Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

P450 enzymes comprise a superfamily of heme-containing monooxygenases that mediate the metabolism of a host of compounds, from the transformation of natural endogenous substances such as steroids and vitamins to the metabolism of exogenous substances such as drugs, solvents, and environmental pollutants (Guengerich, 1996). The expression of these enzymes in various tissues has been studied extensively using antibodies (Gelboin, 1993; Boobis et al., 1996). Knowledge of the primary structures of a great number of the P450 enzymes allows the application of an antipeptide approach whereby antibodies are targeted against synthetic peptides representing a small unique region of an individual P450 enzyme (Boobis et al., 1996). Such antibodies have proved to be highly specific and have been applied in a wide variety of studies (Edwards et al., 1991, 1995, 1998; Debri et al., 1995; Schulz et al., 1998).

Inhibition of enzyme activity by antibodies is an effective method for determining the quantitative contribution of individual P450 enzymes to the metabolism of specific drugs in a particular tissue (Gelboin, 1993). To produce antipeptide antibodies that inhibit enzyme activity, it is necessary to target an appropriate region of the P450 enzyme of interest that is involved in the catalytic process. Although this has been achieved in a number of different studies, it is notable that in almost all cases there is some considerable variation in success (Edwards et al., 1990, 1991; Cribb et al., 1995; Duclos-Vallee et al., 1995; Laurenzana et al., 1995; Nakamura et al., 1995; Adams et al., 1997; Richardson et al., 1997; Wang et al., 1999). In some cases inhibition was incomplete; in some, large amounts of antiserum were needed to cause inhibition; and in others, a large interindividual response between antisera from different rabbits immunized was noted. One possible explanation for these results is that the resultant antibodies often bind only relatively poorly to the target protein in its native conformation. Peptide immunogens may adopt multiple conformations, many of which are not present in the native protein, but only those antibodies that bind to a peptide in a conformation similar to that of the target native protein are useful.

A number of previous studies have reported that antibodies raised against cyclic peptides recognize native proteins with a higher affinity than antibodies raised against linear peptides. In this way, strongly binding antibodies against lysozyme (Arnon et al., 1971), myoglobin (Dorrow et al., 1985), influenza virus hemagglutinin (Schulze-Gahmen et al., 1986; Muller et al., 1990), meningococcal class I outer membrane protein (Christodoulides et al., 1993; Hoogerhout et al., 1995), and HIV-2 envelope glycoprotein (Jrad and Bahraoui, 1997) have been produced. It is suggested that cyclization restricts the flexibility of the immunizing peptide, so its conformation in solution more closely mimics that of the target loop structure in the native protein.

Information on the structure of a proinhibitory site on the surface of CYP2D6 has been gained from epitope mapping studies using antisera from patients with type-1 autoimmune hepatitis that inhibit CYP2D6 activity (Gueguen et al., 1991; Manns, 1991). The sequence corresponding to residues 239 to 271 of CYP2D6 was recognized by all antisera tested with the core region comprising residues 261 to 263 (Gueguen et al., 1991). Models of the structure of eukaryotic P450 enzymes predict that the region encompassing residues 254 to 290 of human CYP2D6 lies between helices G and I (Edwards et al., 1989; Lewis, 1995) and comprises a largely hydrophilic loop region on the surface of the protein. Therefore, such a site may be suitable for targeting by conformationally restricted antibodies. Consequently, in the present study, both cyclic and linear peptides encompassing residues 254 to 290 of human CYP2D6 were used to target antibodies to CYP2D6. The ability of the resultant anticyclic and antilinear antibodies to bind to and inhibit the activity of CYP2D6 was assessed.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. Dimethylformamide and piperidine (both peptide synthesis grade) were obtained from Rathburn Chemicals (Walkerburn, UK). N--(9-Fluorenyl)methoxycarbonyl (Fmoc)²-protected amino acids linked to a Polyhipe/polyamide resin and Fmoc-protected amino acid pentafluorophenyl esters were purchased from Novabiochem (Nottingham, UK). Nucleosil C₁₈ 10-µm HPLC and DB5 0.25-mm high resolution gas chromatography columns were purchased from Jones Chromatography (Hengoed, UK). Sephadex G-15 and Sephadex G-25 were obtained from Pharmacia (Milton Keynes, UK). Keyhole limpet hemocyanin (KLH) was purchased from Calbiochem (Nottingham, UK). All SDS-polyacrylamide gel electrophoresis reagents were obtained from National Diagnostics (Aylesbury, UK), except for ammonium persulfate and prestained molecular weight standards, which were purchased from Sigma (Poole, UK). Protein G conjugated to horseradish peroxidase was obtained from Sigma. Hybond-C nitrocellulose membrane, enhanced chemiluminescence reagent, and Hyperfilm were purchased from Amersham International (Little Chalfont, UK). Immulon-I 96-well microtitre plates were obtained from Dynatech Laboratories (Billingshurst, UK). Samples of microsomal fractions prepared from Sf9 insect cells or human lymphoblastoid cells expressing various human P450 enzymes were purchased from Gentest Corporation (Woburn, MA). Debrisoquine and 4-hydroxydebrisoquine were kind gifts from Roche (Welwyn Garden City, UK). All other chemicals were purchased from Sigma or Merck-BDH (Lutterworth, UK) and were of analytical grade or the best equivalent.

Preparation of Microsomal Fraction from Human Liver. Histologically normal human liver samples from renal transplant donors were obtained from the human tissue bank, maintained in the Section on Clinical Pharmacology at the Imperial College School of Medicine. The livers had been cut into small pieces, snap-frozen in liquid nitrogen, and stored at 70°C. Permission to use these samples in these studies had been obtained from the Local Research Ethical Committee and had Coroner's approval. Microsomal fractions were prepared by differential centrifugation using the method of Boobis et al. (1980). The protein content of each microsomal fraction was determined by the method described by Lowry et al. (1951) using BSA (fraction V) as standard.

Peptide Synthesis. Peptides (Table 1) were synthesized according to the method published previously (Edwards et al., 1991) using a NovaSyn Gem semiautomated peptide synthesizer (Novabiochem). Synthesis was carried out by Fmoc solid-phase chemistry. Polyhipe supports consisting of polydimethylacrylamide functionalized with ethylenediamine, norleucine, and the acid labile linker 4-hydroxymethylphenoxyacetic acid were used. A cysteine residue was incorporated at the N terminus of each linear peptide to allow conjugation to KLH. For cyclic peptides, acetamidomethyl cysteine residues were introduced at the C terminus and N-terminal to the immunizing sequence of the peptide and for conjugation to KLH, the dipeptide Lys-Gly was added to the N terminus of each peptide (Fig. 1).

View this table: [in this window] [in a new window]   TABLE 1 Aligned sequences of CYP2D6 and the various synthetic peptides studied The CYP2D6 amino acid sequence was predicted from its cDNA sequence and is numbered from the N-terminal methionine. The synthetic peptides are numbered with respect to human CYP2D6. The consensus core sequence of the epitope of anti-CYP2D6 antibodies identified by Manns (1991) and Gueguen et al. (1991) is shown in bold. The residues of CYP2D6 identified by Gueguen et al. (1991) as essential for antibody binding are underlined. Cyclic peptides were formed by oxidation of the two cysteine residues in their sequence and then coupled to KLH through the N-terminal lysine residue (Fig. 1). Linear peptides were coupled to KLH through their N-terminal cysteine residue.

View larger version (6K): [in this window] [in a new window]   Fig. 1.   The structure of a cyclized peptide conjugated to carrier protein. Formation of the disulfide bridge was achieved by iodine oxidation as described in Experimental Procedures. Conjugation of the cyclized peptide to carrier protein was achieved using glutaraldehyde as described in Experimental Procedures.

Cyclic Peptide Formation. For the formation of cyclic peptides (Fig. 1) a solution of 5 µmol/ml acetamidomethyl-protected peptide in 15.75 M acetic acid was added to a 50 mM iodine solution containing 15.75 M acetic acid (total volume 22.5 ml). The mixture was stirred for 30 min after which 1 ml of 1 M sodium thiosulfate was added to stop the reaction. The mixture was concentrated by rotary evaporation under reduced pressure and then applied directly to a Sephadex G-15 column (33 × 1.6 cm) equilibrated in 0.5 M acetic acid. The purified products were >90% pure, as determined by reversed phase HPLC under the conditions described previously (Edwards et al., 1991). The success of the oxidative cyclization was also confirmed from the molecular weight of the product as determined by electrospray mass spectrometry.

Coupling to Carrier Protein. Linear peptides were coupled through their N-terminal cysteine to KLH using the cross-linking agent m-maleimidobenzoyl-N-hydroxysuccinimide ester as described by Liu et al. (1979). Cyclic peptides were coupled through their N-terminal lysine to KLH using the cross-linking reagent glutaraldehyde as described by Mao et al. (1989).

Immunization. Antibody production was carried out commercially by Regal Group UK (Great Bookham, UK). Male or female New Zealand White rabbits (3 kg) were immunized with peptide conjugates by repeated injection of 200 µg of conjugate in Freund's adjuvant in a total volume of 1 ml as described previously (Edwards et al., 1991).

Enzyme-Linked Immunosorbent Assay (ELISA). Binding of antibodies to peptides was determined by ELISA under the conditions described previously (Schulz-Utermoehl et al., 1999), except that poly-L-lysine used to coat the wells of microtitre plates was activated by the addition of 100 µl of 25 mM bis(N-succinimidyl) carbonate for 15 min at 37°C instead of activation by glutaraldehyde. The plates were washed once with PBS and then incubated with 100 µl of peptide (10 µg/ml in PBS) for 1 h at 37°C to allow coupling to poly-L-lysine. For investigations into the relative binding of antibodies to microsomal protein from insect cells expressing human CYP2D6, wells of microtitre plates were coated directly with 10 µg/ml of recombinant CYP2D6 in PBS. Nonspecific binding sites were then blocked with 2% (w/v) BSA in PBS for 1 h at 37°C, and antibody binding was determined as described previously (Edwards et al., 1991).

Immunoblotting. SDS-polyacrylamide gel electrophoresis and immunoblot analysis were carried out as described by Edwards et al. (1998), using microsomal protein from liver or from insect or lymphoblastoid cells containing recombinant P450 enzymes. The amount of each recombinant P450 enzyme used was sufficient for detection using this technique (Edwards et al., 1998). The immunoblots were developed using the antipeptide antibodies raised against each CYP2D6 enzyme (Table 1) for 2 h at room temperature. Binding of these antibodies to the target protein was detected after incubation of the nitrocellulose filter with Protein G coupled to horseradish peroxidase (12.5 ng/ml) and visualized by enhanced chemiluminescence and exposure to Hyperfilm.

Debrisoquine 4-Hydroxylase. This activity of hepatic microsomal fraction was determined by the method of Boobis et al. (1983) using a final debrisoquine concentration of 20 µM. Hepatic microsomal fraction (25 µg) was incubated for 30 min at room temperature with varying amounts of preimmune or immune serum (25, 50, 100, and 200 µl) in a total volume of 0.84 ml. Buffer and cofactors were then added, before the reaction was started with the addition of the substrate (final volume 1 ml). After 45 min at 37°C, the reaction was stopped by the addition of 0.2 ml of 1 M sodium hydroxide. ²H₄ 4-Hydroxydebrisoquine in water (1 µg/ml, 100 µl) was added to each sample. Unmetabolized debrisoquine was removed by extraction with chloroform (3× 3 ml). Samples were neutralized by the addition of 1 M HCl (150 µl). 4-Hydroxydebrisoquine and the deuterated internal standard in the samples were then converted to their 3,5-bistrifluoromethylpyrimidinyl derivatives by adding saturated NaHCO₃ (200 µl), redistilled toluene (1 ml), and hexafluoroacetylacetone (50 µl) and heating in a boiling water bath (100°C) for 1 h. After cooling and centrifugation, the upper organic layers present in the samples were transferred to glass vials and evaporated to dryness under nitrogen. Bis(trimethylsilyl)trifluoroacetamide (50 µl) was added to each vial, and the vials were securely capped and allowed to stand at room temperature overnight. This reagent trimethylsilylates the free hydroxyl groups present in the bistrifluoromethylpyrimidinyl derivatives of 4-hydroxydebrisoquine and ²H₄ 4-hydroxydebrisoquine, resulting in improved chromatographic behavior, and is a modification to the original published assay (Boobis et al., 1983). Just before analysis by gas chromatography-mass spectrometry, vial contents were evaporated to dryness under nitrogen, and the residue was reconstituted in n-undecane (50 µl). Aliquots (2 µl) were then injected into the gas chromatograph-mass spectrometer.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

The relative binding of antilinear peptide and anticyclic peptide antibodies to their corresponding linear and cyclic peptides was examined by ELISA (Fig. 2). The anti-CYP2D6_254-273, anti-CYP2D6_261-272, and anti-CYP2D6_278-290 antibodies bound more strongly to their immunizing peptides than to their corresponding cyclic peptides (Fig. 2). On the other hand, the anti-CYP2D6_{cyclic 254-273}, anti-CYP2D6_{cyclic 261-272}, and anti-CYP2D6_{cyclic 278-290} antibodies bound similarly to their respective immunizing peptides and their corresponding linear peptides (Fig. 2). No antibody binding to any of the peptides was detected when preimmune sera were used in ELISA (data not shown).

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Comparison of the binding of antilinear and anticyclic peptide antibodies to linear and cyclic peptides. Wells of microtitre plates were coated with poly-L-lysine and coupled to the peptides CYP2D6cyclic 254-273 (), CYP2D6254-273 (), CYP2D6cyclic 261-272 (), CYP2D6261-272 (), CYP2D6cyclic 278-290 (), CYP2D6278-290 (), and a structurally unrelated peptide VPIGVPHRVTKD () using bis(N-succinimidyl) carbonate. A range of dilutions of anti-CYP2D6254-273 (a), anti-CYP2D6cyclic 254-273 (b), anti-CYP2D6261-272 (c), anti-CYP2D6cyclic 261-272 (d), anti-CYP2D6278-290 (e), or anti-C1Y1P2D6cyclic 278-290 (f) antibody was then added to each series of wells, and antibody binding was determined as described in Experimental Procedures.

All of the antilinear and anticyclic peptide antibodies recognized a single immunoreactive protein, with an apparent molecular mass of 54 kDa in microsomal preparations of samples of human liver containing CYP2D6 (Fig. 3), whereas no bands were detected using preimmune sera (data not shown). The relative intensity of the immunoreactive band obtained with each antibody corresponded to the debrisoquine 4-hydroxylase activity of the four samples (data not shown). The specificity of each of the antibodies for CYP2D6 was also determined by immunoblotting using samples of microsomal fraction prepared from human B-lymphoblastoid or insect cells that expressed a single form of human P450 enzyme. Each of the antibodies recognized a single immunoreactive protein in the sample containing human CYP2D6 (Fig. 3). The reasons for the aberrant migration of recombinant CYP2D6 from the source used have been discussed previously (Edwards et al., 1998). No immunoreactive bands were detected in samples of microsomal fraction from control cells or cells expressing human CYP1A1, CYP1A2, CYP2A6, CYP2B6, CYP2C9-Arg, CYP2C19, CYP2E1, or CYP3A4 (Fig. 3).

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Binding of the anti-CYP2D6cyclic 257-268 antibody to recombinant P450 enzymes. Immunoblot of microsomal protein from lymphoblastoid or insect cells expressing recombinant (r) P450 enzymes. The following amounts of each sample of microsomal protein were applied to each lane: 25 µg microsomal fraction from a human liver sample (HLM), 10 µg of untransfected lymphoblastoid cells (Control M), 10 µg of untransfected insect cells (Control S), lymphoblastoid cells expressing recombinant CYP2A6 (10 µg), CYP2B6 (10 µg), CYP2D6 (10 µg), CYP2E1 (5 µg), and CYP3A4 (10 µg), insect cells expressing recombinant CYP1A1 (2 µg), CYP1A2 (10 µg), CYP2C9-Arg (5 µg), and CYP2C19 (5 µg). The immunoblot was developed with the anti-CYP2D6cyclic 257-268 antibody at a dilution of 1:1000. Antibody binding was determined as described in Experimental Procedures. Only the central section of the blot containing immunoreactive bands is shown. The anti-CYP2D6cyclic 257-268 antibody bound only to CYP2D6; similar results were obtained with each of the antibodies raised against both linear and cyclic peptides, except that the intensity of the immunoreactive bands varied.

The relative intensity of the immunoreactive bands using antibodies raised against linear and cyclic peptides was determined. When compared with antibodies raised against the corresponding linear peptides, the anti-CYP2D6_{cyclic 254-273}, anti-CYP2D6_{cyclic 261-272}, and anti-CYP2D6_{cyclic 278-290} antibodies all bound more strongly to CYP2D6 in hepatic microsomal fractions from human liver and transfected lymphoblastoid cells (Fig. 4). However, the similarity of binding to the respective immunizing peptides (Fig. 2) indicates that there is little variation in the amount of antipeptide IgG in each sample of antiserum.

View larger version (57K): [in this window] [in a new window]   Fig. 4.   Comparison of the binding of anticyclic and antilinear peptide antibodies directed against human CYP2D6. Samples of 25 µg microsomal fraction from four human livers and 10 µg microsomal protein from lymphoblastoid cells expressing recombinant CYP2D6 were subjected to immunoblotting using antibodies raised against either linear (A1-3) or cyclic (B1-3) peptides. Replicate blots were developed with anti-CYP2D6254-273 (A1), anti-CYP2D6cyclic 254-273 (B1), anti-CYP2D6261-272 (A2), anti-CYP2D6cyclic 261-272 (B2), anti-CYP2D6278-290 (A3), or anti-CYP2D6cyclic 278-290. (B3) antibodies using antiserum diluted 1:1000 and immunoreactive bands visualized as described in Experimental Procedures. Under these conditions, the intensity of the immunoreactive bands produced using the anti-CYP2D6254-273 and the anti-CYP2D6278-290 antibodies were too weak to be clearly detected, so blots developed with antiserum diluted 1:500 are shown to demonstrate that binding to CYP2D6 did occur, albeit weakly. In each case only the central section of each immunoblot containing immunoreactive protein is shown. Development of replicate blots with each of the respective preimmune sera produced no immunoreactive bands.

The binding of each of the antibodies to native protein was examined by ELISA. Wells of microtitre plates were coated with microsomal protein from insect cells expressing human CYP2D6. Antibodies against all of the linear and cyclic peptides recognized CYP2D6 (Fig. 5). In comparison with the antibodies raised against linear peptides, the corresponding anticyclic peptide antibodies bound more strongly (10- to 100-fold) to CYP2D6 in ELISA (Fig. 5). Denaturation of the microsomal protein by treatment with 8 M urea had no effect on subsequent binding of any of the antibodies (data not shown).

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Relative binding of antipeptide antibodies raised against various peptides of human CYP2D6 to recombinant CYP2D6. Wells of microtitre plates were coated with microsomal fraction from insect cells expressing recombinant CYP2D6. A range of dilutions of preimmune serum () or antipeptide antibody raised against CYP2D6cyclic 254-273 (), CYP2D6254-273 (), CYP2D6cyclic 257-268 (), CYP2D6cyclic 261-272 (), CYP2D6261-272 (), CYP2D6263-270 (), CYP2D6cyclic 278-290 (), and CYP2D6278-290 () was added to each series of wells, and antibody binding was determined as described in Experimental Procedures. Each point is the mean of two determinations, and the data shown are from a typical experiment representative of two experiments with similar results.

The effect of the antipeptide antibodies on P450-dependent monooxygenase activity of human hepatic microsomal fraction was determined using debrisoquine as the probe substrate, whereas none of the antibodies raised against linear peptides had any appreciable effect on debrisoquine 4-hydroxylation (Table 2), the anti-CYP2D6_{cyclic 254-273}, anti-CYP2D6_{cyclic 257-268}, and anti-CYP2D6_{cyclic 261-272} antibodies progressively inhibited debrisoquine 4-hydroxylation by human hepatic microsomal fraction with increasing amounts of antibody. Compared with preimmune serum, maximum inhibitions of 60 and 75% were achieved with the anti-CYP2D6_{cyclic 254-273} and anti-CYP2D6_{cyclic 261-272} antibodies, respectively (Fig. 6), whereas virtually complete inhibition of enzyme activity (91%) was obtained with the anti-CYP2D6_{cyclic 257-268} antibody (Fig. 6). Furthermore, the volume of antiserum required to produce maximum inhibition was considerably lower for the anti-CYP2D6_{cyclic 257-268} antibody than with the other antisera (Fig. 6).

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Immunoinhibition of debrisoquine 4-hydroxylase activity of human hepatic microsomal fraction by antibodies raised against cyclic peptides. Microsomal protein (25 µg) was preincubated for 30 min at room temperature with varying volumes of antiserum raised against CYP2D6cyclic 254-273 (), CYP2D6cyclic 257-268 (), CYP2D6cyclic 261-272 (), or CYP2D6cyclic 278-290 (). The reaction was then started by the addition of NADPH and debrisoquine and incubated for 45 min at 37°C. The activities shown for each point are relative to that obtained after the addition of the same volume of preimmune serum. Debrisoquine 4-hydroxylase activity was measured as described in Experimental Procedures. The control debrisoquine 4-hydroxylase activity of human hepatic microsomal fraction (with no serum added) was 162 pmol/min/mg protein.

The relative binding of the anti-CYP2D6_{cyclic 257-268} antibody to structurally related peptides was investigated by ELISA to determine the essential residues of the targeted epitope (Fig. 7). Compared with the binding to its immunizing peptide CYP2D6_{cyclic 257-268}, only a slight reduction in binding was found to peptides CYP2D6_{cyclic 254-273} and CYP2D6_{cyclic 261-272}, whereas binding to peptide CYP2D6_263-270 was reduced by approximately 10-fold. No binding to the structurally unrelated peptide CYP2D6_{cyclic 278-290} was detected (Fig. 7).

View larger version (19K): [in this window] [in a new window]   Fig. 7.   Relative binding of the anti-CYP2D6cyclic 257-268 antibody to structurally related peptides. Wells of microtitre plates were coated with poly-L-lysine and coupled to the peptides CYP2D6cyclic 254-273 (), CYP2D6cyclic 257-268 (), CYP2D6cyclic 261-272 (), CYP2D6263-270 (), and CYP2D6cyclic 278-290 () using bis(N-succinimidyl) carbonate. A range of dilutions of anti-CYP2D6cyclic 257-268 antibody was added to each series of wells and antibody binding was determined as described in Experimental Procedures. Each point is the mean of two determinations, and the data shown are typical of two experiments with similar results.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

In this study, a series of anticyclic and antilinear peptide antibodies were targeted against CYP2D6. It was found that cyclic peptides were superior to linear peptides as immunogens for the production of antibodies that bind to CYP2D6 (Table 2). Antibodies raised against the peptides that were constrained by cyclization bound more strongly to CYP2D6 under both native (ELISA) and denatured (immunoblotting) conditions compared with antibodies raised against unconstrained peptides (linear peptides). Indeed, only antibodies against cyclic peptides were able to inhibit the activity of CYP2D6. It has been suggested, based on physiochemical and immunochemical studies, that cyclized peptides are more immunogenic and antigenic than linear peptides, due to the restriction of their conformational flexibility by cyclization (Furie et al., 1975; Nagy et al., 1982).

Although antibodies raised against linear and cyclic peptides bound equally well to their respective immunizing peptides, all of the antibodies raised against linear peptides bound relatively poorly to native CYP2D6. For the purpose of immunization, each linear peptide was coupled through its N terminus to KLH. In this form the peptides would be highly flexible, as each residue within the peptide will have considerable rotational freedom. It is possible that such peptides will exist in a number of different conformations or no stable conformation. Thus, these peptides may not necessarily mimic the conformation of the target region in the native protein. Consequently, antibodies may be directed against irrelevant peptide conformations. It has been shown previously that antibodies directed against linear peptides representing internal sequences of proteins consist of at least two important populations (Edwards et al., 1995). The predominant population binds strongly to the immunizing peptide at an epitope that includes the free terminal residue. However, these antibodies do not bind to the target protein. It is the minor fraction of antibodies that bind to epitopes that exclude the free terminal residue that bind to the protein (Edwards et al., 1995). Therefore, antibodies raised against linear peptides are not optimally targeted for binding to proteins.

Cyclization of peptides greatly restricts their intramolecular movements and effectively locks them into a loop configuration. In this form, each residue is unable to rotate independently of the other residues within the peptide, thus conformational flexibility of the peptide is highly restricted. Provided that the peptide represents a loop region in the target protein, it is suggested that its conformation will closely mimic that of the equivalent region in the native protein. This would explain why antibodies raised against cyclic peptides bind well to the target protein. Interestingly, the antibodies also bound strongly to the denatured protein, suggesting that the conformation of the epitope is similar in native and denatured protein. Also, the cyclic configuration of the peptide renders both of its ends inflexible. Consequently, the strong immunological response to the uncoupled end of linear peptides is avoided. The results of this study support this notion, as antibodies raised against linear peptides bound more strongly to their respective linear peptides, where the C terminus is uncoupled, than their corresponding cyclic peptides, in which the C terminus is coupled through the disulfide linkage. In contrast, antibodies raised against cyclic peptides bound equally to both linear and cyclic versions of the same peptide.

The application of cyclic peptides as immunogens to produce antibodies that bind to a protein requires that a loop region of the protein is targeted. The region of CYP2D6 chosen in this study is predicted from models of the three-dimensional structure of P450 to be a loop region N-terminal to the I-helix (Edwards et al., 1989; Lewis, 1995). This region contains three proline residues that are also conserved in other CYP2D enzymes. Proline residues are frequently associated with loop regions (Leszczynski and Rose, 1986).

The antibodies raised against peptides CYP2D6_{cyclic 254-273}, CYP2D6_{cyclic 257-268}, and CYP2D6_{cyclic 261-272} inhibited debrisoquine 4-hydroxylase activity of human hepatic microsomal fraction by a maximum of 60, 91, and 75%, respectively. However, antibodies raised against the linear peptides CYP2D6_254-273, CYP2D6_261-272, and CYP2D6_263-270 were unable to inhibit enzyme activity of hepatic microsomal fraction, even though two of these antibodies were raised against the same target sequences used in the cyclic peptides. Similarly, Duclos-Vallee et al. (1995) found that an antibody raised against a linear peptide corresponding to residues 251 to 271 of CYP2D6 failed to inhibit the O-demethylation of dextromethorphan of human hepatic microsomal fraction. Although Cribb et al. (1995) were successful in producing an antibody against a linear peptide corresponding to residues 254 to 273 of CYP2D6 that inhibited debrisoquine 4-hydroxylase activity of human hepatic microsomal fraction up to 90%, it was necessary to use a protracted immunization schedule. Furthermore, antisera raised in three other rabbits immunized with the same peptide had only a moderate effect on enzyme activity. The extent of variation in the properties of antibodies produced from different rabbits has not been explored here. However, this has been investigated in a separate study in which groups of rabbits were immunized with cyclic and linear peptide versions of residues 291 to 302 of CYP1A2. It was found that immunization with the cyclic peptide consistently produced higher affinity, proinhibitory antibodies (L. Peters, A.R. Boobis, and R.J. Edwards, manuscript in preparation).

The ability of antibodies raised against cyclic peptides CYP2- D6_{cyclic 254-273} and CYP2D6_{cyclic 261-272} rather than the corresponding linear peptides to inhibit enzyme activity is most likely explained by the stronger binding of the anticyclic peptide antibodies to CYP2D6 compared with the antibodies raised against the corresponding linear versions of the peptides. However, inhibition is not simply a product of the strength of binding of the antibodies, as the binding of anti-CYP2- D6_{cyclic 257-268} and anti-CYP2D6_{cyclic 254-273} antibodies to CYP2D6 are similar and yet the anti-CYP2D6_{cyclic 257-268} antibody is a more potent inhibitor of CYP2D6 activity. Therefore, as expected, the precise epitope to which the antibodies bind is also important for producing inhibition of enzyme activity.

The major epitope for the proinhibitory anti-CYP2D6_{cyclic 257-268} antibody was determined by examining the binding of this antibody to structurally related peptides. Compared with the binding to its immunizing peptide, CYP2D6_{cyclic 257-268}, the antibody bound similarly to the peptide CYP2D6_{cyclic 254-273}, which contains the complete epitope of CYP2D6_{cyclic 257-268}, and to the peptide CYP2D6_{cyclic 261-272}, which lacks the four N-terminal residues of the immunizing peptide. In contrast, binding to the peptide CYP2D6_263-270, which lacks an additional two residues at the N terminus, was reduced by 10-fold. It is unlikely that antibody binding could occur at the C termini of any of these peptides, as these regions do not overlap with the epitope for CYP2D6_{cyclic 257-268}. Therefore, these data indicate that the residues at positions 261 and 262 are involved in the binding of the anti-CYP2D6_{cyclic 257-268} to CYP2D6. Interestingly, Gueguen et al. (1991) determined that the essential part of the epitope of proinhibitory anti-CYP2D6 antibodies present in patients with type-1 autoimmune hepatitis antibodies comprised the tripeptide Thr-Trp-Asp (residues 261-263). This finding was supported by site-directed mutagenesis studies carried out by Yamamoto et al. (1993), where deletion or substitution of the negatively charged aspartate residue (at position 263) by another negatively charged amino acid, glutamate, by a neutral amino acid, asparagine, or by a basic amino acid, arginine, resulted in lack of recognition of CYP2D6 by LKM-1 autoantibodies.

The function of the proinhibitory region is unknown at present. It may involve interference in the interaction of P450 with NADPH-cytochrome P450 reductase and, thereby, electron transfer. Alternatively, antibody binding may result in a change in the conformation of the region around the substrate binding pocket or substrate access channel, making the active site inaccessible to a substrate molecule. These possibilities need to be explored in future studies.

A region adjacent to the I-helix has also been identified as proinhibitory in rat CYP1A1, rat CYP1A2, and human CYP1A2 (Edwards et al., 1990, 1991; Adams et al., 1997). However, alignment based on the primary structure of these sequences with CYP2D6 appears to indicate that the position of the proinhibitory region in CYP2D6 varies from that found in CYP1A1 and CYP1A2 (Murray et al., 1993). On the other hand, it has been proposed by Murray et al. (1993) that alignment based on secondary structure is a more accurate way of comparing P450 enzymes from different families and this method of alignment indicates that the proinhibitory region of CYP1A1, CYP1A2, and CYP2D6 are coincident. To investigate this further, antibodies were raised against the region of CYP2D6 (residues 278-290) that, based on primary structure alignment, is predicted to correspond to the proinhibitory region of CYP1A1 and CYP1A2. Thus, antibodies against the peptides CYP2D6_{cyclic 278-290} and CYP2D6_278-290 were produced. However, neither antibody had any inhibitory effect on debrisoquine 4-hydroxylase activity measured in human hepatic microsomal fraction. Because the anti-CYP2- D6_{cyclic 278-290} antibody bound strongly to its immunizing peptide as well as to CYP2D6, the lack of inhibition of activity cannot be attributed to a poor immune response against the peptide. It appears, therefore, that this site is outside the proinhibitory region of human CYP2D6. These results support the hypothesis that secondary structure alignment of P450 enzymes is preferable to primary structure alignment for identifying functional regions of P450 enzymes.

In conclusion, immunization with peptides conformationally restricted by cyclization to mimic loop regions of CYP2D6 resulted in strongly binding antibodies that when targeted appropriately were able to inhibit CYP2D6-catalyzed activity.