Gut1996;39: 164-171

Gastroprotective effect of ranitidine bismuthcitrate is associated with increased mucus bismuthconcentration in rats

S Tanaka, P H Guth, G Paulsen, J D Kaunitz

AbstractBackground-Antisecretory and bismuthcompounds protect the gastric mucosafrom injury resulting from non-steroidalanti-inflammatory drugs.Aim-To study the mechanism underlyingthe gastroprotective effects of ranitidinebismuth citrate (GG3 11) in rats.Methods-Indomethacin rat injury modeland in vivo microscopy in which acid out-put, surface cell intracellular pH (pHi),gastric mucus gel thickness, and mucosalblood flow were measured simultane-ously.Results-In injury studies, GG311 dosedependently protected against severeinjury induced by indomethacin (60 mg/kg subcutaneously). In in vivo micro-scopic studies, indomethacin significantlydecreased mucus gel thickness andincreased the initial rate ofacidification ofgastric surface cells when the superfusatepH was lowered from 7.4 to 1.0, andimpaired pHi during acid exposure.Indomethacin had no effect on mucosalblood flow or acid output. GG311 alonehad no effect on gel thickness, blood flow,or pHi homeostasis during acid exposure,but improved the initial acidification rateand pHi during superfusion with pH 1.0solutions in the presence ofindomethacin.In separate experiments, indomethacinpretreatment considerably increasedgastric mucus bismuth concentrations inrats given GG311.Conclusions-The gastroprotective effectof GG311 against indomethacin inducedgastric injury is associated with high andprolonged gastric mucus bismuth con-centrations, which may impair protonpermeation across the mucus gel.(Gut 1996; 39: 164-171)

Keywords: stomach, intracellular pH, ranitidinebismuth citrate, gastric mucus, proton diffusion,mucosal blood flow.

Bismuth salts protect the gastric mucosa fromdamage caused by non-steroidal anti-inflam-matory drugs (NSAIDs) by a yet undefinedmechanism.' 2 The most usual explanationsadvanced have suggested that the primarymode of action of bismuth on the gastricmucosal barrier is at the pre-epithelial level.3-5Antisecretory compounds, such as H2 receptor

antagonists, have also been shown to preventduodenal ulcers in patients taking NSAIDs.6 7

Ranitidine bismuth citrate (GG311) is anovel drug that combines two compoundsknown to protect the gastric mucosa: an H2receptor antagonist and a bismuth salt. Thiscompound has been shown to prevent NSAIDinduced gastric ulcers in an animal model8 andin a clinical study.9 However, the mechanismby which GG311 protects gastric mucosa hasnot been clearly explained. The aim of thisstudy is therefore to ascertain the effect of thisnovel compound on gastric defensive mech-anisms, as determined with an establishedgastric corpus injury model induced byindomethacin,10 11 and with a novel systemdeveloped in our laboratory that enables thegastric barrier function to be measured in vivo.

Methods

ANIMALSMale Sprague-Dawley rats weighing 200-250 gwere fasted overnight, but had free access towater. All studies were approved by the AnimalUse Committee of the West Los AngelesVeterans Administration Medical Center.

ASSESSMENT OF GASTRIC MUCOSAL LESIONSRats were killed in a carbon dioxide chamber,the abdomens opened, and the stomachsremoved and opened along the greater curva-ture. Severe lesions were assessed according topreviously described methods.'1 The lesionswere graded and scored by an observer whowas unaware of the drugs administered as fol-lows: petechial lesions= 1, erosions less than 1mm=2, erosions between 1 and 2 mm=3,erosions between 2 and 4 mm=4, and erosionsgreater than 4 mm=actual length in mm. Theindividual lesion scores in each rat weresummed to provide a lesion score for eachanimal.For histological evaluation, three strips of

tissue were taken from across the entireposterior wall (parallel to the limiting ridge):(a) just below the limiting ridge, (b) the mid-corpus, and (c) the distal-corpus. The severityof gastric mucosal injury was evaluated on thesections stained with haematoxylin and eosinusing a modification of established criteria.12Damage was graded as follows: surfacemucous cell damage (surface damage);erosions down to, but not deeper than, the

Medical Service, WestLos Angeles VAMedical Center

Department ofMedicine, Universityof California, LosAngeles

CURE DigestiveDiseases ResearchCenter, Los Angeles,USAJ D KaunitzP H GuthS Tanaka

Research Service,West Los Angeles VAMedical CenterG Paulsen

Correspondence to:Dr J D Kaunitz, CUREDigestive Diseases ResearchCenter, Bldg 1 15, Rm 219,Wadsworth VA MedicalCenter, 11301 Wilshire Blvd,Los Angeles, CA 90073,USA.Accepted for publication27 February 1996

164

Ranitidine bismuth citrate protects gastric mucosa

mucous neck cell area (superficial erosions);and erosions extending down into the parietalcell area (deep erosions). At least 100 glandswere assessed in each specimen. The lesionscore was calculated by multiplying thepercentage of glands with no injury by 0, thepercentage with surface damage by 1, thepercentage with superficial erosions by 2, andthe percentage with deep lesions by 3. Thescores were then added and averaged amongthe three specimens.

IN VIVO MICROSCOPIC MEASUREMENTSAn in vivo microfluorometric technique,described in detail elsewhere,13-17 was used tomeasure intracellular pH (pHi) and mucousgel thickness. The initial acidification rate wascalculated from the fall of pHi during the firsttwo minutes of acid exposure as describedpreviously.14 Gastric mucosal blood flow wasmeasured by laser-Doppler flowmetry, as apercentage of baseline, and acid secretion wasmeasured by back titration.18 In brief, thetechnique entails microscopic observation ofthe gastric mucosa of an urethane anaes-thetised rat under epifluorescent illumination.The mucosa is continuously superfused bymeans of a perfusion chamber. Carbon parti-cles, placed on top of the mucus gel, delineatethe mucus-luminal interface,19 enablingmeasurement ofmucus gel thickness by up anddown focusing. Intracellular pH is measuredby a ratiometric technique using the trappedintracellular dye 5,6 carboxyfluorescein dia-cetate (CF). Image analysis is used to limit thearea of interest for pHi measurements to two tothree surface cells.

MEASUREMENT OF MUCUS AND SERUMBISMUTH CONCENTRATIONThe bismuth concentration in gastroduodenalmucus was measured using a modification ofthe method of Slikerveer et al,20 with mucuscollected according to the method of Mufiozet al.21 Gastric and duodenal mucus weregently scraped with a glass slide and weighed.Blood was collected by cardiac puncture.Mucus and serum were solubilised in 5 MHNO3 for 12 hours at 70°C. Water was thenadded for a final concentration of 0.5 MHNO3. Bismuth concentration was measuredby inductive coupled plasma atomic emissionspectroscopy (ICP) by the UCLA Environ-mental Analysis Laboratory, and expressedas ,ug/g wet weight for mucus bismuth con-centrations or ,ug/l for serum bismuth concen-trations.

EXPERIMENTAL DESIGN - INJURY STUDIESAfter an overnight fast, rats received GG311(1 ml volume) or vehicle (water) by gavage.One half hour later, the animals were treatedwith 60 mg/kg indomethacin (1 ml volume) orvehicle (PEG 400) by subcutaneous injection.Six hours after receiving indomethacin, theanimals were killed and gastric lesions werequantified.

EXPERIMENTAL DESIGN -IN VIVO MICROSCOPYSTUDIESRats were divided into six groups, control,indomethacin, GG311, GG311+indometha-cin, bismuth citrate+indomethacin, andranitidine+indomethacin. GG31 1 (100 mg/kgin 1 ml volume), ranitidine (45 mg/kg),bismuth citrate (57 mglkg) or vehicle (water)were given by gavage. One half hourlater, the animals were treated with60 mg/kg indomethacin (1 ml volume) orvehicle (PEG 400) by subcutaneous injection.Two hours after the injection of indomethacinor vehicle, in vivo microscopic observationwas started, with measurement of acidoutput, mucosal blood flow, surfacecell pHi, and mucus gel thickness. The cham-ber was superfused with pH 7.4 Krebs'solution for the first 30 minutes, thenthe superfusate was changed to pH 1.0solution.

EXPERIMENTAL DESIGN -BISMUTHMEASUREMENTSRats were treated with indomethacin, 60mg/kg or vehicle (PEG 400) subcutaneously,and GG311, 100 mg/kg, orally as describedabove. Two groups were studied: GG311+vehicle, GG311 +indomethacin. Rats werekilled at one, three, six, and 12 hours afterGG311 treatment.

SOLUTIONSThe dye solution consisted of CF, whichwas first dissolved in dimethylsulphoxide andthen diluted with Krebs' solution. The finalconcentration of the solvent was less than1%. Krebs' solution contains (in mM)136 NaCl, 2.6 KCl, 1.8 CaCl2, N-2-hydroxyethylpiperazine-N'-2-ethanesulphonicacid (HEPES) for pH 7.4. For acid superfu-sion, Krebs' solution was (in mM) 36 NaCl,2.6 KC1, 1.8 CaCl2, HEPES, and titrated topH 1.0 with 5 N HC1. Indomethacin was dis-solved in PEG 400. GG311, ranitidine, andbismuth citrate were dissolved in distilledwater.

CHEMICALSCF was purchased from Molecular Probes(Eugene, OR, USA). Indomethacin, bismuthcitrate, ranitidine, and PEG400 were pur-chased from Sigma Chemicals (St Louis, MO,USA). Ranitidine bismuth citrate (GG31 1)was a gift from Glaxo Research andDevelopment (Greenford, UK).

STATISTICAL ANALYSISResults were expressed as means (SEM).Comparisons between the two groups werecalculated by the Student's t test. Multiplegroup comparisons were performed by analysisof variance (ANOVA, factorial or repeatedmeasures) followed by Fisher's contrast.A probability level <0-05 was consideredsignificant.

165

Tanaka, Guth, Paulsen, Kaunitz

0.a0.

0.

0

'a

4-

1250

0200 O

(Q

0._o0

150 =

00

100

~~VohicI*- -10 ~~~~~ - 0 20 -40. 60 80 100tn241,

,

(n-: 0nti (n=)GG 1 1:doe (m-g/g) GG311 dose (mg/kg)

Figure 1: Inhibitory effect of GG311 on gastric damage induced by subcutaneous injection of 60 mg/kg indomethacin.Rats received GG311 or vehicle by gavage 30 minutes before indomethacin (60 mg/kg subcutaneous) treatment. Six hoursafter receiving indomethacin, gastric lesions were assessed as described in Methods. (A) Effect ofGG31 1 on gross gastricdamage. (B) Effect ofGG31 1 on histological gastric damage. Results are the mean of rats per group. In this andsubsequentfigures, vertical error bars represent SEM. *p<0.05, **p<0.01, ***p<0.001 v vehicle treated rats byANO VA.

Results

INJURY MODELSevere lesions were present in indomethacininjected, vehicle pre-treated rats (lesionscore=33-4 (3-3), n=24). GG311 decreasedthe lesion scores in a dose dependent fashion.Lesion scores were 23.5 (7.8) (n=8), 8.0 (4.4)(n=8), and 1-6 (1 1) (n=8) in rats treated with10, 30, and 100 mg/kg GG31 1, respectively(Fig 1A). Lesion scores in rats treated with thetwo highest doses were significantly decreasedfrom the lesion scores in untreated rats. Thehistological lesion score in rats treated withindomethacin alone was 134 (10) (n=8).Pretreatment with 10 mg/kg GG311 signifi-cantly increased the lesion score to 160 (5)(n=8), although pre-treatmnent with GG311, 30and 100 mg/kg decreased the scores to 106 (8)(n=8) and 77 (11) (n=8), respectively (Fig1B). Deep lesions, defined as injury extendinginto the glandular portion of the gastric pitswere significantly prevented by the highest dose(100 mg/kg) of GG311. Deep lesions were

present in 4.7 (1.1)%, 5.3 (1.4)%, 3.0 (1.7)%,and 0 5 (0.3)%, ofglands in rats pretreated with0, 10, 30, and 100 mg/kg GG311, respectively.

IN VIVO MICROSCOPIC STUDY

Acid outputAcid output was significantly suppressed, asexpected, by GG3 11 or ranitidine. Acid outputwas 0-45 (0.05) pumol/min/cm2 in controls(n= 10), 0 1 1 (0-03) pumolmin/cm2 in GG31 1treated rats (n=9), 0-31 (0-06) pmol/min/cm2in indomethacin treated rats (n=9), 0-09(0-03) pRmolmin/cm2 in GG31 1 +indo-methacin treated rats (n=8), 0-39 (0-06)pumolmin/cm2 in bismuth citrate+indo-methacin treated rats (n=6), and 0-13 (0-03)gmol/min/cm2 in ranitidine+indomethacintreated rats (n= 6). Acid output was thussignificantly decreased in groups treated withGG311 or ranitidine (p<0-05 v controls, byANOVA).

120

100

0_-

1080

a

-o-. 10 20. 3-0 40 50Figure 2: Effect ofGG311 and its components on relative gastric mucosal bloodflow. Ratswere pretreated with GG31 1, bismuth citrate (Bis), ranitidine (Ranit) or vehicle bygavage 30 minutes before subcutaneous injection of 60 mg/kg indomethacin (Indo) orvehicle. Two hours after indomethacin or vehicle injection, gastric mucosal bloodflowmeasurement by laser-Dopplerflowmetry was started.

Mucosal bloodflowFigure 2 depicts relative gastric mucosal bloodflow. Relative gastric mucosal blood flow wasnot affected by any treatment during in vivomicroscopic experiments. Blood flow graduallydeclined in all groups to a level of 83.8 (1.4)%(n= 10), 78-1 (4.9)% (n=9), 79-1 (2-5)%(n=8), 90-8 (5.4)% (n=8), 81*7 (4.1)%(n=6), and 88-8 (3.3)O/O (n=6) of baseline incontrol, GG311, indomethacin, GG311+indomethacin, bismuth citrate+indomethacin,and ranitidine+indomethacin treated groups,respectively, measured at 60 minutes. Consis-tent with our prior studies,22 acid superfusiondid not affect mucosal blood flow.

Mucus gel thicknessFigure 3 depicts mucus gel thickness, asmeasured by the fluorometric technique.Indomethacin significantly decreased mucus

166

Ranitidine bismuth citrate protects gastric mucosa

120r-

E..=L

C.-

0

'Eto

U

pH 7L 4 superfusion

Control (n = 10) GG311 (n=9)

100l

.1-

Bis + Indo* (n m 6) Ranit +lndo* (

0 10 20 30 40Figure 3: Effect of GG311 and its components on mucus gel thickness. Rats zwith GG311, bismuth citrate (Bis), ranitidine (Ranit) or vehicle by gavage .before subcutaneous injection of 60 mg/kg indomethacin (Indo) or vehicle. Ttindomethacin or vehicle injection, measurement ofmucus gel thickness was ste*p<0.05 v controls by repeated ANOVA.

gel thickness over the 50 minute meperiod (p<0 05 v control, byANOVA). At time=0 min, gel thi119.8 (5.7) Kim (n=10), 112.0(n=9), 90.0 (5.0) ,um (n=9), 88.9(n=8), 78A4 (7.3) mm (n=6), andmm (n= 6) in the control, GGmethacin, GG31 1 +indomethacincitrate+indomethacin, and raniticmethacin treated groups, respectivepretreatment did not affect gel thickcontrol or indomethacin pre-treatedaccord with our prior studies,22 afusion had no effect on gel thicknes

IntracellularpH and initial acidificatiDuring superfusion with pH 7*4 KrpHi remained stable in the range of'to 7.11 (0.04), with no statisticalbetween any of the groups. During swith pH 1.0 buffer, pHi was significin the indomethacin and raniti(methacin groups than in the con

7.25 r-

7

6.75

C.

6.25Rant + InSdo*;

Indo* (nm 18)l l

0 10 20 30 -o40 50

Figure 4: Effect of GG311 and its components on intracellularpH (pHi) ofAcells. Rats were pretreated with GG31 1, bismuth citrate (Bis), ranitidine (Rxvehicle by gavage 30 minutes before subcutaneous injection of 60 mg/kg indo,(Indo) or vehicle. Two hours after indomethacin or vehicle injection, gastric pmeasurement was started. IntracellularpH was compared among six groups Esuperfusion with pH 7-4 andpH 1 0 solutions by repeatedANOVA. *p<O. (tp<0-05 v indomethacin treated group by repeatedANOVA for the period osuperfusion (20-50 min, inclusive).

(p<005, by repeated ANOVA during acidsuperfusion), and was significantly higher inthe GG31l+indomethacin, and bismuthcitrate+indomethacin groups compared withthe indomethacin group (p<005, by repeatedANOVA during acid superfusion). GG311itself had no significant effect on pHi during

=n 8) acid superfusion in the absence of indo-methacin (Fig 4).The initial acidification rate was calculated

from the initial drop of pHi after the super-fusion of pH 1(0 buffer. Initial acidification

n f6)urate was7o53 (0153)mMb min (n=20),6a63(0'54) miM/min (n=20), 11.75 (0.55)mM/min (n= 18), 8&08 (0.54) mM/min

50 (n= 16), 9.32 (0.98) mM/min (n= 12), andwere pretreated 1034 (0.94) mM/min (n= 12) in the control,30 minutes GG3 11, indomethacin, GG311 +indometha-wo hours afterarted. cin, bismuth citrate+indomethacin, and

ranitidine+indomethacin treated groups,respectively. The acidification rates in the

easurement GG3 11, GG311 +indomethacin, and bismuthrepeated citrate+indomethacin groups were statistically

ickness was similar to that of the controls, and was(4.8) ,um significantly faster in the indomethacin, and

2 (5.6) ,um ranitidine+indomethacin groups (p<0 05L83-5 (6.1) v control, by ANOVA).311, indo-L. bismuthdine+indo- MUCUS BISMUTH CONCENTRATIONSly. GG3 11 Figure 5 shows mucus bismuth concentrationkness ofthe in rats treated with either GG311 alone or1 groups. In GG311 plus indomethacin. Bismuth concen-acid super- trations in the gastric mucus was 7.3 (2.8) ,ug/gIs. wet weight (n= 5) at baseline. In rats treated

with GG311 alone, gastric mucus bismuthconcentrations were 4903 (2519) (n=4),

,on rate 149.6 (53.5) (n=5), 40.8 (12.2) (n=6), and.ebs' buffer, 31.9 (14.3) ,ug/g (n=3) at one, three, six, and6.98 (006) 12 hours, respectively. Indomethacin in-differences creased the bismuth concentration in GG311superfusion treated rats to 6944 (1596) (n=4, NS), 2017,antly lower (456) (n= 5, 13-5-fold increase, p<001), 1788dine+indo- (669) (n=6, 494-fold increase, p<0O05), anditrol group 207-3 (20.2) (n=3, 65-fold increase, p<0l01)

,ug/g at one, three, six, and 12 hours, respec-tively (Fig 5A). Bismuth concentrations in theduodenal mucus was 12-8 (8.5) ,ug/g wetweight (n=5) at baseline. In rats treated withGG311 alone, duodenal mucus bismuth con-centrations were 1133 (608) (n=4), 16.6(12.1) (n=5), 15.6 (10.5) (n=6), and 15-0(5.4) ,ug/g (n=3) at one, three, six, and12 hours, respectively. Indomethacin alsoincreased the duodenal mucus bismuth con-centration to a lesser extent. Bismuth concen-

k11 + Ilndo*t trations in indomethacin+GG311 treated rats{n-I6) were 1296 (687) (n=4, NS), 208.9 (66.9)+| Indo* (n=5, 12-5-fold increase, p<005), 51.2 (7.6)

In -12) (n=6, 3-2-fold increase, p<0.05), and 29.7(12.9) (n=3, NS) ,ug/g at one, three, six, and12 hours, respectively (Fig 5B).Administration of GG311 resulted in low

levels of bismuth absorption. Basal serum,gast"rc surface bismuth concentration was 14.0 (7.0) ,ug/lanit) ormethacin (n=4). In rats treated with GG311 alone,pHi serum bismuth concentrations were 68.0both during (32.3) (n= 3), 173.3 (115.7) (n= 3), and 112-90f acid ' (44 8) (n=3) jig/l at three, six, and 12 hours,

respectively. In indomethacin+GG3 11 treated

n. .

a 1 1 a

167

6.6

Tanaka, Guth, Paulsen, Kaunitz

-

C._

a)

c;

0

s

A

7500 H

5000 H

1000

500Indo + GG3112500 H

B

0 J \GG311 ** * ~~~~~~~~GG311

' 0 U oI0 3 6 12 0 3 6 12

Figure 5: Bismuth concentration in gastroduodenal mucus gel. Rats were pretreated with GG311 by gavage 30 minutesbefore subcutaneous injection of 60 mg/kg indomethacin (Indo) or vehicle. Rats were killed at one, three, six, and 12 hoursafter GG31 1 treatment. Bismuth concentration was measured by inductive coupled plasma atomic emission spectroscopy asdescribed in Methods. (A) Bismuth concentration in gastric mucus gel. (B) Bismuth concentration in duodenal mucus gel.*p<0.05, **p<0.01 v rats treated with GG311 alone by the Student's t test.

rats, serum bismuth concentrations were 97.8(35'3) (n=3), 84.9 (25.1) (n=3), and 37.3(7.1) ,uUg/l (n=3) at three, six, and 12 hours,respectively. There was no statistically signifi-cant difference between the two groups.

DiscussionGG3 11 dose dependently prevented severe

mucosal lesions in indomethacin treated ratsand prevented microscopic injury at the 30 and100 mg/kg doses, and prevented deep injury tothe gastric glands at the 100 mg/kg dose.GG311 (100 mg/kg) significantly suppressedacid output. No treatment affected relativemucosal blood flow. Indomethacin signifi-cantly reduced mucus gel thickness, increasedthe initial acidification rate, and impaired pH,homeostasis during luminal acid superfusion.In indomethacin treated rats, GG311 (100mg/kg) normalised the initial acidification rateto control values without affecting gel thick-ness, and improved pHi homeostasis duringacid superfusion. The most striking and novelfinding was that indomethacin substantiallyincreased at times >one hour the gastric andduodenal mucus gel bismuth content ofGG3 11 treated rats.

Although bismuth compounds have beenused for several centuries to treat gastro-intestinal problems, their mechanism of actionhas remained incompletely understood.Bismuth salts including GG3 11 have beenshown to protect the gastric mucosa againsta variety of injurious stimuli, includingNSAIDs.1 2 8 9 23 This latter effect suggests thatbismuth might exert at least some of its protec-tive effect in a prostaglandin independentfashion. Mechanisms that have been proposedto explain the gastroprotective effects of bis-muth include inhibition of pepsin activity,4binding of epidermal growth factor,24 bindingto ulcers,25 stimulation of prostaglandin syn-thesis and bicarbonate secretion,26 27 antibac-terial activity,28 and binding to mucus.29Antisecretory compounds such as the H2receptor antagonist ranitidine have also beenshown to protect the stomach against injury.30The primary protective action of H2 receptor

antagonists presumably stems from their anti-secretory effect, and not from an enhancementof gastroprotective factors.

In this study, indomethacin reduced mucusgel thickness and impaired pHi homeostasisduring acid exposure. The mucus gel layercovering the epithelial surfaces of gastricmucosa constitutes the first line of mucosaldefence against luminal acid. The thinnedmucus gel in indomethacin treated rats iseither caused by diminished mucus synthesisor increased degradation. In previous studies,indomethacin thinned the adherent mucus gel,and decreased the synthetic rate of gastricmucus31 32 rendering the first possibilityplausible. On the other hand, it has beenshown that gastric mucus content is increasedafter fasting, which is assumed to be due to thedecreased mechanical abrasion of the gastricmucosa.33 Therefore, it is also possible thatgastric hypercontraction, which is a knowneffect of ulcerogenic doses of indomethacin,34 35may increase the degradation of the gastricmucus resulting in the thinned mucus gel.Gastric mucosal blood flow, which has beenshown to affect pHi homeostasis duringsuperfusion with acidic solutions22 is known tobe decreased by indomethacin.36 37 Laser-Doppler flowmetry used in this study is a goodtechnique to measure relative blood flowchanges, but is not suitable for the measure-ment of absolute blood flow because itsaccuracy is dependent on the pressure anddirection of the laser probe.38 Although relativegastric mucosal blood flow was not differentamong all of the groups during the experi-ments, absolute mucosal blood flow may bedecreased in indomethacin treated rats. Thedecrease in absolute blood flow would reducedelivery of bicarbonate needed for the preser-vation of pHi during acid exposure.The explanation most consistent with the

finding that bismuth compounds protected thestomach from indomethacin associated injuryis that bismuth, present in high concentrationsin the gastric mucus gel, decreased its acidpermeability. This impaired permeabilitywas manifest as a slowed acidification rateand improved pHi homeostasis during acid

168

Ranitidine bismuth citrate protects gastric mucosa

superfusion, compared with rats treated withindomethacin alone. We assume that the rateof rapid, initial drop ofpHi (initial acidificationrate) is primarily dependent on the rate of acidpermeation through the mucus gel. We havepreviously shown that the initial acidificationrate is inversely correlated to mucus gel thick-ness - that is, the thicker the gel, the slower thediffusion.39 Other homeostatic mechanismsthat change pH1 such as increased Na+/H+exchange are induced relatively slowly, andthus would have little effect on the initial dropin pHi.40 41 Although GG3 11 decreased acidsecretion, luminal pH was held constant by thehigh flow rate of the superfusate. Furthermore,the small change in acid secretory state of themucosa - that is, from basal to fully inhibited -was unlikely to change pHi regulation, as wehave shown previously in cimetidine treatedrats,22 and also in this study in the ranitidineonly group. On the basis of the foregoingconsiderations, the most plausible explanationfor the normal acidification rate despite athinned mucus gel in the indomethacin+GG3 11 group is decreased proton perme-ability of the adherent mucus gel.The mechanism for the decreased perme-

ability of adherent gastric mucus in thepresence of NSAIDs, and high concentrationofbismuth, however, is unclear. Indomethacin,if anything, increased mucus proton perme-ability in and of itself, as the increase in initialacidification rate (56%) in indomethacintreated rats was greater than would have beenpredicted from the 19% fall in gel thickness.*The initial acidification rate in the GG3 11group was unchanged from control rats,suggesting that GG311 alone had no measur-able effect on mucus proton permeability.Thus, high mucus bismuth concentrations,found only in the indomethacin+GG31 1group at three hour time point, correlated withdecreased proton permeability of adherentgastric mucus. This suggests a direct, probablyphysical interaction between bismuth and themucus gel. The nature of this interaction is notclear, although there is experimental evidencesuggesting that divalent cations such as Mg2+,Ca2+, and Fe2+ can alter the physical proper-ties of adherent mucus.4244 Moreover, Leefound that precipitated gastric mucus glyco-protein complexed with colloidal bismuthsubcitrate in vitro produced a mucus-bismuthcomplex that impeded proton permeability.45These in vitro experiments thus provide aplausible basis for hypothesis that bismuth, inhigh concentrations, changes the physicalstructure of adherent gastric mucus so as todecrease its permeability to protons.The reason why indomethacin increased

gastric mucus bismuth concentrations by thesubstantial extent observed is also not obvious.One possibility is that bismuth might beadhering to damaged areas or an inflammatoryexudate in the gastric mucosa. Indeed, it

*Our previous data demonstrated that the per cent change ininitial acidification rate was 68% of the per cent change inmucus gel thickness. We would thus predict that for a 19%decrease in mucus gel thickness, initial acidification rate wouldincrease by only 13%.

has been shown that bismuth binds toinjured gastric mucosa in a human study45 andexperimental ulcer models.25 In this study,however, severe and histological injury werenearly absent under the conditions in whichmucus bismuth concentrations were studied.Recently, several reports have suggested thatthe pathogenesis of indomethacin inducedgastric injury involves gastric hypercontractionand microvascular disturbance in addition to aprostaglandin deficiency.34 35 46 47 Takeuchiet a146 47 reported that an ulcerogenic dose ofindomethacin induced a pronounced increasein frequency and amplitude of gastric con-tractions, which was associated with oscillatoryfluctuations of mucosal blood flow, andincreased extravasation of Evans blue dye.These findings occurred prior to the appear-ance of gastric lesions. Therefore, it is plausiblethat bismuth was bound to an exudate derivedfrom microvessels before the appearance ofgastric lesions. The second prospect is thatindomethacin slowed gastric emptying, allow-ing more contact time between GG3 11 and thegastric mucosa. This latter hypothesis is notsupported by experimental motility studies,which suggest that gastric emptying, if changedat all, is hastened by indomethacin.48 49Furthermore, gastric hypercontraction in-duced by indomethacin may enhance themechanical contact between GG311 and thegastric mucosa, increasing the diffusion ofbismuth particles into the gastric mucus gel.Another explanation is that indomethacinchanged the chemical composition of thegastric mucus gel in such a manner as toincrease binding to cationic metals. Forexample, indomethacin decreased the amountof lipids in rat gastric mucus.50 The NSAIDaspirin also decreased the density of phospho-lipids and the hydrophobicity of canine gastricmucus.51 52 Indomethacin, by decreasing thehydrophobicity and lipid content of the surfaceof the adherent mucus, may have increased theaccessibility of the mucus to exogenousbismuth. The studies with bismuth citrate andranitidine alone strongly suggest that thebismuth, and not the ranitidine component ofGG3 11 decreases the permeability of gastricmucus.The potentiation of bismuth concentrations

in the gastric mucus by indomethacin hassignificance beyond the observed alterations ofacid permeability and protection againstindomethacin induced gastric mucosal injury.Helicobacter pylori, the organism associatedwith chronic gastritis and recurrent pepticulcer disease, resides in the juxtamucosalregion of the gastric mucus. This organism issensitive to bismuth in vitro,53 althoughclinical trials in which bismuth compoundswere used as monotherapy to eradicate Hpylorihave been disappointing.54 55 Furthermore,H pylori infection is associated with decreasedgastric mucus bismuth concentrations inhuman postmortem specimens treated in vitrowith colloidal bismuth subcitrate.21 This studysuggests that the combination of a bismuthcontaining compound such as GG311 com-bined with an NSAID such as indomethacin,

169

170 Tanaka, Guth, Paulsen, Kaunitz

by virtue of its high and prolonged concen-tration of bismuth in mucus, might be alogical combination for the large scaleeradication ofH pylori in selected clinical pop-ulations.

In conclusion, this study indicates thatbismuth concentrations in the gastric mucusare increased during the inapparent gastricinjury induced by indomethacin, and areassociated with impaired acid permeability ofthe mucus gel. Thus, GG311 protects gastricmucosa from indomethacin induced injuryby strengthening pre-epithelial defencemechanisms in addition to inhibiting acidsecretion. As gastroprotective properties ofbismuth compounds may also result frominhibition of pepsin activity, binding of bileacids and epidermal growth factor, and induc-tion of PG synthesis, and because bismuthcompounds also protect the gastric mucosafrom other noxious stimuli such as ethanol andcold restraint stress,2 23 it is possible that othermechanisms may also be involved in thegastroprotective effect ofGG3 11.

We would like to thank Mr Larry Myers and Mr Jerry Snidowof the Glaxo Research Institute for their helpful suggestions andadvice concerning the design of these experiments.This work was supported by a Veterans Administration Merit

Review Award (JDK), the Glaxo Research Institute (JDK), theCURE Experimental Ulcer and Blood Flow Core of NIHDK41301 (PHG), and a CURE Pilot and Feasibility Award(JDK).

1 Konturek SJ, Kwiecien N, Obtulowicz W, Hebzda Z,Oleksy J. Effects of colloidal bismuth subcitrate onaspirin-induced gastric microbleeding, DNA loss, andprostaglandin formation in humans. ScandJ7 Gastroenterol1988; 23: 861-6.

2 Goldenberg MM, Honkomp U, Burrous SE, CastellonAW. Protective effect of Pepto-Bismol liquid on thegastric mucosa of rats. Gastroenterology 1975; 69: 636-40.

3 Slomiany BL, Nishikawa H, Bilski J, Slomiany A. Colloidalbismuth subcitrate inhibits peptic degradation of gastricmucus and epidermal growth factor in vitro. Am JfGastroenterol 1990; 85: 390-3.

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

5 Shorrock CJ, Crampton JR, Gibbons LC, Rees WDW.Effect ofbismuth subcitrate on amphibian gastroduodenalbicarbonate secretion. Gut 1989; 30: 917-21.

6 Ehsanullah R, Page M, Tildesley G, Wood J. Prevention ofgastroduodenal damage induced by non-steroidal anti-inflammatory drugs: controlled trial of ranitidine. BMJ1988; 297: 1017-21.

7 Robinson MG, Griffin JWJ, Bowers J, Kogan FJ, KogutDG, Lanza FL, et al. Effect of ranitidine on gastroduo-denal mucosal damage induced by nonsteroidal anti-inflammatory drugs. Dig Dis Sci 1989; 34: 424-8.

8 Stables R, Campbell CJ, Clayton NM, Clitherow JW,Grinham CJ, McColm AA, et al. Gastric anti-secretory,mucosal protective, anti-pepsin, and anti-helicobacterproperties of ranitidine bismuth citrate. Aliment PharmacolTher 1993; 7: 237-46.

9 Hudson N, Murray FE, Cole AT, Turnbull GM, LettisGM, Hawkey CJ. Ranitidine bismuth citrate and aspirin-induced gastric mucosal injury. Aliment Pharmacol Ther1993; 7: 515-21.

10 Djahanguiri B. The production of acute gastric ulcerationby indomethacin in the rat. Scand_J Gastroenterol 1969; 4:265-7.

11 Satoh H, Guth PH, Grossman MI. Role of food in gastroin-testinal ulceration produced by indomethacin in the rat.Gastroenterology 1982; 83: 210-5.

12 Itoh M, Paulsen G, Guth PH. Hemorrhagic shock and acidgastric injury in the rat. Gastroenterology 1986; 90:1103-10.

13 Kaneko K, Guth PH, Kaunitz JD. In vivo measurement ofrat gastric surface cell intracellular pH. Am JfPhysiol 1991;261: G548-52.

14 Kaunitz JD, Nishixaki Y, Kaneko K, Guth PH. Effect oforogastric nicotine on rat gastric mucosal gel thickness,surface cell viability, and intracellular pH. J PharmacolExp Ther 1993; 265: 948-54.

15 Nishizaki Y, Guth PH, Kaunitz JD. Isoproterenol enhancesrat gastric mucosal barrier function in vivo. Gastro-enterology 1993; 105: 340-6.

16 Nishizaki Y, Guth PH, Quintero E, Del Rivero M, Bover J,Kaunitz JD. Prostaglandin E, enhances gastric defense

mechanisms against acid injury in uremic rats.Gastroenterology 1994; 107: 1382-9.

17 Tanaka S, Kaunitz JD. Indomethacin does not alter theeffect of pentagastrin on rat gastric defense mechanisms.Peptides 1996; 17: 155-9.

18 Gamer A, Heylings JR. Stimulation of alkaline secretion inamphibian-isolated gastric mucosa by 16,16-dimethylPGE2 and PGF2a. Gastroenterology 1979; 76: 497-503.

19 Kerss S, Allen A, Gamer A. A simple method for measuringthickness of the mucus gel layer adherent to rat, frog, andhuman gastric mucosa: influence of feeding,prostaglandin, N-acetylcysteine and other agents. Clin Sci1982; 63: 187-95.

20 Slikkerveer A, Helmich RB, de Wolff FA. Analysis for bis-muth in tissue by electrothermal atomic absorption spec-trometry. Clin Chem 1993; 39: 800-3.

21 Mufioz DJB, Tasman-Jones C, Pybus J. Effect ofHelicobacter pylori on colloidal bismuth subcitrate con-centration in gastric mucus. Gut 1994; 33: 592-6.

22 Nishizaki Y, Guth PH, Kim G, Wayland H, Kaunitz JD.Pentagastrin enhances gastric mucosal defenses in vivo:luminal acid-dependent and independent effects. Am JfPhysiol 1994; 267: G94-104.

23 Goldenberg MM, Honkomp LJ, Castellion AW. Preventionof bismuth subsalicylate of gastric mucosal lesions inresponse to noxious stimuli in rats. Pharmacol Res Comm1978; 10: 13-20.

24 Slomiany BL, Nishikawa H, Bilski J, Slomiany A. Colloidalbismuth subcitrate inhibits peptic degradation of gastricmucus and epidermal growth factor in vitro. Am 7Gastroenterol 1990; 85: 390-3.

25 Koo J, Ho J, Lam SK, Wong J, Ong GB. Selective coatingof gastric ulcer by tripotassium dicitrato bismuthate in therat. Gastroenterology 1982; 82: 864-70.

26 Konturek SJ, Bilski J, Kwiecien N, Obtulowicz W, Kopp B,Oleksy J. De-Nol stimulates gastric and duodenal alkalinesecretion through prostaglandin dependent mechanism.Gut 1987; 28: 1557-63.

27 Mertz-Nielsen A, Steenberg P, Neumark T, Bukhave K,Rask-Madsen J. Colloidal bismuth subcitrate causes sus-tained release of gastric mucosal prostaglandin E2.Aliment Pharmacol Ther 1991; 5: 127-33.

28 Manhart MD. In vitro antimicrobial activity of bismuthsubsalicylate and other bismuth salts. Rev Infect Dis 1990;12: S11-5.

29 Lambert JR. Pharmacology of bismuth-containing com-pounds. Rev Infect Dis 1991; 13 (suppl 8): S691-5

30 Navarova J, Nasalova V. Effect of H2-receptor antagonistson indomethacin-induced lysosomal enzyme release fromrat gastric mucosa. Methods Find Exp Clin Pharmacol1994; 16: 119-24.

31 Torii A, Kameda H, Kawamura T, Onizawa N, Nozawa H,Ariizumi M, et al. Effect of gastric mucus on the uptake ofthe carcinogen MNNG by gastric mucosal DNA. Jf ClinGastroenterol 1990; 12 (suppl 1): S 116-24.

32 Slomiany BL, Piotrowski J, Ismail A, Rajiyah G, Tamura S,Bielanski W, et al. Protection against alcohol-inducedgastric mucosal injury by nitecapone. Gen Pharmacol1991; 22: 1055-62.

33 Ohara S, Hotta K. Effects of fasting on mucus glycoproteinbiosynthesis in rat stomach. Comp Biochem Physiol 1985;82B: 207-10.

34 Mersereau WA, Hinchey EJ. Relationship between thegastric myoelectric and mechanical activity in the genesisof ulcers in indomethacin-insulin-treated rats. Dig Dis Sci1988; 33: 200-8.

35 Ueki S, Takeuchi K, Okabe S. Gastric motility is an import-ant factor in the pathogenesis of indomethacin-inducedgastric mucosal lesions in rats. Dig Dis Sci 1988; 33:209-16.

36 Gerkens JF, Shand DG, Flexner C, Nies AS, Oates JA,Data JL. Effect of indomethacin and aspirin on gastricblood flow and acid secretion. Jf Pharmacol Exp Ther 1977;203: 646-52.

37 Holm L, Jagare A. Role of prostaglandins in regulation ofgastric mucosal blood flow and acid secretion. Am JPhysiol 1992; 263: G446-51.

38 Kiel JW, Riedel GL, DiResta GR, Shepherd AP. Gastricmucosal blood flow measured by laser-Doppler veloci-metry. AmJ Physiol 1985; 249: G539-45.

39 Engel E, Guth PH, Nishizaki Y, Kasunitz JD. Barrierfunction of the gastric mucus gel. Am J Physiol1995; 269: G994-9.

40 Kaneko K, Guth PH, Kaunitz JD NA+IH+ exchangeregulates intracellular pH of rat gastric cells in vivo.Pflugers Arch 1992; 421: 322-8.

41 Seidler U, Carter K, Ito S, Silen W. Effect of C02 on pHiin rabbit parietal, chief, and surface cells. Am J Physiol1989; 256: G466-75.

42 Crowther RS, Marriott C, James SL. Cation inducedchanges in the rheological properties of purified mucusglycoproteins. Biorheology 1984; 21: 253-63.

43 Formtner JF, Jabbal I, Findlay BP, Forstner GG. Interactionof mucins with calcium, H+ ion and albumin. Mod ProbiPediatr 1976; 19: 54-65.

44 Marriot C, Shih CK, Litt M. Changes in the gel propertiesof tracheal mucus induced by divalent cations. BiorheologAy1979; 16: 331-7.

45 Lee SP. A potential mechanism of action of colloidal bis-muth subcitrate: diffusion barrier to hydrochloric acid.ScandJf Gastroenterol 1982; 17 (suppl 80): 17-21.

46 Takeuchi K, Ueshima K, Hironaka Y, Fujioka Y, MatsumotoJ, Okabe S. Oxygen free radicals and lipid peroxidation inthe pathogenesis of gastric mucosal lesions induced byindomethacin in rats. Digestion 1991; 49: 175-84.

Ranitidine bismuth citrate protects gastric mucosa 171

47 Takeuchi K, Ueki S, Okabe S. Importance of gastricmotility in the pathogenesis of indomethacin-inducedgastric lesions in rats. Dig Dis Sci 1986; 31: 1114-22.

48 Stein J, Zeuzem S, Uphoff K, Laube H. Effects ofprostaglandins and indomethacin on gastric emptying inthe rat. Prostaglandins 1994; 47: 31-40.

49 Kohut A, Majzis J. Effect of allopurinol and superoxide dis-mutase on indomethacin-induced gastric lesions in therat. Physiol Res 1993; 42: 273-6.

50 Slomiany A, Mizuta K, Piotrowski J, Nishikawa H,Slomiany BL. Gastric mucosal; protection by sucralfateinvolves phosphoinositides participation. Int J Biochem1995; 22: 1179-83.

51 Goddard PJ, Kao Y-CJ, Lichtenberger L. Luminal surfacehydrophobicity of canine gastric mucosa is dependent on

a surface mucous gel. Gastroenterology 1990; 98: 361-70.52 Kao Y-C, Goddard PJ, Lichtenberger LM. Morphological

effects of aspirin and prostaglandin on the canine gastricmucosal surface. Gastroenterology 1990; 98: 592-606.

53 McNulty CAM, Dent J, Wise R. Susceptibility of clinicalisolates of Campylobacter pyloridis to 11 antimicrobialagents. Antimicrob Agents Chemother 1985; 28: 837-8.

54 Rauws EAJ, Tytgat GN. Campylobacter pylori: treatmentof gastritis. In: Rathbone BJ, Heatley RV, ed.Campylobacter pylori and gastrointestinal disease. Oxford:Blackwell Scientific, 1989: 225-32.

55 Coghlan JG, Tobin A, O'Morian C. Campylobacter pyloriand ulcer treatment. In: Rathbone BJ, Heatley RV, ed.Campylobacter pylon and gastroduodenal disease. Oxford:Blackwell Scientific, 1989: 232-45.