Vol. 59, No. 10INFECTION AND IMMUNITY, OCt. 1991, p. 3708-37140019-9567/91/103708-07$02.00/0Copyright © 1991, American Society for Microbiology

Efficacy of Enteric-Coated Protease in Preventing Attachment ofEnterotoxigenic Escherichia coli and Diarrheal Disease in the

RITARD ModelTRACEY L. MYNOTT,1* DAVID S. CHANDLER,2 AND RICHARD K. J. LUKE'

School ofAgriculture, La Trobe University, Bundoora, Victoria 3083,' and Victorian Institute ofAnimal Science,Attwood, Victoria 3049,2 Australia

Received 26 April 1991/Accepted 1 July 1991

In this study, we report on a novel approach based on modification of the intestinal surface to preventdiarrhea caused by enterotoxigenic Escherichia coli (ETEC). The removable intestinal tie adult rabbit diarrhea(RITARD) model was used to test the efficacy of an enteric-coated protease preparation (Detach; EnzacorTechnology Pty. Ltd.) in the prevention of bacterial attachment and diarrheal disease caused by colonizationfactor antigen I-positive (CFA/I+) E. coli H10407. Protease was administered orally to rabbits 18 h prior tochallenge with 10" bacteria. Four groups of rabbits were inoculated with different ETEC strains whichproduced different combinations of adhesin and enterotoxin or with sterile phosphate-buffered saline.Occurrence of diarrhea during the subsequent 24-h incubation period was recorded. Oral administration ofprotease was successful in reducing diarrhea and diarrhea-induced death in six of seven (86%) rabbits infectedwith CFA/I+, heat-stable and heat-labile toxin-positive E. coli (H10407). Seven of eight (87%) rabbits notprotected by protease treatment died or developed severe diarrhea. Quantitative analysis of bacterial culturesobtained from the small intestine of rabbits showed a significant (P < 0.001) 2,000-fold reduction in CFU percentimeter of intestine following treatment with protease. The efficacy of protease treatment was 99.5%, withvery wide confidence limits (>0 to 99.9%). The data indicate that the use of protease to prevent ETECdiarrheal disease has considerable potential.

The role of enterotoxigenic Escherichia coli (ETEC) as animportant etiologic agent in human diarrheal disease is wellestablished (31). These organisms are characterized by theirability to produce heat-labile toxin (LT) and/or heat-stabletoxin (ST) (11). Some strains also produce pilus adhesinscalled colonization factor antigens (CFAs). These adhesinspermit attachment of ETEC strains to the intestinal mucosa,thereby facilitating colonization and delivery of enterotoxinto target epithelial cells.CFAs may be of use as vaccine candidates and for this

reason have attracted considerable attention. Those identi-fied to date include CFA/I and CFA/III and two multivariantantigen groups known as CFA/II and CFA/IV. Each of thelast two possesses a combination of three pilus adhesins,designated coli surface antigens (CS). CS1, CS2, and CS3are present on CFA/II E. coli, and CS4, CS5, and CS6 arepresent on CFA/IV E. coli (7, 16, 21, 28). Other putativecolonization factors have also been described previously (5,14, 34, 36). As colonization factors are antigenically distinct,potential vaccines must be multivalent. A prototype ETECvaccine that contains known CFAs and outer membraneantigens associated with ETEC is being developed (32).

Significant protection against diarrhea caused by ETECfollowing vaccination with a combination of formalin andheat-inactivated whole Vibrio cholerae cells and the purifiedB subunit of cholera toxin has been reported recently (4). V.cholerae and ETEC produce heat-labile enterotoxins whichare structurally, functionally, and immunologically similarand consist of a combination of a biologically active Asubunit and five B subunits (12, 15, 22).Other efforts to develop vaccines against ETEC infection

* Corresponding author.

have given variable results (8, 18, 19), and while the prospectof effective immunization against diarrheal diseases of hu-mans is appealing, effective vaccines are not expected to beavailable in the near future (16a). Alternative approaches tothe prevention of ETEC diarrhea are therefore being pur-sued.

Studies with pig ETEC that possess the K88 adhesin haveindicated that the interaction between ETEC and pig brushborder membranes can be modified by the use of proteases(27). Preliminary challenge experiments and field trials (2a)have indicated that an enterically protected protease prepa-ration (Detach; Enzacor Technology Pty. Ltd., Melbourne,Australia) administered orally is able to modify the mucosalsurface of the piglet small intestine such that diarrhea is lesslikely to occur.A number of studies have described similarities in mech-

anisms of pathogenesis of ETEC infection in humans andanimals (11, 31). This has led us to investigate the use ofprotease in the prevention of human diarrheal diseases.Recent experiments (23a) have indicated that binding ofCFA/I and CFA/II to human intestinal mucosa can beprevented in vitro by the use of protease. An effect ofprotease in reducing the binding of LT was also demon-strated. It appears that modification of intestinal mucosa toprevent human diarrheal diseases has considerable potential.This report describes an experiment in which the removableintestinal tie adult rabbit diarrhea (RITARD) model de-scribed by Spira et al. (29) was used to test the efficacy ofexogenous protease (Detach) in reducing attachment ofCFA/I-positive E. coli to rabbit intestinal mucosa in vivo.The work represents a novel approach to preventing attach-ment of ETEC to intestinal cells and therefore preventingdiarrheal disease.

3708

on Septem

ber 9, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PROTEASE AND PREVENTION OF ETEC DIARRHEA 3709

MATERIALS AND METHODS

Animals. Forty-four New Zealand White breed rabbits ofboth sexes from a single breeder were used for the experi-ment. Their weights ranged from 1.5 to 2.7 kg. Animals wereacclimatized to their animal housing facility for at least 1week prior to the start of experimentation.

Bacteria. ETEC strains used in this study were originallyisolated in Bangladesh from patients with diarrhea. StrainH10407 (serotype 078:K80:H11) and a mutant derivative ofthis strain, H10407-P, were kindly provided by D. G. Evans(The University of Texas Medical School at Houston, Hous-ton). Strain E1392/75 7A (serotype 06:K15:H16) was kindlyprovided by B. Rowe (Division of Enteric Pathogens, Lon-don, United Kingdom). Strain H10407 produces both ST andLT and possesses CFA/I. Strain H10407-P produces both STand LT but does not produce CFA/I (7). Strain E1392/75 7Ais a CFA-negative, nontoxigenic spontaneous laboratoryderivative of CFA/II1 E. coli 1392 (17) and has been shownnot to colonize or induce diarrhea in the RITARD model(35).

Stock cultures of all strains were suspended in Trypticasesoy broth (Oxoid) containing 15% (vol/vol) glycerol andstored in multiple aliquots at -80°C. A new aliquot was usedfor each experiment. Bacteria were inoculated onto CFAagar (6) and grown at 37°C overnight. Cultures were har-vested by means of a flamed Pasteur pipette, washed insterile phosphate-buffered saline (0.01 M, pH 7.2; PBS), anddiluted to yield the desired optical density measurements.The bacterial concentration was confirmed by viable cellcount on duplicate blood agar plates after serial dilution inPBS. Prior to the inoculation of rabbits, all cultures werechecked for the presence of CFA/I and LT by a specificenzyme immunoassay (EIA).

Antigens. CFA/I was purified as previously described (6).Purified LT (unnicked) was kindly provided by J. D. Clem-ents (Walter Reed Army Institute of Research, Washington,D.C.).

Antisera. Specific CFA/I antiserum was produced bygiving rabbits three subcutaneous injections with 60 p.g ofpurified CFA/I at intervals of 4 weeks. For the first immu-nization, the antigen was emulsified with Freund's completeadjuvant. Subsequent immunizations were emulsified withFreund's incomplete adjuvant. The animals were bled by theear vein 2 weeks after the final immunization. LT-specificantiserum was similarly produced by injecting 60 ,ug ofpurified LT. The immunoglobulin G fraction was preparedby protein A affinity chromatography with protein A-Seph-arose CL-4B (Pharmacia) as specified by the manufacturer.

Detection of CFAII and LT by EIA. Disposable polystyrenemicrotiter plates (Nunc, Roskilde, Denmark) were used forall assays. CFA/I- or LT-specific immunoglobulin G wasdiluted in sodium bicarbonate buffer (0.1 M, pH 9.6) andadsorbed to wells (100 pl per well) by incubation overnight at4°C. For CFA/I detection, bacteria were diluted to approx-imately 2.5 x 109 cells per ml (A625 = 1.0) in working dilutionbuffer containing PBS (0.01 M, pH 7.2), bovine serumalbumin (0.25%; Fraction V; Sigma), Tween 20 (0.05%;Sigma), disodium salt EDTA (0.0075%), and sodium azide(0.02%). For LT detection, supernatant material was ob-tained following centrifugation (at 12,000 x g for 15 min) ofovernight cultures of bacteria grown in CYE medium (9).The bacterial suspension or culture supernatant was inoc-

ulated into duplicate wells (100 ,ul per well). Bound materialwas detected with urease-conjugated (CSL, Melbourne,Australia) specific immunoglobulin G diluted in working

dilution buffer containing 1% hen egg albumin (Fraction II;Sigma). Urea substrate (0.008% bromocresol purple [Sigma],0.1% urea [Bio-Rad], 0.0075% disodium salt EDTA; pH 4.8,100 ,ul) was used, and a positive reaction was indicated by acolor change from yellow to purple. During each of theabove steps, plates were incubated at 37°C for 30 min.Between each of the steps, supernatant liquid was removedfrom the wells, which were then washed with washing buffer(0.01 M PBS, 0.05% Tween 20). Prior to the addition ofsubstrate, wells were washed with distilled water to removeresidual buffer. Development of purple color was monitoredat A540.RITARD model procedure. The RITARD model developed

by Spira et al. (29) was used with slight modifications. At 18h prior to challenge, half of the rabbits from each group (seeTable 1) were given a single oral dose of 0.42 g of enteric-coated protease granules (Detach containing 25% protease;Enzacor Technology Pty. Ltd.). The protease granules wereplaced in gelatin capsules and administered by placing thecapsules at the back of the throat. Food was withheld afterdosing, but water was made available ad libitum. Beforesurgery, animals were anesthetized with 16 mg of xylazine(Rompun; Bayer) and 100 mg of ketamine (Ketapex; ApexLaboratories Pty. Ltd) intramuscularly. The incision sitewas anesthetized with 2 ml of lignocaine with adrenalin (20mg of lignocaine hydrochloride per ml, 0.01 mg of adrenalin[as bitartrate] [Apex Laboratories Pty. Ltd] per ml). Thececum was exteriorized through a midline incision andligated permanently as close to the ileo-cecal junction aspossible with no. 11 umbilical tape (Ethicon). At this timethe ileum was temporarily obstructed approximately 10 cmproximal to the cecal tie with a slipknot tie with umbilicaltape. Inoculum (10 ml) containing 1011 bacteria diluted insterile PBS or sterile PBS alone was injected into theduodenum. The intestine was returned to the peritonealcavity, which was sutured with one end of the temporary tieleft accessible through the incision. The ileal tie was gentlyremoved 2 h after bacterial challenge, and the rabbits werereturned to their cages, where food and water were availablead libitum. All surviving animals were killed 24 h postchal-lenge by means of barbiturate overdose; some animalsinfected with the CFAII+ bacteria died within the 24 h.Immediately after death, each animal was autopsied and theperitoneal cavity was swabbed for bacterial culture.The E. coli enterotoxin and adhesin combinations were

selected to include a CFA-positive ETEC strain (H10407), aCFA-negative ETEC strain (H10407-P), and a CFA-nega-tive, nontoxigenic strain (E1392/75 7A). A PBS control wasincluded to enable the effect of surgery and protease treat-ment in the absence of bacterial challenge to be monitored.The challenge rate of 1011 bacteria was based on earlierexperiments in which such a dose of organisms was requiredto induce diarrhea in 90% of infected animals (1). In thepresent study, surgery was performed on rabbits in groups ofeight over a period of 4 weeks (two rabbits from each groupeach week; see Table 1). Two weeks later, 12 rabbits werechallenged with H10407.

Monitoring of infection. Rabbits were observed hourly forthe 24-h postchallenge period for diarrhea, weakness, ordeath. Diarrhea was scored as follows: 0, no diarrhea; 1,mild diarrhea with feces softer than normal; 2, moderatediarrhea with at least three watery stools; and 3, severediarrhea with multiple watery stools. Fecal swabs werecollected when feces were passed, and rectal swabs weretaken from rabbits which did not pass feces. Challengeorganisms were identified by means of typical E. coli colony

VOL. 59, 1991

on Septem

ber 9, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

3710 MYNOTT ET AL.

TABLE 1. Diarrheal response in rabbits treated with Detach oruntreated and challenged with different ETEC strains

Strain Adhesin Toxin Treat- Diarrheal Small intestinementa responseb fluid vol (ml)

H10407C CFA/I+ ST+ LT+ D 1/7d 35-50C 7/8 20-105f

H10407-P CFA/1- ST+ LT+ D 1/49 10-50C 1/49 10-50

E1392/75 7A CFA/II- ST- LT- D 0/4 15-50C 0/4 15-50

Nil (PBS) D 0/4 8-12C 0/4 24-40

a D, rabbits treated with Detach; C, untreated rabbits.b Number of animals with diarrhea or death/total number tested. For

H10407-challenged rabbits, P < 0.001: one of seven Detach-treated animalsdied compared with seven of eight untreated animals challenged with the samedose rate.

c Five rabbits were omitted from the analysis because of death not relatedto diarrhea.

d One Detach-treated rabbit died; CFU/cm at S3 was 1.2 x 107.e One rabbit survived infection; CFU/cm at S3 was 5.8 x 109.f The total volume of fluid accumulated in small and large intestines

combined was 130 to 165 ml.g Mild diarrhea (score 1).

morphology and enzyme-linked immunosorbent assay nitro-cellulose replica methods as described previously (20).

Collection of tissue specimens. All animals were sacrificed24 h postchallenge, and the intraluminal fluid of the smallintestine was measured. Large fluid volumes (>60 ml) in thesmall intestine of euthanized or dead rabbits were taken asan indication that diarrhea had been a major contribution todeath (29). Sections (2 by 3 cm) of small intestine werecollected from five sites: duodenum (Si), proximal jejunum(S2), midjejunum (S3), distal jejunum (S4), and ileum (S5).Each segment was opened longitudinally and washed exten-sively in sterile PBS to determine the numbers of stronglyadherent bacteria or left unwashed to determine the totalbacterial numbers present. Quantitative cultures were pre-pared by homogenizing tissue for 1 min in a Sorvall homog-enizer operated at full speed. Serial dilutions were made inPBS, and aliquots (25 ,ul) were plated onto sheep blood agar(5%, vol/vol) and CFA agar. After incubation at 37°C for 18h, the number of CFU per centimeter of tissue was deter-mined. Other specimens were processed promptly for his-tology following fixation in 10% neutral-buffered formalin.After specimens had been embedded in paraffin, they werestained with hematoxylin-eosin stain and tissue Gram stain.

Statistical analysis. Bacterial counts were converted by logtransformation to stabilize variances and analyzed by meansof Genstat V for analysis of variance. Data are expressed asthe mean values ± the standard error of the mean. Theefficacy of Detach (protease) protection was determined byFortran-Finney, a program that determines efficacies (per-cents) from chemotherapeutic tests (10).

RESULTS

Diarrheal response. None of the rabbits challenged withnontoxigenic E1392/75 7A or given 10 ml of sterile PBSdeveloped diarrhea (Table 1). At autopsy, the fluid volumesin the small intestine (pyloric sphincter to the ileo-cecaljunction) ranged from 8 to 12 ml in the PBS-protease-treatedgroup to 24 to 40 ml in the PBS-treated, non-protease-treatedgroup. In rabbits challenged with E1392/5 7A, 15 to 50 ml offluid accumulated. Two of eight rabbits that were challenged

with the CFA-negative, toxigenic H10407-P passed fecesthat were softer than normal (score 1). This was only milddiarrhea and not considered to be important. One rabbit wasin the protease-treated group, and one was in the non-pro-tease-treated group. In rabbits challenged with H10407-P,the fluid volumes in the small intestine ranged from 10 to 50ml. There was no significant difference between the amountsof fluid accumulated by the rabbits challenged with the twoCFA-negative strains.

In total, 20 rabbits were challenged with the enterotoxi-genic CFA/I-positive strain H10407. Five died within 4 h ofchallenge, but none of these had typical acute watery diar-rhea. These rabbits were autopsied, and peritoneal swabswere plated on blood agar and MacConkey agar. Leakage ofbacteria from the inoculation site was deemed to be thecause of death, as colonies of typical E. coli morphologywere isolated from the peritoneum of these rabbits. No suchbacteria were detected in peritoneal swabs from the rabbitswhich died in a diarrhea-related manner or in swabs from theremaining rabbits autopsied at 24 h. These five rabbits wereomitted from the experiment, leaving seven protease-treatedrabbits and eight untreated rabbits challenged with H10407.Of the eight control (non-protease-treated) rabbits chal-

lenged with H10407, seven died or developed severe diar-rhea. Six rabbits died 5 to 13 h postchallenge, and four ofthese died without passing feces. Of the two rabbits thatsurvived the 24-h postchallenge period, one had profusewatery diarrhea 18 h postchallenge while the other did notpass feces and had no clinical signs of infection. At autopsy,the fluid volume in the small intestines of rabbits which didnot pass feces ranged from 20 to 105 ml. The total volume inthe small and large intestine combined, however, rangedfrom 130 to 165 ml (in comparison with 10 to 50 ml in rabbitsinoculated with E1392/75 7A, H10407-P, and PBS). In com-parison, the small intestine contained relatively little fluid (35to 50 ml) in rabbits that had diarrhea at the time of death. Thetotal volume of intestinal fluid was only 55 to 60 ml. A lackof fluid accumulation in rabbits that have passed feces at thetime of death has been reported previously (29).Of the rabbits treated with protease prior to H10407

challenge, only one died. This rabbit died 11 h postchallengeafter passing one loose stool (score 1). Fluid volumes in thesmall and large intestines were 60 and 50 ml, respectively.None of the other six rabbits treated with protease haddiarrhea, and the majority (four of six) had passed formedfeces by 24 h. At autopsy, the contents of the large intestinewere solid and fluid accumulation in the small intestineranged from 12 to 60 ml.

Bacterial adhesion. Quantitative cultures (CFU) were pre-pared from all animals to determine the adhesion of chal-lenge bacteria in different parts of the small intestine. Chal-lenge bacteria were present at all sites, with CFA/I+ H10407being the most effective colonizer. The mean CFU levels ofCFA/I+ bacteria at sampling sites in non-protease-treatedrabbits were lower at S1, S2, S4, and S5 (1.1 x 108, 6.7 x

107, 1.0 x 108, and 4.0 x 108 CFU/cm, respectively) than atS3 (5.5 x 109 CFU/cm). In the analysis which follows, S3cultures are used for comparison. The recovery rates ofdifferent organisms from 1 cm of mucosa are shown in Fig. 1.The number of CFA/I+ bacteria adherent to the mucosa in

the protease-treated rabbits ranged from 1.3 x 104 (minimumcount) to 1.2 x 107 CFU/cm in the rabbit that died (arith-metic mean, 2.6 x 106 CFU/cm). Values for control rabbitschallenged with the same strain were 2,000 times greater (P< 0.001). Figure 2 illustrates the differences in colony countsbetween protease-treated and untreated animals.

INFECT. IMMUN.

on Septem

ber 9, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PROTEASE AND PREVENTION OF ETEC DIARRHEA 3711

CFU/cm

1010

109

108

107

106

105

104

0

9

000

8

U

U

U

U

U

Diarrhoea0

U

8 U

a

00

UU

0U

0U

No Diarrhoea

0

.

Noa0

H10407 H10407-P

0

E1392/757A

cica

PBS

FIG. 1. Range of quantitative cultures at midjejunum (S3) of rabbits challenged with different ETEC strains or given sterile PBS. *, rabbitstreated with Detach 18 h prior to challenge; 0, untreated.

The number of bacteria bound to the small intestinalmucosa of rabbits infected with the CFA/I- strain H10407-Pranged from 1.3 x 104 CFU/cm (minimum count) to 6.6 x107 CFU/cm in a rabbit with mild diarrhea (arithmetic mean,1.8 x . CFU/cm). Counts for strain E1392/75 7A weresimilar, ranging from 1.3 x 104 (minimum count) to 2 x 108CFIJ/cm (arithmetic mean, 3.9 x 107 CFU/cm). There wereno significant differences between bacterial numbers in pro-tease-treated and non-protease-treated animals challengedwith either CFA-negative strain.

In rabbits which received sterile PBS only, there wererelatively few bacteria in the small intestine (1.3 x 104CFU/cm). There were, however, two exceptions. Two con-trol rabbits which did not receive protease had colony countsof 4.6 x 106 and 5.2 x 10' CFU/cm at S3 (small intestinalvolumes of 40 and 35 ml, respectively). One of these rabbitshad 5.1 x 1010 and 4.1 x 1010 CFU/cm at Si and S5,respectively. All colonies observed had typical E. coli mor-phology, and it seems likely that there was cross infectionfrom one or more other bacterial challenge groups. Thesecolonies were not checked for the presence of CFA/I.Williams-Smith and Halls (37) have reported that E. coli arenot isolated as part of the normal rabbit flora, and it istherefore unlikely that the E. coli observed were the result ofovergrowth of endogenous bacteria.To assess the total number of bacteria present on the gut

mucosa, sections were left unwashed. The mean values ofwashed versus unwashed sections from rabbits challengedwith CFA/I+ are shown in Table 2. These results suggestthat the organisms are strongly adherent to the mucosa, asonly 102 bacteria were washed away.

Fecal excretion of bacteria. Fecal swabs were obtained

from rabbits when feces were passed. In all animals thechallenge bacteria were excreted. Rectal swabs were takenat autopsy. The presence of the challenge strain in therectum was apparent in all rabbits, including those that hadnot passed feces prior to termination of the experiment. Inall instances, 100% of the colonies cultured were of thechallenge strain.

Histology. Histological studies (light microscopy) of smallintestinal tissue revealed no mucosal abnormality in any ofthe rabbits. Bacteria were only rarely seen on the mucosa,suggesting that bacteria bound in particular areas rather thanbeing evenly distributed along the mucosa.

DISCUSSION

The decision to investigate the use of protease to modifythe intestinal surface and thereby prevent diarrhea in hu-mans was based on previous prevention of K88+ E. coliinfection in piglets (2a). Several studies have describedsimilarities in mechanisms of pathogenesis of ETEC infec-tion in humans and animals (11, 31). Most ETEC strains ofhuman and animal origin rely on pili for adhesion andsubsequent colonization of the small intestine. Also, diar-rheal disease in both species is dependent ultimately onproduction and efficient delivery of enterotoxin.The interaction between K88 adhesin and its intestinal

receptor is one of the most studied host-pathogen associa-tions, while relatively little is known about the interactionbetween CFA adhesins and human brush border mem-branes. The epithelial receptor for attachment of K88+ETEC to the brush border membrane of piglets is known tobe a multimeric glycoprotein (30). This and other receptors

VOL. 59, 1991

on Septem

ber 9, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

3712 MYNOTT ET AL.

Mean (log ICFU/cm

Site 1 Site 3 Site 5

Mean (log 10)CFU/cm

10-

9

8 -

7-

6-

5-

4-

3-

2-

1

A I

BX, I

Site 1 Site 3 Site 5

Mean (log 1O)CFU/cm

C

T /

10 -

8-

7 -

6

5-

4-

3-

2-

1-

0Site 1 Site 3 Site 5

FIG. 2. Mean bacterial counts at duodenum (site 1), midjejunum (site 3), and ileum (site 5) of rabbits challenged with H10407 (A),H10407-P (B), E1392/75 7A (C), or sterile PBS (D). Each point represents the mean (log1o) standard error CFU per centimeter of tissue.*, rabbits treated with Detach 18 h prior to bacterial challenge; 0, untreated rabbits.

involved in attachment (and colonization) have been shownto be readily inactivated by proteolytic enzymes, includingthose proteases that are normally active in the small intestine(23, 27, 30). This was recently investigated in experimentsobserving the instability of K88 receptor within pig intestines(3a). The variability of receptor activity in intestinal contentswas confirmed by EIA as described previously (3), andinstability of receptor activity could be controlled by addi-tion of trypsin inhibitor to sample collection buffers. Thistechnique was also used to demonstrate an effect of anexogenous enzyme (Detach) on a K88 intestinal glycoproteinreceptor, thereby confirming the disruptive influence ofprotease on the binding of K88 adhesin to intestinal tissue.The efficacy of Detach under field conditions has beenconfirmed by field trials in commercial piggeries where largereductions in diarrhea-induced mortality and the incidenceof diarrhea have been observed.

Recently we developed an in vitro (EIA) technique formonitoring the interaction among brush border preparationsderived from human small intestine, ETEC, and LT. Theeffects of enzymatic treatment indicate that receptors for

TABLE 2. Quantitative cultures of H10407 bacteria boundat five different sites

Small tb Bacterial count (CFU/cm)cintestine sitea TreatmenWashed Unwashed

Si D 5.72 + 0.40 6.96 + 0.34C 7.25 t 0.35 8.13 ± 0.44

S2 D 5.80 ± 0.28 6.90 ± 0.46C 7.30 ± 0.35 8.26 ± 0.31

S3 D 5.56 ± 0.40 7.04 ± 0.50C 7.87 ± 0.60 9.29 ± 0.38

S4 D 6.11 ± 0.38 7.20 + 0.63C 7.66 ± 0.24 9.46 ± 0.29

S5 D 6.66 ± 0.42 8.80 ± 0.58C 7.22 + 0.47 8.93 ± 0.36

a Si, duodenum; S2, proximal jejunum; S3, midjejunum; S4, distal jejunum;S5, ileum.

b D, Detach-treated rabbits; C, untreated rabbits.c Mean loglo + standard error recovered per centimeter of washed and

unwashed small intestine at death or 24 h after RITARD challenge. Each meanreflects data from six or seven rabbits.

Mean (log l)CFU/cm

10:1-g_98 -

7 -

6 -

5-

4-

3-

2-

I1-

10-

9

8 -

7-

6

5-

4-

3-

2-

1

D

. - I~~~~-

Site 1 Site 3 Site 5

INFECT. IMMUN.

'k

on Septem

ber 9, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PROTEASE AND PREVENTION OF ETEC DIARRHEA 3713

CFA/I and CFA/II may be sialoglycoproteins on the intesti-nal brush border membrane. Receptors for CFA/I on eryth-rocytes may also be sialoglycoproteins (2, 13, 25). Binding ofCFA/I, CFA/II, and LT to the human mucosa in the EIA isinterfered with by protease.The concept of surface modification to control human

diarrheal disease has now been tested in vivo by means ofthe RITARD model. This model, devised by Spira et al. (29),has previously been used extensively to study the pathogen-esis of ETEC and V. cholerae infection and to study protec-tive immunity afforded by vaccine candidates (1, 24, 26, 33).The present study has demonstrated that oral administrationof Detach, an enteric-coated protease preparation, wassuccessful in reducing diarrhea and diarrhea-induced deathby 86% (six of seven) in rabbits infected with the CFA/Istrain H10407. Of the eight control rabbits which did notreceive protease, seven (87%) died or developed severediarrhea. One of the non-protease-treated rabbits survivedthe infection and did not develop diarrhea despite thepresence of large numbers of CFA/I1 bacteria adherent tothe mucosa (5.8 x 109 CFU/cm). Resistance of RITARDrabbits to large challenge doses of CFA/I1 bacteria has beenobserved previously (1, 26). This may reflect physiological,genetic, or immune variation. Conversely, one -of the sevenprotease-treated rabbits died. It is possible that stress causedby oral dosing and starvation prior to surgery induced gutstasis, thereby preventing passage of protease through thestomach. Reliable specific protease detection assays arecurrently being developed to enable movement of Detachthrough the gastrointestinal tract to be monitored.Wanke and Guerrant (35) have shown that enterotoxigenic

CFA/II1 bacteria (strain E1392) induce diarrhea in RITARDrabbits and colonize to a level of more than 108.5 CFU/cm2 ofsmall intestine. None of rabbits given CFA-negative 1392-bacteria were colonized with more than 108 CFU/cm2, andnone developed diarrhea. Their report shows that the thresh-old for expression of clinical symptoms of diarrheal infectionwas 108 CFU/cm2 (35). In the present study, rabbits chal-lenged with bacterial strains possessing no known coloniza-tion factors did not develop diarrhea and were colonized tolevels below 5 x 107 CFU/cm. In these rabbits there was nosignificant difference between colonization of protease-treated and non-protease-treated groups. In contrast, non-protease-treated rabbits challenged with the CFA/I+ strainH10407 were colonized to levels well above 107 CFU/cm(arithmetic mean, 6.2 x 109). Seven of eight of these rabbitseither developed severe diarrhea or died in a diarrhea-relatedmanner. Rabbits treated with protease and challenged withstrain H10407 were colonized to levels below 107 CFU/cm(arithmetic mean, 2.6 x 106 CFU/cm), i.e., levels similar tothose observed when the challenge strain produced noknown CFA (Fig. 1). This represents a more than 2,000-foldreduction in CFU per centimeter of tissue. It is apparenttherefore that oral treatment with protease was successful inmodifying the surface of the rabbit mucosa such that colo-nization of CFA/I1 bacteria was significantly (P < 0.001)reduced.

In summary, oral administration of an enteric-coatedprotease preparation was successful in reducing diarrhea anddiarrhea-induced death by 86% (P < 0.001) in rabbits chal-lenged with CFA/I-positive H10407. Quantitative analysis ofthe cultures obtained from the small intestine of these rabbitsshowed a significant (P < 0.001) treatment effect, with agreater than 2,000-fold reduction in colonies per centimeterof intestine. The reduction brings the number of bacteriabelow the threshold required for the development of diar-

rhea. The efficacy of protection based on the ability ofprotease treatment to reduce colony numbers was deter-mined to be 99.5% (>O to 99.9%) (10). Although based on asmall number of animals, the data indicate that the conceptof surface modification to prevent diarrheal disease hasconsiderable potential. Further studies are required to deter-mine whether this concept can be applied to prevent ETEC-induced diarrhea in humans.

ACKNOWLEDGMENTS

We thank Alex Hauler and Barry Heywood for their invaluableassistance and advice. Analysis of the data by Leigh Callinan,Victorian Department of Agriculture Biometric Services, is alsogreatly appreciated.We also acknowledge the support of Enzacor Technology Pty.

Ltd.

REFERENCES1. Ahren, C. M., and A.-M. Svennerholm. 1985. Experimental

enterotoxin-induced Escherichia coli diarrhea and protectioninduced by previous infection with bacteria of the same adhesinor enterotoxin type. Infect. Immun. 50:255-261.

2. Bartus, H., P. Actor, E. Snipes, D. Sedlock, and I. Zajac. 1985.Indications that the erythrocyte receptor involved in enterotoxi-genic Escherichia coli attachment is a sialoglycoconjugate. J.Clin. Microbiol. 21:951-954.

2a.Chandler, D. S., et al. Unpublished data.3. Chandler, D. S., H. M. Chandler, R. K. J. Luke, S. R. Tzipori,

and J. A. Craven. 1986. Screening of pig intestines for K88non-adhesive phenotype by enzyme immunoassay. Vet. Micro-biol. 11:153-161.

3a.Chandler, D. S., T. L. Mynott, R. K. J. Luke, and J. H. Craven.Submitted for publication.

4. Clemens, J. D., D. A. Sack, J. R. Harris, J. Chakraborty, P. K.Neogy, B. Stanton, N. Huda, M. U. Khan, B. A. Kay, M. R.Khan, M. Ansaruzzaman, M. Yunus, M. Raghava Rao, A.-M.Svennerholm, and J. Holmgren. 1988. Cross protection by Bsubunit-whole cell cholera vaccine against diarrhoea associatedwith heat labile toxin-producing enterotoxigenic Escherichiacoli: results of a large scale field trial. J. Infect. Dis. 158:372-377.

5. Darfeuille-Michaud, A., C. Forestier, B. Joly, and R. Cluzel.1986. Identification of a nonfimbrial adhesive factor of anenterotoxigenic Escherichia coli strain. Infect. Immun. 52:468-475.

6. Evans, D. G., D. J. Evans, Jr., S. Clegg, and J. A. Pauley. 1979.Purification and characterization of the CFA/I antigen of entero-toxigenic Escherichia coli. Infect. Immun. 25:738-748.

7. Evans, D. G., D. J. Evans, Jr., W. S. Tjoa, and H. L. DuPont.1978. Detection and characterization of colonization factor ofenterotoxigenic Escherichia coli isolated from adults with diar-rhea. Infect. Immun. 19:727-736.

8. Evans, D. G., D. Y. Graham, D. J. Evans, Jr., and A. Opebun.1984. Administration of purified colonisation factor antigens(CFA/I, CFA/II) of enterotoxigenic Escherichia coli to volun-teers. Gastroenterology 87:934-940.

9. Evans, D. J., Jr., D. G. Evans, and S. L. Gorbach. 1973.Production of vascular permeability factor by enterotoxigenicEscherichia coli isolated from man. Infect. Immun. 8:725-730.

10. Finney, D. J. 1952. Statistical method in biological assay.Charles Griffin & Co. Ltd., Wycombe, United Kingdom.

11. Gaastra, W., and F. K. de Graaf. 1982. Host-specific fimbrialadhesins of noninvasive enterotoxigenic Escherichia colistrains. Microbiol. Rev. 46:129-161.

12. Gill, D. M., D. J. Evans, Jr., and D. G. Evans. 1976. Mecha-nisms of activation of adenylate cyclase in vitro by polymyxin-released heat-labile enterotoxin of Escherichia coli. J. Infect.Dis. 133(Suppl.):S103-S107.

13. Gonzalez, E. A., and J. Blanco. 1985. Comparative study ofinhibition of mannose-resistant haemagglutination caused byCFA/I, CFAIII, K88 and K99-positive Escherichia coli strains.

VOL. 59, 1991

on Septem

ber 9, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

3714 MYNOTT ET AL.

FEMS Microbiol. Lett. 29:115-121.14. Heuzenroeder, M. W., T. R. Elliot, C. J. Thomas, R. Halter, and

P. A. Manning. 1990. A new fimbrial type (PCF 09) on entero-toxigenic Escherichia coli 09:H-LT+ isolated from a case ofinfant diarrhoea in central Australia. FEMS Microbiol. Lett.66:55-60.

15. Holmgren, J. 1981. Actions of cholera toxin and the preventionand treatment of cholera. Nature (London) 292:413-417.

16. Honda, T., M. Arita, and T. Miwatani. 1984. Characterization ofnew hydrophobic pili of human enterotoxigenic Escherichiacoli: a possible new colonization factor. Infect. Immun. 43:959-965.

16a.Levine, M. M. Personal communication.17. Levine, M. M., R. E. Black, M. L. Clements, C. R. Young, C. P.

Cheney, P. Schad, H. Collins, and E. C. Boedeker. 1984.Prevention of enterotoxigenic Escherichia coli diarrhoeal infec-tion in man by vaccines' that stimulate anti-adhesion (anti-pili)immunity, p. 224-244. In E. C. Boedeker (ed.), Attachment oforganisms to the gut mucosa. CRC Press, Inc., Boca Raton, Fla.

18. Levine, M. M., J. B. Kaper, R. E. Black, and M. L. Clements.1983. New knowledge of pathogenesis of bacterial entericinfections as applied to vaccine development. Microbiol. Rev.47:510-550.

19. Levine, M. M., P. Ristaipo, G. Marley, C. Smyth, S. Knutton, E.Boedeker, R. Black, C. Young, M. L. Clements, C. Cheney, andR. Patnaik. 1984. Coli surface antigens 1 and 3 of colonizationfactor antigen II-positive enterotoxigenic Escherichia coli: mor-phology, purification, and immune responses in humans. Infect.Immun. 44:409-420.

20. L6pez-Vidal, Y., and A.-M. Svennerholm. 1986. Hemadsorptionand enzyme-linked immunosorbent assay nitrocellulose replicamethoUds for identification of colonization factor antigen (CFA)-positive Escherichia coli colonies and for isolation of CFA-negative mutants. J. Clin. Microbiol. 24:615-619.

21. McConnell, M. M., L. V. Thomas, G. A. Willshaw, H. R. Smith,and B. Rowe. 1988. Genetic control and properties of colisurface antigens of colonization factor antigen IV (PCF8775) ofenterotoxigenic Escherichia coli. Infect. Immun. 56:1974-1980.

22. Moss, J., J. C. Osborne, Jr., P. H. Fishman, S. Nakaya, andD. C. Robertson. 1981. Escherichia coli heat-labile enterotoxinganglioside specificity and ADP-ribosyltransferase activity. J.Biol. Chem. 256:12861-12865.

23. Mouricout, M. A., and R. A. Julien. 1987. Pilus-mediatedbinding of bovine enterotoxigenic Escherichia coli to calf smallintestinal mucins. Infect. Immun. 55:1216-1223.

23a.Mynott, T. L., et al. Unpublished data.24. Pierce, N. F., W. C. Cra'y, Jr., J. B. Kaper, and J. J. Mekalanos.

1988. Determinants of immunogenicity and mechanisms of

protection by virulent and mutant Vibrio cholerae 01 in rabbits.Infect. Immun. 56:142-148.

25. Pieroni, P., E. A. Worobec, W. Paranchych, and G. D. Arm-strong. 1988. Identification of a human erythrocyte receptor forcolonization factor antigen I pili expressed by H10407 entero-toxigenic Escherichia coli. Infect. Immun. 56:1334-1340.

26. Sack, R. B., R. L. Kline, and W. M. Spira. 1988. Oral immuni-zation of rabbits with enterotoxigenic Escherichia coli protectsagainst intraintestinal challenge. Infect. Immun. 56:387-394.

27. Sellwood, R. 1980. The interaction of the K88 antigen withporcine intestinal epithelial cell brush borders. Biochim.Biophys. Acta 632:326-335.

28. Smyth, C. J. 1982. Two mannose-resistant haemagglutinins onenterotoxigenic Escherichia coli of serotype 06:H16 or H-isolated from traveller's and infantile diarrhoea. J. Gen. Micro-biol. 128:2081-2096.

29. Spira, W. M., R. B. Sack, and J. L. Froehlich. 1981. Simpleadult rabbit model for Vibrio cholerae and enterotoxigenicEscherichia coli diarrhea. Infect. Immun. 32:739-747.

30. Staley, T. E., and J. B. Wilson. 1983. Soluble pig intestinal cellmembrane components with affinities for Escherichia coli K88+antigen. Mol. Cell. Biochem. 52:177-189.

31. Sussman, M. 1985. Escherichia coli in human and animaldisease, P. 7-45. In M. Sussman (ed.), The virulence of Esche-richia coli: reviews and methods. Academic Press, Inc., NewYork.

32. Svennerholm, A.-M., J. Holmgren, and D. A. Sack. 1989.Development of oral vaccines against enterotoxinogenic Esch-erichia coli diarrhoea. Vaccine'7:196-198.

33. Svennerholm, A.-M., Y. L. Vidal, J. Holmgren, M. M. McCon-nell, and B. Rowe. 1988. Role of PCF8775 antigen and its colisurface subcomponents for colonization, disease, and protec-tive immunogenicity of enterotoxigenic Escherichia coli in rab-bits. Infect. Immun. 56:523-528.

34. Tacket, C. O., D. R. Maneval, and M. M. Levine. 1987.Purification, morphology, and genetics of a new fimbrial puta-tive colonization 'factor of enterotoxigenic Escherichia coli0159:H4. Infect. Immun. 55:1063-1069.

35. Wanke, C. A., and R. L. Guerrant. 1987. Small-bowel coloni-zation alone is a cause of diarrhea. Infect. Immun. 55:1924-1926.

36. Wevers, P., R. Picken, G. Schmidt, B. Jann, K. Jann, J. R.Golecki, and M. Kist. 1980. Characterization of pili associatedwith Escherichia coli 018ac. Infect. Immun. 29:685-691.

37. Williams-Smith, H., and S. Halls. 1967. Observations by theligated intestinal segment and oral inoculation methods onEscherichia coli infections in pigs, calves, lambs and rabbits. J.Pathol. Bacteriol 93:499-529.

INFECT. IMMUN.

on Septem

ber 9, 2020 by guesthttp://iai.asm

.org/D

ownloaded from