THE JOURNAL OF INFECTIOUS DISEASES • VOL. 133, SUPPLEMENT © 1976 by the University of Chicago. All rights reserved.
• MARCH 1976
Activation of Heat-Labile Escherichia coli Enterotoxin by Trypsin From the Department of Biological Product Development, Wyeth Laboratories, Philadelphia, Pennsylvania
Ruth S. Rappaport, Jerome F. Sagin, Walter A. Pierzchala, Geraldine Bonde, Benjamin A. Rubin, and Howard Tint
Since the isolation of enterotoxigenic Escherichia coli strains from piglets [1, 2] and humans [3] with acute cholera-like disease, it has been established that many enteropathogenic strains of E. coli, unlike Vibrio cholerae, elaborate both a heat-labile and a heat-stable enterotoxin [4]. Present evidence indicates that the synthesis of both types of enterotoxin may be under the control of a plasmid [5]. These findings have stimulated efforts to isolate the E. coli enterotoxins and to compare their structural and functional properties with those of cholera toxin, the well-characterized diarrhea-producing enterotoxin. Although the heat-labile enterotoxin was first described in 1967 [1], attempts to purify this toxin, which is thought to be closely related to cholera toxin, have been hindered primarily by the inability to produce large amounts of biologically active, cell-free enterotoxin and by a considerable heterogeneity in the size of the active
molecules [6-9]. In the course of our own attempts to solve these problems, it was discovered that trypsin treatment of cell-free filtrates from cultures of enterotoxigenic strains resulted in an elevation of preexisting rabbit skin vascular permeability factor (PF) activity. This observation suggested the possibility that heat-labile E. coli enterotoxins might represent a new class of microbial prototoxins, prior examples of which include the s-toxin of Clostridium periringens type D [10], the i-toxin of Clostridium perjringens type E [11], and the neurotoxin of Clostridium botulinum type E [12]. The purpose of the present investigation was, therefore, to study the biological properties of heat-labile enterotoxin, before and after treatment with trypsin, and to consider the possibility that proteolytic activation of the enterotoxin might play a role in the pathogenesis of the disease. Materials and Methods
This work was supported by the U.S. Army Research and Development Command under research contract no. DADA-17-74-C-4007. The authors are grateful to Mr. Tom McCann for outstanding technical assistance, Mr. Jack Kassarich for radioimmunoassay of growth hormone, and Ms. Terry Schaffer for expert secretarial assistance. Please address requests for reprints to Dr. Ruth Rappaport, Wyeth Laboratories, P.O. Box 8299, Philadelphia, Pennsylvania 19101.
Bacterial strains. Six enterotoxigenic human strains and one nonenterotoxigenic human strain were obtained from Col. Joseph Metzger (Fort Detrick, Maryland). The toxigenic strains, most of which have been described [3, 13, 14], were designated as follows: H334(OI5:Hll), H210 (085:H7), HI06-3(078:H11), H408-3(078:H 12), HI0407(078:Hll), and HI97(078:H11);
841
Downloaded from http://jid.oxfordjournals.org/ at Florida Atlantic University on September 9, 2015
Trypsin-treated, cell-free filtrates derived from enterotoxigenic Escherichia coli, strain H197 (078:Hll), exhibited a fourfold or greater increase in heat-labile vascular permeability factor activity and a lO-fold or greater increase in the ability to stimulate secretion of growth hormone by cultured rat pituitary cells. In contrast, trypsin-treated filtrates were not different from untreated filtrates in their ability to elicit a secretory response in ligated rabbit intestinal loops. However, incubation of culture filtrate in ligated intestinal loops, or with rabbit intestinal fluid (in vitro), resulted in at least a twofold increase in permeability factor that did not occur in the presence of trypsin inhibitor or with heat-inactivated intestinal fluid. Moreover, trypsin inhibitor could reduce the secretory response to culture filtrate. These findings suggest that activation of heat-labile E. coli enterotoxin by host enzymes may playa role in the development of a full pathogenic effect.
842
point. Therefore, the data are expressed in terms of a 4-mm blueing dose (BD4mm) per ml, determined from the reciprocal of that dilution of sample which elicited a 4-mm blueing lesion 18 hr after intradermal inoculation of 0.1 ml. (2) Measurement of ability to stimulate growth hormone secretion by monolayer cultures of rat pituitary cells. In this system, both cholera toxin and enterotoxigenic E. coli culture filtrates dramatically alter the morphology of cultured rat pituitary cells and stimulate their secretion of growth hormone [19]. The ability of control and trypsin-treated culture filtrates to stimulate secretion of growth hormone by pituitary cell monolayers (prepared as described earlier [19]) was determined by double antibody radioimmunoassay of the amount of growth hormone secreted into the culture medium. Serial 10-fold dilutions of control and of trypsin-treated culture filtrates were made in Earle's balanced salt solution. Dilutions ranging from 1: 10 to 1: 1,000 were then each incubated with cell monolayers, in triplicate, for 3 hr at 37 C in a humidified atmosphere of 95% air and 5% CO 2 • Control cell monolayers were incubated with the same dilutions of enterotoxin-production medium (EPM), which, like the culture filtrates, had been treated with trypsin or buffer, and to which trypsin inhibitor had been added. The data are expressed in terms of the minimal effective dose per ml, i.e., the last 10-fold dilution of culture filtrate that stimulated significant (P < 0.05 or P < 0.01) secretion of growth hormone relative to controls. Results were evaluated by analysis of variance, with use of a least significant difference criterion for assessment of differences among means [20]. (3) Measurement of enterotoxicity. The adult rabbit intestinal loop assay [21] was employed for measurement of the secretory response to undiluted and low dilutions of control and trypsintreated culture filtrates, respectively. The assay was performed essentially as described by Pierce [22]; New Zealand albino rabbits, weighing about 2 kg and with ligated intestinal segments 5-8 em in length, were used. Purified cholera toxin (5 ug/rnl) served as the positive control, and 0.067 M phosphate buffer with 0.1 % gelatin (PBG), pH 7.4, the diluent for cholera toxin and E. coli culture filtrate, served as the negative control.
Downloaded from http://jid.oxfordjournals.org/ at Florida Atlantic University on September 9, 2015
the nontoxigenic strain [3] was designated H4084(078:H12). Strain H197 was employed in most studies. In addition, three strains of E. coli K12, two of which contained porcine enterotoxin plasmids and produced both heat-labile and heatstable enterotoxin, were provided by Dr. Stanley Falkow (University of Washington, Seattle, Washington). The strains, which have been described [15], were designated 711F-, 711 (PISS), and 711(P307). Medium and production of enterotoxin. The medium and culture conditions employed for the production of enterotoxin were the same as those described by Evans et al. [16], with the exception that shaker cultures were inoculated with 0.1 ml of a 4-hr, 2 % peptone culture of a particular strain and the cultures were harvested at 48 hr, unless specified otherwise. Culture supernatants were obtained by centrifugation of the cultures in a refrigerated PR-2 International centrifuge at 1,800 g for 1 hr; the supernatants were sterilized by passage through a filter (pore size, 0.45 urn; Millipore Corp., Bedford, Mass.). The cell-free culture filtrates were stored at 4 C for at least two months without noticeable loss in toxicity. Conditions for treatment with trypsin. Unless noted otherwise, 50 ug of trypsin (p-tosylphenylalanyl chloromethyl ketone-treated; Worthington Corp., Freehold, N.J.) in 50 ul of 0.05 M NH 4 HC0 3 , buffer (pH 8) was added per ml of culture filtrate, and the sample was incubated at 37 C for 1 hr. Controls, which were incubated under the same conditions, consisted of culture filtrate to which 50 fAl of 0.05 M NH 4HC03 buffer was added per ml. After incubation 25 ug of lima bean trypsin inhibitor (Worthington) in 25 ul of NH 4 HC03 buffer was added to each ml of trypsintreated and control culture filtrates, Biological assay of enterotoxin activity. (1) Measurement of vascular permeability factor activity. The permeability factor (PF) activity of control and trypsin-treated culture filtrates was assayed by intradermal inoculation of the shaved backs of New Zealand albino rabbits (weight, -2 kg). The conditions employed [17] were essentially the same as those described for cholera toxin by Craig [18] and for heat-labile E. coli enterotoxin by Evans et al. [16], with the exception that serial twofold or fourfold dilutions of sample were titrated to a 4-mm blueing lesion end
Rappaport et al.
Activation of E. coli Enterotoxin
(nil/em).
Neutralization of PF by antitoxin and by ganglioside G m . Twofold dilutions (in PBG, pH 7.4) of control and of trypsin-treated culture filtrates were incubated with equal volumes of appropriate twofold dilutions of either E. coli or cholera antitoxin for I hr at 37 C. The E. coli antitoxin was a goat antiserum to partially purified heatlabile enterotoxin from E. coli HI97; the antitoxin was prepared as described previously [23]. It was assigned a titer of 400 antitoxin units (AU) per ml by titration, in rabbit skin, against one limitof-blueing dose of cholera toxin [24]. The cholera antitoxin was the Swiss Serum Vaccine Institute provisional standard cholera (horse) antitoxin with an assigned titer of 4,470 AU/mi kindly provided by Dr. Carl Miller of the National Institutes of Health, Bethesda, Maryland). After incubation, O.I-ml aliquots of each sample were inoculated intradermally, in duplicate or in triplicate, into each of two rabbit skins. Blueing lesions were measured 18 hr later [17]. The results are expressed in AU/ml, determined from the serum dilution which, in the presence of an equal volume of E. coli culture filtrate, yielded a 4-mm blueing lesion in rabbit skin. In each experiment, one limit-of-blueing dose of cholera toxin was used as a standard. For measurement of PF neutralization by ganglioside G M 1 (galactosyl-N-acetylgalactosaminyl [sialosyl] lactosyl ceramide; GGnSLC) [25, 26J, a similar procedure was employed. Prior to inoculation of rabbit skin, undiluted or twofold dilutions (in PBG, pH 7.4) of control and trypsintreated culture filtrates were incubated for 30 min at 37 C with appropriate serial twofold dilutions of G~n, which contained 12 % silicic acid.
(This preparation was generously provided by Dr. W. E. van Heyningen, Oxford University, Oxford, England.) In this case, the results are expressed in terms of the concentration of ganglioside which, in the presence of an equal volume of E. coli culture filtrate, elicited a 4-mm blueing lesion in rabbit skin. The results of both antitoxin and ganglioside neutralization were adjusted for undiluted culture filtrates. In vivo incubation experiments. Paired I-ml samples of control and trypsin-treated culture filtrate were incubated side by side in a series of ligated intestinal loops for various periods, ranging from a few seconds to 2 hr. The samples were inoculated with a 26-gauge needle in the order of increasing time of incubation. This procedure was followed for convenience and to avoid confusing the samples. When sample inoculations were completed in each rabbit, the opened abdomen was covered with moist towels until the end of the experiment. After incubation for a given time, the fluids were withdrawn with a 5-ml syringe (I9-gauge needle), and the volumes were measured. The fluids were then clarified by centrifugation at 1,000 g for 10 min at 4 C in an International clinical centrifuge (model CL); NaN a (final concentration, 0.02%) was added to the supernatants. Two or three rabbits were employed for each in vivo incubation experiment, and the samples from each rabbit were assayed separately for PF. The results, adjusted for increases in volume, are expressed as percentage of uninoculated (control and trypsin-treated) PF, and they represent the mean of samples incubated in several rabbits. Preparation of rabbit intestinal fluid. To obtain samples of rabbit intestinal enzymes, I-ml aliquots of EPM were incubated in a series of ligated intestinal loops for 30 min in a manner similar to that employed in the in vivo incubation experiments. Samples were incubated both with and without lima bean trypsin inhibitor (0.5 mg/ ml) . After incubation, fluids were withdrawn and centrifuged as described above. After tests for the presence or absence of proteolytic activity, equivalent samples of intestinal fluid from different rabbits were pooled and stored at -20 C. Measurement of proteolytic activity. For measurement of proteolytic activity in rabbit intestinal fluid or in loop-incubated culture filtrate, a prot-
Downloaded from http://jid.oxfordjournals.org/ at Florida Atlantic University on September 9, 2015
All samples, including positive and negative controls, were injected in duplicate into each of two animals. Equivalent dilutions of control and trypsin-treated culture filtrates were inoculated side by side, but the position of various paired dilutions was random. Eighteen hours after incubation of the loops in the closed abdomen, animals were sacrificed by injection of sodium pentobarbital, and the small intestines were removed. The volume of fluid contained in each segment was determined, and the length of the empty segment was measured. The secretory response per loop was expressed as the volume-to-length ratio
843
Rappaport et al.
S44
without withdrawal for the full 18-hr period. At this time, animals were sacrified, and the volumeto-length ratio (ml/cm) of each loop was determined. The response elicited by filtrates incubated for periods of up to 30 min was computed as a percentage of the response elicited by the identical sample that had not been withdrawn, i.e., culture filtrate with or without inhibitor. The results from several animals were pooled. Results
Effect of trypsin on PF activity of cell-free filtrates from E. coli H 197 and other enterotoxigenic strains. Trypsin treatment of cell-free culture filtrate from E. coli H197 resulted in a four- to 20-fold increase in PF activity that became maximal when the concentration of trypsin reached 10 ug/ml and that did not occur in the presence of trypsin inhibitor or with boiled trypsin (table 1). Although the reaction conditions chosen for this study involved incubation of culture filtrates with 50 I-1g of trypsin/ml for 1 hr at 37 C, maximal activation of PF by trypsin occurred within 5 min (table 2). Both preexisting (control) PF and trypsin-activated PF were abolished after heating at 65 C for 30 min (or at 100 C for 15 min), and both were stable at a pH range of 6-10. Table 1. Effect of trypsin on vascular permeability factor (PF) activity present in a representative cellfree, 48-hr culture filtrate from Escherichia coli B197. Fold increase in PF
Trypsin (ug /rnl)
o 0.1 1.0 10.0 50.0 100.0 500.0 50 (boiled 10 min) 50 25 ILg LBTI
+
155 80 92 1,371 1,408 1,324 1,444 102 184
1