Gut, 1990, 31, 1230-1236

Relation of Helicobacter pylori to the human gastricmucosa in chronic gastritis of the antrum

L L Thomsen, J B Gavin, C Tasman-Jones

AbstractThe spatial relations between bacteria and theaffected tissues can indicate pathogenicmechanisms. This study was undertaken todefine the spatial relation of Helicobacterpylori to the human gastric mucosa. Anti-bodies against gastric mucus and rutheniumred were used to stabilise the glycoproteinstructure of the mucus and glycocalyces inantral biopsy specimens from eight patientsinfected with H pylori. The location oforganisms and ultrastructural features wereassessed using systematic scanning and trans-mission electron microscopy: 92 (2)% (mean(SE)) of H pylori were in the pit mucus, and7 (3)% were in the surface mucus; 60 (12)% ofH pylori were close to epithelial cells, withonly 5 (2)% located near the epithelial inter-cellular junctions. Fine filamentous strandsextended between organisms and nearbyepithelial cells, with few organisms in mem-brane to membrane contact. H pylorn were notobserved between, beneath, or within cells ofthe gastric mucosa. The preferred location ofH pylori in the gastric antrum is within the pitmucus close to the epithelial cell surface, withno evidence that they have a direct toxic effecton the mucosa.

shown by the use of ruthenium red" in thepreparation of samples.The research described in this paper used

systematic microscopy to determine the relationofH pylori to the gastric epithelium in infectedhuman gastric biopsy specimens, in which theglycoprotein structure of the mucus and of theglycocalyces had been stabilised.

Methods

SUBJECTS AND SAMPLESBiopsy specimens from the gastric antrum ofeight subjects infected with H pylori werestudied. The subjects were selected frompatients with dyspeptic symptoms attending theendoscopy clinic at Auckland Hospital. Onlypatients not receiving treatment for dyspepticsymptoms, who gave informed consent, andfrom whom the additional specimens could beobtained with reasonable comfort, were includedin this study.Three antral biopsy specimens (one for urease

detection, one for histopathological assessment,and one for electron microscopy) were obtainedfrom each subject from the region within 5 cm ofthe pyloric valve.

Departments ofPathology and Medicine,University ofAucklandSchool of Medicine,Auckland, New ZealandL L ThomsenJ B GavinC Tasman-JonesCorrespondence to:Lindy L Thomsen,Department of Pathology,University of AucklandPrivate Bag, Auckland,New Zealand.Accepted for publication2 January 1990

Organisms have been observed in the gastricmucosa of humans since 1893' but they excitedlittle interest until 1983 when they were isolatedby Warren and Marshall2 and subsequentlynamed Helicobacter pylori.3 Although studies'have shown a striking correlation between thepresence ofH pylori and chronic gastritis, theirrole in its aetiology remains obscure. Neither adirect toxic effect nor a specific immunologicalreaction have yet been proved.The spatial relations between bacteria and the

affected tissues can indicate whether the patho-genic mechanisms are likely to be direct orindirect.7 For example, an intimate associationwith host cells affords advantage for toxin-producing organisms.8 The relation ofH pylori tothe human gastric mucosa, however, has notbeen carefully studied.A major difficulty in determining the in vivo

location of bacteria in the gut is the collapseduring preparation for histological examinationof the highly hydrated mucus and the artifactualdisplacement of the organisms.9'10 Stabilisationof mucus with anti-mucus antibodies preservesthe mucus layer in situ,9 however, and thusprevents misinterpretation of the relation ofbacteria to the mucosa.'0 Similarly, the glyco-calyces at bacterial and epithelial cell surfaces,which also collapse during preparation for histo-logical examination, may be preserved and

UREASE DETECTIONUrease activity in the biopsy specimens wasdetermined using the urea broth test,'2 and acolour change from orange to pink within 24hours indicated a positive result.

HISTOPATHOLOGICAL ASSESSMENTBiopsy specimens were immediately fixed in 4%neutral buffered formaldehyde. They were thenroutinely embedded in paraffin, and sectionedand stained with haematoxylin and eosin' in thehistopathology laboratory at Auckland Hospital.A histological diagnosis for each case was madeby an experienced histopathologist and thechronic gastritis was graded according to thecriteria of Whitehead et al," which are based onthe numbers of inflammatory cells observedrelative to the numbers of inflammatory cellspresent in histologically normal sections. Thegrades were: Normal= normal numbers ofmono-nuclear cells and polymorphonuclear leucocytes.Mild=increased mononuclear cells and veryoccasional polymorphonuclear leucocytes.Moderate=increased mononuclear cells and poly-morphonuclear leucocytes. Severe=markedlyincreased mononuclear cells and polymorpho-nuclear leucocytes, with intraepithelial poly-morphonuclear leucocytes.

Additional sections were stained with the

1230

on 4 July 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.31.11.1230 on 1 Novem

ber 1990. Dow

nloaded from

Relation ofH pylori to gastric mucosa

Giemsa stain'4 and independently examined forthe presence of Helicobacter-like organisms.

ELECTRON MICROSCOPYThe mucus layer was stabilised using a methodsimilar to that described by Bollard et a19 forcolon mucus. Antibodies were raised againstgastric mucus obtained from a histologicallynormal human stomach at necropsy.'I Thepresence ofantibodies to human gastric mucus inthe rabbit serum was confirmed using an enzymelinked immunosorbent assay (ELISA) tech-nique. An aliquot of a homogenate of gastricmucus in phosphate buffered saline (PBS) (1:9)was incubated in the wells of an ELISA plate(Nunc Intermed, Denmark) for 12-18 hours at4°C, followed by incubation with PBS containing2% bovine serum albumin (Serva, Germany) and2% gelatine for two hours at 20°C. Next, thediluted rabbit serum was added and incubatedfor three hours at 20°C, followed by the additionof horseradish peroxidase labelled goat-antirabbit IgG (South Pacific ImmunologicalLaboratories Ltd, New Zealand), and incubatedfor two hours at 20°C. The plate was washedthree times with PBS containing 0 05% tween 20(Serva, Germany) between each incubation step.After the addition of 0-phenyldiamine dihydro-chloride (Sigma, USA), the absorbances of eachwell were read on an ELISA plate reader (EAR,SLT Labinst, Austria) at 486 nm. The resultsfrom the serum of the immunised rabbit wereexpressed as a ratio to those from serum froma control rabbit. Complement activity wasremoved from the gastric mucus antiserumimmediately before use by incubation at 56°C for30 minutes.Each biopsy specimen was carefully placed on

LS

Figure 1: Diagram ofasection through the gastricantral mucosa, showing thethree regions (S,P,G) ofmucus examinedfor thepresence ofH pylori.S=surface region ofmucus,P=pit region ofmucus,G=gland region ofmucus,LS= gastric luminal surface,SE=surface epithelial cells,PE =pit epithelial cells,GE=glandular epithelium,LP=lamina propria.

to dental wax, and the entire mucosal surfacecovered with undiluted rabbit antiserum for onehour at 4°C.The biopsy specimens were then prepared for

electron microscopy using a method based onthat described by Mackie et al." They were fixedby immersion in a 0 5% glutaraldehyde solutioncontaining 0 15% ruthenium red in 0 1 M caco-dylate buffer (pH 7 4) for 30 min at 20°C, then in5% glutaraldehyde with 0-05% ruthenium red inthe same buffer for 18-24 hours at 4°C. Thebiopsy specimens were washed in 0-05%ruthenium red in 0-1 M cacodylate buffer at pH7-4 (three rinses at 10 minute intervals), thenpostfixed for two hours in 0-66% osmiumtetroxide in the same buffer containing 0-1%ruthenium red. After washing a further threetimes (10 minute intervals) in 0 05% rutheniumred in 0 1 M cacodylate buffer (pH 7 4), theywere dehydrated in increasing concentrations(10 minutes each step) of ethanol (50, 70, 90,100%, 2X anhydrous) and two changes of propy-lene oxide (10 minutes each step). Next, theywere placed in a propylene oxide:epoxy resin(EMBED-812, Electron Microscopy Sciences,Fort Washington, USA) (50:50) mixture for twohours, then transferred to a propylene oxide:epoxy resin (25:75) mixture for a further twohours, before being placed in 100% epoxy resinovernight under vacuum. Finally, they wereembedded in flat molds and the resin curedovernight at 60°C under vacuum.

SCANNING ELECTRON MICROSCOPYSamples were prepared for scanning electronmicroscopy using the method described byThorne et al.'6 Sections 2 Fm thick were cutfrom the resin blocks and examined, using a lightmicroscope, for correct orientation before beingmounted on pieces of carbon coated glass andstained with saturated uranium acetate'7 (15 min)and lead citrate'8 (15 min). After mounting onscanning electron microscopy stubs, the sectionswere examined using a back-scattered electrondetector'9 and an accelerating voltage of 30 kV inan ISI DS-130 scanning electron microscope.'6The polarity of the image was reversed to givephotomicrographs that resembled thoseobtained by transmission electron microscopy.The numbers ofH pylon were counted within

the mucus, which was divided into surfacemucus, pit mucus, and gland mucus (Fig 1).

Within each region, the numbers ofH pylon'organisms located close to the epithelial cellsurface were noted. An organism was consideredclose to the epithelial cell surface when all or partof the organism was observed within the zoneoccupied by the microvilli on the surface of theepithelial cells. The numbers ofH pylori locatedwithin 1 [im of the interepithelial cell junctionswere also noted.

TRANSMISSION ELECTRON MICROSCOPYThin sections (90 nm), from selected areas ofinterest identified in the sections examined byback-scattered electron imaging, were cut on aReichert ultramicrotome using glass or diamondknives. These were collected on copper grids,

1231

on 4 July 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.31.11.1230 on 1 Novem

ber 1990. Dow

nloaded from

Thomsen, Gavin, Tasman-Jones

* e% . 'i *TFigure 2: Gastric pitsfrommhe antrum oj SuOjecLs witl stained with uranium acetate'7 and lead citrate,'8moderate chronic gastritiscontaining many H pylori. and examined at magnifications up to 140 000(A) Showing epithelial cells times with a Philips EM 410 LS transmissionofnormal appearance filled electron microscope operated at an acceleratingwith mucus granules (m).Some organisms are grouped voltage of 60 kV.together as in a colony.(Scanning electronmicroscope. Originalmagnification x4200). (B) ResultsShowing epithelial cells with All eight subjects had positive urea broth tests,ragged surfaces (arrow) and Helicobacter-like organisms were observed indepleted ofmucus granules(m). (SEM. Original Giemsa stained sections. Histopathologicalmagnification x3800). assessment showed that three subjects had mild,

four had moderate, and one had severe chronicgastritis. Gastritis was noted as non-active - thatis no polymorphonuclear leucocytes observed -for one subject with mild chronic gastritis, andintestinal metaplasia was noted in one subjectwith moderate chronic gastritis.

SCANNING ELECTRON MICROSCOPYThe gastric mucus in all sections formed acomplete layer over the epithelium and H pyloriwere observed as moderately densely stainedbodies within it (Fig 2). Organisms sectionedlongitudinally were readily distinguished bytheir characteristically curved shape, but thosein cross section were sometimes more difficult toidentify positively. Organisms were observed inseven of the eight biopsy samples examined. Theremaining one showed incomplete intestinalmetaplasia. The organisms were unevenlydistributed within each section (Fig 2A) and

Numbers and distribution ofH pylori in biopsy specimens from seven patients

Location (% oftotal)

NearGrade of Total no Surface Pit Gland Close to intercellulargastritis observed mucus mucus mucus epithelium junctions

Mild 19 0 100 0 0 0Mild 103 10 90 0 64 4Mild 550 14 86 0 90 4Moderate 111 10 90 0 58 11Moderate 280 5 95 0 64 3Moderate 1347 12 85 3 51 6Severe 67 0 100 0 93 9

occasionally they were grouped as in a colony(Fig 2A).The number ofH pylon in each section varied

from 19-1347 (Table). Of the total number ofH pylon counted, 92 (2)% (mean (SE)) werepresent in the mucus within the gastric pits and7 (3)% were in the surface mucus. Only inone subject, that with the greatest number oforganisms, were organisms seen in the glandmucus, and these made up 3% of the total(Table).The proportion of H pylon close to the

epithelial cell surface was 60 (12)% (Table). Insix subjects, the proportion ranged from 51-93%. In the remaining subject, in which thesmallest number of organisms were observed,none were located close to the epithelial cellsurface. In the subject in which organisms wereseen in the gland mucus, 50% of these were closeto the epithelial cell surface. Only 5 (1)% ofH pyloni were located near the epithelial inter-cellular junctions.There was no apparent relation between the

number and location of H pylon organismswithin any region of mucus and the appearanceof the nearby epithelial cells. Large numbers oforganisms were sometimes present in regionswith epithelial cells of normal appearance,packed with mucus granules and with bulgingcell surfaces (Fig 2A), as well as in regions inwhich the epithelial cells had ragged surfaces

Figure 3: Gastric pitsfrom the antral regions ofsubjects withmoderate chronic gastritis. (A) Despite the presence ofH pylori (small arrows) only occasional inflammatory cells arepresent in the lamina propria (large arrow). (SEM. Originalmagnification x2300). (B) Although there are increasednumbers ofinflammatory cells (arrow) in the lamina propria,no H pylori organismis are present in the adjacent mucus.(SEM. Original magnification x3100).

1232

on 4 July 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.31.11.1230 on 1 Novem

ber 1990. Dow

nloaded from

Relation ofH pylori to gastric mucosa

(Fig 2B) or were depleted ofmucus granules (Fig2B).There was no apparent relation between the

number and location of H pylori organismswithin a region of mucus and the inflammatorycell infiltrate within the lamina propria beneath.Normal inflammatory cell numbers were some-

times observed in the lamina propria when largenumbers of organisms were present in theadjacent mucus (Fig 3A), and conversely, denseinflammatory cell infiltration of the laminapropria was sometimes observed when no

organisms were present in the adjacent mucus

(Fig 3B).There was no apparent relation between the

number of H pyloni present and the histologicalgrade of gastritis observed (Table).

TRANSMISSION ELECTRON MICROSCOPYH pylori organisms appeared as densely stainedcurved, oval, or circular bodies within the finelygranular mucus (Fig 4A). The mucus surround-ing some micro-organisms (Fig 4A) and that inthe vicinity of the epithelial microvilli (Fig 5)appeared less granular than the remainder andin one subject it contained round, membranebound, glycocalyceal bodies (Fig 4A). Organismslocated close to epithelial cell surfaces often layentirely within this electron-lucent zone (Fig 5).

Bacterial glycocalyx material was observed atthe surface of all organisms (Fig 4A). Finefilamentous strands of glycocalyx-like materialextended between the cell membranes of theorganisms and between that oforganisms and thenearby epithelial cells (Figs 4B, 6B). Manyorganisms lay close to epithelial cell membranes,but relatively few organisms were in membraneto membrane contact with epithelial cell surfaces(Figs 6A and B).

In some cases, however, where organisms layclose to or contiguous with epithelial cell sur-

faces, focal displacement and depletion of micro-villi had occurred (Figs 5 and 6A). In only oneinstance was there any alteration within theepithelial cell immediately beneath the contactpoint with an organism (Fig 6B).H pylon organisms were never observed

between, beneath, or within any cells of thegastric mucosa.

DiscussionThis investigation has shown that the preferredlocation ofH pylon in the human gastric antrumis close to the surface of the epithelial cells liningthe gastric pits, with no preference for the regionclose to intercellular junctions. Apart from somedistortion and depletion of microvilli, there wasno other evidence that the organisms had causedpathological changes in the mucosal epithelialcells, even when they were in membrane tomembrane contact with each other.

This study seems to be the first systematicinvestigation of the relation ofH pylori to gastricepithelial cells and of their distribution in thegastric mucus. An essential feature of the investi-gation was the antibody stabilisation of thegastric mucus before tissue dehydration andpreparation for electron microscopy. Bollardet al'° have clearly shown distinct differences inthe location of micro-organisms in the colon,depending on the method of tissue preparation.When conventional methods were used themucus layer collapsed during dehydration andorganisms were observed within the crypts andclose to, or in contact with, the epithelium. But,in all samples in which mucus was stabilised, theorganisms were separated from the epitheliumby a thick layer of mucus.The present study also used scanning electron

microscopy and back-scattered electronimaging,'6 '9 which do not appear to have beenused previously to study gastric mucosa. Thescanning photomicrographs used in this studywere produced using reversed polarity so thatthey resembled those produced by transmissionelectron microscopy. The use of back-scatteredelectrons to study resin sections through wholebiopsy specimens was based on the techniquesdeveloped by Thorne et al"6 to study the cochlea.The method allows detailed study of large speci-mens without the loss of orientation associatedwith the small samples required for transmissionelectron microscopy, but the resolution obtainedat higher magnifications (over 10000 x) is not asgood because of the thickness (2 [tm) of thespecimen and the method of image formation.The method enabled accurate counting of thenumber of organisms present rather than thesemiquantitative estimation possible with lightmicroscopy.'2 In only two sections, in whichthere were large numbers of organisms clumpedtogether, was it difficult to count individualmicro-organisms.That most organisms were located in the

gastric pits (mean (SE) 92 (2)%), and most (60(12)%) close to the epithelial cell surface suggeststhat their location was not random. It seems thatthe pit mucus offers some environmental advant-ages, perhaps protection to organisms from the'bactericidal gastric juice.'21H pyloni are motile in a viscous environment,2'

and thus the relatively few organisms observed inthe surface mucus may represent those relocat-ing from one gastric pit to another. Alterna-tively, they may have only recently penetratedthe mucus from the luminal contents, and havebeen in the process of moving towards theirpreferred niche.7The tendency for ,H pylon to lie close to the

epithelial surface was not as marked, with 40%

Figure 4: H pylori in the pitregion ofgastric antralmucosafrom subjects withmoderate chronic gastritis.(A) An organism lying in thefinely granular mucus (M)but separatedfrom it by a lessgranular zone containingglycocalyced bodies (smallarrows). The glycocalyx is aflocculent layer coating thelimiting membrane (largearrow) ofthe organism.(Transmission electronmicroscope. Originalmagnification x40400). (B)An organism close to andlinked with an epithelial cell(E) by radiating strands(small arrow) which link theglycocalyx (large arrows) ofeach ofthem. (TEM.Original magnification x57800).

1233

on 4 July 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.31.11.1230 on 1 Novem

ber 1990. Dow

nloaded from

Thomsen, Gavin, Tasman-Jones

Figure 5: H pylori in the pitregion in the gastric antralmucosa from a subject withmoderate chronic gastritis.The organisms lie close to thesurfaces ofepithelial cells inthe vicinity ofanintercellularjunction (curvedarrow) in a zone which is lessgranular than the mucusabove (M). The epithelialcell membrane and mucusgranules are intact but themicrovilli (arrow) aredisplaced and reduced innumber. (TEM. Originalmagnification x40200).

.V.; .

.40. .

'C

t4.K .

.J.. 4.', -I.

lying outside the zone of the microvilli. Althoughthis latter group may also have includedorganisms relocating to their preferred site, theirrelatively high proportion and the fact that thefew colonies of organisms which were observedalso lay outside the zone of the microvilli argueagainst this. Nevertheless, the zone adjacent tothe epithelial cell surface appears to be thepreferred location of the majority ofH pylori.Only 5% of organisms lay within 1 ,m of

epithelial cell junctions, a proportion muchlower than the 80% reported by Hazell et al,"who counted micro-organisms within 2 ,um ofthe junctions. In the present material, gastricepithelial cells were up to 5 iim in width so that a2 pm zone about each junction would encom-pass 80% of the entire cell surface.

Others20 have also shown many H pylori

organisms, however, often arranged in groups,close to intercellular junctions, which may beattributable to artefactual relocation during pre-paration'" as such groups were never observed inthe stabilised mucus in the present study.Furthermore, those organisms with the closestassociation to epithelial cells, membrane tomembrane contact, were never observed toimpinge on interepithelial cell junctions. Theclaim2' that H pylori accumulate at the inter-cellular junctions to use nutrients available therewas not supported by the present study.Given the wide range in the numbers of

organisms present, it was interesting that thishad no apparent effect on the grade of chronichistological gastritis. This has also been indicatedby others.22-25 Similarly, there was no correlationbetween the location of the organisms and either

1234

on 4 July 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.31.11.1230 on 1 Novem

ber 1990. Dow

nloaded from

Relation ofH pylori to gastric mucosa 1235

-. .*

S

4' 4:

-4'

.#

44.

BFigure 6: H pylori close toepithelial cells (E) in the pitregion ofgastric antralmucosa from subjects withmoderate chronic gastritis.(A) Organisms showingextensive apposition (arrow)ofbacterial and cellmembranes in a regiondevoid ofmicrovilli. (TEM.Original magnification x19100). (B) An organismshowing glycocalycealfilaments (small arrow) andan intraepithelialvacuolation (large arrow)beneath a region ofmembrane to membraneapposition. (TEM. Originalmagnification x58300).

the morphology of adjacent epithelial cells or thedegree of the subjacent inflammatory cellinfiltrate. This suggests that the gastritis associ-ated withHpylon is not due to a direct cytopathiceffect of the organisms on the mucosa.A difficulty with studies of host/pathogen

relations is that the time of initial infection oftenis not known. H pylori infection does not alwaysproduce symptoms,6 and thus the patient maynot present for endoscopic investigation for sometime. It is also possible that the nature of thehost/pathogen relation changes with time. Thepresent study may have included subjects atdifferent phases of the disease. For one sub-ject, however, the time of infection has beenaccurately defined,4 and a biopsy specimen wasobtained two years after self inoculation andincluded in the present study. This sampleshowed severe chronic histological gastritis butvery few organisms present, suggesting that onceestablished the disease may require only minimalnumbers of organisms to persist.

All those biopsy specimens which showedH pylon, also contained mixed inflammatory cellinfiltrates in the lamina propria, although, hereagain, there was no clear quantitative relationbetween those factors. While the presence ofplasma cells and lymphocytes in all cases isconsistent with an immunological response tothe organisms,2627 it seems that such a responseis largely inappropriate or ineffective againstH pylori organisms in the gastric mucosa.The strand like material observed connecting

the glycocalyces ofH pylori with each other andwith epithelial cells may have developed becauseof attractive forces between the glycocalyxmaterial at these two surfaces.28 They arebelieved to be involved in adherence betweenbacteria and epithelial cell surfaces.82829 Whilethe ruthenium red used in the preparation ofsamples would have enhanced visualisation ofthese connections, other forms of connectionsuch as pili may not have been resolved.28Bode et al20 suggested that, after initial

location close to an epithelial cell surface, theadhesion of H pylori to the apical membrane isthe final step in the process of association withthe mucosa. Furthermore, Zafriri et al30 havepostulated that in general, adhesion of bacteria to

cells directs the diffusion of toxins, and enhancesthe pathogenic activity of the organisms.Bode et a120 also believe that adhesion of

H pylon is followed by subcellular alterations tothe affected epithelial cells, including depletionof microvilli, dissolution of mucous granulesresulting in translucent empty structures,and increased generation of sialic acid-richglycoproteins. Observations in the present studydo not support such direct bacterial action.Relatively few organisms were observed lyingvery close to cell surfaces, and such epithelialcells appeared unaltered apart from displace-ment or depletion of some microvilli. Thereasons why a few bacteria develop such a closeassociation with epithelial cells remain obscure.The present study indicates that virtual

membrane-to-membrane contact, accumulationat intercellular junctions, and incorporation intoepithelial cells or the lamina propria are notimportant features of the pathogenic action ofH pylon. It seems likely that physicochemicalalterations in the protective gastric mucus layermay be more important.'5

This research was supported by the Ruth Spencer MedicalResearch Fellowship Trust.We would like to acknowledge the assistance of Dr I Hamilton,

Dr M Miller, Dr J McKay, Mr M A Vanderwee; Mr A Ellis forassistance with preparation ofFigure 1; and Mrs K Siow for typingthe manuscript.

1 Bizzozero G. Ueber die schaulchformigen drusen desmagendarmkanals und die bezienhungen ihres epithels zudem oberflachenepithel der schleimhaut. Arch F Mikr Anast1893; 42: 82.

2 Warren JR, Marshall B. Unidentified curved bacilli on gastricepithelium in active chronic gastritis. Lancet 1983; i: 1273-5.

3 Marshall BJ, Goodwin CS. Revised nomenclature ofCampylobacter pyloridis. IntJ Systematic Bacteriol 1987; 37:68.

4 Morris A, Nicholson G. Ingestion of Campylobacter pyloridiscauses gastritis and a raised fasting gastric pH. Am JGastroenterol 1987; 82: 192-9.

5 Morris A, Maher K, Thomsen L, Miller M, Nicholson G,Tasman-Jones C. Distribution of Campylobacter pylori in thehuman stomach obtained at postmortem. Scand J Gastro-enterol 1988; 23: 257-64.

6 Blaser MJ. Gastric Campylobacter-like organisms, gastritis,and peptic ulcer disease. Gastroenterology 1987; 93: 371-83.

7 Freter R. Mechanisms of association of bacteria with mucosalsurfaces. In: Adhesion and microorganism pathogenicity. (CibaFoundation Symposium 80.) Tunbridge Wells: PitmanMedical, 1981: 36-55.

8 Beachey EH. Bacterial adherence: adhesin-receptor interac-tions mediating the attachment of bacteria to mucosalsurfaces.J Intern Dis 1981; 143: 325-45.

9 Bollard JE, Vanderwee MA, Smith GW, Tasman-Jones C,Gavin JB, Lee SP. Preservation of mucus in situ in the ratcolon. DigDis Sci 1986; 31: 1388-44.

10 Bollard JE, Vanderwee MA, Smith GW, Tasman-Jones C,Gavin JB, Lee SP. Location of bacteria in the mid-colon ofthe rat. Appl Environ Microbiol 1986; 51: 604-8.

11 Mackie EB, Brown KN, Lam J, Costerton JW. Morphologicalstabilization ofcapsules ofgroup B streptococci, types Ia, Ib,II, and III, with specific antibody. J Bacteriol 1979; 138:609-17.

12 Hazell SL, Borody TJ, Gal A, Lee A. Campylobacter pyloridisgastritis. I: Detection of urease as a marker of bacterialcolonization. AmJ Gastroenterol 1987; 82: 292-6.

13 Whitehead R, Truelove SC, Gear MWL. The histologicaldiagnosis ofchronic gastritis in fibreoptic gastroscope biopsyspecimens. J Clin Pathol 1972; 25: 1-11.

14 Gray SF, Wyatt JI, Rathbone BJ. Simplified techniques foridentifying Campylobacter pyloridis. J Clin Pathol 1986; 39:1279.

15 Thomsen L, Tasman-Jones C, Morris A, Wiggins P, Lee S,Forlong C. Ammonia produced by Campylobacter pylon'neutralises H+ moving through gastric mucus. Scand JGastroenterol 1989; 24: 761-8.

16 Thorne PR, Vuicich TE, Gavin JB. Back-scattered electronimaging of sections through the cochlea: a new technique forstudying cochlear morphology. Stain Technol 1987; 62:191-9.

17 Watson ML. Staining of tissue sections for electron micro-scopy with heavy metals. J Biophys Biochemn Cytol 1958; 4:475-8.

18 Sato T. A modified method for lead staining of thin sections.J ElectronMicrosc 1967; 16: 133.

19 Robinson VNE. Imaging with backscattered electrons in ascanning electron microscope. Scanning 1980; 3: 15-26.

on 4 July 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.31.11.1230 on 1 Novem

ber 1990. Dow

nloaded from

1236 Thomsen, Gavin, Tasman-Jones

20 Bode G, Malfertheiner P, Ditschuneit H. Pathogenic implica-tions of ultrastructural findings in Campylobactor pylorirelated gastroduodenal disease. Scandj Gastroenterol 1988;23 (suppl 142): 25-39.

21 Hazell SL, Lee A, Brady L, Hennessy W. Campylobaterpyloridis and gastritis: association with intercellular spacesand adaptation to an environment of mucus as importantfactors in colonization of the gastric epithelium. J Infect Dis1986; 153:658-63.

22 Price AB, Levi J, Dolby JM, et al. Campylobacter pyloridis inpeptic ulcer disease: microbiology, pathology, and scanningelectron microscopy. Gut 1985; 26: 1183-8.

23 Johnston BJ, Reed PI, Ali MH. Campylobacter-like organismsin duodenal and antral endoscopic biopsies: relationship toinflammation. Gut 1986; 27: 1132-7.

24 Rollason TP, Stone J, Rhodes JM. Spiral organisms inendoscopic biopsies of the human stomach. J Clin Patlhol1984; 37: 23-6.

25 Jones DM, Lessells AM, Eldridge 1. Campvlobacter-like

organisms on the gastric mucosa: culture, histological, andserological studies. J Clin Pathol 1984; 37: 1002-6.

26 Morris A, Nicholson G, Lloyd G, Haines D, Rogers A, TaylorD. Seroepidemiology of Campylobacter pyloridis. NZ MedJ1986; 99: 657-9.

27 Wyatt JI, Rathbone BJ, Heatley RV. Local immune responseto gastric campylobacter in non-ulcer dyspepsia. J ClinPathol 1986; 39: 863-70.

28 Costerton JW, Irvin RT, Cheng K-J. The role of bacterialsurface structures in pathogenesis. CRC Crit Rev Microbiol1981; 8: 303-38.

29 Davis CP, Avots-Avotins AE, Fader RC. Evidence for abladder cell glycolipid receptor for Escherichia coli and theeffect of neuraminic acid and colominic acid on adherence.InfectImmun 1981; 34: 944-8.

30 Zafriri D, Oron Y, Eisenstein BI, Ofek I. Growth advantageand enhanced toxicity of Escherichia coli adherent to tissueculture cells due to restricted diffusion of products secretedby the cells. J Clin Invest 1987; 79: 1210-6.

on 4 July 2019 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.31.11.1230 on 1 Novem

ber 1990. Dow

nloaded from