Gut 1997; 40: 505-511 1

Effect of shigella enterotoxin 1 (ShET1) on rabbitintestine in vitro and in vivo

A Fasano, F R Noriega, F M Liao, W Wang, MM Levine

Division ofPediatricGastroenterology andNutritionA Fasano

Center for VaccineDevelopment,University ofMaryland, School ofMedicine, Baltimore,MD 21201, USAA FasanoF R NoriegaFMLiaoW WangMM LevineCorrespondence to:Dr Alessio Fasano,GastrointestinalPathophysiology Unit,Center for VaccineDevelopment, University ofMaryland School ofMedicine, 685W BaltimoreStreet, HSF Facility,Room 465, Baltimore,MD 21201, USA.Accepted for publication20 December 1996

AbstractBackground-Shigella enterotoxin 1 is anovel enterotoxin elaborated by Shigellaflexneri 2a that causes fluid accumulationin rabbit ileal loops and a rise in shortcircuit current in Ussing chambers.Aims-To gain insights into the mechan-ism of action of shigella enterotoxin 1.Methods-Supernatants from geneticallyengineered clones either overexpressingshigella enterotoxin 1 or producing de-letion mutants of the toxin were tested inrabbit ileum both in vitro and in vivo.Results-In rabbit ileum shigella entero-toxin 1 induced an irreversible rise inshort circuit current that was not medi-ated by any ofthe recognised intracellularmediators of secretion. Deletion of90% ofthe A subunit of the holotoxin ablated itsenterotoxicity. In the in vivo perfusionmodel, the toxin induced a time de-pendent decrease in water absorption,whereas no changes were detected in thesegment perfused with supernatants ob-tained from the deletion mutant. Finally,partially purified toxin induced a dosedependent increment in short circuitcurrent that reached its plateau at a toxinconcentration of4x 10 ' M.Conclusions-Shigella enterotoxin 1 in-duces a time and dose dependent in-testinal secretion in the rabbit malmodel, suggesting that it may be respon-sible for the watery phase of Shigellaflexneri 2a infection.(Gut 1997; 40: 505-511)

Keywords: diarrhoea, shigellosis, Ussing chambers,ileum, secretion.

Shigella dysenteriae type I was first isolated in1898 during an epidemic of severe dysentery inJapan. ' Over the subsequent century othershigella species were identified, includingShigella flexneri, Shigella boydii, and Shigellasonnei, and the mechanisms whereby theycause disease have been intensively investi-gated. A cardinal feature of shigella virulenceis the capacity of the organism to invademammalian epithelial cells, followed by celldeath, spread to adjacent cells, and elicitationof an intense inflammatory response withinfiltration by polymorphonuclear leuco-cytes.2" The classic dysenteric stool is scantyand consists of blood and mucus. However,most ill patients develop an initial phase of

watery diarrhoea, that in mild cases mayrepresent the only clinical presentation of thedisease, without evolving to dysentery.7 8We have recently reported the elaboration by

shigella of two novel iron-regulated entero-toxins, shigella enterotoxin 1 (ShETI) and 2(ShET2), that alter electrolyte and water trans-port in rabbit small intestine both in vitro andin vivo.' '° ShETl is a chromosomally en-coded, 55 kDa complex protein9 that isuniversally elaborated by Shigella flexneri 2astrains but only rarely by other serotypes."1Sequencing analysis of the genes encoding thistoxin disclosed the presence of two contiguousopen reading frames (orfs) of 534 (setlA) and186 (setlB) bp respectively, governed by thesame promoter and separated by only 3 bp.ShET2 is a 62-8 kDa single protein that ourgroup originally described in enteroinvasiveEscherichia coli (EIEC) and referred to as EIECenterotoxin (EIET).'2 The gene encoding thistoxin is located on the 140 mDa invasivenessplasmid of shigella and seems to be present inmore then 80% of a wide array of shigellaserotypes examined.'0 Gene sequence analysisshowed that both ShETI and ShET2 aregenetically unrelated to shiga toxin elaboratedby Shigella dysenteriae. Furthermore, the threetoxins also seem to be immunologicallyunrelated.

In this paper we further characterise themechanism of action of ShETl, both in vitroand in vivo. Our data indicate that ShET1induces a time and dose dependent intestinalsecretion in the rabbit animal model, suggest-ing that the toxin may have a role in thepathogenesis of the watery phase of dysentericinfection.

Methods

OVEREXPRESSION OF ShETIThe 1093 bp chromosomal fragment contain-ing the ShETl genes (setl) was cloned in themultiple copy plasmid pBluscript SK aspreviously described.9 The plasmid so ob-tained, psetl, was transformed in DH5a(Fig 1) and overexpression of the ShETIprotein was obtained in iron depleted culturesupernatants.9 More specifically, 100 ml LuriaBertani broth containing 25 ,ug/ml of the ironchelator ethylenediamine-N-N' diacetic acid(EDDA) and 100 ,ug/ml ampicillin were in-oculated with a single colony of DH5a (psetl)and incubated overnight at 37°C with shaking(200 rpm). Bacterial cells were removed by

505

on Septem

ber 25, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.40.4.505 on 1 April 1997. D

ownloaded from

Fasano, Noriega, Liao, Wang, Levine

Xmnl

Scal 2526-

Xmnl

Scal 3618.

T7 promoterKpalApalXholSall (Hincil)AcclHindillEcoRVEcoRI

-PstlSmalBamHISpelXbalEaglNotiScallBstXISaclT3 promoter

3737

Smal setl B

Figure 1: Genetic construction ofDH5ot (psetl) and its deletion mutant DH5ot (pAsetl).Plasmid pBluscript was used to mobilise the 1098 bp Shigellaflexneri 2a chromosomalfragment containing the setlA and setlB orfs. The deletion mutant was engineered bydeleting 90% of the setlA gene (from AA residue 17 to AA redisue 169), while the naturalpromoter and the whole setlB orf were left unaltered.

centrifugation, and the supernatant was filter-sterilised through a 0-22 mm membrane filter.The supernatant was then frozen until used forthe experiments described later.

PURIFICATION OF ShET1Overexpressed ShET 1 was obtained as de-scribed above. The proteins contained in theDH5ot (psetl) supernatant were then precipi-tated with saturated trichloroacetic acid (TCA)(20:80 v/v) and separated by SDS-PAGEelectrophoresis. Resultant bands were trans-ferred to a nitrocellulose membrane and cutinto two strips to be either silver stained tovisualise protein bands or reacted with rabbitantiserum against ShET I1 for western im-munoblotting. The single band correspondingto the 55 kDa ShET1 protein was carefully cutout of the gel and electroeluted using anelectroseparation chamber (Schleicher andSchuell, Keene, NH, USA). Purity of ShET1was established by SDS-PAGE electrophoresisand by western immunoblotting. The amountof electroeluted ShET1 was determined by theBradford method."3

CONSTRUCTION OF A setl DELETION MUTANT

A Asetl allele was obtained by deleting 90% ofsetlA gene (from AA residue 17 to AA residue169), while the natural promoter and the wholeB subunit were left intact (Fig 1). DNA seg-ments were amplified from a library derivedphagemid pF9-1-909 containing the Asetlallele and fused by polymerase chain reaction(PCR). Unique restriction sites were thenintroduced using external primers and wereutilised to clone the Asetl allele into pBluscriptSK. The plasmid so obtained (named pAsetl)was electroporated in DH5cl and supernatantsof this strain (DH5ot (pAsetl)) (Fig 1) wereprepared as reported above and frozen untiltested in the in vitro and in vivo assaysdescribed below. The amount of proteinpresent in DH5cx (pAsetl) was determined bythe method of Bradford'3 and results werecomparable with the total protein concen-tration present in the DH5oc (psetl) super-natant (4 21 [ig/ml v 4 41 pLg/ml respectively).

CONSTRUCTION OF MALTOSE BINDING PROTEIN

(MBP)-ShETI (A SUBUNIT FUSION PROTEIN)AND PREPARATION OF ANTI-A SUBUNIT

POLYCLONAL ANTIBODIES

The ShETI A subunit gene was amplified byPCR using psetl as a template. Fidelity ofPCRamplification was confirmed by DNAsequencing of the plasmid insert. setl A genewas then fused in frame with the maltosebinding protein (MBP) gene, using vectorpMalc2'4 to create a MBP-ShET1 A fusionprotein. The fusion product was expressed inEscherichia coli and obtained by disrupting thecells using a French press. The cell lysate wascentrifuged for 30 minutes at 15 000 g, and thesupernatant was filtered, diluted five times inbuffer A (20 mM Tris HCl pH 7 5, 1 mMEDTA, 0-2 M NaCl, 1 mM NaN3, and 1 mM2-ME), and loaded on to an amylose column(New England Biolabs) at a flow rate of 1 ml/min. The column was washed with fivevolumes of buffer A and the fusion protein waseluted with 10 mM maltose in buffer A.Protein concentrations were assessed with a

506

on Septem

ber 25, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.40.4.505 on 1 April 1997. D

ownloaded from

Enterotoxicity ofShigella enterotoxin 1

Bio-Rad protein assay, and purity was deter-mined by SDS-PAGE followed by Comassie orsilver staining.To raise polyclonal antibodies against

ShETl A subunit, purified MBP-ShET1 Afusion protein was mixed with an equal volumeof complete Freund's adjuvant and injectedinto an adult male New Zealand white rabbit.Two booster doses were given with Freund'sincomplete adjuvant four and eight weeks later,and the animal was then bled.

USSING CHAMBERS

Experiments were carried out as previouslydescribed.'5 Briefly, adult male New Zealandwhite rabbits (body weight 2-3 kg) were killedby cervical dislocation. A 20 cm segment ofileum was removed, rinsed free of the intestinalcontent, opened along the mesenteric border,and stripped of muscular and serosal layers.Eight sheets of mucosa so prepared were thenmounted in lucite Ussing chambers (1I12 cm2opening), connected to a voltage clamp appar-atus (EVC 4000 WPI, Saratosa, FL, USA),and bathed with freshly prepared buffer con-taining (in mM): NaCl, 53; KCI, 5; Na2SO4,30 5; mannitol, 30 5; Na2HPO4, 1-69;NaH2PO4, 03; CaCl2 1-25; MgCl2 1i1;NaHCO3, 25. The bathing solution was main-tained at 37°C with water jacketed reservoirsconnected to a constant temperature circu-lating pump and gassed with 95% 02/5% CO2.Potential difference (PD) and short circuitcurrent (Isc) were measured under short circuitconditions and tissue resistance (Rt) wascalculated as previously described."6 Filteredsupernatants (300 ,ul) obtained from eitherDH5a (psetl) or DH5at (pAsetl) were addedto the mucosal surface and an identical amountwas added to the serosal surface to preserve theosmotic balance. In selected experiments,partially purified ShETl was added only to themucosal side of the rabbit ileum. Variations inPD, Isc, and Rt were then recorded every 10minutes. At the end of every experiment, 0-5mM glucose was added to the mucosal side ofeach chamber. Only those tissues whichshowed an increase in Isc in response toglucose, indicating tissue viability (98% of thetissues tested), were included in the analysis.

IN VIVO PERFUSION ASSAY

Intestinal perfusion was carried out accordingto the method previously described by Sladenet al'7 with minor modifications. More specifi-cally, after a 24 hour fast, 2-5-3 kg adult maleNew Zealand white rabbits were anaesthetisedwith 50 mg ketamine/kg body weight, followedby intramuscular injection of 7 0 mg xylazine/kg body weight. Their body temperature waskept at 37°C by a lamp. The abdominal cavitywas opened by a midline incision and twodistinct segments of jejunum were cannulated.A second cannula was placed 10-15 cm beloweach proximal cannula. The segments wererinsed free of intestinal contents with 0-9%(w/v) NaCl warmed to 37°C. The proximalcannulae were connected by a polyvinyl tube

to a peristaltic pump (model WPI SP220 I),and the two segments were perfused at a rateof 0 4 ml/min with a solution comprising 2-0mmoIIl glucose, 4 0 mmol/l KCI, 25 mmolINaHCO3, 3 0 g/l PEG-4000, and 3 0 ,uCi/100ml '4C-PEG-4000. The solution was madeisotonic by adjustment with NaCl, and the pHfixed at 7-4 by gassing with 95% 02/5% CO2.

Eluates were collected in 20 minute aliquotsfrom the distal cannulae. An initial equilibriumperiod of 30 minutes was allowed, followed bythree consecutive 20 minute collection periodsfor baseline measurement of the net transportof water and electrolytes in each segmentstudied.

Subsequently, one of the two segments wasperfused with the same solution describedabove containing 30 RI/ml DH5c (psetl) super-natant, while the second segment was perfusedwith a similar amount of DH5a (pAsetl)supematant. A second perfusion period (30minute equilibration plus 6X20 minutecollection) was then carried out. At the end ofthe experiment, the animal was killed and thesegments perfused were isolated, measured,dried, and weighed. Water absorption was thencalculated as previously described.'7

Results

INTRACELLULAR MESSENGER(S) OF ShETlENTEROTOXIC ACTIVITYTo establish whether the enterotoxic effect ofShETl was mediated by one of the describedintracellular mediators of intestinal secretion(cAMP, cGMP, and Ca2"), 5 mM theophylline,0-2 mM 8Br-cGMP, and 5X10- M Ca iono-phore A23 187 were tested in Ussing chambersin the presence and absence of DH5a (psetl)supematant. Segments of rabbit ileummounted in Ussing chambers were pairedbased on their Rt. Supernatant obtained fromDH5ot (psetl) (positive control) was added toone of the chambers, while the coupledchamber was exposed to DH5at (pBS) (nega-tive control) supernatant. Once the incrementof Isc induced by DH5a (psetl) reached aplateau, the second messenger to be tested wasadded to the serosal side of both chambers andthe subsequent increment in Isc recorded. Asshown in Fig 2, the Isc changes induced by thethree intracellular messengers tested weresimilar in the ShETI exposed and controlexposed tissues, suggesting that none of themis involved in ShETI secretory action.

LACK OF REVERSIBILITY OF ShETIENTEROTOXIC EFFECTTo establish whether the ShETl enterotoxiceffect was reversible, rabbit ileal mucosa wasfirst exposed to DH5a (psetl) supematant.When the enterotoxic effect of supematant con-taining ShETl reached its plateau, the reservoirwas emptied, rinsed twice with Ringers'solution, and refilled with Ringers' alone. Nosubstantial changes in Isc were seen afterShETI withdrawal (Fig 3), suggesting either anirreversible binding of the toxin with its receptor

507

on Septem

ber 25, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.40.4.505 on 1 April 1997. D

ownloaded from

Fasano, Noriega, Liao, Wang, Levine

120

100

zE 80ET0

E 60

40

20

0cAMP cGMP Ca2+

Figure 2: Peak inlcremizetnts i.n Isc after addition of 5 mmt}lol/ltheophylline (cAMP), 0-2 mnisol/l ,8-bromo-cGMP(cGMP), or 5x 10 mol/l calcium ionophore A23187(Ca7+) to tissues pretreated with either DH5x (psetl)(ShETI positive) (U) or DH5ot (pBS) (ShET1 negative)(E) supernatants. The second messengers were added to theserosal side once the Isc increment induced by ShETJreached the plateau (110-120 min) (n=4 aninials). Barsrepresent SEM.

or a continuous activation of the intracellularsignalling that mediates the enterotoxin effect.

WESTERN I.MM.UNOBLOTS OF DH5ot (psetl) ANDITS DELETION MUTANT DH5CX (pAsetl)Western blotting performed with the rabbitantiserum raised against ShETI9 showed the55 kDa ShETi band in DH5oC (psetl) super-natant, but not in DH5t (pAsetl) supernatant(Fig 4A). Interestingly, the band was presentonly in the supernatant but not in the cell lysateof DH5ot (psetl), suggesting that ShETI is asecreted protein. These results are in keepingwith the fact that the predicted amino acid

140-

120

100

E 80aL I

E 60_ I

40

20

0Le

-20

- uperna

Setl supernatant

Gluc3tant removea

0 50 100 150

Time (min)

Figure 3: Lack of reversibility ofShET1 enterotoxic effect.Supernatants obtainedfrom both DH5ot (psetl) (ShETJpositive) (A) and DH5ot (pBS) (ShET1 negative) (O)were added to rabbit ileal mucosa mounted in Ussingchambers and changes in Isc were followed over time.ShETI induced a time dependent increase in Isc comparedwith the negative control that reached its plateau afteralmost 100 nminutes. Once the toxin was withdrawn fromthe reservoir (arrow), no mnodifications in Isc were observed,suggesting an irreversible binding ofShET1 to its receptor.Addition of 0 5 mM glucose to the mucosal side of the rabbitileumi induced a sinizilar rise in Isc in both the ShETI andnegative control exposed tissues, proving that ShETI didnot affect the tissue viability (n=4 aninials).

sequence of the setl gene features a putativesignal sequence.9 As these experiments wereperformed under denaturing conditions, werepeated the western blotting experiment usingnon-denaturing conditions. The resultsshowed in Fig 4B confirmed that ShETI ispresent in DH5ot (psetl) supernatant, but islacking in both DH5oX (pBS) and DH5oL(pAsetl) supernatants. Finally, to establishwhether DH5oa (pAsetl) expresses the ShETIB subunit, the western blotting was repeatedusing polyclonal antibodies raised against theShETi A subunit and compared with westernimmunoblots performed using polyclonalantibodies raised against the entire holotoxin.9As shown in Fig 4C, the polyclonal antibodiesagainst the holotoxin visualised a 35 kDa bandpresent in both DH5c (psetl) and DH5ot(pAsetl) supernatants, but not in DH5cx (pBS)negative control supernatant. This band wasnot visualised by the polyclonal antibodiesraised against the ShET 1 A subunit, suggestingthat the 35 kDa band may represent a B5aggregate and, therefore, that the B subunit isstill expressed by the deletion mutant.

EFFECT OF shETl DELETION MUTANT IN

USSING CHAMBERSThe apparent molecular size of ShET 1 by bothgene sequencing and western blot analysis9suggests a holotoxin stoichiometry (A,-B,)similar to that of other well established entero-toxins.18 21 To establish whether the ShETl Asubunit plays a pivotal role in ShETl entero-toxicity, supernatants obtained from DH5o-(pAsetl) lacking 90% ofShETl A subunit weretested in Ussing chambers. When added torabbit ileal mucosa, this supernatant failed toinduce the increase in Isc typically seen withDH5o (psetl) supernatant (Fig 5).

IN VIVO PERFUSION EXPERIMENTSWe have previously shown that ShET 1 con-taining supernatants induce a mild, but signifi-cant fluid accumulation in rabbit ileal loops invivo.9 To better define the mechanism ofsecretion of ShET1 and to establish the in vivoeffect of the deletion mutant ofShETl protein,two contiguous segments of jejunum weresimultaneously perfused in the same animalwith a saline solution (period 1), followed bya second period during which solutionscontaining either DH5oc (psetl) or DH5ct(pAsetl) supernatants were perfused. The firstperiod of perfusion was characterised by theactive absorption of water and electrolytes inboth segments (Fig 6). After 60 minutes theaddition of ShETI to the perfusion solutionproduced a significant decrease in water ab-sorption, whereas no changes were detected inthe segment perfused with ShETI proteindeleted of its A subunit (Fig 6).

DOSE-RESPONSE CURVE OF ShET 1ENTEROTOXICITYTo establish whether the enterotoxic effectof ShETI was dose dependent, increasing

I

508

on Septem

ber 25, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.40.4.505 on 1 April 1997. D

ownloaded from

Enterotoxicitvi of Sliigella enterotox-in15

B

B C D E F

A B c

A B C A B C

Figure 4: lttTsteni imimiunoblots ofDH5c- (pseti) anid its deletionl Mlutanlt DH5ic (p \setl). (A) Denaturing gel. Supernatants anid cell ls'ates obtainledfrom DHco (pBS), DH5x (pseti), and DH5cx (p \setl) .w-re separated by SDS-PAGE undcr denaturing conditions, transferred to niitrocellulose, anzddeveloped using rabbit antisernu to the actiVce fiaction obtainedfroi Shigella flexneri 2a strain .I4243avir.' Lane A: DH5a (pBS) cli ll satc; lauie B:DH5u. (pBS) supernatantt; lane C: DH5in (psetl) cell lvsate; lane D: DH.5a (pseti) supernatant; lane E: DH5cx (p \setl) celi lvsate; lane F: DH5cx(pAsetl) supernatant. Arrow sh/lows 55 kDa ShETI band present in D, butt not in the otlei- lanies. (B) Non-denaturing gel. Supcrlatants obtained fromDH5(x (pBS), DH5ca (pseti), anid DH5ca (p Asetl) .were separated by SDS-PAGE under non-denaturnng concditions, tranisferrCed to nitrocellulose, anzddeveloped uisinlg rabbit antiserum to the active fraction obtained fromi Shigella flexneri 2a strain M14243avir.4 Lanie A: DH5cx (pBS,); lane B: DH5cc(psetl); lane C: DH5u (pAsetl). Note the presence of ShETI band in B, buit not A and C. (C) W,esteri imimiunoblot developed zwit/i anti-SIiETJ Asubunit polclonal antibodies. Slupeniatanits obtainedfrom DH5(x (pBS), DH5cx (pset1), and DH5. (pAsetl) wcrer scparated by SDS-PAGE underdenaturinng conditionis, tranisferied to nitrocellulose, anid developed uising eit/let rabbit antisenrun to SlIETJ /olotox-in (lcft) or ShIETI A subunit (rigit).Lane A : DH5cx (pBS); lane B: DH5ct (psetl); lanze C: DH5o. (p Asetl). Note t/le prcscnzcc cf a 35 kDa band (arrozv.) izn lanes B and C z/hlcn antibodicsagainst the liolotoxin . ere utsed, zw-hereas thlis band as not detected ith antibodies against t/lc SlIETI A subun.it.

100 concentrations of purified toxin obtained by

electroelution were added to the mucosal sideof rabbit ileum mounted in Ussing chambers

80 and changes in Isc recorded. ShETl induced

E _ a dose dependent increment in Isc that reacheda its plateau at a toxin concentration of 4x 1O

E M and an ED50 of 9 x 10 " M (Fig 7).

' 40< _ * _ Discussion

Shigella flexieri 2a is one of the most common

20 serotypes causing bacillarv dvsentery world-

wide.'' " The ability to invade and spread_ _ _ within the epithelial cells and thereby to cause

> c> dysenterv has always been considered thecardinal mechanism of pathogenicity ofShigella flexueri. Nevertheless, the clinicalpresentation of shigellosis, with an initial phase

Figure 5: Enterotoxic effectofS/iETJdclcto of watery diarrhoea that in mild cases may

Ussing chambers. Peak incremlents in Isc after addition of represent the only gastrointestinal symptom,suipernatants obtainedfrom DH5cx(pset1) (ShETrpositizVe), its mutant DH5cx (pAsetl) deleted of t/ie S/iETI would predict the existence of enterotoxins.A suibluniit, or the negatiVe con0trol DHca (pBS,). The Further evidence of the involvement of other

wih/IS/iETiconmtaintinegtsupendlarant. ep< (i Ico= d

pathogenic factors (other than associated withanimnals). shigella invasion and multiplication within the

97 -

66 -

45

A

I:c45 -

3.1

.I.

509

on Septem

ber 25, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.40.4.505 on 1 April 1997. D

ownloaded from

Fasano, Noriega, Liao, Wang, Levine

c

i-

a)

80

60

40

20

0 20 40 60

Perfusion solution

Figure 6: In vivo enterotoxicity of LTwo contiguous segments of rabbitfjdescnrbed in the method section (opeDHSix (pAsetl) were then added t(60 minutes the addition of the supersupernatant (A) showed a significasegment perfused with the toxin dele

CN-E

E

cl,

FigconMenio

red

condosdosM.cha

en

Li:arl

da

su

its

re(

vo

lin

ml

ge

T - antigen of S flexneri 2a to Escherichia coliK- 12.29 This vaccine also induced adversereactions in 31% of vaccinees; among the sideeffects, watery diarrhoea was reported.We have recently reported the elaboration of

two distinct enterotoxins by Shigella flexnenr2a,9 "o that may be responsible for the wateryphase of shigellosis. We have shown that, whengrown in iron depleted medium, enteroinvasive

\ - Escherichia coli (EIEC) elaborate an entero-toxin (EIET) that causes fluid accumulation inisolated ileal loops and an electrical response inUssing chambers.'2 Based on the similaritiesknown to exist between EIEC and shigella, 30we investigated the possibility that Shigellaflexneri 2a elaborates a similar toxin. Our group

80 100 120 140 160 180 200 has recently cloned and sequenced the EIETTime (min) gene.30 A virtually identical gene (99% hom-

ology) was found on the 140 mDa invasivenessplasmid of Shigella flexneri 2a and gene probestudies showed that this gene is widely present

Perfusion solution + supernatant in all shigella serotypes.10 Surprisingly, plasmid)H5ot (psetl) and its deletion mutant DH5o. (pAsetl). cured derivatives of Shigella flexneri 2a stillejunum were initiall perfused with the saline solution retain enterotoxic activity, both in vivo and in'n bar). Supernatants obtainedfroni DHSr (psetl) and .9 . . 'o the solution and perfused separately (solid bar). After vitro. The combination of genetic analysis andrnatants, the segment perfused with ShETI containing intestinal electrophysiology allowed us toznt reduction in water absorption compared with the localise the chromosomal genes responsible for?ted of its A subunit (a). *p<005 (n=3 animals). this residual enterotoxicity.9 Analysis of the

sequence of the cloned genes responsible forShET 1 enterotoxic activity disclosed two

50 distinct, yet contiguous orfs, encoding for twoproteins of predicted MW of 7 kDa and 20

40 kDa, respectively. The common A:Bn activebinding subunit motif often found amongbacterial enterotoxins, including cholera toxin

30 (CT),"0 heat labile enterotoxin (LT) of entero-toxigenic E coli,`9 and shiga toxin of Sdysenteriae 1,20 21 may be reflected in these data.

20 - The apparent molecular sizes of active materialL / as predicted by both the gene sequencing and10 the western blot experiments,9 are consistent10

with such stoichiometries based on the sizes ofthe A (20-32 kDa) and B (7-11 kDa) subunits

0 9 8 7 6 of these recognised enterotoxins. By extension,10-lo 10-9 1o-8 10-7 10o-6 10 a holotoxin consistent with a size of 55 kDa

Sheti (M) and an Aj:B5 structure would be predicted by,ure 7: ShETI dose response curve. Increasing these conventions. These tentative configur-icentrations ofpurified ShETI obtained as described in ations also satisfy the usual requirements for'thods were added to the niucosal side of rabbit ileut both a binding and an active domain that allowwunted in Ussing chambers and peak incremenits i71 Iscorded. No significant changes were found up to a toxin the enterotoxin to attach and gain entrance totcentration of 1o8 M. Higher concentrations induced a enterocytes and to initiate events that culmi-e dependent increment in Isc, with 50% of the effective nate in intestinal secretion. The data presented*e (ED -j) at 9X 1J0` M and a niaximal dose of 4xIrs6Further additions of the toxin did not induce significant in this paper suggest that, as for other entero-inges in Isc (n=4 aninials). toxins, the ShETI A subunit is responsible for

the secretory activity, whereas the B subunitsmay be involved in the irreversible binding of

Lterocyte) come from recent vaccine trials. the toxin to the enterocyte receptor. Withndberg and coworkers engineered a live, regular discoveries of additional members of aomatic dependent S flexneri vaccine candi- family of CT-/LT-like toxins among salmon-te by deleting the aroD gene.26 Although ella, pseudomonas, campylobacter, aeromo-ch a strain was able to invade epithelial cells, nas, and other genera, new combinations ofability to grow intracellularly was greatly A:B structures seem likely to emerge.

duced.27 When this vaccine was tested in It is intriguing that none of the currently)lunteers, some of them experienced a self identified intracellular messengers of intestinalniting watery diarrhoea, with no blood or secretion is involved in ShETI enterotoxicity.ucus in their faeces.28 However, the possibility that intestinal sec-Another vaccine candidate was engineered retagogues may act via new intracellulartransfer of the 140 mDa invasiveness plas- modulators is emerging both for bacterial

id from S flexnenr 5 and the chromosomal enterotoxins (RDEC enterotoxin, F Raimondignes encoding the group and type specific 0 and A Fasano, unpublished data) and inflam-

510

- F ^ w ^ ^ on Septem

ber 25, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.40.4.505 on 1 April 1997. D

ownloaded from

Enterotoxicity of Shigella enterotoxin 1

matory mediators such as adenosine and itsnon-metabolised analogues phenylisopropyladenosine (PIA) and N-ethlylcarboxamido-adenosine (NECA).3'- This new finding mayopen unexplored avenues on the intracellularsignalling involved in the intestinal fluid andelectrolyte transepithelial regulation. Experi-ments to confirm this hypothesis are currentlyin progress in our laboratory.

This work was supported by National Institute of Health grantsIROIDK48373 and IROIAI35740 to AF and N01AI45251,lUOlAI35948-01, and ROIAI29471 to MML.

1 Shiga K. Ueber den Dysenterriebacillus (Bacillus dys-eneriae). Zentbl Bakt ParasitKde Abt I Orig 1898; 24:817-24.

2 Makino S, Sasakawa C, Yoshikawa M. Genetic relatednessof the basic replicon of the virulence plasmid in theshigellae and enteroinvasive Escherichia coli. MicrobPathog 1988; 5: 267-74.

3 Sansonetti PJ, Kopecko DJ, Formal SB. Involvement of aplasmid in the invasive ability of S flexneri. Infect Immun1982; 35: 852-60.

4 Hale TL, Sansonetti PJ, Schad PA, Austin S, Formal SB.Characterization of virulence plasmids and plasmid-associated outer membrane proteins in S flexneri, Shigellasonnei, and Escherichia coli. Infect Immun 1983; 40:340-50.

5 Pal T, Formal SB, Hale TL. Characterization of virulencemarker antigen of Shigella spp and enteroinvasiveEscherichia coli. J Clin Microbiol 1989; 27: 561-3.

6 Venkatesan MM, Buysse JM, and Kopecko DJ. Character-ization of invasion plasmid antigen genes (ipaBCD) fromS flexneri. Proc Natl Acad Sci USA 1988; 85: 9317-21.

7 Taylor DN, Echeverria P, Pal T, Sethabutr 0, Saiborisuth S,Srichamorn S, et al. The role of Shigella spp, entero-invasive Escherichia cohl and other enteropathogens ascauses of childhood dysentery in Thailand. J Infect Dis1986; 153: 1132-8.

8 Taylor DN, Echeverria P, Sethabutr 0, Pitarangsi C,Leksomboon U, Blacklow N, et al. Clinical and micro-biological features of Shigella and enteroinvasive Escher-ichia coli infections detected by DNA hybridization. JClin Microbiol 1988; 26: 1362-6.

9 Fasano A, Noriega FR, Maneval RD Jr, Chanasongcram S,Russell R, Guandalini S, Levine MM. Shigella entero-toxin 1: an enterotoxin of Shigella flexneri 2a active inrabbit small intestine. JClin Invest 1995; 95: 2853-61.

10 Nataro JP, Seriwatana J, Fasano A, Maneval DR Jr,Guers LD, Noriega FR, et al. Identification and cloningof a novel plasmid-encoded enterotoxin in enteroinvasiveE coli and Shigella. Infect Immun 1995; 63: 4721-8.

11 Noriega FR, Ming Liao F, Formal SB, Fasano A,Levine MM. Prevalence of Shigella enterotoxin 1(ShET1) among Shigella clinical isolates of diverseserotypes.JtInfect Dis 1995; 172: 1408-10.

12 Fasano A, Kay BA, Russell BG, Maneval DR Jr, LevineMM. Enterotoxin and cytotoxin production by entero-invasive Escherichia coli. Infect Immun 1991; 58:3717-23.

13 Bradford MM. A rapid and sensitive method for thequantitation of micrograms quantities of protein utilizingthe principle of protein dye binding. Anal Biochem 1976;72: 248-54.

14 Ko YH, Thomas PJ, Delannoy MR, Pedersen PL. Thecystic fibrosis transmembrane canductance regulator.Over expression, purification, and characterization of wildtype and DF508 mutant forms of the first nucleotide

binding fold in fusion with the maltose-binding protein._JBiol Chem 1993; 268: 24330-8.

15 Fasano A, Hokama Y, Russell BG, Morris JG Jr. Diarrheain ciguatera fish poisoning: preliminary evaluation ofpathophysiological mechanisms. Gastroenterology 1991;100: 471-6.

16 Field M, Fromm D, McColl I. Ion transport in rabbit ilealmucosaI. Na and Cl fluxes and short circuit current. Am_J Physiol 197 1; 220: 1388-96.

17 Sladen GE, Harries JT. Studies on the effects of unconju-gated dihydroxy bile salts on rat small intestinal functionin vivo. Biochem Biophys Acta 1972; 288: 443-56.

18 LoSpalluto Jn, Finkelstein RA. Chemical and physicalproperties of cholera exo-enterotoxin (choleragen) and itsspontaneously formed toxoid (choleragenoid). BiochemBiophys Acta 1972; 257: 158-66.

19 Clements JD, Flint DC, Klipstein AF. Immunological andphysicochemical characterization of heat-labile entero-toxins isolated from two strains of Escherichia coli. InfectImmun 1982; 38: 806-809.

20 Olsnes S, Reisbig R, Eiklid K. Subunit structure of Shigellacytotoxin. I Biol Chem 1981 ; 256: 8732-8.

21 Seidah NG, Donohue-Rolfe A, Lazure C, Auclair F,Keusch GT, Chretein M. Complete amino acid sequenceof Shigella toxin-B chain. A novel polypeptide containing69 amino acids and one disulfide bridge. Jf Biol Chem1986; 261: 13928-31.

22 Casalino M, YusufMW, Nicoletti M, et al. A two year studyof enteric infections associated with diarrhoeal diseases inchildren in Urban Somalia. Trans R Soc Trop Med Hyg1988; 82: 637-41.

23 Ferrecio C, Prado V, Ojeda A, Cayyazo M, Abrego P,Guers L, Levine MM. Epidiemiologic patterns of acutediarrhea and endemic Shigella infections in children in apoor periurban setting in Santiago, Chile. Am Jf Epidemiol1991; 134: 614-27.

24 Mata U. The children of Santa Manra Cauque: a prospectivefield study of health and growth. Cambridge, MA: MITPress, 1978.

25 Piechaud D, Szturm-Rubistein S, D'Hauteville H, et al.Epidemiologie de la dysenterie bacillaire a Dijibouti.Bulletin de la Societe de Pathologie Exotique et de ses Filiales1971; 64: 37-42.

26 Lindberg AA, Karnell A, Pal T, Sweiha H, Hultenby K,Stocker BA. Shigella flexneri strain for use as a livevaccine. Microb Pathog 1990; 8: 433-40.

27 Karnell A, Stocker BAD, Katadura S, Reinholt FP,Lindberg AA. Live oral auxotrophic Shigella flexneri SFL12 vaccine with a deleted aroD gene; characterization andmonkey protection studies. Vaccine 1992; 10: 389-94.

28 Kamell A, Li A, Zhao CR, Karlsson K, Nguyen BM,Lindberg AA. Safety and immunogenicity study of theauxotrophic Shigella flexneri 2a vaccine SFL 1070 witha deleted aroD gene in adult Swedish volunteers. Vaccine1995; 1: 88-99.

29 Kotloff KL, Herrington DA, Hale TL, et al. Safety,immunogenicity and efficacy in monkeys and humans ofinvasive Escherichia coli K-12 hybrid vaccine candidatesexpressing Shigella flexneri 2a somatic antigen. InfectImmun 1992; 60: 2218-24.

30 Levine MM. Escherichia coli that cause diarrhea: entero-toxigenic, enteropathogenic, enteroinvasive, entero-hemorrhagic, and enteroadherent. 7 Infect Dis 1987; 155:377-89.

31 Barret KE, Dharmsathaphorn K. Mechanisms of chloridesecretion in a colonic epithelial cell line. In: Lebenthal E,Duffey M, eds. Texbook of secretory diarrhea. New York:Raven Press, 1990: 59-66.

32 Barret KE, Cohn JA, Huott PA, Wasserman SI,Dharmsathaphom K. Immune-related intestinal Cl-secretion. II. Effect of adenosine on the T84 cell line. AmJf Physiol 1990; 258: C902-12.

33 Barrett KE, Shin JA, MacMartin KA, Dharmsathaphorn K.A possible role for arachidonic acid in the chloridesecretory response induced by adenosine in T84 cells[abstract]. Gastroenterology 1988; 94: A24.

511

on Septem

ber 25, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.40.4.505 on 1 April 1997. D

ownloaded from