Structure-Binding Relationships for the Interaction between a

A series of vancomycin analogues and tracers were synthesized, and their binding interactions with an anti-vancomycin Fab fragment were evaluated unde...
0 downloads 0 Views 294KB Size
Download PDF
176

Bioconjugate Chem. 1999, 10, 176−185

Structure-Binding Relationships for the Interaction between a Vancomycin Monoclonal Antibody Fab Fragment and a Library of Vancomycin Analogues and Tracers Maciej Adamczyk,* Jonathan Grote, Jeffrey A. Moore, Sushil D. Rege, and Zhiguang Yu Diagnostics Division Organic Chemistry (9-NM), Abbott Laboratories, Building AP 20, 100 Abbott Park Road, Abbott Park, Illinois, 60064. Received November 6, 1998; Revised Manuscript Received December 23, 1998

A series of vancomycin analogues and tracers were synthesized, and their binding interactions with an anti-vancomycin Fab fragment were evaluated under mass transport limiting conditions using surface plasmon resonance detection. Differences observed in binding interactions were utilized to define the vancomycin structural elements critical for antibody recognition. Major structural regions of vancomycin shown to play an important role in anti-vancomycin Fab fragment recognition include two sugar moieties and one chlorinated phenyl ring. The N-methylleucyl residue, the carboxy terminal residue, and residues in the peptide-binding region of vancomycin have minimal impact on the antivancomycin Fab fragment/vancomycin binding interaction. The selection of an antibody with such binding properties plays a critical role in the development of a vancomycin immunoassay that employs stable calibrators and controls.

INTRODUCTION

The definition of a well-characterized immunoreagent has been expanded in recent years (1, 2). Haptens and tracers have been characterized traditionally for their identity and purity. Advances in electrophoresis, HPLC, and ESI and MALDI MS methods have permitted similar characterization of antibodies and immunoconjugates (1, 3, 4). Although important, such characterization is less critical to assay development than the structure-function relationships of the immunoreagents. This includes a more complete understanding of the molecular features critical for specific recognition between ligands (haptens and tracers) and antibodies. These molecular features can be identified through a rigorous evaluation of the binding interactions between antibodies and structurally modified haptens and tracers. A method utilizing automated surface plasmon resonance technology has recently been demonstrated to be useful for studying the binding interactions between a monoclonal antibody Fab fragment and structurally modified ligands (5, 6). This method involves an immobilized ligand and a soluble ligand free in solution competing for binding sites on a single antibody Fab fragment. If high-density immobilized ligand surfaces are employed and the association rate between the antibody Fab fragment and the immobilized ligand is g1 × 105 M-1 s-1, the observed binding will be limited by mass transfer (7, 8). Under such conditions, initial binding rates are directly proportional to the amount of free antibody Fab fragment (not bound to soluble ligand) in solution and independent of the interaction kinetics or analyte-ligand affinity. Measuring concentrations of free antibody Fab fragment in equilibrium solutions with different concentrations of structurally modified ligand allows one to determine binding affinities and relate these results to structural features critical for recognition. * To whom correspondence should be addressed. Phone: (847) 937-0225.Fax: (847)938-8927.E-mail: [email protected].

Vancomycin (1, Figure 1) is the antibiotic drug of choice for the treatment of Gram-positive infections caused by methicillin resistant Staphylococcus aureus (9). It is also the treatment of choice in bacterial infections in patients allergic to β-lactam antibiotics (10). For the safe and effective use of this drug, quantitation of its levels in patient blood is required to maintain therapeutic levels (11, 12). Immunoassay techniques requiring structurally modified vancomycin tracers are frequently employed for this purpose. However, the antibiotic is rich in functionality from which to covalently attach signal-generating moieties. Here, we describe the preparation of several vancomycin analogues and tracers and the binding studies carried out on an automated BIAcore 2000 designed to evaluate the structural regions of vancomycin critical to recognition by an anti-vancomycin Fab fragment and those vancomycin regions which can be structurally modified with minimal impact to the binding interaction. MATERIALS AND METHODS

Protein A Affinity-Pak columns and ImmunoPure Fab preparation kits, BupH phosphate-buffered saline packs, Slide-A-Lyzer 10K MWCO dialysis cassettes, the Micro BCA protein assay kit, and the Immunopure Mouse IgG F(ab′)2 fragment were obtained from Pierce (Rockford, IL). Microconcentrators were obtained from Millipore Corp. (Bedford, MA). Vancomycin (1) was obtained from the Chemical and Agricultural Products Division, Abbott Laboratories (Abbott Park, IL). Anti-vancomycin mAb was obtained from the Abbott cell culture facility (Abbott Park, IL) (13). (NR,N-diacetyl)KdAdA tripeptide was obtained from Peninsula Laboratories (Belmont, CA). Aminocaproate-derivatized (N-acetyl)KdAdA tripeptide was obtained from Research Genetics (Huntsville, Al). Biotin active ester (2) was prepared by the method of Wilchek (14). 6-Carboxyfluorescein active ester (3) was prepared by the method of Adamczyk et al. (15). Acridinium chemiluminescent label (4) was prepared following the general procedure of Mattingly et al. (16, 17).

10.1021/bc980135i CCC: $18.00 © 1999 American Chemical Society Published on Web 02/25/1999

Anti-Vancomycin mAb Recognition

Figure 1. Structure of vancomycin.

Desvancosaminylvancomycin (5), aglucovancomycin (6), and N-acetylvancosaminylvancomycin (7) were prepared by the method of Kannan et al. (18). Ring-2 dechlorovancomycin (8) was prepared by the method of Harris et al. (19). Crystalline degradation product (CDP, 10) was prepared by the method of Marshall (20). Ring-2 dechloroCDP (12) was prepared by the method of Harris et al. (19). All other reagents used in synthesis were obtained from Aldrich Chemical Co. (Milwaukee, WI) and utilized without further purification. Synthesized compounds were purified by HPLC [Waters (Millford, MA) Delta Prep 3000 Preparative Chromatography system equipped with a Lambda-Max 481 UV detector, a model 740 data module, and a 40 × 100 mm µBondapak C18 column]. Analytical HPLC was performed with the same system using an 8 × 100 mm µBondapak C18 column. HPLC columns were eluted with a linear gradient of 5 to 50% CH3CN in 50 mM ammonium formate (solvent A) or isocratically in aqueous CH3CN containing trifluoroacetic acid (v:v:v, CH3CN/H2O/TFA; solvent B) as noted. Elution profiles were recorded at 254 nm. Electrospray ionization mass spectrometry (ESI/MS) was carried out on a PerkinElmer (Norwalk, CT) Sciex API 100 Benchtop system employing the Turbo IonSpray ion source. Protein was analyzed by SDS-PAGE on a Bio-Rad Minigel system (Hercules, CA) utilizing 12.5% polyacrylamide gels (7 cm × 10 cm × 1 mm), followed by staining with Coomassie Blue. Surface plasmon resonance measurements were carried out on a BIAcore 2000 (BIAcore, Inc., Piscataway, NJ) automated system using CM-5 four-channel sensor chips. Reagents for the BIAcore instrument consisted of HBS buffer [10 mM Hepes (pH 7.4), 150 mM NaCl, 3.4 mM EDTA, and 0.05% surfactant P-20], a coupling kit containing N-hydroxysuccinimide (NHS), N-ethyl-N-(3diethylaminopropyl)carbodiimide (EDAC), and 1 M ethanolamine hydrochloride (pH 8.5), all from BIAcore, Inc. Preparation of AglucoCDP (11). Crystalline degradation product (CDP, 10) (20) (100 mg, 0.069 mmol) in 1 N HCl (3 mL) was heated on a water bath for 1 h. The mixture was cooled to ambient temperature, and the solid was filtered off. The crude product was dissolved in saturated NaHCO3, purified by preparative HPLC (solvent A over 20 min; flow rate, 45 mL/min), and lyophilized (45 mg, 57%). Analytical HPLC (solvent A over 20 min; flow rate, 2 mL/min): retention time, 9.4 min, 99%. ESI MS m/z: 1145 (MH+). General Procedure for Preparation of N-Vancosaminyl-Derived Vancomycin Tracers. To a solution of vancomycin (1) (482 mg, 0.33 mmol) in dry DMF (6 mL) were added active ester 2, 3, or 4 (0.36 mmol) and triethylamine (0.91 mL, 6.6 mmol). After stirring for

Bioconjugate Chem., Vol. 10, No. 2, 1999 177

24 h at ambient temperature under N2, the crude reaction mixtures were purified by preparative HPLC and lyophilized. Compound 13 was obtained from 1 and biotin-active ester (450 mg, 81%). Preparative HPLC: solvent A over 20 min; flow rate, 45 mL/min. Analytical HPLC (solvent A over 15 min; flow rate, 2 mL/min): retention time, 9.4 min, 99%. ESI/MS m/z: 1674 (MH+), 1305 (MH+ vancosaminylbiotin), 1143 (MH+ - disaccharide - biotin), 554 (disaccharide + biotin + Na+). Compound 14 was obtained from 1 and 6-carboxyfluorescein-active ester (20 mg, 11%). Preparative HPLC: solvent B, 30:70:0.1; flow rate, 45 mL/min. Analytical HPLC (solvent B, 30:70:0.1; flow rate, 2 mL/min): retention time, 4.0 min, 99%. ESI/MS m/z: 1809 (MH+), 904 (MH22+), 1305 (MH+ - vancosaminylfluorescein), 1143 (MH+ - disaccharide - fluorescein), 665 (disaccharide + fluorescein). Compound 15 was obtained from 1 and acridiniumactive ester (409 mg, 70%). Preparative HPLC: solvent A over 20 min; flow rate, 45 mL/min. Analytical HPLC (solvent A over 20 min; flow rate, 2 mL/min): retention time, 10.4 min, 99%. ESI/MS m/z: 2016 (MH+), 1305 (MH+ - vancosaminylacridinium), 1143 (MH+ - disaccharide - acridinium), 710 (disaccharide + acridinium). General Procedure for Preparation of N-Methylleucyl-Derived Vancomycin Tracers. To a solution of vancomycin (1) (296 mg, 0.20 mmol) in DMSO (10 mL) were added biotin or 10-(3-sulfopropyl)-N-tosyl-N-(3carboxypropyl)acridinium-9-carboxamide (0.20 mmol) and N-hydroxybenztriazole (33 mg, 0.24 mmol). Dicyclohexylcarbodiimide (206 mg, 1.0 mmol) was added, and the mixtures were stirred for 72 h at ambient temperature under N2. The crude reaction mixtures were purified by preparative HPLC and lyophilized. Compound 16 was obtained from 1 and biotin (136 mg, 40%). Preparative HPLC: solvent A over 20 min; flow rate, 45 mL/min. Analytical HPLC (solvent A over 20 min; flow rate, 2 mL/min): retention time, 9.4 min, 99%. ESI/MS m/z: 1675 (MH+), 1531 (MH+ - vancosamine), 1372 (MH+ - disaccharide), 838 (MH22+). Compound 17 was obtained from 1 and 10-(3-sulfopropyl)-N-tosyl-N-(3-carboxypropyl)acridinium-9-carboxamide (190 mg, 35%). Preparative HPLC: solvent A over 20 min; flow rate, 45 mL/min. Analytical HPLC (solvent A over 15 min; flow rate, 2 mL/min): retention time, 12.1 min, 99%. ESI/MS m/z: 2016 (MH+), 1874 (MH+ vancosamine), 1711 (MH+ - disaccharide), 694 (Nmethylleucylacridinium). General Procedure for Preparation of CarboxylDerived Vancomycin Tracers. (A) Compound 9 was prepared according to the method of Sundram and Griffin (21). Briefly, to a solution of vancomycin (1) (500 mg, 0.35 mmol) and hexanediamine hydrochloride (196 mg, 1.04 mmol) in anhydrous DMSO/DMF (v:v, 1:1, 8 mL) at 0 °C were added HBTU (262 mg, 0.69 mmol) and diisopropylethylamine (480 µL, 2.76 mmol). After stirring for 72 h at ambient temperature, the crude reaction mixture was purified by preparative HPLC (solvent B, 17:83:0.0; flow rate, 40 mL/min) and lyophilized (228 mg, 43%). Analytical HPLC (solvent B, 17:83:0.0; flow rate, 2 mL/min): retention time, 7.9 min, 99%. ESI/MS m/z: 1548 (MH+). (B) To a solution of compound 9 (19 mg, 12 µmol) in dry DMF (0.5 mL) were added active ester 2, 3, or 4 (12 µmol) and triethylamine (2 µL, 12 µmol). After stirring for 24 h at ambient temperature under N2, the crude reaction mixtures were purified by preparative HPLC and lyophilized.

178 Bioconjugate Chem., Vol. 10, No. 2, 1999

Compound 18 was obtained from 9 and biotin-active ester (9 mg, 31% based on recovered starting material). Preparative HPLC: solvent B, 20:80:0.05; flow rate, 40 mL/min). Analytical HPLC (solvent B, 20:80:0.05; flow rate, 2 mL/min): retention time, 6.1 min, 99%. ESI/MS m/z: 1797 (M + Na+), 1775 (MH+), 1632 (MH+ vancosamine), 1469 (MH+ - disaccharide). Compound 19 was obtained from 9 and 6-carboxyfluorescein-active ester (4 mg, 26% based on recovered starting material). Preparative HPLC: solvent B, 27:73: 0.05; flow rate, 40 mL/min. Analytical HPLC (solvent B, 27:73:0.05; flow rate, 2 mL/min): retention time, 7.5 min, 99%. ESI/MS m/z: 1929 (M + Na+), 1907 (MH+), 1764 (MH+ - vancosamine), 1600 (MH+ - disaccharide). Compound 20 was obtained from 9 and acridiniumactive ester (3 mg, 35% based on recovered starting material). Preparative HPLC: solvent B, 27:73:0.05; flow rate, 40 mL/min. Analytical HPLC (solvent B, 27:73:0.05; flow rate, 2 mL/min): retention time, 5.8 min, 99%. ESI/ MS m/z: 2137 (M + Na+), 2115 (MH+), 1972 (MH+ vancosamine), 1810 (MH+ - disaccharide). Anti-Vancomycin Monoclonal Antibody Preparation. Anti-vancomycin mAb, which was raised against vancomycin coupled through the carboxylate to thyroglobulin via a 4-aminobutyrate linker (13), was purified on an Affinity-Pak Protein A column according to the manufacturer’s protocol. Briefly, the cell culture medium containing mAb (25.4 mg) was clarified by centrifugation (3500g, 30 min) and filtered through a 0.2 µm filter, and the supernatant was applied to the Affinity-Pak Protein A column which had been equilibrated with 12 mL of IgG binding buffer. After washing with IgG binding buffer, the antibody was eluted with 6 mL of IgG elution buffer to a vial containing 1.0 mL of 1.5 M Tris buffer, pH 9.0. The purified IgG was dialyzed against PBS buffer (20 mM phosphate and 30 mM NaCl, pH 7.2) for 12 h at 4 °C and then concentrated to approximately 10 mg/mL in an Amicon Microcon-50 microconcentration device. The mAb concentration was 12.4 mg/mL as determined by the micro-BCA protein assay (22). The total recovery of the purified mAb was 18.6 mg. Anti-Vancomycin Fab Fragment Preparation. The digestion of anti-vancomycin mAb was carried out with the ImmunoPure Fab preparation kit following the manufacturer’s protocol. Briefly, purified anti-vancomycin mAb (∼12 mg) in digestion buffer (0.5 mL; 20 mM sodium phosphate, 10 mM EDTA, and 30 mM cysteine, pH 7.0) was incubated for 5 h in a 37 °C incubator shaker with immobilized papain in digestion buffer (0.5 mL). Papain-digested anti-vancomycin mAb was then passed through a preequilibrated immobilized protein-A column (2 mL) supplied with the kit to remove undigested mAb and the Fc fragment. The protein-A column was washed with an additional 13 mL of ImmunoPure binding buffer. The column flow-through and column washes containing the Fab fragment were pooled and dialyzed against PBS buffer for 12 h at 4 °C and concentrated in a Centriprep10 concentration device. The concentration of the antivancomycin Fab fragment was 2.0 mg/mL as determined by the micro-BCA method using mouse IgG (Fab′)2 as the standard. Purified anti-vancomycin Fab fragment was characterized by SDS-PAGE and LC/ESI mass spectrometry. The molecular masses of the heavy and light chains of the Fab fragment were 23 986 and 24 033 Da, respectively. All studies described were conducted with the Fab fragment of anti-vancomycin mAb to eliminate the complexity associated with the bivalency of the monoclonal antibody.

Adamczyk et al.

Preparation of Biosensor Surfaces. Immobilization of vancomycin analogue 9 or the aminocaproate-derivatized (N-acetyl)KdAdA tripeptide via amine coupling to the CM-5 sensor chip was performed by a previously described method (6). Briefly, a continuous flow of HBS buffer at 10 µL/min was initiated over the biosensor surface. The carboxymethylated dextran matrix on the sensor surface was activated by a 3.5 min injection of a solution of 0.05 M NHS and 0.2 M EDAC. A solution of 9 or the aminocaproate-derivatized (N-acetyl)KdAdA tripeptide (10 µM) and ethanolamine (990 µM; 1 mM total amine in HBS buffer) was then injected (7 min), followed by a 7 min injection of 1.0 M ethanolamine hydrochloride to block remaining unreacted active ester groups. Blank surfaces were generated under identical conditions omitting the ligand immobilization step. Solution Competition Analysis of the Vancomycin Analogue/Anti-Vancomycin Fab Fragment Binding Interaction. Solution competition studies were carried out on a BIAcore 2000 following the general procedure previously described (6). Biosensor surfaces were regenerated after each injection with successive 1 min pulses of 6, 6, and 1.5 M guanidine hydrochloride. Initial rates of binding of anti-vancomycin Fab fragment to the biosensor surface were measured over a 15 s window beginning 20 s postinjection. Data were evaluated by nonlinear regression analysis using the solution affinity model built into BIAevaluation 3.0 software (BIAcore, Inc). RESULTS

Preparation of Vancomycin Analogues and Tracers. Vancomycin analogues lacking the sugars or the ring-2 chlorine were prepared as previously described (18, 19). Tracers 13-15, containing a derivatized N-vancosaminyl carbohydrate moiety, were prepared in 1181% yield by coupling vancomycin with the NHS active esters of biotin (2), 6-carboxyfluorescein (3), or 10-(3sulfopropyl)-N-tosyl-N-(3-carboxypropyl)acridinium-9carboxamide (4) and purification by preparative HPLC (Scheme 1). Vancomycin tracers 16 and 17, bearing a derivatized N-methylleucyl moiety, were prepared in 40 and 35% yield, respectively, by coupling vancomycin with free biotin or acridinium acid in the presence of N,N′dicyclohexylcarbodiimide and N-hydroxybenztriazole (Scheme 1). Analogue 9, containing an aminoalkyl linker on the free carboxyl functionality of vancomycin, was prepared by the method of Sundram and Griffin (21). Coupling of 9 with the NHS active esters of biotin (2), 6-carboxyfluorescein (3), or 10-(3-sulfopropyl)-N-tosyl-N(3-carboxypropyl)acridinium-9-carboxamide (4), as above, provided mixtures of the carboxyl and N-vancosaminyl carbohydrate derivatized vancomycin tracers. Repeated purifications by preparative HPLC provided pure carboxyl-modified tracers 18-20 in 26-35% yield (Scheme 1). Crystalline degradation product analogues were prepared following literature procedures or the general methodology described for preparation of vancomycin analogues. The structures of all vancomycin analogues and tracers and the CDP analogues utilized in this study are summarized in Table 1. Preparation of an Immobilized Vancomycin Biosensor Surface. Initial attempts to generate a biosensor surface involved immobilization of vancomycin through the primary amine of the vancosamine sugar moiety via amine coupling to the activated carboxymethyl dextran surface of a CM-5 sensor chip. Subsequent binding studies with saturating amounts of anti-vancomycin Fab

Anti-Vancomycin mAb Recognition

Bioconjugate Chem., Vol. 10, No. 2, 1999 179

Scheme 1

fragment showed that these surfaces had minimal binding capacity and a relatively low affinity for the antibody fragment. In contrast, immobilization of vancomycin derivative 9, containing the aminoalkyl linker from the free carboxyl functionality, under identical conditions to the activated carboxymethyl dextran surface provided biosensors with relatively high binding capacity (RUmax ≈ 4000) and high affinity for anti-vancomycin Fab fragment. Additional binding studies of the anti-vancomycin Fab fragment to the immobilized aminoalkylmodified vancomycin surface demonstrated the binding

to be limited by mass transfer and suitable for use in the solution binding studies described below. Determination of Binding Affinities of Vancomycin Analogues and Tracers for Anti-Vancomycin Fab Fragment. The binding affinity of anti-vancomycin Fab fragment for several vancomycin analogues and tracers was determined from solution competition experiments. Initially, known concentrations of anti-vancomycin Fab fragment (0-22 nM) were injected over the aminoalkyl vancomycin biosensor surface, and the initial rate of binding for each anti-vancomycin Fab fragment

180 Bioconjugate Chem., Vol. 10, No. 2, 1999 Table 1. Structures of Vancomycin Analogues and Tracers

Adamczyk et al.

Anti-Vancomycin mAb Recognition

Bioconjugate Chem., Vol. 10, No. 2, 1999 181

Table 1 (Continued)

concentration was determined. A plot of initial binding rate versus concentration of anti-vancomycin Fab fragment was fit using a 4-parameter logistic (general model in BIAevaluation 3.0) providing a standard curve. Several standard curves were generated during the course of these studies and all were identical within experimental error. A fixed concentration of anti-vancomycin Fab fragment (20 nM) was then mixed with 12 concentrations of each vancomycin analogue or tracer and allowed to reach equilibrium. The equilibrium mixtures were individually injected over the aminoalkyl vancomycin biosensor surface, and the concentration of free anti-

vancomycin Fab fragment remaining was quantitated by determination of the initial rate of binding to the biosensor surface as described above. A plot of free antivancomycin Fab fragment versus total concentration of added analogue or tracer provides a competition curve. Figure 2 shows typical competition curves for studies conducted here. The data points represent the experimentally determined concentrations of free anti-vancomycin Fab fragment in equilibrium solutions at a given concentration of soluble analogue or tracer. The curves represent the best nonlinear fit of the data using eq 1 (solution affinity model in BIAevaluation 3.0 software)

182 Bioconjugate Chem., Vol. 10, No. 2, 1999

Adamczyk et al. Table 2. Structure-Binding Relationships for the Interaction between Vancomycin Analogues and the Anti-Vancomycin Fab Fragmenta

Figure 2. Solution competition curves for determination of equilibrium dissociation constants. Ring-2 dechlorovancomycin (8, b); Biotinylated carboxyl-HDAvancomycin tracer (18, O). The fixed concentration of anti-vancomycin Fab fragment was 20 nM.

FabT - A - KD + Fabf ) 2

x

(FabT - A - KD)2 - A × FabT (1) 4

where Fabf is the concentration of free anti-vancomycin Fab fragment in equilibrium solutions, FabT is the total concentration of anti-vancomycin Fab fragment in solution (20 nM), A is the total concentration of added vancomycin analogue or tracer, and KD is the equilibrium dissociation constant for the binding of the vancomycin analogue or tracer to anti-vancomycin Fab fragment in solution. The structure-binding relationships for the interaction between vancomycin analogues and the anti-vancomycin Fab fragment, as measured by changes in the equilibrium dissociation constants (KD), are summarized in Table 2. Vancomycin and the N-acetylvancosaminyl-derivatized vancomycin analogue (7) bind exceedingly tight with KD values being outside the range for which the BIAcore instrument can accurately provide values from solution competition studies. Incorporation of the aminoalkyl linker on the free carboxyl functionality of vancomycin, providing 9, which was used for preparation of the immobilized vancomycin biosensor surface, has minimal effect on the anti-vancomycin Fab fragment recognition in solution. Removal of the ring-2 chlorine atom from vancomycin results in a significant loss in binding recognition by the antibody fragment (cf. 8). Cleavage of one or both of the carbohydrate rings from vancomycin by acid hydrolysis providing 5 and 6, respectively, results in a further sequential loss in binding recognition by the antibody fragment with the loss of each monosaccharide. Anti-vancomycin Fab fragment binding interactions with vancomycin degradation products were extremely weak relative to the native antibiotic binding interaction. Crystalline degradation product (10) binds with a KD of 488 ( 34 nM. Removal of the chlorine atom from the 2-position again results in a significant loss in binding recognition by the anti-vancomycin Fab fragment and, complete hydrolysis of the carbohydrate rings, providing 11, results in a binding interaction with the antivancomycin Fab fragment too weak to be determined by the solution competition studies. The structure-binding relationships for the binding interaction between vancomycin tracers and the anti-

analogue

equilibrium dissociation constant (KD) (nM)

vancomycin (1) desvancosaminylvancomycin (5) aglucovancomycin (6) N-acetylvancosaminylvancomycin (7) ring-2 dechlorovancomycin (8) carboxyl-HDAvancomycin (9) CDP (10) AglucoCDP (11) ring-2 dechloroCDP (12)

e0.2b 587 ( 27 42 000 ( 1000 e0.2b 87 ( 4 0.32 ( 0.04 488 ( 34 g50 000c 6000 ( 100

a Values are average equilibrium dissociation constants obtained from triplicate measurements. b KD was too small to be reliably measured by solution affinity experiments. c KD was too large to be reliably measured by solution affinity experiments.

Table 3. Structure-Binding Relationships for the Interaction between Vancomycin Tracers and the Anti-Vancomycin Fab Fragmenta derivatization site (nM) derivative

N-vancosaminyl

biotin fluorescein acridinium

584 ( 19 76 ( 3 1200 ( 100

N-methylleucyl e0.2b c 25 ( 0.7

carboxyl-HDA e0.2b 204 ( 6 26 ( 0.9

a Values are average equilibrium dissociation constants obtained from triplicate measurements. b KD was too small to be reliably measured by solution affinity experiments. c Not determined.

vancomycin Fab fragment, as measured by changes in the equilibrium dissociation constants (KD), are summarized in Table 3. The binding interactions vary depending on the label contained on the tracer. However, N-methylleucyl- (16 and 17) and carboxyl-HDA-derived tracers (18-20) containing the same label bind the antibody fragment with similar affinities. In contrast, N-vancosaminyl-derived tracers (13-15) containing the equivalent label bind the antibody fragment substantially weaker. Evaluation of a Vancomycin/KdAdA Tripeptide Binding Interaction on Anti-Vancomycin Fab Fragment Recognition. To further evaluate the topology of vancomycin critical for anti-vancomycin Fab fragment recognition, the role of residues located in the peptidebinding pocket of the antibiotic was investigated (Figure 3). For these studies an (N-acetyl)KdAdA tripeptide containing an aminocaproate linker from the amino terminus was immobilized via amine coupling to an activated carboxymethyl dextran surface of a CM-5 sensor chip. Vancomycin (0.5 µM) binds to this surface (Figure 4). However, since the mass of the antibiotic is small, the instrumental response is relatively small (∼50 RU at equilibrium). In contrast, injection of a vancomycin (0.5 µM)/anti-vancomycin Fab fragment (1 µM) complex, in which g99% of the vancomycin is bound by the antibody fragment, results in an approximately 10-fold increase in response due the increased mass of the complex (Figure 4). Anti-vancomycin Fab fragment alone has no affinity for the immobilized tripeptide surface (Figure 4). To further verify that the peptide and antibodybinding pockets of vancomycin were mutually exclusive, the vancomycin/anti-vancomycin Fab fragment solution binding interaction was reinvestigated under the conditions described above in the presence of (Na,N-diacetyl)KdAdA tripeptide (500 µM). The KD value obtained for the vancomycin/anti-vancomycin Fab fragment binding interaction in solution in the presence of (Na,N-diacetyl)-

Anti-Vancomycin mAb Recognition

Figure 3. Proposed model for the binding interaction between vancomycin (1) and a cell wall peptide analogue (NR,N-diacetyl)KdAdA (28). Dashed lines indicate hydrogen bonds.

Figure 4. Binding of vancomycin, anti-vancomycin Fab fragment, and vancomycin/anti-vancomycin Fab fragment complex to aminocaproate-derivatized (N-acetyl)KdAdA tripeptide biosensor surface.

KdAdA tripeptide was identical to the value obtained in the absence of the tripeptide (e0.2 nM). DISCUSSION

Reagent design plays an important role in rational immunoassay development. Specifically, the molecular architecture of an immunogen used for triggering an antigenic response is critical to the production of antibodies exhibiting the desired ligand binding characteristics. Important features of immunogen design can include the attachment site of a spacer arm to the ligand, the structure of the spacer arm, and the method employed for coupling the hapten to a carrier molecule. Likewise, some molecular features of the complementary ligand and labeled tracers will be important for recognition by the elicited antibody. The goal of tracer design is to produce ligands containing a signal-generating moiety, which exhibit diagnostically useful binding characteristics. Two approaches to immunoreagent design have been investigated. The design of homologous immunoreagents involves preparation of an immunogen and a tracer from the same hapten. Alternatively, in design of heterologous immunoreagents the immunogen and tracer are prepared from haptens with distinct points of attachment. In the present study, several vancomycin tracers (homologous

Bioconjugate Chem., Vol. 10, No. 2, 1999 183

and heterologous) were prepared, and their binding interactions with an anti-vancomycin Fab fragment were investigated. Several vancomycin analogues were also prepared and tested to further evaluate the structural regions of the antibiotic critical for recognition by the antibody fragment. The inherent reactivity of vancomycin allows for the preparation of several analogues. Removal of the sugars from vancomycin occurs by mild acid hydrolysis (18, 20). Crystalline degradation product (CDP, 10) is formed in solution at ambient temperature by rearrangement of vancomycin with concomitant hydrolytic loss of ammonia (23). The antibiotic further contains several phenols that are susceptible to oxidative conditions and eight racemizable asymmetric centers. In contrast, the reactivity and complexity of vancomycin makes the preparation of tracers a challenge. As a result, literature reports of vancomycin tracers are limited, and no structural assignments have been provided (24-26). Tracers designed and synthesized here were derivatized on the vancosaminyl, N-methylleucyl, and carboxy terminal residues of vancomycin. The vancosaminyl and N-methylleucyl derivatized tracers (heterologous) were synthesized in good to moderate yield. The carboxyl derivatized tracers (homologous) were obtained in lower yields due to competing derivatization reactions yielding products, which were difficult to separate from the desired material. All tracers were isolated in >99% purity as determined by analytical HPLC. The structural assignments for all tracers were confirmed by mass spectral fragmentation patterns. Aminoalkyl modified vancomycin (9) was immobilized on biosensor surfaces and utilized for all solution competition experiments. Analogue 9 contains two primary amines that can potentially react with the activated carboxymethyl dextran of a CM-5 sensor chip to produce a heterogeneous surface. Results obtained using underivatized vancomycin immobilized surfaces indicate that if coupling does occur via the primary amine of the vancosamine sugar moiety it is minor. Coupling of 9 results in a vancomycin-immobilized surface with high affinity and high-binding capacity for anti-vancomycin Fab fragment. This suggests the majority of coupling occurs via the primary amine of the aminoalkyl linker conjugated to the carboxy terminus of vancomycin. The possible generation of heterogeneous vancomycin biosensor surfaces in this manner does not effect the solution competition experiments, since all studies were conducted under mass transport limiting conditions with concentrations of free antibody fragment in equilibrium solutions being determined from a standard curve. Solution competition studies show the antibody fragment binds vancomycin and the aminoalkyl modified vancomycin (9) with high affinity. However, hydrolysis of the two sugar moieties or removal of the ring-2 chlorine atom from vancomycin results in significant losses in binding affinity. The anti-vancomycin Fab fragment also exhibits minimal binding interactions with CDP analogues. In fact, binding interactions observed with CDP analogues and vancomycin analogues lacking the sugar moieties were outside the diagnostically relevant range for the vancomycin immunoassay (13). This is critical as assay specificity relies on the ability of the mAb to distinguish between vancomycin and its metabolites (i.e., CDP and other structurally similar compounds). The decreased binding recognition of the anti-vancomycin Fab fragment for CDP is attributed to a conformational change which occurs in the vancomycin to CDP transformation due to a ring expansion relieving the strain in

184 Bioconjugate Chem., Vol. 10, No. 2, 1999

Adamczyk et al.

Figure 5. Space filling model of vancomycin showing a proposed anti-vancomycin mAb binding region. Color scheme: carbons (grey), chlorines (green), oxygens (red), and nitrogens (blue). Structure adapted from PDB file 1AA5 using SymApps 5.1 (Bio-Rad Laboratories).

the peptide backbone of the macrocycle (23). Crystalline degradation product (CDP, 10) actually exists as an equilibrium mixture (2:3) of atropisomers differing in the orientation of the chlorine containing ring-2 residue (23). The orientation of the chlorine containing ring-2 residue in the major isomer differs from that in vancomycin by 180°. The minor isomer corresponds to the orientation found in vancomycin. The ring-2 orientation conversion is restricted in vancomycin. This reorientation of the chlorine atom in CDP can partially account for the decreased binding interaction observed with anti-vancomycin Fab fragment. Results obtained for the binding interactions between vancomycin tracers and the anti-vancomycin Fab fragment are consistent with the vancomycin analogue binding studies. The N-methylleucyl and carboxy terminal residues of vancomycin can be structurally modified with minimal impact to the anti-vancomycin Fab fragment binding interaction. In contrast, derivatization of the vancosaminyl sugar residue of vancomycin with biotin, fluorescein, or acridinium labels results in a substantial loss in binding recognition by the antibody fragment. Interestingly, acetylation of the same vancosaminylamine does not affect the binding interaction. This suggests that the antibody-binding pocket may accommodate small structural modifications at the amino functionality and that the amine is not directly involved in the binding interaction. Vancomycin manifests its antibiotic action by binding the carboxy terminal d-Ala-d-Ala peptides of the lipidPP-disaccharide-pentapeptide intermediates involved in the biosynthesis of the bacterial cell wall peptidoglycan (27). This inhibits cross-linking of the chains in the growing cell wall peptidoglycan making the cell susceptible to lysis through osmotic shock. Binding of the carboxy terminal d-Ala-d-Ala peptides to vancomycin has been shown to occur in the concave pocket of vancomycin through five hydrogen bonds between the peptide and the amide backbone of the antibiotic (Figure 3) (28). To investigate the role this region played, if any, in antivancomycin Fab fragment binding recognition, we evaluated the vancomycin/anti-vancomycin Fab fragment binding interaction in the presence of KdAdA model peptides. Direct binding studies of a vancomycin/anti-vancomycin

Fab fragment complex to an aminocaproate-derivatized (N-acetyl)KdAdA tripeptide biosensor surface demonstrated the peptide and antibody fragment could bind vancomycin simultaneously. This observation was further confirmed by determination of the vancomycin/antivancomycin Fab fragment solution binding affinity in the presence of (NR,N-diacetyl)KdAdA tripeptide. The results obtained were essentially identical to the value obtained in the absence of the tripeptide. The selection of an anti-vancomycin mAb that binds vancomycin at a site independent from that of the peptide-binding site is notable. The degradation of vancomycin to CDP is relatively facile and can lead to problems associated with stability of immunoassay components (i.e., calibrators, controls, and tracers) and cross reactivity with degradation products. Certain peptidoglycan analogues have been shown to inhibit the degradation of vancomycin to CDP resulting in reagent stabilization (29). The mAb selected and utilized here not only is highly specific for vancomycin but can also be used in the presence of peptides that stabilize the assay components without interfering with the immunoassay (13). A proposed antibody binding region consistent with all observations is shown in Figure 5. Major structural regions of vancomycin shown to be critical for recognition by the anti-vancomycin Fab fragment include the disaccharide moiety and the ring-2 chlorine atom. Additional structural features of vancomycin critical to recognition by the anti-vancomycin Fab fragment may exist but cannot be evaluated from the results obtained here. CONCLUSION

A series of vancomycin analogues and tracers were synthesized, and their binding interactions with an antivancomycin Fab fragment were evaluated under mass transport limiting conditions using surface plasmon resonance. The results show that two sugar moieties and one chlorinated phenyl ring play important roles in the antibody binding interaction. The N-methylleucyl and carboxy terminal residues have minimal contact with the antibody binding pocket and can be structurally modified to generate tracers with little impact on the binding interaction. Additionally, KdAdA model peptides were

Anti-Vancomycin mAb Recognition

shown to bind to vancomycin at a site independent from that of the mAb-binding site and as a result can be used to stabilize immunoassay components. LITERATURE CITED (1) Henry, C. (1996) FDA, reform, and the well-characterized biologic. Anal. Chem. 68, 674A-677A. (2) Adamczyk, M., Gebler, J. C., Gunasekera, A. H., Mattingly, P. G., and Pan, Y. (1997) Immunoassay reagents for thyroid testing. 2. Binding properties and energetic parameters of a T4 monoclonal antibody and its Fab fragment with a library of thyroxine analog biosensors using surface plasmon resonance. Bioconjugate Chem. 8, 133-145. (3) Adamczyk, M., Buko, A., Chen, Y. Y., Fishpaugh, J. A., Gebler, J. C., and Johnson, D. D. (1994) Characterization of protein-hapten conjugates. 1. Matrix-assisted laser desorption ionization mass spectrometry of immuno BSA-hapten conjugates and comparison with other characterization methods. Bioconjugate Chem. 5, 631-635. (4) Adamczyk, M., Gebler, J. G., and Mattingly, P. G. (1996) Characterization of protein-hapten conjugates. 2. Electrospray mass spectrometry of bovine serum albumin-hapten conjugates. Bioconjugate Chem. 7, 475-481. (5) Rauffer, N., Zeder-Lutz, G., Wenger, R., Van Regenmortel, M. H. V., and Altschuh, D. (1994) Structure-activity relationships for the interaction between cyclosporin A derivatives and the Fab fragment of a monoclonal antibody. Mol. Immunol. 31, 913-922. (6) Adamczyk, M., Johnson, D. D., Mattingly, P. G., Moore, J. A., and Pan, Y. (1998) Immunoassay reagents for thyroid testing. 3. Determination of the solution binding affinities of a T4 monoclonal antibody Fab fragment for a library of thyroxine analogs using surface plasmon resonance. Bioconjugate Chem. 9, 23-32. (7) Karlsson, R., Fagerstam, L., Nilshans, H., and Persson, B. (1993) Analysis of active antibody concentration. Separation of affinity and concentration parameters. J. Immunol. Methods 166, 75-84. (8) Karlsson, R. (1994) Real-time competitive kinetic analysis of interactions between low-molecular-weight ligands in solution and surface-immobilized receptors. Anal. Biochem. 221, 142-151. (9) Kirby, W. M. M. (1981) Vancomycin therapy in severe staphylococcal infections. Rev. Infect. Dis. 3, 5236-5239. (10) Ingerman, M. J., and Santoro, J. (1989) Vancomycin a new old agent. Infect. Dis. Clin. North Am. 3, 641-651. (11) Pryka, R. D., Rodvold, K. A., and Erdman, S. M. (1991) An updated comparison of drug dosing methods. IV. Vancomycin. Clin. Pharmacokinet. 20, 463-476. (12) Ryback, M. J., and Boike, S. C. (1986) Monitoring vancomycin therapy. Drug Intell. Clin. Pharm. 20, 757-761. (13) Adamczyk, M., Brate, E. M., Chiappetta, E. G., Ginsburg, S., Hoffman, E., Klein, C., Perkowitz, M. M., Rege, S. D., Chou, P. P., and Constantino, A. G. (1998) Development of a quantitative vancomycin immunoassay for the Abbott AxSYM Analyzer. Ther. Drug Monit. 20, 191-201.

Bioconjugate Chem., Vol. 10, No. 2, 1999 185 (14) Wilchek, M., and Bayer, E. A. (1990) Biotin-containing reagents. In Avidin-Biotin Technology (M. Wilchek, Ed.) pp 123-138, Academic Press, New York. (15) Adamczyk, M., Fishpaugh, J., and Heuser, K. (1997) Preparation of succinimidyl and pentafluorophenyl active esters of 5- and 6-carboxyfluorescein. Bioconjugate Chem. 8, 253-255. (16) Mattingly, P. G. (1991) Chemiluminescent 10-methylacridinum-9-(N-sulfonylcarboxamide) salts. Synthesis and kinetics of light emission. J. Biolumin. Chemilumin. 6, 107114. (17) Mattingly, P. G., and Bennett, L. (1995) Chemiluminescent acridinium salts, U.S. patent 5,468,646. (18) Kannan, R., Harris, C. M., Harris, T. M., Waltho, J. P., Skelton, N. J., and Williams, D. H. (1988) Function of amino sugar and N-terminal amino acid of the antibiotic vancomycin in its complexation with cell wall peptides. J. Am. Chem. Soc. 110, 2946-2953. (19) Harris, C. M., Kannan, R., Kopecka, H., and Harris, T. M. (1985) The role of chlorine substituents in the antibiotic vancomycin: Preparation and characterization of mono- and didechlorovancomycin. J. Am. Chem. Soc. 107, 6652-6658. (20) Marshall, F. J. (1965) Structure studies on vancomycin. J. Org. Chem. 8, 18-22. (21) Sundram, U. N., and Griffin, J. H. (1995) General and efficient method for the solution- and solid-phase synthesis of vancomycin carboxamide derivatives. J. Org. Chem. 60, 1102-1103. (22) Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner, F. H., Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk, D. C. (1985) Measurement of protein using bicinchoninic acid. Anal. Biochem. 150, 76-85. (23) Harris, C. M., Kopecka, H., and Harris, T. M. (1983) Vancomycin: Structure and transformation to CDP-I. J. Am. Chem. Soc. 105, 6915-6922. (24) Colbert, D. L., Smith, D. S., and Landon, J. (1993) Dissociation of fluorescein-labelled haptens from polyclonal antibodies. The influence of hapten molecular weight. J. Immunol. Methods 161, 269-272. (25) Schwenzer, K. S., Wang, C.-H. J., and Anhalt, J. P. (1983) Automated fluorescence polarization immunoassay for monitoring vancomycin. Ther. Drug Monit. 5, 341-345. (26) Fong, K.-L. L., Ho, D.-H. W., Bogerd, L., Pan, T., Brown, N. S., Gentry, L., and Bodey, G. P. (1981) Sensitive radioimmunoassay for vancomycin. Antimicrob. Agents Chemother. 19, 139-143. (27) Walsh, C. T., Fisher, S. L., Park, I.-S., Prahalad, M., and Wu, Z. (1996) Bacterial resistance to vancomycin: five genes and one missing hydrogen bond tell the story. Chem. Biol. 3, 21-28. (28) Barna, J. C. J., and Williams, D. H. (1984) Structure and mode of action of glycopeptide antibiotics of the vancomycin group. Annu. Rev. Microbiol. 38, 339-357. (29) Harris, C. M., Kopecka, H., and Harris, T. M. (1985) The stabilization of vancomycin with peptidoglycan analogs. J. Antibiot. 38, 51-57.

BC980135I