Analysis of Receptor-binding Site in Escherichia coli Enterotoxin*

Heat-labile enterotoxin produced by enterotoxigenic Escherichia coli and cholera enterotoxin are both composed of A and B subunits. The A subunit is an enzy- matically active ADP-ribosylating subunit, while the B subunit, consisting of 103 amino acids, binds the toxin to a receptor, GM1-ganglioside, on the cell sur-face. A mutant isolated after treatment of E. coli producing heat-labile enterotoxin with N-methyl-N'-nitro- N-nitrosoguanidine produces a B subunit that is unable to bind to ganglioside. This subunit was purified and its primary amino acid sequence was determined. It differed from the native B subunit in only one amino acid at position 33; namely it had aspartate instead of glycine at position 33 from the N terminus. Thus glycine at position 33 from the N terminus of the B subunit is important for binding the B subunit to the ganglioside receptor. Cholera toxin (CT') and Escherichia heat-labile enterotoxin (LT) are causative agents of diarrhea and are structur-ally, functionally, and immunologically similar (1). Both tox-ins cause increase in cyclic AMP in target cells by catalyzing NAD-dependent ADP ribosylation, and are composed of two subunits, A and B. The A subunit has enzymatic activity and activates adenylate

Cholera toxin (CT') and Escherichia coli heat-labile enterotoxin (LT) are causative agents of diarrhea and are structurally, functionally, and immunologically similar (1). Both toxins cause increase in cyclic AMP in target cells by catalyzing NAD-dependent ADP ribosylation, and are composed of two subunits, A and B. The A subunit has enzymatic activity and activates adenylate cyclase in the cells, whereas the B subunit of both CT and LT recognizes the membrane component, GM1-ganglioside, on the cell membrane as a receptor and binds the holotoxin to the target cells, facilitating internalization of the A subunit (2).
The complete amino acid sequences, consisting of 103 amino acid residues, of the B subunits of CT and LT were deduced from DNA analyses and confirmed by amino acid sequence analyses of the purified proteins (3-9).
Although there is some suggestive evidence that the binding site of the B subunit of CT (CT-B) to GM1-ganglioside is in or near tryptophan at position 88 (10-12) or arginine at position 35 from the N terminus of , the precise position of the binding site in the B subunit has not yet been determined. To determine its position, we tried to isolate and *This work was supported by Grant-in-Aid for Special Project Research and Scientific Research from the Ministry of Education, Science and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

MATERIALS AND METHODS
Bacteria-A porcine strain of enterotoxigenic E. coli WT-1 producing L T (14, 15) was used.
Isolation of Tetracycline-sensitive Mutants-Mutagenesis was induced by the method of Koyama (16). A single colony of E. coli WT-1 was picked up, inoculated into 5 ml of minimal medium (16), and incubated with shaking for 8 h at 30 "C. Then 15 pg/ml N-methyl-N'-nitro-N-nitrosoguanidine (NGD) was added and the culture stood for 15 min at 30 "C. The cells were then harvested by centrifugation, washed once with minimal medium, and suspended in 2 ml of minimal medium. Then 0.5 ml of the mutagenized cell suspension was mixed with 3 ml of minimal medium or neopeptone broth (15) and incubated for 3 h at 37 "C with shaking. Penicillin and tetracycline were then added at final concentrations of 1000 pg/ml and 20 pg/ml, respectively, and after incubation for 16 h, 0.1 ml of the culture were plated on brain heart infusion agar (Difco). The plates were incubated at 37 "C for 1 day, and then the sensitivities of the colonies to tetracycline were tested on brain heart infusion agar plates containing tetracycline.
After incubation for 30 min a t 37 "C, the bacteria were collected by centrifugation and washed several times with g-syncase medium (15). Then they were resuspended in 5 ml of g-syncase medium, and a sample of 0.5 ml was added into 5-ml of g-syncase medium. This culture was incubated with shaking at 37 "C for 16 h. After incubation, a sample (0.1 ml) was plated on brain heart infusion agar, containing 40 ,ug/ml of tetracycline and incubated at 37 "C for 2 days. Colonies that grew on the agar plate were picked up and the immunological difference of their L T from that of native L T was examined by the Biken test (18).
Immunological differences were detected as spur formation between the precipitin lines of the parent and mutagenized cells on the Biken agar plates.
Chromatography-Gel filtration was carried out a t 4 "C on a column of Bio-Gel A-5m (Bio-Rad) equilibrated with TEAN buffer, composed of 50 mM Tris-HC1, 1 mM EDTA, 3 mM NaN3, and 0.2 M NaCl (pH 7.4). Material was eluted with the same buffer. Then the material was subjected to DEAE-Sephacel (Pharmacia Fine Chemicals) column chromatography. The B' subunit was eluted with 20 mM Tris-HC1 (pH 7.1), suspended in 20 mM Tris-HC1 (pH 7.1) containing 0.14 M NaCl and applied to anti-LT coupled-Sepharose 4B prepared by the method reported previously (19,20). The column was washed with 0.14 M NaCl containing 20 mM Tris-HC1 (pH 7.11, 0.5 M NaCl containing 20 mM phosphate buffer (pH 7.4), and 6 M urea containing 20 mM phosphate buffer (pH 7.4), and then the B' subunit was eluted with 0.1 M propionic acid containing 6 M urea.
The preparation was further purified on Sephadex G-75 (Pharmacia Fine Chemicals) and eluted with 0.1 M propionic acid containing 6 M urea. Fractions containing the B' subunit from the columns I3 Subunit of E. coli Enterotoxin 8553 were monitored by t.he double gel diffusion test on 1.21 Noble agar in 'I'EAN buffer (pH 7.4).
Iktrrminntion o f Protrin Contrnt-Protein content was determined by the method of Lowry et 01. (21) with crystalline bovine serum alhumin as a standard.
Succinvlation-Cm-R and €3' suhunits were dissolved at concentrations of ahout 3-5 mg of protein/ml in 0.5 M sodium-bicarbonate containing 6 M guanidine HCI. Solid succinic anhydride (60-fold molar excess per amino group) was added and the pH of the solution was maintained at between 8.0 and 9.0 by adding -5 N NaCl as descrihed previously (28).

I'rptidc.
Prrparation-The digests of the B and €3' subunits were each subjected to Gilson high performance liquid chromatography (HPLC) in 0.1% trifluoroacetic acid and the peptides were eluted with a gradient of 0-909; acetonitrile in 0.15 trifluoroacetic acid. Elution of peptides was monitored as absorbance at 220 nm. Fractions of peptides were pooled and lyophilized. Amino Acid Srqurncr Defrrmination-Amino acid sequences from the N terminus were determined by manual Edman degradation (24) using samples of about 0.1 p~ peptides. Phenylthiohydantoin amino ncids were identified by HPLC as described elsewhere ( 2 5 ) .
('irculor I)ichrnism-CD spectra were obtained from 205 to 260 nm with a Jasco J-500. In calculation of the mean residue ellipticity, 101, the mean residue weight was taken to he 103 for the R and R' subunits.

Immunological Character of LT Produced by the Mutant-
A crude extract of cells of the mutant grown in CAYE broth (1.51) was examined by the double gel diffusion test. As shown in Fig. la; LT produced by the mutant formed a precipitin line against anti-R subunit serum hut not against anti-A subunit serum. Fig. l b shows spur formation between the precipitin lines of the B and B' subunits with antisera prepared against LT or the R subunit. These data suggest that the mutant produces only a B subunit (R') that is immunologically different from the B subunit.
Reaction of the R' Subunit with Ganglioside in Gel-The binding ability of the B' subunit to ganglioside was examined by the double gel diffusion test. As shown in Fig. IC, the native R subunit formed a precipitin line against ganglioside mixture type 111 (Sigma Chemical Co.), hut the crude R' subunit did not, suggesting that the ability of the B' subunit to hind to the ganglioside was lost.
Purification of the R ' Subunit-For more detailed analysis of the R' subunit, we tried to purify it. As shown in Table I, the crude sample prepared from cells in 10 liters of CAYE medium as described previously (20) was precipitated with ammonium sulfate (70% saturation).
The precipitate was dissolved in TEAN buffer and subjected to Bio-Gel A-5m column chromatography in TEAN buffer. Although native LT and the R subunit were adsorbed to a Rio-Gel A-5m column (2 X 90 cm), the R' subunit was not adsorbed. Fractions (300 ml) containing the B' subunit in 3 liters of 20 mM Tris-HCI buffer (pH 7.1) were applied to a column of DEAE-Sephacel equilibrated with 20 mM Tris-HCI at a flow rate of 1 ml/min. The R' subunit passed through this column, whereas most other proteins were adsorbed. Fractions (500 ml) containing the R' subunit were mixed with NaCl at a concentration of 0.14 M and applied to an anti-LT immunoglobulin-coupled Sepharose 4R column equilibrated with 0.14 M NaCl at an elution speed of 1 ml/min. After absorption of the R' subunit, the column was washed with 20 mM Tris-HCI (pH 7.1) containing 0.14 M NaCI, 20 mM phosphate buffer Material was eluted with the same buffer. In this way we obtained 0.5 mg of purified R' subunit. Homogeneity of the Purified R ' Subunit-The homogeneity of the purified R' subunit was examined by SDS-polyacrylamide gel electrophoresis. As shown in Fig. 2, when about 5 pg of material was applied, the R ' subunit (tanv 2) gave a single hand on the gel, suggesting that the preparation was almost homogeneous. The mobility of the purified R' subunit was almost the same as that of the R subunit of LT prepared according to the method previously reported (20, 26). T h u s there seemed to he little difference between the molecular weights of the B and R' subunits.
Isoelectric Point-The isoelectric point of the R' subunit was determined by polyacrylamide gel electrophoresis. as shown in Fig. 3. The pI of the R subunit was calculated to he 10.0 and that of the R' subunit as 9.8. UVAbsorption Spectra of the H and R ' Subunits-The UV absorption spectra of the R' and R subunits from 250 t o 330 nm were compared. As shown in Fig. 4 Comparison of the B' and B Subunits by GMl-ganglioside ELISA-The binding abilities of the B' and B subunits to ganglioside were examined in ganglioside ELISA.
As shown in Fig. 5 , the B subunit showed positive absorption at a concentration of 10 ng/ml, but the B' subunit showed no absorption even at a concentration of 10,000 ng/ml in this ELISA test. Thus the binding ability of the B' subunit to GM,-ganglioside was at least 1000 times less than that of the B subunit. A crude sample of the B' subunit, which gave a precipitin line in the Ouchterlony test, did not react with the ganglioside in the ELISA test. Thus the mutant produces only the B' subunit, which cannot bind to GM1-ganglioside. Analysis of Binding of the B' Subunit to Ganglioside by SDS-PAGE-Van Heyningen (27) reported that after treatment with GM1-ganglioside the CT-B subunit did not enter urea-PAGE. We used this principle to compare the binding abilities of the B' and B subunits to ganglioside by SDS-PAGE. As shown in Fig. 6, treatment with ganglioside influenced the mobility of the B subunit but not that of the B' subunit, again suggesting that the B subunit bound to GMlganglioside but the B' subunit did not.
Inhibitory Effects of the B ' and B Subunits on LT Activity of CHO Cells- Fig.  7 shows that the dose-response curve of LT was affected by pretreatment of CHO cells with excess B subunit but not with excess B' subunit: 50% elongations of untreated CHO cells and cells pretreated with the B' subunit were observed with 63 and 100 pg/ml of LT, respectively, but after pretreatment of the cells with excess B subunit, 50% elongation was not observed even with more than 10 ng/ml of LT.
These data suggest again that the B' subunit does not combine with the binding site of LT.
Effects were mixed with the ganglioside mixture (1 mg) and incubated a t 37 "C for 60 min. The samples were treated with dithiothreitol (10 mM) a t 37 "C for 10 min and were analyzed with SDS-PAGE in 12% polyacrylamide containing 1% SDS. 1, B' subunit with ganglioside; 2, B' subunit; 3, B subunit with ganglioside; 4 , B subunit. B Subunit to CHO Cell-CHO cells were incubated with a constant amount of lZ5I-B subunit (75 ng/ml) and different amounts of unlabeled B or B' subunit at 37 "C for 60 min. Then the cells were washed repeatedly and the amount of bound "'1-B subunit was measured. As seen in Fig. 8, unlabeled B subunit (25 ng/ml) inhibited binding of the lZ5I-B subunit, causing almost complete inhibition at a concentration of 28 pg/ml. In contrast, unlabeled B' subunit was not inhibitory even at a concentration of 6 pg/ml. These data also support the conclusion that the B' subunit differs from the B subunit in the region of the binding site of the later to GMl- 3.0 X lo5 cells suspended in 1.5 ml of minimal essential medium culture medium containing 10% calf serum and 1% fetal calf serum and incubated for 30 min a t 37 "C in a C 0 2 incubator. Then, the cells were washed with phosphate-buffered saline containing 1 mg/ml of bovine serum albumin and harvested with a pipette. The cells were collected by centrifugation and their radioactivity was counted. and B' subunit was constructed by the method reported previously (26). The mixture of A and B' subunits in 0.1 M propionic acid containing 6 M urea was dialyzed against TEAN buffer and then analyzed by disc electrophoresis. As shown in Fig. .9, the mobility of the B' subunit (lane B ') was less than that of the A subunit (lane A ) . The reconstituted hybrid toxin gave a band (arrow) with a mobility intermediate between those of the B' and A subunits (lane Hy). As the B' subunit could form a hybrid toxin with the native A subunit, it must have the native conformation necessary for binding to the A subunit and thus not have a significantly different conformation from the B subunit.
As expected, the biological activity of the hybrid toxin, examined by CHO cell assay, was similar to that of the A subunit.
Circular Dichroic Spectra of B and B' Subunits-It was examined whether there was difference in secondary structure between B and B' subunits by obtaining each CD spectrum in the far ultraviolet region. As shown in Fig. 10, the pattern of CD spectrum of B' subunit was very similar to that of B subunit, suggesting that there was no significant change in secondary structure of B' subunit from native one.

HPLC Patterns of Chymotrypsin Digests of the B and B '
Subunits-Samples of 3-5 mg of the B and B' subunits were carboxymethylated, succinylated, and then digested with chymotrypsin as described under "Materials and Methods." The digests were then subjected to HPLC and eluted with a gradient of 0-90% acetonitrile in 0.1% trifluoroacetic acid. As shown in Fig. 11, the elution profiles were similar except that peak 4 of the B' subunit was eluted later than that of the B subunit.
Comparison of Amino Acid Compositions of Peptides of the B and B' Subunits-The amino acid contents of the peptides in peaks 1-10 separated by HPLC were determined (Table  11). Only the amino acid contents of the peptides in peak 4 of A Hy B'

FIG. 9. Hybrid toxin of the native A subunit and B' subunit.
The hybrid was formed as described previously ( the B and B' subunits were found to differ: the content of aspartate was increased and that of glycine was decreased in the peptides of the B' subunit, indicating replacement of glycine by aspartate. As shown in Table 11, the amino acid composition of peptide 3 was the same in the B and B' subunits but different from that reported by Dallas and Falkow (3): peptide 3 contained lysine but not methionine.  A-3300), after hydrolysis of samples in 25% hydrochloric acid at 110 "C for 24 and 72 h (Table 111). The numbers of various amino acid residues in the B and B' subunits are almost identical except for those of glycine and aspartate: the B' subunit has less glycine and more aspartate than the B subunit.

Amino Acid Compositions
Determination of Primary Sequence of Peak 4-We determined the primary sequence of the peptide in peak 4 to be Table IV. This sequence corresponded to residues 28-37 from the N terminus of the B subunit except for Asp* at position 33. This residue is glycine in the B subunit and aspartate in the B' subunit.

DISCUSSION
In this work mutagenesis of E. coli WT-1 (14) was induced with NGD as reported (15) and the revertant mutants were screened by the Biken test. We obtained 10 mutants that produce immunologically nonidentical LT. The mutant described here produces only a B subunit (the B' subunit) that differs immunologically from the native B subunit (Fig. lb).
We demonstrated that this B' subunit could not bind to ganglioside by several methods. 1) The binding ability of the B' subunit to GMl-ganglioside in GMl ELISA (Fig. 5) was at least 1000 times less than that of the native B subunit. 2) SDS-PAGE analysis showed that the B' subunit did not bind to ganglioside (Fig. 6). 3) The B' subunit did not form precip- Fraction 6 (residues 19-27) was contaminated about 10% (estimated from the value for Glu in the case of the B subunit) or 20% (estimated from the value for Glu in the case of the B' subunit) with another peptide (residues 49-67) and so its amino acid composition was calculated as total amino acids in fraction 6/10% or 20% of residues in peptide 49-67.
Values for valine and isoleucine in fraction 7 were lower than those determined from the DNA sequence, because amino acid analysis was performed on a 24-h hydrolysate and the sequence Val-Ile-Ile was not hydrolyzed completely.
e We did not determine the tryptophan content of fraction 10, but this fraction should contain tryptophan and consistent with this, it gave a high peak on HPLC. itin line against ganglioside (Fig. IC). 4) Pretreatment of CHO cells with excess B subunit but not excess B' subunit inhibited the biological activity of L T (Fig. 7). 5) In binding inhibition assay, unlabeled B' subunit did not inhibit the binding of lZ5I-B subunit to CHO cell (Fig. 8). All these data indicate that  (1) 1.66 (2) 1.02 (1) 1.50 (2) \ID' (1) 12 the B' subunit cannot bind to GM,-ganglioside, the receptor for LT. Then we determined the primary sequence of the purified B' subunit. Samples of the B and B' subunits were carboxymethylated, succinylated, and digested with chymotrypsin. The digests were then applied to a HPLC column and eluted with an acetonitrile gradient. All the elution peaks of peptides of the B' subunit were identical with those peptides of the B subunit, except that of peak 4 (Fig. 11). The amino acid compositions of the peptides (Table 11) in these peaks were compared with those of the peptides of the B subunit reported (3). All were the same, except for that of peak 4; peak 4 of the B' subunit contained more aspartate and less glycine than that of the B subunit. The amino acid sequence of the peptide in this peak was determined as Thr-Glu-Ser-Met-Ala-Asp*-Lys-Arg-Glu-Met. This sequence coincides with the sequence of positions 28-37 of the native B subunit reported except at position 33; the B' subunit has aspartate a t position 33 from the N terminus instead of glycine in the B subunit. Thus it is supposed that mutation of GGC encoding glycine to GAC encoding aspartate in the mutant could occur on treatment with NGD.
As a single amino acid exchange at position 33 from the N terminus in the B subunit results in loss of ability to bind to ganglioside, we suppose that position 33 is important for binding activity. We also observed an immunological difference between B and 8' (Fig. l b ) , suggesting that the binding site of the B subunit to GMl-ganglioside is one of the epitopes of B subunit. This 1 amino acid exchange at position 33 of the B' subunit could also cause some conformational change in the molecule that results in loss of ability to bind to ganglioside. However,

(CT)
His Ala the fact that a hybrid toxin could be formed between the B' 11. subunit and native A subunit suggests that the conformation 12. of the B subunit for binding to the A subunit was not affected by the 1 amino acid exchange. Moreover, CD spectrum of B' 13.
subunit was very similar to that of B subunit (Fig. 10). Thus these data suggested that there was no significant conforma-14. tional change in the B' subunit. De Wolf et al. (11,12) reported that the tryptophan residue 15. at position 88 from the N terminus might be near or in the binding site of the B subunit of cholera toxin to GM1-ganglio-16. side. However, this tryptophan residue was not affected in the 17. B' subunit.
The glycine at position 33 of the B subunit is very close to arginine a t position 35, which was reported to be important 19. for CT-B binding to ganglioside (13). Thus positions 33-35 from the N terminus of the B subunit seem to be important 20.
for binding of LT-B and CT-B to ganglioside receptor.