Monoclonal Antibodies That Recognize Minimal Differences in the

Mar 30, 2002 - Immunization of BALB/c mice with a cadmium−chelate−protein conjugate resulted in the isolation of two hybridoma cell lines (A4 and ...
0 downloads 0 Views 226KB Size
Download PDF
408

Bioconjugate Chem. 2002, 13, 408−415

Monoclonal Antibodies That Recognize Minimal Differences in the Three-Dimensional Structures of Metal-Chelate Complexes R. Mark Jones,† Haini Yu,† James B. Delehanty,† and Diane A. Blake*,†,‡ Department of Ophthalmology, Tulane University Health Sciences Center, New Orleans, Louisiana, 70112, and Tulane/Xavier Center for Bioenvironmental Research, New Orleans, Louisiana 70112. Received August 1, 2001; Revised Manuscript Received November 29, 2001

Immunization of BALB/c mice with a cadmium-chelate-protein conjugate resulted in the isolation of two hybridoma cell lines (A4 and E5) that synthesized antibodies with different variable regions, but similar metal-chelate affinity. The ability of these two monoclonal antibodies to interact with 12 different metal-chelate complexes was studied using the KinExA 3000 immunoassay instrument. The two antibodies showed the highest affinity for cadmium and mercury complexes of ethylenediamine N,N,N′,N′-tetraacetic acid (EDTA). The E5 antibody bound to EDTA complexes of cadmium and mercury with equilibrium dissociation constants (Kd) of 1.62 × 10-9 M and 3.64 × 10-9 M, respectively. The corresponding values for the A4 antibody were 14.7 × 10-9 M and 3.56 × 10-9 M. Addition of a cyclohexyl ring to the EDTA backbone increased the affinity of E5 for the metal-chelate haptens, while decreasing the binding of A4 to the same haptens. Based on available crystal structures, molecular models were constructed for five different divalent metal-chelate complexes. The models were compared to determine structural features of the haptens that may influence antibody recognition. Difference distance matrixes were used to identify areas of the metal-chelate haptens that differed in three-dimensional space. Antibody affinity correlated well with the extent of total structural difference for these metal-EDTA complexes.

INTRODUCTION

Half of the 20 amino acids have functional groups that can coordinate metal ions, and the resulting proteinmetal complexes play a number of roles; they serve as electrophiles at active sites, redox centers, and foci for structural stabilization (1). Antibodies that bind metals and metal-containing antigens make up a special class of these metal-binding proteins. There are a number of reports of metal-binding antibodies. Myelomas producing Cu- and Ca-binding immunoglobulins were reported in the mid and late 1970s (2, 3). More recently antibodies directed against glutathione- and chelate-metal complexes of Hg(II), In(III), Cd(II), and Pb(II) have been reported (4-7). These metal-specific monoclonal antibodies (mAbs)1 have potential uses in both medicine and environmental analysis. Some metal-binding antibodies * To whom correspondence should be addressed: Department of Ophthalmology SL-69, Tulane University Health Sciences Center, 1430 Tulane Ave., New Orleans, LA 70112-2699: Tel.: 504-584-2478; Fax: 504-584-2684. E-mail: [email protected]. † Tulane University Health Sciences Center. ‡ Tulane/Xavier Center for Bioenvironmental Research. 1 Abbreviations used: mAb, monoclonal antibody; EDTA, ethylenediamine-N,N,N′,N′-tetraacetic acid; EOTUBE, 4-[N′-(2hydroxyethyl)thioureido]-L-benzyl-EDTA; CDR, complementarity determining regions; BSA, bovine serum albumin; KLH, keyhole limpet hemocyanin; ITCBE, 1-(4-isothiocyanobenzyl)ethylenediamine-N,N,N′,N′-tetraacetic acid; CDTA, trans-1,2cyclohexanediamino-N,N,N′,N′-tetraacetic acid; TMB, 3,3′,5,5′tetramethylbenzidine; ELISA, enzyme-linked immunosorbent assay; PCR, polymerase chain reaction; HBS, HEPES buffered saline, (137 mM NaCl, 3 mM KCl, 10 mM Hepes); IgG, immunoglobulin G.; Kd, equilibrium dissociation constant; Kon, association rate constant; Koff, dissociation rate constant; DM, distance matrix; DDM, difference distance matrix; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis.

have had utility in tumor imaging and treatment (8). Immunoassays for specific metal ions have also been developed using metal-specific monoclonal antibodies as their primary reagent (9-12). These immunoassays, with sensitivities of sub-part per billion for some metal ions, will be useful adjuncts to more traditional analytical methods of metals analysis. Metal-specific monoclonal antibodies also offer a unique opportunity to examine mechanisms of protein-metal interactions. The best studied metal-specific mAb is CHA255, which binds to In(III)-EDTA complexes (13). This antibody was studied in detail via X-ray crystallography and was shown to bind an In(III)-4-[N′-(2hydroxyethyl)thioureido]-L-benzyl-EDTA (EOTUBE) complex through both direct metal coordination and ionic interactions with the chelate molecule. A histidine residue in the third complementarity determining region of the heavy chain (CDRH3) was shown to coordinate the In(III) ion. Contiguous to the histidine, an arginine appeared to ion-pair with one of the carboxylate arms of the In(III)-EOTUBE complex. When the In(III) was replaced with Fe(III), the affinity of CHA255 decreased 20-fold. The histidine in CDRH3 could no longer coordinate the Fe(III) due to changes in the conformation of the glycinate arms of the EDTA chelate. This study emphasized that the direct coordination of the metal by CHA255 was required for high affinity binding. The Cd(II)- and Hg(II)-EDTA specific antibody, 2A81G5, was homology modeled using CHA255 as a template (6). This model revealed a CDRH3 histidine residue in a similar metal-ion-coordinating orientation. Apart from this histidine, the two antibodies shared little CDR homology or binding similarities. The metal-ion selectivity of the CHA255 antibody could not be explained in terms of metal ion size or metal-chelate stability constants (5).

10.1021/bc0155418 CCC: $22.00 © 2002 American Chemical Society Published on Web 03/30/2002

Antibody Recognition of Metal−Chelate Complexes

With the exception of Hg(II) and Mg(II), the metal-ion selectivity and affinity of the 2A81G5 antibody was found to correlate well with the atomic volume of the divalent metal ions (6). Both of these studies gave only cursory regard to the total metal-chelate structure, which is regulated by both ion size and coordination bond lability. In the current study, we report the isolation and characterization of two monoclonal antibodies that showed similar metal-chelate affinities during initial screening studies. These two antibodies were subsequently found to have different N-terminal sequences in both their heavy and light chains. We have compared the ability of these two antibodies to bind to a variety of metal-chelate complexes and hypothesized about recognition mechanisms based on metal-chelate structure. The two antibodies were found to bind Cd(II)- and Hg(II)-chelate complexes with the highest affinity. Molecular modeling of five divalent metal-chelates with known crystal structures also allowed us to focus on differences in the metal-chelate hapten structures that may be important for antibody recognition. EXPERIMENTAL PROCEDURES

Materials. Bovine serum albumin (BSA) (ultrapure) and an immunoglobulin G (IgG) isotyping kit were obtained from Boehringer Mannheim (Indianapolis, IN). Keyhole limpet hemocyanin (KLH) and 1-(4-isothiocyanobenzyl)ethylenediaminetetraacetic acid (ITCBE) were purchased from CalBiochem Corp. (La Jolla, CA). Centricon 30 microconcentrators were products of Amicon, Inc. (Beverly, MA). Metal ions were atomic absorption standards from Perkin-Elmer Corporation (Norwalk, CT). BALB/c AnNtacfBR inbred mice were purchased from Charles River Laboratories (Wilmington, MA). Ribi adjuvant was obtained from Ribi Immunochemicals (Hamilton, MT). HEPES, ethylenediamine-N,N,N′,N′-tetraacetic acid (EDTA), tissue culture medium, L-glutamine, HAT medium supplement, antibiotics, cyanogen bromide activated agarose, and goat anti-mouse IgG coupled to horseradish peroxidase were purchased from Sigma Chemical Co. (St. Louis, MO). Sodium dodecyl sulfatepolyacrylamide gel electrophesis (SDS-PAGE) reagents were obtained from Bio-Rad (Hercules, CA). Pyroglutmate amino peptidase was purchased from Panvera Corporation (Madison, WI). trans-1,2-Diaminocyclohexane-N,N,N′,N′-tetraacetic acid (CDTA) was purchased from Aldrich (Milwaukee, WI). 3,3′5,5′-Tetramethylbenzidine (TMB) microwell substrate was a product of Kirkegaard-Perry Laboratories (Gaithersburg, MD). Mini Problott membranes were obtained from Applied Biosystems (Foster City, CA). Immobilized protein G resin and BCA protein assay were purchased from Pierce Chemical Co. (Rockford, IL) and used according to the manufacturer’s instructions. Protein-Chelate Conjugates. Protein chelate conjugates were prepared as described in (14). A 20 mg/mL solution of protein (BSA or KLH) was prepared in phosphate-buffered saline (137 mM NaCl, 3 mM KCl, 10 mM sodium phosphate), pH 9.5. ITCBE and Cd(II) were added to the BSA solutions to achieve final concentrations of 1.36 mM and 1.4 mM, respectively. ITCBE and Cd(II) were added to KLH solutions in equimolar amounts to achieve a final concentration of 1.5 mM. The reactions were adjusted to pH 9.0 and stirred for 22 h at 25 °C. Unreacted low-molecular weight reagents were removed by buffer exchange using a Centricon 30 device that had been previously treated with EDTA and washed profusely with Hepes-buffered saline (HBS, 137 mM NaCl, 3 mM

Bioconjugate Chem., Vol. 13, No. 3, 2002 409

KCl, 10 mM Hepes), pH 7.4. The conjugates were washed five times with metal-free HEPES (0.1 M, pH 9.0) and twice with metal-free HEPES (0.1 M, pH 7.4). Protein conjugate concentration was determined using the Pierce BCA protein assay. The degree of conjugate substitution was quantified using the trinitrobenzenesulfonic acid method similar to that described in refs 15 and 16. The extent of substitution of free lysine residues was 15.817.0% for BSA conjugates and 15% for the KLH conjugate. Mouse Immunization, Hybridoma Production, and Purification of Monoclonal Antibodies. Five 6-week old BALB/c AnNtacfBR mice were injected intraperitoneally with 50 µg of the Cd(II)-EDTA-KLH conjugate emulsified in Ribi adjuvant at days 0, 21, and then at 2-week intervals. Antibody response of each mouse was determined by indirect ELISA using both metal-free EDTA-BSA and cadmium-loaded conjugate (6). The mice showing highest antibody responses were given a final intraperitoneal boost of 50 µg of the KLH conjugate in phosphate-buffered saline (pH 7.4) 4 days prior to fusion. Mouse spleen cells were harvested, washed in RPMI 1640 medium, and fused with X63-Ag 8.653 myeloma cells. Supernatants from viable, replicating hybridoma clones were collected and screened for antibodies to a variety of metal-chelate complexes by competitive ELISA (described below). Hybridomas synthesizing and secreting metal-specific antibodies were subcloned twice by limiting dilution. Positive hybridoma clones were subcultured and frozen. Ascites fluid was produced in BALB/c mice primed with Freund’s incomplete adjuvant by intraperitoneal injection of 1.27 × 106 hybridoma cells. Ascites fluid was collected 10-15 days after the hybridoma injection. The clone E5 was difficult to grow as an ascites tumor, while clone A4 displayed no problems developing as ascites. The IgG fraction of ascites was isolated by affinity chromatography on immobilized protein G. A portion of the protein G purified antibody was further purified on a Cd(II)EDTA-BSA affinity column. The Cd(II)-EDTA-BSA affinity-purified antibodies were separated under reducing conditions by SDS-PAGE and electroblotted to Mini Problott membranes. The heavy chain N-terminus of antibody E5 was blocked and required deblocking according to the method described in Mozdzanowski et. al prior to electrophoresis (17). The heavy and light chain proteins were excised, and N-terminal sequencing was performed at a commercial facility. The immunoglobulin subclass and light chain isotype of the antibodies were determined using an IsoStrip antibody isotyping kit according to the manufacture’s instructions. Protein concentration was determined using the BCA protein assay kit. Competitive ELISA. Competitive ELISAs were performed essentially as described in ref 11. Cd(II), Hg(II), Mn(II), Cu(II), Zn(II), Fe(III), Ni(II), In(III), Ag(I), Pb(II), Co(III), Mo(VI), and Ca(II) were tested for their ability to inhibit binding to the immobilized Cd(II)EDTA-BSA conjugate in a HBS amended with 0.4 mM CDTA or 4.0 mM EDTA. Determination of Equilibrium Dissociation Constants and Bimolecular Association Rate Constants. The KinExA 3000 (Sapidyne Instruments Inc., Boise, ID) is an automated flow fluorimeter used to determine the amount of antibody with unoccupied binding sites in a mixture of antibody and ligand (18). Binding experiments were performed by coating 200 mg of poly(methyl methacrylate) beads (63-103 µm) (Sapidyne Instruments, Inc.) with 1 mL Cd(II)-EDTA-BSA

410 Bioconjugate Chem., Vol. 13, No. 3, 2002

conjugate (100 µg/mL). The conjugate and beads were rocked for 1 h at 37 °C. The beads were then pelleted and washed three times with 1 mL of HBS. The beads were blocked with 1 mL of 1% BSA in HBS at 37 °C for 1 h and stored in blocking buffer at 4 °C for up to six weeks. Detailed protocols for use of the KinExA instrument are available in ref 18. Data acquisition and instrument control were accomplished using a PC interfaced with the KinExA and software distributed by Sapidyne, Inc. The resulting data was compiled using Microsoft Excel, and equilibrium dissociation constants were obtained from nonlinear regression analysis of the data using a one-site homogeneous binding model contained within SlideWrite Plus software (Version 5.0, Advanced Graphics, Inc, Encinitas, CA). Modeling of Metal-Chelate Conjugates and Distance Matrixes. Crystal structures for Cd(II)-, Hg(II)-, Mn(II)-, In(III)-, Fe(III)-, Ni(II)-, and Zn(II)-EDTA complexes were obtained from the Cambridge Structural Database (19-25). Metal-EDTA structures were displayed on a Silicon Graphics Octane workstation with SYBYL 6.6 molecular modeling software (Tripos Associates, St. Louis, MO). The extended parameter set $TA_DEMOS/metals.tpd was used for metal recognition as provided by Tripos. The crystal structures were geometry-optimized using the standard Tripos molecular mechanics force field with 0.001 kcal/mol energy gradient convergence criterion. Interatomic distances were determined manually from the SYBYL menu using the ANALYSE>DISTANCE option. The atom numbering between the crystal structures varied, so the chelatemetal complexes were aligned and numbered according to ref 26. Distance Matrixes and Difference Distance Matrixes were built using the program Mathematica version 4.1 (Wolfram Research, Champaign, IL). The carbonyl oxygens (O2, O4, O6, and O8) (see Figure 3) were not included in the difference distance matrixes in order to reduce the size and complexity of Figure 4. Due to the planar nature of the carbonyl bond the distance information is mirrored by the metal-coordinating oxygens and thus would be repetitive. RESULTS

Hybridoma Screen. Hybridoma cells were screened for antibodies that bound Cd(II)-EDTA-BSA and 49 positive clones were detected. The positive hybridomas were grouped based on the ability of their antibodies to be inhibited in competitive ELISA. Cells producing antibodies that showed little inhibition by soluble EDTA or Zn(II)-EDTA, but considerable inhibition by Cd(II)EDTA, were selected for further analysis. Cells in a single well showing activity were subcloned by limiting dilution and gave rise to two clones, A4 and E5. Supernatants from these two clones were again screened for their ability to be competitively inhibited by metal-free EDTA and Zn(II) or Cd(II) in a background of 5 mM EDTA. Both antibodies showed little inhibition by metal-free EDTA at concentrations as high as 5 mM (data not shown). As shown in Figure 1, addition of Zn(II) (765 µM)-EDTA inhibited A4 by 54% and E5 by 71%, while Cd(II) (445 µM)-EDTA inhibited 97% of the binding activities of the two antibodies. The initial interpretation of this screening data was that A4 and E5 were identical clones, and the differences in the competitive ELISA screen were due to variations of antibody concentration in the culture supernatants. Isotyping revealed that both mAbs were of the IgG1 subclass with κ light chains.

Jones et al.

Figure 1. Hybridoma supernatant screen by competitive ELISA. The tissue culture supernatants from clones A4 and E5 were diluted in PBS containing 5 mM EDTA and tested by competitive ELISA for their ability to be inhibited by soluble Zn(II) (765 µM) or Cd(II) (445 µM) in a background of 5 mM EDTA. Table 1. N-Terminal Amino Acid Sequence for A4 and E5 Variable Regionsa heavy chain variable region sequence A4 E5

EVQLQQSGAELVKPGASVKLSCSVTA QVQLKESGPGLVAPSQSLSIT light chain variable region sequence

A4 E5

DIVMSQFPSSLAVSTGERVTM DVVMTQTPLSLPVSLGXQAXI

a An X indicates amino acid residues that could not be accurately determined from the data.

N-Terminal sequencing subsequently revealed that the two antibodies differed in amino acid sequence in the first framework region of both their light chain and heavy chain variable regions as shown in Table 1. Although these framework sequences do not define the antigenbinding domains of the antibodies, these data, taken with the differences observed in binding affinities (see below), indicate that A4 and E5 are two different antibodies. Binding Specificity. Binding comparisons of A4 and E5 were initially performed using competitive ELISA (data not shown). The ELISA data identified trends that could be more rigorously studied using the KinExA 3000. The KinExA instrument allows the separation and quantification of unbound antibody in an equilibrium mixture of antibody, ligand, and antibody-ligand complexes. Unbound antibody is removed from the equilibrium mixture by a brief passage through a column of immobilized ligand. The equilibrium mixture is exposed to the column for approximately 240 ms, and thus the immobilized ligand acts merely as a trapping reagent to remove a portion of uncomplexed antibody from the mixture (18). The KinExA instrument generates fluorescence curves as shown in Figure 2A. A portion of the curve (for this study between 300 and 399 s) was integrated, and the concentration of free, unbound antibody in each equilibrium mixture was determined from the integrated fluorescence. The fraction of occupied binding sites on the antibody molecule was taken as:

I0 - Iexp I0 - I∞

(1)

Antibody Recognition of Metal−Chelate Complexes

Bioconjugate Chem., Vol. 13, No. 3, 2002 411

Table 2. Binding (Kd, M) of Chelators and Chelated Metals to Antibodies A4 and E5 A4

E5

chelator

EDTA

CDTA

EDTA

CDTA

metal free Cd(II) Hg(II) In(III) Mn(II) Zn(II) Cu(II) Ni(II) Co(III) Fe(III) Ca(II) Pb(II)

>10-3 (1.47 ( 0.24) × 10-8 (3.56 ( 1.08) × 10-9 (3.11 ( 0.55) × 10-7 (4.71 ( 0.88) × 10-7 (1.10 ( 0.21) × 10-6 (1.11 ( 0.06) × 10-6 (5.39 ( 0.80) × 10-6 (3.04 ( 0.17) × 10-6 (6.03 ( 1.08) × 10-6 >1 × 10-6 >1 × 10-5

(5.27 ( 1.60) × 10-4 (4.20 ( 0.50) × 10-8

>10-3 (1.62 ( 0.14) × 10-9 (3.64 ( 1.33) × 10-9 (1.69 ( 0.30) × 10-7 (1.45 ( 0.09) × 10-7 (1.18 ( 0.13) × 10-6 (1.80 ( 0.23) × 10-6 (1.10 ( 0.50) × 10-6 (1.22 ( 0.40) × 10-6 (4.00 ( 0.89) × 10-6 >2 × 10-6 >3 × 10-5

(2.27 ( 0.34) × 10-8 (1.16 ( 0.34) × 10-9

(6.90 ( 2.00) × 10-7 (6.42 ( 0.45) × 10-6

Figure 2. Determination of equilibrium dissociation constants for A4 and E5 to cadmium-chelate complexes. Panel A: Instrument response traces from the KinExA 3000. A4 inhibition by Cd(II) in a backgroud of 0.4 mM EDTA. Curve 1 corresponds to zero Cd(II) concentration, curve 8 corresponds to a saturating Cd(II) concentration (100 nM), and curves 2-7 correspond to Cd(II) concentrations of 0.5, 2,5, 10, 20, and 50 nM, respectively. Panel B: The fraction of occupied binding sites is determined by integrating a portion of the instrument response traces between 300 and 399 s, graphing against ligand concentration, and normalizing these integrated fluorescence curves. A4 binding to Cd(II)-EDTA is represented by closed circles (b). E5 binding to Cd(II)-EDTA is represented by closed triangles (2), and E5 binding to Cd(II)-CDTA is represented by open triangles (4). Equilibrium dissociation constants are summarized in Table 1.

where I represents the integrated fluorescence and the subscripts 0, exp, and ∞ represent experimental traces corresponding to a ligand concentration of zero, an intermediate concentration, and a saturating ligand concentration, respectively. The fraction of occupied binding sites was plotted versus ligand concentration, and the equilibrium dissociation constant, Kd, was determined from a nonlinear regression fit of the data (example shown in Figure 2B) to the following rectangular hyperbola:

fraction of occupied binding sites ) X/(X + Kd) (2) where X represents the free ligand concentration.

(3.23 ( 0.83) × 10-8 (3.40 ( 0.96) × 10-9

Both antibodies showed negligible affinity for metalfree EDTA as shown in Table 2. Antibody A4 bound most tightly to Hg(II)-EDTA complexes. Replacement of Hg(II) in the EDTA complex with Cd(II) decreased the affinity of A4 by 4-fold. Replacement with In(III) or Mn(II) lowered the Kd 2 orders of magnitude from that observed with the Hg(II)-EDTA complex. All other metal-EDTA complexes tested (Zn(II), Cu(II), Ni(II), Co(III), Fe(III), Ca(II), and Pb(II)) bound A4 more than 200fold less tightly than Hg(II)-EDTA. Of the EDTA complexes, E5 showed the highest affinity for Cd(II)-EDTA and only a 2-fold lower affinity for Hg(II)-EDTA (Table 2). Replacement of Cd(II) with In(III) or Mn(II) in the EDTA complex resulted in an approximate 100-fold decrease in affinity, while all other metals decreased the affinity of E5 for the metal-chelate complex by over 500-fold. The two antibodies bound EDTA complexes of Cd(II), Mn(II), and Ni(II) with differing affinities, but both antibodies bound the EDTA complexes of Hg(II), Zn(II), Co(III), Cu(II), Fe(III), and In(III) with very similar affinities. When the structure of the chelator was changed by substituting CDTA for EDTA, antibody recognition of the metal-free CDTA was measurable (Table 2). A4 recognized metal-free CDTA with just under millimolar affinity, while E5 bound to metal-free CDTA with a surprisingly high affinity (22.7 nM). A4 bound less tightly to the metal-CDTA complexes than to the corresponding metal-EDTA complexes. Except for Mn(II)-CDTA, the average decrease in affinity of A4 for metal-CDTA complexes over metal-EDTA complexes was approximately 2-fold. The decrease for Mn(II)-CDTA versus Mn(II)-EDTA was just over 10-fold. In contrast, E5 bound more tightly to the metal-CDTA complexes than to the corresponding metal-EDTA complexes. E5 bound Cd(II)-CDTA slightly tighter than the Cd(II)-EDTA complex, but recognized Mn-CDTA with a 40-fold higher affinity than Mn(II)-EDTA. E5 also showed a 5-fold increase in affinity for In(III)-CDTA over the In(III)EDTA complex. Association and dissociation rate constant determinations were attempted for the Cd(II), Mn(II), In(III), and Hg(II) metal-chelate complexes and for the metal-free CDTA. The KinExA 3000 was used to determine bimolecular association rate constants in these procedures (18). The concentration of antigen was varied while maintaining a constant, but short reaction time. Antibody was injected into and mixed with a stream of antigen. This mixture was allowed to react for the 11 s it took the stream to flow from the injection point to the observation cell and interact with the bead column. The resulting fluorescent response curves were similar to those in Figure 2A, and a single-exponential function of

412 Bioconjugate Chem., Vol. 13, No. 3, 2002

Jones et al.

Figure 3. Energy-minimized crystal structures of metal-EDTA complexes. EDTA complexes of Cd(II), yellow; Hg(II), red; Mn(II), cyan; Ni(II), magenta; and Zn(II), blue. Panels A and B: Superimposed structures of all five complexes, shown from two different angles of view. Panel C: Superimposed structures of EDTA complexes of Cd(II) and Hg(II). Panel D: Superimposed structures of EDTA complexes of Cd(II) and Ni(II). Bar, 1 Å. Table 3. Measurable Kinetic Constants for Antibodies A4 and E5a ligand

association rate constant, dissociation rate constant, s-1 M-1 s-1

Hg(II)-EDTA Mn(II)-EDTA

Antibody A4 (3.96 ( 0.9) × 106 (6.64 ( 1.7) × 104

0.014 0.031

Hg(II)-EDTA Cd(II)-EDTA

Antibody E5 (2.26 ( 0.2) × 106 (1.73 ( 0.2) × 107

0.008 0.028

a Association rate constants were determined as described in the text. Values for dissociation rate constants were obtained by calculation from the association rate constant and the Kd.

time was fit to each data set (see ref 18 for details). Unfortunately, association rate constants for metal-free CDTA and the In(III)-chelate complexes with both antibodies, Cd(II)-EDTA with A4, and Mn(II)-EDTA with E5 were not measurable by this method. The antibody and metal-chelates had reached equilibrium in the 11 s allowed for the mixing reaction, indicating that the antibody had an on-rate for these ligands that was faster than the operational deadtime of the instrument. The on-rate for Hg(II)-EDTA with A4 was 59-fold higher than that determined for Mn(II)-EDTA as shown in Table 3. The off rate was determined by the identity Kd ) Koff/Kon. A4 dissociated from Mn(II)-EDTA at a rate 2× faster than that determined for Hg(II)-EDTA. The on-rate for E5 was 7.5-fold faster with Cd(II)-EDTA than with Hg(II)-EDTA, but it also dissociated 3.5-fold faster. The bimolecular association rate constants of both antibodies with Hg(II)-EDTA were similar. Difference Distance Matrixes. The ability of the A4 and E5 antibodies to discern subtle differences in the

structures of metal-chelate complexes was further investigated by analyzing the available crystal data for several of these complexes. Figure 3 shows the energyminimized crystal structures for EDTA complexes of Cd(II), Hg(II), Mn(II), Ni(II), and Zn(II). As can be seen from the 1 Å bar at the bottom of the figure, these complexes are structurally very similar. In Panels A and B, all five structures were superimposed and displayed from two different angles of view. A superimposition of the two most structurally similar complexes under study, Cd(II)EDTA and Hg(II)-EDTA, is shown in Panel C, while the superimposed structures of the two most dissimilar complexes, the Cd(II) and Ni(II) complexes, are displayed in Panel D. While comparisons of such three-dimensional projections provided an overall appreciation of the similarities and differences in these metal-EDTA complexes, they were not sufficient to quantify structural differences between the complexes. Distance matrixes (DMs) were therefore developed to compare the molecular structures of the metal-chelate complexes. The DM for one individual complex was subsequently subtracted from that of another complex to yield difference distance matrixes (DDMs), which identified areas in two metal-chelate structures under comparison where differences in the three-dimensional structures occurred. This analysis required that the atoms compared be numbered identically. Metal-EDTA crystal structures obtained from the Cambridge Structural Database all had different numbering schemes; these crystal structures were therefore renumbered according to the scheme used by Nesterova and Porai-Koshits (26), and the numbering scheme used in this analysis is shown in Panel C of Figure 4. The glycinate arms are labeled A-D to aid in their identification. Two examples of the DDMs are shown in Panels A

Antibody Recognition of Metal−Chelate Complexes

Figure 4. Difference distance matrixes. Panels A and B: The difference distance matrixes (DDM) were generated by subtracting the distance matrixes of two individual metal-chelate complexes (see Experimental Procedures), and the values were given a gray scale shade. Each square represents a value corresponding to the distance difference between the two atoms; therefore the lower right square represents the distance difference between the metal and oxygen 7. The entire bottom row of squares is distance differences from the metal atom. The dark dashed line separates the DDM into the two symmetric halves. This line runs through squares of zero value that can be used as a color reference with the range scale. Panel A: DDM for Cd(II)-EDTA minus Hg(II)-EDTA. The differences are all below 0.2 Å. Panel B: DDM for Cd(II)-EDTA minus Ni(II)EDTA. The splotchy pattern reports atom distance differences up to 0.6 Å. Panel C: Distance matrixes require that the atoms compared be numbered identically. Metal-EDTA crystal structures obtained from the Cambridge Structural Database all had different numbering schemes; these crystal structures were renumbered according to the scheme used by Nesterova and Porai-Koshits (26). C1 was established by determining the ethylene carbon that was farthest from the metal ion. The glycinate arms are labeled A-D to aid in their identification.

and B of Figure 4 with the gray scale range (-0.9 Å to +0.9 Å) shown below. Figure 4A is the DDM for Cd(II)EDTA minus Hg(II)-EDTA. The distance differences between atom positions in the Cd(II)-EDTA and Hg(II)EDTA structures all fall below 0.2 Å, indicating that the three-dimensional structures of these two complexes are very similar (see also the superimposed structures in Figure 3C). In contrast, a number of differences are evident in the DDM for Cd(II)-EDTA minus Ni(II)EDTA, shown in Figure 4B. Dark squares indicate distances in the Ni(II)-EDTA complex where the atoms are farther apart than corresponding atoms in the Cd(II)-EDTA structure; lighter squares indicate distances in the Ni(II)-EDTA complex where the atoms are closer together than atoms in the Cd(II)-EDTA structure. This analysis showed that the atoms of arm A (C3/C4/O1) (see Figure 4C) and the atoms of arm D (C9/C10/O7) were closer together in the Ni(II)-EDTA complex than in the Cd(II)-EDTA complex. When comparing the distance differences between the arms, compare the distance between the atoms C3 and C9, C4 and C10, and O1 and O7 (highlighted in Figure 4B by squares with bold edges).

Bioconjugate Chem., Vol. 13, No. 3, 2002 413

Lighter shading at the intersection of the atoms in arm A (C4/O1) with the atoms in arm B (C6/O3) (highlighted in Figure 4B by squares with dotted edges) indicates that these atoms are also closer together in the Ni(II)-EDTA complex than in the Cd(II)-EDTA complex. Worth noting is that O1 does not get closer to C5 or C9. This may indicate a rotation around the C-C bond in the glycinate arms. The dark shading at the intersection of C4 with C1/C2 and O1 with C1/C2 indicates that in the Ni(II)EDTA complex these atoms were farther apart than in the Cd(II)-EDTA complex. This indicates arm A is further from the ethylene backbone in the Ni(II)-EDTA complex than the Cd(II) complex. This is counter-intuitive because Cd(II) has a larger ion size, but may be explained by crowding of the glycinate arms around the Ni(II) ion and forcing arm A to move away from the ion. The distances between the atoms in arm B (C5/C6/O3) with the atoms in arm C (C7/C8/O5) are farther apart from one another in the Ni(II)- compared to the Cd(II)-EDTA complex. These differences are mimicked by the Cd(II)EDTA minus Zn(II)-EDTA difference matrix and similar differences are seen when either the Zn(II)- or Ni(II)EDTA matrix is subtracted from the Mn(II)-EDTA matrix (data not shown). In an effort to relate differences in the molecular shape of the metal-chelate complexes to their ability to be recognized by the A4 and E5 antibodies, the absolute values from one of the symmetric halves of the DDM were summed. This term is representative of the total differences between the two metal-chelate haptens being compared. The structures of five EDTA-chelated divalent metals (Cd, Hg, Ni, Mn, and Zn), available in the Cambridge Structural Database, were used in this comparison. As shown in Figure 5, when the total difference was related to the -log Kd, an inverse relationship was observed. The metal-EDTA complex with the highest affinity for each antibody was used as the standard when the differences were calculated. The coefficient of determination (r2) for the total difference related to the binding of A4 was very strong (r2 ) 0.97), while this relationship was only slightly weaker for E5 (r2 of 0.84) (Figure 5). DISCUSSION

Antigen specificity and antibody binding are believed to reside mainly in the CDRs found in the variable regions of the antibody molecule (27). The two antibodies examined in the present study differ in the N-terminal sequences of their light and heavy chain variable regions (Table 1). The differences in the N-terminal sequence data and the differences in affinity are compelling evidence that E5 and A4 are two different antibodies. Because of the differences in protein sequence and the similarities of activity in the initial screening, these two antibodies were chosen for a comparative binding study. Such a structure/function comparison requires the use of a number of structurally similar ligands to probe the antibody binding activity. The metal-chelate complexes offer a unique system whereby the hapten can be easily modified simply by changing the metal in the chelate complex. Larger hapten modifications can be made by replacing the chelate. For the metals examined in this study, the logarithms of the critical stability constants for EDTA range from 10.8 (Ca2+) to 25.0 (Fe3+), and those for CDTA from 17.5 (Mn2+) to 28.7 (In3+) (28). Under the conditions of our binding assays (400 µM chelator and 0.1 to 50 µM metal ion), it was reasonable to assume that all the metal ions in Table 2 existed as complexes with the chelator, and that the metal-chelate complex was the form that interacted with the antibody. The metal-

414 Bioconjugate Chem., Vol. 13, No. 3, 2002

Jones et al.

Figure 5. Changes in hapten shape correlate with antibody binding. One-half the sum of the absolute value of the DDM was plotted versus the negative logarithm of the equilibrium dissociation constant for the divalent metal ions modeled. Panel A: The relationship for A4 when the metal-EDTA DDMs are subtracted from the highest affinity ligand, Hg(II)-EDTA (r2 ) 0.97). Panel B: The relationship for E5 when the metal-EDTA DDMs are subtracted from the highest affinity ligand, Cd(II)-EDTA (r2 ) 0.84).

chelates used in this study exist in at least four distinct conformations. NMR studies have shown that these conformations arise primarily from the lability of the metal-ligand bonds and that the bond lability is dependent upon the identity of the metal ion (29, 30). For example, the metal-oxygen bond of the Cd(II)-EDTA molecule is labile and rapidly rotates around the nitrogencarbon bond, while the metal-nitrogen bond is longer lived and does not allow for complete dissociation between the chelate and metal (29). This may allow for rapid interconversion between different conformations of the Cd(II)-EDTA complex. In contrast, these NMR studies point out the longevity of both the metalnitrogen and metal-oxygen coordination bonds in the Co(III)-EDTA complex, which may suppress rapid conformational conversion. The crystal structures of Cd(II)-, Hg(II)-, In(III)-, and Mn(II)-EDTA were observed to be in the same stereoisomeric conformation as the In(III)-chelate complex in the CHA255 binding pocket (13, 20-22, 24). The crystal structures of Ni(II)-, Fe(III)-, and Zn(II)-EDTA, all recognized with lower affinity by both A4 and E5 antibodies, were in a second, different stereoisomeric conformation (19, 23, 25). Modeling and geometrical minimizing revealed that these original crystal structures were in the lowest energy conformation (data not shown). If the energy differences are indicative of conformer stability, then the metal-chelates’ ability to exchange between conformations may play a role in how readily they adopt the proper conformation to fit into the antigen binding site of the antibodies. The conformational changes of the chelate arms as they shift to accommodate various metal ions appear to be an important factor in antibody recognition. In our study, the DDM indicated that the major difference between the metal-EDTA haptens was the way in which the chelate arms encircled the metal ion. The strong correlation between the total difference in structure and binding activity for the divalent metal ions with A4 and E5 is striking, especially when the maximum, single distance difference between any of the atoms in the structures was 0.6 Å (Figure 4). The shifting of the chelate arms to accommodate metals was demonstrated when the separate crystal structures of the antibody CHA255 bound to In(III)- and Fe(III)-EDTA were solved (13); when Fe(III) replaced In(III) in the chelate complex, the conformation of the chelate arms around the metal changed and blocked the antibody coordination to the metal by

the histidine, thus lowering the antibody’s affinity. The crystallography data obtained with antibody CHA255 indicate that it is entirely plausible that the additive effects of these small differences can significantly alter antibody affinity. Altering the chelate structure also played a major role in antibody affinity. The two antibodies showed opposite behavior when presented with CDTA-bound metals. In CDTA-metal complexes, the nitrogen becomes more basic due to the cyclohexyl ring; this serves to increase the strength of the nitrogen-metal bond and increase the metal-chelate stability. The stability constants for CDTA complexes are all higher than for their respective EDTA complexes (28). A4 bound less tightly to CDTA metal complexes than to the corresponding EDTA complexes, while E5 increased its affinity for the CDTA complexes. The increased affinity of E5 for the metalfree CDTA could be due to the positive hydrophobic interaction of the cyclohexyl ring with the antigen binding site. The same could be suggested for the affinity of A4 for the metal-free CDTA, but then we would expect an increase in affinity with metal-loaded CDTA. This is not what is observed. The increase in A4 affinity to metalfree CDTA may more likely be due to the preorganizing effect of the cyclohexyl group on the chelate; thus the normally nonstructured glycinate arms, as in EDTA, would now have a structured shape similar to the metalchelated arms in metal-loaded EDTA (31). The decrease in A4’s affinity for metal-loaded CDTA may be due to a negative hydrophobic interaction of the cyclohexyl portion of the chelate with the binding site. The association rate constants may also yield information on how well the metal-chelates fit the binding site. Both antibodies have similar rate constants for Hg(II)EDTA. The on-rate of A4 for Mn(II)-EDTA is almost 60fold less than for Hg(II)-EDTA. The Mn(II) ion, like Co(III), Fe(III), and Ni(II), all have strong metal-oxygen bonds (32). If these smaller ions cannot convert rapidly to a conformation preferred by the antibody, it would most likely reduce the on-rate. Cd(II)-EDTA complexes were shown to have very labile metal-oxygen bonds (29). This fact, taken with the fast on-rate of Cd(II)-EDTA with E5, supports the idea of rapid interconversion assisting in binding to the protein. Metal ion size has an effect on antibody affinity both through ionic radii and via the effect each metal has on the overall chelate complex conformation. This shape is directly related to the ion size and electronic properties

Antibody Recognition of Metal−Chelate Complexes

of the different metal ions. The information obtained from comparisons of the distance matrixes is superior to comparisons of ionic radii, because the DM compares the entire three-dimensional structure of the metal-chelate. We were able to demonstrate that Hg(II)- and Cd(II)EDTA complexes appear very similar in three-dimensional shape, and these structural differences correlate well with antibody binding (Figure 5). Blake et al. used only the difference in ionic radii to correlate the affinity of mAb 2A81G5 to the metal ions, and the difference in ionic radii for Hg(II) from Cd(II) did not correlate well to antibody binding. Studies are in progress to further characterize the structure of the antibody binding sites of A4 and E5 and to develop models that may also correlate the effects of the electronic properties of the metal ions in antibody recognition. ACKNOWLEDGMENT

The authors thank Drs. Thomas C. Bishop and Thomas E. Weise for advice and guidance on molecular modeling and for the use of the computing facilities in the Theoretical Molecular Biology Laboratory. Dr. Robert C. Blake II is acknowledged for his help in the binding studies and for his critical reading of this manuscript. This work was supported by a grant to D.A.B. from the Natural and Accelerated Bioremediation Research (NABIR) program, Biological and Environmental Research (BER), U.S. Department of Energy (grant #DE-FG02-98ER62704). The Theoretical Molecular Biology Laboratory at the Tulane/Xavier Center for Bioenvironmental Research was established under NSF Cooperative Agreement Number OSR-9550481/LEQSF (1996-1998)-SI-JFAP-04 and is also supported by the Office of Naval Research (NO-0014-99-1-0763). LITERATURE CITED (1) Glusker, J. P. (1991) Adv. Protein Chem. 42, 1-76. (2) Baker, B. L., and Hultquist, D. E. (1978) J. Biol. Chem. 253, 8444-8451. (3) Lindgarde, F., and Zettervall, O. (1974) Scand. J. Immunol. 3, 277-285. (4) Wylie, D. E., Lu, D., Carlson, L. D., Carlson, R., Babacan, K. F., Schuster, S. M., and Wagner, F. W. (1992) Proc. Natl Acad. Sci. U.S.A. 89, 4104-4108. (5) Reardan, D. T., Meares, C. F., Goodwin, D. A., McTigue, M., David, G. S., Stone, M. R., Leung, J. P., Bartholomew, R. M., and Frincke, J. M. (1985) Nature 316, 265-267. (6) Blake, D. A., Chakrabarti, P., Khosraviani, M., Hatcher, F. M., Westhoff, C. M., Goebel, P., Wylie, D. E., and Blake, R. C., II. (1996) J. Biol. Chem. 271, 27677-27685. (7) Khosraviani, M., Blake, R. C., II, Pavlov, A. R., Lorbach, S. C., Yu, H., Delehanty, J. B., Brechbiel, M. W., and Blake, D. A. (2000) Bioconjugate Chem. 11, 267-277.

Bioconjugate Chem., Vol. 13, No. 3, 2002 415 (8) Doussal, J. M. L., Chetanneau, A., Gruaz-Guyon, A., Martin, M., Gautherot, E., Lehur, P. A., Chatal, J. F., Delaage, M., and Barbet, J. (1993) J. Nuc. Med. 34, 1662-1671. (9) Wylie, D. E., Carlson, L. D., Carlson, R., Wagner, F. W., Schuster, S. M. (1991) Anal. Biochem. 194, 381-387. (10) Blake, D. A., Pavlov, A. R., and Khosraviani, M. (1998) in Current Protocols in Field Analytical Chemistry, 3C.1.13C1.10, John Wiley and Sons, Inc., New York. (11) Khosraviani, M., Pavlov, A. R., Flowers, G. C., and Blake D. A. (1998) Environ. Sci. Technol. 32, 137-142. (12) Darwish, I. A., and Blake, D. A. (2001) Anal. Chem. 73, 1889-1895. (13) Love, R. A., Villafranca, J. E., Aust, R. M., Nakamura, K. K., Jue, R. A., Major, J. G., Radakrishnan, R., and Butler, W. F. (1993) Biochemistry 32, 10950-10959. (14) Chakrabarti, P., Hatcher, F. M., Blake, R. C., II., Ladd, P. A., Blake, D. A. (1994) Anal. Biochem. 217, 70-75. (15) Cayot, P., and Tainturier, G. (1997) Anal. Biochem. 249, 184-200. (16) Habeeb, A. F. (1966) Anal. Biochem. 14, 328-36. (17) Mozdzanowski, J., Bongers, J., and Anumula, K. (1998) Anal. Biochem. 260, 183-187. (18) Blake, R. C., II., Pavlov, A. R., and Blake, D. A. (1999) Anal. Biochem. 272, 123-134. (19) Solans, X., Gali, S., Font-Altaba, M., Oliva, J., and Herrera, J. (1983) Acta Crystallogr. C39, 438-440. (20) Albinati, A., Gramlich, V., Anderegg, G., Lianza, F., Petter, W., and Bommeli, H. (1998) Inorg. Chem. Acta 274, 219-223. (21) Solans, X., Gali, S., Font-Altaba, M., Oliva, J., and Herrera, J. (1988) Afinidad 45, 243-248. (22) Agre, V. M., Kozlova, N. P., Trunov, V. K., and Ershova, S. D. (1981) Zh. Strukt. Khim. 22, 138-146. (23) Solans, X., Font-Altaba, M., and Garcia-Oricain, J. (1984) Acta Crystallogr. C40, 635-638. (24) Agre, V. M., Sysoeva, T. F., Trunov, V. K., Efremov, V. A., Fridman, A., and Barkhanova, N. N. (1980) Zh. Strukt. Khim. 21, 110-117. (25) Sysoeva, T. F., Starikova, Z. A., Leont′eva, M. V., and Dyatlov, N. M. (1990) Zh. Strukt. Khim. 31, 140-143. (26) Nesterova, Y. M., and Porai-Koshits, M. A. (1982) Koord. Khim. 8, 994-1001. (27) Paul, W. E., Ed. (1999) Fundamental Immunology, 4th ed., pp 58-60, Lippincontt-Raven Publishers, Philadelphia, PA. (28) Martell, A. E., and Smith, R. M., Eds. (1998) NIST Critically Selected Stability Constants of Metal Complexes version 5.0, NIST Standard Reference Data, Gaithersburg, MD 20899. (29) Kula, R. J., Sawyer, D. T., Chan, S. I., Finley, C. M. (1963) J. Am. Chem. Soc. 85, 2930-2936. (30) Day, R. J., and Reilley, C. N. (1964) Anal. Chem. 36, 10731076. (31) Brechbiel, M. W., Gansow, O. A., Pippin, C. G., Rogers, R. D., and Planalp, R. P. (1996) Inorg. Chem. 35, 6343-6348. (32) Rettig, S. J., and Trotter, J. (1973) Can. J. Chem 51, 1303-1312.

BC0155418