Vol. 25, No. 2
INFECTION AND IMMUNITY, Aug. 1979, p. 586-596 0019-9567/79/08-0586/11$02.00/0
Purification and Chemical Charcterization of the Heat-Labile Enterotoxin Produced by Enterotoxigenic Escherichia coli STEVEN L. KUNKEL AND DONALD C. ROBERTSON* Department of Microbiology, University of Kansas, Lawrence, Kansas 66045 Received for publication 21 May 1979
Heat-labile enterotoxin (LT) produced by a human strain of enterotoxigenic Escherichia coli (286C2) was purified to homogeneity from pH extracts of fermentor-grown cells by ultrafiltration, (NH4)2SO4 fractionation, hydrophobic chromatography on norleucine-Sepharose 4B, hydroxylapatite chromatography, and Bio-Gel P-150 filtration. Purified LT preparations exhibited biological activity comparable to that of cholera toxin in four bioassays specific for the two enterotoxins (Y-1 adrenal tumor cells, Chinese hamster ovary cells, pigeon erythrocyte lysates, and skin permeability test). The overall yield of LT protein was 20%, which represented a 500-fold purification over pH extracts. A native molecular weight of 73,000 was determined by gel electrophoresis. The toxin dissociated upon treatment with sodium dodecyl sulfate, pH 7.0, into two components with molecular weights of 44,000 and 30,000. Purified LT preparations were remarkably stable over a wide range of storage conditions, temperatures, and pH's. The biological activity was increased by incubation with trypsin and completely destroyed by pronase and proteinase K, whereas deoxyribonuclease I, ribonuclease, and phospholipase D had no effect. The amino acid composition of purified LT was quite different from that of cholera toxin. Neither carbohydrate nor lipopolysaccharide was present in purified preparations. The purification scheme appeared applicable to LT produced by other human and porcine enterotoxigenic strains, but reflected the amount of LT produced by each strain. These data show that LT and cholera toxin share many common chemical and physical properties, but must be purified by different techniques.
Enterotoxigenic (ENT') strains of Escherichia coli have been implicated as the etiological agent of diarrheal disease in children (45, 47) and adults (48), common traveler's diarrhea (41, 46), and colibacillosis of neonatal animals (25, 30, 51). Clinical isolates of ENT' E. coli from both humans and neonatal animals produce one or two kinds of enterotoxins: either a low-molecular-weight nonantigenic heat-stable molecule or a high-molecular-weight antigenic protein (LT) or both. The mechanism of action of LT is similar to that of cholera toxin, since both enterotoxins activate membrane-bound adenylate cyclase with a subsequent increase in intracellular cyclic adenosine 3',5'-monophosphate in epithelial cells of the small intestine (17, 47). The heat-stable enterotoxin does not stimulate adenylate cyclase (47) but, instead, appears to increase levels of cyclic guanosine 3',5'-monophosphate in intestinal tissue through activation of guanylate cyclase (16). The heatstable enterotoxin produced by a porcine strain of ENT' E. coli has been purified and chemically characterized (1). There were no unique
chemical properties, except the presence of six half-cystine residues which may be linked through disulfide bonds, to explain the heat stability. Although several reports have appeared on the purification of LT (8, 10, 15, 18, 31, 35, 40, 49, 53, 56), there is no general agreement as to the molecular weight and chemical nature of toxin as it is released from whole cells. Molecular weights ranging from 23 x 103 to over 106 have been determined, but, more importantly, the preparations noted above were 103- to 106-fold less active than cholera toxin in bioassays specific for the two enterotoxins. The purification of LT produced by a human strain of ENT' E. coli (286C2) with biological activity almost identical to that of cholera toxin is described in this report. Techniques have been developed to increase yields of cell-associated LT (22, 32) and to maintain the protein in its holotoxin form, where biological activity in whole-cell assays would be facilitated by a binding component. The purification scheme was also applied to the isolation of LT from porcine 586
VOL. 25, 1979
PURIFICATION AND PROPERTIES OF LT
strains, but the yields were about 10-fold lower.
It appears that E. coli LT and cholera toxin share many chemical and physical properties. MATERIALS AND METHODS Bacterial strains. The ENT' E. coli strains were kindly supplied by Harley Moon, National Animal Disease Center, Ames, Iowa, and R. Bradley Sack, Johns Hopkins University School of Medicine, Baltimore, Md.; although several strains were screened in early stages of these experiments, most of the purification work was done with strain 286C2, a human strain, and strain 263, a porcine strain. Preparation of media. The M-9 minimal salts medium containing 10 mM N-tris(hydroxymethyl)methyl glycine (Tricine), 0.5% glucose, and three amino acids (methionine, lysine, and either aspartic acid or glutamic acid) was prepared as described previously (22). The medium for growth in the fermentor was prepared by autoclaving 27 liters of distilled water plus MgCl2 and FeCl2 in the fermentor vessel. The amino acids were dissolved in 3 liters of distilled water, adjusted to pH 7.5 with 5 N NaOH, and sterilized separately by autoclaving. One liter of 20% glucose and 10-fold-concentrated basal salts, pH 7.5, were each autoclaved separately. Before inoculation, the amino acids, basal salts, and glucose were added to the fermentor containing the sterile water and trace salts. Growth conditions. Starter cultures were grown aerobically with shaking (300 rpm) in 250-ml Erlenmeyer flasks containing 50 ml of Trypticase soy broth or defined medium. After 8 h of incubation at 370C, the starter culture was used to inoculate four Fernbach flasks, each containing 1 liter of defined medium. The volume of the starter culture was sufficient to adjust the initial absorbance at 620 nm to 0.05. After 8 h of shaking (300 rpm) at 37°C, the fourth Fernbach flasks were used to inoculate 36 liters of defined medium in an F-50 New Brunswick fermentor. The initial absorbance at 620 nm varied from 0.2 to 0.3. Growth in the fermentor vessel was for 5 h at 37°C with vigorous aeration (2.0 CFM at STP) and stirring (200 rpm). The initial pH of the medium was 7.5, which decreased after 5 h of growth to about 6.8. Cells were collected with a Du Pont Sorvall RC-5 refrigerated centrifuge and an SZ-14 continuous-flow rotor at 18,000 rpm. About 250 g of cells was collected per 40 liters of defined growth medium. Toxin assays. Several biological assay systems were used to examine preparations for LT activity. The Y-1 adrenal tumor cell assay of Donta et al. (9) was used to scan column fractions and to determine the biological activity of purified LT preparations. The adrenal cells were maintained on F-10 medium (GIBCO Laboratories, Grand Island, N.Y.) supplemented with 15% horse serum, 2.5% fetal calf serum, and 50,ug of gentamicin per ml. Assays for enterotoxin activity were performed in 24-well cluster dishes (Costar, Cambridge, Mass.). Each well contained 1 ml of medium and was seeded with 105 cells. After 2 days of growth at 37°C in a humidified atmosphere of 95% air and 5% C02, 0.05 ml of sample was added to each well. Toxin activity was determined by estimating the
587
amount of rounding at 18 h and in some experiments by extracting steroids. Under these assay conditions, 1 ng of either purified LT or cholera toxin induced greater than 90% rounding, and 5 to 10 pg induced 20 to 30% rounding. The pigeon erythrocyte lysate (PEL) assay of Gill and King (21) was also used extensively to complement the tissue culture assay just described. Samples for the PEL assay were preincubated in 0.15% sodium dodecyl sulfate (SDS)-0.02 M sodium phosphate, pH 7.0, for 15 min before the assay (22). The Chinese hamster ovary (CHO) cell assay was performed as described by Guerrant et al. (24), and the skin permeability test was performed by the method of Craig (5). Purification of LT. Cells collected from 40 liters of defined growth medium were suspended in 2 liters of cold 0.12 M tris(hydroxymethyl)aminomethane (Tris)-chloride, pH 6.5, and stirred for 45 min at room temperature. The cells were collected and suspended in 1.6 liters of 0.12 M Tris-chloride, pH 8.5. The pHadjusted cells were incubated at 370C in Fernbach flasks with gentle shaking until the temperature of the culture reached 34WC (about 1 h). The pH was then adjusted again to 8.5 and incubated for an additional 30 min. The cell suspension was centrifuged at 10,000 x g, and the supernatant was concentrated sixfold by ultrafiltration through PM-10 Diaflo membranes (Amicon Corp., Lexington, Mass.). (i) (NH4)2SO4 fractionation. Solid (NH4)2SO4 was added with stirring at 4VC to the ultrafiltration retentate to 90% saturation. After centrifugation, the pellet was extracted with 30% (NH4)2SO4 to yield a protein concentration of about 15 mg/ml. Any particulate matter was removed by centrifugation, and the supernatant was applied to a column of norleucine covalently coupled to Sepharose 4B (hydrophobic chromatography). (ii) Hydrophobic chromatography. A modification of the procedure of Cuatrecasas (7) was used to conjugate norleucine to the activated resin. Largescale preparation of norleucine-Sepharose was based on the following pilot experiment. Five grams of cyanogen bromide in 20 ml of distilled water was added to 20 ml of packed Sepharose 4B. The pH of the suspension was maintained at 11 by the slow addition of 15 to 16 ml of 5 N NaOH. Chipped ice was added to maintain the temperature below 30°C. After 9 to 11 min, the resin suspension was poured through a sintered-glass filter and washed with 1 liter of ice-cold 0.1 M NaHCO3, pH 9.0. The activated Sepharose was added to a solution containing 98 mg (750 mmol) of norleucine dissolved in 20 ml of 0.1 M NaHCO3, pH 9.0. The suspension was gently stirred for 16 h and washed extensively with distilled water. A column (2.5 by 18 cm) of norleucine-Sepharose was equilibrated with 1.25 M (NH4)2SO4-20 mM Tricine, pH 8.0 (run buffer). The supernatant from the 30% (NH4)2SO4 back extraction was loaded onto the column and washed with run buffer until the optical density returned to base line. A linear decreasing salt gradient (800 ml) from 1.25 M (NH4)2SO4-20 mM Tricine, pH 8.0, to 20 mM Tricine, pH 8.0, containing 50% ethylene glycol was used for elution. Fractions (20-ml) were collected at a flow rate of 60 ml/h. Fractions containing LT activity were pooled, dialyzed to remove eth-
588
KUNKEL AND ROBERTSON
ylene glycol, concentrated by ultrafiltration with a PM-10 Diaflo membrane, and dialyzed against 100 volumes of 10 mM N-2-hydroxyethylpiperazine-N'-2ethanesulfonic acid (HEPES), pH 7.5. (iii) Hydroxylapatite chromatography. A column (1.5 by 20 cm) of hydroxylapatite was equilibrated with 10 mM HEPES, pH 7.5, as the run buffer. The partially purified LT recovered from the norleucine column was loaded onto the hydroxylapatite column and eluted with an increasing potassium phosphate gradient (400 ml) from 10 mM HEPES to 10 mM HEPES-400 mM potasium phosphate, pH 7.5. Fractions (10-ml) were collected at a flow rate of 30 nil/h. (iv) Gel filtration. A Bio-Gel P-150 column (1.5 by 85 cm) was equilibrated with 10 mM Tricine, pH 8.0, plus 50 mM NaCl as the run buffer. Fractions (2-ml) were collected at a flow rate of 6 ml/h. Determination of LT recovery by single radial immunodiffusion. The single radial immunodiffusion technique developed by Mancini et al. (37) was used to determine LT recovery at each purification step. The antiserum against purified LT was kindly prepared by P. H. Gilligan of this laboratory as follows. New Zealand white rabbits, 3 to 5 months of age, were injected intradermally at multiple sites on a shaved portion of the back and in one hind footpad with 100 Ag of purified LT in Freund complete adjuvant. The animals were bled from the central ear artery at weekly intervals 3 weeks after immunization. The titer of humoral antibody against 286C2 LT remained high for as long as 15 weeks. The antiserum raised to purified 286C2 LT exhibited a single precipitin band when reacted with crude pH extracts and partially purified preparations. Molecular weight determination. The molecular weight of native LT was determined by comparing the slopes obtained in Ferguson plots with those obtained using proteins of known molecular weights (26). The subunit molecular weights were determined using slab gels prepared as described by Ames (2) with the gel composition of Lugtenberg et al. (36) and in glass tubes (0.6 by 10 cm) by the procedure of Weber and Osborn (58), except that the phosphate concentration was reduced to 0.1 M. Sedimentation coefficient. Analytical ultracentrifugation studies were performed with a Spinco model E analytical ultracentrifuge equipped with phase plate schlieren optics. The sedimentation coefficient for purified LT was determined by the procedure of Chervenka (4). Sulfhydryl group determination. Determination of protein sulfhydryl groups was performed with Ellman reagent (13) by the procedure of Mohler et al.
INFECT. IMMUN.
carboxymethylcysteine (6). Tryptophan was determined by the method of Edelhoch (11). Amino-terminal residue. The procedure used for the identification of the amino-terminal residue(s) was a minor modification of that described by Gray (23). One milligram of dialyzed LT was lyophilized and suspended in 0.5 ml of 0.5 M NaHCO3, and 0.5 ml of dansyl chloride (2.5 mg/ml) was added, followed by incubation at 37°C for 4 h. The hydrolysate was extracted twice with ethyl acetate and suspended in acetone-glacial acetic acid (3:2, vol/vol) for application to polyamide sheets as described by Woods and Wang (59). Solvents used for thin-layer chromatography were water-90% formic acid (200:3, vol/vol) in the first dimension and n-heptane-n-butanol-glacial acetic acid (3:3:1, vol/vol/vol) in the second dimension. Stability studies. (i) Enzyme treatment of LT. Several hydrolytic enzymes (pronase, trypsin, proteinase K, deoxyribonuclease, ribonuclease, and phospholipase D) were incubated with purified LT at LT/ enzyme ratios (wt/wt) of 1:50, 1:10, 1:1, and 10:1. The procedure of Efling et al. (12) was used for protein degradation by proteinase K. Protease activity was terminated after 60 min by adding either phenylmethylsulfonyl fluoride or soybean trypsin inhibitor. (ii) pH treatment of LT. Hydrochloric acid (1 N) or sodium hydroxide (1 N) was added with a microsyringe to 1-mi samples of LT (22 pg/ml). The pH was determined by spotting on litmus paper until the following values were reached: 1, 3, 5, 7.2, 10, and 12. The samples were then incubated at 37°C for 60 min and adjusted to pH 7.2 before assay. (iii) Temperature treatment of LT. Purified LT was stored at a concentration of 2.2 mg/ml at -70, -20, and 4°C for 2 months and assayed for biological activity. Samples (1-mi) of purified LT (22 ,ug/mi) were heated at 50, 60, 70, 80, and 90°C for 30 min. The samples were cooled to room temperature and assayed for LT activity. Ganglioside treatment of LT. The procedure of Van Heyningen et al. (55) was used to examine the effects of mixed gangliosides (Sigma type III) on the biological activity of purified LT. Mixed gangliosides were prepared in 0.1 M phosphate buffer, pH 7.2, containing 0.2% gelatin and added to LT at LT/ganglioside (wt/wt) ratios of 1:100, 1:50, 1:25, 1:10, 1:1, and 100:1. The solutions were then incubated at 370C for 30 min and assayed for LT activity. Protein and carbohydrate determination. Protein was determined by the absorbance at 280 and 260 nm (57). Total carbohydrate was determined by the phenol-sulfuric technique, with glucose as the standard, and qualitative detection of lipopolysaccharide was by the colorimetric assay for 2-keto-3-deoxyocton(39). Amino acid analysis. Before hydrolysis, samples ate (29). Lipopolysaccharide of Salmonella typhimuwere extensively dialyzed against distilled water, trans- rium was a positive control. ferred to hydrolysis ampoules, and lyophilized. The Reagents used. All medium components and reampoules were degassed and hydrolyzed at 110°C for agents employed throughout this study were pur24 to 72 h. After hydrolysis, excess HC1 was removed, chased from Sigma Chemical Co., St. Louis, Mo., and the residue was dissolved in citrate buffer, pH 2.2. unless otherwise indicated. The electrophoresis reThe samples were applied to a Beckman 120C amino agents and Bio-Gel P-150 (100- to 200-mesh) were acid analyzer. The observed threonine and serine val- purchased from Bio-Rad Laboratories, Richmond, ues were divided by 0.95 and 0.90, respectively, to Calif. Thin-layer chromatography plates were purcorrect for destruction during hydrolysis (52). Half- chased from Brinkmann Instruments Inc., Westbury, cystine residues were determined as cystine or as S- N.Y. Purified cholera toxin was kindly provided by J.
PURIFICATION AND PROPERTIES OF LT
VOL. 25, 1979 Peterson, University of Texas, Galveston.
RESULTS Purification of LT. The procedures developed for the purification of E. coli LT are summarized in Fig. 1. Yields of cell-associated LT were increased by growing cells in a defined medium (22) and extracting fermentor-grown cells under conditions which rapidly released a significant amount of LT (32). The pH extract from 80 liters of cells was concentrated approximately sixfold with an Amicon TC5E ultrafiltration unit with PM-10 Diaflo membranes and fractionated by the addition of (NH4)2SO4 to 90% saturation. After stirring for 1 to 2 h at 40C, the precipitate was removed by centrifugation at 10,000 X g and extracted with 30% (NH4)2SO4. The 30% (NH4)2SO4 extract was applied to a column of norleucine covalently coupled to Sepharose 4B (hydrophobic chromatography). The LT activity, quantitatively retained by the norleucine column, was eluted free from contaminating lipopolysaccharide, shown by the absence of 2-keto-3-deoxyoctulosonic acid (KDO) with a linear decreasing (NH4)2SO4 gradient and a linear increasing gradient of 50% ethylene glycol (Fig. 2). It should be noted that LT was one of the most hydrophobic proteins in the pH extract and that a 15- to 20-fold purification was routinely observed. The LT activity in pH extracts of a porcine strain, 263, eluted at a slightly higher concentration of (NH4)2SO4, but LT activity in preparations from most strains eluted between concentrations of 0.5 and 0.25 M (NH4)2S04 (Fig. 2). Addition of ethylene glycol was necessary to prevent spreading of toxin activity. The LT activity produced by some strains of ENT' E. coli, which precipitated below 40% (NH4)2SO4, did not bind to the hydrophobic
589
column. Since these preparations were high in KDO color, it seemed likely that LT was tightly completed with outer membrane components. The amount of soluble LT protein back extracted by 30% (NH4)2SO4 from either a 0 to 90% (NH4)2SO4 cut or a 40 to 90% (NH4)2SO4 cut varied widely from strain to strain and determined whether sufficient amounts could be purified for chemical characterization. The final yield of LT was not dependent on the amount of protein applied to the hydrophobic column since up to 5 was loaded with little effect on overall recovery. The hydrophobic column fractions with LT activity were pooled, dialyzed, and concentrated. Before hydroxylapatite chromatography, the partially purified preparation was dialyzed against 50 to 100 volumes of 10 mM HEPES, pH 7.5. The LT activity eluted from hydroxylapatite as a symmetrical peak at 150 mM phosphate, PH EXTRACT (CELL-ASSOCIATED LT OF CELLS)
FROM
80 LITERS
ULTRAFILTRATION USING PM-10 DIAFLO MEMBRANES
(NH4)2S04 (90Z SATURATION) HYDROPHOBIC CHROMATOGRAPHY SEPHAROSE
ON
NORLEUCINE-
HYDROXYLAPAT I TE CHROMATOGRAPHY
4
GEL FILTRATION ON BIO-GEL P-150 FIG. 1. Purification scheme for E. coli LT.
E. D
z
Z z ) zt
;
FRACTION
FRACTION
FIG. 2. Fractionation of 286C2 LT by hydrophobic chromatography, using norleucine-Sepharose 4B. The solid line reflects absorbance at 280 nm, and the broken line represents Y- 1 adrenal cell activity detected at 4 h with 25 I&d of a 1:1K00 dilution of each fraction. The slashes indicate those fractions with maximum rounding activity but not further diluted. The gradient was started at the point indicated by the arrow.
590
KUNKEL AND ROBERTSON
INFECT. IMMUN.
pH 7.5 (Fig. 3), using a linear 0 to 400 mM 90 phosphate gradient. Fractions containing LT acso tivity were pooled, concentrated to about 2 ml 70 with an Amicon stirred cell, and applied to a 60 z Bio-Gel P-150 column (1.5 by 85 cm). Although some rounding activity appeared in the void 10 IS 2 5 3 S 4 20 volume of the column, the biological activity was -j FRCIO so " associated primarily with the second major protein peak (Fig. 4). Peak fractions were pooled, with about 10% of the LT activity on either side of the peak not included. Finally, the pooled gel filtration fractions were concentrated and stored FIG. 4. Gel filtration of 286C2 LT on Bio-Gel Pat 2 to 4 mg/ml. 150. The lines were as described in the legend to Fig. The complexity and relative purity of LT at 2. each stage of the purification scheme are shown in Fig. 5. Note that LT was the major protein eluted from the norleucine-Sepharose column and the predominant band eluted in the second peak from hydroxylapatite. A final gel filtration step was necessary to yield a preparation estimated to be 95 to 98% pure. The gels were overloaded with 150 ,ug of protein to detect traces of impurities. The recovery of LT from strain 286C2 at each stage of the purification is shown in Table 1. The minimal LT activity that did not adsorb to the hydrophobic column is probably associated with lipopolysaccharide since most of the KDO color eluted in the first peak. Since the predominant immune response in rabbits is directed against the binding component of the toxin (P. H. Gilligan and D. C. Robertson, unpublished observations), the radial immunodiffusion assay detected both the B component and holotoxin in pH extracts. Therefore, the yield of biologically active holotoxin is likely higher than that shown in Table 1. Molecular weight. The native molecular weight of LT was determined by the method of Hedrick and Smith (26). The mobility of LT and several protein standards in 6, 8, 10, and 12% 0
us
CD
04
.. .. *.
,00
190 70 a
60 Z 0
50C 30
£
FRACTION
FIG. 3. Fractionation of 286C2 LT by hydroxylapatite chromatography. The lines were as described in the legend to Fig. 2.
FIG. 5. Polyacrylamide gels of LT fractions: (A) (NH4)2S04 precipitate ofpH extract; (B) pooled fractions after hydrophobic chromatography; (C) pooled fractions eluted from hydroxylapatite; (D) final preparation after Bio- Gel P-iSO0 gel filtration.
acrylamide gels was determined, and a slope was calculated by plotting the log mobility versus the percent acrylamide concentration (Fig. 6). The molecular weight of 73,000 was derived from the relationship between the slope and molecular weight of protein standards. Sedimentation velocity experiments run at a protein concentra-
VOL. 25, 1979
PURIFICATION AND PROPERTIES OF LT
TABLE 1. Summary of LTpurification from EN7' E. coli strain 286C2 Total LT pro-
Step
protein (mg) 7,800
teina
(mg) 81
Recov- Fold puery ( rifiCation 100
pH-released LT (ultrafiltration with PM10 Diaflo membranes) 90% (NH4)2SO4 frac61 75 3,168 tionation 196 31 37 Hydrophobic chromatography 30 23 28 Hydroxylapatite chromatography 17 17 21 Bio-Gel P-150 filtration a Determined as LT antigen by Mancini assays.
2.5 40 260
459
- 2.6
"L 24
z 0 22
/lli-e Ph..plint ...
20
(D 1.8
--oi.. S.,-wAlb-t.
wi 1.6 LL 1.4 0 1.2 hi a.
0 -i
591
acidic pH treatment suggests that the A fragment is more stable than the holotoxin and, also, that heating at 650C for 30 min may not be sufficient to remove all LT activity from crude preparations. In addition, storage of LT at -70 and -20OC for months did not affect activity. Since these preparations consisted of purified extracellular LT released by pH adjustment, there was no need for activation with proteases, although the biological activity was enhanced by trypsin digestion. Pronase and proteinase K completely destroyed the biological activity of purified LT. In contrast to protease treatment, there was no effect on LT activity by deoxyribonuclease I, ribonuclease, and phospholipase D. Purified preparations contained neither carbohydrate nor lipopolysaccharide, as shown by a negative test for the presence of KDO. Nondissociated LT did not possess surface sulfhydryl groups, as no reaction occurred with dithiobisnitrobenzoic acid (DNTB-Ellman reagent). Furthermore, DNTB did not react with purified LT in the presence of 8 M urea. Identification of amino-terminal resi-
1.0 *;
*
*
.
.
.
.
*
*
95 11.54 13.5 7.5 MOLECULAR WEIGHTX 10
35
55
FIG. 6. Determination of molecular weight of 286C2 LT by disc gel electrophoresis.
^ ,,§
..
S
*:
w}
tion of 4 mg/ml (50,000 rpm, 20°C) did not reveal any polydispersity in purified LT preparations. The s2ow value was found to be 5.65. No attempt was made to calculate a molecular weight from these limited data. .Easily The purity of LT after Bio-Gel P-150 gel filtration was further assessed by SDS-gel elec3"I trophoresis. Upon incubation in neutral SDS at ,wF:; room temperature, two stained bands corresponding to molecular weights of 44,000 and 30,000 were observed (Fig. 7). The species with a molecular weight of 30,000 had activity in the PEL assay and likely corresponds to the same fragment synthesized by a porcine ENT' strain, 1362 (32). The 44,000-molecular-weight component dissociated upon heating in SDS to a species which corresponded to a molecular weight of 11,000. This subunit is probably the binding component of LT. LT purified from a porcine strain of ENT' E. coli (strain 263) had a similar subunit structure (Fig. 7). Stability and chemical characteristics of LT. As shown in Table 2, LT retained a portion FIG. 7. SDS-gel electrophoresis of purified LT of its biological activity at extremes of both pH preparations: (A) 286C2 LT, nonreduced; (B) 286C2 and temperature. The biological activity of LT LT, heated and reduced; (C) 263 LT, nonreduced; observed in the PEL assay after heating and and (D) 263 LT, heated and reduced. :. :?