NAD(P)H oxidase mediates the endothelial barrier dysfunction

Am J Physiol Lung Cell Mol Physiol 286: L37–L48, 2004.
First published June 13, 2003; 10.1152/ajplung.00116.2003.
CALL FOR PAPERS
Oxidant Signaling in Lung Cells
NAD(P)H oxidase mediates the endothelial barrier
dysfunction induced by TNF-␣
Nancy Gertzberg,1 Paul Neumann,1 Victor Rizzo,1 and Arnold Johnson1,2
1
Center for Cardiovascular Science, Albany Medical College, and 2Research
Service, Stratton Veterans Affairs Medical Center, Albany, New York 1228
Submitted 16 April 2003; accepted in final form 5 June 2003
Gertzberg, Nancy, Paul Neumann, Victor Rizzo, and Arnold
Johnson. NAD(P)H oxidase mediates the endothelial barrier dysfunction induced by TNF-␣. Am J Physiol Lung Cell Mol Physiol 286:
L37–L48, 2004. First published June 13, 2003; 10.1152/ajplung.
00116.2003.—We tested the hypothesis that the NAD(P)H oxidasedependent generation of superoxide anion (O⫺
2 䡠) mediates tumor
necrosis factor-␣ (TNF)-induced alterations in the permeability of
pulmonary microvessel endothelial monolayers (PMEM). The permeability of PMEM was assessed by the clearance rate of Evans
blue-labeled albumin. The NAD(P)H oxidase subcomponents p47phox
and p22phox were assessed by immunofluorescent microscopy and
Western blot. The reactive oxygen species O⫺
2 䡠 was measured by the
fluorescence of 6-carboxy-2⬘,7⬘-dichlorodihydrofluorescein diacetatedi(acetoxymethyl ester), 5 (and 6)-chloromethyl-2⬘,7⬘-dichlorodihydrofluorescein diacetate-acetyl ester, and dihydroethidium. TNF treatment (50 ng/ml for 4.0 h) induced 1) p47phox translocation, 2) an
increase in p22phox protein, 3) increased localization of p47phox with
p22phox, 4) O⫺
2 䡠 generation, and 5) increased permeability to albumin.
p22phox antisense oligonucleotide prevented the TNF-induced effect
on p22phox, p47phox, O⫺
2 䡠, and permeability. The scrambled nonsense
oligonucleotide had no effect. The TNF-induced increase in O⫺
2 䡠 and
permeability to albumin was also prevented by the O⫺
2 䡠 scavenger
Cu-Zn superoxide dismutase (100 U/ml). The results indicate that the
activation of NAD(P)H oxidase, via the generation of O⫺
2 䡠, mediates
TNF-induced barrier dysfunction in PMEM.
antisense; edema; mRNA; permeability; transcription
TNF (34, 35) INDUCES AN INCREASE in pulmonary vascular permeability in vivo (2, 15, 20, 34), in the isolated lung (14), in
pulmonary arterial endothelial cell monolayers (6, 12, 26), and
in pulmonary microvessel endothelial monolayers (2, 8). The
mechanism for the TNF-induced increase in permeability of
the microvessel endothelial monolayer is not clear. We previously showed that the TNF-induced increase in permeability is
mediated by protein kinase C-␣ (PKC-␣) (8) and endothelial
nitric oxide synthetase (eNOS) (2). Yet, TNF causes the
PKC-␨-dependent activation of NAD(P)H oxidase in pulmonary endothelial monolayers (30). Concerning NAD(P)H oxidase, our previous data indicate a possible role for NAD(P)H
oxidase-dependent O⫺
2 䡠 in TNF-induced endothelial dysfunction, because Cu-Zn superoxide dismutase (SOD) prevents the
TNF-induced 1) decrease in endothelium-dependent dilation
(7), 2) increase in oxidized glutathione (27), 3) latent decrease
Address for reprint requests and other correspondence: A. Johnson, Mailstop 151, 113 Holland Ave., Stratton VA Medical Center, Albany, NY 12208
(E-mail: jmurd@msn.com).
http://www.ajplung.org
in 䡠NO (18), 4) generation of peroxynitrite (ONOO⫺) (6, 27),
and 5) activation of activator protein (AP)-1 (10).
NAD(P)H oxidase, recently explored in endothelium, is an
enzyme that produces superoxide anion (O⫺
2 䡠) via the reaction
⫹
⫹
2O2 ⫹ NAD(P)H 3 2O⫺
2 䡠 ⫹ NADP ⫹ H (21–23, 32). The
NAD(P)H oxidase enzyme complex is composed of a membrane-bound flavocytochrome b558 and the cytosolic activating factors Rac-2, p40phox, p47phox, and p67phox (21–23, 32).
The membrane-bound flavocytochrome b558 consists of two
heme redox centers and the subcomponents p22phox and
gp91phox. The translocation of Rac-2, p40phox, p47phox, and
p67phox to the flavocytochrome b558 is generally considered
the signature event for activation of NAD(P)H oxidase and the
subsequent generation of O⫺
2 䡠 (21–23, 32). The activation of
NAD(P)H oxidase in response to TNF is theorized to occur by
an increase in PKC-mediated phosphorylation of p47phox,
which causes its translocation to the peripheral membrane
(21–23, 32). This notion is supported by our previous work that
indicates the TNF-induced increase in pulmonary endothelial
permeability is dependent on PKC-␣, a potential activator of
NAD(P)H oxidase (8). However, the role of NAD(P)H oxidase-dependent generation of O⫺
2 䡠 in the TNF-induced pulmonary microvessel endothelial barrier dysfunction is not known.
Therefore, we hypothesize that the increased endothelial permeability in response to TNF is mediated by NAD(P)H oxidase-dependent generation of O⫺
2 䡠.
MATERIALS AND METHODS
Pulmonary Microvessel Endothelial Cell Culture
Bovine lung microvessel endothelial cells (BLMVEC) derived
from fresh calf lungs were obtained at 4th passage (Vec Technologies,
Rensselaer, NY). The preparations were identified by Vec Technologies as pure populations by 1) the characteristic “cobblestone” appearance as assessed by phase contrast microscopy, 2) the presence of
factor VIII-related antigen (indirect immunofluorescence), 3) the uptake of acylated low-density lipoproteins, and 4) the absence of
smooth muscle actin (indirect immunofluorescence).
For all studies, BLMVEC were cultured from 4 to 12 passages (2,
8) in medium containing Dulbecco’s modified Eagle’s Medium
(DMEM; GIBCO-BRL, Grand Island, NY) supplemented with 20%
fetal bovine serum (FBS; Hyclone Laboratories, Logan, UT), 15
␮g/ml endothelial cell growth supplement (Upstate Biotechnology,
Lake Placid, NY), and 1% nonessential amino acids (GIBCO-BRL)
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked “advertisement”
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
L37
L38
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
and maintained in 5% CO2 plus humidified air at 37°C. A confluent
pulmonary microvessel endothelial monolayer (PMEM) was reached
within two to three population doublings, which took 3–4 days.
Treatments
TNF treatment. Highly purified recombinant human TNF from
Escherichia coli (Calbiochem-Novabiochem, La Jolla, CA) in a stock
solution of 10 ␮g/ml was used. The endotoxin level was ⬍0.1 ng/␮g
of TNF as determined by standard limulus assay (2, 8). We previously
showed that boiling TNF for 0.75 h blocks the effect of TNF in our
system (6, 7, 16, 18), which indicates no endotoxin contamination. A
TNF dose of 50 ng/ml was used throughout this study because our
previous studies indicated that this dose causes a PKC-␣- and 䡠NOdependent increase in permeability of PMEM (2, 8).
SOD. SOD (Sigma, St. Louis, MO), an enzyme that catalyzes the
dismutation of superoxide radicals to hydrogen peroxide (H2O2) and
molecular oxygen (7, 18, 29), was used to verify the role of O⫺
2 䡠.
PMEM were pretreated with SOD (100 U/ml) 0.25 h before its
co-incubation with TNF. We previously showed that heat-inactivated
SOD (autoclaving SOD for 0.75 h) blocked its protective effects,
which verified the antioxidant specificity of SOD (7, 18, 27).
4,5-Dihydroxy-1,3-benzenedisulfonic acid. 4,5-Dihydroxy-1,3-benzenedisulfonic acid (Tiron, Sigma), a cell-permeable superoxide trap, was
also used to verify O⫺
2 䡠-mediated alterations by TNF. We previously
showed that Tiron prevents the TNF-induced latent decrease in 䡠NO and
increase in peroxides in response to TNF (18). PMEM were pretreated
with Tiron (1 mM) 0.25 h before its co-incubation with TNF.
Treatment medium. For all studies, incubation of PMEM with TNF
and corresponding controls was performed with phenol-free DMEM
(pf-DMEM, GIBCO-BRL) supplemented with 10% FBS to avoid the
potential antioxidant effect of phenol as well as its spectrophotometric
interference.
Antisense Oligonucleotides
centrifugation to separate the silica-coated endothelial cell plasma
membranes from the whole cell homogenate. Silica-coated endothelial
cell monolayers were scraped and pooled in 1 ml of HEPES-buffered
sucrose (HBS) containing a cocktail of protease inhibitors [0.25 M
sucrose, 25 mM HEPES at pH 7.0, 10 ␮g/ml leupeptin, 10 ␮g/ml
pepstatin A, 10 ␮g/ml o-phenanthroline, 10 ␮g/ml 4-(2-aminoethyl)
benzenesulfonyl fluoride, and 50 ␮g/ml trans-epoxysuccinyl-L-leucinamido (4-guanidino) butane]. After sample homogenization, cell
homogenates were mixed with 102% (wt/vol) Nycodenz (Life Sciences) with 20 mM KCl to make a 50% final solution. The Nycodenzhomogenate solution was layered over a continuous 55–70% Nycodenz gradient containing 20 mM KCl and HBS and centrifuged in a
Beckman SW 55 rotor at 15,000 rpm for 0.5 h at 4°C. The resulting
plasma membrane pellet was resuspended in 0.5 ml of MBS.
Western analysis. Protein concentrations of the whole cell homogenate or purified luminal membrane fractions were determined by
bicinchoninic acid analysis (Pierce). Equivalent amounts of protein
from each sample were prepared and separated by SDS-PAGE (10%
gels) followed by electrotransfer to nitrocellulose filters. Immunoblotting with antibodies against p47phox (Santa Cruz) served to detect
p47phox protein present in each cell subfraction. Briefly, filters were
incubated in blocking solution (Tris-buffered saline, 0.5% Tween 20,
5% nonfat dry milk) for 1.0 h, followed by a 2.0-h incubation in
primary antibody diluted in blocking buffer. After extensive washing,
membranes were incubated for 1.0 h with anti-rabbit horseradish
peroxidase. Membranes were extensively washed, and p47phox signal
was detected with enhanced chemiluminescence substrate (ECL, Amersham). Western blots were scanned and digitized (Molecular Dynamics, Sunnyvale, CA). Densitometric quantification of immunoblots with Imagequant software (Molecular Dynamics) enabled direct
comparisons between control and TNF-stimulated cells. We normalized results by arbitrarily setting the densitometry of control cells
to 1.0.
Assay of p22phox and p47phox by Immunofluorescence Cell Staining
p22phox antisense. Expression of p22phox was inhibited with a
p22phox antisense oligonucleotide complementary to bovine p22phox
nucleotides 1,165–1,189 (25). The antisense and a corresponding
scrambled-nonsense oligonucleotide sequences were 5⬘-CCG-CGTGTC-GAG-GTC-CAT-GCA-GAC-C-3⬘ and 5⬘-GAC-GAC-GTGACT GAC-CTG-GTG-CCG-C-3⬘, respectively. Both oligonucleotides were second generation chimeras, consisting of phosphorothioated DNA (9 nucleotides) and 2⬘-O-methylated RNA (16 nucleotides)
(Oligos Etc., Wilsonville, OR), to achieve improved specificity of
binding to mRNA and nuclease resistance.
Transfection of oligonucleotides in PMEM. Transfection of oligonucleotides was accomplished with the cationic liposome carrier
Lipofectin (GIBCO-BRL) by our standard technique (2, 8). Oligonucleotides from a 40-␮M stock solution were added to serum-free
DMEM containing Lipofectin (6.6 ␮l/ml) for a final concentration of
200 nM, then added to PMEM, and incubated at 37°C for 4.0 h. The
Lipofectin/oligonucleotide-containing medium was removed before
experimental treatment.
Detection of p47phox Protein by Immunoblot
Purification of endothelial luminal plasma membranes. To determine subcellular distribution of p47phox in response to TNF treatment,
we isolated endothelial cell luminal plasma membranes as previously
described (24, 31). After the 4.0-h stimulation with TNF, we rapidly
cooled cells to 4°C by rinsing them with ice-cold MES-buffered saline
(MBS, 20 mM, pH 6.0) to stabilize enzymatic activity and diffusion
of molecules within the lipid bilayer of the plasma membrane.
Subsequently, the cells were incubated with a positively charged
colloidal silica solution for 10 min at 4°C. Subsequent cross-linking of
the silica particles by incubation with polyacrylic acid (0.1%) serves
to create a stable adherent silica pellicle that permits its purification by
AJP-Lung Cell Mol Physiol • VOL
BLMVEC (1 ⫻ 104/0.20 ml of culture medium) were plated on
18-mm coverslips inside a 35-mm culture dish, incubated at 37°C for
2 h to allow attachment, and then grown to confluence in an additional
2 ml of culture medium. The PMEM were treated as indicated,
washed with Dulbecco’s phosphate-buffered saline without ions
(DPBS, GIBCO-BRL), fixed with 3.7% formaldehyde solution at
room temperature (RT) for 20 min, and then permeabilized with 1%
Triton X-100 (Sigma) in DPBS at RT for 5 min. The cells were
washed with DPBS and then blocked in 10% normal goat serum
(NGS, GIBCO-BRL) at RT for 1 h. PMEM were incubated with either
a rabbit polyclonal anti-human p22phox peptide antibody or a rabbit
polyclonal anti-bovine p47phox antibody (provided by Mark Quinn,
Marsh Laboratory, Bozeman, MT), at a 1:1,000 dilution in 10% NGS
at RT for 1 h then washed sufficiently. The secondary antibody, Alexa
Fluor 488-labeled goat anti-rabbit IgG (Molecular Probes) was added
at a 1:1,000 dilution in 10% NGS and incubated at RT for 1 h and then
washed sufficiently. The nuclei were stained with 2.5 ␮g/ml propidium iodide (Molecular Probes) at RT for 5 min then washed 1⫻
with DPBS. The coverslips were mounted on clean glass slides with
Permafluor mounting media (Thermo Shandon, Pittsburgh, PA). The
PMEM were visualized with a Spot RT color camera (Diagnostic
Instruments, Sterling Heights, MI) mounted on an Olympus IX70
inverted microscope (Olympus America, Melville, NY) equipped for
phase, light, and fluorescence detection. Images were captured at
⫻100 magnification with an exposure time of 8 s and downloaded into
Spot RT imaging software (Diagnostic Instruments).
The quantification strategy for the fluorescent images is as follows.
PMEM were visualized and quantified with confocal microscopy with
the Leica Confocal System TCS SP2 (Leica Microsystems, Exton,
PA). There were four separate studies with six treatment groups per
study. All fields were selected by random movement of the micro286 • JANUARY 2004 •
www.ajplung.org
L39
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
scope stage to another area within an intact endothelial monolayer. Six
entire fields per treatment group were analyzed with one image per
field. We normalized all treatment groups for fluorescent intensity by
initially adjusting the settings for noise, brightness, and contrast, as
determined by the slide with the maximum fluorescence.
Detection of Oxidant Generation
The oxidation of 5 (and 6)-chloromethyl-2⬘,7⬘-dichlorodihydrofluorescein diacetate-acetyl ester (CM-H2DCFDA; Molecular
Probes, Eugene, OR) or 6-carboxy-2⬘,7⬘-dichlorodihydrofluorescein
diacetate-di(acetoxymethyl ester) (C-H2DCFDA, Molecular Probes)
was used to assess intracellular reactive oxygen species (ROS) (5, 9,
9a). Cleavage of the ester groups by intracellular esterases and
oxidation by ROS results in intracellular dichlorofluorescein (DCF)
derivatives, which are highly fluorescent. In addition, the oxidation of
dihydroethidium (DHE, Molecular Probes) to ethidium was used to
specifically assess O⫺
2 䡠. Upon oxidation, ethidium intercalates within
the cell’s DNA, becoming brightly fluorescent, and is considered a
relatively specific measurement of O⫺
2 䡠 (5, 9, 9a).
Fluorescence detection of ROS in intact PMEM. DCF fluorescence
was measured in a Wallac 1420 Victor2 multilabel counter (Wallac,
Turku, Finland) using excitation and emission wavelengths of 490 and
535 nm, respectively. BLMVEC (1 ⫻ 104/0.20 ml of culture medium)
were plated in 96-well microplates and grown to confluence. Before
treatment, PMEM were washed 2⫻ with Hanks’ buffered saline
solution (HBSS, GIBCO-BRL), loaded with 20 ␮M CM-H2DCFDA
(in HBSS 100 ␮l/well), and incubated at 37°C for 0.5 h. PMEM were
then washed 1⫻ with HBSS and 1⫻ with pf-DMEM ⫹ 10% FBS to
remove free probe. Then, fresh pf-DMEM ⫹ 10% FBS (100 ␮l/well)
was added, and a baseline fluorescence reading was taken before
treatment. Fluorescence measurements were taken at 0.0, 0.5, 1.0, 2.0,
3.0, and 4.0 h during the treatments. Results are presented as percent
change from baseline by the formula [(Ftexp ⫺ Ftbase)/Ftbase ⫻ 100],
where Ftexp ⫽ fluorescence at any given time during the experiment
in a given well and Ftbase ⫽ baseline fluorescence of the same well.
The difference in DCF fluorescence generated in the presence and
absence of SOD or Tiron was used as a further verification of
O⫺
2 䡠-dependent pathways (7, 18, 27).
Fluorescence detection of O2⫺䡠 in PMEM lysate. BLMVEC (1 ⫻
5
10 /2.0 ml of culture medium) were plated in six-well culture dishes
and, upon reaching confluence, were treated as indicated. After
treatment, PMEM were washed 1⫻ with pf-DMEM ⫹ 2% FBS and
Fig. 1. TNF induces an increase in p22phox
immunofluorescence, which is prevented by
p22phox antisense. Shown are representative
micrographs of pulmonary microvessel endothelial monolayer (PMEM) p22phox immunofluorescence (green Alexa Fluor 488).
Nuclear fluorescence is red propidium iodide, and images are ⫻100 magnification.
PMEM groups were incubated with media
alone or TNF for 4.0 h. The groups are no
oligonucleotide, anti-p22phox oligonucleotide, and scrambled-nonsense oligonucleotide (Scrambled). 3, p22phox fluorescence.
AB, antibody.
AJP-Lung Cell Mol Physiol • VOL
286 • JANUARY 2004 •
www.ajplung.org
L40
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
incubated with 10 ␮M DHE in pf-DMEM ⫹ 2% FBS (1 ml/well) at
37°C for 0.5 h. PMEM were then washed on ice 2⫻ with ice-cold
DPBS to remove free probe and scraped with 1 ml of ice-cold DPBS
into microtubes. The cells were centrifuged at 8,000 g for 10 min at
4°C, and the pellet was resuspended in 0.5 ml of ice-cold DPBS. The
cell suspensions were sonicated on ice for 15 s, and 100 ␮l/well
sonicate were added to Falcon 96-well black microplates (Becton
Dickinson, Franklin Lakes, NJ) in quadruplicate. Probe-free cell
sonicate was used for blanks. Ethidium fluorescence was measured in
a Wallac 1420 Victor2 multilabel counter (Wallac) using excitation
and emission wavelengths of 490 and 605 nm, respectively. Fluorescence was presented as percentage of control by the formula [Ftexp/
Ftcontrol], where Ftexp ⫽ fluorescence at any time after treatment in a
given lysate and Ftcontrol ⫽ fluorescence of the respective untreated
control group.
Microscopic detection of ROS-generated fluorescence in intact
PMEM. BLMVEC (1 ⫻ 105/2.0 ml of culture medium) were plated in
35-mm dishes and, upon reaching confluence, were treated as indicated. We performed detection of ROS by incubating the PMEM with
1.0 ␮M C-H2DCFDA diluted in pf-DMEM ⫹ 2% FBS at 37°C for 20
min. PMEM were then washed once in DPBS, fixed in a 3.7%
formaldehyde solution for 20 min at RT, and then washed twice with
DPBS. The PMEM fluorescence was visualized with a Spot RT color
camera mounted on an Olympus IX70 inverted microscope. Images
were captured at ⫻20 magnification with an exposure time of 8 s and
downloaded into Spot RT imaging software. Immunofluorescence was
quantified by computerized densitometry with Sigma Scan Pro (SPSS
Scientific Software, San Rafael, CA). There were four separate studies
with six treatment groups per study. All fields were selected by
random movement of the microscope stage to another area within an
intact endothelial monolayer. Six entire fields per treatment group
were analyzed with one image per field.
Assay of Endothelial Permeability
Nuclepore Track-Etch Polycarbonate Membranes (13-mm diameter, 0.8-␮m pore size; Corning Costar, Cambridge, MA) were coated
with gelatin (type B from bovine skin; Sigma) as previously described
(2, 7), mounted on modified Boyden chemotaxis chambers (9-mm
inner diameter; Adaps, Dedham, MA) with MF cement no. 1 (Millipore, Bedford, MA), and sterilized by ultraviolet light for 12.0–24.0 h.
BLMVEC (1 ⫻ 105/0.50 ml of DMEM) were plated on the gelatinized
membranes and allowed to reach confluence.
The experimental apparatus for the study of transendothelial transport in the absence of hydrostatic and oncotic pressure gradients has
been described (2, 8, 26). In brief, the system consists of two
compartments separated by a microporous polycarbonate membrane
lined with the endothelial cell monolayer as described above. The
upper, luminal compartment (700 ␮l) was suspended in the lower,
abluminal compartment (25 ml). The abluminal compartment was
stirred continuously for complete mixing. The entire system was kept
in a water bath at a constant temperature of 37°C. The fluid height in
both compartments was the same to eliminate convective flux.
Endothelial permeability was characterized by the clearance rate of
Evans blue-labeled albumin as previously described (2, 6, 8). A buffer
solution containing HBSS (GIBCO-BRL), 0.5% bovine serum albumin (Sigma), and 20 mM HEPES (Sigma) was used on both sides of
the monolayer. The luminal compartment buffer was labeled with a
final concentration of 0.057% Evans blue dye in a volume of 700 ␮l.
The absorbance of free Evans blue dye in the luminal and abluminal
compartments was always ⬍1% of the total absorbance of Evans blue
in the buffer. At the beginning of each study, a luminal compartment
sample was diluted 1:100 to determine the initial absorbance of that
compartment. Abluminal compartment samples (300 ␮l) were taken
every 5 min for 1.0 h. The absorbance of the samples was measured
in a SpectraMax Plus microplate spectrophotometer (Molecular Devices, Sunnyvale, CA) at 620 nm. The clearance rate of Evans
AJP-Lung Cell Mol Physiol • VOL
blue-labeled albumin was determined by least-squares linear regression between 10 and 60 min for the control and experimental groups.
Assay of Cell Viability
Trypan blue exclusion. BLMVEC (1 ⫻ 105/2.0 ml of culture
medium) were seeded in 35-mm dishes and grown until confluent.
After respective drug treatments, PMEM were washed with DPBS and
followed by treatment with 0.05% trypsin (0.2 ml) for 1.0 min at
37°C. The cells were resuspended in DPBS (1 ml), and an aliquot of
cell suspension (50 ␮l) was combined with 0.08% trypan blue (50 ␮l)
for 3 min. Cells were counted in 10 ␮l of the mixture with a
hemocytometer. Cell viability was defined by the following formula:
Cell viability ⫽ (cells excluding trypan blue/total cells) ⫻ 100 (10).
Statistics
A one-way analysis of variance was used to compare values among
the treatment groups. If significance was noted, a Bonferroni multiplecomparison test was used to determine significant differences between
the groups. A t-test was used when appropriate. Each PMEM well and
flask represent a single experiment. There were 5–10 samples (n) per
group, in which each sample was from a separate study, unless noted
otherwise. All data are reported as means ⫾ SE. Significance was at
P ⬍ 0.05.
RESULTS
Cell Viability of PMEM: Trypan Blue Exclusion and
Cell Counts
We previously showed that TNF, SOD, and Tiron do not
affect cell viability (2, 6, 8, 10, 27). In the present study,
treatment with p22phox oligonucleotides ⫾ TNF (50 ng/ml) did
not affect PMEM viability. The trypan blue exclusion was
similar among the groups (i.e., control: 98.5 ⫾ 0.5% vs.
p22phox antisense: 97.5 ⫾ 0.5% vs. p22phox antisense ⫹ TNF:
97.0 ⫾ 0.5% vs. scrambled: 98.5 ⫾ 0.5% vs. scrambled ⫹
TNF: 99.0 ⫾ 0.5%).
Fig. 2. TNF induces an increase in mean p22phox immunofluorescence, which
is prevented by p22phox antisense. The mean pixel intensity of the p22phox
fluorescence of the micrographs for each treatment in Fig. 1 is shown. PMEM
groups were incubated with media alone or TNF for 4.0 h. The groups are no
oligonucleotide (Con), anti-p22phox oligonucleotide, and scrambled-nonsense
oligonucleotide. *p22phox fluorescence different from respective group without
TNF; #p22phox fluorescence different from TNF.
286 • JANUARY 2004 •
www.ajplung.org
L41
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
TNF Induces an Increase in p22phox That Is Prevented by
p22phox Antisense
In the representative study of Fig. 1, PMEM were incubated
with TNF for 4.0 h both with and without pretreatment with
either antisense oligonucleotides to p22phox or scrambled oligonucleotides and then assayed for p22phox fluorescence. We
measured the p22phox protein to show that the antisense to
p22phox can prevent the upregulation of p22phox protein during
the response to TNF. In PMEM treated with TNF alone or
scrambled oligonucleotides ⫹ TNF, there was an increase in
p22phox fluorescence (green) compared with their respective
controls; note the peripheral localization of p22phox fluorescence in these groups. In PMEM treated with anti-p22phox ⫹
TNF, there was no increase in p22phox fluorescence compared
with anti-p22phox controls, nor was there notable peripheral
localization of p22phox fluorescence in these groups. The
nuclear morphology (red) was similar among all the groups.
There was no significant fluorescence in the TNF group
stained with only secondary antibody and without propidium iodide. In Fig. 2, quantification of the digitized
images indicates that TNF induced an increase in total
cellular p22phox fluorescence that was prevented by antip22phox, whereas the scrambled oligonucleotide had no
effect. The data from Figs. 1 and 2 show that TNF induces
an increase in p22phox protein. The increase in p22phox
protein and the peripheral localization of the p22phox protein
were specifically prevented by anti-p22phox.
TNF Induces Activation of NAD(P)H Oxidase
PMEM were incubated with or without TNF for 4.0 h (Fig.
3). Total cell lysate and membrane fractions were assayed for
p47phox protein by Western immunoblot. The intracellular
Fig. 3. TNF induces an increase in and alters
localization of p47phox. PMEM groups were
incubated with media alone or TNF for 4.0 h.
A: representative (n ⫽ 2) p47phox Western
blot of total lysate and membrane fractions in
the control and TNF-treated PMEM. B: representative micrograph of PMEM p47phox
immunofluorescence (green Alexa Fluor
488) . The treatment groups are no oligonucleotide, anti-p22phox oligonucleotide, and
scrambled-nonsense oligonucleotide. Nuclear fluorescence is red propidium iodide,
and images are shown at ⫻100 magnification. 3, p47phoxfluorescence; *different
from control group.
AJP-Lung Cell Mol Physiol • VOL
286 • JANUARY 2004 •
www.ajplung.org
L42
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
Fig. 4. TNF induces an increase in mean p47phox immunofluorescence, which
is prevented by p22phox antisense. The mean pixel intensity of the p47phox
fluorescence of the micrographs for each treatment in Fig. 3 is shown. PMEM
groups were incubated with media alone or TNF for 4.0 h. The groups are no
oligonucleotide, anti-p22phox oligonucleotide, and scrambled-nonsense oligonucleotide. *p47phox fluorescence different from respective group without
TNF; #p47phox fluorescence different from TNF.
distribution of p47phox protein was measured because increased
localization of p47phox in the peripheral membrane is a signature
event for NAD(P)H oxidase activation. The representative (n ⫽ 2)
Western blot shows TNF increased the membrane p47phox/total
p47phox ratios compared with the control group. We assayed the
immunofluorescence of p47phox protein to further explore the
NAD(P)H oxidase activation in response to TNF. In the representative micrographs of Fig. 3B, the control group shows faint
p47phox fluorescence (green) diffusely distributed within the cell.
In the TNF and scrambled ⫹ TNF groups, there are increases in
intensity of p47phox fluorescence compared with the respective
control and scrambled groups. In the anti-p22phox ⫹ TNF group,
there is no increase in p47phox fluorescence compared with the
anti-p22phox group. The nucleus (red) does not exhibit any significant change in morphology among the groups. PMEM treated
with only secondary antibody reveals no significant fluorescence
in any group. In Fig. 4, quantification of the digitized images
indicates that TNF induced an increase in total cellular p47phox
fluorescence that was prevented by anti-p22phox, whereas the
scrambled oligonucleotide had no effect.
The data of Figs. 3 and 4 demonstrate that TNF treatment
causes increased localization of p47phox to the peripheral membrane and indicates, therefore, NAD(P)H oxidase activation in
response to TNF. In addition, digital analysis of the images
indicates that TNF induced an increase in p47phox protein that
was specifically prevented by anti-p22phox.
p22phox Antisense Prevents the TNF-Induced Increase in
Colocalization of p47phox with p22phox
Figures 5 and 6 show the apical-to-basal p22phox and p47phox
fluorescence along the z-axis to further demonstrate the effect
of TNF on the distribution of p22phox and p47phox protein. In
Fig. 5A, the micrograph indicates there was greater p22phox
fluorescence in the TNF group compared with the control
group. Moreover, in the control group, the p22phox fluorescence
was primarily between the cells, whereas the p22phox fluorescence was more diffuse in the TNF group. In Fig. 5B, the
micrograph indicates there was greater p47phox fluorescence in
the TNF group compared with the control group. Moreover, in
the TNF group, there was greater p47phox fluorescence between
the cells compared with the diffuse fluorescence in the control
group. In Fig. 5C, the increase in p22phox and p47phox fluorescence was maximally localized to similar z-fractions in the
TNF group, indicating increased localization of p47phox with
p22phox in response to TNF.
Figure 6, A and B, shows the comparisons of z-axis distribution for p47phox and p22phox fluorescence to assess the effect
of the anti-p22phox on the response to TNF. Figure 6A indicates
that in the anti-p22phox ⫹ TNF group, there was no change in
localization of p47phox with p22phox compared with the antip22phox group. Figure 6B indicates that in the scrambled ⫹
TNF group, the localization of p47phox with p22phox increased
compared with the scrambled group, a result similar to the TNF
group. The results of Figs. 5 and 6 indicate that TNF induces
an increase in localization of p47phox with p22phox in PMEM.
Moreover, the data indicate that p22phox antisense specifically
prevents the TNF-induced increase in localization of p47phox
with p22phox.
p22phox Mediates Generation of ROS in Response to TNF
In Fig. 7, PMEM grown in 96-well microplates were assessed for the inhibited DCF fluorescence generated from
CM-H2DCFDA during a 4-h incubation with TNF to indicate
the temporal generation of ROS. In separate studies, PMEM
were pretreated with SOD before TNF incubation to verify that
the DCF fluorescence was dependent on the generation of O⫺
2 䡠.
In the SOD-inhibited control group, there was no significant
change in fluorescence during the 4.0-h study period. In the
SOD-inhibited TNF group, there was a time-dependent increase in DCF fluorescence, which was significantly greater at
4.0 h than both its baseline value and the corresponding 4.0-h
value in the SOD-inhibited control group. PMEM were also
pretreated with the cell-permeable antioxidant Tiron for 0.25 h
before TNF incubation to verify that the DCF fluorescence was
indeed sensitive to changes in O⫺
2 䡠. In the Tiron-inhibited
control group, there was an increase in DCF fluorescence,
which became significant at 4.0 h compared with baseline. In
the Tiron-inhibited TNF group, there was an increase in DCF
fluorescence that became significant at 2.0 and 4.0 h compared
with baseline and the corresponding 4.0-h value in the Tironinhibited control and SOD-inhibited TNF groups. Thus Fig. 7
indicates that TNF causes the generation of ROS, which is
inhibited by the anti-O⫺
2 䡠 agents SOD and Tiron.
Figure 8A shows a representative micrograph, and Fig. 8B
shows the mean pixels of the digitized images using DCF on
Fig. 5. TNF increases localization of p47phox with p22phox assessed by z-distribution of p22phox and p47phox immunofluorescence.
PMEM were incubated with media alone or TNF for 4.0 h. A: representative x/z micrographs of PMEM p22phox immunofluorescence. B: representative x/z micrographs of PMEM p47phox immunofluorescence. The p22phox and p47phox fluorescence is green
(Alexa Fluor 488), nuclear fluorescence is red (propidium iodide), and image magnification is ⫻100. C: p47phox and p22phox mean
immunofluorescence per section of the PMEM z-axis sections. 3, p22phox or p47phox fluorescence; *fluorescence different from
p47phox; #fluorescence different from respective group without TNF.
AJP-Lung Cell Mol Physiol • VOL
286 • JANUARY 2004 •
www.ajplung.org
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
AJP-Lung Cell Mol Physiol • VOL
286 • JANUARY 2004 •
www.ajplung.org
L43
L44
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
Fig. 6. Anti-p22phox prevents the TNF-induced increase in localization of p47phox with
p22phox. PMEM groups were incubated with
media alone or TNF for 4.0 h. A: p47phox and
p22phox mean immunofluorescence per z-axis
section of the x/z micrographs of PMEM in
the p22phox antisense oligonucleotide group.
B: p47phox and p22phox mean immunofluorescence per z-axis section of the x/z micrographs of PMEM in the scrambled-nonsense
oligonucleotide group. *fluorescence different from p47phox; #fluorescence different
from respective group without TNF.
Fig. 7. TNF causes the generation of reactive oxygen species (ROS) detected
with 5 (and 6)-chloromethyl-2⬘,7⬘-dichlorodihydrofluorescein diacetate-acetyl
ester (CM-H2DCFDA). Dichlorofluorescein (DCF) fluorescence generated
from CM-H2DCFDA was assessed in 96-well plates. Shown is the mean
superoxide dismutase (SOD)-inhibited and 4,5-dihydroxy-1,3-benzenedisulfonic acid (Tiron)-inhibited fluorescence of DCF in wells treated with TNF and
measured repeatedly over 4.0 h. RLU, relative light units; Exp, experimental;
bas, baseline; *fluorescence different from 0.0 h; #fluorescence different from
respective group without TNF.
AJP-Lung Cell Mol Physiol • VOL
fixed PMEM, to verify that the SOD-inhibited fluorescence
was not specific to any variant of the DCF assay condition. In
the TNF group, there was an increase in fluorescence at 4.0 h
compared with the control group. In the SOD ⫹ TNF group,
the fluorescence was the same as the SOD group and less than
the TNF group. In addition, some PMEM were pretreated with
antisense to p22phox or the scrambled oligonucleotide before
the TNF incubation to show that the NAD(P)H oxidase system
mediates the increase in fluorescence in response to the TNF.
In the anti-p22phox ⫹ TNF group, the fluorescence was the
same as the anti-p22phox group and less than the TNF group.
The scrambled oligonucleotide had no effect on the TNFinduced increase in fluorescence. The data from Fig. 8 verify
that the TNF-induced increase in ROS 1) is dependent on O⫺
2 䡠,
2) is consistent between different assays using DCF, and 3) is
mediated by NAD(P)H oxidase.
Figure 9 shows PMEM incubated with vehicle or TNF (50
ng/ml) for 4.0 h and assayed with the specific O⫺
2 䡠 fluorescent
probe DHE to verify O⫺
䡠-dependent
generation
of ROS. In
2
separate studies, PMEM were pretreated with SOD for 0.25 h
before the TNF incubation to further verify that the DHE
fluorescence was due to the generation of O⫺
2 䡠. Also, some
PMEM were pretreated with antisense to p22phox or the scrambled oligonucleotide before the TNF incubation to show that
the NAD(P)H oxidase system mediates the increase in fluorescence in response to the TNF. In the TNF group, there was an
increase in DHE fluorescence compared with the control
group. In the SOD ⫹ TNF group, the DHE fluorescence was
286 • JANUARY 2004 •
www.ajplung.org
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
L45
Fig. 8. p22phox antisense prevents the generation of
ROS detected with 6-carboxy-2⬘,7⬘-dichlorodihydrofluorescein diacetate-di(acetoxymethyl ester) (CH2DCFDA). DCF fluorescence generated from
C-H2DCFDA was assessed in fixed cells. A: representative images of the DCF fluorescence of cells
fixed following a 4.0-h incubation with or without
TNF or SOD. B: mean fluorescence of DCF of
PMEM groups fixed after incubation with media
alone or TNF for 4.0 h. The treatment groups are no
oligonucleotide, SOD, anti-p22phox oligonucleotide,
and scrambled-nonsense oligonucleotide. *Fluorescence different from respective group without TNF;
#fluorescence different from the TNF group.
similar to the SOD group and less than in the TNF group. In the
anti-p22phox ⫹ TNF group, the fluorescence was similar to the
anti-p22phox group and less than in the TNF group. The
scrambled oligonucleotide had no effect on the TNF-induced
increase in fluorescence. The results shown in Fig. 9 confirm
that TNF causes the p22phox-dependent generation of O⫺
2 䡠.
TNF-Induced Increase in Permeability Is Mediated by
NAD(P)H Oxidase and O⫺
2䡠
As shown in Fig. 10, the PMEM were incubated with vehicle
or TNF (50 ng/ml) for 4.0 h. Some PMEM were pretreated
with antisense to p22phox or the scrambled oligonucleotides for
4.0 h before TNF incubation, with the purpose of showing that
AJP-Lung Cell Mol Physiol • VOL
the NAD(P)H oxidase system mediates the barrier dysfunction
in response to the TNF. The permeability of Evans blue
albumin was assayed and is shown in Fig. 10. Albumin flux
was increased in the TNF and scrambled ⫹ TNF groups
compared with the respective control and scrambled groups. In
the anti-p22phox ⫹ TNF group, there was no increase in
albumin flux compared with the control and anti-p22phox
group. In separate studies, PMEM were incubated with Cu-Zn
SOD (100 U/ml) throughout the TNF incubation to show that
the increased permeability is due to the generation of O⫺
2 䡠 in
response to TNF. The albumin clearance rate increased in the
TNF group; however, there was no increase in the albumin
clearance rate in the Cu-Zn SOD-treated group. The data
286 • JANUARY 2004 •
www.ajplung.org
L46
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
Fig. 9. p22phox antisense prevents the generation of O⫺
2 䡠. The figure shows the
mean ethidium fluorescence generated from dihydroethidium in lysate of
PMEM treated with media alone or TNF for 4.0 h. The treatment groups are
no oligonucleotide, SOD, anti-p22phox oligonucleotide, and scrambled-nonsense oligonucleotide. *Different from respective group without TNF; #different from TNF group.
indicate that the activation of NAD(P)H oxidase, via O⫺
2 䡠,
mediates the pulmonary endothelial barrier dysfunction induced by TNF.
DISCUSSION
The present study indicates that in pulmonary microvessel
endothelium NAD(P)H oxidase mediates the increase in O⫺
2䡠
and albumin permeability in response to TNF. NAD(P)H
oxidase mediates the response to TNF because: 1) TNF caused
increased membrane localization of p47phox, 2) TNF caused the
increased localization of p47phox with p22phox, 3) TNF induced
the release of O⫺
2 䡠, 4) TNF caused barrier dysfunction, and 5)
both SOD and antisense to p22phox prevented the effects of
TNF. The data indicate that the cellular effects of the antisense
for p22phox were specific for p22phox protein because the
scrambled oligonucleotide and Lipofectin had no effect on the
response to TNF. Finally, there was similar cell viability
among all the groups. We previously showed that TNF induces
a O⫺
2 䡠-dependent 1) decrease in endothelium-dependent dilation (7), 2) increase in oxidized glutathione (27), 3) latent
decrease in 䡠NO (18), and 4) generation of ONOO⫺ (6, 17, 27)
because the anti-O⫺
2 䡠 agents Cu-Zn SOD, Tiron, and/or N-acetylcysteine prevented these effects of TNF, and the inactive
SOD had no protective effect. Our data confirm the results of
others that demonstrate that O⫺
2 䡠 mediates the increase in lung
endothelial permeability in response to inflammatory mediators such as endotoxin (9a). However, the present study is the
first to demonstrate that NAD(P)H oxidase is essential for
TNF-induced pulmonary microvessel endothelial barrier dysfunction in vitro.
The movement of p40phox, 47phox, p67phox, and Rac from
cytosol to membrane and their formation with membranebound p22phox and gp91phox are generally considered the
events required for activation of the NAD(P)H oxidase (21–23,
32). The translocation of p47phox into the cell membrane
compartment is considered a signature for the activation of
NAD(P)H oxidase (21–23, 32). The present study indicates
that TNF induced the activation of the NAD(P)H oxidase
because the Western blot analysis showed translocation of
p47phox to the peripheral luminal membrane. Moreover, the
confocal microscopy showed increased localization of p47phox
with p22phox in response to TNF. Finally, the immunofluorescent images of p22phox and p47phox indicate that the membraneassociated p22phox is more apparent than the membrane-associated p47phox. The compartmental regulation of p47phox and
p22phox may be entirely different because p47phox normally resides in both cytoplasm and membrane, whereas p22phox primarily
resides in membrane (21–23, 32). Thus an overall increase in
p47phox will be manifested as increases in both cytoplasm and
membrane as opposed to the overall increase in p22phox, which
will be manifested as an increase mostly in the membrane. The
Fig. 10. The inhibition of NAD(P)H oxidase prevents barrier dysfunction induced by TNF. PMEM
were co-incubated with media alone, SOD, and TNF
for 4.0 h. In separate studies, PMEM were preincubated for 4.0 h with media alone, p22phox antisense
oligonucleotide, or scrambled-nonsense oligonucleotide followed by incubation with media alone or
TNF for 4.0 h. PMEM were assayed for endothelial
albumin clearance rate. *Different from control
group; #different from TNF group.
AJP-Lung Cell Mol Physiol • VOL
286 • JANUARY 2004 •
www.ajplung.org
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
antisense to p22phox prevented the 1) peripheral localization of
p22phox, 2) TNF-induced increase in p22phox, and 3) increased
localization of p47phox with p22phox. The prevention of the increased localization of p47phox with p22phox by antisense to
p22phox is consistent with the theory that membrane p22phox is
essential for the integration of p47phox into the NAD(P)H oxidase
complex (21). Our data agree with other studies indicating that
TNF activates the NAD(P)H oxidase pathway in pulmonary
endothelium (5, 9) and that NAD(P)H oxidase activity can be
inhibited by antisense to p22phox (3, 25).
⫺
TNF caused generation of O⫺
2 䡠 because the anti-O2 䡠 agents
SOD and/or Tiron inhibited the TNF-induced increase in fluorescence of CM-H2DCFDA, C-H2DCFDA, and DHE. Moreover, the effect of the TNF was independent of our O⫺
2 䡠 assay
conditions because the TNF-induced increase in O⫺
2 䡠 was
detected with intact PMEM, fixed PMEM, and PMEM lysate.
Finally, DHE is a more specific indicator of O⫺
2 䡠 than CMH2DCFDA and C-H2DCFDA (25). NAD(P)H oxidase mediated the TNF-induced increase in O⫺
2 䡠 because antisense to
p22phox prevented the TNF-induced increase in fluorescence of
C-H2DCFDA and DHE, and the scrambled oligonucleotide
had no effect. Also, SOD prevented the TNF-induced increase
in fluorescence of C-H2DCFDA and DHE, which is consistent
⫺
with the notion that SOD is a “O⫺
2 䡠 sink” that binds the O2 䡠
generated by the membrane NAD(P)H oxidase because SOD is
not cell permeable (27). Yet, our data do not rule out the fact
that other pathways such as xanthine oxidase (33), eNOS (29),
and mitochondrial enzymes (13) can mediate generation of
ROS (e.g., O⫺
2 䡠, H2O2, 䡠OH) in response to TNF because Tiron,
which is cell permeable, had a greater inhibitory effect than
SOD on CM-H2DCFDA fluorescence, and CM-H2DCFDA
may not be completely specific for 䡠O2 (27, 29).
The treatment with antisense to p22phox did not decrease
constitutive p22phox protein, but it did prevent the TNF-induced increase in p22phox protein. We used a dose and time of
antisense treatment that would only prevent the TNF-induced
increase in p22phox protein to avoid parenthetic adaptive effects
that may occur in response to decreases in an essential constitutive protein. Moreover, we wanted to study the role of the
NAD(P)H oxidase only during the response to TNF. We used
this approach in our previous study using anti-PKC-␣ to
deplete PKC-␣ (8). We previously showed that TNF caused a
prolonged increase in PKC-␣ protein (8), which supported the
endothelial barrier dysfunction in response to TNF, a result
similar to our present conclusion about p22phox. De Keulenaer
et al. (3) showed in smooth muscle that TNF causes an increase
in p22phox mRNA. Interestingly, treatment with antisense to
p22phox also prevented the TNF-induced increase in p47phox
protein. The mechanism for the increase in p22phox and p47phox
protein is not a focus of the present study; however, it has been
shown that ROS mediate the activation of transcription factors
(e.g., AP-1) and gene expression in response to TNF (5, 10,
30). In addition, a preliminary study (n ⫽ 2, data not shown)
indicates that treatment of PMEM with SOD prevents the
TNF-induced increase in p47phox protein, implicating O⫺
2 䡠 in
the regulation of p47phox protein. Therefore, antisense to
p22phox may have prevented the increase in p47phox protein by
preventing the generation of O⫺
2 䡠 and the resultant downstream
effect of ROS on gene and protein expression (5, 10, 30).
The mechanism for the activation of NAD(P)H oxidase in
response to TNF was not investigated in the present study. We
AJP-Lung Cell Mol Physiol • VOL
L47
previously indicated that both PKC and/or PKC-␣ activation
occurs in response to TNF and mediates the 1) latent O⫺
2 䡠mediated decrease in 䡠NO (18), 2) increase in ONOO⫺ (27),
and 3) increase in glutathione oxidation (27). Frey et al. (9)
showed that PKC-␨ regulates TNF-induced activation of
NAD(P)H oxidase in endothelial cells. In response to TNF, the
PKC-mediated phosphorylation of the NAD(P)H oxidase components such as p47phox can induce the translocation of p47phox
to the peripheral membrane and the formation of the NAD(P)H
oxidase enzyme complex (3, 9). Thus probable downstream
targets for PKC activation are pathways leading to generation
of ROS via NAD(P)H oxidase. Finally, other mechanisms for
NAD(P)H activation in response to TNF can include activation
of the mitogen-activated protein kinase pathways (11, 25).
The present results show that TNF caused an increase in
protein permeability, which is consistent with our previous
studies (2, 6, 8). The compartmentalization of O⫺
2 䡠, via
NAD(P)H oxidase, within the cell peripheral membrane is a
primary event responsible for the barrier dysfunction in response to TNF, because p22phox antisense and SOD completely
blocked the increase in permeability. In separate ongoing
studies of ours (n ⫽ 3), antisense to p47phox also prevented the
TNF-induced increase in permeability (control: 0.274 ⫾ 0.021
⬵ anti-p47phox ⫹ TNF 0.320 ⫾ 0.013 ␮l/min; control: 0.274 ⫾
0.021 ⬍⬍ TNF: 0.392 ⫾ 0.018, scrambled p47phox ⫹ TNF:
0.379 ⫾ 0.031 ␮l/min; P ⬍ 0.05). Our previous work indicates
that TNF-induced barrier dysfunction is mediated by PKC-␣
(8), eNOS (2), and 䡠NO (6); in addition, our present study finds
that NAD(P)H oxidase and O⫺
2 䡠 mediate barrier dysfunction.
Moreover, our previous studies (6, 27) and the literature (29)
indicate a role for ONOO⫺ in pulmonary endothelial injury,
which is a paradigm that integrates the mediators 䡠NO and O⫺
2 䡠.
䡠NO and O⫺
䡠
can
react
with
each
other
at
a
diffusion-limited
2
rate to form the potent reactive oxygen/nitrogen molecule
ONOO⫺ (29). ONOO⫺ is known to cause endothelial injury
and to increase endothelial permeability (29). The role of
ONOO⫺ in TNF-induced endothelial injury is now under
intense investigation in our laboratory.
It is known that reactive nitrogen (e.g., 䡠NO, ONOO⫺) and
oxygen (O⫺
2 䡠, H2O2, and 䡠OH) species mediate alterations in the
cytoskeleton and extracellular matrix that may ultimately have
an impact on the barrier dysfunction in response to TNF. There
are pathways that link reactive nitrogen and oxygen species
with the cytoskeleton and permeability, such as direct oxidation and nitration of cytoskeletal protein (e.g., actin, tubulin; 1,
3, 4, 6), modulation of kinase pathways (e.g.., PKC, ERK, Src,
and Rho; 11, 19, 25, 36), reorganization of cytoskeletal proteins (e.g., actin; 12, 28), and increases in protease activity
(e.g., metalloproteases; 26). The interpretation of the z-axis
images of p22phox and p47phox integrated with the increase in
p47phox in the luminal membrane suggests that the TNFinduced increase in localization of p47phox with p22phox occurs,
at least in part, at the apical membrane within the boundaries of
the cell-cell interface. The cell-cell interface is an area in which
cell-cell adhesion occurs. Actin, in association with zonula
occludens protein ZO-1, claudins, occludins, and zonula adherence proteins catenins and cadherins, is a primary component of the mechanism for cell-cell adhesion (11, 19, 25, 36).
Therefore, the results support the notion that a potential mechanism for the O⫺
2 䡠-induced increase in permeability is the
generation of ONOO⫺ followed by alteration of cytoskeletal
286 • JANUARY 2004 •
www.ajplung.org
L48
TNF-INDUCED PERMEABILITY AND NAD(P)H OXIDASE
targets such as nitration of actin, resulting in alterations in cell
adhesion and cell shape and increases in paracellular permeability (6, 12, 27).
In summary, our study indicates for the first time that TNF
induces increased pulmonary microvascular endothelial permeability, which is dependent on NAD(P)H oxidase-mediated
generation of O⫺
2 䡠. The development of strategies that target
ongoing NAD(P)H oxidase activity may provide therapy for
TNF-mediated acute lung injury.
GRANTS
This work was supported by National Heart, Lung, and Blood Institute
Grants RO1-HL-59901-02 (A. Johnson) and RO1-HL-66301-01 (V. Rizzo).
REFERENCES
1. Aslan M, Ryan TM, Townes TM, Coward L, Kirk MC, Barnes S,
Alexander CB, Rosenfeld SS, and Freeman BA. Nitric oxide-dependent
generation of reactive species in sickle cell disease. Actin tyrosine nitration induces defective cytoskeletal polymerization. J Biol Chem 278:
4194–4204, 2003.
2. Bove K, Neumann P, Gertzberg N, and Johnson A. Role of ecNOS
derived NO in mediating TNF-induced endothelial barrier dysfunction.
Am J Physiol Lung Cell Mol Physiol 280: L914–L922, 2001.
3. De Keulenaer GW, Alexander RW, Ushio-Fukai M, Ishizaka N, and
Griendling KK. Tumour necrosis factor alpha activates a p22phox-based
NADH oxidase in vascular smooth muscle. Biochem J 329: 653–657,
1998.
4. Eiserich JP, Estevez AG, Bamberg TV, Ye YZ, Chumley PH, Beckman JS, and Freeman BA. Microtubule dysfunction by posttranslational
nitrotyrosination of A-tubulin: a nitric oxide-dependent mechanism of
cellular injury. Proc Natl Acad Sci USA 96: 6365–6370, 1999.
5. Fan J, Frey RS, Rahman A, and Malik AB. Role of neutrophil
NAD(P)H oxidase in the mechanism of tumor necrosis factor-alphainduced NF-kappa B activation and intercellular adhesion molecule-1
expression in endothelial cells. J Biol Chem 277: 3404–3411, 2002.
6. Ferro TJ, Gertzberg N, Selden L, Neumann P, and Johnson A.
Endothelial barrier dysfunction and p42 oxidation induced by TNF-␣ are
mediated by nitric oxide. Am J Physiol Lung Cell Mol Physiol 272:
L979–L988, 1997.
7. Ferro TJ, Hocking DC, and Johnson A. Tumor necrosis factor-␣ alters
pulmonary vasoreactivity via neutrophil-derived oxidants. Am J Physiol
Lung Cell Mol Physiol 265: L462–L471, 1993.
8. Ferro T, Neumann P, Gertzberg N, Clements R, and Johnson A.
Protein kinase C-␣ mediates endothelial barrier dysfunction induced by
TNF-␣. Am J Physiol Lung Cell Mol Physiol 278: L1107–L1117, 2000.
9. Frey RS, Rahman A, Kefer JC, Minshall RD, and Malik AB. PKC␨
regulates TNF-induced activation of NAD(P)H oxidase in endothelial
cells. Circ Res 90: 1012–1019, 2002.
9a.Gao X-P, Standiford TJ, Rahman A, Newstead M, Holland SM,
Dinauer MC, Liu QH, and Malik AB. Role of NAD(P)H oxidase in the
mechanism of lung neutrophil sequestration and microvessel injury induced by Gram-negative sepsis: studies in p47phox⫺/⫺ and
gp91phox⫺/⫺ mice. J Immunol 168: 3974–3982, 2002.
10. Gertzberg N, Clements R, Jaspers I, Ferro TJ, Neumann P, Flescher
E, and Johnson A. Tumor necrosis factor-␣ induced activating protein-1
activity is modulated by nitric oxide-mediated protein kinase G activation.
Am J Respir Cell Mol Biol 22: 105–115, 2000.
11. Goebeler M, Kilian K, Gillitzer R, Kunz M, Yoshimura T, Brocker
EB, Rapp UR, and Ludwig S. The MKK6/p38 stress kinase cascade is
critical for tumor necrosis factor-alpha induced expression of monocytechemoattractant protein-1 in endothelial cells. Blood 93: 857–865, 1999.
12. Goldblum SE, Ding X, and Campbell-Washington J. TNF-␣ induces
endothelial cell F-actin depolymerization, new actin synthesis, and barrier
dysfunction. Am J Physiol Cell Physiol 264: C894–C905, 1993.
13. Hennet T, Richter C, and Peterhans E. Tumor necrosis factor-alpha
induces superoxide anion generation in mitochondria of L929 cells.
Biochem J 289: 587–592, 1993.
14. Hocking DC, Phillips PG, Ferro TJ, and Johnson A. Mechanisms of
pulmonary edema induced by tumor necrosis factor-␣. Circ Res 67:
68–77, 1990.
AJP-Lung Cell Mol Physiol • VOL
15. Horvath CJ, Ferro TJ, Jesmok G, and Malik AB. Recombinant tumor
necrosis factor increases pulmonary vascular permeability independent of
neutrophils. Proc Natl Acad Sci USA 85: 9219–9223, 1988.
16. Johnson A and Ferro TJ. Tumor necrosis factor-␣ augments pulmonary
vasoconstriction via inhibition of nitrovasodilator activity. J Appl Physiol
73: 2483–2492, 1992.
17. Johnson A and Ferro TJ. Nitrovasodilator repletion increases TNF-␣induced pulmonary edema. J Appl Physiol 80: 2151–2155, 1996.
18. Johnson A, Phelps DT, and Ferro TJ. Tumor necrosis factor-␣ decreases pulmonary artery endothelial nitrovasodilator via protein kinase C.
Am J Physiol Lung Cell Mol Physiol 267: L318–L325, 1994.
19. Johnson A, Phillips P, Hocking D, Tsan MF, and Ferro TJ. A protein
kinase inhibitor prevents pulmonary edema in response to H2O2. Am J
Physiol Heart Circ Physiol 256: H1012–H1022, 1989.
20. Johnson J, Meyrick B, Jesmok G, and Brigham KL. Human recombinant tumor necrosis factor-␣ infusion mimics endotoxemia in sheep.
J Appl Physiol 66: 1448–1454, 1989.
21. Leusen JHW, Verhoeven AJ, and Roos D. Interactions between the
components of the human NAD(P)H oxidase: a review about the intrigues
in the phox family. Front Biosci 1: 72–90, 1996.
22. Li JM, Mullen AM, Yun S, Wientjes F, Brouns GY, Thrasher AJ, and
Shah AM. Essential role of the NAD(P)H oxidase subunit p47(phox) in
endothelial cell superoxide production in response to phorbol ester and
tumor necrosis factor-alpha. Circ Res 90: 143–50, 2002.
23. Murphy HS, Yu C, and Quddus J. Functional expression of NAD(P)H
oxidase p47 in lung microvascular endothelial cells. Biochem Biophys Res
Commun 278: 584–589, 2000.
24. Oh P and Schnitzer JE. Immunoisolation of caveolae with high affinity
antibody binding to the oligomeric caveolin cage. Toward understanding
the basis of purification. J Biol Chem 274: 23144–23154, 1999.
25. Parinandi NL, Kleinberg MA, Usatyuk PV, Cummings RJ, Pennathur
A, Cardounel AJ, Zweier JL, Garcia JGN, and Natarajan V. Hyperoxia-induced NAD(P)H oxidase activation and regulation by MAP kinases
in human lung endothelial cells. Am J Physiol Lung Cell Mol Physiol 284:
L26–L38, 2003.
26. Partridge CA, Jeffrey JJ, and Malik AB. A 96-kDa gelatinase induced
by TNF-␣ contributes to increased microvascular endothelial permeability. Am J Physiol Lung Cell Mol Physiol 265: L438–L447, 1993.
27. Phelps DT, Ferro TJ, Higgins PJ, Shankar R, Parker DM, and
Johnson A. TNF-␣ induces peroxynitrite-mediated depletion of lung
endothelial glutathione via protein kinase C. Am J Physiol Lung Cell Mol
Physiol 269: L551–L559, 1995.
28. Phillips PG, Lum H, Malik AB, and Tsan MF. Phallacidin prevents
thrombin-induced increases in endothelial permeability to albumin. Am J
Physiol Cell Physiol 257: C562–C567, 1989.
29. Pryor WA and Squadrito GL. The chemistry of peroxynitrite: a product
from the reaction of nitric oxide with superoxide. Am J Physiol Lung Cell
Mol Physiol 268: L699–L722, 1995.
30. Rahman A, Anwar KN, and Malik AB. Protein kinase C-␨ mediates
TNF-␣-induced ICAM-1 gene transcription in endothelial cells. Am J
Physiol Cell Physiol 279: C906–C914, 2000.
31. Schnitzer JE, Oh P, Jacobson BS, and Dvorak AM. Caveolae from
luminal plasmalemma of rat lung endothelium: microdomains enriched in
caveolin, Ca2⫹-ATPase, and inositol trisphosphate receptor. Proc Natl
Acad Sci USA 92: 1759–1763, 1995.
32. Souza HP, Souza LC, Anastacio VM, Pereira AC, Junquera ML,
Krieger JE, da Luz PL, Augusto O, and Laurindo FR. Vascular
oxidant stress early after balloon injury: evidence for increased NAD(P)H
oxireductase activity. Free Radic Biol Med 28: 1232–1242, 2000.
33. Terada LS, Willingham IR, Guidot DM, Shibao GN, Kindt GW, and
Repine JE. Tungsten treatment prevents tumor necrosis factor-induced
injury of brain endothelial cells. Inflammation 16: 13–19, 1992.
34. Tracey KJ, Lowry SF, and Cerami A. Cachectin/TNF␣ in septic shock
and septic adult respiratory distress syndrome. Am Rev Respir Dis 138:
1377–1379, 1988.
35. Vilcek J and Lee TH. Tumor necrosis factor: new insights into the
molecular mechanisms of its multiple actions. J Biol Chem 266: 7313–
7316, 1991.
36. Wojciak-Stothard B, Entwistle A, Garg R, and Ridley AJ. Regulation
of TNF-alpha-induced reorganization of the actin cytoskeleton and cellcell junctions by Rho, Rac and Cdc42 in human endothelial cells. J Cell
Physiol 17: 150–165, 1998.
286 • JANUARY 2004 •
www.ajplung.org