Gut 1997; 40: 175-181

Superoxide anions produced by inflammatory cellsplay an important part in the pathogenesis of acidand pepsin induced oesophagitis in rabbits

M J Naya, D Pereboom, J Ortego, J 0 Alda, A Lanas

AbstractBackground-Reactive oxygen metab-olites have been associated with gastro-intestinal injury.Objective-To investigate whether muco-sal reactive oxygen metabolites areinvolved in acid and pepsin induced oeso-phagitis, and if so, which specificmetabolites.Methods-The effects of free radicalscavengers and the anti-inflammatorydrug ketotifen on rabbit oesophagitisinduced by acidified pepsin were studied.Isolated oesophageal cells were obtainedbefore and after oesophageal injury andthe generation of superoxide anion andhydrogen peroxide was analysed by flowcytometry. The presence ofinflammatorycelis was determined by indirect immuno-fluorescence with a mouse antirabbitCDllb antibody.Results-Of the free radical scavengerstested, superoxide dismutase, which reactswith the superoxide anion, significantlyreduced oesophagitis, whereas catalase,which reacts with hydrogen peroxide, hadonly a mild effect and dimethylsulphoxidehad no effect. Ketotifen significantlyreduced the inflammation and alsoprevented the induction of oesophagitis.Isolated cells obtained from the oeso-phageal mucosa after acidified pepsinexposure generated increased amounts ofsuperoxide anions, which were mainlyproduced by CDllb positive cells.Conclusion-Reactive oxygen metab-olites, especially superoxide anion, pro-duced by inflammatory cells play asignificant part in the genesis of oeso-phagitis induced by acid and pepsin inrabbits and might be a target for futuremedical therapy.(Gut 1997; 40: 175-181)

Keywords: oesophagitis, superoxide anion, free radical,inflammation, acid, pepsin.

Reflux oesophagitis results from contact of theoesophageal epithelium with gastric juice,which contains hydrochloric acid and pepsin.'The intrinsic mechanisms of acid and pepsininduced oesophageal mucosal damage are notwell known. Recent studies have shown thatacid damages oesophageal epithelial cells byinducing cell oedema, cell acidification, andnecrosis2 3 after affecting the lipid bilayers ofthe apical cell membranes, their intercellularjunction structures, and overwhelming the

ability of the acid extruding membrane ex-changers and the activity of the Na+/K+ATPase. Other potential mechanisms ofmucosal damage have not yet been explored.There is substantial evidence that oxygen

derived free radicals play an important part inthe pathogenesis of the injury of various tissuesincluding the digestive system.4 5 The involve-ment of oxygen derived free radicals, such asthe superoxide anion, hydrogen peroxide, andhydroxyl radical, have been well established inthe pathogenesis of ischaemic injury of thegastrointestinal mucosa and in other models ofmucosal damage induced by non-steroidalanti-inflammatory drugs,6 ethanol,7 haemor-rhagic shock,8 feeding restriction stress,9platelet activating factor'0 and Helicobacterpylori. " In all these models the presence of acidand pepsin increased and potentiated themucosal damage.'2 13 However, it is not clearwhether oxygen free radicals are involved in thepathogenesis of oesophagitis induced by acidand pepsin. Potential sources of free radicalsinclude the activated inflammatory cells, thehypoxanthine-xanthine oxidase system, thedisrupted mitochondrial electron transportsystem, the metabolism of arachidonate via thelipoxygenase pathway, and vascular endothelialcells.4 From a theoretical point of view, severalof them could be involved in the pathogenesisof mucosal damage, including the lipoxygenasepathway and the activated inflammatory cells.Oesophageal mucosal cells metabolise ara-chidonic acid via both the cyclo-oxygenase andlipoxygenase pathways'4 '5 and the presence ofinflammatory cell infiltrates in the oesophagealmucosa, lamina propria, and muscularismucosae, usually accompanies epithelialerosions and ulceration in reflux oesophagitis.'The objective of this investigation was,

therefore, to examine the role of inflammatorycells and the superoxide anion and hydrogenperoxide generation in the mechanisms ofoesophageal injury induced by acid and pepsinand to determine whether scavenging radicalsand the anti-inflammatory drug ketotifenprotects against this damage. Finally we soughtto determine the cell source of the free radicalrelease.

MethodsThis investigation comprised both in vivo andin vitro studies.

IN VIVO STUDIESAll animal studies were carried out at theService of Biomedicine and Biomaterials of the

GastroenterologyServicesM J NayaA Lanas

Pathology ServicesJ Ortego

Department ofPhysiology, UniversityofZaragoza andHospital ClinicoUniversitario,Unidad Mixta deInvestigaci6n,Zaragoza, SpainD PereboomJ 0 AldaCorrespondence to:Dr Angel Lanas,Servicio de AparatoDigestivo, Hospital ClinicoUniversitario, 50009Zaragoza, Spain.Accepted for publication20 September 1996

175

on October 10, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.175 on 1 F

ebruary 1997. Dow

nloaded from

Naya, Pereboom, Ortego, Alda, Lanas

University of Zaragoza, officially inscribed as a"Research Establishment" for the adequatehusbandry and use of all research animalsunder the Good laboratory practices norms. NewZealand white rabbits weighing 2-5-3-5 kgeach were studied. The animals wereanaesthetised with intramuscular ketamineKCI (75 mg/kg) and intraperitoneal 20%urethane (40 mg/kg) before all studies.

Experimental modelThe oesophagitis model is a modification ofone described by Lillemoe et al, 6 which hasbeen described elsewhere.'7 In brief, aftersedation, the rabbits underwent laparotomyand neck dissection. The oesophagus wascannulated in the neck at the pharyngo-oesophageal junction and in the abdomen atthe gastro-oesophageal junction with plastictubing (internal diameter 2 mm) that wassecured in place with umbilical ligatures. Theoesophagus was then perfused with 50 ml ofthe various damaging solutions at a flow rateof 10 ml/min via a recirculatin system using aperistaltic pump (Microtube Pump, MP-3.Eyela, Tokyo Rikakikai, Japan). The tempera-ture of the perfusate was maintained at 37°Cby a thermoregulator (P-Selecta, Barcelona,Spain). To induce oesophageal damage,acidified pepsin (normal saline acidified to pH2-0 with 1-0 N HC1+2000 units/ml pepsin;Sigma Chemical Co, St Louis, MO, USA) wasperfused for 80 minutes in each animal(exposed period) via a recirculating system. Toevaluate ion flux rates after damage, theexposure period was followed by another 40minute period in which acidified saline (normalsaline acidified to pH 2 0 with 1'0 N HCI) wasperfused (flux period). In each experiment afive minute washout period with normal salinewas performed between the two differentsolutions perfused.

Aliquots of the flux solution were taken fromthe reservoir (50 ml) at the beginning and afterthe perfusion period for the later analysis ofpH, K+, and haemoglobin (see below). Afterthe completion of the flux period, each animalwas killed with an intracardiac bolus of pento-barbitone and the oesophagus removed andopened longitudinally for pathological exam-ination.

Experimental protocolSix different groups of six to eight animals eachwere included in this part of the study. In allof them the same protocol of induction ofmucosal damage (described earlier) wasperformed. However, different substances weregiven 10 minutes before and during the ex-posure period, when an intraluminal acidifiedpepsin solution was perfused. Unless otherwisespecified, all drugs were given intravenously.These substances were 1 mg/kg dimethyl-sulphoxide (DMSO); 1 mg/kg intraperitonealdimethylsulphoxide (one bolus); 90 000 units/kg catalase; 6 mg/kg superoxide dismutase(SOD); 0-05 mg/kg ketotifen; and saline asplacebo. All substances were finally dissolved

and injected in 50 ml saline. Catalase andDMSO were obtained from Sigma ChemicalCo (St Louis, MO, USA) and SOD(Orgoteina-Ontosein) from Tedec-Zambaletti(Madrid, Spain). The SOD used is a liverbovine metaloprotein which has been widelyused to evaluate the role of superoxide anionin different experimental models.7 1218 Keto-tifen (Zasten) was obtained form Sandoz-Pharma (Barcelona, Spain).

Indicators ofdamageMacroscopic and microscopical changes -Immediatley after removal, the oesophaus wasexamined for gross changes, fixed in 10%formalin, and photographed. Histologicalexamination was performed on sections takenfrom the proximal, central, and distal portionsof the oesophagus and stained with haema-toxylin and eosin. The extent of both macro-scopic and microscopical mucosal damage wasgraded by two independent observers unawareof the treatments given. The macroscopic andmicroscopical oesophagitis indices were de-termined following previously describedcriteria.'7 19 20 Macroscopic: O=normal appear-ance; 1=hyperaemia involving >50% of thetissue; 2=non-confluent mucosal or sub-mucosal haemorrhage; 3=confluent intramuralhaemorrhages or erosions. Microscopical:O=normal oesophagus; 1=submucosal oedemaor separation of epithelial layers or vascularcongestion alone or in combination; 2focalareas of intramural haemorrhage or partialepithelial loss or inflammation alone or incombination; 3=large areas of haemorrhageand/or complete epithelial desquamation. Theseverity of infiltration by inflammatory cellswas assessed subjectively from the followingscale: O=no infiltration; 1 =mild infiltration;2=pronounced infiltration.2'Assessment of mucosal barnier function - As

previously described,'7 mucosal permeabilitywas assessed by the calculation of fluxes of H+(,ueq/h), K+ (,ueq/h). H+ concentrations weredetermined by measuring the pH (Crison,micropH 2001, Barcelona, Spain) and K+ con-centrations were assayed by a flame photo-meter (Eppendorf Genatebau, Netherler HinzGMBH, Hamburg, Germany). Mucosalbleeding was quantified as the total amount ofhaemoglobin shed into the perfusate in boththe exposure and flux periods by a colorimetricmethod (Quimica Clinica Aplicada SATerragona, Spain).

IN VITRO STUDIES

Cell isolationIsolated cells from the oesophageal mucosawere obtained according to a procedure pre-viously described22 with small modifications. Inbrief, the oesophagus from overnight fastedNew Zealand white rabbits (3-3 5 kg) wasremoved after a lethal intravenous injection of3 ml sodium pentobarbitone (50 mg/kg). Themuscle layer was stripped from the submucosawhile tissues were bathed in a 4°C Ringer's

176

on October 10, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.175 on 1 F

ebruary 1997. Dow

nloaded from

Superoxide anion and oesophagitis

solution previously gassed with 5% CO2 and95% 02. The standard Ringer solution con-

tained 82-2 mM NaCl, 4 0 mM KCI, 18 mMCaCl2, 0-8 mM MgCl2, 0-8 mM NaSO4, 17-8mM NaHC03, 08 mM NaH2PO4, 115 mMglucose, and 0-2% bovine serum albumin. Themedium was equilibrated with 95% 02-5%CO2 and pH was adjusted to 7-4. The oeso-phagus was then transferred to a low calciumRinger solution (Ca++ low medium contained01 mM CaCl2 and no MgCl2), containing 1mg/ml trypsin (Sigma Chemical Co) and 0 I%collagenase (Worthington, New Jersey, NJ,USA; digestion medium). After 30 minutes,the mucosa was cut into small pieces andtransferred to a new digestion medium in a

shaking water bath at 37°C for another 30minute period. The digestion medium was

continuously oxygenated with 95%0 2 and 5%CO2. During this period the tissue was

subjected to repetitive pipetting for severalminutes. Finally, the tissue fragments and cellswere collected by centrifugation and re-

suspended in Ringer's solution allowing thetissue fragments to settle and leaving isolatedmucosal cells suspended in solution. Thissuspension was collected and then filteredthrough 100 ,um Teflon mesh. A small aliquotof cells was removed for a cell count. Cells were>90% viable as determined by flow cytometryusing the fluorochromes rhodamine'23 andpropidium iodide to measure live and deadcells respectively.23

Cytometric determinationsThe cellular content of superoxide anion andhydrogen peroxide, cell mortality, and popu-

lation studies were carried out by flow cyto-metry with an EPICS Elite (Coulter, Hialeah,FL, USA) equipped with an Argon Laser (ILT550, Salt Lake City, UT, USA). All deter-minations were carried out with 10 000 cells.

Intracellular superoxide anion - *°2- was

measured by intracellular ethidium bromideformation from hydroethidine as previouslydescribed.24 Briefly, the separated cells were

resuspended at 1 X 1 o6 cell/ml in Hank'sbalanced salt solution medium with 20 ,uMhydroethidine and incubated for 20 minutes;the transformation of ethidium bromide by*O2 was measured with a DL filter of 625 nm.

Intracellular hydrogen peroxide - Hydrogenperoxide was determined as rhodamine'23formation from the oxidation of dihydrorho-damine'23.25 Oesophageal isolated mucosalcells were resuspended at 1 X 106 cell/ml inHank's balanced salt solution and incubatedfor 20 minutes with 2 ,uM dihydrorho-damine123 after which the rhodamine123fluorescence was recorded at 510-540 nm witha band pass filter.

Determination of the cell populations of theoesophageal mucosa - The oesophageal isolatedmucosal cells were cytometrically studied bysize and granularity and two populations were

clearly separated. The histological analysis ofthese populations obtained by cytometricsorting showed that in one of them there was

a predominance of squamous cells and that in

the other area of greater granularity the cellpopulation was composed of smaller cells fromthe oesophageal basal layers. Moreover, nomast cells could be identified either in isolatedcells or in histological specimens (toluidineblue stain).The presence of monocytes, macrophages,

and neutrophils in isolated oesophagealmucosal cells was determined by indirectimmunofluorescence. The cells were firstlylabelled with a mouse antirabbit CD1 lb(Labgen, Barcelona, Spain) antibody for 45minutes at 4°C and then with a goat antimouseIg Gl-FITC (Southern Biotechnology Asso-ciates Inc, Birmingham, AL, USA) for 45minutes at 4°C. FITC fluorescence was deter-mined at 525±+ 15 nm with a band pass filter.

In vitro experimental protocolSuperoxide anion, hydrogen peroxide, and cellmortality were measured in isolated cellsobtained from the oesophageal mucosa ofrabbits with and without previous 80 minutesof oesophageal acidified pepsin perfusion. Inboth types of isolates simultaneous measure-ments of superoxide anion and CD 1 lb positivecells were performed. Simultaneous measure-ments of hydrogen peroxide in CD1 lb positivecells could not be undertaken because thefluorochromes used to determine the presenceof hydrogen peroxide and CD 1 lb receptorshave the same wavelength. The cytometricmeasurements were performed in isolatedmucosal cells as a group and individually ineach population identified by flow cytometry.

DATA ANALYSIS

Results are expressed as mean (SEM). Thedifferences between means were evaluated forsignificance (p<0 05). After logarithmic Box-Cox transformation, continuous data wereanalysed by analysis ofvariance (ANOVA) withpost hoc comparison where appropriate. Two-group comparisons were conducted withStudent's unpaired t test (two tailed). Cate-gorical data were analysed by Fisher's exacttest. Cell mortality and CD1 lb labellingstudies were analysed by Kruskal Wallis testand Mann-Whitney U test.

Results

IN VIVO STUDIES

Effects of radical scavengers on acid and pepsininduced oesophageal mucosal damageExposure of the oesophageal mucosa to acidi-fied pepsin for 80 minutes induced severe grossand microscopical mucosal damage (Fig 1).Mucosal barrier function was also severelyaffected, showing loss of hydrogen ions fromthe lumen and a positive flux of potassium andhaemoglobin into the oesophageal lumen (Figs2 and 3). The parenteral administration ofeither intraperitoneal or intravascular DMSO,a scavenger of hydroxyl radicals, during theexposure period failed to modify this mucosal

177

on October 10, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.175 on 1 F

ebruary 1997. Dow

nloaded from

Naya, Pereboom, Ortego, Alda, Lanas

MicroscopicM Macroscopic

n = 6-8~ * p <0.05

**p < 0.001C

c/io 0Control DMSO Catalase SOD Ketotifen

Figure 1: Effects of different free radical scavengers andketotifen on the oesophageal mucosal damage induced byacidified pepsin in rabbits (DMSO=dimethylsulphoxide;SOD=superoxide dismutase).

M K+ H+

n = 6-81100 p <0Q05850 **p<0.01

az 600 | * **

xi 350°

o 1001501 Ii _

400

650-Control DMS0 Catalase SOD Ketotifen

Figure 2: Effects of differentifree radical scavengers andketotifen on the function of rabbit oesophageal mucosalbarrier damaged by the intraluminal perfusion of acidifiedpepsin in rabbits (DMSO=dimethylsulphoxide;SOD=superoxide dismutase).

150

- 125-

E 100-c.0 750

o 50Ea)m 25

n = 6-8** p < 0.01

0

Control DMSO Catalase SOD Ketotifen

Figure 3: Effects of differentfree radical scavengers andketotifen on the oesophageal mucosal bleeding induced byacidified pepsin in rabbits (DMSO=dimethylsulphoxide;SOD=superoxide disniutase).

damage. However, catalase, a scavenger ofhydrogen peroxide radicals, induced a weakreduction in mucosal injury induced byacidified pepsin. All the indicators of damagewere reduced but the differences were onlystatistically significant for macroscopic damage(Figs 1-3). By contrast, treatment with SOD,a scavenger of superoxide anion radicals,greatly reduced the mucosal injury induced byacidified pepsin, and all indicators of damagewere significantly improved (Figs 1-4).

Effects of ketotifen on acid and pepsin inducedoesophageal mucosal damageThe parenteral administration of ketotifenduring the exposure period of the oesophagealmucosa to acidified pepsin also induced a

dramatic improvement of all indicators ofmucosal damage. The effect was similar to thatobserved with SOD treatments (Figs 1-3). Theeffects of ketotifen on acid and pepsin inducedoesophageal mucosal damage were associatedwith a significant decrease in the presence ofinflammatory cells in the mucosa and laminapropia of the oesophagus (Fig 5).

IN VITRO STUDIES

Isolated cells from the oesophageal mucosaobtained after the in vivo rabbit oesophagealexposure to acidified pepsin for 80 minutesinduced a significant increase (300%) in genera-tion of superoxide anions compared with controlexperiments (isolated cells without previous invivo acidified pepsin exposure; Table I). The cellpopulations identified by flow cytometry showedsimilar increases in the production of superoxideanions, but the high correlation obtainedbetween the superoxide anion content and theimmune CD1 lb staining suggest that mostsuperoxide anions were produced by the inflam-matory cells that were equally distributed in theepithelial populations identified by flowcytometry (Fig 6). The percentage of CD1 lbpositive cells isolated from the oesophagealmucosa increased after the rabbit oesophagealexposure in vivo to acidified pepsin from 2 97(1 08) to 6 06 (0 97)% (p<001). Hydrogenperoxide production by isolated cells from thispreparation did not differ from production bycontrol cells (Table II).

DiscussionThe data obtained in this study show thatgeneration of superoxide anions is involved inthe pathogenesis of acid and pepsin inducedoesophageal damage in rabbits. The admin-istration of SOD, a scavenger of this radical,significantly reduced the oesophageal damageinduced by acidified pepsin. Other free radicalscavengers were either weaker or did not offerany significant mucosal protection. Involve-ment of oxygen free radicals in mucosal injuryhas usually been assessed indirectly, but wealso directly evaluated and confirmed thepresence of superoxide anion in cells isolatedfrom the oesophageal mucosa after acidexposure by measuring ethidium bromide,which is intracellularly formed by the oxidationof hydroethidine in the presence of thesuperoxide anion. This study also suggests thatthe presence of inflammation is associated withoesophageal damage as the anti-inflammatorydrug ketotifen, which is not a free radicalscavenger, was as effective as SOD in theprevention of oesophageal damage induced byacid and pepsin, and that activated anti-inflam-matory cells (neutrophils, or monocytes, ormacrophages) are one of the main sources ofthis free radical, although other potential, butprobably minor, sources were not excluded.Oxygen free radicals are detrimental to the

integrity of biological tissues and mediate theirinjury. The mechanisms of damage involvelipid peroxidation, which destroys cell mem-branes with the release of intracellular

178

1:

on October 10, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.175 on 1 F

ebruary 1997. Dow

nloaded from

Siuperoxi'de anion anidoesopliagiti1s

'A

hB:t.: . - -As, - ,>; - ......~~....Figuire 4: M1icrophotographs of oesophageal speciniens in control experinients (A and B)anzd ani'mals treated .it/i superoxide dismnziutase (C). Sniperoxide disniutase redutced thzcMnicroscopical mnucosal daniiage and inflammatioon inldniced by acidified pepsiml.(A) Oesophageal epitthelial ulceration and oedemna, vascular congestionI, haemnorrhage, andpolvmnorphonuclear cell infiltrationz; (B) Detail of the polvmnorph/onuclcar cell infiltration;(C) Smnall suipeificial epitlielial loss with oedeniia and z ascuilar congestion. Nopolvmnomphonuclear cell infiltrationm .as Seemi in ttins specinMein.

None Mild M Marked

n = 6-8 *p < 0.05

100o<a)

co

CZ

50

oControl DMSO Catalase SOD Ketotifen

Figure 5: Effects of diffcerent frec radical scaven)ges andketotifecn oni tlhe presence of inflammatory cells in the mnnztcosaand lamiina propn7a of t/ic oesophags after the intraluminalpefiusion of acidified pepsin in rabbits (DAISO=dimethlv sulpihoxide; SOD=supcwxtide dismnutas ).

ITABLE II Aleas urenment of hvdro-gen pe ox-.ide productionby cellflow cvtomietrvin i'solated cells obtained from theocsophageal m)lucosa before a)ld after thc perfusion ofacidified pepsin in vi. o

C-ll popudlationl Ititltoutr dantag. 18 itlr daiiagc

Total 5.59 A1 74' 6 36 '3 03)Population A 7 51 2 70' 6 95 4 15)Population B 4 25 18 44 (54 1 99)

The generation of hydrogen peroxide is expressed as the mean(SEM) of arbitrars units of fluorescence of rhodamine n infour experiments. In population A there x-as a predominanceof squamous cells and in population B a predominance of basalcells.

generation of the superoxide anion as a mech-anism of damage is well established in differentmodels of acute and chronic injury in thestomach and intestine,4 13 but it is not wellestablished whether this radical is involved inthe pathogenesis of acid and pepsin inducedoesophageal damage. In a recent report byWVetscher et al,I oesophagitis induced bvduodenogastro-oesophageal reflux after duo-denojejunal ligation in rats was prevented orsignificantly reduced bv the blockade of thesuperoxide anion with SOD, but it was notreduced with catalase, dimethvlthiourea, allo-purinol, or vitamin E succinate. Treatmentwith catalase reduced lipid peroxidation by506% but was not sufficient to inhibit oeso-phagitis on gross appearance. In a differentmodel of oesophagitis, induced by perfusing

A

as

C:co '1a)as-

Y -

aso0~

Q ,)-E

400 n = 6

* p < 0.01300

200

100

0CD1lb (+) cells CD1lb ( cells

TIABLE I .lcasutreinent of s'npcoxide anion prodnction by,cell flow- cvtomnetrltin i'solated cells obtained front tuzeoeSopliageal muicosa before and after tiie pctfisnion (ifacidified pepsi i1n vico

C'.ll popultati() lf'oitulrli dawaclfa rdt'itildawiia-Total 45-48 (25 43' 152 50 (207 3 *

Population A 84 46 (4070) 2'55TO087 5)*Population B 4427 120 04) 121 13 (4 83'

The generation of superoxide anion is expressed as the mean(SEMI) of arbitrarx units of fluorescence of ethidium bromidein four expenrments. In population A there wi as apredominance of squamous cells and in population B apredominance of basal cells.*p<0 0O.

components, such as lvsosomal enzymes,leading to further tissue damage. The radicalsalso promote mucosal damage bv causingdegradation of the epithelial basement mem-brane components, complete alteration of thecell metabolism, and DNA damage.) The

B 4

3- 02-(CDl 1 b-)

2 ~~~~~~~HEHE 02-(CD11 b+)

1 \ ~~~~~HE-

C {1,A HE 02- tota

TotalLgFITC

Ftgniire 6: (A) Snperooxide anlioni production (meaasmured asarbittrarr iunti of fluorc sicnce osf /tvdroethiditn=HE) bviSolatcd CDI lb (+) anid CD] lb (-) cells obtaincd fronltrabbit oesop/lageal mnuiicosa after t/zc in vivo intraluminm alpcr-fnion (f acidified pcpsin. (B) ExamnplIc of onc (7f t/hcc.xpcrimntcts pcfrfcsncd: (1) Total flutorcscentcc obtatnic'd in1all cell populations aftic t/te addition of t/ic labellecd (FITC)CDJ lb antibody. (2) Fluiorescccncc correiponding toin7tracellutlar superoxide anion (HE) produiccd by the w/tolcceil population. (3) Fluiorcescenicc of ittraccillular sipLc )ox dicanlionz produiecd bs' CD/ lb (+) clls. (4) No significantfl uosrescence corresponding to inttraccllular- sipc10ox'Sic/c isgenerated tint CD] lb (-) cells.

I17i9

on October 10, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.175 on 1 F

ebruary 1997. Dow

nloaded from

Naya, Pereboom, Ortego, Alda, Lanas

controlled amounts of acid and pepsin for fixedperiods, we have shown both directly andindirectly that superoxide anions were directlyinvolved in the oesophageal damage inducedby acid and pepsin in rabbits. In our model,catalase was much less effective than SOD inpreventing or reducing the mucosal damage,and generation of hydrogen peroxide was not

detected in mucosal cells isolated afterdamage, but we cannot exclude that smallamounts of this radical might be produced invivo. The involvement of these two radicalsseems plausible because SOD efficiently andspecifically scavenge the superoxide radical bycatalysing its dismutation to hydrogenperoxide and oxygen.27 Hydrogen peroxide,which is a relatively non-toxic substance can bescavenged by catalase to form water andmolecular oxygen.27

Hydroxyl radicals are formed by the reactionof superoxide anions with hydrogen superoxidevia the Haber-Weiss reaction. This reactionproceeds at too slow a rate to be of physio-logical relevance. However, this reaction isaccelerated by the presence of transition metalssuch as iron. In these conditions the generationof hydroxyl radicals, which are highly reactiveand cytotoxic, is particularly harmful. In our

model, parenteral DMSO, a scavenger ofhydroxyl radicals, did not have a significanteffect on oesophageal damage, suggesting thatthis radical is not involved in the pathogenesisof acid and pepsin induced oesophagitis.Similar results were reported by Westcher etal,'5 who did not find any significant effect withdimethylthiourea alone, another hydroxylradical scavenger, on rat oesophagitis.However, the hydroxyl radical is extremelyreactive and it is unlikely that DMSO in vivoexperiments will reach the site of generation ofthis radical in sufficient concentration to pre-vent hydroxyl radical induced local damage andtherefore, despite our findings and Westcher'sdata, a role for hydroxyl radicals in the patho-genesis of acid and pepsin induced oesophagitiscannot be definitely excluded.An important question related to this investi-

gation concerns the source of the oxygen

radicals produced during oesophageal injuryinduced by acid and pepsin. Our in vitrostudies suggest that the inflammatory infiltrateis the main source of generation of superoxideanions. As measured by flow cytometry, iso-lated oesophageal cells obtained after the per-fusion of acidified pepsin showed increasedgeneration of superoxide anions. Pretreatmentwith a monoclonal antibody directed againstsubunits of the granulocyte adherence glyco-protein CD 1 lb showed that most of the super-oxide anion was generated in CDl lb positivecells, which include neutrophils, monocytes,and macrophages in rabbits. Eosinophils are

not labelled by the CD1 lb antibody in rabbitsand, therefore, we cannot exclude the possi-bility that this type of cell, usually present as

a small percentage of the inflammatory infil-trate, may contribute to the generation of super-oxide anions. In a recent study, the indirectmeasurement of reactive oxygen species byluminol chemiluminescence from human oeso-

phageal biopsy specimens showed an increasedproduction of these compounds in patients withoesophagitis that was partially reduced by the invitro treatment with the myeloperoxidaseinhibitor azide and catalase, suggesting thatneutrophils were not the only source of mucosalluminol chemiluminescence.2"

In our model, ketotifen also significantly anddramatically reduced the oesophageal damageinduced by acidified pepsin. The effect wassimilar to that obtained with the perfusion ofthe superoxide anion scavenger SOD. Keto-tifen is an anti-inflammatory drug, usuallyknown as a "mast cell stabiliser" with otherimportant effects that include the inhibition ofneutrophilic migration,29 30 the reduction ofeosinophil viability,3' and inhibition of cellNADPH oxidase,32 and the inhibition of leuko-trienes, platelet activating factor, prostaglandinE2, and tromboxane B2 concentrations in in-flamed tissues.3335 Previous studies have alsoshown that treatment with ketotifen eitherreduced or totally prevented mucosal damagein different models of experimental colitis,33Clostridium difficile toxin A induced enteritis inrat ileum and ethanol and gastric mucosaldamage induced by non-steroidal anti-inflammatory drugs in rats and humans.29 33 3The presence of increased amounts of neutro-phils and macrophages in the oesophagealmucosa after damage suggests the involvementof soluble chemotactic mediators causing mi-gration of inflammatory cells into the mucosa.In other models of damage, mast cells areinvolved in both the induction and amplifi-cation of the inflammatory proceSS33-36 byreleasing soluble mediators. Therefore, theeffects of ketotifen on this model of acid andpepsin induced oesophagitis could be due to aneffect on resident oesophageal mast cells. How-ever, mast cells, which are present in the adultNorth American opossum and humans, arevery rare in rabbit oesophageal mucosa,37 38

suggesting that most effects of ketotifen in thismodel of oesophagitis might be mast cell inde-pendent and probably related to other drugactions (for example, inhibition of neutrophilicmigration). Our histological studies haveshown a significant reduction in the presenceof inflammatory cells only in those experimentsin which rabbits were treated with ketotifen.The results of this study demonstrate that

oxygen derived free radicals, especially thesuperoxide anion, are involved in the patho-genesis of acid and pepsin induced oeso-phagitis in rabbits, and that the inflammatoryinfiltrate - namely, neutrophils, macrophages,and monocytes - are the main source of thisradical and the drugs that modulate this in-flammatory response or radical scavengersdramatically reduce the acid and pepsininduced injury. Because oesophagitis inhumans is often characterised by epithelialerosion, ulceration, and accompanying inflam-matory infiltrate,' these results may be usefulin the devleopment of future treatment ofreflux oesophagitis.

We thank Dr Tomas Casuella for his help in the cytologicalanalysis and Clara Tapia for technical assistance. This study wassupported by grants 92/0670 and 91/0613 from the Fondo de

180

on October 10, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.175 on 1 F

ebruary 1997. Dow

nloaded from

181Superoxide anion andoesophagitis

Investigaciones Sanitarias de la Seguridad Social (FIS). Part ofthis work was presented at the 1995 meeting of the AmericanGastroenterological Association.

1 Kahrilas PJ, Hogan WJ. Gastroesophageal reflux diseases.In: Sleisenger MH, Fordtran JS, eds. Gastrointestinaldiseases. London:WB Saunders, 1993: 378-401.

2 Orlando RC. Esophageal epithelial defences against acidinjury. AmJ' Gastroenterol 1994; 89: 48-52.

3 KhalbussWE, Marousis CG, SubramanyamM, Orlando RC.Effect of HCI on transmembrane potentials and intra-cellular pH in rabbit esophageal epithelium. Gastro-enterology 1995; 108: 662-72.

4 Grisham MB, Hernandez LA, Granger DN. Xanthine oxi-dase and neutrophil in filtration in intestinal ischemia. AmJPhysiol 1986; 251: G567-74.

5 Tarnasky PR,Livingston EH, Jacobs KM, Zimmerman BJ,Guth PH, Garrick TR. Role of oxyradicals in cold waterimmersion restraint-induced gastric mucosal injury in therat.DigDis Sci 1990; 35: 173-7.

6 Wallace JL, Arfors KE, Mcknight GW. A monoclonalantibody against the CD18 leukocyte adhesion moleculeprevents indomethacin-induced gastric damage in therabbit. Gastroenterology 1991; 100: 878-83.

7 Terano A, Hiraishi H, Ota S, Shiga J, Sugimoto T. Role ofsuperoxide and hydroxyl radicals in rat gastric mucosalinjury induced by ethanol. Gastroenterol Jrpn 1989; 24:488-93.

8 Yasue N, Guth PH. Role of exogenous acid and re-transfusion in hemorrhagic shock induced gastric lesionsin the rat. Gastroenterology 1988; 94: 1135-43.

9 Nakamura K, Aoike A, Rokutan K, Hosokawa T, Koyama K,Kawai K. The role of oxygen radicals in the pathogenesisof gastric mucosal lesions induced in mice by feeding-restriction stress. Scand J Gastroenterol 1989; 24 (suppl162): 47-50.

10 Binnaka T, Yamaguchi T, Hirohara J, Hiramatsu A,Mizuno T, Sameshima Y. Gastric mucosal damageinduced in rats by intravenous administration of platelet-activating factor. Scand Jf Gastroenterol 1989; 24 (suppl162): 67-70.

11 Perminder SP, Colin JG, Meron RJ. A radical view of thestomach: the role of oxygen-derived free radicals and anti-oxidants in gastroduodenal disease. Eur J GastroenterolHepatol 1995; 7: 265-71.

12 Stein HJ, Espluges J, Whittle BJ, Bauerfeind P, Hinder RA,Blum AL. Direct cytotoxic effect of oxygen radicals on thegastric mucosa. Surgery 1989; 106: 318-24.

13 Matsumoto T, Moriguchi R, Yamada H. Role of poly-morphonuclear leucocytes and oxygen-derived freeradicals in the formation of gastric lesions induced byHCl/ethanol, and a possible mechanism of protection byanti-ulcer polysaccharide. J Pharm Pharmacol 1993; 45:535-9.

14 Stein BE, Schwartzman ML, Carroll MA, Stahl RE,Rosenthal WS. Role of arachidonic acid metabolites inacid-pepsin injury to rabbit esophagus. Gastroenterology1989; 97: 278-83.

15 Jimenez P, Esteva F, Piazuelo E, Garcia MA, Lanas A.Prostaglandins and arachidonic acid metabolites pro-duced by esophageal mucosal cells: implications formucosal defense. Eur Surg Res 1994; 26: 35-6.

16 Lillemoe KD, Johnson LF, Harmon JW. Alkaline eso-phagitis: a comparison of the ability of components ofgastroduodenal contents to injure the rabbit esophagus.Gastroenterology 1983; 85: 621-8.

17 Lanas A, Sousa FL, Ortego J, Soria J, Esteva F, Blas JM,et al. Aspirin renders the esophageal mucosa more per-meable to acid and pepsin in rabbits by different mechan-isms. EurJ3 Gastroenterol Hepatol 1995; 7: 1065-72.

18 Wetscher GJ, Hunder PR, Bagchi D, Perdikis G,Redmond EJ, Glaser K, et aL Free radical scavengers

prevent reflux esophagitis in rats. Dig Dis Sci 1995; 40:1292-6.

19 Schweitzer EJ, Bass BJ, Johnson LF, HarmonJW. Sucralfateprevents experimental peptic esophagitis in rabbits.Gastroenterology 1985; 88: 611-9.

20 Tay HP, Chaparala RC, Harmon JW, HueskenJ, Saini N,Hakki FZ, et al. Bismuth subsalicylate reduces pepticinjury of the oesophagus in rabbits. Gut 1990; 31: 11-6.

21 Trevethick MA, Clayton NM, Strong P, HarmanIW. Doinfiltrating neutrophils contribute to the pathogenesis ofindomethacin induced ulceration of the rat gastricantrum? Gut 1993; 34: 156-60.

22 Layden TJ, Agnone LM, Schmidt LN, Hakim B,Goldstein JL. Rabbit esophageal cells possess an Na+, H+antiport. Gastroenterology 1990; 99: 909-17.

23 Lanas A, Nerin J, Esteva F, Sainz R. Nonsteroidal anti-inflammatory drugs and prostaglandin effects on pep-sinogen secretion by dispersed human peptic cells. Gut1995; 36: 657-63.

24 Rothe G, Valet G. Flow cytometric analysis of respiratoryburst activity in phagocytes with hydroethidine and 2',7'dichlorofluorescein. JLeukoc Biol 1990; 47: 440-8.

25 Rothe G, Oser A, Valet G. Didydrorhodamine 123: a newflow cytometric indicator for respiratory burst activity inneutrophil granulocytes. Naturwissenchaften 1988; 75:354-5.

26 Halliwell B. Reactive oxygen species in living systems:source, biochemistry and role in human disease. Am J7Med 1991; 91: 14-22.

27 Mutoh H, OtaS, Hiraishi H, Ivey KJ, Terano A, Sugimoto T.Reduced glutathione protects cultured gastric mucosalcells from suckling rats against acid. Am J Physiol 1991;261: 65-70.

28 Olayaee M, Sontag S, Salman W, Schnell T, Mobarhan S,Eiznhamer D, et al. Mucosal reactive oxygen speciesproduction in oesophagitis and Barrett's oesophagus. Gut1995; 37: 168-73.

29 Pathoulakis C, Karmeli F, Kelly CP, Eliakim R, Joshi MA,O'Keane CJ, et al. Ketotifen inhibits Clostridium difficiletoxin A induced enteritis in rat ileum. Gastroenterology1993; 105: 701-7.

30 Shiratori Y, Takada H, Hai K, Kiriyama H, Mawet E,Komatsu Y, et al. Effect of antiallergic agents on chemo-taxis of neutrophils by stimulation of chemotactic factorreleased from hepatocytes exposed to ethanol. DigDis Sci1994; 39: 1569-75.

31 Hossain M, Okubo Y, Sekiguchi M. Effects of various drugs(staurosporine, herbimycin A, Ketotifen, theophylline,FK506 and cyclosporin A) on eosinophil viability. Arerugi1994; 43: 711-7.

32 Hojo M, Hamasaki Y, Fujita I, Koga H, Matsumoto S,Miyazaki S. Effects of antiallergy drugs on fMet-Leu-Phe-stimulated superoxide generation in human neutrophils.Ann Allergy 1994; 73: 21-6.

33 Eliakim R, Karmeli F, Okon E, Rachmilewitz D. Ketotifeneffectively prevents mucosal damage in experimentalcolitis. Gut 1992; 33: 1498-503.

34 Eliakim R, Karmeli F, Chorev M, Okon E, Rachmilewitz D.Effect of drugs on colonic eicosanoid accumulation inactive ulcerative colitis. Scand J Gastroenterol 1992; 27:968-72.

35 Eliakim R, Karmeli F, Rachmilewitz D. Ketotifen old drug,new indication: reduction of gastric mucosal injury. ScandJf Gastroenterol 1993; 28: 202-4.

36 Kubes P, Kanwar S, Niu XF, Gaboury JP. Nitric oxidesynthesis inhibition induces leukocyte adhesion viasuperoxide and mast cells. FASEBJ7 1993; 7: 1293-9.

37 Barclay RL, Dinda PK, Morris GP, Paterson WG. Morpho-logical evidence of mast cell degranulation in an animalmodel of acid-induced esophageal mucosal injury. Dig DisSci 1995; 40: 1651-8.

38 Norris HT, Zamcheck N, Gottlieb LS. The presence anddistribution of mast cells in the human gastrointestinaltract at autopsy. Gastroenterology 1963; 44: 448-55.

on October 10, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.175 on 1 F

ebruary 1997. Dow

nloaded from