pigs (110) and rabbits (18, 205) results inincreased capillary filtration coefficient within 30

min, leading to pulmonary edema. Similarly, superfusion of ROS onto the surface of brain pial venulesresults in dose-dependent decreases in transendothelial electrical resistance, indicating barrier impairment (175). In some studies, the increased permeability is accompanied by increased capillarypressure (17, 110), suggesting that both factors maybe important in promoting tissue edema formationThe oxidant-mediated vasopressor effect appears tobe associated with activation of tyrosine kinase(108), although the definitive signaling mechanismsresponsible are not clear. However, other studiehave indicated that edema formation in response to

oxidant stress can occur in the absence of increasedcapillary pressure (17, 18, 205). These observationsindicate that oxidant-mediated tissue edema maylikely be attributed predominantly to increased vascular endothelial permeability and that an accompanying capillary pressure rise may function to accelerate the transvascular fluid flux.

This notion is further supported by in vitro studies inwhich direct treatment of cultured endothelial celmonolayers with ROS increased transendothelial permeability. The incubation of endothelial monolayerswith xanthine/xanthine oxidase (208) or glucose/glucose oxidase (which generates H2O2) (97) increased the

transendothelial flux of tracer albumin; the durationand severity of the increase corresponded positivelywith increased ROS generation. Furthermore, directaddition of H2O2increased albumin flux and decreasedtransendothelial electrical resistance (172, 214). Interestingly, comparison of the effects of H2O2with those ohypochlorous acid (HOCl), a more reactive oxidanthan H2O2, showed that HOCl causes a faster andgreater increase in albumin permeability (172).

The vascular endothelial barrier dysfunction iswell documented in ischemia-reperfusion injurywhich is associated with toxic amounts of ROS generated primarily by activated leukocytes and endo

thelial cells. In experimental models of brain ischemia-reperfusion injury, the treatment with a permeant H2O2scavenger and sodium tungstate (whichinactivates xanthine oxidase) (177) reduced brainedema, and treatment with polyethylene glycol superoxide dismutase or polyethylene glycol catalaseattenuated the increased blood-brain transfer o[14C]urea (11). Similarly, perfusion of superoxidedismutase or a xanthine oxidase inhibitor (allopurinol) protected against increases in vascular permeability in ischemia-reperfusion models of the rabbitlung (99, 116) and hamster cheek pouch (51).

C720 INVITED REVIEW


3/23

STRUCTURAL BASIS OF OXIDANT-MEDIATED

BARRIER DYSFUNCTION

ROS, like other edemagenic mediators (e.g., throm-bin, histamine, and TNF-) cause intercellular gapformation, cell shape change, and actin filament reor-ganization (Fig. 1 and Table 1) (147). These morpho-logical features implicate impaired cell-cell adhesion,and, consequently, impaired intercellular junctions, asa primary determinant of increased paracellular per-meability. The endothelium, like the epithelium, hasadherens and tight junctions, both of which are criticalfor maintaining a restrictive barrier.

Mechanisms For a Restrictive Barrier:Role of Cell Junctions

The tight junction consists of the transmembraneprotein occludin (61) in association with several cyto-plasmic proteins [i.e., zonula occludens (ZO)-1, ZO-2,ZO-3] that establish a link with the actin cytoskeleton(Fig. 1) (54). In endothelial (93, 118) and epithelial cells(248), a decreased occludin level correlates with in-

creased permeability. Morphologically, occludin mayrepresent the intramembranous strands observed withfreeze-fracture replicas of tight junctions that seal theouter membrane leaflets and effectively impart thebarrier property to the endothelium and epithelium(155). Although tight junctions provide the primarybarrier for the endothelium, their function, assembly,and organization are believed to be dependent on func-tional cadherins of adherens junctions (80, 137, 187,239).

The endothelial cell adherens junction contains theintegral protein VE-cadherin (cadherin-5), a member ofthe cadherin family of glycoprotein adhesion mole-

cules, which is anchored to the actin cytoskeletal net-work by several intermediate proteins (i.e., -,-, and-catenins) (Fig. 1) (81, 91, 168). This linkage to thecytoskeleton positions and stabilizes VE-cadherin tothe junctional site (167, 225). Cadherins form homo-typic adhesion between cells in a Ca2-dependent man-ner via interactions of the extracellular amino termini.On the basis of crystal structure analysis of the cad-herin family, a tripeptide HAV sequence residing inthe first of five tandem repeats (110 amino acidseach) of the amino terminus was found to be the com-mon cell-cell recognition site shared among differentcadherin subtypes (148). However, VE-cadherin lacksthis tripeptide sequence in the amino terminus, sug-

gesting that its recognition function is comprised of adifferent amino acid sequence (103).The amino and carboxy regions of the cadherin mol-

ecule contain sequences that regulate endothelial bar-rier function. Antibodies directed against the cadherinamino terminus have been shown to increase solutepermeability and impair formation of adherens junc-tions of bovine aortic endothelial cell monolayer (4, 85).Fujimori and Takeichi (59) transfected a mutant N-cadherin (an amino terminus deletion) into keratino-cytes that inhibited cell-cell adhesion, confirming theimportance of this region for homotypic adhesion.

The cadherin cytoplasmic domain, containing thebinding site for -catenin, which links to the actincytoskeleton (111), may also be a critical region inconferring a restrictive barrier for high-molecularweight molecules. Using Chinese hamster cells (whichlack cadherin and have high monolayer permeability)Navarro and coworkers (167) made stable transfectants expressing either wild-type or mutant VE-cad

herin with truncation of the cytoplasmic domain. Thewild-type transfectants had lower permeability thawas comparable to that of endothelial cells, whereasmutant transfectants had high permeability. Howevercell-cell aggregation was not impaired in either transfectant (167). Interestingly, transfection of a carboxyterminus-truncated N-cadherin also did not impaircell-cell adhesion or aggregation (59). These findingssuggest that the cadherin amino terminus is requiredfor cell-cell adhesion, whereas the carboxy terminusmay function to promote a tight restrictive barrier byanchorage to the actin cytoskeletal network.

Relation of Impaired Cell Junctionsto Barrier Dysfunction

At present, there is limited information regardingthe effects of oxidant stress on the function and organization of adherens and tight junctional proteins. Inone study, in which rat lungs were perfused with H2O2the increased extravasated FITC-albumin was not associated with altered distribution of ZO-1 as detectedby immunogold localization (182). However, in an invitro study, H2O2 treatment of human umbilical veinendothelial cells resulted in the redistribution of occlu-din and dissociation from ZO-1, which was associatedwith phosphorylation on serine residues of occludin

(119). Interestingly, Kevil et al. (117) reported thatH2O2treatment of endothelial cells promotes cadherininternalization as a possible mechanism for reducingcadherin expression. Thus disruption of endotheliajunctions occurs during oxidant stress and likely provides the basis for increased endothelial permeability(Fig. 1B).

The impairment in expression and organization othe adherens and tight junctional proteins in responseto ROS shows features similar to those caused by otherinflammatory mediators. For example, histamine (69or TNF-(23) can cause reduction or fragmentation ofZO-1 distribution on the cell periphery of microvascu

lar endothelial cells. With adherens junctions, perfusion of rat mesentery venules with TNF- combinedwith IFN- results in increased albumin permeabilityand disorganization of VE-cadherin at the junctionasite (247). The functional significance of cadherin inbarrier function was further investigated with the useof ECV304 cells, which are bladder carcinoma cellslacking cadherin. In this study, ECV304 were transfected with cDNA encoding cadherin-5 or E-cadherinand the transendothelial resistance in response to histamine was determined. The results indicated that thehistamine-mediated decreased resistance occurs pre

C721INVITED REVIEW


4/23

dominantly at the cell-cell junction and not the cell-matrix junction (Fig. 2) (242). These are the first re-ports establishing a direct functional relationship be-tween cadherin and mediator-induced endothelialpermeability changes.

Actin Filaments in Barrier Function Regulation

The actin filaments of vascular endothelium are or-

ganized into a diffuse network of short microfilamentslocalized in the cell cortex and into several types ofprominent microfilament bundles. These bundles ofmicrofilaments in in vitro cell cultures and in vivovessels undergo dynamic changes in response to phys-iological and pathological stresses, including shearstress, vascular pressure, and wound healing processes(62, 241, 244, 246). Increases in endothelial permeabil-

ity in response to edemagenic mediators are accompanied by intercellular gap formation, cell shape changeand reorganization of actin microfilament bundles. Thetypical morphological pattern of actin reorganizationassociated with endothelial barrier dysfunction is increased stress fiber density and reduction or loss of thecortical actin band (also designated as the dense peripheral band) (25, 50, 110, 143). Furthermore, in stud

ies in which endothelial cells were treated with cytochalasins, which depolymerize actin filamentsendothelial permeability was increased (120, 209)whereas in cells treated with phallacidin compoundswhich stabilize actin filaments, mediator-induced increases in endothelial permeability were diminished(5, 181). These studies clearly suggest that regulationof endothelial barrier function is dependent on the

C722 INVITED REVIEW


5/23

highly dynamic network of actin filaments. Although itis not known precisely how actin may modulate endo-thelial barrier function, some possible cellular sites ofregulation include actin-myosin contraction, intercel-lular junction assembly/disassembly, cell-matrix adhe-sion, and transmembrane signaling.

Role of stress fibers and cortical filaments.The actinfilaments of both the cortical actin band and stressfibers are associated with -actinin, myosin, and tro-pomyosin (76); thus the overall cell contraction mayengage both filament networks. There is evidence thatstress fiber formation is dependent on myosin lightchain (MLC) phosphorylation, which activates myosin

ATPase, leading to conformational changes that pro-

mote myosin filament assembly. These events result incross-linking of filamentous actin, increased contractile force generation, which drives the formationof stress fibers, and focal adhesions (27). Althoughthere is evidence supporting a role of endotheliacontraction in regulation of endothelial barrier function (66, 73, 163), it is not yet clear how this contraction-driven stress fiber formation is utilized by theendothelium to alter permeability. The finding thatactin microfilaments of the cortical actin band arelinked directly to junctional sites suggests that contractile forces generated from these filaments maysomehow affect junctional organization and function

However, this hypothesis remains to be tested.

Table 1. Oxidant stress remodels endothelial actin filament network

Endothelial Cell Type Oxidant Stress Actin Filament Network Referenc

PPAEC Xanthine/xanthine oxidase Disruption of actin filaments 208Intercellular gap formation

CPAEC 020 mM H2O2 Disruption of actin filaments 92Cell rounding

100 M Increased stress fibers 110Decreased cortical actin band

010 mM H2O2 Increased stress fibers 255Decreased cortical actin band

95% O2 Increased stress fibers 180Decreased cortical actin band

H2O2, HOCl Intercellular gap formation 172HAEC Reoxygenation Increased total F-actin 42

Increased actin to cell peripheryBAEC 10 and 100 M H2O2 Increased stress fibers 140

Decreased cortical actin bandBPMEC Reoxygenation Increased stress fibers 144

Decreased cortical actin band00.5 mM H2O2 Increased actin to cell periphery 213

Intercellular gap formation

PPAEC, porcine pulmonary artery endothelial cell; CPAEC, bovine pulmonary artery endothelial cell; HAEC, human aorta endothelial cellBAEC, bovine aorta endothelial cell; BPMEC, bovine microvascular artery endothelial cell.

Fig. 1. Subcellular endothelial sites regulated by oxidant stress.A: normal structure of the vascular endothelium.B: endothelium under oxidant stress. Site 1, lateral cell membrane: as shown in A, the lateral cell border containsthe tight junctional protein occludin (blue) in association with intermediate proteins such as zonula occludens-1(ZO-1; green), which form the linkage to cortical actin filaments; the adherens junction proteins VE-cadherin inassociation with -, -, and -catenins (green), which associates with actin-binding proteins (blue); and otherproteins in association with junctions, such as platelet endothelial cell adhesion molecule 1 (PECAM-1). As shownin B, oxidant stress causes phosphorylation (P) and redistribution of occludin, decreased levels of VE-cadherin(possibly mediated by endocytosis), disruption of linkage to cortical actin filaments through catenins and actin-binding proteins, and phosphorylation of PECAM-1. Site 2, actin filament network: as shown in A, cortical actinfilaments are shown to directly link to ZO-1 at the tight junction and to actin-binding proteins such as vinculin and-actinin (blue), which link the catenins (green) to VE-cadherin (red). Adhesion molecules (red) expressed on theendothelial luminal surface, in particular intercellular adhesion molecule-1 (ICAM-1), may also associate withcortical actin filaments. Actin stress fibers are known to associate with several intermediate proteins (green;including paxillin, focal adhesion kinase), which bind with integrins, forming focal adhesions at the basalendothelial surface. In B, oxidant stress causes reduction of the cortical actin band but increased formation ofstress fibers (see Table 1). The decreased cortical actin band may result in destabilization of the tight and adherensjunctions, leading to dissociation and redistribution of the proteins. Reactive oxygen species (ROS) have beenshown to result in increased phosphorylation of myosin light chain (MLC), which would promote stress fiberformation and cell contraction. Site 3, surface expression of adhesion molecules: as shown in A, endothelial cellsexpress constitutively ICAM-1, ICAM-2, and vascular cell adhesion molecule-1 (VCAM-1), which are members ofthe IgG superfamily of adhesion molecules (green) (see Table 2). ICAM-1 has been shown to bind directly to actinfilaments. P-selectin is (red) constitutively expressed on the cell surface and is stored preformed in Weible-Paladebodies, whereas E-selectin expression is upregulated transcriptionally and therefore requires de novo synthesis. Asshown inB, ROS may functionally upregulate adhesion molecule by altering the conformation and/or clustering ofexisting adhesion molecules through action of the cortical actin filaments. Both ICAM-1 and ICAM-2 are likelyregulated through such a mechanism. ROS also induce a quantitative upregulation of adhesion molecules byextrusion of preformed pools (i.e., P-selectin) and by de novo synthesis such as that with ICAM-1, VCAM-1, and P-and E-selectins. NF-B, nuclear factor-B.

C723INVITED REVIEW


6/23

Increasing evidence supports a relationship betweenthe endothelial barrier dysfunction and the loss orreduction of the cortical actin band. One importantfunction of the cortical actin band is the assembly andlocalization of adherens junctions to the lateral cellborder. In epithelial cells, treatment with cytochala-sins or expression of V12rac (constitutively activatedrac) has resulted in disruption of the cortical actinband, accompanied by redistribution of cadherin-cate-nin protein localization at the cell border to perinuclearGolgi vesicles (184). The reformation of cortical actinband resulted in the return of the cadherin-catenin

proteins to the cell border (184). In endothelial cells,sphingosine-1-phosphate, a specific phospholipid li-gand of the endothelial differentiation gene receptorproteins, causes dramatic increases in actin stress fi-bers and cortical actin band, which are associated withlocalization of VE-cadherin and -, -, and -cateninsto cellular junctions (135). Furthermore, with the ex-pression of mutant forms of RhoA and Rac, the sphin-gosine-1-phosphate-mediated promotion of actin fila-ments and junctions was inhibited (135). Theseobservations suggest that the localization of adherensjunctions to the cell border is highly dependent on

functional actin filament bundles, particularly the cortical actin band. Thus a conceivable mechanism bywhich edemagenic mediators increase paracellularpermeability is via disruption of the cortical actinband, leading to disassembly and loss of the junctionsfrom the cell border. This hypothesis is consistent withreports of loss or disruption of the adherens and tightjunctional proteins after exposure to a variety oedemagenic mediators, including histamine, TNF-vascular endothelial growth factor (VEGF), and thrombin.

Actin-binding proteins.The dynamic assembly and

organization of actin filaments into three-dimensionastructures that are fundamental for cell activities arecontrolled by actin-binding proteins (222). Actin-binding proteins are regulated by a multitude of signalingmolecules, including phosphoinositides (60, 158)cAMP-dependent protein kinase (87, 107), proteinkinase C (PKC) (179, 220), and Ras-related smalGTPases (173). Several other actin-binding proteinssuch as profilin (217), -actinin (60), and vinculin (71also bind to and are regulated by phosphatidylinosito4,5-bisphosphate (PIP2). Furthermore, the Rho GTPasesappear to be the underlying foundation for the remod

Fig. 2. A: histamine (1 105 M) did not significantly change the total (top) or the cell-cell resistance (middle) ofnontransfected ECV304 cells. There was a slow 10% fall in cell-matrix resistance ( bottom).B: histamine (1 105

M) decreased total and cell-cell resistances of ECV304 transfected with pLKneo-E-cadherin. There was a smalldecrease in cell-matrix resistance after histamine. C: histamine (1 105 M) decreased the total, cell-cell, andcell-matrix resistances of ECV304 transfected with pLKNeo-cadherin-5 (VE-cadherin). D: histamine decreasedtotal, cell-cell, and cell-matrix resistances of HUVEC. [Reprinted from Ref. 242.]

C724 INVITED REVIEW


7/23

eling of the actin filament network. Several actin-binding proteins and enzymes that regulate actin-bind-ing proteins have been identified to be targets of theRho GTPases (14). For example, LIM kinase catalyzesthe phosphorylation and thereby inactivates cofilin, anactin-depolymerizing protein, in arac-dependent man-ner, resulting in accumulation of filamentous actin atthe cell periphery (252). A direct target of RhoA is

Rho-kinase, which phosphorylates Thr-564 of theezrin/radixin/moesin proteins, interfering with theiractin filament/plasma membrane cross-linking func-tion (154). In addition, one of the isozymes (PIP5K)responsible for formation of PIP2is regulated by RhoA(13). It will be important to fully understand the mech-anisms by which the Rho GTPases are regulated and,in turn, how they regulate actin dynamics in the endo-thelial cell.

Tensegrity model.The coupling of the actin cytoskel-eton network to cell shape changes has been explainedon the basis of the cellular tensegrity concept (106),and therefore this conceptual model may provide someunderstanding of the mechanical forces driving the cellshape changes associated with increases in endothelialpermeability. The model is based on the premise thatthe dynamic cytoskeleton is an interconnected contin-uous network that can transmit tension. Cell-gener-ated tensile forces, such as those generated by actin-myosin, are transmitted as tension (centripetallydirected) to discrete resistive elements (e.g., cellularjunctions, focal adhesions) that oppose this tension.Thus the balance between tensile and resistive forcesdetermines the overall cell shape. However, a changein the tensile or resistive tension shifts this cellularbalance of forces, resulting in integrated changes incell shape, that is, the cell becomes more flattened or

rounded. For example, a decrease in the resistive force(e.g., through loss of cell-cell adhesion) would shift thebalance of cellular forces to favor tensile forces, result-ing in more rounding of the cell. In endothelial cells,there is no clear relationship between cell shapechange per se and barrier function; it is known onlythat shape changes do occur in association with alteredpermeability. However, the tensegrity model points tothe importance of the actin cytoskeleton as a key inte-grator of global cellular forces, which must somehowimpact on all phases of the permeability response.

Oxidants Remodel Actin Filament Network

Endothelial cells respond to oxidant stress withstriking changes in their cell shape and remodeling ofthe actin filament network (Table 1 and Fig. 1). Directtreatment of cultured endothelial cells with H2O2 orxanthine/xanthine oxidase consistently results in cellrounding and formation of intercellular gaps (92, 172,208, 213). Evidence to date indicates that the oxidant-mediated endothelial cell shape change occurs in asso-ciation with increased density of actin stress fibers andloss or disruption of the cortical actin band (110, 140,180, 255). We further observed that exposure of bovinemicrovascular endothelial cells to a period of hypoxia

followed by reoxygenation induces the production osufficient amounts of O2

to increase actin stress fiberformation and increase endothelial permeability (144)Using a replication-deficient adenovirus vector, Crawford et al. (42) overexpressed superoxide dismutase inhuman aortic endothelial cells and found that thereoxygenation-induced actin reorganization was inhibited. These findings provide evidence that the actin

cytoskeleton is a target of oxidant stress action.One mechanism of oxidant-mediated regulation o

actin is through the actin-binding proteins. Filamin isa nonmuscle cross-linking actin-binding protein thatconnects actin microfilaments to membrane glycoproteins and appears to be sensitive to oxidant stressHastie et al. (86) observed that in human umbilicavein endothelial cells treated with H2O2, filamin rapidly translocated from the membrane to the cytosolfollowed by changes of actin microfilament reorganization and intercellular gap formation, a sequence oevents suggestive of a possible causal relation betweenfilamin and barrier impairment. Hastie et al. (88) observed that H2O2-activated generation of phospholipase D (PLD) metabolites may mediate filamin redistribution and actin filament reorganization. There isalso evidence that ROS such as H2O2 can directlyoxidize actin thiols (in particular, the sulfhydryl groupof Cys-374), which likely alters the actin carboxy terminus, leading to impairment of actin subunit interactions as well as binding with actin-binding proteins(43).

Another characteristic of oxidant stress in endothelial cells is a rapid fall in cellular ATP levels (92, 218)resulting from inactivation of the glycolytic and mitochondrial pathways of ADP phosphorylation (203). Theoxidant-mediated decreased ATP level is accompanied

by actin microfilament disruption (92) and increasedendothelial permeability (97). The F-actin microfilament disruption can be mimicked by decreasing theATP level with glucose depletion and metabolic inhibition (92, 127). The metabolically induced ATP depletion also is accompanied by increased permeability (29151, 188). Additionally, metabolic inhibition potentiated the oxidant-mediated reorganization of F-actinmicrofilaments in endothelial cells (44). Although theprecise relationship of decreased ATP with F-actinreorganization and increased permeability is unclearthese observations indicate that decreased ATP may bea critical component of oxidant stress-induced barrierdysfunction.

OXIDANT-MEDIATED SIGNALING MECHANISMS

IN ENDOTHELIAL BARRIER FUNCTION

There is convincing evidence that low subcytotoxicconcentrations of ROS function as important physiological signaling molecules mediating basic cellularactivities such as cell growth and differentiation (Fig3) (112, 164). ROS are ubiquitous and short lived, andthey likely regulate signaling cascades at multiplepoints. For example, the phospholipases PLA2, PLCand PLD are activated by ROS, which in turn generate

C725INVITED REVIEW


8/23

a host of cellular messengers and cofactors that regu-late further downstream cellular activities, includingprotein kinases and phosphatases (164). Additionally,ROS may directly modulate activity of the downstreammolecules. There is also evidence suggesting that sub-lethal levels of oxidant stress inactivate tyrosine andserine/threonine protein phosphatases, contributing toincreased activities of protein kinases in several sig-

naling pathways (240). The cellular targets modulatedby ROS are highly dependent on the lipid solubility,cellular site of oxidant generation, and half-life of theparticular oxidant. We next present some emergingdata regarding how ROS modulate several key signal-ing pathways that are implicated in the regulation ofpermeability and endothelial cell-leukocyte interac-tions. Overall, these reports suggest likely complexmultiple regulatory mechanisms that control endothe-lial barrier function under oxidant stress conditions.Further research is needed at the molecular level todelineate the precise role and contribution of the signaltransduction systems involved.

Altered Intracellular Ca2 Regulation

Abundant evidence indicates that oxidant stress ofthe endothelium increases the intracellular calciumconcentration ([Ca2]i) (46, 214, 234), which correlateswith increased endothelial permeability (208, 214). Di-rect treatment of endothelial cells with H2O2increased[Ca2]i within minutes, and the increase appeared tobe sustained (214, 234). The chelation of extracellularCa2 (214) as well as treatment with the PLC inhibitorU-73122 (234) inhibited in part the oxidant-inducedincreased [Ca2]i, suggesting that Ca

2 influx and mo-

bilization, respectively, contributed to the overall increased [Ca2]i.

The increased influx mediated by xanthine/xanthineoxidase has been recently reported to be inhibited byan anion channel blocker, Ni2 (an inorganic membrane Ca2 channel blocker), dithiothreitol, and inhibitors of the Haber-Weiss reaction (15). These observations suggest that Ca2 influx occurred through

membrane Ca2 channels that were regulated by OHgeneration. The mechanism of mobilization of Ca2

from intracellular stores is likely through activation ofPLC. The treatment of pulmonary artery endotheliacells with H2O2 results in the hydrolysis of PIP2, producing diglycerides, phosphatidic acid, and inositophosphates (48, 211). The generated inositol 1,4,5trisphosphate (IP3) is a known ligand of the IP3-activated Ca2 channel on endoplasmic reticulum for Ca2

mobilization (55). However, both peroxide (78) and OH(134) also have been shown to inhibit the endoplasmicreticular Ca2-ATPase, causing increased [Ca2]i. Normally, Ca2-ATPases function to lower [Ca2]i by

transporting Ca2

from the cytosol into endoplasmicreticulum. There is some evidence that the inhibitionwas the result of the OH modifying key sulfhydrygroups on the Ca2-ATPase molecule (134).

The inorganic Ca2 entry blocker LaCl3inhibited theincrease in albumin permeability of porcine pulmonaryartery endothelial cells in response to ROS generatedby xanthine/xanthine oxidase (208), yet it was ineffective in inhibition of the H2O2-mediated increased permeability of bovine pulmonary microvascular endothelial cells (214). These contrasting findings may beattributable to the species difference of cells. Alterna-

Fig. 3. Oxidant stress mediates acute endothelial dysfunctional responses through several pathways.1) ROS mayactivate phospholipases [i.e., phospholipase C (PLC), phospholipase D (PLD), and phospholipase A2(PLA2)], whichin turn generates a multitude of cellular messengers and cofactors that are critical in the regulation of actin-binding proteins and adhesion molecules expressed on the lateral and luminal membrane surface. S/T, serine/threonine; Tyr, tyrosine. 2) ROS may modulate cellular messengers through direct and indirect oxidativemodification of the signals themselves or their regulating cofactors, which in turn will affect the downstreamtargets, actin-binding proteins and adhesion molecules.3) ROS may alter the structure and function of structural

and proteins and adhesion molecules by direct oxidative modification. Additionally, ROS depletes cellular ATP,which has profound effects on the cell cytoskeleton. These acute endothelial responses to ROS are proposedmechanisms that lead to decreases in cortical actin band, increased stress fibers, increased endothelial surfaceadhesiveness (upregulation of luminal adhesion molecules), and loss/disassembly of tight and adherens junctions.

C726 INVITED REVIEW


9/23

tively, they may be related to different regulatorymechanisms of permeability utilized by endothelialcells from conduit vs. microvascular vessels. Endothe-lial cells from different tissue sites are known to bedistinct in their nutritional requirements and re-sponses to growth and migration stimuli (254), antiox-idant profiles (231), surface expression of glycoproteins(22), and Ca2 regulation (37, 115). These observations

point to our incomplete understanding regardingmechanisms by which the rise in [Ca2]i regulatesendothelial permeability.

Although increased [Ca2]ihas been correlated withendothelial barrier dysfunction in response to ROS aswell as other edemagenic agents (38, 67, 89, 115, 145,198, 214, 223), the underlying cellular and molecularbasis remains to be resolved. The current understand-ing is that the increased permeability consists of Ca2-dependent and -independent components, and an in-creased [Ca2]i alone is likely not sufficient to fullypromote increases in permeability. The Ca2-depen-dent component of the permeability response may in-volve regulation of myosin light chain kinase (MLCK)(66, 212), the classic PKC isoforms (28, 235), and theCa2-inhibitable adenylate cyclase (221).

Myosin Light Chain Kinase

Recent findings indicate that H2O2 treatment of en-dothelial cells increases MLC phosphorylation (142,255), suggesting that endothelial contraction plays animportant role in the oxidant stress-induced endothe-lial barrier dysfunction. One important basis of in-creased vascular endothelial permeability is the acti-vation of endothelial contraction. Permeability-increasing mediators such as thrombin and histaminealso increase MLC phosphorylation (66, 147, 163),which precedes the onset of endothelial contraction.Furthermore, maximal phosphorylation has beenmaintained during sustained contraction (73). Acti-vated MLCK directly phosphorylates Thr-18 andSer-19 of MLC, which are related to isometric tensiondevelopment and increased actin polymerization in en-dothelial cells (73). Inhibition of MLCK with ML-7,which competes with ATP for binding to the kinase,reduces both the thrombin- and histamine-induced in-creased MLC phosphorylation and endothelial perme-ability (66, 163). These findings support the thesis thatactin-myosin contraction increases centripetally di-rected tension that opposes and overcomes the centrif-

ugal tethering forces, resulting in a decrease or loss ofcell-cell adhesion and the subsequent barrier dysfunc-tion. Additionally, inflammatory mediators may inac-tivate myosin-associated protein phosphatases, furtherpromoting MLC phosphorylation (53, 210). ROS havebeen reported to inactivate serine/threonine proteinphosphatases (240). However, increasing evidence in-dicates that increased endothelial permeability canalso be independent of increased MLC phosphorylation(67, 163, 178, 207), implicating the involvement ofother regulatory mechanisms such as PKC, tyrosinekinases, and Rho GTPase proteins.

Protein Kinase C

Evidence for PKC in oxidant-mediated barrier dysfunction. Several lines of evidence implicate the involvement of PKC in mediating the increased vascularendothelial permeability in response to oxidant stressIn guinea pig lungs pretreated with H-7 (a PKC inhib-itor acting on the catalytic site of the enzyme), the

increase in pulmonary capillary filtration coefficient inresponse to perfusion of H2O2 was inhibited (110). Incultured monolayers of pulmonary microvascular endothelial cells, H-7 or calphostin C (which inhibits theregulatory site of the enzyme) prevented the H2O2induced increase in albumin permeability (213). Furthermore, the increases in permeability were accompanied by reorganization of actin cytoskeleton, which wasalso inhibited by the PKC inhibitors (110, 213). Inanother study, PKC inhibitors blocked both the H2O2induced increases in endothelial permeability as wellas increased MLC phosphorylation (142).

Oxidants activate PKC. ROS activate PKC by at leastthree mechanisms. One mechanism is a direct oxida

tive modification of the regulatory domain, resulting inincreased PKC activity independent of Ca2 and phospholipids (74). However, the specific site(s) of oxidativemodification of the enzyme has yet to be identified. Asecond mechanism by which ROS activate PKC isthrough mediators and cofactors generated secondaryto activation of PLC, PLD, and PLA2. In endotheliacells, ROS cause activation of PLC and PLA2 (35, 48211) as well as PLD (165), leading to production opotent activators of PKC [i.e., Ca2, diacylglycero(DAG), free fatty acids, and phosphoinositides] (139164, 171, 224). Finally, oxidant-mediated activation oPKC also may occur through phosphorylation of dis-tinct residues of different PKC isoforms. The treatmenof COS-7 cells, which had been transfected with individual PKC isoforms, with H2O2 resulted in tyrosinephosphorylation and catalytic activation of the classicPKC (cPKC) isoforms , I, and ; the novel PKC(nPKC) isoforms and; and the atypical PKC (aPKCisoform (126). Further evaluation of PKC- showedthat the phosphorylation was restricted to tyrosineresidues, particularly Tyr-512 and Tyr-523, in the car-boxy terminal half of the enzyme that are critical forPKC activation (126). The upstream events responsiblefor the tyrosine phosphorylation of PKC isoforms areunknown. ROS are known to activate several membersof the Src family of protein tyrosine kinases (PTK) in a

wide range of cell types, including endothelial cells (316, 84, 183, 202). The PTK implicated in regulation oPKC and endothelial barrier function remain to beelucidated.

Mechanisms of PKC in regulation of barrier functionPKC is a family of serine/threonine protein kinasesconsisting of at least 12 known isoforms, which areclassified primarily by cofactor requirements for activation and function (122, 139, 171). The cPKC isoforms, I, II, and are activated by Ca2, phosphatidylserine, and DAG, whereas the nPKC isoforms , , and are activated by phosphatidylserine and DAG

C727INVITED REVIEW


10/23

and do not require Ca2. The aPKC isoforms , , and are activated by phosphatidylserine but do not re-quire DAG or Ca2. With some isoforms, cis-unsatur-ated free fatty acids (e.g., arachidonic, linoleic, andoleic acid) generated by PLA2-mediated hydrolysis ofphospholipids enhance the DAG-dependent activationof the isoforms (139, 171). These isoforms are distrib-uted to different intracellular sites, are selectively sen-

sitive to PKC inhibitors and downregulation, and havedifferent selectivity for substrates, suggesting that theisoforms have unique functions (139). In endothelialcells, PKC- and - isoforms are the primary Ca2-dependent isoforms, whereas the novel isoforms andas well as the atypical isoforms and comprise thepredominant Ca2-independent isoforms (28, 83, 235,237, 256).

The endothelial adherens and tight junctions andassociated proteins can exist as phosphoproteins (41,72, 189, 219, 233), suggesting that their function maybe regulated by the phosphorylation state of the pro-teins. Phosphorylation by PKC is a well-documentedmechanism for mediator-induced increases in perme-ability, although the precise molecular events remainundefined. Pretreatment of cultured endothelial cellmonolayers with inhibitors of PKC (i.e., H-7 and stau-rosporine) attenuates the increased albumin perme-ability in response to thrombin (149, 220). Further-more, the PKC inhibitor calphostin C has been shownto prevent thrombin-induced loss of-catenin,-cate-nin, and p120 proteins of the cadherin immunocomplex(191), suggesting that PKC-mediated phosphorylationregulates shifts in intracellular pools of cadherin/cate-nin. Current evidence indicates that increases in endo-thelial permeability are regulated by the classic PKCisoforms. Expression of full-length antisense PKC-

cDNA into endothelial cells had been shown to reducethe protein levels of endogenous PKC-, accompaniedby inhibition of the phorbol 12-myristate 13-acetate-mediated increase in endothelial permeability (235).However, thrombin-induced increases in permeabilityas well as [Ca2]ihave been potentiated by expressionof antisense PKC-(235), suggesting that PKC-maydownregulate signaling pathways activated by thethrombin receptor. Recently, the TNF--mediated in-crease in endothelial permeability was correlated to aprolonged activation pool of PKC- (56). The roles ofother PKC isoforms in endothelial barrier functionregulation remain to be determined.

Mitogen-Activated Protein Kinase

The mitogen-activated protein (MAP) kinase path-way can be activated by reactive oxidant stimuli suchas H2O2 or ischemia-reperfusion and is a key cellularresponse to oxidant stress (2, 82, 238). The upstreamregulator of MAP kinase, p21ras, has been shown to bea direct target of ROS and is believed to be a centralmechanism by which redox stress signals are transmit-ted (133). MAP kinases are proline-directed serine/threonine kinases that are activated by dual phosphor-ylation on threonine and tyrosine residues in response

to a variety of proliferation and stress stimuli (113)The MAP kinase family consists of three distincgroups, the extracellular signal-regulated kinase(ERK), the c-Jun amino-terminal kinase (JNK), andthe p38 MAP kinase. The ERK pathway is implicatedin the endothelial barrier dysfunction caused by phorbol esters, functioning downstream of Ras activation(232).

The involvement of MAP kinase in oxidant-inducedendothelial barrier dysfunction has been little studiedalthough the p38 kinase pathway is known to contribute to actin filament reorganization in response tooxidant stress (79, 104, 105). Recent work by Kevil andcoworkers (119) showed that the H2O2-mediated endothelial increase in permeability was accompanied byERK1/2 phosphorylation, occludin redistribution, andphosphorylation. Furthermore, these H2O2-mediatedactivities were inhibited by PD-98059, an ERK kinase(MEK) inhibitor.

Tyrosine Kinases

Increasing evidence indicates that the PTK pathwayis important in mediating endothelial barrier dysfunction. The promotion of phosphotyrosine content bytreatment of endothelial cells with tyrosine phosphatase inhibitors (i.e., phenyl arsine oxide and sodiumorthovanadate) increases solute permeability (236253). Several edemagenic mediators may increase endothelial permeability through tyrosine kinase-mediated phosphorylation of junctional proteins. VEGFtreatment (1590 min) of endothelial cells has resultedin tyrosine phosphorylation of VE-cadherin,-cateninplakoglobin (-catenin), p120, and platelet endotheliacell adhesion molecule-1 (PECAM-1) from human umbilical vein endothelial cells (52) and occludin-1 and

ZO-1 from bovine retinal endothelial cells (10). Although the functional significance of tyrosine phosphorylation is not clear, this may provide a means totarget the junctional proteins for degradation (236)Furthermore, tyrosine residues of paxillin and focaadhesion kinase (FAK) have been phosphorylated inbovine coronary venular endothelial cells stimulatedwith histamine or phorbol esters; the tyrosine kinaseinhibitor damnacanthal abolished this phosphorylation as well as the increased permeability (253).

Endothelial cells stimulated with H2O2 showed increased tyrosine phosphorylation of proteins at focacontacts (77, 230). Interestingly, inhibition of tyrosinephosphorylation with genistein prevented H

2

O2

-mediated increases in tyrosine phosphorylation, intercellular gap formation, and monolayer permeability (30)implicating the involvement of PTKs in the oxidantmediated permeability response.

Rho GTPases

The family of Rho GTPases are essential in the relayof signals to the actin cytoskeleton involved in a multitude of cell activities such as cell adhesion, motilitypolarity, cell cycle progression, and apoptosis. The family consists of 14 distinct members (Rac13, Cdc42

C728 INVITED REVIEW


11/23

TC10, RhoAE, G, and H, and Rnd1 and 2), of whichCdc42, Rac1, and RhoA are well characterized (13).Studies in fibroblasts indicate that RhoA causes stressfiber and focal contact formation, Cdc42 regulatesfilopodia formation, and Rac1 regulates the formationof lamellipodia.

RhoA promotes MLC phosphorylation in endothelialcells as in other cell types, suggesting a possible mech-

anism by which Rho GTPases regulate barrier function(8, 53, 65, 121). However, the involvement of RhoGTPases in the regulation of endothelial barrier func-tion remains unclear and controversial. In some stud-ies in which RhoA was inactivated in endothelial cells,the thrombin-mediated barrier dysfunction was re-duced (9, 53), whereas in another study there was noinhibition (31). Rho GTPases may be crucial for theregulation and function of adherens junctions in endo-thelial cells (24, 135). Indeed, the expression of mutantforms of RhoA and Rac was shown to inhibit the for-mation of junctions in endothelial cells (135). Althoughthere is little information regarding the effects of ROS

on Rho GTPase function, ROS are known to dramati-cally remodel the endothelial actin cytoskeleton, sug-gesting that the Rho GTPases may mediate these actinresponses. Further studies are needed to understandfully the mechanisms by which this important modu-lator of the cell cytoskeleton regulates endothelial bar-rier function under oxidant stress conditions.

LEUKOCYTE EXTRAVASATION AND OXIDANT STRESS

As mentioned earlier (EFFECTS OF OXIDANT STRESS ONVASCULAR ENDOTHELIUM), numerous experimental modelshave documented increased leukocyte extravasation in

response to oxidant stress (100, 190, 200). Further-more, using intravital microscopy, Gaboury et al. (63)showed that infusion of hypoxanthine/xanthine oxi-dase into the rat mesenteric circulation increased thenumber of both rolling and adherent leukocytes.

Current Understanding of Leukocyte Extravasation

It is now recognized that the endothelium plays akey role in the control of leukocyte adhesion and trafficking. Leukocyte extravasation, a vital host-defenseresponse, consists of precisely coordinated series ointeractions between leukocytes and endothelial cellsthat direct leukocytes out from the circulation to ex

travascular tissue sites. At sites of inflammation, infection, or injury, local mediators activate circulatingleukocytes and/or the endothelium, whereby the leukocytes bind to endothelial cells with low-affinity inter-mittent adhesions (rolling) on the endothelial surfaceThese initial events are believed to be mediated by theselectin family of adhesion molecules (L-selectin onleukocytes, P- and E-selectins on endothelial cells) thatinitiate rolling of leukocytes on the endothelial surface(Fig. 4) (157). In endothelial cells, P-selectin is storedwithin the Weibel-Palade bodies and can be rapidlyrecruited to the cell surface upon activation, whereasE-selectin expression requires de novo synthesis byactivation of the E-selectin gene. The initial adhesionis followed by development of a more firm adhesionbetween leukocytes and endothelial cells, which is postulated to be required for subsequent transmigrationThe firm adhesion is primarily mediated by activationof the2-integrin CD18 (i.e., of PMN and macrophagesor the 1-integrin CD29 (i.e., of lymphocytes) and/orupregulation (functional or quantitative) of the endothelial counter receptors, the immunoglobulin (IgGsupergene family of adhesion molecules including intercellular adhesion molecule (ICAM)-1 and ICAM-2and vascular cell adhesion molecule (VCAM)-1 (216)The firm adhesion of leukocytes onto the vascular endothelial surface is regarded to be an essential step

leading to transmigration through the intercellularjunctions (diapedesis). Thus successful recruitment oleukocytes to sites of host defense requires a concertedseries of complex interactions between the leukocyteand endothelium. However, under pathological condi

Fig. 4. General model of leukocyte extravasation. Asites of inflammation or infection, local mediators activate circulating leukocytes and/or the endothelium, inducing rolling of leukocytes on the endothelial surfaceThese initial events are mediated by the selectin family

of adhesion molecules (L-selectin on leukocytes, P- andE-selectins on endothelial cells) (157). Rolling leukocytes, in turn, develop a more firm adhesion with endothelial cells, which is mediated by activation of th2-integrin CD18 (i.e., of PMN and macrophages) or1-integrin CD29 (i.e., of lymphocytes) and/or functional/quantitative upregulation of the endothelial counterreceptors ICAM-1, ICAM-2, and VCAM-1 (216). Thefirm adhesion leads to transmigration through the intercellular junctions (diapedesis) and into the extravascular matrix.

C729INVITED REVIEW


12/23

tions, in which localized stimuli such as cytokinescause excessive and prolonged leukocyte and endothe-lial cell activation responses, excessive extravasationmay occur, resulting in tissue damage (136). Duringacute inflammation, PMN are mobilized within min-utes to hours upon stimulation, whereas monocytes arelocalized to sites of infection within approximately aday. Although the general principals of leukocyte traf-

ficking are similar, it is becoming increasingly appar-ent that the underlying cellular and molecular mech-anisms are highly dependent on the inflammatorystimulus, the specific leukocyte population, and thetissue involved.

Basis for Endothelial-Leukocytic Firm Adhesion

Oxidants promote leukocyte adhesion to endothelium.There is clear indication that oxidant stress promotesincreased leukocyte adhesion to vascular endothelium(63, 141, 176). This increased leukocyte adhesion hasbeen demonstrated to be attributable to ICAM-1- andselectin-dependent adhesion mechanisms in several

experimental systems. In the rat mesenteric circula-tion, increased rolling and adherent PMN occurred inresponse to hypoxanthine/xanthine oxidase infusionand were inhibited by antibodies directed againstCD18 or P-selectin (63). In human umbilical vein en-dothelial cells, treatment with H2O2increased expres-sion of P-selectin within 1 h and was sustained for upto 3 h, and antibody to P-selectin abolished PMN ad-hesion to the endothelial cells (176). Similarly, oxidant-mediated increased PMN adhesion to endothelium wasassociated with increased expression of ICAM-1 (141),and antibody to ICAM-1 inhibited PMN adhesion (141,206). This upregulation of ICAM-1 has been shown tobe associated with the activation of transcriptionthrough AP1/Ets elements within the promoter (195,196). Furthermore, oxidant stress may induce multiplephases of response by adhesion molecules. Oxidantstress produced by a redox imbalance (i.e., reducedglutathione) has been shown to cause a biphasic in-crease in PMN adhesion to human umbilical vein en-dothelial cells, with peak responses at 15 and 240 min,both inhibited by antibodies to P-selectin or ICAM-1and correlated to increased surface expression of theadhesion molecules (124). Overall, these results sug-gest that leukocyte adhesion is mediated by existingadhesion molecules on the endothelium and by de novosynthesis (Fig. 1). In contrast, others (29) reported a

lack of transcriptional activation and expression ofICAM-1 in response to H2O2, suggesting an absence ofde novo synthesis of ICAM-1 or that ICAM-1-indepen-dent adhesion pathways exist (see below). These dif-ferent findings underscore the incomplete understand-ing of the molecular basis of leukocyte adhesion andtransmigration under oxidant stress conditions.

Recent studies stress two factors important in theunderstanding and interpretation of cellular mecha-nisms of leukocyte extravasation. Nishio and cowork-ers (170) reported that in hyperoxia-exposed rat lungs,P-selectin was sparsely expressed in arterioles,

whereas ICAM-1 was expressed significantly invenules and capillaries, which was associated withfirm adhesion in the capillaries but not in arterioles orvenules. These findings indicate regional differences inboth expression and function of endothelial surfaceadhesion molecules, suggesting possible regional susceptibility to oxidative stress and/or intrinsic differences in expression. Furthermore, oxidant-responsive

transcription factors are regulated in a cell type-specific manner. We recently demonstrated that oxidantinduction of ICAM-1 and the PMN chemoattractanIL-8 is cell-type dependent and associated with differential activation and binding of activator protein-1(AP-1) and nuclear factor (NF)-B (130, 131, 192)H2O2 selectively induced ICAM-1 expression in endothelial cells and IL-8 in epithelial cells (130, 193). H2O2induction of IL-8 mRNA in epithelial cells was concen-tration dependent and corresponded to H2O2inductionof AP-1 binding to the IL-8 promoter (131). Thus ROSin contrast to cytokines, activate IL-8 and ICAM-1gene expression in a cell type-specific manner, andoxidant signaling appears to be an important mecha

nism for differential gene expression in endothelial andepithelial cells. These studies point to the importantnotion that regulation of leukocyte adhesion and transmigration is governed by intrinsic factors that differamong regions within the tissue, different tissues, andcell types.

Regulation of ICAM-1 function.Despite the currentcontrasting findings regarding the regulation oICAM-1 expression by ROS, it is generally agreed thatthe most important adhesion molecules in mediatingfirm adhesion of leukocytes to the endothelial surfaceafter the initial rolling events are members of the IgGfamily (i.e., ICAM-1, ICAM-2, VCAM-1). Both ICAM-1

and ICAM-2 are constitutively expressed on endothelial cells, but only ICAM-1 can be upregulated transcriptionally (195) (Fig. 1, Table 2). Intracellular vesic-ular pools of ICAM-1 have been detected in synoviaendothelial cells, in which cytokines can induce therelease and surface expression of the stored ICAM-1(33). Functional upregulation of the ICAMs can alsooccur through changes in conformation and/or distribution of the constitutively expressed forms, resultingin increased affinity (206). Evidence indicates thatICAM-1 is anchored to the actin cytoskeleton throughassociation with -actinin (20, 34). Both ICAM-1 andICAM-2 are found to associate with ezrin, a proteinknown to link the actin cytoskeleton to the cell mem

brane (90). The Rho GTPases, central in regulation ofactin organization and function, have been implicatedin the function of these adhesion molecules. In a studyby Wojciak-Stothard et al. (243), the inhibition of RhoAactivity with C3 transferase or N19RhoA (dominantnegative RhoA) resulted in inhibition of clustering ofE-selectin, ICAM-1, and VCAM-1 and reduction of ad-hesion of monocytes to activated endothelium. Thesefindings support a mechanism for stable adhesionbetween the endothelium and leukocytes in which theactin filament network and its associated regulatorymolecules function to redistribute/assemble clustering

C730 INVITED REVIEW


13/23

of adhesion molecules. The specific role(s) of endothe-lial actin filaments and Rho GTPases in the regulationof leukocyte adhesion and transmigration by ROS re-main to be shown.

ICAM-1-independent adhesion.It is increasingly ap-parent that ICAM-1-independent adhesion pathwaysexist (26, 57, 150, 161, 245). For example, double-knockout mice deficient in P-selectin and ICAM-1showed a complete lack of PMN emigration into the

peritoneum duringStreptococcus pneumoniae-inducedperitonitis, whereas PMN emigration was not inhib-ited in S. pneumoniae-induced pneumonia, indicatingan ICAM-1-independent response in the lung (26).However, in pneumonia induced with either Esche-richia coli lipopolysaccharide or Pseudomonas aerugi-nosa, CD18-deficient PMN, but not wild-type PMN,showed impairment in emigration, whereas in pneu-monia induced with S. pneumoniae, emigration ofCD18-deficient and wild-type PMN was similar, indi-cating stimulus specificity of the responses within thesame tissue (161). This latter study is further sup-ported by a recent in vitro study in which PMN trans-endothelial migration was determined in response to

three chemoattractants: fMLP, leukotriene B4 (LTB4),or IL-8 (150). These authors reported that function-block-ing antibody to CD18 decreased70% of fMLP-mediatedand 20% of LTB4-mediated PMN transmigration buthad no effects on IL-8-mediated transmigration. Interest-ingly, the study also showed that both the LTB4- andIL-8-mediated PMN transmigrations were insensitiveto blockade with antibodies to P-selectin, E-selectin,and CD29. These findings indicate that activation ofICAM-1-mediated adhesion in endothelial cells ishighly dependent on the inflammatory stimulus andthat ICAM-1-independent pathways exist for leukocyte

adhesion and transmigration. It is not known whetheroxidant stress of endothelial cells may also mobilizeICAM-1-independent adhesion pathways.

Redox-Sensitive Transcription Factors

The final targets of signal transduction pathways arenuclear transcription factors. AP-1 and NF-B are two

major redox-sensitive transcription factors that areactivated through the PKC and MAP kinase pathwaysand mediate endothelial cell gene responses to oxidantstress (Fig. 5). AP-1 is a large family of basic leucinezipper (bZip) transcription factors that are induced bygrowth factors, tumor-promoting phorbol esters, andregulatory cytokines [reviewed by Karin et al. (114)]AP-1 consists of two subfamilies, Jun (members include c-Jun, JunB, and JunD) and Fos (members include c-Fos, FosB, Fra1, and Fra2), that bind a commonDNA recognition site as either Jun homodimers or asmore stable Jun/Fos heterodimers. Jun and Fos canalso form heterodimers with the cAMP response ele

ment binding protein/activating transcription factor(CREB/ATF) transcription factor family (185, 194).AP-1 binding activity in vitro is regulated by the

redox status of a single conserved cysteine residue inthe DNA binding domains of Jun and Fos (1, 169). Thiscysteine must be in the reduced state for DNA bindingto occur, and changing it to a serine residue results inan increase in AP-1 binding activity that is no longerredox regulated (174). An AP-1 regulatory proteincalled Ref-1, has been identified and shown to regulateAP-1 binding activity in vivo through its redox effectson the regulatory cysteine residue (250, 251). Ref-1activity itself is regulated by a redox mechanism in

volving thioredoxin and by phosphorylation by caseinkinase II (58, 94).The three MAP kinase pathways regulate AP-1 ac

tivity both by increasing the transcription of the junandfos genes and by phosphorylation of newly synthesized AP-1 transcription complexes. c-Fos gene transcription is activated by the cooperative interaction oElk-1 with the serum response factor (SRF). Phosphorylation of Elk-1 by ERK, JNK, and p38 MAP kinasesincreases ternary complex formation with the SRFthereby activating c-Fos gene transcription. The activation of c-Jun gene transcription is mediated by twoAP-1 binding sites within the jun promoter that areconstitutively occupied by heterodimers of c-Jun andATF-2. JNK and p38 kinases phosphorylate c-Jun andATF-2 on specific serine residues that are required toactivate transcription of the jun gene (40). Newly synthesized AP-1 is also regulated by the MAP kinasesJNK phosphorylates c-Jun, whereas ATF-2 is phosphorylated by both JNK and p38.

The NF-B family of transcription factors is composed of two groups of structurally related, interactingproteins that bind B elements as dimers and whoseactivity is regulated by subcellular location (reviewedin Ref. 19). NF-B family members of the first group

Table 2. Endothelial adhesion moleculesin leukocyte trafficking

Adhesion Molecule SynonymConstitutive (C),

I nducible (I) Major Ligands

Expression on Luminal Surface

SelectinsP-selectin CD62P C,I PSGL-1

E-selectin CD62E I PSGL-1, ESL-1ImmunoglobulinsICAM-1 CD54 C,I LFA-1 (CD11a/CD18);

Mac-1 (CD11b/CD18);gp150/95 (CD11c/

CD18)ICAM-2 CD102 C LFA-1; Mac-1VCAM-1 CD106 C,I 4/1 (CD29, VLA-4);

4/7

Expression on Lateral Membrane

ImmunoglobulinsPECAM-1 CD31 C CD31, v/3

Junctional proteinsCadherin C Cadherin, integrinOccludin C Occludin

PSGL-1, P-selectin glycoprotein; ESL-1, E-selectin ligand; ICAM,intercellular adhesion molecule; VCAM, vascular cell adhesion mol-ecule; PECAM, platelet endothelial cell adhesion molecule.

C731INVITED REVIEW


14/23

include NF-B1 (p50) and NF-B2 (p52), which aresynthesized as precursor proteins. The second groupincludes Rel A (p65), Rel B, and c-Rel, which aresynthesized as mature proteins containing one or morepotent transactivation domains. Each NF-B subunithas distinct binding and transactivation propertiesand when combined as homo- or heterodimers, theyform transcriptional activators with distinct activities

(129). In resting cells, latent NF-B is complexed to aclass of cytoplasmic retention proteins called inhibitorsof NF-B (IB). Signals that induce NF-B activityresult in the phosphorylation of IB on specific serineresidues, marking the protein inhibitor for ubiquitination and subsequent proteolytic degradation by theATP-dependent 26S proteasome complex. The newlyreleased NF-B is then free to move to the nucleusbind to its recognition site, and activate gene transcription.

The NF-B signaling pathway is activated by theproinflammatory cytokines TNF- and IL-1, whichare the major cytokine inducers of gene expression in

endothelial cells. These cytokines increase intracellular ROS, and their effects can be suppressed by thiolantioxidants such as N-acetyl-L-cysteine (NAC) andpyrrolidine dithiocarbamate (PDTC) (186, 196). TNF-activates NF-B through a kinase-mediated phosphorylation cascade involving a high-molecular-mass kinase complex (159). Cytokine activation of the multiprotein IB kinase complex, which contains two kinasecomponents, the IB kinase (IKK)/heterodimer andNF-B-inducing kinase (NIK), results in the serinephosphorylation of IB and the subsequent activationof NF-B (249). Blockage of IB phosphorylation byserine proteases or IB degradation by inhibition of the26S proteasome has been shown to suppress the TNF-

induction of gene expression (36, 109).In addition to phosphorylation events, intracellular

thiol redox status appears to be a critical determinantof NF-B activation. The major intracellular thiol, thetriamino acid peptide glutathione (GSH), inhibits theserine phosphorylation of IB, which is required forNF-B activation (39). NAC, a precursor of GSH, alsohas been shown to inhibit NF-B activation (204)Thus, at high levels of cytosolic GSH, NF-B is notactivated. However, NF-B binding activity is alsoinhibited by high concentrations of glutathione disulfide (GSSG), the oxidized form of GSH, suggesting thatoptimal cellular redox ratios of GSH to GSSG are

required for effective NF-B activation (47, 64) andpossibly explaining the cell type-specific redox regulation of NF-B. Indeed, GSH-depleting agents such asdiamide and buthionine sulfoximine (BSO) alter theintracellular redox state and inhibit oxidant-inducedactivation of NF-B as well as AP-1 (i.e., c-Jun/ATF-2(124, 197). Recently, it was demonstrated that modulation of the GSH/GSSG ratio in endothelial cellproduces a biphasic PMN-endothelial cell adhesionresponse mediated by transcription-dependent and -independent surface expression of different endotheliacell adhesion molecules (124).

Fig. 5. Intracellular ROS signals are transduced to specific tran-scription factor signaling pathways by redox-sensitive signaltransducers including the GTP binding proteins Ras and Rac,protein kinase C (PKC), the ratio of glutathione (GSH) to reducedglutathione (GSSG), and the intracellular thiol, thioredoxin. Thetranscription factor AP-1, composed of jun and fos heterodimers,is activated by the three mitogen-activated protein (MAP) kinase

pathways, which lead to the transcriptional activation of the junand fos genes as well as the phosphorylation of the Jun and Fosproteins. The j un gene is activated by the j un/ATF-2 hetero dimerbinding to the ju npromoter. Jun is activated by serine phosphor-ylation induced by c-Jun NH2-terminal kinase (JNK), and ATF-2is activated by p38 phosphorylation. The c-Fos gene is activated bythe transcription factor Elk-1 binding to the f os promoter. Elk-1,an Ets family transcription factor, is activated by extracellularsignal-regulated kinase (ERK) phosphorylation. jun a nd fos com-bine to form heterodimers that bind to AP-1 sites within genepromoters. NF-B is retained in the cytoplasm complexed by theinhibitor protein IB. Activation of cytokine receptors such astumor necrosis factor- (TNF-) and interleukin-1 (IL-1) activatea cascade of kinases that lead to the phosphorylation of IB, whichleads to the degradation of IB by the 26S proteasome. Thedegradation of IB allows NF-B (p50/p65 heterodimers) to trans-locate to the nucleus and bind to elements in gene promoters.NF-B activation has been shown to be modulated by redox-sensitive signal transducers such as glutathione, thioredoxin, andPKC. STAT (signal transducers and activators of transcription)proteins are also retained in the cytoplasm and dimerize whenphosphorylated by cytokine receptor-associated kinases. Dimer-ization allows STAT proteins to translocate to the nucleus andbind responsive elements in gene promoters. STAT activation andbinding activity is modulated by redox-sensitive signal transduc-ers such as glutathione and the hydroxyl radical. The threeredox-regulated transcription factors, AP-1, STAT, and NF-B, canfunction independently or act in concert to cooperatively activatetranscription from oxidant-responsive genes such as ICAM-1 andIL-8.

C732 INVITED REVIEW


15/23

Another physiologically important thiol that contrib-utes to the regulation of NF-B activity is thioredoxin,an oxidoreductase with potent antioxidant properties(201). Thioredoxin increases NF-B binding activitythrough the reduction of a redox-sensitive cysteineresidue in the DNA binding domain (128, 160). Thecysteine residue must be maintained in a reduced stateto allow DNA binding. Paradoxically, antioxidants

such as -lipoate, NAC, and PDTC inhibit oxidant-mediated NF-B activation induced by TNF-, phorbolester, or H2O2(204), suggesting that the redox regula-tion of NF-B, like that of AP-1, is complex and medi-ated by both oxidant and antioxidant mechanisms.

Recently, in addition to AP-1 and NF-B, oxidantstress has been shown to modulate the activity of thetranscription factor STAT (signal transducers and ac-tivators of transcription), a family of transcription fac-tors activated by tyrosine phosphorylation signalstransmitted from cytokine receptors such as interfer-ons and interleukins (Fig. 5). Receptor activation ini-tiates via receptor-associated kinases the phosphoryla-tion of specific tyrosine residues within the STATtransactivation domain, which allows for the dimeriza-tion and nuclear translocation of STAT transcriptionfactors. Nuclear localization of STAT allows thedimeric factors to interact in a sequence-specific man-ner with the IFN- response sequence (GAS) and theinterferon-stimulated response element (IRE). Thebinding of STAT to these promoter elements activatestranscription of cytokine responsive genes such asICAM-1 in endothelial cells.

The reactive oxidant H2O2 activates the STAT sig-naling pathway, promoting the nuclear translocationand binding of STAT-1 and STAT-3 to their respectivepromoter elements (32, 215). Also, H2O2stimulates the

activity of the known STAT kinases JAK2 and TYK2.Moreover, intracellular oxidant stress activates theSTAT pathway, because oxidized low-density lipopro-tein or depletion of intracellular glutathione enhancesSTAT-1 and STAT-3 binding activity (156).

Transmigration Through Endothelial Junctions

Relation to barrier dysfunction.The increased endo-thelial permeability caused by ROS may facilitate leu-kocyte transmigration (117, 119). Most current evi-dence indicates that leukocyte extravasation involvesmigration of the leukocyte between endothelial cells(i.e., through intercellular junctions). Studies of the

relationship between leukocyte diapedesis and endo-thelial barrier function suggest that these events areassociated with some degree of barrier impairment (70,227). In one study, the addition of PMN (at a PMN-to-endothelial cell ratio of 10:1) in response to fMLP orLTB4caused a rapid (within minutes) transendothelialresistance drop that preceded PMN transmigration(70), suggesting that barrier impairment may lead totransmigration. In support of this notion are studiesdemonstrating that antibody to VE-cadherin increasesendothelial permeability as well as PMN transmigra-tion (75, 98). In contrast, in another study, no change

in transendothelial resistance was detected for up to 30min after addition of PMN (using the same PMN-toendothelial cell ratio as reported above) in response tofMLP (101). Furthermore, in a study in which albuminflux and transendothelial migrations of lymphocyteswere monitored continuously, increased transmigration occurred after the increased albumin clearancerate had returned toward baseline (49). Therefore, the

notion that leukocyte transmigration requires increased endothelial permeability remains somewhacontroversial.

Interestingly, several studies have demonstratedthat leukocyte adhesion and transmigration remodeendothelial junctions. In an immunofluorescencestudy, adhesion of bovine coronary venular endotheliacells with C5a-activated PMN resulted in loss of VEcadherin and -catenin, increased stress fibers, andintercellular gap formation (227). In this same study(227) phosphorylation of VE-cadherin and -cateninwas accompanied by endothelial permeability. Moreover, adhesion of PMN to TNF--activated human umbilical vein endothelial cells also resulted in loss o

VE-cadherin and its associated proteins as detected byimmunofluorescence (6, 45). Corresponding immunoprecipitation studies corroborated this loss of adherensjunctional proteins; however, there was a concern thatthese latter findings were caused by PMN proteasesreleased upon detergent lysis of cells (162). A morerecent immunofluorescence confocal microscopic studyby Allport and coworkers (7) found that under flowconditions, monocyte transmigration induces focal reversible loss of VE-cadherin and -catenin. The authors hypothesized that the VE-cadherin complex wasdisrupted locally only by active transmigration andthat the complex resealed after diapedesis. The finding

of reversible junctional remodeling bolsters the argument that loss of adherens junctional proteins accompanied the leukocyte adhesion and transmigrationevents.

PECAM-1 is another adhesion molecule importantfor leukocyte extravasation and has been implicated inthe regulation of leukocyte transmigration in oxidantstress. PECAM-1 is located at the endothelial lateracell border and also is expressed by leukocytes; thus itis able to form homotypic as well as heterotypic adhesions. Superfusion of H2O2 onto rat mesenteric microvasculature results in increased leukocyte extravasation inhibitable by antibody to PECAM-1 (200)Similarly, antibody to PECAM-1 inhibits the increased

transmigration of monocyte-like HL-60 cells across endothelial cells activated by the oxidant tert-butylhydroperoxide (190).

Endothelial signaling events involved in transmigration.The oxidant-mediated increased [Ca2]ias well asphosphorylation of MLC in endothelial cells may provide critical regulatory signals in the regulation oleukocyte transmigration. Transient increases in endothelial [Ca2]i occur in response to PMN adhesion toand transmigration across fMLP- or IL-1-activated human umbilical vein endothelial cells (102). Abrogationof the [Ca2]i increase inhibits the PMN transmigra

C733INVITED REVIEW


16/23

tion as well as the increase in permeability, but notPMN adhesion. Furthermore, PMN adhesion andtransmigration in response to chemoattractants fMLPor LTB4 resulted in phosphorylation of endothelialMLC, and pharmacological inhibition of MLCK de-creased transmigration (68, 95, 199). The importanceof MLC-mediated contraction is further supported byfindings that stabilization of actin filaments with phal-

loidin or jasplakinolide inhibits leukocyte transmigra-tion across the endothelium (12, 98). These studiessuggest a model by which leukocyte-endothelial inter-actions during adhesion and transmigration eventssignal the activation of endothelial cells, resulting inendothelial contraction, junctional remodeling, and im-paired barrier, conditions that would facilitate trans-migration. Importantly, the studies implicate the in-volvement of Ca2 and MLC, both of which areelevated by oxidant stress, suggesting that these sig-nals may be key regulatory mechanisms in transmi-gration.

Oxidant stress has been shown to stimulate theproduction and release of platelet-activating factor(PAF) in endothelial cells (138). Once released extra-cellularly, PAF becomes bound to the endothelial cellsurface and mediates PMN adhesion (138). PAF is apotent phospholipid autocoid implicated in a number ofpathophysiological conditions, including inflammation,ischemia-reperfusion injury, and shock (125). The ad-ministration of a PAF-receptor antagonist (WEB-2086)or a PAF antagonist (BN-52021) abolishes leukocyterecruitment mediated by perfusion of xanthine/xan-thine oxidase in the intact lung (63) and pancreasmicrovasculature (100), supporting the idea that PAFproduction is a key component contributing to oxidant-mediated endothelial dysfunction. Recent reports have

indicated that oxidant stress produces PAF-like phos-pholipids from phosphatidylcholine in greater abun-dance and with greater potency than PAF (152, 228). Itis evident that PAF as well as PAF-like phospholipidsare critical factors in the pathophysiology of vascularendothelial dysfunction under oxidant stress condi-tions. It will be important to understand the mecha-nisms by which ROS regulate the production and re-lease of these phospholipid mediators.

CONCLUSIONS AND FUTURE PERSPECTIVES

Oxidant stress, a pervasive condition of increasedamounts of ROS, constitutes a serious pathophysiolog-

ical factor for a wide variety of vascular-based disor-ders. It is apparent that its effects on the vascularendothelium are multifactorial, and two significantconsequences are endothelial barrier dysfunction andincreased adhesion for leukocytes. At present, our un-derstanding of how ROS increase endothelial perme-ability and leukocyte extravasation is not well defined.Furthermore, despite much available information re-garding signaling pathways that are activated by ROS,the causal relations between the signals generated andthe functional response have not been rigorously inves-tigated. Several aspects of oxidant stress-mediated en-

dothelial dysfunction should be emphasized for futureinvestigation.1) For example, what are the molecularmechanisms by which ROS disassemble/assemble thetight and adherens junctions? The importance of theRho GTPases in regulation of the actin cytoskeletonis evident. Does oxidative stress regulate the RhoGTPases? Furthermore, oxidant-mediated remodelingof the actin cytoskeleton is implicated in the loss of

junctional organization and stability, but this hypothesis has not been vigorously tested. Several proteinkinases have been identified to regulate barrier function; yet specific roles and targets of these kinases (andphosphatases) remain unresolved. 2) An exciting areaof study is defining the molecular communication orinteractions between the endothelial cell and the transmigrating leukocyte under oxidant stress conditionsThis is particularly important in light of the fact thatmechanisms governing leukocyte extravasation arehighly stimulus, and likely tissue, specific. It has beenshown that the adherent or transmigrating leukocytecan induce endothelial cell activation responses, including remodeling of endothelial junctions and theactin cytoskeleton. The functional significance and theregulatory mechanisms engaged by these interactionsduring oxidant stress have yet to be determined. 3) Asshown by Kokura et al. (124), a redox imbalance inendothelial cells results in both a transcription-independent and -dependent surface expression of differentendothelial cell adhesion molecules, suggesting thaoxidant stress induces acute and chronic phases oleukocyte adhesion to the endothelium. The control ofthese different facets of leukocyte adhesion likely engages different regulatory mechanisms. Future studiesare needed to identify and define the molecular eventsby which the endothelium regulates leukocyte adhe

sion and barrier function in the context of acute (minutes) vs. chronic (more than several hours) oxidanstress. This information is important in fully understanding the mechanisms of oxidant stress-inducedendothelial dysfunction.

This work was supported by National Institutes of Health GrantHL-62649 (to H. Lum) and AR-45835 (to K. A. Roebuck).

REFERENCES

1. Abate C, Patel L, Rauscher FJ, and Curran T. Redoxregulation offos and jun DNA-binding activity in vitro.Scienc249: 11571161, 1990.

2. Abe J, Kusuhara M, Ulevitch RJ, Berk BC, and Lee JDBig mitogen-activated protein kinase 1 (BMK1) is a redox

sensitive kinase. J Biol Chem 271: 1658616590, 1996.3. Abe J, Takahashi M, Ishida M, Lee JD, and Berk BC. c-Sr

is required for oxidative stress-mediated activation of big mitogen-activated protein kinase 1 (BMK1). J Biol Chem 2722038920394, 1997.

4. Alexander JS, Blaschuk OW, and Haselton FR. An Ncadherin-like protein contributes to solute barrier maintenancein cultured endothelium. J Cell Physiol 156: 610618, 1993.

5. Alexander JS, Hechtman HB, and Shepro D. Phalloidinenhances endothelial barrier function and reduces inflammatory permeability in vitro. Microvasc Res 35: 308315, 1988.

6. Allport JR, Ding H, Collins T, Gerritsen ME, and Luscinskas FW. Endothelial-dependent mechanisms regulate leukocyte transmigration: a process involving the proteasome and

C734 INVITED REVIEW


17/23

disruption of the vascular endothelial-cadherin complex at en-dothelial cell-to-cell junctions. J Exp Med 186: 517527, 1997.

7. Allport JR, Muller WA, and Luscinskas FW. Monocytesinduce reversible focal changes in vascular endothelial cadherincomplex during transendothelial migration under flow. J CellBiol 148: 203216, 2000.

8. Amano M, Mukai H, Ono Y, Chihara K, Matsui T, Hama-jima Y, Okawa K, Iwamatsu A, and Kaibuchi K. Identifi-cation of a putative target for Rho as the serine-threoninekinase protein kinase N. Science 271: 648650, 1996.

9. Amerongen GV, Draijer R, Vermeer MA, and van Hins-bergh VM. Transient and prolonged increase in endothelialpermeability induced by histamine and thrombin. Role of pro-tein kinases, calcium, and RhoA.Circ Res83: 11151123, 1998.

10. Antonetti DA, Barber AJ, Khin S, Lieth E, Tarbell JM,and Gardner TW. Vascular permeability in experimental di-abetes is associated with reduced endothelial occludin content:vascular endothelial growth factor decreases occludin in retinalendothelial cells. Penn State Retina Research Group. Diabetes47: 19531959, 1998.

11. Armstead WM, Mirro R, Thelin OP, Shibata M, Zucker-man SL, Shanklin DR, Busija DW, and Leffler CW. Poly-ethylene glycol superoxide dismutase and catalase attenuateincreased blood-brain barrier permeability after ischemia inpiglets. Stroke23: 755762, 1992.

12. Asako H, Wolf RE, Granger DN, and Korthuis RJ. Phal-

loidin prevents leukocyte emigration induced by proinflamma-tory stimuli in rat mesentery. Am J Physiol Heart Circ Physiol263: H1637H1642, 1992.

13. Aspenstrom P.Effectors for the Rho GTPases.Curr Opin CellBiol 11: 95102, 1999.

14. Aspenstrom P.The Rho GTPases have multiple effects on theactin cytoskeleton. Exp Cell Res 246: 2025, 1999.

15. Az-ma T, Saeki N, and Yuge O. Cytosolic Ca2 movements ofendothelial cells exposed to reactive oxygen intermediates: roleof hydroxyl radical-mediated redox alteration of cell-membraneCa2 channels.Br J Pharmacol 126: 14621470, 1999.

16. Barchowsky A, Munro SR, Morana SJ, Vincenti MP, andTreadwell M. Oxidant-sensitive and phosphorylation-depen-dent activation of NF-B and AP-1 in endothelial cells. Am JPhysiol Lung Cell Mol Physiol269: L829L836, 1995.

17. Barnard JW, Patterson CE, Hull MT, Wagner WWJ, andRhoades RA. Role of microvascular pressure in reactive oxy-

gen-induced lung edema. J Appl Physiol 66: 14861493, 1989.18. Barnard ML and Matalon S. Mechanisms of extracellular

reactive oxygen species injury to the pulmonary microvascula-ture. J Appl Physiol 72: 17241729, 1992.

19. Barnes PJ. Nuclear factor-kappa B.Int J Biochem Cell Biol 29:867870, 1997.

20. Barton WW, Wilcoxen SE, Christensen PJ, and Paine R.Association of ICAM-1 with the cytoskeleton in rat alveolarepithelial cells in primary culture. Am J Physiol Lung Cell MolPhysiol271: L707L718, 1996.

21. Beckman JS, Beckman TW, Chen J, Marshall PA, andFreeman BA. Apparent hydroxyl radical production by per-oxynitrite: implications for endothelial injury from nitric oxideand superoxide.Proc Natl Acad Sci USA 87: 16201624, 1990.

22. Belloni PN and Nicolson GL. Differential expression of cellsurface glycoproteins on various organ-derived microvascularendothelia and endothelial cell cultures. J Cell Physiol 136:

389398, 1988.23. Blum MS, Toninelli E, Anderson JM, Balda MS, Zhou JY,

Donnell LO, Pardi R, and Bender JR. Cytoskeletal rear-rangement mediates human microvascular endothelial tightjunction modulation by cytokines. Am J Physiol Heart CircPhysiol273: H286H294, 1997.

24. Braga VM.Small GTPases and regulation of cadherin depen-dent cell-cell adhesion. Mol Pathol 52: 197202, 1999.

25. Brett J, Gerlach H, Nawroth P, Steinberg S, Godman G,and Stern D. Tumor necrosis factor/cachectin increases per-meability of endothelial cell monolayers by a mechanism involv-ing regulatory G proteins. J Exp Med 169: 19771991, 1989.

26. Bullard DC, Qin L, Lorenzo I, Quinlin WM, Doyle NA,Bosse R, Vestweber D, Doerschuk CM, and Beaudet AL.

P-selectin/ICAM-1 double mutant mice: acute emigration oneutrophils into the peritoneum is completely absent but inormal into pulmonary alveoli. J Clin Invest 95: 178217881995.

27. Burridge K, Chrzanowskawodnicka M, and Zhong CLFocal adhesion assembly. Trends Cell Biol 7: 342347, 1997.

28. Bussolino F, Silvagno F, Garbarino G, Costamagna CSanavio F, Arese M, Soldi R, Aglietta M, Pescarmona GCamussi G, and Bosia A. Human en

oxidant stress and endothelial cell dysfunction

Documents