Escherichia coli enterotoxin. Purification and partial characterization.

Enterotoxin, a diarrheagenic protein elaborated by pathogenic Escherichia coli strains has been isolated from the supernatant of fermenter cultures of E. coli strain P263, a porcine enteropathogen. Purification steps involving Bio-Gel agarose A-5m, Sephadex G-75 chromatography, and preparative isotachophoresis were used in the isolation. The resulting product appears to be pure according to immunoelectrophoretic, disc electrophoretic, ultracentrifugal, and immunologic criteria. The entertoxin has an apparent molecular weight of 102,000 as judged by gel filtration and sodium dodecyl sulfate polyacrylamide gel electrophoresis, and its isoelectric point is 6.90. The isolated product is highly active in inducing experimental diarrhea in adult rabbits and piglets. It also elicits, in small dosage, a marked increase in adenylate cyclase activity in broken cell preparations of cat heart tissue. The enterotoxin activity is acid-labile and is destroyed by heating at 65 degrees for 30 min. It is suggested that the heat-stable enterotoxin material is derived from heat-labile enterotoxin by forming a complex with endotoxin or capsular material present in the culture supernatant.


Enterotoxigenic
Escherichia coli strains isolated from both humans and animals have been reported to produce an enterotoxin, of which two forms are generally recognized, one heat-labile and the other heat-stable. Both types of enterotoxin are controlled by extrachromosomal genetic factors that can be transferred by sexual conjugation to certain other bacteria, and it has been postulated that these apparently dissimilar enterotoxins are probably two different forms of what is essentially the same enterotoxin (l-3). The severe losses of water and electrolytes which occur in E. coli infection appear to be caused by this toxin whose action is mediated by stimulation of adenylate cyclase activity in the epithelial cells of the small intestine with a consequent increase in the intracellular concentration of cyclic adenosine 3':5'-monophosphate (4,5). In this respect the heat-labile E. coli enterotoxin shows some characteristics of Vibrio cholerae enterotoxin (6). However, the V. cholerae enterotoxin has been isolated in a pure state and has been subjected to extensive physical and chemical characterization, whereas the E. coli enterotoxin has resisted efforts aimed at purification and characterization (7,8). The methods that have been successfully used for the study of the pathophysiological effect of V. cholerae enterotoxin were, therefore, applied to elucidate the biological activities of the E. coli enterotoxin itself (9)(10)(11)(12)(13)(14)(15)(16)(17). It seems evident that the purification and characterization of E. coli enterotoxin is necessary to fully evaluate its role in disease and to study the antitoxic immunity, and this is expected to contribute to the development of rapid in vitro tests for the recognition of E. coli enteropathogens. This paper describes the purification and partial characterization of the heat-labile enterotoxin from a strain of E. coli reported to be associated with diarrhea1 disease in pigs MATERIALS AND METHODS Goats were injected in multiple subcutaneous sites near the axillary and pelvic lymph nodes with 10 ml of the emulsified antigen. After 6 weeks each goat received another 10 ml of the antigen and monthly booster doses thereafter.
All animals were bled 10 to 12 days after each booster. A pool derived from the same goats prior to immunization served as control serum.
Immunochemical Techniques-Laurel1 antigen-antibody crossed electrophoresis (31) was carried out by the method of Clarke and Freeman (32) as described in the LKB instrumentation manual. After electrophoresis the gel was pressed, washed with 0.1 M NaCl and water, pressed, and dried. The staining and destaining procedure was identical with that described above for polyacrylamide gels.

Sedimentation
Velocity Experiments-These were performed using schlieren optics. For the determination of sedimentation coefficients in buffer, single sector cells were used at rotor speeds of 56,000 rpm. The values of the sedimentation coefficients were corrected to the density and viscosity of water at 20" (34) The homogenate was filtered through nylon gauze and centrifuged at 1,500 x g for 10 min. The supernatant was discarded and the pellet washed with sucrose buffer, resuspended, and recentrifuged.
The second 1,500 x g pellet was taken up in about one-fourth of the original volume of sucrose buffer, and the total volume was measured.
An amount of 2.00 M sucrose buffer was added such that the final concentration of sucrose was 1.40 M. The mixture was homogenized once more with one or two strokes to make an evenly divided suspension which was put into centrifuge tubes. A small volume of 0.25 M sucrose buffer was layered over the top, and the tubes were spun in at 1500 x g for 15 min, and the supernatant material was added to 5 ml of ethylether saturated with water; the mixture was thoroughly shaken, and after separation of the layers by centrifugation the upper ether phase was removed and discarded. The ether extraction was repeated two additional times. The aqueous solutions were kept in a 60-70" water bath and concentrated to dryness in a stream of air.
By adding a known amount of tritiated CAMP tracer to one sample the appropriate correction for extraction recovery (usually 90 to 98%) was made for each set of experiments.
The residue was dissolved in 1.0 ml of 0.2 M acetate buffer, pH 4.0, and aliquots were used for the CAMP assay according to the procedure provided by the manufacturer.

Enterotoxin
Zsolation-All steps were performed at 4" unless otherwise noted.
Step 1: Bio-Gel A-5m Gel Filtration-A Step 2 The gel was prepared from stock solutions according to Davis (22), with the use of Tris-phosphate, pH 8.1 (leading electrolyte), as gel buffer, polymerized by photopolymerization only (Philips mercury B 0.9 AH4 lamp). The elution rate was 25 ml/hour, and the eluate from the polyacrylamide column was carried to a fraction collector via an LKB Uvicord II ultraviolet absorptiometer, which provides a primary record of the separation. The final evaluation of the fractions (7 ml) was carried out by the in uitro assay for exterotoxin with myocardial adenylate cyclase preparations (Fig. 3). The experiments were performed with a constant power supply. A current of 10 ma was applied in every run. The temperature of the cooling water was 10". The activity peak was pooled and concentrated by vacuum dialysis against Tris-t-aminocaproate buffer, pH 8.9.
Step 4: Preparative Zsotachophoresis II-The power of preparative isotachophoresis is considerably increased by performing two consecutive runs with either a different ratio of the molarities of leading ion and counterion or a different leading electrolyte (38). The contaminant-free products from five of the initial isotachophoresis runs were combined and subjected a second time by preparative isotachophoresis.
The sample, 6 to 7 ml, was mixed with 0.6 g of sucrose and was diluted to 20 ml with Tris-t-aminocaproate buffer, pH 8.9. The systems and conditions were the same as in the first preparative isotachophoresis except that the leading electrolyte was changed to Tris-2-(N-morpholino)ethanesulfonic acid, pH 6.2. The final material from the isotachophoresis column was concentrated by vacuum dialysis and subjected to gel filtration on Sephadex G-50 to separate the low molecular weight Ampholine carrier ampholytes from the isotachophoretically fractionated protein. The purification scheme is summarized in Table I. Estimation of Purity-The specific adenylate cyclasestimulating activity of the enterotoxin presented in Table I in broken cell preparations from myocardial tissue was about 10,000 times that of the crude supernatant, with an overall recovery of 58%. The purified enterotoxin resulting from this procedure appears homogenous by polyacrylamide gel electrophoresis in three different systems: by the standard procedure, FIG. 3 in sodium dodecyl sulfate gels, and in 8 M urea gels (Fig. 4).
Isoelectric focusing (pH 3 to 10) profiles of the enterotoxin also suggest a single component in each case (Fig. 5, A and B). The isoelectric point was determined to be 6.90 for the isoelectric focusing experiment in a sucrose density gradient and 6.85 for isoelectric focusing in thin layer polyacrylamide gel, respectively.
The ultracentrifugal pattern of the purified enterotoxin shows a single symmetrical peak. The sedimentation coefficient at a concentration of 4.5 mg/ml was measured, and a value of .s!&~ = 5.4 S was obtained.
The purity of the preparation was also tested by means of crossed immunoelectrophoresis (Fig. 6).
Stability of Enterotoxin-The Escherichia coli enterotoxin from strain P263 is labile at low pH, especially in the absence of glycerol. The stability of the pure enterotoxin at 37" gradually decreased when the pH was lowered from 7 to 4 at which point it was rapidly inactivated in the absence of glycerol. If glycerol was not present in the crude extracts, and if the pH was below 6, then over 60% of the enterotoxigenic activity was lost in crude extracts stored for more than 1 day at 4". We were unable to reverse this loss in activity.
The enterotoxin is labile at higher temperatures. The stability of the protein at pH 7.0 gradually decreased when the temperature was raised. The enterotoxigenic activity of the material was entirely destroyed by heating at 65" for 30 min. The toxin was heat-labile in that the activity was affected at 50" and almost entirely destroyed at 60". The addition of glycerol did not prevent the inactivation.
On the other hand, the pure enterotoxin retained 85 to 100% of its biological activity after 6 months of storage at -20" in 0.1 M Tris-HCl, pH 8.0, 0.5 mM EDTA, 20 mM 2-mercaptoethanol, and 10% glycerol.
Molecular Weight-The molecular weight of the native enterotoxin was determined by measuring the elution volume on a column of Sephadex G-200. The toxin was found to have a molecular weight of 102,000. Polyacrylamide gel electrophoresis in sodium dodecyl sulfate can be used as a criterion of homogeneity and can also provide information on subunit molecular weight. A value of 102,000 was found when this method was applied to either the native or the reduced and carboxymethylated protein. This value is in agreement with the molecular weight determined by gel filtrations. These data suggest that the enterotoxin is composed of one polypeptide chain.
Biological Actiuities-The purified E. coli enterotoxin preparation was employed to investigate the response in the ligated intestinal loop test and in the adenylate cyclase assay as a function of concentration. As noted in Fig. 7 both assay systems produced a near linear dose response curve in the range employed in the present study with 0.65 pug and 4.75 fig for the 50% dose or 1 unit of toxin in the adenylate cyclase assay system and the ligated intestinal loop assay, respectively. In order to compare the activity of Vibrio cholerae enterotoxin in both test systems titration of known amounts of choleragen was performed in the ligated loop test and in the adenylate assay. Essentially the same response curve in both systems was obtained (results not shown). The 50% dose of V. cholerae enterotoxin in these assays was found to be 0.32 bg in the adenylate cyclase assay system and 1.24 fig in the ligated intestinal loop assay. Consequently, the respective toxins from E. coli and V. cholerae seem to elicit a very similar quantitative response in these test systems,

Immunological
Titration of Adenylate Cyclase-stimulating Activity of E. coli Enterotoxin--For the assay of neutralizing antibody more than 1 unit of toxin is required to allow titration of residual activity. The amounts of toxin used here have been 2 units/ml of the loop assay and 3 units for the adenylate cyclase assay, respectively. The toxin solution was mixed with equal volumes of serial dilutions of inactivated serum, incubated for 1 hour at 37" with constant agitation, and titrated for residual activity. A neutralization coefficient (NC) was calculated for each dilution by dividing the mean volume length ratio x by 2.66, the calculated upper asymptote of the dose response curve (Fig. 7), and subtracting this value from 1:  Assuming that this kind of calculation is applicable to the adenylate cyclase test system, the corresponding values are 9, the mean picomoles of CAMP accumulated under the conditions of the test system, and 950 pmol of CAMP, the upper asymptote of the dose-response curve in this system. NC has a range of 0 to 1, with 0 representing no neutralization and 1 complete neutralization as limits. When log NC was plotted against log milliliters of antiserum, the points were best fitted by a slightly curved line, but the departure from linearity appeared to be sufficiently small over the range NC = 0.4 to 0.8 that half-neutralization, NC = 0.5, could be reasonably well interpolated from a straight line.
In the titration illustrated in Fig. 8, two units of toxin were used in the ileal loop assay, and NC = 0.5 indicates the neutralization of 1 unit of toxin. From the interpolation of the 50% point a value of 2173 units per ml as the antitoxin titer has been determined.
For the adenylate cyclase system 3 units of toxin were used; NC = 0.5 indicates in this experiment the neutralization of 2 units, and 3076 units per ml have been found as antitoxin titer.
E. coli Enterotoxin-Endotoxin Interaction-Approximately 65% of the activity of the concentrated sterile broth filtrate remained after heating at 65" for 30 min (Table I). This finding is at variance with results reported in the literature.
The heat-stable form of E. coli enterotoxin was reported to be dialyzable and to pass through a Diaflo PM-30 membrane (33). It was, therefore, anticipated that the enterotoxin heat-stable activity would readily pass through the Diaflo membrane during our initial concentration procedure. However, when the concentrated broth filtrate was subjected to Bio-Gel agarose 5m chromatography heat-stable enterotoxic activity could be found in the first peak material (Fig. 9).
This finding indicates that heat-stable enterotoxic activity is contained in high molecular weight material which is clearly separated from the heat-labile material. Chemical analyses of this material indicated the presence of carbohydrate.
A positive Shwartzman reaction (39) and a positive E-toxate test (limulus amebocyte lysate) indicated the existence of endotoxin. The percentage of the analyzed carbohydrate in this peak was around 40%. The approximate molecular weight exclusion limit for polysaccharides on agarose A-5m is 5 x 10'. Since the material was largely carbohydrate and appeared in the void volume of the column it can be concluded that this material probably has a molecular weight of 1 to 5 x 10'.
To separate the enterotoxic component from the endotoxin, the first peak material (300 mg dry weight) from the agarose A-5m column was treated with 1% Na dodecyl-SO, in 0.1 M (NHJHCO,, pH 8.0. Fig. 10 shows the elution pattern of this material from an agarose A-5m column (2.5 x 85 cm) equilibrated with 0.1% Na dodecyl-SO, in 0.1 M (NHJHCO,, pH 8.0. The appearance of four peaks was evidence that this reagent caused dissociation of the material contained in the first peak. (range 3 to 10) were set in the gel. Gels were removed and fixed in 5% trichloroacetic acid after focusing for 150 min at 50 ma (decreasing to 28 ma) at a voltage of 210 volts (increasing to 1080 volts). After the run, the pH gradient established was determined with a combined microsurface electrode.
The materials of all four peaks were incubated separately in concentrated urea and were then freed of the detergent by an anion exchange resin. The resulting dodecyl sulfate-free protein fractions were renatured from urea solution by standard procedures (40).
Only the material of the second peak from the agarose-Na dodecyl-SO, column showed enterotoxic activity. It was devoid of endotoxin and carbohydrate under the conditions of the test systems applied (20,21). In contrast to the starting material, it now proved to be heat-sensitive; the activity was completely destroyed by heating at 65' for 30 min. The first peak material did not reveal any enterotoxic activity, and the analyzed carbohydrate and protein were comparable to the amount of these substances in endotoxin. The third and fourth peaks were devoid of any activity and free of endotoxin. The renatured protein fractions with heat-labile enterotoxic activity were subjected to the above mentioned purification procedure for the heat-labile enterotoxin. A protein was isolated with virtu- ally identical physicochemical and biological characteristics as determined for the heat-labile enterotoxin. It is, therefore, tempting to conclude that the heat-stable enterotoxin is derived from the heat-labile enterotoxin by forming a complex with endotoxin present in the culture medium. In order to substantiate this assumption an excess of endotoxic material was incubated with the heat-labile enterotoxin and then chromatographed on an agarose A-5m column. The material eluted in one single peak in the void volume of the column and again showed heat-stable activity.

DISCUSSION
It is now well established that many strains of E. coli elaborate one or more toxic factors which fulfill the definition of an enterotoxin; that is, a factor which causes a net movement of fluid and electrolyte from 'plasma to gut lumen when given by the intraluminal route. The data which have been gathered thus far, however, present a confusing panorama of a number of enterotoxic factors produced by enteropathogenie E. coli strains of different serological types and different host adaptations, possessing widely differing chemical, physical, and biological properties. After surveying this busy scene one longs for the simplicity of a single antigenic type of cholera toxin elaborated by only two antigenic types of cholera vibrio.
Among the E. coli enterotoxins reported so far, some are resistant to heating, causing one to doubt whether they are even protein in nature (41). Furthermore, a given strain usually produces both heat-labile and heat-stable enterotoxin under the same cultural conditions (l-3, 33). Chromatographic separation has been difficult to impossible, with enterotoxic activity associated with a wide spectrum of molecular sizes (8,42). The underlying problem depends on the uncertainty of the bioassay of toxic activity. Without a reproducible assay system which can localize and quantitate enterotoxic activity, none of the physical or chemical properties can be ascertained, and quantitative studies are impossible. Therefore, the first aim was to develop a test which is easy to perform, highly reproducible, superior in sensitivity to the current methods, and economical. In all the work discussed the stimulation of enterotoxin-sensitive adenylate cyclase in broken cell preparations from cat myocardial tissue was employed as an accurate measure of enterotoxin concentration (35). As a first step in the efforts to purify the heat-labile enterotoxin the initial period of culture was reduced from 18 to 6 hours. Thus. the cells were harvested before entering the stationary phase, thereby reducing the content of somatic antigen as well as other large molecular species in the crude filtrate and minimizing the protein degradation in E. coli cells. Although protein breakdown occurs relatively slowly in exponentially growing cells, the average rate of degradation increases 4-to 6-fold when cells enter the stationary phase. To protect the bacterial products from proteolytic degradation during ultrafiltration a protease inhibitor was added. Of the various compounds tested, pentamidin isethionate appeared to be the most effective and most selective inhibitor of protein breakdown during the concentration procedure (18). The purification of the heat-labile E. coli enterotoxin results in a protein that appears to be homogenous as indicated by immunoelectrophoretic, disc electrophoretic, ultracentrifugal, and immunologic criteria. The protein has been purified lO,OOO-fold with a yield of 58% and is devoid of carbohydrate (20). Chromatographic and electrophoretic studies suggest that the enterotoxin has an apparent molecular weight of 102,000 and consists of one single polypeptide chain. The isoelectric point has been determined to be 6.90. These findings are at variance with the results obtained by other investigators. Evans et al. (44) have found a molecular weight of E. coli heat-labile enterotoxin in the range of 20,000. However, this material was not obtained from the supernatant of an enterotoxin-producing strain, but by the induced release from intact E. coli cells by polymyxin B. Although the exact mechanism of action of polymyxin for liberating cell-bound enterotoxin is not clear, it is at least conceivable that this procedure could lead to an increased proteolytic activity and consequently to a degraded form of the enterotoxin.
On the other hand, Jacks et al. (8) recently reported that the enterotoxic component from two porcine E. coli strains was eluted from a Sephadex G-200 column in the void volume. The molecular weight assigned to this material was greater than 5 x IO'. However, their preparation contained 45.8% carbohydrate and 9.3% protein and showed endotoxic activity by a positive Shwartzman reaction. They could not separate the enterotoxic component from the endotoxin; in fact, their data indicated that the two components are closely associated and that the enterotoxic activity resides in material of a protein nature. These authors and others (42) did not find any indication of enterotoxic activity within the fractionation range of the gel filtration columns in use. The data presented in this paper clearly show the dissociation of enterotoxic material from endotoxin or capsular material containing a polysaccharide structure, and the subsequent isolation of heat-labile enterotoxin from this material to the pure state. In addition, the incubation of an excess of endotoxin with heat-labile enterotoxin results in a complex which is heat-stable and has a high molecular weight. From these results it is tempting to conclude that the heat-stable enterotoxin is a complex derived from the heat-labile enterotoxin and endotoxin or capsular material present in the culture supernatant. This finding would also explain the observation of a high molecular weight species of enterotoxin as reported by Jacks et al. (8). This conclusion gains support from genetic evidence by Gyles et al. (43). They could show that Ent+ strains producing both the heat-labile and heat-stable enterotoxin carry only one well characterizable type of plasmid. Their studies seem to indicate that a single plasmid can code for both heat-stable and heat-labile toxin and that the plasmid, therefore, is carrying either two distinct toxin genes or that the two toxin species are variants of the same chemical entity. It is interesting to note that they also described a second class of plasmid, which seems to be present in strains producing only heat-stable, but not heat-labile enterotoxin.
Whether this plasmid is related to the small molecular weight heat-stable enterotoxin discussed by Gyles (33) is unclear. However, it seems not impossible that there exist two genetically and chemically unrelated types of heat-stable enterotoxin. Further studies are necessary to clarify this problem.
The isolated E. coli enterotoxin is highly active in inducing experimental diarrhea in adult rabbits and piglets. It also elicits, in minute amounts, a dramatic increase in adenylate cyclase activity in broken cell preparations of cat heart tissue. The E. coli enterotoxin is antigenic. A goat antiserum directed against purified enterotoxin neutralized all adenylate cyclasestimulating activity as measured in the ileal loop assay and in the adenylate cyclase test system. In this respect the heatlabile E. coli enterotoxin shows a resemblance with V. cholerae enterotoxin despite their marked structural differences (7). The availability of purified material should facilitate more definitive studies on the chemical composition and nature of E. coli enterotoxin. Its mode of action and its potential role in the production of immunity can now be further investigated.