Neutrophil induced injury of rat pulmonary alveolar epithelial cells

Neutrophil-induced injury of rat pulmonary

alveolar epithelial cells.

R H Simon, … , P D DeHart, R F Todd 3rd

J Clin Invest.

1986;

78(5)

:1375-1386.

https://doi.org/10.1172/JCI112724

.

The damage to pulmonary alveolar epithelial cells that occurs in many inflammatory

conditions is thought to be caused in part by phagocytic neutrophils. To investigate this

process, we exposed monolayers of purified rat alveolar epithelial cells to stimulated human

neutrophils and measured cytotoxicity using a 51Cr-release assay. We found that

stimulated neutrophils killed epithelial cells by a process that did not require

neutrophil-generated reactive oxygen metabolites. Pretreatment of neutrophils with an antibody

(anti-Mo1) that reduced neutrophil adherence to epithelial cells limited killing. Although a variety

of serine protease inhibitors partially inhibited cytotoxicity, we found that neutrophil

cytoplasts, neutrophil lysates, neutrophil-conditioned medium, purified azurophilic or

specific granule contents, and purified human neutrophil elastase did not duplicate the

injury. We conclude that stimulated neutrophils can kill alveolar epithelial cells in an oxygen

metabolite-independent manner. Tight adherence of stimulated neutrophils to epithelial cell

monolayers appears to promote epithelial cell killing.

Research Article

Find the latest version:

Neutrophil-induced Injury

of Rat

Pulmonary

Alveolar

Epithelial Cells

Richard H.Simon,Peter D.DeHart, and RobertF.ToddIII

Pulmonary&Critical Care MedicineDivision, and the Hematology and Oncology Division, Department ofInternal Medicine, University ofMichiganSchoolofMedicine, Ann Arbor, Michigan 48109

Abstract

Thedamagetopulmonary alveolarepithelial cells that occurs

inmanyinflammatoryconditions isthoughttobe caused inpart by phagocytic neutrophils. Toinvestigatethis process,we

ex-posed monolayersofpurifiedrat alveolarepithelialcells to

stim-ulated humanneutrophilsand measuredcytotoxicity usinga

"Cr-releaseassay.We foundthat stimulatedneutrophilskilled

epi-thelialcellsbyaprocessthat did notrequire neutrophil-generated

reactive oxygenmetabolites. Pretreatment ofneutrophils with

an antibody (anti-Mol)that reduced neutrophil adherence to

epithelialcellslimitedkilling. Althoughavarietyofserine pro-tease inhibitors partially inhibitedcytotoxicity, wefound that

neutrophil cytoplasts, neutrophil lysates, neutrophil-conditioned medium, purified azurophilic orspecific granule contents,and

purifiedhumanneutrophilelastasedid notduplicatetheinjury. Weconclude thatstimulatedneutrophilscankill alveolar

epi-thelialcells inanoxygenmetabolite-independentmanner.Tight

adherence of stimulatedneutrophilstoepithelialcellmonolayers

appears topromote epithelialcellkilling. Introduction

Injuryto alveolar epithelial cells isa prominent featurein a

variety oflung diseases. In the adult respiratory distress

syn-drome,destruction ofalveolarepithelialcells has been notedto beafrequentandearly finding (1, 2).Loss of thepermeability

barrier formedbyalveolarliningcells islikelytobeanimportant

factorcontributingto thealveolarfloodingwhichoccursinthis

condition(3). Althoughtheprecisemechanisms oflung injury

intheadultrespiratorydistresssyndromeremain to bedefined,

manyinvestigatorshavesuggestedthat stimulatedneutrophils

are animportantcauseoflung damageinatleastsomecasesof

thedisease(4-13).

Toelucidateonepotentially importantmechanism of injury,

westudiedthecytotoxiceffect ofstimulated humanneutrophils

upon isolated rat alveolarepithelialcells.Usinganinvitro

cy-totoxicityassay,wefound thatstimulatedneutrophilskill

epi-thelialcells ina mannerthat appearstorequireintimate

neu-trophil-epithelialcellcontactbut doesnotinvolve neutrophil-generatedoxygenmetabolites.

Methods

Materials.Elastase(type I), trypsin (type III), soybean trypsininhibitor

(type II-S), deoxyribonucleaseI(type

HI),

bovineserumalbumin(A6003),

Addressreprint requeststoDr.Simon, Pulmonary& CriticalCare

Med-icineDivision,3916TaubmanCenter,AnnArbor,MI48109.

Receivedforpublication21January1986 and inrevisedform7July

1986.

Griffonia

simplicifoliaIlectin, phorbolmyristateacetate

_(PMA),V

zym-osan, the calcium ionophoreA23187,

formylmethionylleucylphenylal-anine (FMLP), Triton X-100, superoxide dismutase(S8254), a,-anti-trypsin(A6150),N-chlorosuccinimide,aprotinin (Al 153),

phenylmeth-ylsulfonylfluoride (PMSF), orcein-elastin,heparin(210-6),protamine

sulfate (P4020),Ficoll (F4375), and dextran (D7265) were obtained from Sigma Chemical Co. (St. Louis, MO). The PMA (1 mg/ml), PMSF (1 M),A23187 (2 mM), and FMLP(2 mM) were dissolved in dimeth-ylsulfoxide and diluted to the appropriateconcentrations in aqueous buffers. Zymosan was boiled for 5minand washed three times in normal

saline.It wasthensuspendedinfreshlyprepared human serum at 10 mg/ml andincubated at 370C for 30 min. The treated zymosanwas

then washed three times in normal saline and resuspended in the desired buffer. Azocoll was obtained fromCalbiochem-BehringCorp.(LaJolla,

CA). To alter the

a1-antitrypsin

so as toblock itsability to inhibitelastase,

thea1-antitrypsinwasoxidizedwithN-chlorosuccinimide (14)according tothe methodof Shechter et al. (15). In particular,4 mg of

aI-antitrypsin

was incubated atroom temperature with 0.1 mg ofN-chlorosuccinimide in 1 mlof 0.05 M Tris-HCl buffer, pH 8.5. After 20

min,

thesample

was dialyzed to remove unreacted N-chlorosuccinimide. The

a,-anti-trypsintreated in this manner lost100% of its ability to limit the digestion of Azocollby porcine pancreatic elastase.

Catalase was obtainedfromWorthington Biochemical Corp. (Free-hold, NJ) andfromSigmaChemical Co. To inactivate the catalase, the enzyme was carboxymethylatedaccording to the method of Crestfield andcolleagues (16). Loss of activity was confirmed using the assay of Bergmeyer (17). Ketamine was obtained from Park-Davis(MorrisPlains,

NJ). Dulbecco'smodifiedEagle'smedia, penicillin, streptomycin,

am-photericin B,Hanks' balanced saltsolution,andfetal bovine serum were obtainedfrom Gibco(Grand Island, NY). Dextran sulfate (500,000 dal-tons) was obtainedfromPharmacia (Uppsala,Sweden).Human

neutro-philelastase waspurchased fromElastinProducts(Pacific,MO). Poly-brene was obtained from Aldrich Chemical Co. (Milwaukee, WI). Methoxysuccinylalanylalanylprolylvalylmethylchloroketone

(MeoSuc-ala-ala-pro-val-CH2Cl) wasobtained from Enzyme SystemsProducts

(Livermore,CA).

Na2[5Cr]04

(500 mCi/mg) and[3,4,5-3H(N)]leucine (147mCi/mM),both insterile normal saline, were obtained from New England Nuclear (Boston, MA).Anti-Mol (18, 19),anti-Mo5(20), and anti-Ia (9-49) (21)antibodies,all monoclonal andofthe IgG2a subclass, were generated and characterized as previously reported.

Alveolarepithelial cell isolation procedure. TypeIIalveolar epithelial

cells were isolated from specific pathogen-free Fischer 344 male rats(22).

Inbrief,theanimals were anesthetized intraperitoneally with ketamine, thelung vasculature wasperfused,andthe airways were lavaged with Hanks'balanced saltsolutionlacking calcium and magnesium, and the alveolarepithelialcells were releasedbytheintratrachealinstillation of

anelastase and trypsin mixture.Afterneutralizing the proteases, mincing thelungs,andremovingthetissue pieces by filtration, the cellular

sus-pension

wasincubated with

Griffonia simplicifolia

Ilectin which

agglu-tinatesmacrophagesbut not type IIepithelialcells.Afterremoving ag-glutinated cells byfiltrationthrough nylon mesh, the cells were suspended in Dulbecco'smodified Eagle's medium containing 10% fetal bovine serum, 100,000U/literpenicillin, 100 mg/literstreptomycin,and0.25 1.Abbreviations used in thispaper: D-PBS, Dulbecco's

phosphate-buff-eredsaline;FMLP,formylmethionylleucylphenylalanine; MeoSuc-ala-ala-pro-val-CH2C1, methoxysuccinylalanylalanylprolylvalylmethylchloro-ketone; PMA, phorbolmyristateacetate;PMSF,phenylmethylsulfonyl fluoride.

Neutrophil-induced Injury ofPulmonary Epithelial Cells 1375 J.Clin. Invest.

©TheAmerican

_Society

for Clinical

_{Investigation,}

Inc.

mg/literamphotericin B. The suspension was added to tissue culture platesand incubated for 48 h in a 5% C02/air atmosphere at370C.After 48h, thecells formedconfluentmonolayers. Thecells were >95% viable by trypan blueexclusion criterion and were 94±2% (mean±SD) type II cellsbyphosphine staining (23). The cells were joined together by tight

junctions,containedcytoplasmiclamellarbodies,andincorporated pal-mitateinto disaturated phosphatidylcholine (22).

Endothelial cells. Bovine pulmonary artery endothelial cells were providedby Dr. Una S. Ryan(University ofMiami School of Medicine) (24). Inour laboratory, the cells weremaintainedin the same medium astheepithelialcells. Inpreparation for cytotoxicity experiments,

en-dothelialcellsintheir fifthto tenth passage were released by exposure totrypsinandethylenediaminetetraacetic acid in Hanks' balanced salt

solution lackingcalcium andmagnesium,pH 7.4.The cells were then suspended inmediumandadded to the wells of a 96-well plate at the sametime that theepithelialcells were added. The plates were incubated

for48 hin 5%C02/air at 370C duringwhich time the cellsformed confluentmonolayers.

Neutrophil isolation. Human neutrophils were isolated from hepa-rinized(10U/ml) peripheralvenous blood byFicoll-Hypaquedensity

centrifugation, dextransedimentation, and hypotonic lysis of residual erythrocytes(25).The cells were >95%viableby trypan blueexclusion criterion.The cells werewashed and maintained in Hanks' balanced salt

solution lacking calciumandmagnesium untilimmediately beforeuse

(<1h). The cells were thencentrifugedat 75gfor10 min and resuspended in Dulbecco'sphosphate-buffered salinecontaining1

_mg/ml

glucose and

0.1 mg/mlbovineserumalbumin(D-PBS). Neutrophilsfromsubjects

with chronic granulomatous disease were madeavailableby Dr.John T. Curnutte(Scripps Clinicand Research Foundation, LaJolla,CA).

Theseneutrophilsfailed to produce detectablesuperoxidewhen stimu-latedby avariety ofagents includingPMA. Neutrophil lysates were preparedbysonicating (model 300,ArtekSystemsCorp.,Farmingdale,

NY)suspensions ofneutrophils forthree20-sbursts. Supernatants of

PMA-stimulatedneutrophilswereobtained byplacing neutrophilsinto wells of 96-well plates inD-PBS,allowing themtosediment onto the bottom of theplatefor 15min,and then

adding

25 ng/mlPMA.The plates wereincubatedat37°Cinhumidifiedair for variousperiods of

time. The cell-free media was removed and the supernatant used in the

cytotoxicity

studies.Purified

neutrophil azurophilic

and

specific granules

wereprovidedby Dr. Curnutte. Thegranulesweredisrupted by freeze-thawingthreetimes and

sonicating

fortwo20-s bursts.

Neutrophil

cy-toplastswereformedandisolated bythemethodofRoosetal.

(26).

Cytotoxicity assay. Neutrophil-induced

cytotoxicity

wasmeasured

bya

5"Cr-release

assay.Targetcellsweresuspendedinmediaandplaced

into 96-wellflat-bottomed tissuecultureplates(Costar,Cambridge, MA). After32 hofincubationat37°Cin 5%C02/air,0.8MuCi

Na2[5"Cr]04

wasadded per well and the cells incubatedforanadditional 16 h. The monolayers were then washed threetimes with D-PBSandvarious reac-tants wereaddedtothewells. If the

cytotoxic

mechanismof

neutrophils

was to betested,neutrophilsweremixed withappropriateinhibitors in 0.15 mlof D-PBSand then addedtoeach well. The

plates

wereincubated

at37°Cinhumidified airfor15minduringwhich time the

neutrophils

sedimentedontothe

epithelial

cell

monolayer.

Next,a

neutrophil

stimulus wasgently added in a volume of 0.05 ml and theplateswerereturned to theincubatorforvaryingtimes. Todetermine theextentof cellular

death, 0.1mlofmedium was removed from the wells and theradioactivity

wasmeasured(TmAnalytic, ElkGroveVillage, IL).Ifthe0.1-mlaliquot

was removed with minimalagitation, the measured cytotoxicity was

thesamewhetherornotthemediawas

centrifuged

andonlythe super-natant was counted (P>0.5, n=8). Next,0.1mlof 1% TritonX-100

wasadded to the wells and the plateswereincubated for 5 min.After

mixing the contents of the well,0.1ml was removed and theradioactivity

wasmeasured.Usingtheresults of these two measurements, it was pos-sible to calculate both the amount of

51Cr

that wasoriginallyboundby

the cells andthe percentage thatwasreleasedinto the mediaduringthe incubationperiod.Byincludingwellscontaining epithelialcells that were exposedtoD-PBSwithout furtheradditions,thebackground release of

5"Cr

could be measured.Usingthisinformation,the percentage ofspecific

5"Cr

released fromneutrophil-exposed cells was calculated by: percentage ofspecific

5"Cr-release

=(A - B)/(C - B)X 100, where A is the counts per minute(cpm) released into the medium of the test sample, B is the cpmreleased from control, unexposed

cells,

and Cisthe total cpm initially bound tothe cells at thebeginningof the experiment.

Cytotoxicitywasalso measured by the release of tritium from epi-thelial cells previously incubated with [3H leucine. After alveolar epiepi-thelial cells wereisolated and cultured for 32 h, 0.2 MCi

[3H]leucine

were added toeach well.Afteranadditional 16 hofculture,the monolayers were rinsed three times with D-PBS and exposed to neutrophils. The percent releaseof tritiumwasmeasuredusingthe method described for

5"Cr

release, but the 0.l-ml aliquots were added to aqueous counting scintiflant (Amersham Corp., Arlington Heights, IL) and the radioactivity was measured in ascintillation counter using H# quench correction (LS 7500, Beckman Instruments, Inc., Fullerton, CA).

Neutrophil adherence. Neutrophil adherence to epithelial cell

mono-layers wasperformed using

5"Cr-labeled

neutrophils. Neutrophils were

isolatedbythe methodoutlinedabove andincubatedwith 8.0uCi/ml

of

Na2[5Cr]04

in Hanks'balanced salt solution lacking calcium and magnesium for 1 h at 37°C. The labeled neutrophils were washed thrice andresuspended in D-PBS.Confluentmonolayersofepithelialcells in 24- or 96-well plates were washed three times with D-PBS and neutrophils were added.After15minofincubation in room air at 37°C, PMA (25

ng/mlfinalconcentration)wasadded. The plate was then incubated for 30minwithout agitation. Themedium in each well was removed and the wells weregentlywashedtwicewith0.2 mlofD-PBS.Themedium

andwashes werepooledand theradioactivitymeasured.Epithelialcell monolayers remainedvisually intactafterthewashes.Alternatively,the amountofradioactivitythatremainedadherent to theepithelialcell monolayer was measuredafterwashingthe wellstwice withD-PBS and adding 1% Triton X-100 to solubilizeneutrophil-associated

5"Cr.

In ad-dition, the amount of51Cr that leaked from PMA-stimulated neutrophils

intothemediaduringtheincubation periodwasmeasured(6.5%) and subtracted from the total counts that were contained in themediaand washes.Using these measurements, the percentage of neutrophils that adhered to theepithelialcellmonolayerswascalculated.

Electronmicroscopy. Neutrophilswereallowed tosedimentfor 15 min at 1 g onto monolayersofepithelialcellscontainedin 35-mm culture dishes.AfteraddingPMA to onesetof cells andincubatingfor 30min,

anequalvolumeof 4%glutaraldehydein0.2Mcacodylatebuffer,pH 7.4, was added.After3 hat4°C,the monolayers were gently washed

twice with0.1 Mcacodylatebuffer, pH 7.4,andpostfixed with2%osmium

tetroxide for 1 h at4°C. The cells weredehydrated while stillin the culturedisheswithgradedalcohol steps andinfiltrated with100%

alcohol/

eponmixtures.Thesampleswereembedded in 1-mm-thick epon and thensnappedfree from thetissueculturedishes.Thinsectionswerecut,

stained with uranylacetateand leadcitrate,and observed inaPhilips

400 transmission electronmicroscope (Philips Electronic Instruments,

Inc.,Mahwah, NJ).

Assays.Neutrophil degranulationwasmeasuredbymonitoring

ly-sozyme(27)andmyeloperoxidase (28)release.

Neutrophils

were

sus-pendedinD-PBSwithorwithoutvariouspotential inhibitorsand stim-ulatedwith PMA. Thesame amountofPMA per

neutrophil

wasused as inthecytotoxicity experiments. After30 min(lysozyme)or4h (my-eloperoxidase)incubationat37°C,the cells werecentrifugedat8,000g for 2 min and the enzyme activities of the supernatantwasmeasured. The amount ofsuperoxidereleased in 30minfrom stimulatedneutrophils

was measured by the method of McCord and Fridovich(29).Hydrogen

peroxidecontained in medium wasmeasuredbythemethod of Ruch

etal.(30).Elastaseactivityofazurophilic granuleswasmeasuredusing

orcein-elastin as substrate and standardizedagainstporcine pancreatic

elastase(31).Nonspecificproteaseactivitywasmeasured

against

Azocoll

usingtheprocedure outlined in the Calbiochem manual(La Jolla,CA).

Proteinconcentration was measuredbythe methodof Lowryet al.(32) usingbovine serum albuminasthestandard.

Statisticalanalysis. For agiven dayonwhich

cytotoxicity

assays

butusuallythree or more differentdays.The amount of

5"Cr

released

from epithelial cells exposed to agiven number of neutrophils varied morefromday to daythan betweenreplicatesonthe sameday.To pool resultsfrom experiments performed on separatedays, the datafor a given condition was expressed as the percentage of specific"Crrelease relative to thepositive control on that day(epithelialcellsexposed to

PMA-stimulated neutrophilswithoutinhibitors).

Analysis ofvariancewith the Newman-Kuelsmultiplerange test was used todeterminethe statistical ificance ofdifferencesbetween means ofvariousgroups(33).Ifgroups werecomparedonlywith asinglecontrol

condition,thenDunnett's test wasemployed.A0.05 orless probability of type I error wasconsidered statisticallysignificant.

Results

When monolayers of

"Cr-labeled

epithelial cells were exposed toPMA-stimulated neutrophils, there was increased extracellular release

of

"Cr.

The amount

of

"Cr

released

increased in

a

time-dependent

fashion

approaching

a

plateau

at

8-10

h

(Fig.

1).

Epithelial

cells

exposed

to

medium alone spontaneously

released

22.4±4.9% (mean±SD,

n =

15) of

their

initially

incorporated

"Cr

after 10hofincubation. After 10hof exposure to media alone, the

monolayers appeared

normal

when visualized

by

phase-contrast

microscopy, and 97.2±0.8% (SD, n = 8) of the

cells

were

viable by

trypan

blue exclusion criterion. The

addition of 1% fetal bovine serum to the media did not decrease the

background

release of

"Cr

(P

>

0.5,

n=8).

The percentage

ofspecific

"Cr

released

(defined in

Methods) from epithelial cells exposed to 5 X

10'

unstimulated

neutrophils or to 25 ng/ml PMA without

neutrophils

was notincreased (P

>

0.5)

overthat of cells

exposed

to

D-PBS alone (Table I).

The amount

of

"Cr

released

from epithelial cells exposed

to

PMA-stimulated neutrophils

was

dependent

upon

the number

of

neu-trophils

added to the

wells

(Fig.

2).

As

another method

todetect

cytotoxicity,

wemeasured re-lease

of

radioactivity from epithelial cells which had been

pre-viously incubated in

['H]leucine-containing

media. We

found

thatthe pattern

of

'H

released from epithelial cells exposed

to

stimulated

neutrophils

was

similar

to

the

pattern

of

"Cr

release

(Fig. 3).

Although

the

majority of

our

studies were performed using

PMA as the neutrophil activator, we also measured the cytotoxic

effect of neutrophils

stimulated

by

other

agents (Table I). In

TableI

Dependence of Cytotoxicity

on

Neutrophil

Stimulus

%specific

Additionstoepithelialcells n* "Cr releaset P§

PMN"1

12 3.0±0.9

NS¶

PMA(25 ng/ml) 12 2.6±1.1 NS

+PMN 24 44.6±3.2 <0.01

A23187** (1 X 10-7 M) 8 0.3±1.6 NS

+PMN 7 15.2±2.2 <0.01

Serum-treated zymosan (1 mg/ml) 12 0.3±1.2 NS

+ PMN 12 18.4±1.7 <0.01

FMLP(1 X

10-1

M) 4 2.5±0.9 NS

+PMN 4 -1.0±1.4 NS

* Number of determinations.

t Resultsare

expressed

asthepercentageof

specific

5"Cr

released

(mean±SE)frommonolayersof 3.2 X104epithelialcells after 10 h of

incubation.

§Probability that the percent specific

5"Cr

released is different from that of epithelial cells exposed to D-PBS alone.

115X

101

neutrophils (PMN)wereaddedtoeach well and allowedto

sedimentontotheepithelial monolayersfor 15 min before the stimu-lus wasadded.

¶ No statisticallysignificant difference. ** The calciumionophoreA23187.

particular, wefound that the calcium ionophore A23187 and serum-treated zymosan were abletoinduce neutrophilsto cause

increased

51Cr release from epithelial cells. The

chemotactic

peptideFMLP didnotinduce increased

5"Cr

releaseatup to 1

X 10-5M concentration.

Inanefforttodeterminewhether neutrophil-derivedoxygen

metabolites were involved in the injury process,we added

su-peroxide dismutase (310 U/ml) and catalase

(9,424 U/ml)

to

the tissue culture wells atthe time of additionofneutrophils. Superoxide dismutase, which rapidlyconvertssuperoxideto hy-drogen peroxide and

02,

causednoreduction (P>0.5) in the

percentageof specific

5"Cr

release (Table II). Catalase, which convertshydrogen peroxide to oxygen and water, caused a 26%

50-LU

J U) 4

LU

U

X

a:

I D

U II

4 U)

UL 4 u J LU LU

0. w U)_In

z

-u o U 0 lU I 0 U

40+

30+

20+

10+

o0

TIME (hours)

Figure 1. Kinetics of

"Cr

release from injured epithelial cells. Epithe-lial cells (3.2X _{104) as monolayers in wells of 96-well plates were}

ex-posed to 5X

_10'

neutrophilsstimulatedwith 25ng/mlPMA.

Epithe-lialcellcytotoxicity was measured by the percentage of originally bound

"Cr

released into the medium. The data is compiled from three experiments and expressed as the mean±SE.

-NONE 4 5

PMN PER WELL (log) 6

Figure 2. Dependence of epithelial cell cytotoxicity on number of stimulatedneutrophils. Epithelialcells(3.2X 104)asmonolayers in wells of96-well plates were exposed to varying numbers of neutrophils stimulated with 25ng/mlPMA.The percentage ofspecific "Crrelease

(mean±SE)wasmeasuredafter10 hof incubation.The datais

com-piled from three experiments and expressed as the mean±SE.

cn

I--zoo

ow

z _j

uCL

Cc

a.

IC

A

TIME (hours)

504

cn

ZC]4

Uc

Dw

0.

40-

30-

20-

1

0-B

4

NONE

NONE 5

PMN PER WELL

6

(log)

Figure 3.Comparison of[3H]ieucine(- ) and

Na2["Cr]04

( ) as

indicatorsofepithelialcellkilling byPMA-stimulated(25 ng/ml)

neu-trophils(PMN). The data arefromasingle representativeexperiment with each conditionperformedinquadruplicateand expressed as the

mean±SE.(A)Kinetics of isotopereleasefromcells treated as inFig.

1.(B)Dependence ofisotopereleaseonPMNnumberasinFig.2.

reduction (P

<

0.01) in

"Cr

release. However, when the

catalytic

activity

of catalase

was

destroyed

by

carboxymethylation,

a

sta-tistically significant

(P

<

0.01) degree of

protection

remained.

There

was no

difference (P

>

0.2) in the

"Cr

released

from

epithelial

cells exposed

to

neutrophils

in the

presence

of active

or

inactive

catalase.

To

evaluate

further the role of

neutrophil-derived

oxygen

metabolites,

we

utilized

neutrophils

from

subjects

with

chronic

granulomatous disease. These

neutrophils

fail

to

release

oxygen

metabolites when

stimulated with

PMA

but

degranulate

nor-mally

(34).

Neutrophils

were

available

from

two

chronic

gran-ulomatous disease

subjects,

one

of

whom donated

neutrophils

ontwo

occasions.

PMA-stimulated

neutrophils

from

the

chronic

granulomatous disease

subjects

caused an

increase (P

<

0.001)

in

the percentage

of

specific

"Cr

released

from

epithelial

cells

compared with cells exposed

to

buffer

alone. The amount

of

5'Cr

released

from

epithelial

cells

exposed

to

chronic

granulo-matous

disease

neutrophils

was

compared with

the amount re-leased

after

concurrentexposure

of

epithelial

cells

to

stimulated

neutrophils from

normal donors (Table

II).

On two

of

three

occasions,

there was no

difference (P

>

0.5)

in

"Cr

released

from

epithelial

cells exposed to

neutrophils from

normal subjects

or

from patients with chronic

granulomatous

disease.

On one

test

day,

the percentage

of

"Cr

released

from

epithelial

cells

incubated

with

neutrophils from

a

patient with

chronic

granu-Table IL Role ofOxygen Metabolites in

PMN-inducedDamageofEpithelialCells

Additions to epithelial cells no Cytotoxicityt P§

%control

Normal PMN + PMA 100

+SOD" 12 102.3±9.8 NS¶

+CAT** 16 74.0±6.3 <0.001

+ inactive CAT 15 82.8±4.0 <0.01

CGD-laPMNt*+PMA 7 49.9±8.9 <.025

CGD-lbPMN+PMA 4 100.2±7.3 NS

CGD-2 PMN+PMA 4 107.8±9.3 NS

* Number ofdeterminations.

tThecytotoxicity caused by a given condition was measured by the percentage ofspecific

31Cr

releasedfrom 3.2 X 104 epithelial cells cul-tured asconfluentmonolayers in wells of 96-well plates. Results are expressedasmean(±SE)percentcytotoxicitycaused by agiven condi-tion comparedwithaconcurrentlyrunpositivecontrol,i.e., epithelial

cellsexposed to 5 X 105 neutrophils (PMN) stimulated with 25 ng/ml PMA.

§Probabilitythat thedifference from positivecontrol(normal PMN +PMA)isdue tochance.

11

_Superoxide

dismutase(SOD),310U/ml.

¶Nostatisticallysignificant difference frompositive control. **Catalase (CAT), 4,924 U/ml.

ifPMN(5X

10'/well)

from twopatients(1and 2) withchronic gran-ulomatousdisease(CGD). Patient I wastested on twooccasions

(a andb).

lomatous disease

(CGD-la)

was

49.9%

of

that

induced by normal

neutrophils.

The cells

from this

same

patient when tested

on

another

day caused the

sameamount

of

injury

as

cells

from

a

normal

individual. The

reason

for the

variability in the cytotoxic

effect of

chronic

granulomatous

disease

neutrophils relative

to

that

of normal

neutrophils

isunknown. It may be dueto

tem-poral

variation

in the

chronic

granulomatous

disease

patient's

neutrophils

or to

differences in

neutrophils

among

normal

vol-unteer

donors.

No difference in the clinical condition of the

chronic

granulomatous

disease

patient

wasevidentonthetwo

test

days.

We

also tested the

ability

of

neutrophil cytoplasts

to

injure

epithelial

cells.

It

has been

previously

demonstrated that

neu-trophil cytoplasts

generate

normal

amounts

of

oxygen

metab-olites when stimulated with

PMAbut because

they

are

greatly

depleted

of

granules, they

release

negligible quantities

of granule

constitutents

(26).

When

epithelial

cells were

exposed

to 2.5 X

106

neutrophil cytoplasts

stimulated with 25

ng/ml

PMA,

no

statistically

significant

increase in the percentage of

specific "Cr

release

( 1.1±3.2

[mean±SE],

n=

8)

was seen.To

verify

thatour

stimulated

neutrophil cytoplasts

were

generating

oxygen

metab-olites,

we measured the

superoxide

dismutase-inhibitable

re-duction of

ferricytochrome

C in wells

containing

2.5 X 106

cy-toplasts stimulated with

PMA. In the absence of

PMA,

the

cy-toplasts

released 9.5 nmol

superoxide/106

cellsover 30 min

_(n

= 4). In the presence

of

25

ng/ml PMA,

theamountof

superoxide

releasedincreasedto25.7

nmol/106

cells

over30min.

Based on the

combination

of

results

using superoxide

dis-mutase,

catalase,

neutrophil cytoplasts,

and

chronic

granulo-matous

disease

neutrophils,

weconclude that inoursystem,a

is

independent ofneutrophil-generated

oxygen

metabolites.

Be-cause reports

by other

investigators

have demonstrated that stimulated

neutrophils

can killisolated endothelial cells via ox-idant mechanisms (35-38), we hypothesized that our alveolar epithelial cells might be more resistant to oxidant

challenge.

To test

this

possibility,

we exposed

bovine pulmonary

artery en-dothelial cells and rat alveolar epithelial cells to

increasing

con-centrations

of

hydrogen

peroxide and measured the cytotoxic effect using our standard assay. Using the constructed dose-re-sponsecurves, wefound that the concentration of hydrogen per-oxide necessary to cause a 50% specific

5"Cr

release after 10 h ofincubation was 14 times higher for epithelial cells (0.56 mM) than for

endothelial cells (0.04

mM).

Havingdemonstrated that epithelial cells were more resistant to

hydrogen peroxide-induced killing compared

with endothelial cells, we questioned whether this resistance was associated with agreater

ability

of

epithelial

cells to

metabolize hydrogen

per-oxide.

To answer

this

question,

we

added

200

_,d

of 50

gM

hy-drogen peroxide in D-PBS to wells of a 96-well plate containing

monolayers

of alveolar epithelial or

endothelial

cells. The hy-drogen peroxide concentrations ofthe media were sampledover

time and

decay

curves

constructed (Fig.

4). The

half-life of

hy-drogen

peroxide

in

the

media

of wells

containing epithelial

cells

(48.9 min)

wasshorter than

that

of

the

endothelial

cells(108.2 min),

indicating

that

epithelial

cells were able to metabolize

hy-drogen

peroxide

at a greater rate

than endothelial

cells. To evaluate the role

of neutrophil-derived

proteolytic en-zymes,

antiproteases

were added to the media. In

particular,

soybeantrypsin inhibitor,

a1-antitrypsin,

aprotinin, PMSF, and

MeoSuc-ala-ala-pro-val-CH2Cl

were

individually

added.Soybean

trypsin

inhibitor,

a,-antitrypsin,

and

aprotinin

reduced

epithelial

cell

killing

as measured by

51Cr

release (Table III).

PMSF

ac-centuate

at

1.0

m

vestigate4

cells at c

the

incul

1

z:LL

0

I-i

z Lii U

z

0

u

0

CI

Figure 4.

andepith addedto

tained in added20 idewasn

surement

Table

III. Role

of

Proteasesand Serum in PMNDamage

of Epithelial

Cells

Additionstoepithelialcells n* Cytotoxicityt P§

%control

PMN + PMA 100

+ a,-antitrypsin 0.1 mg/ml 16 90.4±6.7 NS"

0.3mg/ml 8 93.8±9.1 NS 1.0mg/ml 12 55.6±4.1 <0.05

+SBTI¶ 0.1 mg/ml 12 88.9±6.3 <0.05

0.3mg/ml 8 86.2±5.1 <0.05 1.0mg/ml 12 45.8±4.0 <0.01

+aprotinin 0.1mg/ml 12 95.4±3.7 NS

0.3mg/ml 12 86.0±3.0 NS

1.0mg/ml 16 47.2±6.6 <0.05

+PMSF** 0.1 mM 20 128.2±8.8 <0.05

0.3 mM 16 118.7±7.7 NS

1.0 mM 20 76.4±6.9 <0.05

+

_FBS#f

1%(vol/vol) 16 69.2±8.2 <0.01 10% (vol/vol) 16 41.8±7.5 <0.01

* Number of determinations.

$Cytotoxicity was measured and data expressed as in Table II. § Probability that thedifferencefrompositivecontrolisdue tochance.

1Nostatistically significant difference from positive control (PMN +PMA).

¶ Soybean trypsin inhibitor (SBTI). **Phenylmethylsulfonyl fluoride (PMSF). $$ Fetal bovineserum(FBS).

i neutrophil-induced injury at 0.1 mM but protected was noted. As a control condition, 1 mg/ml bovine serum al-M. The reason

for this biphasic

response was not in- bumin was added and no reduction in 51Cr release was seen d.

MeoSuc-ala-ala-pro-val-CH2Cl

was toxic to

epithelial

(data not shown). Fetal bovine serum which contains several

oncentrations above 0.01

mM.When

it

wasadded to antiproteases also inhibited epithelial cell killing (Table III). bation media at

0.01

mM, no

epithelial

cell

protection

To determine whether the protective effect of aI-antitrypsin

was

dependent

upon its

ability

to

inhibit

_{neutrophil elastase,}

aI-antitrypsin

was

oxidized with

N-chlorosuccinimide,

a process

which

destroys

its elastase-inhibitory activity

(14).

We then added

either native

or

oxidized

aI-antitrypsin

to

the

cytotoxicity

assay ENDOTHELIAL

system. We found that the protective effect of

1

mg/ml native

Lal-antitrypsin

(65.1±5.5

[SE]

percent

inhibition of cytotoxicity)

was not

decreased

(P

>

0.5,

n =

₈₎

when

it

was

_{replaced by}

oxidized

a1-antitrypsin

(62.2±4.9

percent

inhibition).

\

PITHELIAL

Experiments

werealso

performed

to test

the

effect of

com-EPITHELAL

bining antioxidants and antiproteases. In separate experiments, we found that the reduction in killing associated with 1.0 mg/ mlsoybean

trypsin

inhibitor

(29.9%

inhibition)

was not

further

reduced (P > 0.5, n = _{4) when}

_0.1

_{mg/mlcatalase was added}

0-

-

\°(25.9%

inhibition). Similarly, the reduction in cytotoxicity

as-sociated

with

1

mg/ml

a,-antitrypsin

(36.4%

inhibition)

wasnot

0 ₁₀ 20 30 40 50 60 _{further reduced}

_(P>

_{0.5, n =4) when}

_0.1

_mg/ml

_{catalase was}

added(32.0%

inhibition).

TIME

(MI N)

Another class of

neutrophil products

which may

potentially

cause

epithelial

cell

injury

arethe granule

cationic proteins

(39-Kinetics

_mnof

Hydrogen peroxide

metabolism by

endothelias

43).

Toevaluate

this

possibility, epithelial

cells were exposed to

elial

monolayers. Hydrogen

peroxide

(50

MuM)

in

D-PBS

was

stmltdnurpisnthpeeceotwpoyinhpan

confluentmonolayers of endothelial or epithelial cells con- stimulated

neutrophils

in the presence oftwo

polyanions, heparin

96-wellplates. After various times, 50

Ml

oftheoriginally and dextran sulfate. It has been previously reported that

poly-0

_M1

wasremoved and the concentration of hydrogen perox- anions inhibit cationic

protein-dependent

microbicidal

_activity

neasured. Each point represents the mean of triplicate mea-

(39, 40, 42).

Both

heparin

and dextran

sulfate limited

the amount

s.

_{of epithelial}

_cell

_lysis

_{(Table IV). To determine whether}

Table IV.Inhibition ofPMN-induced DamageofEpithelial Cells by Polyanions

Additions toepithelial cells n* Cytotoxicityt P§ %control

PMN+PMA 100

+heparin 0.02 U/ml 8 84.6±5.8 <0.01 0.2 U/ml 8 83.1±5.9 <0.01

2.0U/ml 16 36.0±4.1 <0.01 + dextransulfate 0.001 mg/ml 8 63.9±8.0 <0.01 0.01 mg/ml 12 35.8±4.2 <0.01 0.1 mg/ml 12 31.6±6.5 <0.01 + dextran 0.001mg/ml 12 98.2±6.6

NS"

0.01 mg/ml 12 86.4±7.7 NS

0.1mg/ml 11 81.5±8.5 NS

* Number of determinations.

tCytotoxicitymeasured andresults expressedasin Table II. §Probabilitythat thedifference from

positive

control (PMN+PMA)

isdue tochance.

Nostatistically significant differencefrompositivecontrol.

tion by dextran sulfate

was

dependent

upon

the

electrical charge

of the molecule,

cytotoxicity

assays were

performed using

un-charged

dextran. We

found that the

neutrophil-induced

cyto-toxicity

was not

diminished

in the

presence

of

uncharged dextran.

Because

the above studies

suggested that neutrophil granule

contents

might be

causing

the

injury,

we

tested

the cytotoxic

effect of

various

neutrophil fractions.

To

determine whether

neutrophil

lysates

contained

cytotoxic

factors, epithelial

cells

were

exposed

to

neutrophils which had been

disrupted by

son-ication for

30

s.

No

significant

increased

"Cr

release

was seen

(Table V). Because

many

of

the

neutrophil

proteases

and

cationic

proteins

are

contained

in

azurophilic granules,

we

exposed

monolayers

of

epithelial

cells

to

the

contents

of

purified

azur-ophilic granules (Table V).

When

epithelial

cells

were

incubated

with

up to

0.1 mg/ml

azurophilic

granule

protein,

no

significant

cellular

lysis

was

detected.

Similarly,

when

epithelial

cells

were

exposed

to

0.1 mg/ml

specific

granule

protein,

no

increased

cy-totoxicity

was

observed.

It

is

possible

that

neutrophil azurophilic

granules

do

contain

the

agents

responsible for

the

observed

ep-ithelial

cell

injury,

but

the

agents are

inactivated

during

prepa-ration.

To

partially

address

this

possibility,

we

measured the

ability

ofthe

isolated

azurophilic granules

to

digest

two

different

substrates.

Using

porcine pancreatic

elastase

asa

standard

for

comparison,

we

found the

presence

of

proteolytic

activity

in

our

preparations

of

azurophilic granules.

The

specific activity

was

2.1±0.2

(SD) U/mg

granule

protein

toward

elastin-orcein

and

9.4±2.3 U/mg granule

protein

toward

Azocoll (a substrate

that

detects

a

broad

range

of

protease

activities).

We

also used

purified human neutrophil

elastase

in

our

cy-totoxicity

assay. We

found

that

elastase in

the amount

contained

in

2 X

106

neutrophils

(44) failed

to

increase

the percentage

of

specific "Cr

release

(Table V).

In

addition

to

the

proteases

con-tained within

the

granular fractions of

neutrophils,

protease

ac-tivity has

also been

found associated with neutrophil

plasma

membranes

(45).

We were not

able

to

implicate

plasma

mem-brane-associated

proteases as a cause

of

increased

"Cr

release

in

that

lysed

neutrophils

(Table V)

and

neutrophil

cytoplasts

did

not

injure

the cells.

Table V Cytotoxic Effect ofPMN-derived Products

%specific Additions toepithelialcells n* 5'Crreleaset P§

Lysed PMN

5X 10' PMN1I 12 1.6±i.9 NST

2 X10 PMN 12 1.2±1.4 NS

PMNazurophilic granule contents

2

Ag,

8 X 105

PMNII

20 -1.5±1.1 NS

20

_;g,

8 X 106PMN 24 -1.3±1.3 NS PMNspecific granule contents

10 g, 8X 106 PMNI 4 0.6±0.8 NS

20 ug, 1.6 X 107PMN 4 -0.2±0.8 NS

Supernatants of PMA-stimulatedPMN

5X 10 PMN** 20 4.4±1.1 <0.01

2X 106PMN 12 3.6±1.4 <0.05 Humanneutrophilelastase

1.7

_;&g,

2 X 106PMNII 8 -0.8±1.1 NS

* Number ofdeterminations.

t Resultsare

expressed

asthe

percentage

of

specific

"Cr released

(mean±SE)frommonolayers of3.2XI0O epithelialcellsafter10 hof incubation.

§Probabilitythatthe percentage ofspecific

"Cr

releasedisdifferent from thatofepithelialcellsexposedtoD-PBSalone.

11Numberof neutrophils (PMN)that wouldcontain thegiven weight of granule

protein

thatwasaddedtoeach well.

¶Nostatistically significant difference.

** Numberof PMN thatwerestimulatedtogenerate supernatant that wasaddedtoeach well.

To

determinewhether

a

stable

cytotoxic

factor

was

released

into

the media by

PMA-stimulated neutrophils, epithelial cells

were

exposed

tosupernatants

from wells in which

PMA-stim-ulated

neutrophils

were

incubated for 10 h.

A

statistically

sig-nificant,

although quantitatively small, increase in the

percentage

of

specific

"Cr

release

was

noted (Table V). The inability of

stimulated

neutrophil

supernatants to cause

substantial

cyto-toxicity

also

demonstrates that

neutrophil-induced

killing

was not

caused

by

depletion

of

a

vital

factor

from the media.

To

determine whether

a

metastable

cytotoxic

factorwasreleased

by

neutrophils

but inactivated

during

the

10-h

incubation,

neu-trophils

were

placed

into wells and after

sedimenting

for15

min

were

stimulated with

PMA.

After

an

additional 15

or

30

mi,

the

supernatants were transferredtowells

containing "Cr-labeled

epithelial

cells

and the

cytotoxicity

wasmeasured after 10 h of

incubation. No

significant

increase in

the

percentage

of

specific

"Cr

releasewasnoted

(P

>

0.2,

n=

4).

We

tested the

possibility

that

protection

caused

by

the

above-mentioned

agentswas

due

to

reductions

in

neutrophil viability.

Neutrophils

preincubated

with

"Cr

werestimulated with 25

ng/

ml PMA

in

the presence of the various

antiproteases

and

poly-anions.

None

of

the

inhibitors

caused increased extracellular

release of

5'Cr

when

tested

at

the

highest

concentrations used in the

cytotoxicity

studies

(P

>

0.4).

We

also considered

that

the

protection

associated with the

previously

mentioned

agents

might

be due to

inhibition of

neutrophil

activation.

Wefound that the

release of

superoxide (23.8 nmol/106

cellsover30

min)

from PMA-stimulated

neutrophils

was notreduced

by

the

experiments (P

>

0.05).

Similarly,

the

percentage

of lysozyme

released

after PMA stimulation (31.8% in 30

min,

n =

8)

was not

reduced by the

presence

of

the

inhibitors

(P

>

0.5). PMA,

which is

known to

be

a

relatively

poor

agonist

for

myeloper-oxidase

secretion, caused

11.9%

of

neutrophil myeloperoxidase

tobe

released after

4h

of incubation (n

=

4).

None

of

the

in-hibitors

reduced

this by

>1.5%.

During

the

cytotoxicity studies using intact

neutrophils, the

neutrophils

were

allowed

to

sediment

onto

the

epithelial

cell

monolayers

before

PMA was

added. If

PMA was not

added, few

of the

neutrophils adhered

during

an

additional 30 min of

in-cubation (Table VI).

In contrast,

if

PMA was

added,

most

of

the

neutrophils adhered

by 30

min. To

determine whether

neu-trophil adherence

was an

important

factor in the

cytotoxic

pro-cess, a

monoclonal

antibody (anti-Mol)

was

used because it

inhibits

neutrophil adhesive interactions (46). Anti-Mol does

not alter

activation of neutrophils that have been exposed

to

soluble stimuli

(19).

To

determine whether

anti-Mo 1

antibody

would

inhibit

neutrophil adherence

to

epithelial monolayers,

5"Cr-labeled

neutrophils

were

pretreated with anti-Mol

antibody

for 10 min. The

neutrophil suspension

was

then

transferred

to

a

24-well plate

containing

monolayers of

epithelial

cells.

After

15

_min,

PMA

(25 ng/ml)

was

added

and

neutrophil

adherence

was measured

after

an

additional 30

min as

described in

Meth-ods.

We

found that anti-Mol

antibody

reduced the

adherence

of neutrophils

to

epithelial

cell

monolayers (Table

VI). By

in-direct

immunofluorescence

analysis,

we

found

that

anti-Mol

antibody, which is known

to

bind

to

neutrophils,

did

not

bind

to

the

epithelial cells.

Therefore,

we

believe that the

ability

of

anti-Mo 1 antibody

to

inhibit

neutrophil

adherence

to

epithelial

cells

was

due

to

its

effect

on

neutrophils

and

not on

epithelial

cells.

To

determine whether the inhibition might be

a

nonspecific

effect of

antibody,

parallel

experiments

were

performed using

two

other

IgG2a monoclonal

antibodies,

anti-Mo5

and an

anti-Ia.

Anti-Mo5 antibody

binds

to

human

neutrophils but does

not

alter

aggregation (47)

or

activation (19). The anti-Ia

antibody

does

not

bind

to

neutrophils.

We

found that Mo5 and

anti-Ia antibodies did

not

inhibit

neutrophil adherence

to

epithelial

cell monolayers (Table VI).

To

demonstrate

the

alteration

visually

in

cell-cell contact

TableVI.PMNAdherencetoEpithelial Cell Monolayers

AdditionstoPMN n* PMNadherencet P§

%ofadded PMN

None 4 3.2±1.2

PMA 6 90.6±2.0

PMA+

anti-Mo1ll

6 50.7±7.4 <0.001

PMA+anti-Mo5 5 89.4±2.4

NS¶

PMA+anti-Ia 5 91.6±0.9 NS

*Numberofdeterminations.

f Percentage (mean±SE) of3 X

106

neutrophils(PMN)that adhered toepithelialcell monolayers in wells of 24-wellplates 30 min after

ad-ditionof 25 ng/mlPMAorcontrolbuffer.

§Probabilitythat thedifference inPMNadherence under the given

condition compared withPMAaloneis dueto chance.

11Purifiedmonoclonal

antibodies

(5

qg/ml)

were

preincubated

with PMNfor10 minasdescribed inMethods.

¶No

_{statistically}

_significant

difference.

causedby the

anti-Mo 1 antibody, neutrophils

were

allowed

to

sediment onto epithelial

cells for 15

min

inthe

presence

of either

anti-Mo 1

or

anti-Mo5 antibody.

PMA was

then added,

and after 30min

of incubation,

the cells were

fixed by gently

adding an

equal volume

of

4% glutaraldehyde

(Fig. 5).

In cultures

con-taining anti-Mo 1

antibody, the neutrophils appeared rounded

and had

limited

contact

with

the

underlying epithelial

cell monolayer. When

anti-Mo5 antibody

was

used

as a

control

con-dition, the PMA-stimulated neutrophils had

an

extensive

area

of

contact

with

the

epithelial cell monolayer.

The

effect of

anti-Mo 1 antibody

on

neutrophil-induced

ep-ithelial

cell

cytotoxicity

was

tested in

our

standard

assay.

Neu-trophils

were

pretreated

with antibody for

10 min

and then the

cellular suspension

was

added

to

96-well plates containing

5"Cr-labeled

epithelial

cell

monolayers.

After 15

min,

PMA was

gently

added, and the

percentage

of

specific 'Cr release

was

measured

after

10 h. We

found that

anti-Mo 1

antibody caused

a

dose-dependent

reduction in

5"Cr

released

from

epithelial cells (Table

VII). As control

conditions,

the

effects of anti-Mo5

and

anti-Ia

were also tested. These two

antibodies

causedonly

slight

inhi-bition

of

neutrophil-induced injury of epithelial cells. The

dose-response

experiments

were

performed using monoclonal

anti-bodies contained in

mouse

ascitic fluid. The results

were

con-firmed using purified antibodies

at one

concentration

(Table VII).

To

determine

the

critical time during which

adherence

is

important,

we exposed

epithelial

cells to

neutrophils

that

had

been

stimulated with

PMA at

various times before

being

trans-ferred

to

the

epithelial

cell-containing

wells

(Table VIII).

Com-pared with

our

usual

protocol in which

neutrophils

were

allowed

to

sediment

onto

the

monolayers before the

PMA was

added,

we

found

that

cytotoxicity

was

less if the

PMAwas

added

si-multaneously

with the

neutrophils. The longer the time interval

between

stimulation

and

transfer of

neutrophils

to

the

wells, the

less

cytotoxicity

was

noted. If 10 min

elapsed, the

amount

of

5'Cr release

was not greater

than that of unexposed

epithelial

cells.

Because

neutrophil

adherence

appeared to be

important

for

cytotoxicity,

we

questioned whether

some

ofthe protection noted

in

experiments using

inhibitors might be

due to

alterations in

adherence

rather than

the

intended mechanism. To

test

this

possibility, neutrophils

were

added

to

epithelial

cells

in

the pres-ence

of the various inhibitors.

After

allowing 15

min

for the

neutrophils

to

sediment

onto

the

epithelial

cell

monolayer,

phorbol

myristate

etate

(25

ng/ml) was added.

After

an

additional

30

min

of

incubation,

the

percentage

of neutrophil adherence

was

measured

as

outlined in

Methods. As noted

previously,

we

found

that relatively few unstimulated

neutrophils adhered

to

the

epithelial cell monolayer (Table IX).

In contrast,

the

majority

of

neutrophils

stimulated with

PMA

adhered

to

the

monolayer.

The

antiproteases soybean

trypsin inhibitor, a,-antitrypsin,

aprotinin,

and

fetal

bovine

serum

reduced the adherence

of

neu-trophils

tothe

epithelial

monolayer.

Thereduced adherence

as-sociated with

fetal bovine

serum was notdueto

species

differ-encesbetween the serum and cells because

similar reductions

in adherence were noted

with

human or rat serum

(data

not

shown).

The reduced adherence was notdue to a

nonspecific

effect of

protein

in

that

PMA-stimulated

neutrophils

adhered

normally in the presence

of bovine

serum

albumin.

The

poly-anions heparin and

dextran

sulfate were associated with

small

but

statistically insignificant

reductions

in

neutrophil

adherence

to

epithelial

cells.

*:

*6'.

. E..

*A s, . ..

*,

'He';§."'

.,_h

V,,+

oSX.

Figure5.Transmissionelectron

micro-graphs

ofPMA-stimulated

neutrophils

(N)onmonolayers ofalveolar

epithelial

cells

(E).

(A) Neutrophils

pretreated with

5

tg/ml

anti-Mo5

antibody

as acontrol condition.This

antibody

does notaffect

neutrophil

adherenceor

cytotoxicity (X

2,800). (B)

Neutrophils pretreated with5

,tg/ml

anti-Mol antibodywhich reduces

neutrophil

adherence and

cytotoxicity

to-wardepithelialcells.Vacuoles (v)in

epi-thelial cells contained lammelar bodies

whichwerewashed outduring

processing

(22)

(x

4,850).

Discussion

Neutrophil-induced killing ofalveolar epithelial cells bya

mech-anism independent of neutrophil-generatedoxygenmetabolites

hasnotbeen previously described. In this studywe measured

epithelial cell damage by monitoring the extracellular release of

5'Cr

from previously labeled cells. This technique has been

pre-viously used as a measurementof cellulardeath by many

in-vestigators, some of whom are cited forotherreasonsin this

paper(35-38, 40, 42,48-50). In this study, weconfirmedthe

presenceof cellular damage byanadditional technique. Cells

werepreincubated with

[13H]leucine

toallowaradioactivemarker

tobeincorporated into cellular macromolecules. The pattern

of extracellular leakage of 3H from epithelial cellsexposed to

stimulated

neutrophils closely approximated

that of

51Cr

sug-gesting

that bothmethodswere

measuring

loss of

epithelial

cell

integrity.

Neutrophil-mediated

injury

of

epithelial

cells could be in-duced

by

several

neutrophil

stimuli.

Although

PMAwas

asso-ciated

withthegreatestamountofepithelial cell

injury,

wefound

thatthe calcium

ionophore

A23 178and serum-treatedzymosan

werealso abletostimulate neutrophilstocauseincreased

51Cr

release from

epithelial

cells.

Thus,

theprocesses

leading

to

epi-thelial injuryarenotduetounique PMA effects butcanalso be initiated by another, unrelated soluble stimulus (A23187) and byaparticulate stimulus (serum-treated

zymosan).

Suttorp

and Simon

(49) previously reported

that mouse

peritoneal neutrophilswereabletokill monolayers of L2 cells.

.:b

*

de.

.0. AO,

i.

A

,

jk KS~~~ijl~~'

t1

41:

0

.1

i..:....

....e.

"s,.

*

i~~~~~~~~~.;

.:.

V aI I

B

O.

_.I

Table VII. Inhibition ofPMN-inducedEpithelial Cell Cytotoxicity byAnti-Mol Antibody

Additions toepithelialcells n* Cytotoxicityt P§ %control

PMN+PMA 100

+ anti-Mol

1:5,0001,

8 74.1±8.4 NS¶

1:1,000 12 49.8±11.6 <0.01 1:200 12 24.8±10.6 <0.01

5 tg/ml 4 41.0±9.9 <0.01 + anti-MoS 1:5,000 8 101.9±12.7 NS

1:1,000 12 89.2±12.9 <0.01 1:200 12 87.3±10.1 <0.01 5

_Ag/ml

4

98.1±9.7

NS +anti-Ia 1:5,000 8 98.9±7.8 NS

1:1,000 12 90.8±17.2 <0.05

1:200 12 79.1±6.7 <0.01

5 tg/ml 4 104.2±5.3 NS

* Number ofdeterminations.

tCytotoxicitymeasured and resultsexpressedasin TableII.

§Probability that thedifferencefrompositivecontrol(PMA+PMN)

is due to chance.

11 Dilution of immunemouseascitesorconcentration ofpurified

monoclonalantibody added toPMNsuspension.

I Nostatistically significant difference.

L2

cells

are

epithelial

cells

derived

from a clone of rat type II pneumocytes

(51).

It was

concluded that the

injury

was

caused

by

oxygen

metabolites because

catalase,

ascavenger of

hydrogen

peroxide,

was

able

toprotect

the

epithelial

cells.

Heat-inactivated

catalase

was not

effective.

In our

experiments,

we

found

that catalase had

only

weak

protective

effect

which

persisted

when the

catalase

was

inactivated by carboxymethylation.

We

did

not use

heat-inactivated catalase because

we

found that catalase

pre-cipitated

upon

heating. The killing of epithelial

cells

by

neutro-TableVIII.

Dependence

ofCytotoxicity on Time between PMN

Activationand Their

Transfer

ontoEpithelialCell Monolayers

Time* %specific5'Crrelease4 P§

min

-15"1 34.5±2.9

0 28.5±3.0 <0.05

2 14.5±3.3 <0.001

5 16.1±1.4 <0.001

10 1.3±1.0 <0.001

30 2.9±0.9 <0.001

* S x

_10'

neutrophils (PMN) were stimulated with 25 ng/ml PMA in

testtubes andtransferred to wells containing 3.2 X 104 epithelial cells aftervaryingtimes.

t

Results are expressed as the percentage of specific

5"Cr

released

(mean±SE,n=_{4)from epithelial cells after 10 h of exposure to}

stim-ulatedneutrophils.

§ Probability that the difference from"-15min" is due to chance. 11In accordance with our standard cytotoxicity assay protocol,

unstim-ulatedneutrophils were added to wells andallowed to sediment for 15 min, after which PMA was added.

TableIX.Neutrophil AdherencetoEpithelial CellMonolayers

AdditionstoPMN n* PMNadherencet P§

%of total added

None 28 9.5±0.8

PMA 83 90.2±3.7

+catalase(0.1 mg/ml) 16 78.3±5.6 <0.01

+SBTI"l(1.0 mg/ml) 14 72.3±6.8 <0.01

+

a,-antitrypsin

(1.0mg/ml) 15 70.0±6.0 <0.01

+aprotinin (1.0mg/ml) 14 71.4±4.7 <0.01

+PMSF¶(I mM) 16 84.2±2.4 NS**

+fetal bovineserum

(10%,vol/vol) 16 31.8±4.8 <0.01

+bovineserumalbumin

(1 mg/ml) 4 95.0±5.3 NS

+heparin (2 U/ml) 16 85.9±2.0 NS

+dextran sulfate(0.01 mg/ml) 16 81.5±1.8 NS

+dextran(0.01 mg/ml) 16 88.6±2.0 NS

* Number ofdeterminations.

fSuspensionsof 5X I05neutrophils (PMN) containingthedesignated

inhibitorswereaddedtoconfluentmonolayersofepithelialcells

con-tained in wells ofa96-wellplate.After 15

min,

PMA(25ng/ml)was

added.After 30 minof incubation,PMNadherencewasmeasuredas

indicated in Methods.

§Probabilitythat thedifferencefrom PMN+PMAis duetochance. 1"Soybeantrypsininhibitor(SBTI).

IPhenylmethylsulfonyl fluoride (PMSF).

** Nostatistically significant difference.

phils from chronic granulomatous disease patients provided

ad-ditional strong

evidence

that

neutrophil-produced

oxygen me-tabolites were not

required

for the

cytotoxic

effect in our system.

Neutrophil cytoplasts, which release

oxygen

metabolites

when stimulated, were also ineffective inkilling epithelial cells. One

possible explanation for the differences between

our

results

and

those of

Suttorp and Simon is that the

sources

of effector and

target

cells

were

different.

We used

primary

cultures of rat

al-veolar epithelial cells and human peripheral

blood

neutrophils

as

opposed

to

their

use

of

L2

cells and elicited

mouse

peritoneal

neutrophils. Although L2 cells

are

morphologically similar

to type II

epithelial cells, they differ significantly

in

biochemical

properties (52).

Other

investigators

have demonstrated that in

some

exper-imental

systems,

neutrophils

can

kill

cultured

endothelial

cells

using

oxygen

radicals (35-38).

Because we

found

that

neutrophil-generated oxidants

did not

play

a

major role

in

the

killing of

alveolar

epithelial cells,

it would appear

that the

epithelial

cells weremore

resistant

to exogenous

oxidants.

We

confirmed

this

by demonstrating

thatthe concentration

of

hydrogen peroxide

necessary to kill

50%

of the target cells was 14 times

higher

for

epithelial

than for

bovine pulmonary

artery

endothelial

cells. The alveolar

epithelial

cells also

metabolized exogenously added

hydrogen

peroxide

at a

faster

rate than

did

the

endothelial

cells.

Varani

and

colleagues (38)

have

studied

these

endothelial

cells anddemonstrated that oxygen

metabolites

were

involved

in the process

of

neutrophil-induced injury.

Ayars and colleagues

(50)

alsostudied the effects of

PMA-stimulated neutrophils

on alveolar

epithelial

cells. Incontrast

to our

results, they did

not

find

neutrophil-induced

killing

as