THE JOURNAL OF INFECTIOUS DISEASES • VOL. 133, SUPPLEMENT © 1976 by the University of Chicago. AIl rights reserved.

MARCH 1976

Mechanism of Activation of Adenylate Cyclase in Vitro by Polymyxin-Released, Heat-Labile Enterotoxin of Escherichia coli From the Department of Biology, Harvard University, Cambridge, Massachusetts; and the Program in Infectious Diseases and Clinical Microbiology, University of Texas Medical School, Houston, Texas

D. Michael Gill, Doyle J. Evans, Jr., and Dolores G. Evans

Heat-labile enterotoxic material released from Escherichia coli by polymyxin B activates the adenylate cyclase of pigeon erythrocyte ghosts in a time- and concentration-dependent manner. The activation requires nicotinamide adenine dinucleotide, adenosine triphosphate, and another component of the erythrocyte supernatant. The active species has a molecular weight of about 23,000-24,000 daltons, is inhibited by antibodies to the toxin of Vibrio cholerae, and is not irreversibly denatured by sodium dodecyl sulfate. Thus in many respects the active species from E. coli behaves the same as peptide Al of cholera toxin.

Analysis of molecular events during the activation of adenylate cyclase by Vibrio cholerae toxin has been facilitated by use of the pigeon erythrocyte lysate system. Although many details remain obscure, it has been shown that peptide A, of cholera toxin is the active species, that A, almost certainly catalyzes an intracellular reaction involving adenylate cyclase as one substrate, and that this reaction depends on nicotinamide adenine dinucleotide (NAD), adenosine triphosphate (ATP), and a soluble protein [1-3]. Although V. cholerae and E. coli enterotoxins are clearly different proteins, a considerable degree of homology is suggested by the similarities in their action. Both toxins cause diarrhea by elevating adenylate cyclase activity in the intestinal mucosa [4, 5], and both can activate adenylate cyclase in other vertebrate tissues [6]. Both toxins show a lag when acting on whole cells [7]. Both are inactivated by antisera to cholera toxin or to choleragenoid [8]. Moreover, Dorner and Mayer [9] have shown that E. coli toxin is also able to activate adenylate cyclase in vitro (cat myocardial membranes). Thus the possibility existed that the E. coli and V. cholerae toxins might have similar This work was supported by grant no. GB-13217 from the National Science Foundation. Please address requests for reprints to Dr. D. Michael Gill, Department of Biology, Harvard University, 16 Divinity Ave., Cambridge, Massachusetts 02138.

modes of intracellular action. In this paper we present preliminary evidence, obtained with use of erythrocyte lysates, that the modes of action of the two toxins are indeed similar, if not identical. We have identified an active species of E. coli toxin that is similar in size to A, of cholera toxin and that shows similar requirements in vitro for NAD, ATP, and cell supernatant. Such differences as do exist between the two toxins probably relate to those portions of the molecules that are involved with recognition of cell surfaces and the insertion of the active peptides into the cell. Materials and Methods

Cells of E. coli strain H-10407, from a 6-hr culture, were washed and treated with polymyxin-B as described earlier [8]. The material used in this work was a pool of rabbit-skin reactive fractions [10] derived by passage of concentrated polymyxin "extract" through a column of Agarose A-0.5m. This fraction contains principally peptides of about 20,000-40,000 daltons (rnodev -> 35,000 daltons) (see figure 3). The preparation contained approximately 500 [tg of protein/mI. The toxicity, as measured by blueing in rabbit skin, corresponded to that obtained with approximately 0.1 [tg of cholera toxin/nil or about 2 [tg of cholera toxin subunit Ayrnl. From the results presented in this paper, the in vitro activity of the preparation corre-



sponded to that of approximately 10 ftg of cholera toxin/nil (less if the cholera toxin had been activated by preincubation with dithiothreitol and sodium dodecyl sulfate [3]). Cholera toxin was prepared under contract for the National Institute of Allergy and Infectious Diseases by Dr. Richard A. Finkelstein (the University of Texas Southwestern Medical School, Dallas, Tex. [11]. Antitoxin to choleragen was a gift from Dr. R. O. Thomson (Wellcome Laboratories, Kent, England). Antisera to choleragenoid was a gift from Dr. Finkelstein. Activities were determined by incubation at 37 C of portions of the toxin solutions with pigeon erythrocytes that had been freshly lysed by freezing and thawing in one packed-cell volume of buffer (0.13 M NaCt, 0.01 M N-2-hydroxyethylpiperazine-N'-2-ethane sulfonic acid [HEPES], pH 7.3) [1]. After 30 min the activation was stopped by addition of cholera antitoxin; immediately thereafter adenylate cyclase activity in the incubation mixtures was determined [1]. The unit of adenylate cyclase activity recorded in the figures and tables is pmol of adenosine 3': 5'-cyclic phosphate (cyclic AMP) accumulated per hr by the catalytic activity of the ghosts of Iftl of erythrocytes (packed cell volume measured before lysis). Results

Despite the evident impurity of the E. coli enterotoxin preparation used, it was possible to answer certain questions relating to the mode of action of E. coli toxin and its relation to cholera toxin. The following similarities were observed. (1) As is the case for cholera toxin [l, 3], E. coli enterotoxin caused a progressive and dosedependent rise in the membrane-bound adenylate cyclase activity of concentrated lysates of pigeon erythrocytes. The rise started with no apparent lag (figure 1). A mixture of the two toxins had a greater effect than did either toxin alone (table 1). However, when the amount of cholera toxin was sufficient for very rapid activation of all the adenyl ate cyclase present, E. coli toxin had no further effect. Thus the change in the cyclase that is responsible for its elevated activity may be the same in both cases. With both toxins the activated adenyl ate cyclase was more responsive to

Gill et al.






Minutes Figure 1. Activation of adenylate cyclase in vitro by Escherichia coli enterotoxin. The numbers on the right indicate the concentration (ug of crude protein/rnl) of E. coli toxin used.

epinephrine than was the adenylate cyclase of control ghosts (table 1, values in parentheses). (2) Activation of erythrocyte lysate adenylate cyclase by cholera toxin depends on the presence of certain soluble components of the erythrocyte cytoplasm, notably NAD, ATP, and a soluble protein [2, 3]. E. coli toxin requires the same compounds. Its effect was increased by addition Table 1. Activation of adenylate cyclase in vitro by toxins of Vibrio cholerae and Escherichia coli. Adenylate cyclase activity* Cholera toxin (ug/rnl)

o I 3 10 30 100

Cholera toxin alone 2.2 (16) 14

20 28 34 46


Cholera and E. coli toxinst

12.8 (38) 16 22 38 44 47

* Expressed as pmol of adenosine 3':5'-cyclic phosphate (cyclic AMP) accumulated/hr by ghosts of I !J,l of erythrocytes. Activities were measured after incubation of portions of an erythrocyte lysate for 30 min at 37 C with the amounts of toxin shown. Values in parentheses represent adenylate cyclase activities measured in the presence of 0.1 mM epinephrine. t The amount of E. coli toxin used corresponded to 50 ug of protein/mI.

E. coli Enterotoxin Activity in Vitro


of ATP or NAD and was reduced by hydrolysis of the endogenous NAD or ATP of the lysate or by omission of soluble erythrocyte proteins (table 2). (3) Antisera to cholera toxin completely prevented the activity of both toxins. Similarly, an affinity column consisting of cholera antitoxin linked to Sepharose 4B adsorbed both cholera toxin and the active species of E. coli toxin (table 3). We were somewhat surprised to find that antisera to choleragenoid, supposedly devoid of antibody to subunit A, also removed active material from the preparation of E. coli toxin. A possible reason for this is given in the Discussion. (4) The same antisera do not cause a decrease in adenylate cyclase activity once the activity has been raised by incubation with either toxin. We have argued elsewhere that the lack of reversibility by antitoxin reflects an enzymic mode of action of cholera toxin, whose continued presence next to the cyclase is apparently not reTable 2. Dependence of the activities of Vibrio cholerae toxin and Escherichia coli enterotoxin on nicotinamide adenine dinucleotide (NAD), adenosine triphosphate (ATP), and erythrocyte supernatant. Adenylate cyclase activityt

Experiment no., composition of assay mixture* 1. 2. 3. 4. 5. 6. 7.

Control (ghosts and supernatant) Control 0.2 mM NAD Control 5 mM ATP Supernatant omitted Preincubated alone Pre incubated NADase Preincubated ATPase

+ +

+ +

Cholera toxin

E. coli

No toxin 1.5 1.5 1.5 1.5 1.3 1.3 0.9

4.2 5.6 8.2 1.5 4.0 1.7 1.2

4.2 8.5 9.4 1.9 3.2 1.4



* The table records adenylate cyclase activities measured after incubation of portions of a reconstituted lysate (washed ghosts, 10 ul; erythrocyte supernatant, 40 ul) for 30 min at 37 C with 1 ul of cholera toxin (0.5 ug/rnl) that had been activated by preincubation with 1 mM dithiothreitol, 0.5% sodium dodecyl sulfate [3], or 1 1-11 of crude E. coli toxin. For experiment 4, the supernatant was replaced by 40 ul of NaCI-HEPES buffer [1] containing 0.2 mM NAD and 5 mM ATP. For experiments 5, 6, and 7, the reconstituted lysate was preincubated for 5 min at 37 C with no additives, with neurospora NADase (0.1 unit/ml), or with apyrase (potato ATPase, 0.1 mg/ ml), respectively, before incubation with the toxins. t Expressed as pmol of cyclic AMP accumulated/hr by ghosts of 1 ul of erythrocytes.

Table 3. Absorption of Escherichia coli and Vibrio cholerae toxins by insolubilized antisera to choleragen and choleragenoid (anticholeragen and anticholeragenoid) . Adenylate cyclase activity* Absorbant None Insolubilized anticholeragen Insolubilized anticholeragenoid

E. coli toxin

Cholera toxin

2.8 0.4 0.5

22 3.4 4.8

NOTE. Five-tenths milliliter of 0.85% NaCl containing 0.1 ml of E. coli toxin or 10 I-Ig of cholera toxin were each passed twice over 0.3-ml columns consisting of cholera antitoxin or anticholeragenoid that had been linked to cyanogen bromide-activated sepharose 4B by standard procedures. The equilibrium capacity of the columns was approximately 10 ug of cholera toxin. Untreated toxin was incubated for 30 min with lysed erythrocytes at final protein concentrations of 7 ug/ml (E. coli, crude) and 7 ug/rnl (cholera, pure); volumes equivalent to these amounts of protein were used for the columnabsorbed material. * Expressed as pmol of cyclic AMP accumulated/hr by ghosts of 1 ul of erythrocytes.

quired to maintain the activated state [3]. The argument holds equally for E. coli toxin. (5) The crude E. coli toxin was fractionated by electrophoresis on polyacrylamide gels containing 0.1 % sodium dodecyl sulfate. Gels were cut into slices, the slices were extracted, and the extracts were assayed for ability to activate adenylate cyclase (figure 2). A single active peak was found at about the same position as peptide A, of cholera toxin (-- 23,500 daltons). The resolving power of this method is such that the active species must differ in size from A, by 3,000 daltons at most and may have the same size. Without prior reduction of disulfide bonds, peptide A, of cholera toxin remains linked to peptide A 2 and therefore migrates more slowly on gels. The active species of the E. coli toxin preparation, however, migrates at the same rate whether reduced or not; presumably the active portion of the molecule is not connected by disulfide bridges to any other sizeable peptide. We attempted to identify the active species on stained gels by determining which peptides were absent after adsorption with insolubilized cholera antitoxin (figure 3) . The band corresponding most closely in migration rate to the active peptide (Ad of cholera toxin was largely removed by

Gill et al.




Slice Figure 2. Sizes of active species of Vibrio cholerae toxin and of Escherichia coli enterotoxin. Polyacrylamide gels (10%) containing 0.1% sodium dodecyl sulfate were used; samples were reduced with dithiothreitol. (0--0) = 10 flg of cholera toxin. The activity peak coincides with peptide A l . 200 ul of E. coli toxin. Migration was from left to right. After electrophoresis, the gels were cut into 2-mm slices. Slices were extracted overnight at room temperature (- 24 C) with 0.2 ml of 0.13 M NaCl containing 1 mg of serum albumin/ml, 0.01 % NaNg, 10 mM N-2-hydroxyethylpiperazine-N'-2-ethane sulfonic acid (HEPES), pH 7.3. Extracts of each slice (10 ul) were incubated for 1 hr with 40 fll of pigeon erythrocyte lysate. The graphs record the adenylate cyclase activities obtained.

(.--.) =

the column. Unfortunately, we cannot conclude definitively that this band corresponds to the active material in E. coli enterotoxin, for it did not disappear entirely and several other bands also declined in intensity. It is, of course, possible that several of the peptides seen on sodium. dodecyl sulfate gels are derived from different portions of E. coli toxin. Discussion The ability of cholera toxin to activate adenylate cyclase in vitro resides in one of its component peptides, A l . A small amount of A, is liberated from the complete toxin at the start of the in vitro assay; a much greater response is obtained by denaturation and reduction of the toxin to release A l quantitatively before the start of the assay [3].

We have now shown that material released from polymyxin B-treated enterotoxic E. coli is able to activate adenylate cyclase in vitro in a manner very similar to activation by peptide A l of cholera toxin. The mechanism whereby the cyclase is activated appears to be the same, the nature of the change appears to be the same, and the responsible peptide appears very similar; its size is about the same as that of peptide A h and it is inactivated by antibodies to cholera toxin. In fact, no substantial difference has been found between the active peptides of the two toxins. The possibility exists, therefore, that the two peptides may not merely be very similar but may, in fact, be identical. E. coli enterotoxin is neutralized not only by antisera to cholera toxin [12], which may contain antibodies directed against both the A and B subunits of the cholera toxin, but also by antisera to choleragenoid (anti-B) [8], which inhibits the activity of isolated A only weakly and probably nonspecifically. Thus, in addition to its

I I \ \





' .....

Figure 3. Gel electrophoresis of crude Escherichia coli enterotoxin. The solid line represents 0.1 ml of toxin precipitated with four volumes of acetone at -20 C. Dashed line represents 0.1 ml of toxin mixed with 0.3 rnl of 0.85% NaCl and passed thrice over a 0.3-ml column bed of insolubilized cholera antitoxin. The column was then washed with 0.5 ml of 0.85% NaCl. Combined eluates were precipitated with acetone as above. Precipitates were collected by centrifugation, redissolved in a solution containing 0.1% sodium dodecyl sulfate and 50 mM dithiothreitol, and fractionated by electrophoresis on 10% polyacrylamide gels in the presence of 0.1% sodium dodecyl sulfate. Migration was from left to right. Gels were stained with Coomassie blue and scanned at 540 nm. A band that co-migrates with Al of cholera toxin is marked with an arrow.

E. coli Enterotoxin Activity in Vitro

close similarity to peptide A, of cholera toxin, E. coli enterotoxin probably also shares some determinants with the B component of cholera toxin. The relation to B (the portion of cholera toxin concerned with cell surface recognition) does not appear close, however, for there are marked differences between the binding of the two toxins to cells [13]. In the present experiments we found that the activity of E. coli enterotoxin in vitro was inhibited, and the active principle was adsorbed, by antitoxin to choleragenoid. A simple explanation of this observation would be that our preparation of E. coli enterotoxin contained at least two noncovalently bonded peptides, one of which is analogous to A of cholera toxin and the other of which is less closely analagous to B of cholera toxin. It is clear that further experiments are needed to determine the origin of the material that we studied and to determine its relation to the highermolecular-weight E. coli enterotoxin described by others [14, 15]. We would particularly like to know whether the active peptide here identified is- a precursor or a derivative of the larger species. References

1. Gill, D. M., King., C. A. The mechanism of action of cholera toxin in pigeon erythrocyte lysates. J. Biol. Chern. 250:6424-6432, 1975. 2. Gill, D. M. The involvement of nicotinamide adenine dinucleotide in the action of cholera toxin in vitro. Proc. Natl. Acad. Sci. U.S.A. 72:20642068, 1975. 3. Gill, D. M. Multiple roles of erythrocyte supernatant in the activation of adenyl ate cyclase by Vibrio cholerae toxin in vitro. J. Infect. Dis. 133 (Suppl.) : S55--S63, 1976.


4. Evans, D. 1., Jr., Chen, L. C., Curlin, G. T., Evans, D. G. Stimulation of adenyl cyclase by Escherichia coli enterotoxin. Nature [New Biol.] 236: 137-138, 1972. 5. Guerrant, R. L., Pierce, N. F., Ganguly, U., Greenough, W. B., III, Wallace, C. K. Mechanism of action of an Escherichia coli enterotoxin [abstract]. J. Clin. Invest. 51:39a, 1972. 6. Hewlett, E. L., Guerrant, R. L., Evans, D. J., Greenough, W. B., III. Toxins of Vibrio cholerae and Escherichia coli stimulate adenyl cyclase in rat fat cells. Nature (Lond.) 249:371-373, 1974. 7. Evans, D. G., Evans, D. 1., Jr., Pierce, N. F. Differences in the response of rabbit small intestine to heat-labile and heat stable enterotoxins of Escherichia coli. Infec. Immun. 7:873-880, 1973. 8. Evans, D. J., Jr., Evans, D. G., Gorbach, S. L. Polymyxin B-induced release of low-molecular weight, heat-labile enterotoxin from Escherichia coli. Infec. Immun. 10:1010-1017, 1974. 9. Dorner, F., Mayer, P. Escherichia coli enterotoxin: stimulation of adenylate cyclase in broken-cell preparations. Infec. Immun. 11:429-435, 1975. 10. Evans, D. J., Jr., Evans, D. G., Gorbach, S. L. Production of vascular permeability factor by enterotoxigenic Escherichia coli isolated from man. Infec. Immun. 8:725-730, 1973. 11. Finkelstein, R. A., LoSpalluto, J. J. Production of highly purified choleragen and choleragenoid. J. Infect. Dis. 121 :563-572, 1970. 12. Gyles, C. L., Barnum, D. A. A heat-labile enterotoxin from strains of Escherichia coli enteropathogenic for pigs. 1. Infect. Dis. 120:419-426, 1969. 13. Pierce, N. F. Differential inhibitory effects of cholera toxoids and gangliosides on the enterotoxins of Vibrio cholerae and Escherichia coli. J. Exp, Med. 137:1009-1023, 1973. 14. Dorner, F., Jaksche, H. Escherichia coli enterotoxin: purification, partial characterization, and immunological observations. J. Infect. Dis. 133 (Suppl.) :SI42--S156, 1976. 15. Finkelstein, R. A. Isolation and properties of heatlabile enterotoxins, from enterotoxigenic Escherichia coli. J. Infect. Dis. 133(Suppl.):SI20--S137, 1976.

Mechanism of activation adenylate cyclase in vitro by polymyxin-released, heat-labile enterotoxin of Escherichia coli.

THE JOURNAL OF INFECTIOUS DISEASES • VOL. 133, SUPPLEMENT © 1976 by the University of Chicago. AIl rights reserved. • MARCH 1976 Mechanism of Activ...
396KB Sizes 0 Downloads 0 Views