Vol. 57, No. 10INFECTION AND IMMUNITY, OCt. 1989, p. 2947-29520019-9567/89/102947-06$02.00/0Copyright © 1989, American Society for Microbiology

Host Response to Escherichia coli Heat-Labile Enterotoxin via TwoMicrovillus Membrane Receptors in the Rat Intestine

BORIS V. ZEMELMAN, SHU-HEH W. CHU,* AND W. ALLAN WALKERThe Combined Program in Pediatric Gastroenterology and Nutrition, The Children's Hospital and Massachusetts General

Hospital, and Department ofPediatrics, Harvard Medical School, Boston, Massachusetts 02115

Received 16 March 1989/Accepted 22 May 1989

The responsiveness of enterocytes to Escherichia coli heat-labile enterotoxin (LT) was studied in the smallintestine of 6- to 7-week-old rats. Dose-effect analysis showed the dose required for a 50% maximal LT-inducedsecretory response to be at 8 nM. After the well-documented glycolipid GM1 receptor was blocked with thecholera toxin B subunit, LT still activated the second messenger cascade, measured in terms of heightenedcellular adenylate cyclase activity, and caused fluid to be secreted into ligated intestinal loops. Furthermore,Scatchard analysis of binding kinetics suggested that LT bound to two receptor sites on the intestinalmicrovillus membrane. The toxin also bound to delipidated membrane but was competitively inhibited by agalactose-specific lectin, RCA60, suggesting that the additional receptor is a galactoglycoprotein. Western blotanalysis of toxin binding to membrane proteins revealed a group of binding components around 85 to 150kilodaltons. When measured at 2.2 nM LT, approximately 70% of LT-binding activity took place through ahigh-affinity (Kdl, 0.38 nM) GM1 receptor and 30% of LT-binding activity took place through a low-affinity(Kd2, 3.3 nM) glycoprotein receptor. These results suggest that LT functions through two microvillusmembrane receptors in the mature rat small intestine.

Enterotoxigenic Escherichia coli produces the heat-labileenterotoxin (LT) that causes diarrhea in humans (31) and inanimals (34). The E. coli LT affecting humans is a 91,400-dalton protein (6) whose structure, biological activity, andantigenic determinants closely resemble those of Vibriocholerae enterotoxin (CT) (5, 8).Like CT (9, 26), LT is a heterohexamer (13) made up of

five B subunits, which bind to the glycolipid GM1 receptor(17) on the surface of the intestinal epithelium, and a singleA subunit, which crosses the cell membrane, ADP ribosy-lates the Gs subunit of the GTP-binding regulatory protein(14), activates adenylate cyclase (EC 4.6.1.1) (10), anddisrupts ion transport across the epithelial membrane, caus-ing net fluid secretion (11).However, unlike CT, LT binds an additional glycoprotein

receptor on the surface of rabbit (15, 19, 30) and human (18)intestine. The glycoprotein receptor has not been previouslyobserved in the rat intestine. Furthermore, a dose-effectrelationship between the purified LT and its natural hosttarget tissue, the small intestine, has not been adequatelycharacterized.

In this study we have attempted to quantify the respon-siveness of the rat intestine to LT. We have also tried tocharacterize the glycoprotein receptors and to assess theircontribution in the diarrheal effect of the toxin. The ratswere chosen because they were useful models in our earlierinvestigation of the development of the intestinal host de-fense against CT (3, 22, 33). Previous studies have indicatedthat a mature intestine is less susceptible than an immatureintestine to CT infection (3, 33). The data obtained in thisstudy may be used to compare the relative sensitivity of thehost to CT and LT and to provide information for subse-quent developmental studies.

* Corresponding author.

MATERIALS AND METHODSAnimals. Female Sprague-Dawley rats of the CD strain

were purchased from the Charles River Breeding Laborato-ries, Inc. (Wilmington, Mass.). Animals were housed in ananimal room with a 12-h light-dark cycle and fed Purina ratchow (Ralston Purina Co., St. Louis, Mo.) and water adlibitum.

Bacterial toxins. LT isolated from a genetically con-structed strain, MM294(pWD600) (23), which was clonedfrom a human enterotoxigenic strain of E. coli (H74-114)(12), was generously donated by R. A. Finkelstein (Depart-ment of Microbiology, University of Missouri, Columbia).CT and the pentamer of cholera toxin B subunit (cholera-genoid; CT-B) were purchased from List Biological Labora-tories (Campbell, Calif.).

Intestinal loop assay. Female 6- to 7-week-old rats, weigh-ing 150 to 175 g, were fasted for 18 h. The animals wereanesthetized with ether. Two 5-cm loops were ligated on theproximal small intestine, and two were ligated on the distalsmall intestine. One loop in each pair was injected with toxinin 0.5 ml of phosphate-buffered saline (136 mM NaCl, 2.6mM KCl, 8 mM Na2HPO4, and 1.5 mM KH2PO4 [pH 7.3]),containing 0.1% bovine serum albumin; the other loop wasinjected with phosphate-buffered saline-bovine serum albu-min buffer (control). Net fluid accumulation in the twotoxin-treated loops was determined 4 h after the toxin wasadministered and is expressed as microliters per centimeterof gut length. The dose required for a 50% maximal response(ED50) was estimated by using a dose-effect analysis (2)program (Elsevier-Biosoft, Cambridge, United Kingdom) onan IBM personal computer. In blocking the toxin-inducedfluid secretion, a 100-fold excess of CT-B was used. The CTB subunit was introduced into the appropriate loops 3 to 5min before the injection of LT.

Adenylate cyclase assay. Intestinal mucosa was scrapedfrom each loop and homogenized in 0.1 M Tris hydrochlo-ride buffer (pH 7.4) containing 10 mM MgCl2 and 1 mMEDTA. The homogenates were centrifuged at 105,000 x g

2947

on April 9, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

2948 ZEMELMAN ET AL.

for 1 h. The pellets were suspended in the same buffer byhomogenization and used for the enzyme assay. The incu-bation mixture (in a total volume of 0.1 ml) contained 30 mMTris hydrochloride (pH 7.4), 4 mM MgCl2, 0.1 mM EDTA, 3mM ATP, an ATP-regenerating system (10 mM creatinephosphate and 0.2 mg of creatine phosphokinase per ml), 10mM theophylline, and 10 to 50 pLg of protein. The reactionmixture was incubated at 37°C for 10 min and stopped byheating at 100°C for 3 min. The amount of cyclic AMPproduced was determined with a cyclic AMP assay kit fromAmersham Corp. (Arlington Heights, Ill.). Protein concen-trations were determined by the method of Lowry et al. (24)with bovine serum albumin as a standard.MVM preparation and delipidation. Microvillus membrane

(MVM) was prepared from the proximal and distal smallintestine by MgC12 precipitation by a modification of themethod of Hause et al. (16) as previously described (4), but1 mM phenylmethylsulfonyl fluoride (a protease inhibitor)was added to the buffer used. Membrane preparation used inthis study yielded an approximately 18-fold increase insucrase activity (7). MVM lipid was extracted twice, eachtime with 20 volumes of chloroform-methanol-water (4:8:3,vol/vol/vol) (35). The mixture was centrifuged at 2,000 x gfor 10 min. The resulting delipidated MVM pellets along withnative MVM preparations were suspended by homogeniza-tion in 10 mM Tris hydrochloride containing 1 mM EGTA(pH 7.4) and used for the toxin-binding assay.

Toxin iodination. The toxin was iodinated with lodo-beads(Pierce Chemical Co., Rockfold, Ill.) as described by Mark-well (25). The labeled toxin was separated from the unre-acted radioactive iodine on a Sephadex G-25 column (PD-10;Pharmacia Fine Chemicals, Uppsala, Sweden). The concen-tration of labeled toxin was determined by A280. The averagespecific activity of [125I]LT was 640 Ci/mmol.

Toxin-binding assay. Total and nonspecific binding of LTto MVM was determined by incubation of MVM (0.5 ,ug ofprotein) with various concentrations of [1251]LT in the ab-sence and presence of 550 nM unlabeled LT in a bindingbuffer (pH 7.4) containing 25 mM Tris hydrochloride, 135mM NaCl, 1 mM EDTA, and 0.1% bovine serum albumin.The reaction (in a total volume of 50 ,u) was incubated at37°C for 20 min and terminated by quenching with 0.2 ml ofice-cold binding buffer, followed by rapid suction filtrationwith a microsample filtration manifold with nitrocellulosefilters (type BA-85; Schleicher & Schuell Co., Keene, N.H.).Filters were then rinsed 10 times with 0.2 ml of ice-coldbinding buffer and counted in a gamma counter (Pharmacia,Gaithersburg, Md.). Specific binding was determined bysubtracting nonspecific binding from total binding. The bind-ing affinity, estimated by the apparent dissociation constant(Kd), and the maximal binding (Bmax) were calculated usingthe least-square fit computer program LIGAND of Munsonand Rodbard (28) and modified (Elsevier-Biosoft, Cam-bridge, United Kingdom) for use on the IBM personalcomputer.

Detection of a glycoprotein receptor by Western blots.MVM was separated by sodium dodecyl sulfate-polyacryla-mide gel electrophoresis by the method of Laemmli (21)under reducing conditions on a 5 to 15% gradient acrylamidegel and transferred by electroblotting (36) onto a BioTrace0.45-,um nylon matrix (Gelman Sciences, Inc., Keene,N.H.). The membrane was incubated with blotto buffer (5%[wt/vol] nonfat dry milk, 50 mM Tris hydrochloride [pH 7.8],2 mM CaCl2, 0.01% antifoam, 0.05% Triton X-100) at roomtemperature for 2 h to block nonspecific binding sites (20).The membrane was subsequently probed with [125I]LT (107

200

zP 150

yl100

50

z

1 2 3 4TIME (hours)

FIG. 1. Time-course changes of LT-induced fluid secretion in theproximal and distal small intestine of rats. LT (25 jig/ml) in 0.5 ml ofphosphate-buffered saline containing 0.1% bovine serum albuminwas injected into 5-cm loops of 6- to 7-week-old rats.

cpm/ml) at 4°C for 2 h and was washed at room temperaturewith blotto (four times, 20 min each), phosphate-bufferedsaline (four times, 5 min each), and H20 (twice, 5 min each).After the membrane was dried, it was autoradiographed byplacing it against Kodak XAR-5 film (Eastman Kodak Co.,Rochester, N.Y.).

RESULTS

Time course and dose response of LT-induced fluid secre-tion. In the time-course analysis of fluid secreted into ligatedintestinal loops, fluid began to accumulate 2 h after theinjection of toxin and reached a steady state in 4 h (Fig. 1).A lag phase occurred because the LT A subunit entered theepithelial cell from the intestinal lumen and had to translo-cate to the basal-lateral membrane, where it constitutivelyactivated the Gs protein, leading to induction of adenylatecyclase. There was no significant different in the net amountof fluid secreted into the proximal and the distal loops inresponse to LT. Data obtained from both gut regions werepooled to construct the dose-response curve (Fig. 2). Theresults show that the LT-induced secretory response is dosedependent. The maximal effective dose, 55 nM LT (5 ,ug/ml),led to a net accumulation of 200 RI of fluid per cm in theligated intestinal loops 4 h after toxin injection. The ED50was estimated from the dose-response curve to be 8 nM.

Inhibition of fluid secretion by CT B subunit. An attemptwas made to block the effect of LT on the epithelium ofligated intestinal loops. CT-B, the binding subunit of the

z0 200F=5

2 - 1501

1001

i 505z

0.11 1.1 11

[LT] (nM)

1101100

FIG. 2. Dose-response curve of LT-stimulated fluid secretionand the blocking effect by CT B subunit in the rat small intestine.Net fluid accumulation was determined 4 h after toxin exposure in 6-to 7-week-old rats. Each point represents the mean standard error

of two or three assays. *, Significantly (P < 0.05) different betweenblocked and nonblocked intestinal loops by Student's t test.

*-*proximal / - --0- -Odistal i

/f/

//- O~~~~~

* stimulated (LT)O blocked (LT+CTB)

INFECT. IMMUN.

on April 9, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

E. COLI HEAT-LABILE TOXIN AND RAT INTESTINE 2949

.t20 * m_ basal (none)

Zt ___ stimulated (LT)IC

15.r- blocked (LT+CTB)

lo.0 i 150

~~~~E5~ ~ ~ ~~55 6

1.1 2.8 55 5

[LT] (nM)FIG. 3. LT-induced adenylate cyclase activation and the block-

ing effect by CT B subunit in the rat small intestine. Enzyme activitywas measured in the particulate fraction of mucosal homogenatesafter 4 h of incubation with toxin in vivo. Each bar represents themean plus standard error of at least three mucosal samples.

cholera toxin, specifically binds the microvillus receptor,ganglioside GM1, but has no secretory effect on the intes-tine. In fact, an excess of this subunit completely inhibitedthe CT effect on the intestine when 60 nM of CT, a dose (3)known to produce maximal secretory response (205 ± 48,ul/cm over 4-h period), was used. If LT only recognized theganglioside GM1, an excess of CT-B would inhibit its action,and no fluid would be expected to accumulate in the ligatedloop. However, if LT recognized a different or an additionalmembrane receptor, net fluid accumulation in the loopshould be reduced or uninhibited. We noted a reduction inthe amount of fluid that accumulated in loops pretreated withCT-B subunit (Fig. 2). Furthermore, the results show a moreeffective inhibition of the LT effect (-70%) at lower toxindoses (-1 to 5 nM), a less effective inhibition (-30%) athigher doses (-20 to 40 nM), and no inhibition beyond themaximal effective doses (>55 nM). These findings suggestthat LT recognized GM1 as well as an additional receptor(s).

Dose-dependent activation of adenylate cyclase. The abso-lute adenylate cyclase activity in control and LT-stimulatedintestinal loops was difficult to determine since variationfrom animal to animal was substantial. However, whenloops from the same animal were compared, a consistenttwo- to threefold induction of adenylate cyclase activity wasseen in loops treated with LT (Fig. 3). Adenylate cyclaseactivity also seemed to respond to an increase in the toxindose, with the maximum activity (20 pmol of cyclic AMPproduced per min pre mg of membrane protein) reached at 55nM LT (5 jig/ml). This toxin dose also led to maximal fluidsecretion.

Effect of CT B-subunit blocking on adenylate cyclase acti-vation. We assayed the adenylate cyclase activity of ligatedloops where the toxin effect was partially inhibited by apretreatment with CT-B (Fig. 3). The cyclase activity inthese loops was below the levels in loops treated with LTalone. At the same time, this enzyme activity was signifi-cantly above the basal level. In fact, the pattern of adenylatecyclase activity in loops treated with LT and with LT plusCT-B was similar to the pattern of net fluid accumulation inthese loops. At lower toxin doses, CT-B was able to main-tain adenylate cyclase activity close to the basal level,whereas at higher doses, the inhibition of activities was lesssuccessful, and near-maximal cyclase activity resulted.These findings are not surprising if we consider that adenyl-ate cyclase activation leads to altered ion flux across theintestine, which results in fluid secretion. The fluctuations in

30.

E v

0 2 4 6 8 10 12

[LT] (nM)FIG. 4. Saturation binding of LT to MVM of the proximal small

intestine of 6- to 7-week-old rats. Each point represents the mean ofthree membrane preparations.

adenylate cyclase activity support a multireceptor modelaccording to which, when GM1 has been blocked, LT at highdoses acts through an alternate low-affinity receptor(s) toactivate the enzyme.

Binding of LT to MVM. MVM preparations and 125I-labeled LT were used to study the binding kinetics of toxinreceptors in the rat small intestine. The binding of LT toMVM was dose dependent and saturable (Fig. 4). Scatchardanalysis (32) of the binding data (Fig. 5) suggests the exist-ence of two classes of receptors (Kdl, 0.38 nM; Bmaxi, 12.5pmol/mg of protein; Kd2, 3.3 nM, Bmax2, 17.5 pmol/mg ofprotein). The high-affinity receptor corresponds to the gan-glioside GM1.

Binding of LT to delipidated MVM. To estimate thecontribution of the additional receptor sites, glycolipids wereextracted from the MVM. Comparison of the binding of LTto native and delipidated membrane fractions showed thatGM1 was responsible for approximately 70 to 80% of LTbinding (Table 1). The results also indicated that the specificbinding of LT to delipidated MVM, like its binding to nativeMVM, was dose dependent (Table 1), with 2.6 pmol of LTbound per mg of protein at a low toxin dose (0.55 nM) and5.2 pmol of LT bound at a high toxin dose (2.2 nM). Thus,these data suggest that, in addition to GM1, LT may bebinding a single glycoprotein receptor or a family of recep-tors that exhibits low affinity for the toxin.

Competition analysis of LT binding to delipidated MVM.Ricinus communis agglutinin (RCA60), a galactose-specificlectin, successfully competed with radiolabeled LT for bind-ing to delipidated MVM (Fig. 6). Preincubating the mem-brane with various concentrations of the lectin prevented thebinding of [125I]LT (2.2 nM) in a dose-dependent fashion. At

1IEz00

0.5

0.4-

0.3

0.2- l

0.1 -

0.0---0.0 0.1 0.2 0.3 0.4 0.5

BOUND (nM)

FIG. 5. Scatchard plot analysis of LT binding to MVM for Fig. 4.

VOL. 57, 1989

on April 9, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

2950 ZEMELMAN ET AL.

TABLE 1. Comparison of the specific binding of [125I]LT tonative and delipidated MVM of rat proximal small intestine

[(25I]LT bound to microvillus membraneaLT concn (pmol/mg of protein)

(nM) Delipidated/nativeNative Delipidatedrairatio

0.55 11.4 ± 2.3 2.6 ± 0.4 22.82.2 17.2 ± 3.0 5.2 ± 0.9 30.2

a Means ± standard errors of triplicate determinations.

1.2 mg of RCA60 per ml, LT binding constituted 30% ofcontrol. This result suggest that the additional receptorsite(s) for the toxin is galactoglycoprotein(s).Western blot analysis of LT binding. Sodium dodecyl

sulfate-polyacrylamide gel electrophoresis was probed with[1251I]LT. Autoradiography identified bands in two regions ofthe gel (Fig. 7). A prominent band with low molecular massat a dye front corresponding to ganglioside GM1 was presentin all of the membrane preparations. Several smaller bandswith molecular masses around 85 to 150 kilodaltons wereprobably membrane glycoproteins. More binding compo-nents were detected in the distal MVM than in the proximalMVM prepared from the mature intestine, but no proteinband was detected in a 2-week-old neonatal intestine.

DISCUSSION

The findings presented in this study show that a functionalgalactoglycoprotein heat-labile enterotoxin receptor site ex-ists on the rat intestinal epithelium. Similar studies con-ducted on rabbit (15, 19, 30) and human (18) intestine havealso pointed to the existence of such a site in addition to thewell-documented ganglioside GM1 receptor.The toxin, which causes secretory diarrhea by activating

an adenylate cyclase-dependent second messenger pathwayin the enterocyte, is known to act through the GM1. The datain this study show that the diarrheal effect of the toxin inligated intestinal loops persisted even when the GM1 hadbeen blocked. It has also been observed that the fluidsecretion is dose dependent in both blocked and nonblockedloops. However, the pattern of fluid secretion is different inthe two systems. In the loops blocked with CT B subunit,less fluid was secreted at the lower toxin doses; whereas at

120*

1004z 280-00:_I 60-

9-'-'40-20~

200 400 600 800 1000 1200 1400

[RCAW] fi&/ml)

FIG. 6. Inhibition of [125I]LT binding to delipidated MVM byRCA60. Delipidated MVM was prepared from the proximal smallintestine of 6- to 7-week-old rats. Various concentrations of RCA60were preincubated with 0.5 ,ug of membrane protein at 37°C for 30min. The binding of ['25I]LT to membrane was assayed at a toxinconcentration of 2.2 nM. Each point represents the mean of tripli-cate determinations.

1 2 3 4

200-

92-

69-

46-

30-

14- di GM

FIG. 7. Western blot analysis of ['25I]LT binding to the MVM ofrat small intestine. The MVM preparation (100 p.g of protein) wasseparated by sodium dodecyl sulfate-polyacrylamide gel electropho-resis on a 5 to 15% gradient gel under reducing conditions, trans-ferred to a nylon filter, and probed with [125I]LT. An autoradiographof the filter is shown. Lanes: 1, 2-week proximal gut; 2, 2-weekdistal gut; 3, from 6- to 7-week proximal gut; 4, from 6- to 7-weekdistal gut. The arrows indicated MVM components that bound to[125ILT.

the higher doses, similar amounts of fluid collected in bothtypes of loops. These findings suggest that a glycoprotein LTreceptor(s) might have a lower affinity for the toxin than theGM1. Scatchard analysis supports this conclusion. Becauseof the difference in receptor affinity, naturally occurringquantities of toxin most likely are affecting the epitheliumprimarily through the GM1 in the rat intestine.

This study also indicates that, in causing fluid secretion,LT activates the adenylate cyclase second messenger cas-cade. Dose-dependent activation of adenylate cyclase wasseen in the mucosal samples of CT-B-blocked and non-blocked intestinal loops. In all instances, however, thenonblocked samples exhibited higher adenylate cyclase ac-tivity.The receptor-binding studies also support a two-receptor

model. Scatchard analysis of LT binding to the native MVMwas characteristic of two receptors, whereas a similar bind-ing of CT to the MVM of rat intestine suggested only onereceptor (3, 22). LT bound specifically to the delipidatedMVM, suggesting that the additional LT receptor was aglycoprotein. Furthermore, galactose-specific lectin (RCA60)inhibited LT binding to delipidated MVM, suggesting thatthe additional receptor has galactose residues essential fortoxin binding. Finally, Western blot analysis of LT bindingproved that the toxin associated with membrane proteins ofa large molecular mass (85 to 150 kilodaltons) along with thesmall molecular glycolipid GM1.The significance of the multi-band pattern of toxin-binding

proteins remains to be elucidated. However, there is someindication that the pattern depends upon the age of theanimal and the region of the small intestine. These dataprobably can be explained by the previous finding that theexpression of cell-surface galactose onMVM is developmen-tally (29) and regionally (27) regulated in the rat smallintestine. The multi-band pattern might possibly reflect thelectinlike binding of LT to MVM. Such binding characteris-tics have been noted with the pertussis toxin binding toChinese hamster ovary cells (1). Since the membrane com-ponents were separated under reducing conditions, it is alsopossible that the bands might represent subunits of a singlepolypeptide receptor that binds LT.The ratio of glycolipid GM1 and glycoprotein receptors for

LT in the small intestine appears to vary from species to

p..

p.

INFECV. IMMUN.

IMP-

on April 9, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

E. COLI HEAT-LABILE TOXIN AND RAT INTESTINE 2951

species. For example, about 50% of the receptor sites for LTin the human intestine are glycoprotein (18), whereas inrabbit intestine, glycoprotein receptor sites appear to out-number the GM1-binding sites by more than 10-fold (19).Our data show that, in rat intestine, about 30% of thereceptor sites are contributed by delipidated MVM (glyco-protein) when measured with 2.2 nM LT. Comparison ofthese data indicates that the ratio of the two toxin receptorsin humans is much closer to that in rats than to that inrabbits. Thus, the rat seems to serve as a better animalmodel than the rabbit for the study of LT interaction with theenterocyte.The biological efficacy of LT was compared with that of

CT (3), in terms of the relative sensitivity of the mature ratintestine to the two enterotoxins. The data show that, on amolar basis, LT (ED50, 8 nM) was more effective than CT(ED50, 38.9 nM) in triggering a stimulus-secretory response,although both LT and CT bound to GM1 with equally highaffinity (Kd, 0.38 and 0.42 nM, respectively). Furthermore,when the ratio of Kd to ED50 was used as an estimate ofcoupling efficiency for initial receptor binding to the finalsecretory response, the coupling efficiency was higher forLT (5%) than for CT (1%). Therefore, the difference in thehost responsiveness to these two toxins seemed to be relatedto postreceptor events, not to the initial receptor occupancy.A similar finding was previously noted for the increasedsensitivity in the host responsiveness to CT of the immaturesuckling rat (3). Whether developmental changes alter thehost responsiveness of LT is under investigation.

ACKNOWLEDGMENTS

This work was supported by Public Health Service grants DK-37521 and HD-12437 from the National Institutes of Health.We thank Ilene Ely for her technical assistance, R. A. Finkelstein

(University of Missouri) for supplying the E. coli LT used in thisstudy, and D. C. Robertson (University of Kansas) for use of the LTfrom E. coli 286-C2 in a pilot fluid secretion study.

LITERATURE CITED1. Brennan, M. J., J. L. David, J. G. Kenimer, and C. R.

Manclark. 1988. Lectin-like binding of pertussis toxin to a165-kilodalton Chinese hamster ovary cell glycoprotein. J. Biol.Chem. 263:4895-4899.

2. Chou, T. C., and P. Talalay. 1984. Quantitative analysis ofdose-effect relationship: the combined effect of multiple drugsor enzyme inhibitors. Adv. Enzyme Regul. 22:27-55.

3. Chu, S. W., I. G. Ely, and W. A. Walker. 1989. Age andcortisone alter host responsiveness to cholera toxin in thedeveloping gut. Am. J. Physiol. 256:G220-G225.

4. Chu, S. W., and W. A. Walker. 1988. Development of thegastrointestinal mucosal barrier: changes in phospholipid headgroups and fatty acid composition of intestinal microvillusmembranes from newborn and adult rats. Pediatr. Res. 23:439-442.

5. Clements, J. D., and R. A. Finkelstein. 1978. Demonstrating ofshared and unique immunological determinants in enterotoxinsfrom Vibrio cholerae and Escherichia coli. Infect. Immun.22:709-713.

6. Clements, J. D., R. J. Yancey, and R. A. Finkelstein. 1980.Properties of homogeneous heat-labile enterotoxin from Esche-richia coli. Infect. Immun. 29:91-97.

7. Dahlqvist, A. 1968. Assay of intestinal disaccharidases. Anal.Biochem. 22:99-107.

8. Dallas, W. S., and S. Falkow. 1979. The molecular nature ofheat-labile enterotoxin (LT) of Escherichia coli. Nature (Lon-don) 227:406-407.

9. Eidels, L., R. L. Proia, and D. A. Hart. 1983. Membranereceptors for bacterial toxins. Microbiol. Rev. 47:596-620.

10. Evans, D. J., L. C. Chen, G. T. Curlin, and D. G. Evans. 1972.

Stimulation of adenyl cyclase by Escherichia coli enterotoxin.Nature (London) 236:137-138.

11. Field, M. 1979. Intracellular mediators of secretion in the smallintestine, p. 83-91. In H. J. Binder (ed.), Mechanisms ofintestinal secretion. Alan R. Liss, New York.

12. Geary, S. J., B. A. Marchlewicz, and R. A. Finkelstein. 1982.Comparison of heat-labile enterotoxins from porcine and humanstrains of Escherichia coli. Infect. Immun. 36:215-220.

13. Gill, D. M., J. D. Clements, D. C. Robertson, and R. A.Finkelstein. 1981. Subunit number and arrangement in Esche-richia coli heat-labile enterotoxin. Infect. Immun. 33:677-682.

14. Gill, D. M., and S. H. Richardson. 1980. Adenosine diphos-phate-ribosylation of adenylate cyclase catalized by heat-labileenterotoxin of Escherichia coli: comparison with cholera toxin.J. Infect. Dis. 141:64-70.

15. Griffiths, S. L., R. A. Finkelstein, and R. R. Crithley. 1986.Characterization of the receptor for cholera toxin and Esche-richia coli heat-labile toxin in rabbit intestinal brush borders.Biochem. J. 238:313-322.

16. Hauser, H. K., R. M. Howell, C. Dawson, and D. E. Bowyer.1980. Rabbit small intestinal brush border membrane prepara-tion and lipid composition. Biochim. Biophys. Acta 602:567-577.

17. Holmgren, J. 1973. Comparison of the tissue receptors forVibrio cholerae and Escherichia coli enterotoxins by means ofgangliosides and natural cholera toxoid. Infect. Immun. 8:851-859.

18. Holmgren, J., M. Lindblad, P. Fredman, L. Svennerholm, andH. Myrvold. 1985. Comparison of receptors for cholera andEscherichia coli enterotoxins in human intestine. Gastroenter-ology 89:27-35.

19. Holmgren, J., P. Fredman, M. Lindbald, A.-M. Svennerholm,and L. Svennerholm. 1982. Rabbit intestinal glycoprotein recep-tor for Escherichia coli heat-labile enterotoxin lacking affinityfor cholera toxin. Infect. Immun. 38:424-433.

20. Johnson, D. A., J. W. Gautsch, J. R. Sportsman, and J. H.Elder. 1984. Improved technique utilizing nonfat dry milk foranalysis of proteins and nucleic acids transferred to nitrocellu-lose. Gene Anal. Tech. 1:3-8.

21. Laemmli, U. K. 1970. Cleavage of structural protein duringassembly of the head of bacteriophage T4. Nature (London)222:680-685.

22. Lencer, W. I., S. W. Chu, and W. A. Walker. 1987. Differentialbinding kinetics of cholera toxin to intestinal microvillus mem-brane during development. Infect. Immun. 55:3126-3130.

23. Leong, J., A. C. Vinal, and W. S. Dallas. 1985. Nucleotidesequence comparison between heat-labile toxin B-subunit cis-trons from Escherichia coli of human and porcine origin. Infect.Immun. 48:73-77.

24. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall.1951. Protein measurement with the Folin phenol reagent. J.Biol. Chem. 193:265-275.

25. Markwell, M. 1982. A new solid-state reagent to iodinateproteins. I. Conditions for the efficient labeling of antisera.Anal. Biochem. 125:427-432.

26. Middlebrook, J. L., and R. B. Dorland. 1984. Bacterial toxins:cellular mechanisms of action. Microbiol. Rev. 48:199-221.

27. Morita, A., S. Miura, R. H. Erickson, M. H. Sleisenger, andY. S. Kim. 1986. Comparison of brush border membrane glyco-proteins and glycoenzymes in the proximal and distal rat smallintestine. Biochim. Biophys. Act 883:506-516.

28. Munson, P. J., and D. Rodbard. 1980. Ligand: a versatilecomputerized approach for characterization of ligand-bindingsystems. Anal. Biochem. 107:220-239.

29. Ozaki, C. K., S. W. Chu, and W. A. Walker. 1989. Develop-mental changes in galactosyltransferase activity in the rat smallintestine. Biochim. Biophys. Acta 991:243-248.

30. Pierce, N. F. 1973. Differential inhibitory effects of choleratoxoids and ganglioside on the enterotoxins of Vibrio choleraeand Escherichia coli. J. Exp. Med. 137:1009-1023.

31. Sack, R. B. 1975. Human diarrheal disease caused by entero-toxigenic Escherichia coli. Annu. Rev. Microbiol. 29:333-353.

32. Scatchard, G. 1949. The attractions of proteins for small mole-

VOL. 57, 1989

on April 9, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

2952 ZEMELMAN ET AL. INFECT. IMMUN.

cules and ions. Ann. N.Y. Acad. Sci. 51:660-672.

33. Seo, J. K., S. W. Chu, and W. A. Walker. 1989. Development ofintestinal host defense: an increased sensitivity in the adenylatecyclase response to cholera toxin in suckling rats. Pediatr. Res.25:225-227.

34. Smith, H. W., and S. Hall. 1967. Studies on Escherichia colienterotoxin. J. Pathol. Bacteriol. 93:531-543.

35. Svennerholm, L., and P. Fredman. 1980. A procedure for thequantitiative isolation of brain ganglosides. Biochim. Biophys.Acta 617:97-109.

36. Towbin, H., T. Staehelin, and J. Gordon. 1979. Electrophoretictransfer of proteins from polyacrylamide gels to nitrocellulosesheets: procedure and some applications. Proc. Natl. Acad. Sci.USA 76:4350-4354.

on April 9, 2021 by guest

http://iai.asm.org/

Dow

nloaded from