Cultured lung fibroblasts isolated from patients
with idiopathic pulmonary fibrosis have a
diminished capacity to synthesize
prostaglandin E2 and to express
cyclooxygenase-2.
J Wilborn, … , R M Strieter, M Peters-Golden
J Clin Invest.
1995;
95(4)
:1861-1868.
https://doi.org/10.1172/JCI117866
.
Prostaglandin E2 (PGE2) inhibits fibroblast proliferation and collagen synthesis. In this
study, we compared lung fibroblasts isolated from patients with idiopathic pulmonary
fibrosis (F-IPF) and from patients undergoing resectional surgery for lung cancer (F-nl) with
respect to their capacity for PGE2 synthesis and their expression and regulation of
cyclooxygenase (COX) proteins. Basal COX activity, assessed by quantitating
immunoreactive PGE2 synthesized from arachidonic acid, was twofold less (P < 0.05) in
F-IPF than F-nl. In F-nl, incubation with the agonists PMA, LPS, or IL-1 increased COX activity
and protein expression of the inducible form of COX, COX-2, and these responses were
inhibited by coincubation with dexamethasone. By contrast, F-IPF failed to demonstrate
increases in COX-2 protein expression or COX activity in response to these agonists. Under
conditions of maximal induction, COX activity in F-IPF was sixfold less than that in F-nl (P <
0.05). Our data indicate that F-IPF have a striking defect in their capacity to synthesize the
antiinflammatory and antifibrogenic molecule PGE2, apparently because of a diminished
induction of COX-2 protein. This reduction in the endogenous capacity of F-IPF to
down-regulate their function via PGE2 may contribute to the inflammatory and fibrogenic response
in IPF. Moreover, we believe that this represents the first description of a defect in COX-2
expression in association with a human disease.
Research Article
Cultured Lung
Fibroblasts
Isolated from
Patients with Idiopathic
Pulmonary
Fibrosis
Have
a
Diminished
Capacity
to
Synthesize Prostaglandin E2
and
to
Express
Cyclooxygenase-2
Jerome Wilbom,* Leslie J. Crofford,* Mane D. Burdick,* Steven L. Kunkel,§Robert M.Strieter,* and MarcPeters-Golden*
Divisions of *Pulmonary and Critical Care Medicine and
*Rheumatology,
Department of Internal Medicine, and §Departmentof Pathology, University of Michigan and Department ofVeteransAffairsMedical Centers,AnnArbor, Michigan48109Abstract
Prostaglandin E2 (PGE2) inhibits fibroblast proliferation and collagen synthesis. In this study, we compared lung fibroblastsisolated frompatientswithidiopathic pulmonary fibrosis (F-IPF) and from patients undergoing resectional surgeryforlungcancer(F-nl)with respect to theircapacity
forPGE2 synthesis and their expression and regulation of
cyclooxygenase (COX) proteins. Basal COX activity,
as-sessed by quantitating immunoreactive PGE2 synthesized
from arachidonic acid, was twofold less (P < 0.05) in F-IPF than F-nl. InF-nl, incubation with the agonists PMA, LPS,orIL-1increased COXactivityandprotein expression
of the inducible form ofCOX, COX-2,and theseresponses
were inhibited by coincubation with dexamethasone. By contrast, F-IPF failed to demonstrate increases in
COX-2 protein expression or COX activity in response to these agonists. Under conditions of maximalinduction, COX
ac-tivityin F-IPFwassixfold less than thatin F-nl(P< 0.05).
Our data indicate that F-IPF haveastrikingdefect intheir
capacity to synthesize the antiinflammatory and
antifibro-genic molecule PGE2, apparently because of adiminished
induction of COX-2 protein. This reduction in the
endoge-nous capacity of F-IPFto down-regulate their function via PGE2 may contribute to the inflammatory and fibrogenic response in IPF. Moreover, webelieve that thisrepresents
the firstdescription ofadefect in COX-2expression in
asso-ciation withahuman disease.(J. Clin. Invest. 1995.
95:1861-1868.) Key words: arachidonic acid *
Iipopolysaccharide-eicosanoids * interleukin-1l prostaglandin H synthase
Introduction
Idiopathic pulmonary fibrosis (LPF)1 is the most common of
the interstitiallung diseases,a groupof disorders characterized
AddresscorrespondencetoMarcPeters-Golden,M.D.,Division of
Pul-monaryandCritical CareMedicine, 3916 Taubman Center, University
ofMichigan Medical Center,AnnArbor, MI 48109-0360. Phone: 313-936-2612;FAX:313-764-4556.
Receivedfor publication 29 July 1994 andinrevisedform 24 Octo-ber 1994.
1.Abbreviations used inthispaper: COX, cyclooxygenase(PGH
syn-thase); EIA, enzyme-linked immunoassay;F-IPF,lung fibroblasts
iso-lated from patients with IPF; F-nl, normal lung fibroblasts;IPF,
idio-pathic pulmonary fibrosis; PLA2, phospholipase A2; RT,reverse
tran-scriptase.
TheJournal ofClinicalInvestigation, Inc. Volume95, April 1995, 1861-1868
by inflammatory injury and fibrosis of the lung parenchyma. Fibrosis is generally considered to be an irreversible finding and is the hallmark of IPF. This fibrotic component of IPF
is characterized by both a striking increase in the number of
fibroblastsand thedepositionof fibroblast-derivedextracellular
matrixproteins, especially collagen ( 1).Fibroblastproliferation and collagen synthesis are therefore of central importance in
IPF, and the modulation of these processes byeffector
mole-cules such ascytokines (2-4), growth factors(5, 6),and
eico-sanoids (7) is of great interest.
Amongthose factorsthatdownregulate fibroblastfunction, prostaglandin E2(PGE2)has been themostextensively studied. Thislipid mediator has been showntodecrease fibroblast prolif-eration (8, 9)andreducecollagen levels by inhibitingits synthe-sis (10, 11) and promoting its degradation (12). Since PGE2
isamajor eicosanoid product of fibroblasts ( 13 -15 ), it is plau-sible to speculate that this molecule might regulate fibroblast function in anautocrine fashion.
Theinitial steps in the metabolism of arachidonic acid (AA)
to PGE2 and otherprostaglandins (PGs) are catalyzed by the enzyme PGH synthase, orcyclooxygenase (COX) (16).Two
distinct isozymes of COX, COX-1 and COX-2, have been iden-tified. WhereasCOX-1 is constitutively expressedin many tis-sues(17), COX-2 has been shown to be inducibleby a variety of stimuli, such as PMA, IL-1, and LPS, in anumber of cell types ( 18-21). In Swiss 3T3 fibroblasts, for example, increases inCOXactivity in response toIL- I andPMAhave been shown
tocorrelate with increases in COX-2 protein (21).
We undertookthis study todetermine whether the eicosa-noid profile and the regulation of PG synthesis in fibroblasts isolated from patients with IPF (F-IPF) differed from those observed in normal human pulmonary fibroblasts (F-nl). Our dataindicate that, although F-nl andF-IPFhad identical eicosa-noidprofiles, F-IPF exhibited significantly less COX activity at baseline than F-nl. Moreover, when cells from the two popula-tionswerestimulated with IL-1, PMA, or LPS, F-nl responded
by increasingtheir PGE2 production overbaseline, but F-IPF
did not. In cells from both groups, these metabolic responses correlated withtheir capacitiestoup-regulate the expression of COX-2protein.
Methods
Patientpopulations.TheIPFstudy group consisted of nine previously untreatedpatientsfrom the University of Michigan Specialized Center of Research in Interstitial Lung Disease project. These patients displayed
radiographicand clinicalfindings consistent with IPF; pathologic con-firmation of[PFwasmadebyopenlungbiopsy. Therewerethree males and sixfemales,with amean±SEM age of55.2±14.9yr (range
28-69 yr).Allpatientswerecurrentnonsmokers. Thedegreesof
using publishedcriteria (22), and the mean inflammation and fibrosis scores from thethree specimens were calculated for each patient. The study group from whichF-ni were isolated consisted of eight patients undergoing pulmonary resection for bronchogenic carcinoma. There were five males and three females, with a mean age of54±15.1 yr (range 39-72 yr). One was a current smoker, five were former smokers (ceased at least 2 mo before surgery), and two were patients who never smoked. F-nl were isolated from areas of lung parenchyma that were consideredhistopathologically normal. The experimental protocol was approved by the University of Michigan Medical Center Institutional Review Board for Approval of Research Involving Human Subjects. The number of patients from whom cells were taken for each experiment isas indicated inthe text and figure legends.
Isolation and culture of pulmonary fibroblasts. Fibroblasts were isolated from open lung tissue biopsy specimens using methods pre-viously described (23). Briefly, lung tissue samples were isolated under sterile conditions, and1-mm3fragments from each of the three sampled segments werepooled and placed in DME (Gibco Laboratories, Grand Island, NY) with 10% LPS-free FCS (HyClone Laboratories, Inc., Lo-gan, UT). Fibroblasts proliferatingfromthese specimens were grown to80% confluence and serially passed in the same medium. At the fifth passage, fibroblasts wereseeded into96-wellplates (0.25 X 105 cells perwell) for metabolic studies, 60-mm dishes (4 x 105 per dish) for immunoblot analysis, and 175-mm flasks(IO'perflask) for RNA analy-sis.Confluent cells werewashed and rendered quiescent by culturing in serum- and LPS-free DME. They were then cultured for 1-72 h in the presence or absence of LPS from Escherichia coli (serotype 01ll:B4, Sigma ChemicalCo., St. Louis, MO) (100 ng/ml), human recombinantIL-l1, (Collaborative Biomedical Products,Bedford,MA)
(2.5 ng/ml),PMA(SigmaChemicalCo.)(50nM), and/or dexametha-sone(SigmaChemicalCo.)(1 MM)beforeharvesting. Cellswere cul-tured at370Cin ahumidifiedatmosphereof 5% CO2in air.
Separation of eicosanoids. Cellularlipidsof fibroblasts cultured in 96-wellplateswereprelabeled by including 1 MCi of[3H]AA (spact
76-100 Ci/mmol; Du Pont-New England Nuclear, Boston, MA) in the medium for 16 h. Unincorporated label wasremovedby washing
cellswith DME, and cells werethenincubated for 30 min with iono-phore A23187 (5MM).Radiolabeled free AA and its eicosanoid metab-oliteswereextracted from themedium,andpooled lipidextractsfrom four wells were subjected to reverse-phase HPLC as previously de-scribed(24). Radioactivitywasdeterminedon-lineusingaRadiomatic
Flo-One Beta detector(PackardInstrumentCo.,DownersGrove, IL).
DeterminationofPGE2accumulation. Aftera16-or72-h incubation in serum-free medium, cumulative PGE2 synthesis from endogenous
AA was determined byassaying supernatants with anenzyme-linked
immunoassay(EIA; Cayman Chemical,AnnArbor, MI)accordingto
the manufacturer's instructions.
DeterminationofCOX and PGE isomerase activities.PGE2
accumu-lation reflects the actions of bothphospholipase A2 (PLA2)andCOX. Tobypass PLA2andthereforeestimatemaximal COX metabolic
capac-ity,cellswereincubatedasalreadyoutlined for 1, 16,or72h, washed,
andincubated for 30 min in DMEcontainingexogenous AA(10 MM).
PGE2formation in the supernatantwasthenquantitated by EIA. Prelimi-nary experiments indicated that this dose of AA was saturating, as
indicated bymaximal PGE2synthesis (data not shown). Inasimilar
fashion,PGE isomeraseactivitywasdetermined aftera15-min incuba-tion with exogenous PGH2and subsequent quantitation ofPGE2. All assayswere performed induplicate, and activities were expressed as
PGE2 formed in nanograms permicrogramof cellularprotein. Protein concentrationwasdeterminedbyamicrotiterplatemodification(Pierce Biochemical, Rockford, IL) oftheBradford method (25) usingBSA
as astandard.
Preparation ofcrudecelllysates.F-nlorF-IPFwereharvested from 60-mm plates by scraping witharubberpoliceman in 250Ml of ice-coldhomogenizingbuffer(50mMpotassium phosphate,0.1 MNaCl,
2 mMEDTA, 1 mMDTT,0.5 mMPMSF,60
pg/ml
soybean trypsininhibitor, pH 7.1). The cell suspension was sonicated onice using a
model 250 sonifier (Branson Ultrasonics Corp., Danbury, CT) at a power level of1, 20% duty cycle for 1.5 min.
Immunoblot analysis. Aliquots ofF-nland F-IPF lysatescontaining 20Mgof total protein were combined 1:1 (vol/vol) with SDS sample buffer (125 mM Tris-HCl, 4% SDS, 20% glycerol, 10%
/13-mercaptoeth-anol, pH 6.8) and immediately boiled for3min. They were then
sub-jectedtoSDS-PAGE on 10% acrylamide gels overlaid with a 5% stack-ing gel by the method of Laemmli (26). High molecular weight rainbow markers (Amersham Corp., Arlington Heights, IL) and COX-1 and COX-2 standards were run in parallel in each gel. TheCOX-l standard
(Oxford Biomedical Research, Inc., Oxford, MI) was purified from sheep seminal vesicle, and the COX-2 standard (generously provided by D. DeWitt, Michigan State University, East Lansing, MI) consisted of microsomes from cos cells transfected with the cDNA for murine COX-2(27). Proteinsweretransferredtonitrocellulose (Bio Rad Labo-ratories, Richmond, CA) using a Trans-Blot Cell (Hoefer Scientific Instruments, San Francisco, CA), and membranes were blocked for 1 h with 10% nonfat dry milk reconstituted with 0.1% Tween in Tris-buffered saline (50mM Tris-HCl, 150 mM NaCl,pH 7.5) (TBST).
The membranes werewashedfor 5 min in TBST and then incubated for 1 h with either anti-COX-1 antiserum(1:5,000 dilution)or anti-COX-2 antiserum (1:300 dilution).The former is arabbitpolyclonal antibodyraisedagainst sheepseminal vesicleCOX(28),and the latter isarabbit polyclonal antibodyraisedagainsta 17-amino acidpeptide
derived from the- murine COX-2 sequence that is notpresent in COX-1 (29).After another 15-min wash inTBST,the membranewas incu-batedfor 1 h with horseradishperoxidase-conjugatedgoat anti-rabbit
IgGsecondary antibody (1:10,000forCOX-l and1:5,000forCOX-2) followedbythree5-min washes in 0.3% TBST and three 5-min washes inTBST. Detectionwasaccomplished usingtheECL enhanced chemilu-minescence method(Amersham Corp.) accordingtothemanufacturer's directions. Multipleexposureswere obtained for eachmembrane,and thedensities of bandscorrespondingtoCOX-Iand COX-2were
quanti-tatedbyvideodensitometry ofappropriately exposed autoradiographs usingimage analysissoftware from Scion(Frederick, MD).To compare the steady-statebasal levels ofCOX-1 and COX-2 among all the F-nl and F-IPFpatients,allsampleswereloaded inarandom fashiononthe
samegelandprocessedinparallel;banddensities, expressedinarbitrary
densitometricunits,weredetermined. Toanalyzetheeffects ofagonists
on COX protein levels,different experimental samples from a given
patient wereloadedonthesamegelandprocessedinparallel,and the densities of bands from agonist-treated samples were expressed as a
percentage of that from the control(unstimulated) sample.
RNApreparationandanalysis. TotalRNA waspreparedfrom cul-tured fibroblastsbytheTri-Reagentmethod(MolecularResearch Cen-ter,Inc., Cincinnati, OH).ForPCRanalysisofRNA,cDNAwas
pre-paredbyreversetranscriptionof 1 Mgof each RNAsampleina50-Ml
reaction containing 50 mM Tris-HCl, pH 8.3, 40 mM KCl, 6 mM
MgCl2, 1 mM DTT, 0.4 mMdNTPs,2 MM random hexamerprimers
(Perkin Elmer-Cetus, Norwalk, CT),0.2U/MlRNaseinhibitor(Perkin Elmer-Cetus),and8U/Ml Moloney-murineleukemia virusreverse
tran-scriptase (RT)(LifeTechnologies,Inc., Bethesda, MD).Thereaction mixtures wereincubatedat roomtemperature for 10 min, at420C for 30 min, andat95TCfor 5 min. The cDNAswerethen dilutedto 100
Ml,
and thesame cDNAmixtureswereused in all PCRs. PCRsperformedin
a50-Mlreaction volumecontaining5 Mlof eachcDNA, 10 mM
Tris-HCl, pH8.3,50 mMKCl,25 mMMgCl2,50MMeach ofdATP,dCTP,
dGTP, and dTTP, 5 Ml of
[a-32P]dCTP
(3,000 Ci/mmol; DuPont-NewEngland Nuclear),and 0.05MlofTaqpolymerase(Perkin
Elmer-Cetus).Theprimersusedwere(a)humanCOX-1,sense5'-TGC CCA GCT CCTGGC CCG CCGCTf-3'andantisense 5 '-GTG CAT CAA CAC AGG CGC CTCTTC-3';(b)humanCOX-2,sense5'-TTCAAA TGA GAT TGTGGGAAAATTGCT-3'and antisense 5 '-AGA TCA TCTCTG CCTGAGTATCTT-3';and (c) G3PDH, sense5'-CCA
CCC ATG GCA AAT TCC ATG GCA-3' and antisense 5 '-TCT AGA CGG CAG GTC AGG TCC ACC-3'. All primer pairs amplified a
fragment that crossed an intron, thereby distinguishing cDNA from
18- Figure 1. PGE2
accumu-*T
F-ni 111 lation over 16 h inF-nl5;
| F-IPF and F-IPF. BothF-niand:0
X 12- F-IPF were grown to80%confluence. They
N
W 6 werethenincubated for
0
a. 16hinserum-free
me-dium. Aliquots of the
su-0 pernatants wereusedto
F-nI F-IPF quantitate PGE2 byEIA. Results areexpressed in nanograms of PGE2 permicrogramof total cellproteinperwell.Each
barrepresents mean±SEM fromn =3. *P<0.05versuscorresponding
F-nl valueby unpaired t test.
Theconditions foramplification were as follows: COX-2, 950Cfor 2 min for 1 cycle, 950Cfor 1 min, 600C for 1 min,720Cfor 1 min for 35 cycles, and 720Cfor 7 min for 1 cycle; and COX-I and G3PDH,
95°C for 1 min,60°C for 1 min, and 72°C for 1 min for 25 cycles.
Bothcycle programswerepreceded by2 min at the stated denaturation temperature and followed by 7 minat72°C. Cyclecurvestudies between 25and 35cycles confirmed that, for the amounts of cDNAbeing
ampli-fied, the reactions had not reached the plateau of theamplificationcurve
for any primer pair. Negative controlsperformed withnoRNAadded to the reverse transcription reaction or no RTyielded nodetectable fragmentswith either primer pair.
Data analysis. Where applicable, data are expressed as mean±SEM. Paired or unpaired Student's t tests were used to analyze differences in enzymeactivity or steady-state COX protein levels. Linear correlation analysis was used to assess the correlation between PGE2 synthetic
capacity and fibrosis scores. Significancewasinferred from a P value
< 0.05.
Results
ConstitutivePGE2 accumulation in cultures of F-nl and F-IPF. To assess the capacity of F-nI and F-IPFto produce PGE2 in the absenceofanagonist, equal numbers of cellswerecultured
in serum- andLPS-free DMEM for 16 h, and PGE2 released intothemedium wasquantitated. F-IPFproduced6.5-fold less immunoreactive PGE2 per
1ug
of protein than did F-nl (P< 0.05,n =3 for F-nI and F-IPF)(Fig. 1). The proteincontent
did not differsignificantly betweenF-nlandF-IPF(- 130
pg
ofcrude lysateprotein per60-mm dish at80% confluence forboth F-nl andF-IPF).
Eicosanoid profile in F-nl and F-IPF. PGE2 synthesis from endogenous AA reflects the sequential activities ofPLA2 and COX.Todetermine whether thedefect inF-IPFPGB2synthesis reflected adefect in PLA2-dependentAArelease, we assessed thereleaseof radiolabeled products from prelabeled cells
stimu-lated with A23187 for 30 min. Completely separating freeAA andall its metabolites by HPLC further permittedus to
deter-mine whether the two cell populations metabolized AA into
different eicosanoid products and, in particular, whether there was adifferential capacity forthese cells to synthesize
antiin-flammatory PGandproinflammatory leukotrienes.As shown in theexperiment depicted in Fig. 2, which is representativeof a totaloftwoindependent experiments, bothcell types produced a similar profile of eicosanoids upon stimulation, with PGE2
and free AA being the major products. Neithercell type
pro-ducedany detectableproducts of the 5-, 12-,or15-lipoxygenase
pathways. These data clearly establish thatF-IPF were capable
of releasing endogenous AA; however, the ratio of PGE2tofree
a-._
*t
._o
0-0
0
400-PGE2
F-nl
300
-0--FIPF 8 712
0 20 40 60 E6 100 120
time
(min)
Figure2. Radiolabeled eicosanoidprofileofionophore-stimulatedF-nl and F-IPF. Cellularlipidsof fibroblastswereprelabeledwith[3H]AA
duringa 16-hculture in 96-wellplates. Unincorporatedlabelwas
re-moved, and cellswerewashed and then incubated for 30 min with 5
juMA23187.Radiolabeledproductsinpooledextractsfromfourwells
wereseparated by HPLC,andradioactivitywasdeterminedvia on-line detection. Peaks ofradioactivitywereidentifiedonthe basis of
comigra-tion with authentic standards. Similar resultswereobtained inaseparate
experimentusinganotherpairof F-nl and F-IPF.
AA was strikingly lower (- 18-fold in the example shown in Fig. 2) in F-IPF than in F-nl. This indicates that theprimary
limitation inPGE2 synthesis was inutilization of released AA
by the COXpathway of F-IPF.
COX activity in unstimulated F-nl and F-IPF. To focus
more directly onthisapparentdefect in COX metabolism, we
added exogenous AA in order to bypass PLA2, thereby
as-sessing maximal COX activity by quantitating PGE2 formation.
Becauseboth cellproliferation (30) andserum(31)can
influ-encePGsynthesis and COX activity,westudied cellsatvarious time points up to 72 h after the removal of serum to ensure
that anyresidual effects ofserum had abated. There were no
significant differences in COX activity over the 72-h culture period (i.e., at 1, 16, or72 h) for either fibroblast population (Fig. 3). Interestingly, COX activitywas 2.5-fold lower in F-IPF ascomparedwith F-nl (P<0.05)atalltimepoints studied (Fig. 3). All subsequent experiments wereperformedat 16 h.
ls-F-nl
2Figure
3. COXactivity
F-c
IPF in F-nl andF-IPF.
Sub-confluent F-nl and F-IPF
10 were cultured in
serum-free medium for1,16,or
cm 72h, washed,and then
*o
5-*1 * incubated with 10ILM
AAfor 15 min.
Aliquots
of the cellsupernatants
0 1 16 7 h wereusedto
quantitate
PGE2 by ETA.Resultsare
expressed in nanograms ofPGE2permicrogramof total cellproteinper well. Each bar represents
mean±SEMfromn =4. *P< 0.05 versuscorrespondingF-nl value
A
COX-1
69
k-I
IPF IPF nl nI IPF
69kD-l*
B
200-100 eo e
.82()0-@2 F-nl
F-Il
-S
COX-2
std. IPF IPF nI ni
69
kD-
:Uj
c
COX-1 at
std. 0
01-CY
A_
0_
(
T
COX-1 IPF ni std.
Mil
au
F-nI
50- t
F-IPF*
t
40-
t
30-
20-1:
0**
I
I
I~~~~
cont LPS LPS/dex PMA PMA/dex IL-1 Figure 5. COXactivity in stimulatedF-nIand F-IPF. Cells were incu-bated for 16 h in the presence or absence(control;cont.) of 100ng/ mlLPS (n= 7for F-nl; n = 8 forF-LPF),50 nM PMA (n =5 for bothF-nIandF-IPF),or2.5 ng/mlIL-i,6(n= 3 for both nl and F-IPF) and 1
/iM
dexamethasone (dex). Cells were washed and then incubated with 10MM
AAfor 15 min.Aliquots of the supernatants were used toquantitatePGE2 by EIA. Results areexpressedin nano-gramsofPGE2 permicrogramof total cellprotein per well. Each bar represents mean±SEM. *P<0.05versuscorresponding F-nivalueby unpairedttest;tP <0.05 versus unstimulated control valueby pairedttest.
Figure 4. Immunoblot analysis of COX proteins in unstimulated F-ni
and F-IPF. Cells were culturedfor 16 h inserum-freemedium and harvestedfor immunoblotanalyses asdescribedin Methodsusing20
sgof crudelysate proteinper lane.(A)Autoradiograph depicting COX-1 expressionin cellsfromeightnlandeight IPF patients. Migration of
amolecularweight markerisindicatedbythearrows.COX-1 standard was run in theright lanes. (B) Densitometric analysis of COX-1 bands inF-ni andF-IPFfrom theaboveautoradiograph. Data areexpressed
asarbitrary densitometric units,andeachbar representsmean±SEM.
(C) Autoradiograph depicting COX-2 expressionin cellsfrom threenI
and three IPFpatients. Migration ofamolecularweightmarker is indi-catedbythe arrow. ACOX-1 standardwas runin theright lane,anda
COX-2 standard in theleft.
The formation of PGE2 from AA requires the actions of not
only COX, but also PGE isomerase. To exclude thepossibility
thatdifferences in PGE2 synthesis wereattributable to
differ-ences inPGE isomerase activitybetween thetwo cell
popula-tions, PGE2 formation from exogenously supplied PGH2 (5
ILsM)
wasdetermined. Both F-ni and F-IPFsynthesized equally large
amountsof PGE2 from exogenouslysupplied PGH2 (396±84.3
ng per jg ofproteinforF-nIand 473±67.8 ng per
qg
ofproteinfor F-IPF) (P > 0.05, n = 3). Moreover, this 40-fold (in
F-nl) to 100-fold (in F-IPF) greater capacity forproduction of PGE2 from exogenous PGH2 than from exogenousAAconfirms thatCOX itself is the rate-limiting step within this metabolic pathway. Our finding that differences between F-nl and F-IPF in PGE2 production from exogenous AA were attributable to
differences in theactivityof COX itselfwasthereforeexpected.
Immunoblotanalysis ofCOX proteins in unstimulated fi-broblasts. The expressionof COX-1 and COX-2proteins was
examined in F-nl and F-IPF cultured for 16 h in serum- and LPS-free medium. Crude lysates from both cell populations
weresubjectedtoimmunoblotanalysisusing polyclonal antisera
specific for
COX-l
and COX-2. As shown in Fig. 4A, mostof the cellsamples in both groups (n = 8 foreach) expressed COX-1 in the unstimulated state. Densitometric analysis
re-vealed that COX-1 expression in thetwopopulationswas simi-lar(Fig.4B). Thiswas an unexpected finding, given the 2.5-fold lower COX activity in unstimulated F-IPF as compared with unstimulated F-nl, and ledus toconsider that differences in COX-2 expression might be responsible for this difference in COX activity in the basal state. We found, however, very low levels of basal COX-2 expression in both groups. After prolonged exposure of autoradiographs, COX-2 was detectable insomesubjects,butnodifferenceswereapparentbetween the twopopulations (Fig.4 C).
COX activityinstimulatedfibroblasts. Inview of the reduc-tion in COXactivity inF-IPFin the basal state,we next
exam-ined COXactivityin stimulated cellsfrom thetwopopulations.
Previous studies have shown that variousagonists can induce COX-2 protein and that coincubation with the glucocorticoid
dexamethasonecanprevent this induction. Cellsweretherefore incubated for 16 h with medium alone or in the presence of LPS (100 ng/ml),PMA(50 nM),or
IL-1,8
(2.5 ng/ml), after which COXactivitywasassessedbyquantitating PGE2 formed from exogenous AA. As shown inFig. 5, F-nl increased their activity approximatelytwo- tothreefold in responseto stimula-tion with each of these threeagonists. Bycontrast,F-IPF exhib-itednoincrease in COXactivity with any of theagonists. Thus,in the stimulatedstate the difference in maximal COXactivity
between F-nl and F-IPF was even more pronounced: F-IPF
exhibited sixfold less maximal COX activitythan didF-nl (P
< 0.05 for LPS, PMA, and IL-1 stimulation). As expected,
dexamethasonepreventedtheagonist-inducedincrease inCOX
activity in F-nl. SinceF-IPF exhibitedno induction of COX-2, it was not surprising that these cells were unaffected by dexamethasone.
Immunoblotanalysis of COX proteins instimulated
C
69 kD -_
F-nl
F-IPF
Figure6.Immunoblotanalysisof COXproteinsin stimulated F-nl and F-IPF.Cellswereincubated for 16 h in the absence(control; cont)or
pres-enceof 100ng/mlLPS,2.5ng/ml
IL-1,B,or 1
ttM
dexamethasone (dex).Crudelysate protein (20,g)was
sub-jectedtoimmunoblotanalysis for COX-2proteinasdescribed in Meth-ods.(A) Representative
autoradio-graphwithmigrationof molecular
weightmarkersindicated bythe arrows.ACOX-2 standard was run in therightlane.(B) Densitometric
analysisofCOX-2 bands inF-nIand F-IPFincubated with medium alone
(control; cont),LPS(LPS),and LPS
plusdexamethasone(LPS/dex). Rel-ativeCOX-2 levels areexpressedas apercentageof control andrepresent mean±SEM fromseven nl andeight
69 kD IPF. *P < 0.05 versus F-nl by un-pairedttest;and tp < 0.05versus LPSby pairedttest.(C)
Representa-tiveautoradiograph showingCOX-2 induction inresponseto 2.5 ng/ml IL-1.
cells stimulated with LPS, PMA, and IL-1. A representative autoradiograph shown in Fig. 6 A reveals that steady-state COX-2 protein expression was increased by LPS in F-nl, and this inductionwaspreventedby coincubation with dexamethasone. Such induction of COX-2 proteinwasnotfound in F-IPF. Den-sitometric analysis of COX-2 protein expression (Fig. 6 B) indicates that in F-nl (n = 5), relative COX-2 expression
in-creased approximately twofold (P <0.05versuscorresponding controlvalue) inresponsetoLPS and decreased below baseline control levels (P < 0.05) when cells were coincubated with LPS plus dexamethasone. Bycontrast,COX-2wasnotinduced inF-IPF inresponsetoLPS(n=5).Moreover,LPS-stimulated
F-IPF expressed twofold less COX-2 than did LPS-stimulated F-nl. Inasimilar fashion, IL-1 induced COX-2 protein expres-sion substantially overthe control level inF-nl, butnot in F-IPF(Fig. 6 C). Similar induction of COX-2 in F-nl butnotin F-IPF was observed with PMA treatment (data not shown). Agonist stimulation hadno effecton COX-1 levels, and there was nosignificant difference in COX-1expression between the
two populations inthe stimulatedstate (datanotshown).
Expression of COX-2 mRNA inF-nland F-IPF. To deter-mine whether the defect responsible for the lack of COX-2 protein expression in F-IPF was pretranslational, we isolated total RNA from F-nl and F-IPF that had been stimulated with orwithout 100 ng/ml LPS. Fig. 7 demonstrates that cells from
each individual patient examined were representative of the
entiregroupwithrespect totheirCOX activity (top panel) and COX-2 protein expression (middle panel), in that F-nl (Fig.
7 A) but not F-IPF (Fig. 7 B) exhibited increases with LPS stimulation. RT-PCR analysiswasperformedontheseF-nl and F-IPF in the basal and LPS-stimulated states. COX-2 mRNA
in basal F-IPF was detectable only after 35 PCR cycles and
failed to increase afterincubation with LPS (Fig. 7 B, bottom panel), mirroring the results obtained for COX-2 protein ex-pression under thesametreatmentconditions (Fig. 7B, middle
panel). Abundant mRNA was detected in F-nl at 35 cycles (Fig. 7A, bottom panel); in fact, COX-2was detectable after only25cycles inF-nl. When normalizedtoG3PDH mRNA, a 2.8-fold increase in the COX-2 PCR fragmentwasobserved in
F-nltreated with LPS ascompared with control, whereas there was noincrease in COX-2 transcript after LPS treatmentin F-IPF. In sharpcontrast tothese striking differences in basal COX-2 mRNA expression, basal COX-I levelswere similar inF-nl and F-IPF (datanotshown).
Correlation between COX activity and fibrosis. To
deter-mine whether there was a correlation between COX activity andthe degree of pulmonary fibrosis, COX activities in both basal and LPS-stimulated fibroblasts from patients with IPF wereplotted againstmean fibrosisscores. As shown in Fig. 8, a statistically significant inverse relationship existed between thePGE2synthetic capacity of LPS-stimulated F-IPF and fibro-sisscores (r= 0.804). A similar correlation (r= 0.774)was noted for unstimulated F-IPFaswell (datanotshown).
Discussion
Fibroblasts have been studied extensively as targetcells in fi-broticdisorders, includingIPF.Lessis known about the possible role of fibroblastsas effector cells actively participating in the evolution of IPF. Because fibroblasts areknown to synthesize PGE2, and this PG has the ability to downregulate fibroblast proliferation and collagen synthesis, we hypothesized that F-IPF might exhibitadiminished capacity for PGE2 synthesisas compared withF-nl. This hypothesiswastestedinprimary
cul-tures of fibroblasts isolated from lung biopsy specimens from patients with IPF and from nonfibrotic control patients undergo-ing resection for bronchogenic carcinoma. Our study yielded several major findings. (a)F-IPFexhibited diminished capacity
to synthesize PGE2 from endogenous AA in both basal and ionophore-stimulated conditions. (b) Similar reductions in
A
LPS92.5 kD P 69 kD-*
F-nl
LPFS COX-2
4-92.5 kO
4-69kD
F-IPF
U
U
El
B
200-B c( 2 150-x E
0 °
O-e
a
-CD
50-
O-cont. LPS LPS/dex
*
Figure7. COX activity and expression of COX-2 protein andmRNA in F-nl and F-IPF. Cells from a patient without (A) and with (B) IPF were incubated for 16 h in the absence (-) or presence ( + ) of100
ng/ml LPS.COX activity was assessed as described in Methods(-LPS, open bars; +LPS, hatched bars), and expressed in nanograms of PGE2 per microgram of total cell protein per well (top pan-els). Crudelysate protein (20
,g)
was subjected to immunoblotanalysis for COX-2 protein as described in Methods(middle panels), and RT-PCR (35 cy-cles)analysis of COX-2 mRNA expression wasper-formedasdescribed in Methods(bottompanels).
PGE2synthesis from exogenous AA (but not exogenous PGH2) suggest a defect at the level of the COX enzyme itself. (c) Unlike F-nl, F-IPFwereincapable of increasing COX activity after pretreatment with the agonists LPS, PMA, and IL-143. (d) This inability to augment COX metabolic activity correlated with a failure to increase steady-state levels of COX-2 protein and mRNA. (e) Finally, a significant inverse correlation was observed between the PGE2 synthetic capacity of F-IPF and semiquantitative fibrosis scores of lung tissue from which these
F-IPF were obtained.
Several methodological features ofourstudy deserve com-ment.First, cells from three separate lung segmentswerepooled
to obtain theF-IPF used. In the interstitial lung diseases, the degrees of pulmonary inflammation and fibrosis are well known
tobe heterogeneous in distribution (1). Although pooling cells
CL
w
a..
0) c
9-
3-0 5 10
fibrosisscore
15 20
Figure8.Correlationbetween COXactivity and fibrosisscore.Mean pathologicscoresfromthreebiopsy specimensareplotted'against COX activity (expressedasnanogramsofPGE2permicrogram ofprotein) in theLPS-stimulatedstatefor each ofseven lPFpatients. The relation-ship between PGE2 synthetic capacity and fibrosis is described by the equationy =9920.8 - 362.83x;r= 0.804.
from different regions minimizes the possibility of sampling errors, it also averages out cellular heterogeneity, thereby tend-ing to underestimate the magnitude of abnormalities in the most severely affected areas. Second, because alterations in eicosa-noidprofiles have been reported to accompany serial passage of fibroblasts(32), we chose to study cellsatthe fifth passage, reasoning that this was a sufficient passage number to ensure purity, but not so great that differences present in vivo might be lost. Nonetheless, the differences in maximal COX activity between F-nl and F-IPF thatweobserved havepersisted through the 12th passage (n = 2 for both cell populations; data not
shown). Third, because both the proliferativestateand the se-rum itself can influence PG synthesis and COX activity, we studiedquiescent cellsat various timepoints up to 72 h after removalof serum. The defects in COX activity in F-IPF were observedatalltimepoints.
Weinitially observed that constitutive PGE2 accumulation in F-IPFwas 6.5-foldlower than that in F-nI when cells were
studiedat16 h inculture. We therefore performed HPLC analy-sistodetermine whether there were any differences in the eico-sanoidprofilesbetween the twocell types that wouldaccount
for this finding. A previous report indicated that human skin fibroblasts failed to synthesize 5-lipoxygenase metabolites of
AA (33). However, since leukotrienes have the capacity to stimulate fibroblastproliferation and collagen synthesis (34), itwasneverthelessof interest to determine whetherthese metab-olites were synthesized by F-IPF. In fact, HPLC analysis de-tectednolipoxygenase products elaborated by fibroblasts from eitherpatientgroup. Rather, in both F-IPF and F-nl the major
productsobservedwerefreeAAandPGE2. Whatwas striking
was themarkedly reducedcapacityof F-IPFtometabolize the releasedAAtoPGE2,ascompared with F-nl. Although we did
notcompare thetwopopulations for theirquantitativecapacities
torelease AA, HPLC analysis of prelabeled cells clearly indi-cated that F-IPF were capable of releasing free fatty acid, thereby implicating adefect distal toPLA2. This does not
ex-cludeapossibleconcomitant reduction inAAlevels thatmight beavailable as substratefor COX (i.e., decreased deacylation
orincreasedreacylation)in F-IPFascomparedwith F-nl. This
possibilitymust beexploredin future studies.
0
0
0
Thecapacityfor PGH2 metabolismwasfargreater than the cells' capacity forPGH2formation, suggestingthat COX is the rate-limiting enzyme. Moreover, no differences in PGE iso-merase activity were detected between the two cell types in stimulated (datanot shown) orbasal states. But F-IPF did
ex-hibit 2.5-fold less COX activity than F-nl in the basal state.
Despite this, wedetected no differences in steady state levels
of COX-1 proteinand noappreciableCOX-2protein expression
in the basal state. Although the mechanism for reduced basal COXactivityin F-IPF has notbeen elucidated in this investiga-tion, several possibilities exist. First, COX-1 enzyme activity
in F-IPF may be less per molecule than that in F-nl,reflecting either intrinsicenzymemodificationsordifferences in modulat-ing factors. Second, the subcellular localization of COX-1 in F-IPF and F-nl might differ, resulting in decreased access of the enzyme in F-IPF to endogenous and exogenous substrate.
Third, differences in constitutive expression of COX-2 might
exist in the two populations, but may be below the limits of
sensitivity of the Western blotting procedure used. Using RT-PCR, wedidfind that whereasCOX-1 mRNA levels were simi-lar, COX-2 mRNA levels were indeed lowerin F-IPF than in
F-nl in the basal state. Identifying the mechanism responsible for this reduction in basal COX activity in F-IPF will require furtherinvestigation.
COX activity in F-nl wasincreased above baseline by the
addition of the agonists LPS, PMA, andIL-1p6 to the culture medium. Evidence that this increase inactivitywasdueto
COX-2induction wasprovided bythe observations that the increase in COX activity correlated with COX-2 protein and mRNA
expression, that there was no increase in COX- 1 expression
with agonist stimulation, and that the increases in activityand COX-2protein expressionwerebothprevented bycoincubation with dexamethasone. In contrast to F-nl, F-IPF wererefractory
to increases in COX activity orCOX-2 protein expression by
these same stimuli. Thatrefractoriness was observedfor three unrelated agonists stronglysuggests that the defect resides not atthe level ofareceptor orsignaling mechanism, but at a more
distal efferentstep in enzymesynthesis. RT-PCR resultssuggest a failure to increase steady-state mRNA levels upon agonist
stimulation. The molecular mechanism underlying this lackof
mRNAinducibility remains tobedetermined, but couldreflect
enhanced turnover rates forthe transcriptorreduced
transcrip-tion rates. This in turn mightresult from anacquired alteration
in theCOX-2promoter,fromaninabilitytoalternatively splice aprematuretranscript (35),orfrom alterations in theregulation
oractions oftranscription factors. It will beofgreatinterestin
future studies to determine whether other lung cells share this defect in COX-2 inducibility identifiedin F-IPF.
To ourknowledge, this is thefirst demonstrated association ofa lack of inducibility of COX-2 with a human disease. A great deal of attention has been focused onthe roleofCOX-2 ininflammation (36), and thedevelopment of selective inhibi-tors ofCOX-2 offersthe prospect ofantiinflammatory efficacy withlessgastrointestinal and renal toxicitythan current nonse-lective COX inhibitors (37). However, it is also recognized that certain PGs, including PGE2, can exertpotent inhibitory actions on leukocytes (38, 39), lymphocytes (40, 41), and
fibroblasts (8, 9, 11,42), whichparticipate inthe processes of
inflammatory injuryandrepair.Induction ofCOX-2byagonists
present in the inflammatory milieuof thelung might represent
animportant mechanism bywhichfibroblastscanincreasetheir
PGE2 synthetic capacity and therebylimit cellularproliferation,
collagen synthesis, andproductionofinflammatory mediators. A defect in this homeostatic process may promote or sustain
inflammation and fibrosis in the lung.In this regard,it is
intri-guing that, despitethe smallsample size,weobserveda
signifi-cantinverse correlation between thedegreeof fibrosis and COX
activity (PGE2 synthetic capacity) in fibroblasts isolated from IPFpatients. Of note, Boroketal. (43)foundsignificantlyless
PGE2 in lavage fluid obtained from patients with IPF than in normal subjects. These data therefore highlight the potential
detrimentalconsequences thatmightresult frompharmacologic inhibition ofCOX-2, at least during the reparative phases of
inflammatory injuryin thelung. Finally, ourobservations may havetherapeutic ramifications.AugmentationofPGE2levels in thelung viaaerosolizationofastablePGE2 analog (43)orvia
transfectionof the cDNAencodingCOX- 1orCOX-2(44)has thepotential to reduce fibrosis and ultimately to improve the
prognosis for this oftendevastating disease.
Acknowledgments
WethankBingTan, RobertMcNish, HollyEvanoff,and JanetHampton
for technical assistance, Drs. Mark Orringer and Richard Whyte for
providing lung tissue specimens, and Drs. William Smith and David DeWittfor kindly providing the COX-1 and COX-2 antisera,
respec-tively.
This work was supported by grants from the National Institutes of Health
(ROl-HL47391
and Specialized Center of ResearchP50-HL46487). M. Peters-Golden is therecipientofaCareerInvestigator
Award fromthe American Lung Association. J. Wilborn is therecipient
ofaMinority InvestigatorResearch SupplementAward from the
Na-tionalHeart, Lung, and Blood Institute.
References
1.Panos, R.,and T.King.1991.Idiopathic pulmonaryfibrosis. In Immunologi-callyMediatedPulmonaryDiseases.J. P.LynchIII and R. A.DeRemee,editors. J. B.Lippincott, Philadelphia,PA. 1-39.
2.Jordana, M.,M. Newhouse,and J.Gauldie. 1987. Alveolar macrophage-andperipheralbloodmonocyte-derivedfactors modulateproliferationofprimary linesof humanlungfibroblasts. J. Leukocvte Biol. 42:51-60.
3.Elias, J.,B.Freundlich,J.Kern,and J. Rosenbloom. 1990.Cytokine
net-worksintheregulationof inflammation and fibrosis in thelung.Chest. 97:1439-1445.
4. Gauldie, J., M. Jordana, and G. Cox. 1994. Cytokines and pulmonary fibrosis. Cvtokines. 4:931-935.
5.Clark, J.,D.Madtes,and G.Raghu.1993.Effects ofplatelet-derived growth factor isoformonhumanlungfibroblastproliferationandcollagengene expres-sion. Exp. Lung Res. 19:327-344.
6.Fine, A.,and R.Goldstein. 1986. The effect oftransforming growth
factor-,6oncellproliferationandcollagenformationby lungfibroblasts. J. Biol. Chem. 262:3897-3902.
7. Elias,J., M.Rossman,R.Zurier, andR.Daniele. 1985. Humanalveolar macrophageinhibition oflungfibroblastgrowth:aprostaglandin-dependent
pro-cess.Am. Rev.Respir.Dis. 131:94-99.
8.Bitterman, P.,M.Wewers,S.Rennard,S.Adelberg,andR.Crystal. 1986. Modulation of alveolarmacrophage-drivenfibroblastproliferation byalternative macrophagemediators. J. Clin.Invest.77:700-708.
9.Elias,J. 1988. Tumor necrosis factor interacts with interleukin-l and inter-feronstoinhibit fibroblastproliferationviaprostaglandin-dependentand indepen-dent mechanisms.Am. Rev.Respir.Dis. 138:652-658.
10.Goldstein, R.,andP.Polgar.1982. Theeffect and interaction ofbradykinin andprostaglandinsonproteinandcollagen production by lungfibroblasts. J. Biol. Chem. 257:8630-8633.
11. Kom, J.,P.Halushka,andE.Leroy.1980. Mononuclear cell modulation of connective tissue function: suppression of fibroblastgrowthbystimulationof endogenousprostaglandin production. J. Clin. Invest. 65:543-554.
12.Baum, B.,S.Moss,R.Breul,and R.Crystal. 1980.Effect ofcyclicAMP
onthe intracellulardegradation ofnewlysynthesized collagen. J. Biol. Chem. 255:2843-2847.
radioimmunoas-say before andafter separation by high pressure liquid chromatography. Prosta-glandins Med.3:295-304.
14. Korn, J., D. Torres, and E. Downie. 1983. Fibroblast prostaglandinE2 synthesis. J. Clin.Invest. 71:1240-1246.
15. Shindo,N., T. Saito, and K. Murayama. 1988. Rapid quantification of 11 prostanoids bycombined capillary column gas chromatography and negative ion chemicalionization mass spectrometry: application to prostanoids released from normal humanembryonic lung fibroblasts W138 in a culture medium. Biomed. Environ. Mass.Spec. 15:25-32.
16. Smith, W., and L. Mamett. 1991. Prostaglandin endoperoxide synthase: structure andcatalysis. Biochem. Biophys. Acta. 1083:1-17.
17. O'Neill, G., and A. Ford-Hutchinson. 1993. Expression of mRNA for cyclooxygenase-l and cyclooxygenase-2 in human tissues. FEBS (Fed Eur. Bio-chem. Soc.) Lett. 330:156-160.
18.Fu, J.-Y., J. L. Masferrer, K. Seibert, A. Raz, and P. Needleman. 1990. The induction and suppression of prostaglandin H2 synthase (cyclooxygenase) in humanmonocytes. J. Biol. Chem. 265:16737-16740.
19. Jones, D., P. Carlton, T. McIntyre, G. Zimmerman, and S. Prescott. 1993. Molecular cloning of human prostaglandin endoperoxide synthase type 2 and demonstration of expression in response to cytokines. J. Biol. Chem. 268:9049-9054.
20.Masferrer,J., K.Siebert,B.Zweifel, and P. Needleman. 1992. Endogenous glucocorticoids regulate an inducible cyclooxygenase enzyme. Proc. Natl. Acad. Sci. USA.89:3917-3921.
21. Kujubu, D., S. Reddy, B. Fletcher, and H. Herschman. 1993. Expression of theproteinproduct of the prostaglandin synthase-2/TIS1O gene in mitogen-stimulated Swiss 3T3cells. J. Biol. Chem. 268:5425-5430.
22. Watters, L., T. King, M. Schwarz, J. Waldron, R. Stanford, and M. Reuben. 1986. A clinical, radiographic, andphysiologicscoring system for the longitudinal assessmentofpatients with idiopathic pulmonary fibrosis. Am. Rev. Respir. Dis. 133:97-103.
23.Rolfe, M., S. Kunkel, T. Standiford, S. Chensue, R. Allen, H. Evanoff, S. Phan, and R. Strieter. 1992.Pulmonaryfibroblastexpressionof interleukin-8: a modelfor alveolarmacrophage-derived cytokine networking.Am. J.Respir. Cell. Mol. Biol. 5:493-501.
24.Peters-Golden, M., R. W. McNish, R. Hyzy, C. Shelly, and G. B. Toews. 1990. Alterations in the pattern of arachidonate metabolism accompany rat macro-phage differentiation in the lung. J. Immunol. 144:263-270.
25.Bradford, M. 1976. Arapidandsensitive method for thequantitationof microgram quantities ofproteinutilizing the principleofprotein-dye binding. Anal.Biochem. 72:248-254.
26.Laemmli,U. 1970. Cleavage of structuralproteins duringtheassembly of the head ofbacteriophage T4. Nature (Lond.). 227:680-685.
27.Meade, E., W. Smith, and D. DeWitt. 1993. Differential inhibition of prostaglandin endoperoxide synthase (cyclooxygenase) isozymes by aspirinand other non-steroidalanti-inflammatory drugs.J.Biol. Chem. 268:6610-6614.
28.Smith, W., and T. Rollins. 1982. Characteristics of rabbit anti-PGH syn-thaseantibodies and use inimmunochemistry.MethodsEnzymol. 86:213-222.
29. O'Sullivan,M., E. Huggins, Jr., E. Meade, D. DeWitt, and C. McCall. 1992.Lipopolysaccharide induces prostaglandin H synthase-2 in alveolar macro-phages. Biochem.Biophys. Res. Commun. 187:1123-1127.
30. Hong, S., G. Patton, and D. Deykin. 1979. Arachidonic acid levels in cellularlipids determine the amount of prostaglandins synthesized during cell growth in tissueculture. Prostaglandins. 17:53-59.
31. O'Banion, M., H. Sadowski, V. Winn, and D. Young. 1991. A serum-andglucocorticoid-regulated4-kilobase mRNA encodes a cyclooxygenase-related protein. J. Biol. Chem. 206:23261-23267.
32. Polgar, P., and L. Taylor. 1980. Alterations in prostaglandin synthesis during senescence of human lung fibroblasts. Mech. Ageing Dev. 12:305-313.
33. Bernd, M., R. Ludwig, E. Zenzmaier, H. Gleispach, and H. Esterbauer. 1984.Characterization of lipoxygenase metabolites of arachidonic acid in cultured humanskin fibroblasts. Biochim. Biophys. Acta. 795:151-164.
34. Baud, L., J. Perez, M. Denis, and R. Ardaillou. 1987. Modulation of fibroblastproliferationbysulfidopeptide leukotrienes: effect of indomethacin. J. Immunol. 138:1190-1195.
35. Diaz, A., A. Reginato, and S. Jimenez. 1992. Alternative splicing of human prostaglandin G/Hsynthase mRNA and evidence of differential regulation of the resultingtranscripts by transforming growth factor ,, interleukin1I3,and tumor necrosis factor a. J. Biol. Chem. 267:10816-10822.
36. Vane, J. 1994. Towards a betteraspirin.Nature(Lond.).367:215-216. 37.Masferrer,J., B.Zweifel,P.Manning,S.Hauser,K.Leahy,W.Smith,P. Isakson, and K. Seibert. 1994. Selective inhibition of inducible cyclooxygenase-2in vivo isantiinflammatory andnonulcerogenic. Proc.Natl. Acad. Sci. USA. 91:3228-3236.
38.Ham,E.,D.Soderman,M.Zanetti,H.Dougherty,E.McCauley,and F. Kuehl, Jr. 1983. Inhibition by prostaglandins ofleukotriene B4 release from activatedneutrophils. Proc.NatI.Acad. Sci. USA. 80:4349-4353.
39.McLeish, K., G. Stelzer, and J. Wallace. 1987. Regulationof oxygen radical releasefrom murineperitonealmacrophages bypharmacologicdoses of PGE2. Free Radical Biol. Med. 3:15-20.
40.Oppenheimer-Marks, N.,A.Kavanaugh,and P.Lipsky. 1994. Inhibition of thetransendothelialmigrationof human Tlymphocytesby prostaglandinE2. J. Immunol. 152:5703-5713.
41.Betz, M., and B. Fox. 1991.ProstaglandinE2 inhibitsproductionof Thl lymphokines but not of Th2lymphokines. J. Immunol. 146:108-113.
42. Goldstein, R., and A. Fine. 1986. Fibrotic reactions in the lung: the activation of thelungfibroblast.Exp. LungRes. 11:245-261.
43. Borok, Z., A. Gillissen, R. Buhl, R. Hoyt, R. Hubbard, T. Ozaki, S. Rennard,and R.Crystal.1991.Augmentationof functionalprostaglandinE levels on therespiratoryepithelialsurfacebyaerosoladministration ofprostaglandinE. Am.Rev.Respir. Dis. 144:1080-1084.
44.Conary, J.,R.Parker,B.Christman,R.Faulks,G.King,B.Meryick,and K.Brigham. 1994. Protection of rabbitlungsfrom endotoxininjury byin vivo hyperexpression of the prostaglandin G/H synthase gene. J. Clin. Invest. 93:1834-1840.