Biochimica et Biophysica Acta, 1159 (1992) 169-178 © 1992 Elsevier Science Publishers B.V. All rights reserved 0167-4838/92/$05.00
A study of hydrogen exchange of monoclonal antibodies: Specificity of the antigen-binding induced conformational stabilization Paola Rizzo, Caterina Tinello, Antonello Punturieri and Hiroshi Taniuchi Laboratot3, of Chemical Biolo~', National b~stitutes of Diabetes and Digestive and Kidney Diseases, National b~stitutes of Health, Bethesda, MD (USA) (Received 21 February 1992)
Key words: Hydrogen exchange; Monoclonal antibody; Fab fragment; Antigen induced stabilization; Antigen recognition
Amide-hydrogen exchange of three anti-yeast iso-l-cytochrome-c IgG monoclonal antibodies and the Fab, prepared from one of them, were studied by infrared spectrophotometry in the presence and absence of the deuterated immunogen and evolutionarily related species (the deuterated immunogen contained a population of a dimer. Each subunit of the dimer appeared to bind to the antibodies in a manner similar to the monomer). The number of hydrogens of the antibodies whose exchange was suppressed on binding to the immunogen was found to exceed that estimated for the residues shielded by the immunogen. Analysis of the data suggests that such suppression of hydrogen exchange occurs mainly for the Fab domains, but not for the Fc. One of the antibodies showed two distinct classes of amide-hydrogens. Class-1 hydrogens (approx. 36/site) exchange faster than class 2 (approx. 37/site). The exchange of class-1 hydrogens was suppressed by binding to the immunogen, but not to the evolutionarily related species. The exchange of class-2 hydrogens was suppressed by binding to the evolutionarily related species, as well as to the immunogen. Thus, the suppression of exchange of class-1 hydrogens appears to occur by some kind of conformational stabilization, the mechanism of which differentiates between the deutcrated immunogen and the evolutionarily related species. Evidence suggests that the trans-interactions of the Fab domains may modulate the hydrogen exchange. If it is assumed that the antigen-binding strengthens the trans-interactions in such a way that the exchange of the slower exchanging hydrogens is suppressed, this could explain the suppression of exchange of class-2 hydrogens.
Introduction Liberti and colleagues  have shown, in their hydrogen exchange studies of F(ab') 2 fragments of sheep and rabbit antibodies, that ligand binding not only shields the residues at the binding site, but also stabilizes a structural region of the F(ab') 2 away from the binding site. Zavodszky and colleagues  have concluded, based on amide-hydrogen exchange measurements of rabbit antibodies, that antigen-binding increases the conformational stability of antibodies. If a
Correspondence to: H. Taniuchi, Laboratory of Chemical Biology, National Institutes of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892, USA. Abbreviations: mAb, monoclonal antibody; V L and CL, variable and constant domains, respectively, of the light chain of immunoglobulins; V u, CHI, CH2, and CH3 variable and the 1st, 2nd and 3rd constant domains, respectively, of the heavy chain; Fab, the antigenbinding fragment containing V L, C L, V H and CH1; Fc, the fragment containing 2 CH2 and 2 CH3; ELISA, enzyme-linked immunosorbent assay.
long-range interaction between the antigen-antibody interface and the antibody domains is assumed, such decrease in energy of the antibody conformation should, in turn, increase the affinity to the antigen. Thus, we have thought (cf., Ref. 3) that the antigen-antibody interaction may not be confined to the antigenantibody interface and that it might spread to distant regions such as those interior of the V e and V H. In relation to this idea, the studies of Scharff and colleagues are interesting [4,5]. They have shown that substitution of a residue of a variable region of antibodies which is remote from the antigen-binding site abolishes the antigen-binding ability. We have thought that this might be another aspect of the phenomenon in which distant regions influence the antigen-antibody interaction. In light of this view, to know more about the structural regions of antibodies which are stabilized on antigen-binding, we have studied amide-hydrogen exchange of monoclonal IgG antibodies to yeast iso-l-cytochrome c 2-96-12, 4-74-6 and 4-128-6  and the Fab fragment of mAb 4-74-6 in the presence and absence
170 of the deuterated immunogen (yeast iso-l-cytochrome c) and evolutionarily related species (the deuterated immunogen contained a population of a dimeric form. The subunits of the dimer appeared to bind to the monoclonal antibodies in a manner similar to the monomer). For monoclonal antibodies, different rates of hydrogen exchange represent different residues or different groups of residues. On the other hand, for polyclonal antibodies, homologous residues or homologous groups of residues of different populations could show different rates of hydrogen exchange. Therefore, monoclonal antibodies may exhibit changes of hydrogen exchange rates on antigen-binding which are difficult to detect for polyclonal antibodies. Monoclonal antibody 2-96-12 has an epitope which contains Asp-65 of yeast iso-l-cytochrome c and crossreacts with yeast iso-2, Candida, tuna, pigeon, chicken, rabbit, rat, dog, bovine and horse cytochromes c . Monoclonal antibody 4-74-6 shows an epitope which contains Leu-63 a n d / o r Ash-67 a n d / o r Asn-68 of yeast-iso-l-cytochrome c and cross-reacts with none of the above evolutionary related species . Monoclonal antibody 4-128-6 possesses an epitope which contains Glu-93 of yeast-iso-l-cytochrome c and cross-reacts with yeast iso-2-cytochrome c. The cross-reaction of this monoclonal antibody with Candida, tuna and horse cytochromes c was marginal or undetectable . The cross-reaction with yeast iso-l-apocytochrome c was undetectable for mAb 2-96-12 and 4-128-6 and very weak for 4-74-6 . In this report, we show that the number of amidehydrogens of mAb 4-74-6, whose exchange rate decreases on immunogen-binding, is significantly greater than an estimated number if shielding of the antigenbinding site were solely responsible for the decrease in the exchange rate. We also show that the total number of hydrogens of mAb 4-74-6 whose exchange is suppressed is essentially accounted for by the number of hydrogens of the Fab whose exchange is suppressed on binding to the immunogen. Then, we show with mAb 2-96-12 that there are two distinct classes of amide-hydrogens whose exchange is suppressed on binding to the immunogen. These two classes of hydrogens, one exchanging faster than the other, respond completely differently to binding to the evolutionarily related species in terms of a decrease of the exchange rate or lack of it. A preliminary account of this work has been published elsewhere . Materials and Methods
Preparation and purification of monoclonal antibodies Cells of each of hybridoma cell lines 4-74-6, 4-128-6 and 2-96-12 were grown to log phase as previously described . Ascites of B a l b / c female mice containing monoclonal antibodies were prepared according to the protocol of Goding  using pristane treatment and intraperitoneal injection of the hybridoma cells. The supernatant solution of ascites was loaded on an affinity column (0.8 × 9.3 cm) containing yeast iso1-cytochrome c coupled to CH-Sepharose 4B (LKBPharmacia) at 8°C (a maximal capacity, approx. 20 mg of the monoclonal antibodies). The column had been equilibrated with phosphate buffered saline (6.66 mM, pH 7.4). After washing with the phosphate buffered saline, the column was eluted using 0.1 M glycine-HCI (pH 2.5). The eluent containing the monoclonal antibodies was immediately neutralized by adding 1 M Tris-HCl (pH 8.5) (final concentration, 0.119 M) and then lyophilized. The monoclonal samples thus obtained were homogenous based on polyacrylamide gel electrophoresis .
Preparation of the Fab fragments The methods of Good et al. , Harlow and Lane  and Andrew and Titus  were used to construct a protocol. Briefly, a sample of purified mAb 4-74-6 was dialyzed against 0.1 M sodium phosphate (pH 7.5, 8°C, 19-20 h) and then concentrated to approx. 5.1 mg protein/ml (based on absorbance at 280 nm, Ref. 13) using Centricon 30 tubes (Amicon). To 4 ml of the concentrated solution 160 ~1 of 20 mM EDTA, 160 p~l of 1 M cysteine and 9 #1 of 31 m g / m l papain (Worthington Biochemical) were added. After incubation at 37°C for 2 h, 480 ~1 of 0.5 M iodoacetamide was added. After 15 min, the mixture was subjected to affinity chromatography (see above) to separate the Fab from the Fc (the yield of the Fab was 12.2 mg). To ensure complete removal of undigested monoclonal antibodies, chromatography of the Fab sample was carried out on a Sephacryl 5-200 (LKB- Pharmacia) column (4 × 120 cm) at 8°C using 0.1 M sodium phosphate (pH 7.01). Electrophoresis showed that the sample thus purified was homogenous (Fig. 1). The exception is that a small amount of smaller molecular weight materials than the Fab are seen in the pattern (Fig. 1). Since they were bound to the affinity column, they were likely to contain the antigen-binding site and, therefore, not to contain the Fc. Thus, this contamination was ignored.
Preparation of deuterated cytochromes c Materials Samples of yeast iso-1 (Saccharomyces cereuisiae), horse, tuna and bovine cytochromes c from Sigma were purified as previously described .
Yeast iso-1, horse, tuna, and bovine cytochromes c were fully deuterated, so that their hydrogen exchange was transparent in the measurement described below. The samples were deuterated by incubation with DzO
171 (Fluka) in the presence of 3 M guanidine HCI (cf., Ref. 14) at approx, p D 5 for 1 h at 23°C. This was followed by gel-filtration, using Sephadex G-25 M-pD-10 columns (LKB-Pharmacia) and D20. Aliquots of this desalted solution, each containing a suitable amount (e.g., 0.44 mg) of the deuterated samples, were lyophilized. Complete deuteration of the samples was confirmed by the observations that the ratio of amide II absorbance to amide I (see below) did not change after incubation with D 2 0 for 24 h (the ratio ranged from 0.109 to 0.179 depending on the cytochrome c species). Based on the ELISA , the affinity of the deuterated yeast iso-l-cytochrome c sample was found to be equal to the nondeuterated sample for mAb 4-74-6 and greater, and slightly greater than the non-deuterated species for mAb 2-96-12 and 4-128-6, respectively, using the purified monoclonal samples. The mAb 4-74-6 and 4-128-6 samples cross-reacted with neither deuterated nor non-deuterated horse cytochrome c . The
0 X v
5 (..) 111 ._J
18.4 14.3 Fig. 1. T h e Fab sample prepared from m A b 4-74-6 is shown by electrophoresis on 8% polyacrylamide gel in the SDS-Tris buffer system based on the protocol of Laemmli . Lane 1, mAb 4-74-6 digested with papaim lane 2, molecular weight markers (BRL); lane 3, the Fab purified by affinity-chromatography, followed by gel-filtration on Sephacryl S-200. The details are described in Materials and Methods. Proteins and fragments were stained with Coomassie blue .
mAb 2-96-12 sample cross-reacted with tuna and horse cytochromes c regardless of whether the cytochromes c were deuterated or not . The deuterated horse cytochrome c sample was homogenous by get filtration on Sephadex 50 F (Pharmacia) and exhibited a 695 nm absorption band with the same absorptivity as that of the non-deuterated sample. This band is indicative of the Met-80S-heme Fe 3÷ bond . The deuterated tuna and bovine cytochrome c samples exhibited 97 and 88%, respectively, of the absorptivity of the 695 nm band of the non-deuterated samples. These two deuterated samples showed a blue shift by approx. 1 nm or less of a )tma x of a Soret band. The tuna sample also showed a weak absorption band around 650 nm (cf., Ref. 17). The deuterated yeast iso-l-cytochrome c sample showed a decrease in absorptivity, by approx. 42%, of the 695 nm band, a significant extinction of an absorption band around 650 nm and a blue shift by approx. 1 nm of a ,h.max of a Soret band. The sample apparently contained approx. 40% of a dimer (based on the amount of monomer), presumably consisting of two monomers linked by an inter-chain disulfide bond involving Cys107. The dimer was identified by comparison of the gel filtration patterns of the deuterated and non-deuterated samples with that reported by Bryant et al.  for a disulfide dimer of yeast iso-l-cytochrome c. The latter disulfide dimer has been shown  to retain the 3-D structure resembling the monomer. The deuterated yeast iso-l-cytochrome c sample was used without removal of the dimer. This is justified for the following reason: Based on the report of Bryant et al.  and the result of ELISA described above, the present dimer is likely to retain the 3-D structure resembling the monomer. The optical properties of the deuterated sample described above indicate that the heme crevice of the dimer is perturbed (cf., Ref. 19). However, these optical properties are very similar to those for one of the substituted yeast iso-l-cytochrome c studied by Ramdas et al. , that contained Thr in place of Pro-76. Thus, by analogy to the mutant , the overall conformational perturbation of the dimer is assumed not to be large * The epitope of each of the three monoclonal antibodies used is spatially removed from Cys-107 . Therefore, the monomer subunits of the dimer are likely not to significantly interfere with each other for binding to any one of the monoclonal antibodies used. This hypothesis is supported by the ELISA, showing that the affinity of the deuterated yeast iso-l-cytochrome c sample to any one of the three monoclonal
* It is likely that the m o n o m e r population also contained a perturbed heme crevice. However. based on a similar line of reasoning the overall conformational perturbation should be small.
172 antibodies is not weaker than the non-deuterated sample. Furthermore, mAb 4-74-6 was apparently saturated with the deuterated yeast iso-l-cytochrome c sample which bound in a 1 : 1 molar ratio to the antigen-binding site, ignoring the dimer population (see below). The concentration of deuterated yeast iso-l-cytochrome c is based on the sum of the monomer and the subunit.
Hydrogen-deuterium exchange Amide-hydrogen-deuterium exchange was measured by infrared s p e c t r o p h o t o m e t r y [20-24], using a Perkin-Elmer Model 1420 Ratio Recording Infrared Spectrophotometer and CaF 2 cells of 0.1 mm path length (Spectra-Tech). The purified monoclonal samples (see above) were dialyzed against 12.5 mM NaC1 at 8°C for 17 h. After concentration, the dialyzed solution was divided into aliquots, each containing a suitable amount (e.g., 2.82 mg) of protein and 5.5/xmol of NaCI. To each of these a[iquots, placed in microtubes, 110 #1 of 0.1 M sodium phosphate (pH 7.0) was added. The mixtures were lyophilized. For the experiments in the absence of cytochrome c, 110 ~1 of DzO was added to one of the dried samples to initiate the hydrogen exchange. For the experiments in the presence of the antigen, one of the deuterated and dried samples of appropriate cytochrome c (see above) was dissolved in 110/,tl of D 2 0 and added to one of the monoclonal samples. The zero time is the instant at which D , O or the deuterated cytochrome c / D 2 0 was added. A reference was a D~O solution containing the same components, except antibodies or antibodies and deuterated cytochrome c. Spectra of an experimental solution were recorded between 1250 and 1800 cm-~ at 23-25°C as a function of time (Fig. 2). The recording was repeated at a given time point (45 s per recording, 8 cm -~ resolution according to manufacturer's specification). In a typical experiment the numbers of recording at a given time point are: 1 between 3 and 8 rain; 2 - 4 between 12 and 48 rain; 4 - 5 between 1 and 5 h; 5 - 6 between 6 and 24 h and 6 - 7 between 48 and 72 h. Amide I (AI) and amide II (AII) absorbances were determined at 1640 cm ~  and 1545 cm-~ , respectively, after correction for the background absorbances (Fig. 2). Then, the average value for A I I / A I at a given time point t was used as the value for A I I / A I at time t. Both experimental and reference ceils were removed from a compartment of an infrared spectrophotometer each time the instrument was being reset for a second recording. This resulted in no appreciable increase in the temperature of the experimental solution after repeating the recording up to 8 times, as judged by touching the cells. The experiments for the Fab were also similarly carried out with the following exception. A solution of