Proc. Natl. Acad. Sci. USAVol. 93, pp. 1259-1264, February 1996Microbiology

Helicobacter pylori attachment to gastric cells inducescytoskeletal rearrangements and tyrosinephosphorylation of host cell proteinsELLYN D. SEGAL*t, S. FALKOW*t, AND L. S. TOMPKINS*t§*Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA 94305; tDigestive Disease Center, Stanford UniversitySchool of Medicine, Stanford, CA 94305; §Department of Medicine, Stanford University School of Medicine, Stanford, CA 94305; and tRocky MountainLaboratory, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, MT 59840

Contributed by Stanley Falkow, October 18, 1995

ABSTRACT The consequences of Helicobacter pylori at-tachment to human gastric cells were examined by transmis-sion electron microscopy and immunofluorescence micros-copy. H. pylori attachment resulted in (i) effacement of mi-crovilli at the site of attachment, (ii) cytoskeletal re-arrangement directly beneath the bacterium, and (iii)cup/pedestal formation at the site of attachment. Double-immunofluorescence studies revealed that the cytoskeletalcomponents actin, c-actinin, and talin are involved in theprocess. Immunoblot analysis showed that binding ofH. pylorito AGS cells induced tyrosine phosphorylation of two host cellproteins of 145 and 105 kDa. These results indicate thatattachment of H. pylori to gastric epithelial cells resemblesthat of enteropathogenic Escherichia coli. Coccoid H. pylori,which are thought to be terminally differentiated bacterialforms, are capable of binding and inducing cellular changes ofthe same sort as spiral H. pylori, including tyrosine phos-phorylation of host proteins.

Helicobacterpylori has succeeded where many other pathogenshave failed. It can establish itself in an environment wherethere is little competition from other microorganisms and canremain in its niche for decades before its host exhibits anyserious effects. Although spiral bacteria were observed ingastric biopsies for years, H. pylori was not cultivated oridentified until just 12 years ago, when it was discovered thatthis bacterium was the causal agent of type B gastritis, pepticulcers, and gastric cancer (for a review, see ref. 1).

H. pylori is a spiral, Gram-negative rod that attaches spe-cifically to gastric epithelial cells lining the antrum of thestomach. It is highly motile by means of one or more polar-sheathed flagella, each containing a terminal bulb. Its shapeand motility permit the microbe to maneuver easily throughthe gastric mucus layer. Gastric biopsies show H. pylori withinand beneath the mucus layer, in close proximity to the surfaceof the gastric epithelial cells, attached to the gastric cells, and,occasionally, within mucus-secreting gastric cells. Coccoidforms of H. pylori have been observed both in vivo and in vitro(2-4). The coccoid form is considered to be nonviable byculture; it is not clear whether it serves any function in thepathogenesis of infection.The nature of H. pylori attachment to cells has been con-

troversial. Some investigators have reported that H. pyloribinding, which causes microvilli effacement, actin rearrange-ment, and pedestal formation (5-9), is similar to that observedfor "attaching and effacing" Escherichia coli (EPEC) (10).Other workers assert that H. pylori attachment does not resultin pedestal formation or actin rearrangement (11).We report here that H. pylori attachment to AGS cells

clearly is associated with cytoskeletal rearrangement. We

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

further demonstrate that phosphorylation of host cell proteinsoccurs in the immediate vicinity of bacterial attachment.Immunoblot analysis shows that binding of H. pylori to AGScells induces tyrosine phosphorylation of two host cell proteinsof 145 and 105 kDa.

MATERIALS AND METHODSBacterial Strains and Cell Lines. H. pylori strain 87A300,

which is a human clinical isolate that produces the vacuolatingcytotoxin (vacA) and cytotoxin-associated protein (cagA), wasobtained from the State of California Department of HealthServices, Berkeley, CA. It was grown as described (5). Briefly,H. pylori was passaged on either 5% sheep blood plates (TSAII; BBL) or on plates of brucella agar (Difco) to which hadbeen added 5% fetal bovine serum (FBS; Gibco). The plateswere incubated in a BBL GasPak jar containing an anaerobicgas pack (without a catalyst) or in a 5% C02/95% airincubator. Liquid cultures of H. pylori were grown by suspend-ing at least one-quarter of a 2- to 4-day-old plate of H. pyloriinto 30 ml of brucella broth/5% FBS. The flask was placed intoa GasPak jar that contained a GasPak anaerobic systemenvelope (without a catalyst) and was grown with agitation (80rpm) at 37°C. To obtain H. pylori cultures that were coccoid inmorphology, but were still culturable, an overnight liquidculture was grown as described above, placed in a standard37°C incubator, and grown for an additional 2 days withshaking. AGS cells (ATCC CRL 1739, a human gastric ade-nocarcinoma epithelial cell line) were grown in Dulbecco'smodified Eagle's medium (DMEM)/10% FBS.

Antibodies. Rabbit polyclonal anti-H. pylori antibodies wereproduced against whole heat-killed H. pylori strain 87A300.Monoclonal anti-talin and monoclonal anti-a-actinin wereobtained from Sigma Immuno Chemicals (St. Louis, MO).Monoclonal anti-phosphotyrosine antibody was purchasedfrom Upstate Biotechnology (Lake Placid, NY) and Trans-duction Laboratories (Lexington, KY). Anti-mouse IgG crys-talline tetramethylrhodamine isothiocyanate conjugate andanti-rabbit IgG fluorescein isothiocyanate conjugate wereobtained from Sigma Immuno Chemicals (St. Louis, MO).Working dilution of the polyclonal anti-H. pylori antibody was1:100. Working dilutions of all other antibodies were asdetermined by the manufacturer.Immunofluorescence (IF). For IF studies, 5 x 106 AGS cells

were grown on 12-mm glass coverslips (Bellco, Vineland, NJ).Monolayers were washed two times with PBS, and 1 x 107bacteria were added per well in a final vol of 200 ,ul ofDMEM(multiplicity of infection, 2). The plates were incubated. Atappropriate time points, the wells were aspirated, washed fivetimes with PBS (pH 7.4) to remove nonadherent bacteria, and

Abbreviations: EPEC, enteropathogenic Escherichia coli; IF, immu-nofluorescence.tTo whom reprint requests should be addressed.

1259

Dow

nloa

ded

by g

uest

on

June

10,

202

1

Proc. Natl. Acad. Sci. USA 93 (1996)

processed for IF. Fluorescence microscopy was performed ona Nikon Optiphot. Laser scanning confocal microscopy wasperformed with a Bio-Rad MRC 600.

Actin Staining. Actin condensation was revealed by usingrhodamine phalloidin, which labels filamentous actin (6).Transmission Electron Microscopy. H. pylori-infected AGS

monolayers were washed two times in PBS and fixed for 15-35min at 4°C in 2% gluteraldehyde. The monolayer was thenrinsed two times with PBS for 2 min each and incubated in 1%OS04 for 30 min at room temperature. The cells were washedtwo times with double-distilled H20 for 10 min each time andpostfixed in 1% aqueous uranyl acetate for 15 min. Uponwashing the cells two times with double-distilled H20, thesamples were dehydrated as described (5). The samples wereinfiltrated in ethanol/poly/bed812 (1:1) for 5 min and thenpoly/bed812 (100%) for 5 min. For embedding, a gelatincapsule was filled with a plastic embedding medium and themonolayer was drained of 100% resin. The capsule wasinverted on top of the area over the cells. The coverslipconnecting the inverted gelatin capsule was placed in a 60°Coven for 24 hr to polymerize. Appropriate sections wereobtained with an Ultra Cut microtome and examined in aPhilips 201 transmission electron microscope.Immunoblot Analysis. AGS cells were cultured overnight in

DMEM. The cells were washed once with PBS (pH 7.4) and4 ml of fresh medium was added to each dish. H. pylori (5 X107) were removed from a liquid overnight culture and addedto 5 x 106 AGS cells. After incubation in a 5% C02/95% airincubator for 2 hr, cell lysates were made with RIPA buffer andstored at -80°C until needed. Whole-cell lysates of H. pyloriwere made by pelleting the bacteria and suspending the pelletin an equal amount of PBS and 2x SDS lysis buffer (250 mMTris HCl, pH 6.8/4% SDS/20% glycerol/0.002% bromophe-nol blue/10% 2-mercaptoethanol). SDS/PAGE and electro-transfer were performed as described (5). Anti-phosphoty-rosine binding and detection were performed with the ECLsystem (Amersham, Buckinghamshire, England) according tothe manufacturer's instructions.Immunogold Labeling. Grids were floated, section side

down, on PBS (pH 7.4) for 10 min and then washed succes-sively with 0.1 M glycine for 10 min, 2% bovine serumalbumin/PBS for 10 min, and washed four times with PBS for5 min each before fixation with 2% glutaraldehyde for 1 min.Finally, the grids were washed four times for 1 min each withdouble-distilled H20 and then the sections were stained with1% uranyl acetate and lead citrate.

RESULTS

Binding ofH. pylori to Human Gastric Cells Results in ActinRearrangement and Pedestal Formation. AGS cells, grown toa subconfluent monolayer on a glass coverslip, were placed ina perfusion chamber mounted on the stage of a Nikon Diaphot200 microscope. Motile, spiral-shaped H. pylori were added ata multiplicity of infection of 1. Observed by Nomarski videomicroscopy, initial contact and attachment occurred veryquickly (within a few minutes). Attachment invariably oc-curred at the aflagellated end or nose of the bacterium.Subsequently, the entire length of the bacterium was seen inintimate contact with the AGS cell. After attachment, manybacteria were observed to be internalized by the host cell (datanot shown).The attachment of H. pylori to cultured human gastric

adenocarcinoma cells (AGS; ATCC CRL 1739) was studiedfurther by transmission electron microscopy. Figs. 1 and 2illustrate the interaction of H. pylori strain (87A300) to AGScells after 2 and 24 hr of incubation, respectively. Attachmentof H. pylori resulted in (i) effacement of microvilli at the siteof attachment, (ii) actin condensation directly beneath thebacterium, and (iii) pedestal formation at the site of attach-

*7.ag ij.4.

FIG. 1. Transmission electron microscopy of H. pylori 87A300attached to AGS cells. Attachment was for 2 hr. (x 12,000.)

ment. After attachment, we also commonly observed thebacterium to make a transition from the spiral form to thecoccoid form. This transition did not occur in bacterial culturesincubated alone, suggesting that the conversion might beinduced in some way by cell contact. Coccoid forms could bereadily distinguished from spiral forms cut transversely.

Intracellular bacteria were frequently observed (Fig. 2).Numerous steps of internalization of the coccoid form by thecells also could be seen, including some bacteria that weretotally engulfed and appeared to be enclosed within a cyto-plasmic vacuole. Intracellular spiral forms were occasionallyobserved and were usually located within defined membrane-bound vacuoles (data not shown). In general, intracellular H.pylori did not appear to replicate, and many degenerate

attachedto*AGS cells. Atacme twas *tfor 24h. ;'10

' A , ' w *;v*

4~~j4 ~* 4t<

FiG.2.~~wTasmsso elecron mirscp oH.yln8A0

atahe o G cls. A+ttahen a for24, hr.(X 10.)

1260 Microbiology: Segal et al.D

ownl

oade

d by

gue

st o

n Ju

ne 1

0, 2

021

Proc. Natl. Acad. Sci. USA 93 (1996) 1261

bacterial forms were observed intracellularly within hours ofuptake (12).The attachment process between H. pylori and gastric cells

was examined at a more intimate level by double immunoflu-orescence. AGS cells infected with either spiral or coccoid H.pylori cultures were fixed and stained with polyclonal anti-H.pylori antibody to identify Helicobacter associated with the cellsand with the fluorochrome dye rhodamine phalloidin, whichbinds to actin filaments. As demonstrated in Fig. 3, bacteria(stained green) colocalized with actin filaments (stained red),which produced a yellow color. Attachment of H. pyloriresulted in distinct actin foci surrounding each attached bac-terium. Similar results were obtained for attachment of spiralH. pylori (data not shown).

Actin rearrangement within the host cell occurred directlybeneath the site of attachment of H. pylori, forming a very finecondensed structure concentric to the bacterium (Figs. 1 and3). A transverse section of the host cell revealed that rear-rangement can take on the appearance of a wave (data notshown). Viewed from above, the bacterium is observed to bewithin a circle of actin. The cytoskeletal rearrangementsinduced by attachment with H. pylori became apparent within20 min of attachment, although the changes progressed withadditional passage of time (2 hr postattachment). The tem-poral factor may be related to the observations that (i) coccoid

FIG. 3. Confocal IF of coccoid H. pylori 87A300 attached to AGScells (0.25-,um image). (A) Field stained for H. pylori. (B) Field stainedfor actin. (C) Merged image ofA and B, in which H. pylori appearsgreen, actin appears red, and where colocalization occurs yellow isproduced. Imaged on a Bio-Rad MRC 1000 confocal microscope.

H. pylori appeared to induce a stronger cytoskeletal responsethan spiral H. pylori, and (ii) spiral H. pylori, when attached toa gastric cell surface, tended to become coccoid after a fewhours.

Involvement of Additional Cytoskeleton Elements in Bind-ing of H. pylori to Gastric Cells. To determine whetheradditional host cytoskeletal elements were affected by thebinding of H. pylori to gastric cells, AGS cells, which had beenexposed to H. pylori, were probed with anti-a-actinin oranti-talin monoclonal antibodies. a-Actinin functions as across-linker for actin filaments, while talin, a component offocal adhesions, has been postulated to play a role in signaltransduction by linking actin filaments to transmembranereceptors. Both of these proteins have been shown to beinvolved in cytoskeletal rearrangement resulting from theattachment of EPEC (13) and Yersinia invasion-mediateduptake (14). Double IF showed that both a-actinin and talinwere concentrated at the site of H. pylori to gastric epithelialcells.

Induction of Tyrosine Phosphorylation of Host Cell Pro-teins upon Attachment of H. pylori. To investigate whethertyrosine phosphorylation of host cell proteins is associatedwith attachment of H. pylori, AGS cells were exposed to H.pylori and stained with a monoclonal antibody against phos-photyrosine and polyclonal anti-H. pylori antibodies. As shownin Fig. 4, we were clearly able to demonstrate colocalization ofbacteria with phosphotyrosine, as evidenced by a localizedincrease in tyrosine-phosphorylated proteins at the site of H.pylori attachment.

To determine which host cell proteins were phosphorylatedupon H. pylori attachment, extracts were made of AGS cellsalone and after H. pylori binding. These extracts were sepa-rated by SDS/PAGE, transferred to nitrocellulose, and im-munoblotted with a monoclonal antibody against phosphoty-rosine (Fig. 5). The anti-phosphotyrosine antibody identifiedtwo host proteins, a major one of 145 kDa and a minor of 105kDa (not seen in this photograph but seen in a darkerexposure), as being induced in AGS cells postattachment of H.pylori. Phosphorylated proteins of these molecular masseswere not seen in an extract of H. pylori alone, indicating thatthe induced tyrosine-phosphorylated proteins were of host cellorigin. Colocalization of phosphorylated proteins and attachedbacteria was confirmed by immunogold labeling. Fig. 6 showsthat 2 hr postinfection, attachment of H. pylori resulted in thejuxtaposition of phosphotyrosine that contained proteins andH. pylori.

DISCUSSIONWe have investigated the mechanism of adhesion and theconsequences of attachment by H. pylori to human gastric cells.H. pylori attachment to AGS cells is strikingly similar to EPEC(10) and is characterized by effacement of microvilli, pedestalformation, cytoskeletal rearrangement, and phosphorylationof host proteins at the site of adherence.

Transmission electron microscopy confirms the finding thatattachment of H. pylori to gastric cells can induce pedestalformation (Figs. 1 and 2) and that H. pylori attachment to thecell is correlated with the loss of microvilli at that site (Fig. 1).The effacement observed here is consistent with the pathologyseen in gastric biopsies of patients and previous in vitro studies(9, 11, 15). Pedestal formation and actin filament rearrange-ment have been noted by others using primary and establishedgastric epithelial cells (9) but not by those using nongastric cells(11).

Pedestal formation describes the creation of an uprightsupport, constructed of host cell material, beneath an attachedbacterium and is considered to be one of the hallmarks ofEPEC attachment. Not all bound Helicobacter were observedto be associated with these structures, nor is it known at this

Microbiology: Segal et al.D

ownl

oade

d by

gue

st o

n Ju

ne 1

0, 2

021

Proc. Natl. Acad. Sci. USA 93 (1996)

A B C

207

139

84

FIG. 4. Confocal IF of spiral H. pylori 87A300 attached to AGScells (0.3-,um image). (A) Field stained for H. pylori. (B) Field stainedfor phosphotyrosine. (C) Merged image ofA and B, in which H. pyloriappears green, phosphotyrosine appears red, and where colocalizationoccurs yellow is produced. Imaged on a Bio-Rad MRC 1000 confocalmicroscope.

time why only a portion of the attached bacteria induced thechange. We observed, however, the H. pylori that had becomecoccoid in shape were more likely to induce pedestal formationthan spiral H. pylori, which when attached appeared to becomefused with the epithelial cell membrane. The heterogeneity inpedestal formation induction may be related to the mechanismof attachment of H. pylori to gastric cells. Several candidateadhesins have been identified in H. pylori (16-18). Thus, theattachment and effacement phenotype may be the result ofseveral factors acting cooperatively. Genes homologous to E.coli eae have been detected in H. pylori (11).Attachment of H. pylori to AGS cells induced tyrosine

phosphorylation of two host cellular proteins (Fig. 5). Theamount of a 145-kDa protein was greatly increased, while asecond protein of 105 kDa was moderately increased. Theidentities of the 145- and 105-kDa proteins are not currentlyknown. Confocal immunofluorescent microscopy (Fig. 4) andimmunogold labeling (Fig. 6) illustrated that attached bacteriaare in intimate contact with tyrosine-phosphorylated proteinsclearly distinct from focal adhesion sites. This suggests that H.pylori adhesion triggers signal transduction. Cytoskeletal com-ponents, responding to transduction signals induced by extra-cellular signals, can be affected in both structure and functionby tyrosine phosphorylation. Fischer et al. (19) showed thatoverexpression of a tyrosine phosphatase in BHK cells caused

FIG. 5. Immunoblot analysis of AGS cells infected with H. pyloriand probed with anti-phosphotyrosine antibody. Lane A, extract ofwhole H. pylori; lane B, extract of AGS cells; lane C, extract of AGScells to which H. pylori had attached for 2 hr. Molecular size markers(kDa) are indicated on the left.

actin filaments to become resistant to cytochalasin-induceddisassembly, suggesting that the tyrosine phosphatase mightact on the membrane-associated structures (focal adhesions)where actin bundles terminate. Talin and vinculin, part of focaladhesions, are phosphorylated on tyrosyl residues and arethought to be associated with the cytoplasmic domain ofintegrin receptors, which also contain phosphotyrosine [for areview see Burridge et al. (20)].Although we do not know the role of tyrosine phosphory-

lation of host cell proteins in the pathogenesis of gastricdiseases associated with H. pylori infection, we speculate thatthis event is involved in the inflammatory response, which is aninvariable component of gastritis, ulcers, and, presumably,gastric cancer. Signal transduction stimulated by adherencemight generate a cytokine response in the acute phase orinitiate oncogenic transformation as a long-term effect.The piracy of eukaryotic protein tyrosine kinases by bacte-

rial pathogens has been identified previously. EPEC (21) andSalmonella typhimurium (22) both encode genes (cfin andinvA, respectively) that modify host phosphorylation, leadingto invasion of the bacterium into the host cell. Yersiniapseudotuberculosis contains a similar gene, inv (23), and anadditional gene, yopH, which through dephosphorylation pro-vide an antiphagocytotic phenotype (24, 25). EPEC inducestyrosine phosphorylation of three eukaryotic proteins (21), allapparently cytoskeletal associated. The major protein is a90-kDa protein, and the two minor proteins are 39 and 72 kDa.The identity of these proteins is currently not known. Listeriamonocytogenes has been shown to induce the tyrosine phos-phorylation of two isoforms (42 and 44 kDa) of the mitogen-activated protein kinase (26).Our use of an established cell line (albeit a human gastric

epithelial cell line) to study H. pylori attachment is not ideal.However, a comparison of adhesion of H. pylori to surfacemucus cells from different origins shows that adhesion of H.pylori with human cells of gastric origin in vitro resembles thatseen in vivo (27). This was true whether the gastric cells were

1262 Microbiology: Segal et al.D

ownl

oade

d by

gue

st o

n Ju

ne 1

0, 2

021

Proc. Natl. Acad. Sci. USA 93 (1996) 1263

FIG. 6. Transmission electron microscopy of immunogold labelingof AGS cells infected with H. pylori for 2 hr. Anti-phosphotyrosineantibody (diluted 1:50 in 2% bovine serum albumin/PBS) was incu-bated for 1 hr. The grid was washed three times for 5 min each in 2%bovine serum albumin/PBS and incubated with goat anti-mouse IgGconjugated to 10-nm gold beads (Ted Pella, Redding, CA) diluted 1:10in 2% bovine serum albumin/PBS for 30 min. (x27,000.) Large, brightspot represents an accumulation of polyphosphate, which is commonin H. pylori.

a primary culture derived from stomach biopsies or cells froma cell line of human adenocarcinoma (AGS).H. pylori is generally considered to be an extracellular

pathogen, although there have been reports of intracellular H.pylori both in vivo and in vitro (2, 12, 28). We and others haveobserved intact viable bacteria within a host cell along withdegraded forms (29, 30). The mechanism of uptake has beenpostulated to be receptor-mediated endocytosis (12) followingadherence to a coated pit. THe relevance and fate of intra-cellular H. pylori is still unknown, although a study has showna correlation between high bacterial density of colonization,intracellular bacteria, and severe epithelial damage (ulcer-ation) (28).

H. pylori is known to convert from its spiral, motile form toa coccoid form that is nonmotile and nonculturable. In vitro,this conversion generally occurs when a bacterial culture is old,but coccoid forms are also seen in gastric biopsies. Inductionof a shape change from spiral to coccoid upon binding to thegastric cell line HGT1 has been observed by Neiman-Simhaand Megraud (8). A recent study from Eaton et al. (31) statedthat the coccoid form of an H. pylori strain was unable toestablish infection in the gnotobiotic piglet model. However,coccoid forms are aflagellate and it has been shown previouslythat a nonmotile strain was not able to cause infection in thismodel (32). Using a BALB/c mouse model with intragastricinoculation of fresh and laboratory passaged strains with spiralor coccoid morphology, Cellini et al. (33) demonstrated thatwhile neither form of the passaged strain was able to colonize,both forms of the fresh strain colonized and produced gastricalterations. All measured responses produced by the coccoidinoculum were the same as those produced by the spiral formsbut with a delayed time course. Additionally, Cellini and

colleagues further observed spiral organisms in gastric biopsiesof mice infected with coccoid cultures and obtained culturableH. pylori from the mouse stomach, indicating that the coccoidforms reverted to the spiral form during infection.The results of our studies indicate that the coccoid form of

H. pylori is fully capable of attaching to gastric cells and inducesthe same cytoskeletal changes seen upon attachment of spiralH. pylori, including induction of host cell phosphotyrosineactivity. Thus, the coccoid form might serve as the infectiousform in environmental sources such as water, whereas thespiral form appears as the predominant type in the gastricmucus. Therefore, the potential pathogenicity of the coccoidform should be addressed.

The transmission electron microscopy and immunogold labelingwere done with the help of Nafisa Ghori. This research was supportedby National Institutes of Health Grant AI23796 (to L.S.T.) andDigestive Disease Center Grant of the National Institute of Diabetesand Digestive and Kidney Diseases (DK 38707) (to E.D.S., S.F., andL.S.T.) from the Public Health Service.

1. Cover, T. & Blaser, M. (1995) ASM News 61, 21-26.2. Wyle, F. A., Tarnawski, A., Schulman, D. & Dabros, W. (1990)

J. Clin. Gastroenterol. 12, Suppl. 1, S99-S103.3. Catrenich, C. E. & Makin, K. M. (1991) Scand. J. Gastroenterol.

Suppl. 181, 58-64.4. Bode, G., Mauch, F. & Malfertheiner, P. (1993) Epidemiol. Infect.

111, 483-490.5. Segal, E. D., Shon, J. & Tompkins, L. S. (1992) Infect. Immun. 60,

1883-1889.6. Faulstrich, H., Zobeley, S., Rinnerthaler, G. & Small, J. V. (1988)

J. Mus. Res. Cell Motil. 9, 370-383.7. Rudmann, D. G., Eaton, K. A. & Krakowka, S. (1992) Infect.

Immun. 60, 2121-2124.8. Neiman-Simha, V. & Megraud, F. (1988) Infect. Immun. 56,

3329-3333.9. Smoot, D. T., Resau, J. H., Naab, T., Desbordes, B. C., Gilliam,

T., Bull-Henry, K., Curry, S. B., Nidiry, J., Sewchand, J., Mills-Robertson, K., Frontin, K., Abebe, E., Dillon, M., Chippendale,G. R., Phelps, P. C., Scott, V. F. & Mobley, H.L.T. (1993) Infect.Immun. 61, 350-355.

10. Moon, H. W., Whipp, S. C., Argenzio, R. A., Levine, M. M. &Giannella, R. A. (1988) Infect. Immun. 41, 1340-1351.

11. Dytoc, M., Gold, B., Louie, M., Huesca, M., Fedorko, L., Crowe,S., Lingwood, C., Brunton, J. & Sherman, P. (1993) Infect.Immun. 61, 448-456.

12. Evans, D. G., Evans, D. J. & Graham, D. Y. (1992) Gastroenter-ology 102, 1557-1567.

13. Finlay, B. B., Rosenshine, I., Donnenberg, M. S. & Keusch, G. T.(1992) Infect. Immun. 60, 2541-2543.

14. Young, V. B., Falkow, S. & Schoolnik, G. K. (1992) J. Cell Biol.116, 197-207.

15. Hessey, S. J., Spencer, J., Wyatt, J. I., Sabala, G., Rathbone, B. J.,Axon, A. T. R. & Dixon, M. F. (1990) Gut 31, 134-138.

16. Evans, D. G., Evans, D. J. & Moulds, J. T. (1988) Infect. Immun.56, 2896-2906.

17. Boren, T., Falk, P., Roth, K. A., Larson, G. & Normark, S. (1993)Science 262, 1892-1895.

18. Valkonen, K. H., Wadstrom, T. & Moran, A. P. (1994) Infect.Immun. 62, 3640-3648.

19. Fischer, E. H., Charbonneau, H. & Tonks, N. K. (1991) Science253, 401-406.

20. Burridge, K., Fath, K., Kelly, T., Nuckolls, G. & Turner, C. (1988)Annu. Rev. Cell Bio. 4, 487-525.

21. Rosenshine, I., Donnenberg, M. S., Kaper, J. B. & Finlay, B. B.(1992) EMBO J. 11, 3551-3560.

22. Galan, J. E., Pace, J. & Hayman, M. J. (1992) Nature (London)357, 588-589.

23. Rosenshine, I., Duronio, V. & Finlay, B. B. (1992) Infect. Immun.60, 2211-2217.

24. Bliska, J. B., Guan, K., Dixon, J. E. & Falkow, S. (1991) Proc.Natl. Acad. Sci. USA 88, 1187-1191.

25. Rosqvist, R., Bolin, I. & Wolf-Watz, H. (1988) Infect. Immun. 56,2139-2143.

26. Tang, P., Rosenshine, I. & Finlay, B. B. (1994) Mol. Biol. Cell 5,455-464.

Microbiology: Segal et aL

Dow

nloa

ded

by g

uest

on

June

10,

202

1

1264 Microbiology: Segal et al.

27. Nilius, M., Bode, G., Buchler, M. & Malfertheiner, P. (1994) Eur.J. Clin. Invest. 24, 454-459.

28. Chen, M., Lee, A. & Hazell, S. (1992) Lancet 339, 1120-1121.29. Wyle, F. A., Tarnawski, A., Dabros, W. & Gergely, H. (1990) J.

Clin. Gastroenterol. 12, S99-103.30. Andersen, L. P., Blom, J. & Nielsen, H. (1993) APMIS 101,

61-72.

Proc. Natl. Acad. Sci. USA 93 (1996)

31. Eaton, K. A., Catrenich, C. E., Makin, K. M. & Krakowka, S.(1995) J. Infect. Dis. 171, 459-462.

32. Eaton, K., Morgan, D. R. & Krakowka, S. (1989) Infect. Immun.57, 1119-1125.

33. Cellini, L., Allocati, N., Angelucci, D., lezzi, T., Di, C. E.,Marzio, L. & Dainelli, B. (1994) Microbiol. Immunol. 38,843-850.

Dow

nloa

ded

by g

uest

on

June

10,

202

1