Immunochemical Analysis of the Structure of Diphtheria Toxin Shows All Three Domains Undergo Structural Changes at Low pH*

Diphtheria toxin is a bacterial protein that undergoes a physiologically critical conformational change at low pH. This change involves a partial unfolding event form- ing a molten globule-like structure, which exposes hydrophobic regions and which allows the toxin to insert into, and translocate across, membranes. In this report, antibody binding was used to examine the regions of the toxin that undergo structural changes at low pH. Monoclonal antibodies specific to the catalytic (C), transmem- brane (T), and receptor-binding (R) domains of diphtheria toxin were prepared and isolated. In addition, the binding of anti-peptide antibodies raised against pep-tides in the C and T domains to toxin was examined. Anti-C monoclonals and antipeptide antibodies were found to bind preferentially to low pH-treated toxin rel- ative to native toxin. Anti-T and anti-R monoclonal binding ranged between preference for native toxin and preference for low pH-treated toxin. These results suggest that the C domain becomes more exposed to solu- tion at low pH, and that both the T and R domains of the B chain undergo major conformational changes at low pH. Based on these results, a model in which low pH induces several coordinated changes in intra- and inter- domain interactions is suggested. The participation of the R domain in these changes is of particular signifi- cance because it suggests that the R domain plays a more important role in low pH-induced changes than previously realized.

Diphtheria toxin is secreted by Corynebacterium diphtheriae as a single polypeptide of 535 residues (M r 58,348) (1). The polypeptide chain is easily cleaved by proteases into the 193residue A chain (M r 21,167), which is located at the aminoterminal end of the protein, and the 342-residue B chain (M r 37,199) at the carboxyl-terminal end. The A chain's function is to catalyze the transfer of the ADP moiety of NAD ϩ to elongation factor-2. This inhibits protein synthesis and causes cell death (2). The B chain has at least two functions: to aid in translocation of the toxin across a lipid bilayer and in binding to the cellular receptor (2).
The crystal structure of diphtheria toxin at neutral pH shows that toxin comprises three domains: a catalytic (C), 1 a trans-membrane (T), and a receptor-binding (R) domain (3). The catalytic domain is identical to the A chain and largely comprises ␤-strands and short helices. The T domain is the aminoterminal half of the B chain and largely comprises long helices, several of which are hydrophobic. The R domain is the carboxyl-terminal half of the B chain and comprises largely ␤-strands.
The toxin first binds to the cell and then enters by receptormediated endocytosis (4). Once endocytosis has transported the toxin to endosomes, the acidic lumen of the endosomal vesicle triggers a conformational change in the toxin causing it to insert into the membrane bilayer (4). After insertion into the bilayer, the A chain translocates across the bilayer and is released into the cytoplasm where it catalyzes ADP-ribosylation.
In previous studies we have found that diphtheria toxin undergoes a conformation change to a hydrophobic, molten globule-like structure at low pH (5,6). When toxin is exposed to low pH at 23°C in the presence of lipid vesicles, only the B chain undergoes unfolding. However, when toxin is exposed to low pH at higher temperatures, both the A and B chain unfold (7).) 2 We chose anti-toxin antibodies to further examine diphtheria toxin structure in order to identify the regions of the toxin that undergo conformational changes at low pH. To do this, these antibodies were examined for their ability to bind to the different conformations of diphtheria toxin. The results suggest that all three domains undergo some type of structural change at low pH. Combined with the results of the following paper (39), this study suggests that all three domains may play direct roles in the membrane translocation step.

EXPERIMENTAL PROCEDURES
Materials-ImmunoPlate MaxiSorp, flat-bottom 96-well microtiter plates were purchased from Nunc. Goat anti-mouse Ig conjugated to alkaline phosphatase was purchased from Fisher Biotech. Goat antirabbit Ig conjugated to alkaline phosphatase was purchased from Southern Biochemical Associates (Birmingham, AL). Horse anti-diphtheria toxin Ig was purchased from Connaught Laboratories (Swiftwater, PA). Rabbit anti-horse Ig conjugated to alkaline phosphatase was purchased from Sigma. Gelatin was purchased from Difco. Sigma 104 phosphatase substrate (5-mg tablets), BCIP, and NBT were purchased from Sigma. PhastSystem SDS-polyacrylamide gel electrophoresis gels and supplies were obtained from Pharmacia Biotech Inc. Nitrocellulose membranes were purchased from Schleicher & Schuell. Low molecular weight standards were purchased by Electran (Poole, United Kingdom).
Immunization of Mice-Formalinized diphtheria toxin was prepared * This work was supported by National Institutes of Health Grant GM 31986. 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.
§ To whom correspondence should be addressed. 1 The abbreviations used are: C, catalytic; T, transmembrane; R, receptor-binding; ApUp, adenosine-3Ј-phosphate 5Ј-uridine phosphate; BCIP, 5-bromo-4-chloro-3-indolyl phosphate; ELISA, enzyme-linked immunosorbent assay; Ig, immunoglobulin; NBT, 2,2Ј-di-p-nitrophenyl-5,5Ј-diphenyl-3,3Ј-[3,3Ј-dimethoxy-4,4Ј-diphenylene] ditetrazolium chloride or nitro blue tetrazolium; PBS, phosphate-buffered saline; BSA, bovine serum albumin. 2 We have changed the names we established previously (7) for the conformations of toxin exposed to low pH and then pH neutralized from RЈ and RЉ to LrЈ and LrЉ (where L stands for low pH-treated, and r for reversed to neutral pH), respectively. This was done to avoid confusion between the R domain and R conformation. by incubating native toxin with 0.2% (v/v) formalin for 1 week at room temperature (8) or unformalinized free dimer toxin was prepared by incubating toxin at low pH and 41°C for 30 min. The toxin sample was mixed with the adjuvant RAS (Ribi Immunochemicals Research, Hamilton, MT), and injected into BALB/c mice every 2 weeks for 2 months. The hybridomas were produced from these mice by the Tissue Culture/Hybridoma Lab, Microbiology Department, SUNY, Stony Brook by standard procedures (9).
Screening Hybridomas-Supernatants from hybridomas were screened by ELISA and Western blotting. The ELISA protocol used was similar to that described below, except that the microtiter plate was incubated at 37°C and supernatants were not preincubated with toxin in solution. After preliminary experiments, six hybridomas (secreting antibodies we named anti-(␣)-C1, anti-(␣)-C2, anti-(␣)-T1, anti-(␣)-T2, anti-(␣)-R1, and anti-(␣)-R2) were chosen for further study and then subcloned by limited dilution in HyClone fetal bovine serum. All except ␣-C1 were derived from mice injected with the formalinized toxin.
Antibody Isotyping-Antibody class and subclass were determined using a mouse-hybridoma subtyping kit purchased from Boehringer Mannheim as per the manufacturer's instructions. ␣-C1 antibody was found to be a IgG 2b , and the other monoclonal antibodies chosen were IgG 1 with light chains.
Monoclonal Antibody Purification-The antibodies were purified from supernatants of subclones grown in fetal calf serum using the QUICKMab affinity resin (Sterogene Bioseparations Inc., Arcadia, CA). The supernatants were initially syringe-filtered and allowed to bind to about 1.5 ml of anti-mouse affinity resin. (Fresh resin was used for each antibody to prevent cross-contamination.) The antibodies were eluted from the column with 2 ml of ActiSep solution and then washed with a pH 4.5 buffer. Fractions containing antibody as detected by antibody blotting were pooled and concentrated to about 2 ml, followed by a dialysis against 1 liter of PBS (10 mM sodium phosphate, 150 mM NaCl, pH 7.2 (unless otherwise noted)) overnight at 4°C. Antibody purity was determined by SDS-polyacrylamide gel electrophoresis using the PhastSystem (Pharmacia). The only contaminant appeared to be some bovine serum albumin (BSA). For the worst cases, ␣-C1 and ␣-T2, there was 20% BSA contaminant, as judged by stain intensity. Antibody concentration was assayed with the bicinchoninic acid protein assay kit (Sigma). The approximate antibody concentrations after correction for BSA were 0.14 mg/ml for ␣-C1, 0.052 mg/ml for ␣-C2, 0.25 mg/ml for ␣-T1, 0.25 mg/ml for ␣-T2, 0.12 mg/ml for ␣-R1, and 0.16 mg/ml for ␣-R2. (The ␣-C2 value is only a rough estimate due to its low concentration.) Antibodies were stored at 4°C.
Diphtheria Toxin-Partially purified diphtheria toxin was obtained from Connaught Laboratories (Ontario, Canada). Toxin was further purified as described previously (7,11). Free monomer diphtheria toxin and toxin bound to ApUp was diluted in 5 mM Tris, 1 mM Na 2 EDTA, 0.02% NaN 3 , pH 7.0. The protein concentration of the stock solutions were determined spectroscopically (6). Isolated C domain was prepared as described previously (12). Isolated T domain and isolated R domain were kind gifts from the laboratory of Drs. R. John Collier, Hang-Jun Zhao, and Wei Hai.
Western Blotting Analysis-Free monomer toxin, isolated T domain, and isolated R domain were diluted to a final concentration of 0.1 mg/ml into 50 mM Tris-Cl, 1 mM Na 2 EDTA, 2.5% (w/v) SDS, 0.01% (w/v) bromphenol blue, pH 8.0, and 7.5% (v/v) 2-mercaptoethanol (loading buffer). 4 l of the sample was run on a gradient 10 -15% PhastGel, using a PhastSystem apparatus, and transferred to nitrocellulose using the transfer accessory. Blots were incubated with purified antibodies, and after washing, with goat anti-mouse Ig conjugated to alkaline phosphatase. Color development was performed using NBT and BCIP.
Competition ELISA Protocol-A 96-well microtiter plate was coated with 1 g of a 10 g/ml solution of toxin in the LrЈ conformation in 50 mM carbonate buffer, pH 9.6 (coating buffer), for 1 h at room temperature. (LrЈ toxin was prepared by incubating 80 g/ml of ApUp-bound monomer toxin in 20 mM acetate, 300 mM NaCl, pH 4.5, at room temperature for 30 min.) Control wells were prepared devoid of toxin. Wells were then washed at least three times with 200 l of 25 mM Tris-Cl, 150 mM NaCl, and 2.5 mM KCl, pH 7.2 (washing buffer). 200 l of a 1% (w/v) gelatin in PBS was added to each well and then incubated and washed as above. 100 l of a mixture of PBS-diluted antibody and toxin preincubated for 2 h was added to each well, incubated 1 h and washed as above. 100 l of goat anti-mouse Ig conjugated to alkaline phosphatase (diluted 1:1000 with 15 mM Tris-Cl, 150 mM NaCl, pH 7.2) was then added, and the sample was incubated and washed as above. Absorbance at 410 nm minus 490 nm was measured after at least 20 min of incubation with p-nitrophenyl phosphate using a Dynatech MR 600. Under our experimental conditions, color production was found to be linear in time, and nearly linear with the concentration of primary antibody added to the wells. Each sample was prepared in duplicate. The absorbance of the control well was always found to be negligible when compared to the absorbance of the sample wells. It should be cautioned that 50% inhibition of color production may not be equivalent to 50% inhibition of antibody binding to the wells because of the slight non-linearity of color with antibody concentration noted above, and the possibility that some IgGs bind toxin in solution at one antigen binding site and adhere to the toxin on the wells with the other site.
Dependence of Antibody Binding to Toxin as a Function of pH-3 l of 0.57 mg/ml free monomer toxin was incubated with 62 l of various pH buffers using 20 mM sodium acetate, 300 mM NaCl below pH 5 or 20 mM Na 2 HPO 4 , 300 mM NaCl above pH 5. Sample pH was reversed to pH 7.1-7.5 with 150 l of PBS and/or 200 mM Na 2 HPO 4 and incubated at room temperature for 15 min. An aliquot of each sample was added to PBS-diluted antibody and PBS to give a final volume of 320 l. The final toxin concentrations were 2.5 g/ml for ␣-C1, 0.75 g/ml for ␣-C2, 1.0 g/ml for ␣-T1, ␣-T2, and ␣-R1, and 1.5 g/ml for ␣-R2. The final antibody concentrations were 0.65 g/ml for ␣-C1, 0.98 g/ml for ␣-C2, 0.20 g/ml for ␣-T1, 0.77 g/ml for ␣-T2, 0.56 g/ml for ␣-R1, 0.25 g/ml for ␣-R2. A competition ELISA protocol was performed as described above on 100-l aliquots of these mixtures. Absorbance values in the presence of toxin divided by values in the control samples lacking toxin were calculated at each pH.
Similar experiments were performed with anti-peptide antibodies, with a final toxin concentration of 5.0 g/ml, and final antibody concentrations of 4.69 g/ml for ␣-C 141-157 and 9.38 g/ml for ␣-T 224 -237 . For antipeptides, absorbance values in the presence of toxin at each pH divided by the average of the values for the control samples (which were largely pH-independent) lacking toxin.
Dependence of Toxin Binding to Antibodies on Temperature-Samples of free monomer toxin (3.5 l of 0.57 mg/ml) were incubated with 35 l of 20 mM sodium acetate, 300 mM NaCl, pH 4.4, at various temperatures for 30 min, then neutralized by the addition of 214 l of thermally equilibrated PBS and incubated for an additional 15 min. The samples were then incubated at room temperature for 5-10 min and diluted to 320 l with PBS and antibody as above. The final toxin concentration was 2 g/ml. The final antibody concentrations were 0.65 g/ml for ␣-C1 and 0.20 g/ml for ␣-T1. A competition ELISA protocol was then performed as described above.
Dependence of Binding to Antibodies on Toxin Concentration-Toxin was prepared in the native (N) and low pH-treated (LrЈ and LrЉ) conformations. The LrЈ and LrЉ conformations were prepared by incubating 2.2 l of 0.93 mg/ml free monomer toxin with 35 l of a 20 mM acetate, 300 mM NaCl, pH 4.4, at 23°C (LrЈ conformation) or 37°C (LrЉ conformation) for 30 min. Samples were neutralized with 215 l of PBS at the same temperature and incubated for 15 min. Native toxin was prepared by initially mixing 35 l of 20 mM acetate, 300 mM NaCl, pH 4.4, with 215 l of PBS and then adding 2.2 l of 0.93 mg/ml free monomer toxin. Various aliquots of each sample were then diluted to 100 l with a 1:5.8 (v/v) mixture of 20 mM, sodium acetate, 300 mM NaCl, pH 4.4, and PBS and then to 320 l with PBS and antibody as above. Final antibody concentrations were the same as in the versus pH section. The antibody binding was assayed by competition ELISA as described above. Similar experiments were performed with anti-peptide antibodies, except that different toxin concentrations and buffer volumes were used. Again final antibody concentrations were as in the versus pH section. The concentration of toxin (g/ml) that yields 50% inhibition (IC 50 ) in Table  I was obtained from the graph of percent absorbance of p-nitrophenyl versus concentration of toxin, extrapolating when necessary.
Trypsin Digestion of Native Toxin-10 l of 1.46 mg/ml free dimer toxin was incubated with 10 l of a 56 g/ml fresh trypsin solution (dissolved in PBS) for 17 min at room temperature. The trypsin-free control contained 10 l of dimer toxin and 10 l of PBS. 2.5 l of loading buffer was added to a 10-l aliquot of the reaction mixture or a 10 l aliquot of the control sample. 1 l of each sample was run on a high density SDS gel using the PhastSystem apparatus and silver-stained or transferred to a nitrocellulose membrane and developed with ␣-C1, ␣-C2, or ␣-C 141-157 .
Antibody Binding to Isolated Domains-The binding of antibodies to isolated toxin domains was measured using the competition ELISA (see above). ␣-C1 binding to isolated C domain, whole native toxin, and whole toxin in the LrЈ conformation was compared. Isolated native C domain was prepared by mixing 2 l of 1.29 mg/ml C domain and 249 l of PBS. The LrЈ conformation toxin was prepared by mixing 3.5 l of ApUp-bound monomer with 35 l of 20 mM sodium acetate, 300 mM NaCl, pH 4.4, for 30 min at room temperature, and then neutralizing with 214 l of PBS. The same acetate and PBS were premixed and then added to the toxin to obtain native toxin. Aliquots were diluted to 120 l with the same mix of buffers and then to 320 l with PBS and antibody as above. Final antibody concentration was 0.48 g/ml.
␣-T 224 -237 binding to isolated T domain in the native and LrЈ conformation was compared. The LrЈ conformation was prepared by adding 5 l of 1.4 mg/ml T domain to 87.5 l of 20 mM sodium acetate, 300 mM NaCl, pH 4.4, for 30 min at room temperature and then neutralized with 257 l of PBS. The same acetate and PBS were premixed and then added to the T domain to obtain native T domain. Aliquots were diluted to 225 l with the same mix of buffers and then to 320 l with PBS and antibody as above. Final antibody concentration was 4.7 g/ml.
␣-R2 binding to isolated R domain in the native and LrЈ conformation was compared. The LrЈ conformation was prepared by adding 27.5 l of 0.2 mg/ml R domain to 110 l of 20 mM sodium acetate, 300 mM NaCl, pH 4.4, for 30 min at room temperature and then neutralized with 230 l of 1:3.6 (v/v) mixture of 200 mM Na 2 HPO 4 and PBS. The same acetate and PBS were premixed and then added to the R domain to obtain native R domain. Aliquots of both preparations were then made up to 210 l with the same mix of buffers and then diluted to 220 l with PBS and antibody as above. Final antibody concentration was 0.045 g/ml.

RESULTS
Epitope Mapping-Western blotting was used to epitope map monoclonal antibody binding sites on toxin molecules using reduced toxin (which gives bands for the A and B chain as well as whole unnicked toxin), isolated T domain, and isolated R domain (Fig. 1). Three types of antibodies were found: anti-A chain antibodies, which bound to whole toxin and A chain (C domain); anti-T domain antibodies, which bound whole toxin, B chain, and isolated T domain; and anti-R domain antibodies, which bound to whole toxin, B chain and isolated R domain. Two antibodies against each domain (named ␣-C1, ␣-C2, ␣-T1, ␣-T2, ␣-R1, and ␣-R2) were chosen for further study. In addition, the binding of the previously prepared (10) anti-peptide antibodies ␣-C 141-157 and ␣-T 224 -237 to toxin was studied. 3 Higher resolution epitope mapping was performed with trypsin-cleaved toxin. Incubating native toxin with trypsin results in the cleavage of the C domain between Lys-39 and Ser-40 (12). Western blotting of trypsin-cleaved toxin showed monoclonal anti-C domain antibodies did not bind to the 40 -193 fragment (Fig. 2), although as expected ␣-C 141-157 did bind. This suggests part of the epitope for ␣-C1 and ␣-C2 lies on residues 1-39, although not all of the epitope may be within this region.
Epitopes were also mapped with cysteine cleavage reagent 2-nitro-5-thiocyanobenzoic acid. Antibody binding to cleavage products of mutant toxins in which a single cysteine was substituted for either Glu-162 or Ser-337 was examined. Anti-C antibodies bound to the 1-161 fragment, but not to the 162-535 fragment, and anti-T domain antibodies bound to the 1-336 fragment but not the 337-565 fragment (data not shown). This suggests that the anti-C domain antibodies bind between residues 1 and 161, and the anti-T domain antibodies bind between residues 194 and 336.
Additional data on epitope locations comes from the results of antibody binding to octapeptides that cover the entire sequence of the B chain (10). The ␣-R1 antibody binding profile shows an affinity for octapeptides with sequences identical to the loop region between the T and R domain and residues 486 -500 (RB8) (data not shown). The binding profile of ␣-R2, together with hydroxylamine cleavage patterns (data not shown), suggests ␣-R2 binds residues 454 -465 (RB6).
The Effect of pH-induced Changes in Toxin Conformation on Antibody Binding: Competition ELISA-Diphtheria toxin undergoes a conformational change at pH 5.3 that plays a critical role in its entry into cells (4) (see Introduction). Above pH 5 the toxin remains in the native (N) state, but after exposure to low pH at 23°C, the toxin undergoes a partial unfolding process that results in its taking on a hydrophobic, membrane-inserting conformation (5)(6)(7)(13)(14)(15)(16)(17)(18)(19)(20). In this conformation, the B chain has undergone a partial unfolding process but the C domain remains folded (7). In order to further characterize the changes that the toxin undergoes at low pH, the pH dependence of antibody binding to toxin was studied. As with many antibodies, anti-toxin binding at low pH was found to be too weak to measure. Therefore, binding was examined after low pH was reversed to neutral. Under these conditions previous studies have shown the toxin maintains the structural changes that occur at low pH (7). (This low pH-treated state is called the LrЈ conformation. 2

)
To measure the dependence of antibody binding on the pH at which toxin is incubated, toxin was preincubated at various pH values in solution and its antibody binding measured by a competition ELISA after pH neutralization. In this assay the amount of antibody binding to the toxin in solution is measured by its inhibition of antibody binding to toxin-coated wells (21). The amount of antibody bound to the wells is assayed by the amount of p-nitrophenyl absorbance generated after incubation 3 ␣-C 141-157 was generated against a peptide corresponding to residues 141-157 of the toxin sequence, but mainly binds to residues 145-149. ␣-T 224 -237 was generated against a peptide corresponding to residues 224 -237 of the toxin sequence, but mainly binds to residues 231-236 (10). of the wells with an alkaline phosphate conjugated to an antimouse or anti-rabbit Ig and addition of p-nitrophenyl phosphate. The more antibody bound to the toxin in solution, the less binds to the wells, and the lower the absorbance value measured. 4 The degree of inhibition of antibody binding to ELISA plates by toxin in solution was quantitated through the decrease in ELISA absorbance in the presence of toxin relative to the controls lacking toxin.
The Effect of pH-induced Changes in Toxin Conformation on Antibody Binding: Anti-C Domain Antibodies- Fig. 3A illus-trates effect of the pH at which toxin is incubated upon anti-C binding. ␣-C1, ␣-C2, and ␣-C 141-157 behave similarly in that their binding is stronger to low pH-treated toxin, with a sharp transition in binding near pH 5, the value at which the low pH-induced change in toxin conformation has been shown to occur (5,6,20). Antibody binding to toxin can be quantified by the comparison of the concentration of toxin sufficient to inhibit the ELISA color reaction by 50% (IC 50 ) (Figs. 4, A and B, and  5A). The IC 50 values derived from these experiments are given in Table I and show the difference in binding to low pH-treated (LrЈ conformation) and native toxin is greatest in the case of ␣-C1 (about 30-fold), but is also significant for the other anti-C antibodies (about 2.5-fold).
The Effect of pH-induced Changes in Toxin Conformation on Antibody Binding: Anti-T Domain Antibodies-Similar experiments were performed with anti-T domain antibodies. Fig. 3B illustrates the effect of pH at which toxin is incubated upon anti-T binding. ␣-T1, ␣-T2, and ␣-T 224 -237 each react differently with native toxin and low pH-treated toxin. ␣-T1 shows no significant dependence on the pH at which toxin was preincubated. This is confirmed by the observation that IC 50 s for native and low pH-treated (LrЈ) toxin binding by ␣-T1 are the  Table I). ␣-T2 antibody binds to native toxin more strongly than low pH-treated toxin with about a 7-fold preference for native toxin (Fig. 4D and Table I). Again, there is a transition at pH 5 between the strongly and weakly bound toxin conformations. In contrast, ␣-T 224 -237 binds to the low pH-treated toxin more strongly (13-fold) than to native toxin ( Fig. 5B and Table I), although again with a transition at pH 5 (Fig. 3B).
The Effect of pH-induced Changes in Toxin Conformation on Antibody Binding: Anti-R Domain Antibodies- Fig. 3C illustrates the effect of pH at which toxin is incubated upon anti-R binding. pH has an opposite effect on ␣-R1 and ␣-R2 binding to toxin. ␣-R1 binds native toxin more strongly than low pHtreated (LrЈ) toxin (Fig. 4E), with a 5-fold preference for native toxin (Table I). ␣-R2 binds low pH-treated toxin more strongly than native toxin (Fig. 4F), with a 2.2-fold preference for the former ( Table I).
The Effect of pH-induced Changes in Toxin Conformation on Antibody Binding: Binding to Isolated Domains-Binding of several antibodies to isolated domains was also measured. Isolated T and R domains appear to take on native and low pH-induced conformations similar to those they have in whole toxin (22,23).) 5 Therefore, antibody binding to native and low pH-treated domains could be studied. The isolated domain binding of ␣-T and ␣-R antibodies that prefer binding the low pH-treated whole toxin was measured to see if the increased binding at low pH is due to increased exposure of their epitopes due to a loss of interdomain interactions, or due to changes within the domain. Fig. 6 shows ␣-T 224 -237 binding to isolated T domain and ␣-R2 binding to the R domain have the preferential binding to low pH-treated protein seen in whole toxin, supporting the latter proposal.
Isolated C domain takes on a conformation similar to that in whole toxin (24,25) and undergoes only reversible changes at low pH, and thus can only be studied in the native state. ␣-C1 binding to isolated native C domain was also measured and found to be much more similar to the strong binding of C domain in low pH-treated whole toxin rather than the weaker binding of C domain in native whole toxin (Fig. 6).
The Effect of Thermally Induced Conformational Changes on Antibody Binding to Toxin-Above 26°C at low pH (4.5) diphtheria toxin undergoes a change in conformation to a structure in which, in addition to the partial unfolding seen in the B chain at low pH, the C domain partly unfolds (7). After pH neutralization, this conformation is called the LrЉ conformation. The relative role in translocation of the LrЉ toxin and the above-studied LrЈ toxin, which predominates at lower temperatures, remains unresolved (26). Therefore, it was of interest to examine the difference between these conformations by antibody binding.
The competition ELISA was used to examine the binding of ␣-C1 and ␣-T1 to toxin preincubated at various temperatures (Fig. 7). A transition is seen close to 26°C for both antibodies, showing that antibody binding responds to the LrЈ to LrЉ conformational change.
An ELISA competition assay was then performed with all of the antibodies using toxin in the LrЉ conformation. The IC 50 values obtained are given in Table I and show that all the antibodies reacted 2-9-fold more strongly with the LrЈ toxin than with the LrЉ toxin. We attribute this partly to steric blocking arising from aggregation of the toxin hiding epitopes in the LrЉ conformation (see "Discussion").

Using Monoclonal Antibodies to Evaluate Diphtheria Toxin
Structure-Binding of anti-diphtheria toxin monoclonal antibodies to the toxin has previously been studied in other laboratories (27)(28)(29)(30)(31)(32)(33) which have found that monoclonal antibodies that bind to various sites on the toxin can be obtained. The studies of Zucker and Murphy (29,30) tried to designate function to a particular area of the toxin by examining which functions were blocked by antibodies, but did not examine the effect of pH on toxin behavior.
A study by Rolf and Eidels (33) did identify some monoclonals that bind to low pH-treated toxin, but the number of pH values chosen for binding studies were insufficient to determine whether the pH-dependent change in binding they observed corresponded to the conformational transition at pH 5.  b These numbers can be used to calculate toxin/bound antibody ratios at IC 50 by approximating that half the antibody molecules are bound to toxin at IC 50 . Antibody concentrations given under "Experimental Procedures" were converted to nM assuming a molecular mass of 150 kDa. The toxin/bound antibody (molecule/molecule) ratios for toxin in the tightest binding conformation (lowest IC 50 ) are 7/1 for ␣-C1, 1/1 for ␣-C2, 26/1 for ␣-T1, 2/1 for ␣-T2, 2/1 for ␣-R1, 13/1 for ␣-R2, 14/1 for ␣-C 131-147 , and 0.7/1 for ␣-T 224 -237 . A ratio of 2/1 suggests tight and stoichiometric toxin binding to each antibody site. A ratio of 1/1 suggests tight, stoichiometric binding to antibodies where blocking of one site on the antibody prevents binding to the ELISA wells. The antibodies with ratios Ͼ2/1 require excess toxin to saturate binding sites on the antibodies and prevent binding to wells. Furthermore, the dependence of binding on pH was weak, perhaps reflecting the sensitivity of the immunoprecipitation method used.
In this study, antibodies to each domain of the toxin have been identified. These studies show that the binding of monoclonal and antipeptide antibodies can respond strongly to conformational changes in diphtheria toxin. Such antibodies should be useful tools for establishing whether toxin mutants fold and undergo conformational changes similar to those in whole toxin.
Just as important, these studies provide some new insights into the mechanism of diphtheria toxin insertion into membranes. One central observation in this study is that changes in all three domains occur at low pH. One obvious question is what changes are occurring in each of the domains? This is complicated to answer because several factors can influence antibody binding. A difference in antibody binding to two conformations could result from a difference in exposure of an epitope, or a change of an epitope into a conformation that is more weakly or tightly bound. Furthermore, a change in exposure of an epitope could be due to a change in the position or conformation of a sequence of a neighboring domain, or intermolecular aggregation.
Low pH-induced Changes in C, T, and R Domain Structure in the LrЈ Conformation-Despite these difficulties, the changes in antibody binding to toxin do provide some interesting details to what is occurring at low pH. Let us first consider the behavior of the toxin in the LrЈ conformation. Previous studies have shown that although there is some partial unfolding of the B chain in this conformation the C domain (A chain) remains folded (7). Therefore, the increased binding to low pH-treated toxin found for all three anti-C antibodies examined indicates there is increased exposure of the epitopes on the C domain to solution after low pH treatment. This suggests that the degree of contact between the C domain and the B chain (T plus R domains) must decrease at low pH. A related possibility is that the C domain undergoes a small conformational change that promotes antibody binding. In fact, a small conformational change in residues 66 -78 and 169 -176 has been seen in isolated C domain relative to whole toxin by crystallography (25), and is believed to be due to the loss of C domain interactions with the T domain. It is possible that our monoclonal antibodies bind C domain in whole toxin exposed to low pH more tightly due to a similar loss of C domain-T domain interactions. However, a more general increase in exposure at low pH is also very likely because the ␣-C 141-157 , which does not involve the residues that change structure upon loss of T domain interaction, also binds more tightly at low pH. This conclusion is consistent with evidence that the interactions between the C and R domain are lost at low pH (34).
Additional changes occur in the T and R domain. The observation that the differences between the binding of some antibodies to these domains in the native and low pH-treated states are seen with isolated domains as well as whole toxin indicate that they are not solely due to changes in interdomain interactions, and probably reflect conformational changes within each of these domains at low pH. It is likely that there are conformational changes in these domains due to partial unfolding and exposing hydrophobic surfaces at low pH (12,22). This conformational change would decrease the recognition to low pH-treated toxin of antibodies such as ␣-R1 because the epitope would become hidden or change conformation. The same conformational change could cause increase antibody binding to low pH-treated toxin for antibodies such as ␣-T 224 -237 and ␣-R2 due to increased exposure or due to cooperative binding to low pH induced toxin aligomers (although see below).
Toxin Behavior in the LrЉ Conformation-For all of the antibodies tested, it appeared that binding to toxin in the LrЉ conformation was less than binding to the LrЈ conformation. It is likely that in solution strong aggregation of the toxin in the LrЉ conformation generally decreases antibody binding to this conformation of the toxin. Previous studies have shown that the hydrophobic behavior of diphtheria toxin at low pH induces its aggregation in solution (6,35). The observation that membrane insertion, which should eliminate the presence of aggregates held together by hydrophobic interactions, greatly increases antibody binding to the LrЉ conformation (see accompanying paper (39)) supports the proposal that aggregation in solution influences antibody binding. In contrast, aggregation does not seem to be a major influence on antibody binding to toxin in the LrЈ conformation because the antibody binding to LrЈ toxin in solution and inserted in model membranes is generally similar (compare Table I of the present FIG. 6. Assay of antibody binding to isolated toxin domains using competition ELISA. Panel A contains native whole toxin (q), LrЈ whole toxin (E), and isolated native C domain (ϩ). Panel B contains isolated native (q) and LrЈ (E) T domain. Panel C contains isolated native (q) and LrЈ (E) R domain. See Fig. 4 for details. Antibodies used were: A, ␣-C1; B, ␣-T 224 -237 ; and C, ␣-R2.

FIG. 7.
Assay of antibody binding to toxin incubated at various temperatures using competition ELISA. The inhibition of antibody binding to ELISA plates obtained by preincubating antibodies with toxin (treated at pH 4.5 and various temperatures) in solution was measured through the decrease in absorbance. The absorbance relative to that in the absence of toxin in solution (defined as 100%) is shown. The antibodies used were ␣-C1 (q), and ␣-T1 (E). report to Table II in Ref. 39). 6 A Model for the Changes in Diphtheria Toxin at Low pH and Implications for Membrane Insertion-Based on the changes observed in diphtheria toxin at low pH, a schematic model summarizing the changes in toxin structure leading to membrane insertion can be proposed. The first stage would involve the loss of interactions between C and R domains, proposed by Bennett and Eisenberg (34). The next step would involve the major conformational changes observed in both the T and R domain, greatly increasing their hydrophobicity (Fig. 8). These conformational changes suggest that interactions between all three domains would break or become much weaker. This would also result in sites on all three domains becoming exposed to the aqueous environment after treatment with low pH, which could aid in membrane insertion. The unfolding of the C domain into a hydrophobic conformation, which requires slightly more extreme conditions, may also occur at this time. Thus a number of coordinated changes may play a role in the membrane insertion process.
The observation of low pH-induced conformational changes in the R domain is important because it implies that the R domain plays an important role in insertion, rather than just functioning in binding to the cellular receptor (36 -38). This is reinforced by the recent observation that the R domain becomes hydrophobic at low pH (23), and the results in the accompanying report, showing that the R domain inserts into model membranes in a fashion such that some sites translocate across the bilayer. 6 Comparison of Table I in this report and Table II in the accompanying paper (39) generally shows a similar IC 50 for LrЈ conformation toxin in solution compared to LrЈ toxin externally added to vesicles for several monoclonal antibodies. Comparison of results with and without lipid vesicles present is valid because the same monoclonal antibody concentrations were used in the solution and membrane-inserted toxin experiments, and other conditions were very similar except for a slight variation in salts used.
FIG. 8. Schematic figure of diphtheria toxin conformational changes at low pH. A change in conformation is shown by a change in outline shape. The figure shows that in the low pH-treated LrЈ conformation, there is an increase in exposure of sites to solution as well as a change in T and R domain conformations. A small conformational change seen in isolated C domain may also occur (25). In the LrЉ conformation, an additional unfolding of the C domain conformation occurs, as shown in previous studies (7).