Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Clinical Chemistry 48:12 2208 –2216 (2002) Proteomics and Protein Markers Epitope Mapping of Antibodies against Prostatespecific Antigen with Use of Peptide Libraries Jari Leinonen,* Ping Wu, and Ulf-Håkan Stenman Background: Prostate-specific antigen (PSA) is the most important marker for prostate cancer, but PSA concentrations determined by various assays can differ significantly because of differences in specificity of the antibodies used. To identify epitopes recognized by various monoclonal antibodies (MAbs) to PSA, we have isolated peptides that react with the paratopes of these. Methods: Six anti-PSA MAbs representing three major epitope groups were screened with five cyclic phage display peptide libraries. After selection, the peptide sequences were determined by sequencing of the relevant part of viral DNA. Binding of the phage peptides to the MAbs was monitored by immunoassay. Results: For each MAb, several paratope-binding peptides with distinct sequence motifs were identified, but only ⬃10% showed similarity with the PSA sequence. Some of these correctly predicted the location of the epitopes. By sequential panning of the library with two closely related MAbs, we identified peptides reacting equally with both MAbs. When analyzed against a large panel of PSA MAbs, the peptides generally showed restricted specificity toward the MAb used for selection, but some peptides bound to several related MAbs. Conclusions: Most of the cyclic peptides selected with PSA MAbs are specific for the MAb used for selection and do not resemble any sequence on the antigen. Peptides reactive with two MAbs recognizing the same epitope can be obtained by sequential panning. This method can be used to predict the location of some epitopes, but additional methods are needed to confirm the result. © 2002 American Association for Clinical Chemistry Prostate cancer is the most common malignancy of males and the second leading cause of cancer death in the US Department of Clinical Chemistry, Helsinki University Central Hospital, FIN-0029 Helsinki, Finland. *Address correspondence to this author at: Department of Clinical Chemistry in Biomedicum, Helsinki University Central Hospital, Haartmaninkatu 8, FIN-00290 Helsinki, Finland. Fax 358-9-4717-1731; e-mail [email protected]. Received May 27, 2002; accepted August 30, 2002. (1 ). Prostate-specific antigen (PSA)1 is a serine protease secreted by the epithelial cells of the prostate (2 ). It is a reliable marker for monitoring treatment of prostate cancer, but it is also widely used for early diagnosis and screening (3, 4 ). Serum PSA is frequently increased in patients with benign prostatic hyperplasia, and in a screening setting, only approximately one-third of those with increased PSA have prostate cancer (4 ). PSA exists in circulation in various molecular forms, including several PSA-inhibitor complexes and isoforms of free PSA, and the use of assays for these forms can improve the clinical accuracy of PSA determination (5–9 ). The PSA concentrations measured by different assays vary significantly, which can be attributable to differences in standardization or specificity of the antibodies used toward various molecular forms of PSA (10 ). Furthermore, some PSA antibodies can also recognize human kallikrein 2 (hK2), which may affect the measured PSA concentrations. hK2 shows ⬃80% identity with PSA at the amino acid level (11–13 ), and like PSA, hK2 is a prostate-derived proteinase that has been shown to be a useful marker for prostatic disease (14, 15 ). PSA has been shown to contain six major epitope regions, which are exposed to a varying extent in different forms of PSA (16 ). To develop assays with highly defined specificity, it is important to know the epitopes of the antibodies used in the assay. Use of antibodies with defined specificities facilitates the development of assays that produce comparable results and the design of assays specific for variant forms of the antigen. Although the epitopes of PSA have been studied extensively (16 –22 ), the locations of only a few of the epitopes have been localized on the primary sequence of PSA. Amino acid residues forming linear epitopes can be identified by synthetic peptide epitope mapping, in which short chemically synthesized peptides covering the whole coding sequence of the antigen are used to identify the regions interacting with the monoclonal antibodies (MAbs) (23 ), 1 Nonstandard abbreviations: PSA, prostate-specific antigen; hK2, human kallikrein-2; MAb, monoclonal antibody; IFMA, immunofluorometric assay; and TBS, Tris-buffered saline. 2208 Clinical Chemistry 48, No. 12, 2002 as also described for PSA (19, 20 ). An alternative method is to screen peptide libraries to identify paratope-binding peptides, i.e., peptides binding to the antigen-binding site, and to compare the structures of these with that of the antigen (21, 24 –26 ). We studied whether paratope-binding peptides to MAbs against PSA isolated by screening cyclic phage display peptide libraries can be used to predict the structure of PSA epitopes. Materials and Methods antibodies and proteins The MAbs 5E4, 9C5, and 4G10 were developed with the following protocol. BALB/c mice were immunized with 10 –30 g of purified PSA by intraperitoneal injection with Freund’s complete adjuvant. PSA was purified from human seminal fluid as described previously (27 ). A booster dose of 10 g was given after 4 weeks, and then additional boosters of 100 and 150 g were given 1 and 2 days after the first booster, respectively. After the final booster, the splenic lymphoid cells of the mice were fused with the mouse myeloma cell line P3x63-Ag8.653 (American Type Culture Collection). The fused cells were harvested in HAT medium supplemented with interleukin-6 for 4 weeks, after which the cells were harvested in HT medium. PSA antibody production was screened by use of an immunofluorometric assay (IFMA) in microtiter wells coated with rabbit anti-mouse immunoglobulin. Briefly, 25 L of hybridoma supernatants plus 200 L of IFMA assay buffer (6 ) were incubated for 1 h in the wells, after which the wells were washed twice and 10 ng of Eu3⫹labeled PSA in 200 L of assay buffer was added. After additional incubation for 30 min, the wells were washed four times, and 200 L of enhancement solution (PerkinElmer-Wallac) was added. After incubation for 5 min, the fluorescence was measured. The MAbs produced were purified by affinity chromatography using protein G columns (Amersham Biosciences). Anti-PSA MAbs H117 and H50 were obtained from Abbott Diagnostics. Antibody 5A10 was from PerkinElmer-Wallac. The characteristics of the ISOBM workshop panel of anti-PSA MAbs have been described in detail elsewhere (16 ). Anti-phage MAb was a kind gift from Dr. Petri Saviranta (University of Turku, Turku, Finland). Anti-phage polyclonal antibody was purchased from Amersham Biosciences. Peptides were synthesized by standard Merrifield solid-phase chemistry. The phage peptide libraries were a kind gift from Dr Erkki Koivunen (University of Helsinki, Helsinki, Finland). selection of phage peptides Screening of the phage display peptide libraries was performed essentially as described previously (28 ). Libraries with the structures CX7C, CX8C, CX10C, CX3CX3CX3C, and CX3CX4CX2C, where C is cysteine and X is any of the 20 naturally occurring amino acids, were used. A pool of these libraries containing 1011–1012 infectious particles was screened with each MAb. Briefly, 2209 a pool of phage libraries was added to the wells coated with a MAb and incubated for 3 h at 22 °C during the first round of panning and for 1 h during subsequent rounds. The phage solution was removed, and the wells were washed with Tris-buffered saline (TBS; 50 mmol/L TrisHCl, pH 7.8, 0.15 mol/L NaCl) containing 5 mL/L Tween 20. Bound phage were eluted with 0.1 mol/L glycine buffer, pH 2.2, and neutralized with 1 mol/L Tris base. The eluted phage were amplified by infection of Escherichia coli K91 kan cells and purified by precipitation with polyethylene glycol. Sequential panning with two MAbs was performed with MAb pairs known to react with the same antigenic region on PSA, including MAb pairs 5E4/H117, 9C5/H50, and 4G10/5A10. For example, the primary library was panned for two rounds using MAb 5E4, after which the eluate was amplified and panned for one round with MAb H117. After three rounds of selection and amplification, the peptide sequences were determined by sequencing the relevant part of the viral DNA as described previously (28 ). Sequencing was performed with an ABI 310 Genetic Analyzer and Dye Terminator Cycle Sequencing Core Kit (PE Applied Biosystems), with the oligonucleotide 5⬘CCCTCATAGTTAAGCGTAACG-3⬘ as a primer. The amino acid sequence of each peptide isolated was analyzed for homology with PSA and other proteins by the BLASTP program using the SwissProt database. The three-dimensional model of PSA established by Villoutreix et al. (29 ) was visualized with the RasMol program (30 ). IFMAs The solid-phase antibodies used in the IFMA were coated on microtitration wells at a concentration of 5 mg/L in TBS for 16 h at 22 °C. The solution was then discarded, and the wells were saturated with 10 g/L bovine serum albumin in TBS for 3 h at 22 °C. The antibodies used as tracers were labeled with a Eu3⫹ chelate as described previously (28 ). The assay buffer was TBS, pH 7.7, containing 33 mol/L bovine serum albumin and 1 mol/L bovine globulin. The binding of individual phage clones to MAbs was tested by phage IFMA. In this assay, 15 L of phage (109–1010 infectious particles) and 200 L of assay buffer were added to wells coated with each MAb. After incubation for 1 h, the wells were washed and filled with 200 L of assay buffer containing 50 ng of a Eu3⫹-labeled anti-phage MAb recognizing the M13 coat protein of the phage. After incubation for 60 min, the wells were washed four times, and enhancement solution (Perkin-ElmerWallac) was added. Fluorescence was quantified with a 1234 DELFIA Research fluorometer (Perkin-Elmer-Wallac). The detection limit was defined as the fluorescence signal obtained with phage containing unrelated peptide ⫹3 SD. The binding specificity of the phage peptide to the antigen-binding regions of the mAbs was confirmed by 2210 Leinonen et al.: Epitope Mapping of PSA by Peptide Libraries showing that PSA inhibited the binding of phage to antibodies. Briefly, 15 L of each phage clone and 0.3–300 ng of PSA purified from seminal fluid were pipetted into a well coated with each mAb. After incubation of phage and PSA for 2 h, the protocol for phage IFMA was followed. The binding specificity of synthetic peptides was confirmed by showing that they competed with PSA for binding to antibodies. Briefly, 0 –120 g of peptide in 200 L of assay buffer was pipetted into a well coated with MAb. After incubation for 1 h, 5 ng of PSA was added and incubated for 30 min. Inhibition of PSA binding was quantified by PSA IFMA as described (31 ). reactivity of the phage with the isobm workshop panel of MAbs Phage isolated by panning with the six PSA MAbs were tested for binding to 39 MAbs from the ISOBM workshop for characterization of antibodies against PSA by phage IFMA. According to the ISOBM classification, the six MAbs used for screening of the phage libraries belonged to epitope groups 1, 3, and 6 (16 ), which in this study are referred to as I, III, and VI, respectively. The ISOBM workshop MAbs tested included MAbs 26, 33, 68, 74, 77, 78, 85, 209, 213, 216, 223, and 230 from epitope group I; MAbs 31, 36, 57, 63, 64, 66, 72, 82, 84, 88, 89, 212, 224, and 229 from epitope group III; and MAbs 27, 29, 34, 56, 65, 67, 75, 79, 210, 214, 218, 221, and 225 from epitope group VI. Results identification of antibody-binding peptides The MAbs studied were chosen on the basis of crossinhibition characteristics to include three pairs of closely related antibodies. The peptides isolated with the six MAbs contained characteristic motifs and consensus amino acids (Table 1), but only 9 of 100 peptides displayed sequence identity with PSA, defined as at least three identical amino acids within a window of four amino acids. Binding of the peptides to MAbs was inhibited by PSA, showing that they were paratope specific (Fig. 1). The epitopes of MAbs 5E4 and H117 have been mapped to the N-terminal region of PSA (amino acid residues 3–11) by synthetic peptide mapping and crossinhibition studies (16, 20 ). However, most of the peptides developed against these antibodies resembled neither PSA nor each other. This indicated that the peptides recognized structures in the paratope that do not participate in antigen binding or that the epitopes were different and mainly determined by the three-dimensional structure of PSA. To identify peptides reacting with two MAbs recognizing the same epitope, we developed a sequential panning strategy. The peptides identified by this strategy using MAbs 5E4 and H117 did not resemble the peptides selected separately (Table 1). Two peptides, CPSVDGGWTC (VI-13 and VI-20) and CHSACSKHCFVHC (VI-15 and VI-22), were obtained irrespective of the order in which the MAbs were used for selection. In peptides Table 1. Amino acid sequences of peptides selected from phage display peptide libraries with MAbs against PSA.a MAb(s) Peptide Sequence 5E4 5E4 5E4 5E4 and H117 5E4 and H117 5E4 and H117 H117 H117 and 5E4 H117 and 5E4 H117 and 5E4 5A10 and 4G10 5A10 and 4G10 5A10 and 4G10 5A10 and 4G10 4G10 4G10 and 5A10 4G10 and 5A10 4G10 and 5A10 4G10 and 5A10 4G10 and 5A10 H50 H50 H50 H50 H50 and 9C5 H50 and 9C5 H50 and 9C5 H50 and 9C5 H50 and 9C5 9C5 9C5 and H50 9C5 and H50 9C5 and H50 VI-1 VI-9 VI-12 VI-13 VI-14 VI-15 VI-18 VI-20 VI-21 VI-22 I-8 I-9 I-15 I-16 I-19 I-30 I-31 I-32 I-33 I-35 III-1 III-8 III-9 III-12 III-14 III-15 III-17 III-20 III-25 III-28 III-37 III-38 III-39 CPVDFDFLC CPRDFEFLC CPADFEFLC CPSVDGGWTC CKSMDGGWTC CHSACSKHCFVHC CTWHWSPEEC CPSVDGGWTC CHSACSKHCFVYC CHSACSKHCFVHC CRPWGSSRC CKSWGSSRC CNLYKVGC CHPYKVGC CSVYPFWHC CRSYPFWMC CVSYPFWKC CAVFPFWRC CVWWWGC CAPWWGC CSGIAPWLC CGPGIDSWVC CGPGIRSWVC CDYMPLVDNC CEGIAVWLC CTRYSGYWVC CTQMNGYWIC CLYDHLC CILLYNDCC CFFGNWNFC CWNWNLHISC CYFKNWNFC CVRSCISDCELRC a The sequences of the peptides analyzed further are shown. The peptides were isolated by panning with each MAb or by sequential panning, in which the library was first panned using one MAb, after which the eluate was panned with another related MAb. The peptide sequences were determined by sequencing the relevant part of the viral DNA. MAbs 5E4 and H117 belong to epitope group VI, MAbs 5A10 and 4G10 to group I, and MAbs H50 and 9C5 to group III (16 ). The sequences of all peptides isolated (n ⫽ 100) are available as a data supplement with the online version of this article at http://www.clinchem.org/ content/vol48/issue12/. VI-13 and VI-14, the motif GGWTC was similar to the sequence GGWEC comprising residues 3–7 in the NH2 terminus of mature PSA (Fig. 2). Furthermore, in peptides VI-15 and VI-21, the motif CSKH was similar to the sequence CEKH comprising residues 7–10 in PSA (Fig. 2). Thus, the structures of these peptides suggest that the epitope recognized by MAbs 5E4 and H117 comprises residues 3–10 in mature PSA. All phage peptides isolated by use of MAb 5E4 alone contained the structure PXDFXFL. This motif was not present in peptides selected by MAb H117 alone, which contained the motif WXXXPXE (Table 1). Among MAb 5E4-selected peptides, peptides Clinical Chemistry 48, No. 12, 2002 Fig. 1. Inhibition of phage binding to anti-PSA MAb by PSA. We incubated 15 L of each phage clone (1011–1012 infectious particles) with increasing concentrations of PSA (0 – 400 nmol/L) in a well coated with MAb H117. After washing, the binding of phage was monitored by an Eu3⫹-labeled anti-phage antibody. The reactivity is shown for phage clones VI-13 (E), -14 (䡺), -15 (Œ), and -21 (‚) obtained by sequential panning using MAbs 5E4 and H117. The background of ⬃1000 cps has been subtracted. VI-9 and VI-12 contained the sequence EFL corresponding to PSA amino acid residues 140 –142; a similar sequence, DFL, was present in four other peptides, and the FL sequence was present in all 12 peptides. A synthetic peptide (corresponding to the sequence of phage peptide 2211 VI-12) containing the EFL sequence inhibited the binding of PSA with MAb 5E4 in a dose-dependent manner (Fig. 3), confirming that this peptide and PSA bind to same site on the MAb. Among MAb H117-selected peptides, peptide VI-18 contained the PEE sequence (residues 138 –140 in PSA), and the PXE motif was present in three of the four peptides. Thus, these MAbs recognize the same amino acids in region 3–10 and different amino acids from the region 138 –142 (PEEFL; Fig. 2). Interestingly, in the three-dimensional model of PSA (29 ), the PEEFL sequence is located close to the N-terminal epitope (Fig. 4). These results suggest that amino acids 138 –142 form part of a noncontinuous epitope together with amino acids 3–10 in the N-terminal region. Group I MAbs 5A10 and 4G10 bind to an epitope specific for free PSA, which in PSA-serpin complexes is covered by the inhibitors (16, 32 ). Some MAbs in this group have been shown to bind to linear epitopes comprising amino acids 80 –91 in PSA, which form part of the so-called kallikrein loop (16, 20 ), but most of them react only weakly with reduced, intact PSA, suggesting that they are mainly conformation dependent. MAbs 5A10 and 4G10 inhibit each other efficiently, but MAb 4G10 has been shown to cross-react with complexed PSA approximately two- to fivefold more than does 5A10 (33, 34 ), suggesting that the epitopes of these MAbs are not identical. When we selected with 4G10 and sequentially with 4G10 and 5A10, we found the motifs YPFW, WWW, and PWW (Table 1), but these are not found in PSA. Fig. 2. Amino acid sequence and epitope structure of PSA. Residue 1 indicates the N-terminal residue of the active form of PSA. The sequences and codes of the peptides used to assign the epitopes are shown above the PSA sequence with the identical residues underlined. Previously determined locations for the epitope groups of MAbs included in the present study are as follows: group VI, amino acid residues 3–11; group I, residues 80 –91 (16, 20 ), 145–148, and 204 –208 (21 ); group III, residues 151–155 (20 ). Active site residues His-41, Asp-96, and Ser-189 are underlined. 2212 Leinonen et al.: Epitope Mapping of PSA by Peptide Libraries Fig. 3. Inhibition of PSA binding to MAb 5E4 by a synthetic peptide. We incubated 0 – 600 mg/L of chemically synthesized peptide CPADFEFLC (VI-12 in Table 1) in a well coated with MAb 5E4 for 30 min, after which 0.4 ng of PSA was added and incubation was continued for an adiditional 30 min. After washing, the inhibition of PSA binding was quantified by PSA IFMA (31 ). The columns represent mean values of duplicate measurements ⫾ SD (error bars). However, when we selected with MAb 5A10 and sequentially with 5A10 and 4G10, we frequently found the WGS and TVXXAW motifs. Peptide I-9 contained the sequence SWGSSRC, which is similar to the PSA sequence SWGSEPC (amino acids 204 –210 in PSA; Fig. 2). Furthermore, WGS, WG, or GS motifs were present in 7 of 22 peptides isolated. In agreement with this, Michel et al. (21 ) have found by screening peptide libraries that the SWG site forms part of the epitope of MAbs specific for free PSA. Amino acids 204 –207 (SWGS) form the edge of the groove containing the active site residues of PSA (29 ) opposite the region 80 –91 on the other side of the groove. Peptide I-16 contained the sequence HPYKV, which is similar to the PSA sequence HPQKV (amino acids 164 –168 in PSA), which is located adjacent to amino acids 204 –207 and borders to the predicted binding site of group III MAbs (Fig. 2). It may thus form part of the epitope for MAbs specific for free PSA. Further support for the location of the epitope as predicted by phage display peptides can be obtained by analyzing whether the predicted epitope regions differ between PSA and hK2 and accounting for whether the MAb reacts with PSA only or with both PSA and hK2. hK2 shares 80% identity with PSA (11–13 ), and many PSA antibodies are known to recognize hK2, showing that several epitopes in PSA and hK2 are identical (16, 35 ). Thus, the location of the epitope of MAb 5A10 in the 164 –168 region in PSA is further supported by the facts that this region differs between PSA and hK2 (HPQKV and YSEKV, respectively) and that MAb 5A10 Fig. 4. Molecular model of PSA (29 ) showing with “spacefill” the epitopes for MAbs belonging to groups I and VI as identified by the paratope-binding peptides. The dark gray residues constitute amino acids 3–10 (GGWECEKH), similar to peptides VI-13, -14, -15, and -21, and amino acids 138 –142 (PEEFL), similar to peptides VI-12 and VI-18 obtained by panning with group VI MAbs. The white residues constitute amino acids 164 –168 (HPQKV), similar to peptide I-16, and amino acids 204 –207 (SWGS), similar to peptide I-9 obtained by panning with group I MAbs. In addition, amino acids 81–90, previously identified as the binding site for group I MAbs (20 ), are shown in white, and the active site amino acids His-41, Asp-96, and Ser-189 are indicated with dots. does not react with hK2 (16 ). Thus, these results suggest that the epitopes for MAbs belonging to group I may comprise residues on three different loops, i.e., amino acids 80 –91 (20 ), 164 –168, and 204 –207. MAbs H50 and 9C5 bind to epitopes located remote from the active site and are considered to be mainly conformation dependent (16 ). These MAbs show highly similar binding characteristics, but only H50 shows strong reactivity with hK2 (16, 36 ). The peptides isolated by panning with MAb H50 and sequentially with MAbs H50 and 9C5 frequently contained the structures GIXXW or GYW, whereas in peptides isolated by use of MAb 9C5 and sequentially with MAbs 9C5 and H50, the NWN motif was found repeatedly. Peptides III-8 and III-9 (H50 selection) contained the sequences PGIDS and PGIRS, respectively, which are similar to the sequence PGDDS in PSA (amino acids 89 –93; Fig. 2). This region partially overlaps amino acids 80 –91 in PSA, which is the previously identified binding site for MAbs specific for free PSA (20 ). Furthermore, MAb H50 binds strongly to both PSA and hK2, but the amino acid sequence 89 –93 in these differs (PGDDS and PDEDS in PSA and hK2, respectively). Thus, it is unlikely that the PGDDS sequence is part of the epitope for these MAbs. In peptide III-12 (H50 selection), the sequence PLVDN is similar to the PSA sequence 2213 Clinical Chemistry 48, No. 12, 2002 PLVCN (amino acids 192–196 in PSA; Fig. 2). This sequence is identical in PSA and hK2, and it is adjacent to epitope region 151–155 in the three-dimensional model of PSA, which previously has been predicted to be a binding site for MAbs belonging to this epitope group. Thus, these results suggest that the epitopes for group III MAbs comprise at least two different loops, i.e., amino acids 151–155 (20 ) and 192–196. reactivity of the peptides with isobm MAbs To study whether the epitopes found were recognized by other MAbs, we studied the reactivity of 19 peptides with a panel of MAbs from the ISOBM workshop. The peptides studied were isolated by sequential panning using MAb pairs 5E4/H117 (group VI), 4G10/5A10 (group I), and H50/9C5 (group III). The reactivity of three group VI peptides showing similarity with amino acids 3–10 in PSA was studied with 13 group VI MAbs (Table 2). Seven MAbs bound to these peptides, suggesting that amino acids 3–10 in PSA form part of the epitope recognized by most MAbs in this group. Six of the MAbs did not bind any of these peptides, suggesting that these amino acids are not a major part of their epitopes or that the motif is not present in correct conformation for these MAbs. The reactivity of six group I peptides (I-8, I-15, I-30, I-31, I-32, and I-35) was studied with 14 group I MAbs. These peptides showed highly restricted specificity and reacted only with the primary selection MAb, i.e., peptides I-8 and I-15 reacted only with MAb 5A10, and Table 2. Reactivity of group VI MAbs with peptides isolated by panning sequentially with MAbs 5E4 and H117.a Peptide MAb VI-13 VI-14 VI-15 PSA-Eu3ⴙ 27 29 34 56 (H117) 65 67 75 79 (5E4) 210 214 218 221 225 9831 ⫺ 476784 472197 ⫺ 20575 ⫺ 337648 ⫺ 29145 ⫺ ⫺ 1266 ⫺ ⫺ ⫺ 485780 ⫺ ⫺ ⫺ 246281 ⫺ 1391404 ⫺ ⫺ ⫺ 12730 ⫺ 14329 396210 ⫺ 65068 ⫺ 217960 ⫺ 53310 ⫺ ⫺ 2072 32 410 209 062 310 266 330 600 303 059 55 636 37 636 378 440 348 434 389 005 327 034 259 856 377 816 a The binding of phage was measured by IFMA using each of the PSA MAbs on solid phase and anti-phage antibody as tracer. The results are expressed in europium fluorescence units (cps) and represent mean values of duplicate measurements. For comparison, the response (cps) obtained by adding 10 ng of europium-labeled PSA to wells coated with each MAb is shown. A result is indicated by (⫺) when the response was below the detection limit of the assay (1125 cps; defined as binding of phage containing an unrelated peptide to MAbs ⫹3 SD). The assay CVs of phage binding IFMAs were 5–20%. peptides I-30, I-31, I-32, and I-35 reacted only with MAb 4G10. The reactivity of eight group III peptides was tested with 14 MAbs from epitope group III (Table 3). Some of the peptides reacted with several MAbs within this epitope group. Most of them showed unique binding patterns, but MAbs 72 and 229 may have closely related specificity because they reacted only with peptide III-15. MAb 224 reacted only with peptides III-15, -20, and -25, obtained with MAb H50 as primary selection antibody, whereas MAb 89 reacted only with peptide III-37, obtained with MAb 9C5 as primary selection MAb. Four MAbs did not bind to any of the peptides. Discussion The main antigenic regions of PSA recognized by mouse antibodies have been studied extensively (16, 17, 20, 22 ), but not much is known about the fine structure of the epitopes. In this study, we isolated peptides binding specifically with the paratope regions of six PSA MAbs and studied their reactivity with a large panel of PSA antibodies. Alignment of the peptides with the PSA sequence revealed several binding motifs, but most of the isolated peptides did not resemble any part of PSA. These may represent mimotopes, i.e., antigenic mimics, or they may be structurally unrelated to PSA, reflecting the polyspecific nature of the binding region of an antibody (37 ). Thus, although MAbs are usually considered to be specific for a certain antigen, the paratope can also bind structures that do not resemble the original immunogen. Even if MAbs have rarely been shown to bind strongly to unrelated natural antigens, cross-reactivity with a totally unrelated protein has been observed with an antibody prepared against a synthetic peptide (38 ). The polyspecificity can be explained by the fact that the paratope consists of ⬃20 amino acids, but only a few of these participate in antigen binding, and water molecules can fill the remaining space between the paratope and epitope. Thus, the same paratope can contain several subsites for different epitopes, and amino acids in the paratope can form strong bonds with peptides unrelated to the immunogen (37, 39 ). The polyspecific nature of paratopes toward unrelated peptides has also been demonstrated by Keitel et al. (40 ) and Kramer et al. (41 ), who used x-ray structural data of three peptides complexed with an anti-p24 (HIV-I) MAb. In our study, only ⬃10% of the peptides showed similarity with PSA. Some of these corresponded to earlier identified epitopes, whereas others contained new motifs adjacent to earlier described ones. Thus, they may well form part of the epitope. Antibodies 5E4 and H117 identified the same N-terminal epitope (amino acids 3–10) as was found in previous studies for MAbs in this group (20 ), but they also appeared to react with amino acids in the 138 –142 region. These antibodies also react with hK2 (16 ), and regions 3–10 and 138 –142 are identical in PSA 2214 Leinonen et al.: Epitope Mapping of PSA by Peptide Libraries Table 3. Reactivity of group III MAbs with peptides isolated by panning sequentially with MAbs H50 and 9C5.a Peptide MAb 31 36 57 (H50) 63 64 66 72 82 (9C5) 84 88 89 212 224 229 a III-14 III-15 III-17 4571 ⫺ 69 591 1141 ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ 4767 ⫺ 14 927 2467 ⫺ ⫺ 2684 ⫺ ⫺ ⫺ ⫺ ⫺ 6128 2808 10 417 ⫺ 31 185 ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ III-20 III-25 III-37 III-38 III-39 PSA-Eu ⫺ ⫺ 466 040 ⫺ 1157 ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ 1855 ⫺ ⫺ ⫺ 144 901 15 737 ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ 2978 ⫺ ⫺ ⫺ ⫺ 17 555 ⫺ ⫺ ⫺ 101 174 ⫺ ⫺ 1463 ⫺ ⫺ ⫺ 6172 ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ 205 955 ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ 108 687 ⫺ ⫺ ⫺ ⫺ ⫺ ⫺ 289 969 213 806 110 126 256 147 147 172 213 319 157 672 215 199 186 125 26 488 36 888 192 330 71 934 166 244 The assay was as described in footnote a of Table 2. and hK2 (11, 12 ), further supporting the involvement of these residues in the epitopes. Results from other groups have suggested that most of the MAbs that bind both PSA and hK2 recognize nonconformational linear epitopes (35 ). The N-terminal part of PSA appears to form a fairly long linear N-terminal epitope (residues 3–10) and thus may be the major mediator of binding of MAbs in this group. However, regions 3–10 and 138 –142 reside within 15–20 Å from each other, and amino acids within this distance can be in contact with the paratope (37 ). Thus, it is conceivable that the epitopes of some group VI MAbs comprise residues 3–10 and 138 –142. MAbs 5E4 and H117 bind peptides similar to PSA 3–10 and 138 –142, but the epitopes around amino acids 138 –142 are overlapping rather than identical. Parhami-Seren et al. (42 ) also identified overlapping epitopes for two related MAbs against streptokinase by screening peptide libraries. Interestingly, in previous synthetic peptide epitope mapping studies, MAb H117 did not react with any synthetic linear peptide derived from the PSA sequence (20 ), although it has been shown to compete efficiently with other group VI MAbs (16 ). Thus, it is likely that the synthetic peptides used by Piironen et al. (20 ) could not adopt the correct conformation required for strong binding with MAb H117. We also found some paratope-specific peptides showing sequence similarity with PSA, but the suggested epitope locations were not in agreement with those in previous studies. In a single peptide (III-8), the sequence identity was not compatible with previous epitope assignments, suggesting that the sequence similarity was attributable to chance. A key factor in the protocol for isolating peptides resembling the linear epitope in the NH2 terminus of PSA was the introduction of sequential panning with two MAbs. With this method, peptides reacting with both selection MAbs (5E4 and H117) were obtained. However, with MAb pairs 9C5/H50 and 5A10/4G10, no peptides reactive with both MAbs were found, which suggests that the epitopes for these MAbs do not contain common residues although they block the binding of each other. Synthetic peptide mapping studies have suggested that the binding sites for MAbs specific for free PSA reside in the regions 50 – 69 (19 ) and 80 –91 (20 ). In the study of Piironen et al. (20 ), all seven free-PSA MAbs reacted strongly with peptides containing the 84 –91 sequence. Some binding of MAbs was observed to peptides containing the sequence 50 – 67, but strong reaction with these peptides was observed for only two MAbs recognizing both free and complexed PSA. Furthermore, Jette et al. (43 ) identified the sequence PSA 51–55 as an epitope for a “total PSA specific” MAb by use of both synthetic peptide epitope mapping and peptide library screening. Michel et al. (21 ) suggested on the basis of peptide library screening that the epitopes of MAbs specific for free PSA include amino acid residues from PSA regions 53–59, 145–148, and 204 –207. Our results also suggested that amino acids 204 –208 may form part of the epitope for such MAbs, but no peptides resembling the two other sequences were found. In the three-dimensional model, these amino acids form a relatively wide loop, and it is possible that the cyclic peptides used in our study are too tightly coiled to mimic such a loop. No peptides that bound strongly to MAb H50 were found by Piironen et al. (20 ) by use of synthetic peptide epitope mapping, but the binding sites of related MAbs were mapped to the 151–155 region. We did not find any peptides with similarity to this region by screening with the two related MAbs, H50 and 9C5. However, a peptide showing similarity with the 192–196 region was identified after screening with MAb H50, and this region is adjacent to region 151–155 in the three-dimensional model of PSA. Furthermore, the 192–196 region is identical in PSA and 2215 Clinical Chemistry 48, No. 12, 2002 hK2, and MAb H50 is reactive with both PSA and hK2 (16 ). reactivity of the peptides with isobm MAbs Previous studies on reactivity of paratope-binding peptides with MAbs showing highly similar specificity suggested that peptides generally show restricted specificity toward the selection MAb (44 ). This was also the case with peptides to group I MAbs, which were specific for free PSA, i.e., each peptide bound only to the MAb used to isolate it. The peptides selected with group III MAbs also showed restricted specificity. Most of them bound to 1–3 of the 14 MAbs specific for this region. In contrast, the peptides selected with group VI MAbs recognized many antibodies. Two peptides bound to one-half (7 of 13) of the MAbs included. These results suggest that amino acids 3–10 form an immunodominant region that can be mimicked by the fairly short cyclic peptides. The restricted specificity of the peptides to MAbs in other groups may indicate larger variability in epitope specificity. Thus, although some epitope regions can be subgrouped by use of paratope-binding peptides, the general utility of this approach for epitope identification is limited. Many peptides bearing no relationship with the immunogen (PSA) bound strongly to the paratopes. Apparently these binding specificities are not displayed by natural human antigens, as evidenced by the very high specificity of MAbs when used for clinical purposes. However, the unexpected reactivity of a MAb to BRCA1 with semenogelin may be an example of such polyspecificity (38 ). A less dramatic but more common demonstration of this polyspecificity is the nonspecific inhibition of antigenantibody binding exerted by proteins. This phenomenon contributes to the well-known matrix effect that affects all immunoassays (45 ). The nonspecific interactions between various proteins in clinical samples and the paratope must be much weaker that that of the specifically selected peptides, and it can usually be eliminated by diluting the sample (45 ). In conclusion, this study shows that screening with peptide libraries can be used to define both linear and noncontinuous epitopes, but the assignment of an epitope region needs to be confirmed by complementary methods. Furthermore, although MAbs are highly specific when used in clinical assays, the paratope region can bind strongly to several unrelated peptide structures. However, strong nonspecific binding may be revealed only by phage display peptides, whereas it appears to be very rare in natural samples. This work was supported by grants from the Academy of Finland, Sigfrid Jusélius Foundation, and the Finnish Cancer Society. We thank Liisa Airas and Maarit Leinimaa for expert technical assistance. References 1. Wingo PA, Tong T, Bolden S. Cancer statistics, 1995. CA Cancer J Clin 1995;45:8 –30. 2. Wang MC, Valenzuela LA, Murphy GP, Chu TM. Purification of a human prostate specific antigen. Invest Urol 1979;17:159 – 63. 3. Stamey TA, Yang N, Hay AR, McNeal JE, Freiha FS, Redwine E. Prostate specific antigen as a serum marker for adenocarcinoma of the prostate. N Engl J Med 1987;317:909 –16. 4. Catalona WJ, Smith DS, Ratliff TL, Dodds KM, Coplen DE, Yuan JJ, et al. Measurement of prostate-specific antigen in serum as a screening test for prostate cancer. N Engl J Med 1991;324: 1156 – 61. 5. Stenman UH, Leinonen J, Alfthan H, Rannikko S, Tuhkanen K, Alfthan O. A complex between prostate-specific antigen and ␣1-antichymotrypsin is the major form of prostate-specific antigen in serum of patients with prostatic cancer: assay of the complex improves clinical sensitivity for cancer. Cancer Res 1991;51: 222– 6. 6. Leinonen J, Lovgren T, Vornanen T, Stenman UH. Double-label time-resolved immunofluorometric assay of prostate-specific antigen and of its complex with ␣1-antichymotrypsin. Clin Chem 1993;39:2098 –103. 7. Zhang WM, Finne P, Leinonen J, Vesalainen S, Nordling S, Stenman UH. Measurement of the complex between prostatespecific antigen and ␣1-protease inhibitor in serum. Clin Chem 1999;45:814 –21. 8. Zhang WM, Finne P, Leinonen J, Vesalainen S, Nordling S, Rannikko S, et al. Characterization and immunological determination of the complex between prostate-specific antigen and ␣2macroglobulin. Clin Chem 1998;44:2471–9. 9. Mikolajczyk SD, Marker KM, Millar LS, Kumar A, Saedi MS, Payne JK, et al. A truncated precursor form of prostate-specific antigen is a more specific serum marker of prostate cancer. Cancer Res 2001;61:6958 – 63. 10. Stenman UH, Leinonen J, Zhang WM. Problems in the determination of prostate specific antigen. Eur J Clin Chem Clin Biochem 1996;34:735– 40. 11. Schedlich LJ, Bennets BH, Morris BJ. Primary structure of a human glandular kallikrein gene. DNA 1987;6:429 –37. 12. Lundwall Å, Lilja H. Molecular cloning of human prostate specific antigen cDNA. FEBS Lett 1987;214:317–21. 13. Chapdelaine P, Paradis G, Tremblay RR, Dube JY. High level of expression in the prostate of a human glandular kallikrein mRNA related to prostate-specific antigen. FEBS Lett 1988;236:205– 8. 14. Darson MF, Pacelli A, Roche P, Rittenhouse HG, Wolfert RL, Young CY, et al. Human glandular kallikrein 2 (hK2) expression in prostatic intraepithelial neoplasia and adenocarcinoma: a novel prostate cancer marker. Urology 1997;49:857– 62. 15. Finlay JA, Evans CL, Day JR, Payne JK, Mikolajczyk SD, Millar LS, et al. Development of monoclonal antibodies specific for human glandular kallikrein (hK2): development of a dual antibody immunoassay for hK2 with negligible prostate-specific antigen crossreactivity. Urology 1998;51:804 –9. 16. Stenman UH, Paus E, Allard WJ, Andersson I, Andres C, Barnett TR, et al. Summary report of the TD-3 workshop: characterization of 83 antibodies against prostate-specific antigen. Tumour Biol 1999;20:1–12. 17. Pettersson K, Piironen T, Seppala M, Liukkonen L, Christensson A, Matikainen MT, et al. Free and complexed prostate-specific antigen (PSA): in vitro stability, epitope map, and development of immunofluorometric assays for specific and sensitive detection of free PSA and PSA-␣1-antichymotrypsin complex. Clin Chem 1995; 41:1480 – 8. 18. Nilsson O, Peter A, Andersson I, Nilsson K, Grundstrom B, Karlsson B. Antigenic determinants of prostate-specific antigen 2216 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. Leinonen et al.: Epitope Mapping of PSA by Peptide Libraries (PSA) and development of assays specific for different forms of PSA. Eur Urol 1997;31:178 – 81. Corey E, Wegner SK, Corey MJ, Vessella RL. Prostate-specific antigen: characterization of epitopes by synthetic peptide mapping and inhibition studies. Clin Chem 1997;43:575– 84. Piironen T, Villoutreix BO, Becker C, Hollingsworth K, Vihinen M, Bridon D, et al. Determination and analysis of antigenic epitopes of prostate specific antigen (PSA) and human glandular kallikrein 2 (hK2) using synthetic peptides and computer modeling. Protein Sci 1998;7:259 – 69. Michel S, Deleage G, Charrier JP, Passagot J, Battail-Poirot N, Sibai G, et al. Anti-free prostate-specific antigen monoclonal antibody epitopes defined by mimotopes and molecular modeling. Clin Chem 1999;45:638 –50. Black MH, Grass CL, Leinonen J, Stenman UH, Diamandis EP. Characterization of monoclonal antibodies for prostate-specific antigen and development of highly sensitive free prostate-specific antigen assays. Clin Chem 1999;45:347–54. Worthington J, Morgan K. Epitope mapping using synthetic peptides. In: Wisdom GB, ed. Peptide antigens, 1st ed. New York: Oxford University Press, 1994:181–216. Cortese R, Monaci P, Luzzago A, Santini C, Bartoli F, Cortese I, et al. Selection of biologically active peptides by phage display of random peptide libraries. Curr Opin Biotechnol 1996;7:616 –21. Craig L, Sanschagrin PC, Rozek A, Lackie S, Kuhn LA, Scott JK. The role of structure in antibody cross-reactivity between peptides and folded proteins. J Mol Biol 1998;281:183–201. Chargelegue D, Obeid OE, Hsu SC, Shaw MD, Denbury AN, Taylor G, et al. A peptide mimic of a protective epitope of respiratory syncytial virus selected from a combinatorial library induces virus-neutralizing antibodies and reduces viral load in vivo. J Virol 1998;72:2040 – 6. Zhang WM, Leinonen J, Kalkkinen N, Dowell B, Stenman UH. Purification and characterization of different molecular forms of prostate-specific antigen in human seminal fluid. Clin Chem 1995;41:1567–73. Wu P, Leinonen J, Koivunen E, Lankinen H, Stenman UH. Identification of novel prostate-specific antigen-binding peptides modulating its enzyme activity. Eur J Biochem 2000;267:6212–20. Villoutreix BO, Getzoff ED, Griffin JH. A structural model for the prostate disease marker, human prostate-specific antigen. Protein Sci 1994;3:2033– 44. Sayle RA, Milner-White EJ. RASMOL: biomolecular graphics for all. Trends Biochem Sci 1995;20:374. Leinonen J, Zhang WM, Stenman UH. Complex formation between PSA isoenzymes and protease inhibitors. J Urol 1996;155:1099 – 103. 32. Lilja H, Christensson A, Dahlen U, Matikainen MT, Nilsson O, Pettersson K, et al. Prostate-specific antigen in serum occurs predominantly in complex with ␣1-antichymotrypsin. Clin Chem 1991;37:1618 –25. 33. Bellanger L, Andres C, Seguin P. Epitope mapping of 53 antibodies against prostate-specific antigen. Tumour Biol 1999;20(Suppl 1):18 –23. 34. Leinonen J, Zhang W-M, Paus E, Stenman U-H. Reactivity of 77 antibodies to prostate specific antigen (PSA) with isoenzymes and complexes of PSA. Tumor Biol 1999;20(Suppl 1):28 –34. 35. Finlay JA, Day JR, Rittenhouse HG. Polyclonal and monoclonal antibodies to prostate-specific antigen can cross-react with human kallikrein 2 and human kallikrein 1. Urology 1999;53:746 – 51. 36. Wang TJ, Linton HJ, Rittenhouse HG, Wolfert RL. Western blotting analysis of antibodies to prostate-specific antigen: cross-reactivity with human kallikrein-2. Tumour Biol 1999;20(Suppl 1):75– 8. 37. Van Regenmortel MH. The antigen-antibody reaction. In: Price C, Newman D, eds. Principles and practice of immunoassay. New York: Stockton Press, 1997:13–34. 38. Angelopoulou K, Borchert G, Melegos DN, Lianidou E, Lilja H, Diamandis EP. Characterization of the BRCA1-like immunoreactivity of human seminal plasma. Urology 1999;54:753– 62. 39. Weininger R, Richards F. Combining regions of antibodies. In: Atassis M, ed. Immunochemistry of proteins. New York: Plenum, 1979:123– 66. 40. Keitel T, Kramer A, Wessner H, Scholz C, Schneider-Mergener J, Hohne W. Crystallographic analysis of anti-p24 (HIV-1) monoclonal antibody cross-reactivity and polyspecificity. Cell 1997;91:811– 20. 41. Kramer A, Keitel T, Winkler K, Stocklein W, Hohne W, SchneiderMergener J. Molecular basis for the binding promiscuity of an anti-p24 (HIV-1) monoclonal antibody. Cell 1997;91:799 – 809. 42. Parhami-Seren B, Keel T, Reed GL. Sequences of antigenic epitopes of streptokinase identified via random peptide libraries displayed on phage. J Mol Biol 1997;271:333– 41. 43. Jette DC, Kreutz FT, Malcolm BA, Wishart DS, Noujaim AA, Suresh MR. Epitope mapping of prostate-specific antigen with monoclonal antibodies. Clin Chem 1996;42:1961–9. 44. Harris SL, Craig L, Mehroke JS, Rashed M, Zwick MB, Kenar K, et al. Exploring the basis of peptide-carbohydrate crossreactivity: evidence for discrimination by peptides between closely related anti-carbohydrate antibodies. Proc Natl Acad Sci U S A 1997;94: 2454 –9. 45. Stenman U-H. Immunoassay standardization. In: Price C, Newman D, eds. Principles and practice of immunoassay. New York: Stockton Press, 1997:245– 68.