Chemical and Immunochemical Studies on the Receptor Binding Domain of Cholera Toxin B Subunit*

The contributions of various amino acids to the struc- ture and function of cholera toxin B subunit were assessed with quantifiable, chemically conservative, reversible derivatizations, and sensitive assays of ac- tivity. A panel of mon~lonal antibodies was employed to monitor the conformational integrity of modified protein and help distinguish the direct from indirect effects of chemical derivatization. We describe a novel monoclonal antibody, which competes with the receptor G M ~ for binding to cholera toxin B subunit, and use this reagent to help identify critically located residues. Our data support the hypothesis that par- ticipates directly in binding GMl. In addition, we pro-pose a dual role for first, these basic residues maintain an electrostatic attraction vital to receptor recognition; second, at least 1 lysine resides near the receptor binding domain and may interact with G M ~ . The suggest, variance vital the structure and function

11 Recipient of the John A. Hartford Foundation Fellowship. To whom reprint requests should be addressed. and CysS6. The B subunit's specific receptor is a glycosphingolipid, the monosialoganglioside GM1. l The dissociation constant for this interaction is approximately 10"' M (11). The oligosaccharide moiety of GM~, devoid of ceramide (OS-GMJ, exhibits similar affinity for CT-B (12,13). Studies reveal that five OS-&, molecules can simultaneously bind the B pentamer (13).
Although much is known about CT-B, the chemical basis of its biologic functions has not been fully elucidated. Specifically, the amino acids which mediate receptor recognition remain largely undefined. Consequently, CT-B constitutes an interesting model with which not only moIecular pathogenesis, but also p~~~n -c a r b o~y~a t e interaction can be examined.
Physical studies of the B subunit have shown that the binding of G M~, but not related gangliosides, induces a shift in the fluorescence emission spectrum maximum of cholera toxin from 342 nm to 330 nm (14). This "blue shift," which results from interaction between the B subunit and the oligosaccharide moiety of GM1 (15,16), reflects a conformational change affecting the single tryptophan residue of CT-B. Using solute quenching and fluorescence energy transfer studies, De Wolf et al. (17) found that this tryptophan is located in a positively charged microenvironment near the G M~ binding site.
Several investiga~rs have modified CT-B with chemical reagents designed to derivatize specific amino acids. Lonnroth and Holmgren (18) assessed the effects of 20 such modifications on the antigenicity, toxicity, and receptor binding activity of holotoxin. Markel et al. (19) found that lysine-directed acylation abolished the GM1 binding properties of CT-B, but reduction and alkylation of the disulfide bond had no significant effect. Duffy and Lai (ZO), using the ~~d i n o -s p~~c reagent cyclohexanedione, implicated A r e as critical to the antigenicity and function of CT-B. Support for the involvement of tryptophan in receptor recognition stems from a study by De Wolf et aL (21) employing 2,4-dinitrophenylsulfenyl chloride.
Chemical modification studies indeed provide important data regarding the role of various amino acids in the structure and function of proteins. However, interpretation of such data is limited by a fundamental ambiguity-difficulty in distinguishing the direct effects of site-specific derivatizations from allosteric effects, conformational changes, and side-reactions. This ambiguity is compounded by failure of the above-cited chemical studies to demonstrate one or more of the following: (a) reversibility, confirming the chemical specificity of modification; (b) preservation of higher order protein structure, thereby eliminating conformational distortion as a possible cause of biologic inactivity; (c) functional sequelae of derivatization with sensitive assays which differentiate between minor decreases in binding affinity and major d i s~p t i o n of activity. The present study examines the contribution of various amino acids to the receptor binding function of CT-B using quantifiable, reversible derivatizations and sensitive assays of binding. A panel of monoclonal antibodies is employed as reporter groups to monitor the conformational integrity of modified protein and thus help differentiate between the direct and indirect effects of chemical derivatization. In addition, we use a monoclonal antibody that competes with (& to identify residues located near the receptor binding domain of CT-B.
Preparation of OS-GM1-The oligosaccharide moiety was cleaved from GMI by ozonolysis and alkaline fragmentation, purified by ionexchange chromatography, and assessed by thin layer chromatography according to the method of Wiegandt and Bucking (23) as modified by Fishman et al. (13). The concentration of OS-GxI was determined by the thiobarbituric acid assay of Aminoff (24) as adapted by Ledeen and Yu (25). P r e~r~~o n of M o~c~~ A~~~~~ ~~~) -M e t h~s for the preparation of mAbs 32D3,40D9,4E2,22C6, and 15Cll were previously described (26). Two additional clones were obtained with different screening procedures for the same hybridoma bank. mAb 40B10 (isotype IgG1) was detected by screening for clones producing antibodies which bound toxin but not the Gvl-toxin complex, and mAb 35G8 (isotype IgG1) by screening for cross-reactivity with toxin from V. cholerae strain 569B but not 3082.
Protein M~~f~a~~-T~t o p h a n y l residues were formylated in HCl-saturated formic acid accordmg to Previero et a aE. (27), as modified by Holmgren (28). Formic acid was distilled prior to use. The extent of derivatization was calculated from the absorhance at 298 nm using a molar extinction coefficient of 4880 cm" (27). Deformylation was conducted by incubation in 0.2 M phosphate buffer at pH 11.5 for 30 min or 0.5 M ammonium chloride, pH 9.6, containing 8 M urea, overnight. Quantitative citraconylation of lysine was performed with a 100-fold excess of anhydride (moles/mole amino group) in 0.2 M borate buffer, pH 8.2,O "C (29). After 1 h, the solution was divided in half. One portion was dialyzed against borate buffer, at pH 8.0, 0 "C. The second portion was dialyzed against 0.2 M formate buffer, pH 2.8, 25°C for 3 h to effect decitraconylation and then against borate buffer, pH 8.0. Free amino groups were determined by TNBS assay (30). Irreversible acylation was conducted with acetic or succinic anhydride (from 0.01 to 100 mol/mol of amino group) in borate buffer, pH 8.2, 0 "C for 30 min (31). Modified protein was precipitated with 6% cold trichloroacetic acid, washed with cold acetone, dried in vacuo, and resuspended in borate buffer at pH 8.0.
The extent of modification was assessed by TNBS assay (30). Reductive alkylations were done at 0 "C in 0.2 M borate buffer,.pH 9.0, with formaldehyde or acetone and sodium borohydride accordmg to Means and Feeney (32). The reaction was terminated after 1 h by tricbloroacetic acid precipitation, and free amino groups were assayed as discussed above. Arginine-specific derivatization was accomplished with cyclohexanedione in 0.2 M borate buffer after Patthy and Smith (33) as performed by Duffy and Lai (20). Extent of m~f i c a t i o n was controlled by varying reaction conditions (Table II) and determined by amino acid analysis. Free arginine concentration was analyzed following bydrolysis conducted at 110 "C for 24 h in the presence of 1% mercaptoacetic acid (33). Reversal of modification was accomplished by incubation in 0.5 M hydroxylamine, pH 7.0, overnight. Acylation of tyrosine was performed with a 60-fold molar excess of acetylimidazole in borate buffer at pH 7.5 according to the procedure of Riordan et al. (34) for determination of free and buried tymsyl residues. The reaction was quantitated by spectrophotometry (623 = 1160 M" cm"). Reduction and alkylation of the disulfide bond, based on the procedure of Crestfield et ai. (351, was conducted in 6 M guanidine HCl or 8 M urea at p H 8.1 with dithiothreitol (100 mol/ mol of Cys) and iodoacetic acid or iodoacetamide (2.5-fold molar excess over dithiothreitol). Denaturant was removed by dialysis against 4, 2, 1, and 0.5 M guanidine HCI or urea, and then 0.1 M phosphate buffer, pH 7.5. The extent and specificity of this reaction was assessed by amino acid analysis. Protein concentrations were determined by absorbance using an extinction coefficient at 280 nm of 0.956 ml/mg/cm (36), Lowry assay (371, and amino acid analysis. Assays-The receptor binding function of modified protein was assessed by competitive r a d i o b m~g assay. Polyvinyl microtiter wells (96 per plate) were incubated with 100 .d of c7Mf (2 gg/ml) in phosphate-buffered saline at pH 7.4 (PBS) for 1 h, and then washed four times with 0.2% (w/v) BSA in PBS. Serial dilutions of modified protein were mixed with approximately 10,000 cpm of '=I-labeled CT-B, 10 mCi/mg (38), and the resulting solution was added to the wells. After 2 h, the wells were washed and clipped from the plate, and bound radioactivity was determined by y-counting. The ability of modified protein, at concentration X, to block binding of '%CT-B to solid-phase is expressed as Receptor binding activity is defined as the ratio of IC5,, (native CT-B) to ICs0 (modified CT-B) multiplied by loo%, where IC60 is the concentration of competitor which produces a 50% reduction of counts bound. The antigenicity of derivatized protein was assessed by a modification of the assay described above: CT-B (5 pg/ml) was coated onto microtiter wells by overnight incubation in PBS. The wells were washed with PBS/BSA, serial dilutions of modified protein were added, and monoclonal antibody (MOO0 dilution of culture Supernatant) was introduced. After 12 h at 37 "C, the wells were washed and approximately 50,000 cpm of '251-protein A (40 mCi/mg) was added. One h later, plates were washed and bound radioactivity was determined. The ability of modified protein to block the binding of monoclonal antibody to solid-phase CT-B is expressed as per cent maximum binding (see above).

A n~~d~s Recognize at Least Five Dis-
tinct ~o n f o r~t~n -~e~~w e Epitopes on CT-3-The preparation and binding specificity of five anti-CT-B monoclonal antibodies have been reported (26). Two additional mAbs (40B10 and 15Cll) were obtained by alternative screening methods (see "Experimental Procedures"). These antibodies recognize at least three different epitopes based on crossreaction patterns with the CT homologues hLT and pLT produced by e n t e~t o~g e n i c strains of ~s c~r~~i a coli ( Table  I). Cross-reactivity with chemically modified CT-B indicates the presence of two additional epitopes (discussed below and summarized in Table I). Several observations suggest that these antigenic determinants are composed of residues noncontiguous in sequence but juxtaposed in space by virtue of protein folding. First, none of these mAbs binds any of eight

cross-reactivity of the anti-CT-B monoclonal antibodies
The antigenicity of the CT homologues hLT and pLT produced by enterotoxigenic strains of E. coli, and chemically modified CT-Bs were assessed by competitive radiobinding assay (see "Experimental Procedures") or solid-phase radiobinding assay as described below. Polyvinyl microtiter wells were sensitized with antigen (5 pg/ ml in PBS overnight), washed with 0.2% (w/v) BSA in PBS, and incubated with serial dilutions of monoclonal antibody for 12 h at 37 "C. The wells were washed and approximately 50,000 cpm of lSI-labeled protein A was added. One h later, the plates were washed and bound radioactivit~ was determined bv 7-counting.

~o n~l o n a l antibody
HCl/formic acid Citraconic anhydride CH20/NaBH4 Cyclohexanedione Acetylimidazole Dithiothreitol/ICH&OOH Cross-reactivity was determined by competitive radiobinding assay. Cross-reactivity was determined by solid phase radiobinding assay.
' CT-B was bound to solid-phase GM,.
* Minus signifies less than 5% binding in comparison to native CT-B.
synthetic peptides or three proteolytic fragments; even though these peptides together encompass essentially the entire primary structure of CT-B with considerable overlap. Also, none of the peptides alone or in combination blocks mAb recognition of CT-B. Second, denaturation of CT-B by treatment with 2% (w/v) sodium dodecyl sulfate and 0.5 M 2mercaptoethanol at 100 "C for 5 min attenuates or abolishes the binding of all seven mAbs in Western blot analysis (data not shown). Third, reduction and c~~~e t h y l a t i o n of the cystine residue results in loss of all mAb binding sites (see below). Therefore, these antibodies are presumed to bind determinants that depend upon the conformational integrity of the B subunit. They are used in this study as reporter groups to monitor changes in higher-order structure that might occur as a result of chemical modifications. Modifications that preserve the binding of most or all of these reagents are operationally defined as conformation-conse~ative. mAb 40310 Competes with the ~l~o s~~~r~ Moiety of GM1 for Binding to CT-B-mAb 4OB10 does not recognize receptor-bound toxin ( Table I), suggesting that this antibody's epitope resides near the G M 1 binding site. To explore this possibility, we conducted a competitive radiobinding assay between each mAb and the oligosaccharide moiety of G M~ (Fig. 1). Nanomolar concentrations of OS-GMI block the binding of only 40B10. Whie competition of this nature could result from steric hindrance or allosteric interaction, the small size of OS-GM1 (M, -1,000) and the preservation of the other six mAb binding sites imply that 40B10 recognizes amino acids residues near or within the receptor binding domain of CT-B. We, therefore, examined the effects of protein modification upon not only the binding of GM1 but also 40B10, as an additional method by which critically located residues may be identified. ~~i~u~n of T~p t o p~n -T h e B subunit contains a single tryptophan (Fig. 2 ) whose p~i c i p a t i o n in receptor 'The eight synthetic peptides correspond to the following CT-B residues: 1-9, 9-21, 26-38, 40-50, 50-61, 62-73, 73-86, and 86-97. The three proteolytic fragments (residues 1-37, 38-68, and 69-101) were produced by cyanogen bromide cleavage of reduced and carboxymethylated CT-B.  Table I) at 1:5000 dilution of culture supernatant, and the resulting solutions were added to polyvinyl microtiter wells containing CT-B in the solid phase. After 12 h at 37 "C, the wells were washed and approximately 50,000 cpm of lZ51-protein A was added. One h later, the plates were washed and bound radioactivity was determined by y-counting. Nanomolar concentrations of OS-GM~ competitively inhibited the binding of mAb recognition has been implicated by physical studies (14)(15)(16)(17).
Modification of this residue with 2,4-nitrophenylsulfenyl chloride by De Wolf et al. (21) abolished Gwl binding activity. They, however, could neither reverse this loss of activity nor discount the possibility that derivatization adversely affects protein conformation. Treatment of protein with HC1-saturated formic acid results in the quantitative formylation of tryptophan without affecting other amino acid side chains or the peptide backbone (27). The reaction is monitored spectropho~metrically, since the near ultraviolet spectrum of N-formyl tryptophan is shifted toward longer wavelengths. Fig. 3A illustrates the increase in absorbance at 298 nm accompanying formylation of CT-B. The derivatization is complete in 2 h, as estimated with a molar extinction coefficient Ez$s = 4880 cm-l. Defor-mylation and the concomitant reversal of spe~rophotometric changes occur at high pH (Fig. 3B). Formylated CT-B does not bind GM1, while deformylated protein regains partial activity (Fig. 4A). Complete functional reversal was not achieved owing, perhaps, to degenerative changes during incubation at high pH. Nevertheless, partial restoration of activity confirms the importance of tryptophan in binding GMl. Formylation reversibly disrupts the binding of mAb 40B10 (Fig. 4B), but does not affect binding of the other monoclonal antibodies (Table I). These findings suggest that tryptophan does indeed reside near the receptor binding domain of CT-B and that functional p e~r b a t i o n s associated with formylation result not from gross conformational distortion, but rather the direct and local effects of covalent derivatization.
Modification of Lysine-The B subunit contains 9 lysines (Fig. 2). Modification of lysyl residues with a 100-fold molar excess of succinic anhydride in guanidine HCl by Markel et al. (19) abolished receptor binding activity. This acylating agent, however, is not absolutely specific for amino groups and reacts, even near neutral pH, with tyrosyl, histidyl, cysteinyl, seryl, and threonyl residues (39, 40). Moreover, fully succinylated CT-B gains approximately 20 negative charges and may be unable to renature following exposure to guanidine HC1. Here, we examine the importance of lysine with a reversible modification conducted under conditions of physiologic pH and ionic strength.
The amino groups of CT-B react quantitatively with citraconic anhydride in 0.2 M borate buffer, pH 8.0, as determined by TNBS assay (data not shown). Decitraconylation of the N-acyl derivative occurs spontaneously at low pH. 0-Acyl groups that may form as side reactions are stable under these conditions (39). Fig. 5 depicts the effects of modification on 10 function. Citraconylated CT-B does not bind GMI; low pH incubation restores activity. Since deacylation is specific for N-substitution, the impact of side-reactions can be discounted, demonstrating the contribution of lysine to receptor recognition.

Lys-Leu-Cys-Val-TrpAsn-Asn..Lys-Thr-Pro-His-Ala-lle-Ala-Ala-Ile-Ser
The B subunit is positively charged at neutral pH (41); in contrast, GM1 contains a sialic acid residue and is negatively charged. Lysine, as the predominant cationic residue (Fig. 2)) may play a vital role in the binding event by maintaining electrostatic attraction. To expfore this possibility, CT-B was modified with agents which exert differing effects on the charge of primary amino groups. Reductive alkylation with acetone or formaldehyde in the presence of sodium borohydride yields the corresponding monoor dialkylamine and preserves charge. Acylation with acetic or succinic anhydride converts lysyl residues to neutral or anionic derivatives, respectively. In each case, modification was conducted with an increasing molar ratio of reagent to protein and resulted in progressive loss of free amino groups, as monitored by TNBS assay. Fig. 6 depicts receptor binding activity of the CT-B derivatives as a function of amino groups modified. Reductive isopropylation does not alter the charge on primary amino groups and has no effect on activity. M~f i c a t i o n with succinic anhydride reverses this charge and leads to the sequential loss of function. Acetylation, which neutralizes the charge, also inactivates CT-B; but to achieve the same reduction of activity, about twice as many acetyl as compared to succinyl substitutions are required. For example, both 32% succinylated and 58% acetylated CT-B experience an average charge change of approximately -6 and both exhibit 100-fold decreased activity. Since the acetyl and isopropyl groups are of roughly equal mass, the significance of charge relative to steric effects is substantiated. These data suggest that electrostatic attraction, as maintained by lysine, participates critically in the receptor binding event.
The anti-CT-B monoclonal antibodies were screened for cross-reactivity with lysine-modified protein (Table I). Reductive methylation and isopropylation disrupt the binding of 40B10, but no other mAb. Therefore, a lysyl residue may constitute part of the epitope of this antibody and reside near the receptor binding domain of CT-B.
M~i f~a t w n of Arginine-The B subunit contains three arginyl residues at positions 35, 67, and 73. Duffy and Lai (20) treated CT-B with 2,4-cyclohexanedione and found that derivatization of second of the 3 residues to react, correlates with loss of receptor binding activity, as determined by an Ouchterlony-type double-diffusion analysis. The double-diffusion analysis, however, does not measure binding directly, relying instead upon visualization of stable CT-B/ were mixed with A, about 10,000 cpm of lZ5I-labeled CT-B and allowed to compete for binding to G M~; or, B, mAb 40B10 (1:5000 dilution of culture supernatant) and allowed to compete for binding to native CT-B. See "Experimental Procedures" for details. was mixed with about 10,000 cpm of '=I-labeled CT-B and allowed to compete for binding to Garl (see "Experimental Procedures"). GM1 precipitin bands. Thus, any factor affecting binding energy, valence, or solubility of the precipitate can influence results. Moreover, this assay is not quantitative.
We reacted CT-B with cyclohexanedione under conditions that resulted in the progressive modification of arginyl residues (Table 11), and monitored function by competitive radiobinding assay. Fig. 7 depicts the results of this experiment. Function is not abruptly lost upon reaching any specific degree of derivatization. Instead, activity declines continuously with per cent arginine modified, indicating that any of the 3 residues, alone or in combination (e.g. through their joint contribution to electrostatic charge), may participate in receptor recognition. Accordingly, the effects of cyclohexanedione upon the B subunit are not, with certainty, attributable to modification of Arg?'. This conclusion can be recon-  -) or succinic ( U ) anhydride converts lysines to neutral or acidic residues, respectively. Reactions were conducted with an increasing molar ratio of reagent to protein and resulted in progressive loss of free amino groups, as monitored by TNBS assay. Receptor Binding Activitydefined as the ratio of ICs0 (native CT-B) to ICs0 (modified CT-B) where IC% is the concentration of competitor which produces 50% reduction of counts bound in a competitive radiobinding assay (see "Experimental Procedures")-is depicted as a function of amino groups modified.

TABLE I1
Arginine-specific modification of CT-B with 2,4-cyclohexanedione CT-B (1 mg/ml in 0.2 M borate buffer) was derivatized with various concentrations of cyclohexanedione (CHD) at pH between 9.0 and 10.5. Per cent arginine remaining was determined by amino acid analysis following hydrolysis conducted at 110 "C for 24 h in the presence of 1% mercaptoacetic acid.  (Table I). This finding is in concordance with Markel et ul. (19) who employed tetranitromethane.

DH
Modification of the Disulfide Bridge-Reduction and carboxymethylation was conducted with iodoacetamide or iodoacetic acid in guanidine HC1 after Crestfield et al. (35). The reaction was quantitative and did not affect other residues, as determined by amino acid analysis. Modification abolishes receptor binding activity (Fig. 8) and antigenicity ( Table I), suggesting that the disulfide bond is critical to the maintenance of conformation and function. In control experiments, exposure of CT-B to guanidine HC1 per se does not affect activity. Markel et aL (19) argue against the importance of this bond, finding that reduced and alkylated CT-B retains partial receptor binding function in a competitive radiobinding assay. However, their derivatized protein manifests less  (Table 11). Receptor Binding Activity (see "Experimental Procedures") is depicted as a function of arginine modified. (34) for the determination of surface exposed tyrosyl residues. The disulfide bridge was reduced and carboxymethylated with dithiothreitol and iodoacetic acid in guanidine HCl ( U ) after Crestfield et al. (35). Modified proteins were mixed with about 10,000 cpm of '2SI-labeled CT-B and allowed to compete for binding to hl (see "Experimental Procedures"). U , native CT-B. than 5% activity in absolute terms: IC, (native CT-B)/IC,, (modified CT-B)-see "Experimental Procedures." Traces of unreacted protein, whose presence is undetectable by amino acid analysis, could account for this finding.

DISCUSSION
The results of protein modification studies can be difficult to interpret. Chemical reagents seldom display absolute specificity for a single amino acid side chain. Even well described procedures yield atypical reactions due to the unique microenvironment of certain residues within native protein. Furthermore, the direct and local effects of derivatization are often indistinguishable from allosteric influence or the consequences of conformational disruption. This report presents an approach-employing reversible, quantifiable reactions, sensitive assays, and immunologic structural probes-which seeks to minimize the limitations inherent to protein modification investigations.
We describe an interesting monoclonal antibody (40B10) which competes with GM1 for binding to CT-B and enlisted the aid of this reagent to identify residues located near the receptor binding domain. We found that formylation of the single tryptophan residue in CT-B reversibly disfipts binding of both GM3 and 40B10 but preserves structural integrity, as monitored by a panel of conformation-sensitive mAbs. This result supports the contention of De Wolf et al. (21) that tryptophan participates directly in the receptor binding event. Lysine is seen to play a dual role. These residues appear to maintain an electrostatic attraction between toxin and receptor without which binding can not occur. In addition, at least 1 lysyl residue is within the mAb 40B10 epitope and may also interact directly with GMl. We repeated the arginine-specific modification performed by Duffy and Lai (20) but can not attribute the functional sequelae to derivatization of Arg5. Finally, our data suggest that a superficial tyrosyl residue is inessential, but the intramolecular disulfide bridge is vital to the structure and function of CT-B.
Ledley et al. (42) have identified a sequelae similarity between the B chain of CT and thyroid-stimulating hormone, follicle-stimulating hormone, luteinizing hormone, and human chorionic gonadotropin-all proteins which bind cell surface receptors. The region of similarity occurs around Cysg of CT-B. This homology, whatever its evolutionary basis, suggests that the conserved locus may be critical for function. Our data underscore the importance of several amino acids (Trpse and possibly Lys'l, L y P , and/or Lysgl) which reside near Cys". Thus, a model may be proposed wherein the intramoleular disulfide bond juxtaposes in space residues which jointly participate in receptor recognition.
A hypothesis by De Wolf et al. (17) proposes that the GM1 binding site resides at the interface of adjacent polypeptide chains of the B pentamer. This possibility was not acidressed here but could be examined with reconstitution studies, as demonstrated by h b e y and Schachman (43). Two different inactive derivatives (e.g. formylated and cyclohexanedionetreated CT-B) are mixed under conditions promoting random interchain association. Partial restoration of activity following renaturation (complementation) would provide evidence that the receptor binding domain is shared between monomers.
The protein modification studies described in this report have implicated various amino acid residues as critical to the structure and function of cholera toxin B subunit. These findings can be explored with site-directed mutagenesis, designed to alter specific residues without affecting secondary structure. A high resolution depiction of the binding interaction must await solution of three-dimensional structure by crystallography.