Journal of Immunological Methods. 140 (1991) 249-258

249

© 199t Elsevier Science Publishers B.V. 0022-1759/91/$03.50 ADONIS 002217599100218E JIM05978

A crossreactive antipeptide monoclonal antibody with specificity for lysyl-lysine L e v o n M. K h a c h i g i a n i , G e n e v i e v e Evin -, ' F r a n c i s J. M o r g a n 3 D w a i n A. O w e n s b y a n d C o l i n N. C h e s t e r m a n 1 t Department of Hematology. Unit'ervtty of New South Wales. Prince of Wales Hospital. Randwick. NSW 2031. Australia, 2 Mental Health Research hTstitute of Victoria. Parkcille. Vic. 3052. Australia. and ~Department of Biochemistry. La Trobe Unirersity, Bundoora, lOc. 3083. Australia

(Received 3 December 1991).revised received 19 February 1991.accepted 20 March 19911

Synthetic peptides meeting certain guidelines have been used as immunogens to generate antibodies with predefined specificity. We have raised and characterized using established methods a monoclonai antibody against a synthetic peptide corresponding to the 18-amino acid carboxyterminai sequence (A194-2111 of the platelet-derived growth factor ( P D G F ) A chain expressed by the U343 human glioma cell line. This antibody was generated in order to carry out structure-function studies on this region of P D G F whose biological significance is not yet clear. Anti-PDGF-AI94-211 was found to be a low titre, IgMK molecule, with a K d of 2.8 × 10 -7 M. When antibody reactivity was tested with parent P D G F - A A L (A chain homodimer containing a carboxyterminal extension) significant binding was observed. Suprisingly, 125I-PDGF-AAs, consisting of truncated A chains but lacking the extension was also bound. Moreover, poly-L-lysine, /3-thromboglobulin, P D G F - A I 9 4 - 2 1 1 , and myoglobin competed dose-dependently with Iz~I-PDGF-AA L for antibody. ~-~~l-bovine serum albumin was also bound. Examination of the primary sequence of proteins and peptides bound by the antibody revealed only one shared structural motif: a lysyi-lysine moiety, Selected small synthetic peptides containing this and other sequences were used as potential competitors of L'51-PDGF-AI94-211 in antibody binding. Lysyl-lysyl-glycyl-glutamine and lysyl-lysine competed, whereas lysyl-leucine did not. These results suggest that as few as two amino acid residues constitute a functional antigenic determinant and contrast with most previous estimates of the minimum number of residues required. Furthermore. we show that guidelines governing the design of synthetic peptides for their use as antigens to produce monoclonal antibodies of predetermined specificity may be unreliable. Key wt,rds: Synthetic peptid¢: Ant pep', de monoc ona mtibodv Radioimmunoassay;Platelcl tier, cd ~r~,wthfactor

Corrc,spondenc¢" to: C.N Chesterman. Department ol Hematology, Prince of Wales tlospital, Randwick, NSW. 21131, Austr~,lia. Abbret'iations: PDGF, platelet-dcrived grov,'th factor; HPLC, high performance liquid chromatography: PEG,

polyethylene glycol; KLH, keyhole-limpet hemocyanin; BSA, btwine serum albumin; DMSO. dimethyl sulfoxide; SDS. sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. Sequential amino acid residues are represented by single letter codes.

250 Introduction

Monoclonal antibodies directed against specific peptide sequences of native proteins have proved to be of enormous utility in a variety of biological applications such as immunopurification (Chung and Rhee, 1984), immunocharacterization (Sheiton et al., 1990) and immunodetection (Sen et al., 1983; Ahman et al., 1984). A major strategy for developing such monoclonal antibodies involves using as immunogens synthetic peptides having not only a primary sequence identical to a preselected peptide sequence of the native protein but also other specific characteristics (Shinnick et al., 1983). In general, the resulting antibody can be considered to possess predetermined specificity for the chosen sequence of the native protein (Lerner, 1982). However, unexpected crossreactivity with epitopes on unrelated proteins and peptides compromises the value of the antibody as a unique indicator of the peptide sequence originally selected. When crossreactivity is detected it usually reflects substantial structural or sequence homology among the crossreacting species. During the course of generating and characterizing monoclonal antibodies directed against specific peptide sequences of platelet-derived growth factor (PDGF), we observed unexpected crossreactivity of one antibody with several unrelated peptides and proteins. The crossreactivity resulted from a remarkably small shared antigenic recognition site rather than from significant primary sequence homology. The basis for this crossreactivity is reported. In order to orient the reader with respect to the nomenclature, a brief description of PDGF is provided. PDGF, a potent mitogen for cells of mesenchymal origin, is a naturally occurring basic dimeric glycoprotein that exists in several distinguishable forms (Heldin et al., 1988). The basis for this heterogeneity is that two structurally and functionally distinct forms of the component monomeric polypeptide chains of PDGF are recognized. Termed A and B, these distinct chains may combine to form the homodimers PDGF-AA or PDGF-BB, or the heterodimer PDGF-AB, each form having characteristic biological properties (Bywater et al., 1988; Sturani et al., 1989).

Furthermore, microheterogeneity of the A-chain occurs, and two forms have been described. The A s chain consists of 196 amino acid residues, whereas the AL chain coramences with the same initial 193 residues but continues with a unique 18 amino acid extension at the carboxy terminus (A194-211) (Collins et al., 1987). Only one form of the A chain appears to predominate in a specific cell type. For example, PDGF-AA L appears to be the major species expressed by a human glioma cell line U343 in culture (Bonthron et al., 1988), whereas the human osteosarcoma cell line U20S expresses primarily PDGFAA s (Tong et al., 1987). The precise biological significance of the carboxyterminal extension of the A L-chain is not well understood (Rorsman et al., 1988; Maher et al., 1989; Matoskova et al., 1989). To facilitate structure-function studies on the A194-211 moiety we developed a monoelonal antibody directed against a synthetic peptide identical to the octadecapeptide carboxyterminal sequence of the A L chain.

Materials and methods

Materials Recombinant PDGF-AA L and PDGF-AA s were generous gifts of Dr. C.-H. Heldin, Uppsala, Sweden. Bis-diazotised benzidine, polyethylene glycol (PEG, mol. wt. 1300-1600), 2,4,10,14-tetramethylpentadecane, keyhole-limpet hemocyanin (KLH), bovine serum albumin (BSA, fraction V), myoglobin and synthetic peptides glycyl-tyrosine, leupeptin, lysyl-lysyl-glyeyl-glutamine(KKGE, fllipotropin fragment 88-91), lysyl-lysine (KK), and lycyl-leucine (KL) were purchased from Sigma, St. Louis, MO. Insulin (100 U / m l ) was obtained from Commonwealth Serum Laboratories, Melbourne. /3-thromboglobulin was prepared as described previously (Begg et al., 1978). Na125I was obtained from Amersham, and dimethyl sulfoxide (DMSO) was purchased from BDH Chemicals, Victoria.

Peptide synthesis The amino acid sequence of peptides of PDGF were derived from the cDNA sequence of the A chain expressed by the U343 cell line (Betsholtz

251 et al., 1986). Peptide PDGF.A194-211 (YGRPRESGKKRKRKRI,KPT) and two control peptides, PDGF-A 184-200 (LNPDYREEDTGRPRESG) and MR-361-380 (YSPDTQEKGAQEVPFPKTEEV) were synthesized by the automated t-Boc method. PDGF-A184-200 overlaps with PDGF-A194-211 by 7 of its 17 residues, and MR-361-380 represents a segment of the unrelated mineraiocorticoid receptor sequence. A tyrosine residue was introduced at the amino terminus of PDGF-A194-211 in order to allow subsequent radioiodination and coupling. Peptides were purified by HPLC on a reverse-phase Cis column (Waters, Milford, MA) using a gradient of 0-50% acetonitrile/H20 in 0.1% trifluoroacetie acid. Amino acid analysis was carried out to estimate the mass of peptide and ensure correct residue composition. Radioiodination of peptides and proteins was carried out using the chloramine-T method (Greenwood et al., 1963).

Conjugation Bis-diazotised benzidine (BDB) was used to couple PDGF-A194-211 with KLH as described previously (Bassiri and Utinger, 1972). This method was chosen because BDB links peptides and proteins via tyrosine residues. In contrast to linkage to internal residues, coupling to the tyrosine residue at the amino-terminus would facilitate the adoption by the peptide of its natural conformation when presented to the immune system. The resultant conjugate was insoluble and contained approximately 1570 molecules of peptide to one molecule of KLH.

Immunizations Female BALB/c mice (6-8 weeks old) were injected intraperitoneally with conjugate c,~ntaining 0.6 /~g peptide in complete Freund's adjuvant, and boosted three times with similar concentrations of peptide in incomplete Freund's adjuvant at 2-5 week intervals. Free peptide (0.6 /.~g)'in saline was injected intravenously 4 days prior to fusion.

Monoclonal antibody production Mice whose bleeds revealed the strongest reactivity with peptide using radioimmunoassay were sacrificed and splenocytes fused with Sp2/0-Agl4

plasmacytomas at a 5:1 ratio by the method of K6hler and Miistein (1975). To increase the efficiency of hybridoma formation the 47% P E G / 7.5% DMSO mixture was filtered (0.22 p.m Millipore) and adjusted to pH 7.8 prior to use (Rathjen and Underwood, 1985). The hybridomas were resuspended in Dulbecco's modified Eagle's medium containing 20% fetal calf serum, and 50 p.g/ml streptomycin, 50 IU/ml penicillin, 100 /~M hypoxanthine, 0.4 /xM aminopterin and 16 /.tM thymidine. The cells were distributed into 96 well plates (2 x 105/well) and 24 well plates (2 × 106/well) and grown in a humidified incubator with an 8% CO 2 and 92% air gas phase (Borrebaeck, 1984). Supernatants were assayed after 14 days. Hybridomas whose supernatants showed the greatest antipeptide activity were cloned three times by the limiting dilution method. Ascites fluid was generated by injecting 1 x 106 hybridoma cells into mice primed 4 days previously with 0.5 ml pristane (2,4,10,14-tetramethylpentadecane).

Monoclonal antibody purification and isotype determination Ascites fluid was clarified by centrifugation, diluted 1/2 with buffer, and applied to a Sepharose-6B (Pharmacia-LKB, North Ryde) gel filtration column in 0.1 M sodium phosphate buffer, pH 7.4. Peaks with antipeptide antibody activity were identified using radioimmunoassay and SDS-PAGE, and protein concentrations determined using the Coomassie Protein Assay Reagent (Fierce, Rockford, IL). The concentration of the antibody in the pooled eluate was 1.0 mg/ml. The isotype of the antipeptide antibody was determined using the Misotest Kit (Commonwealth Serum Laboratories), the mouse typer kit (Bio-Rad Laboratories), and Ouchterlony immunodiffusion. The antibody was found to be lgM with a K light chain.

Radioimm:moassay Polystyrene microwells (lmmunolon 2, Dynatech) were each coated with 50/.tl of 50/zg/ml sheep anti-mouse immunoglobulin(Silenus Laboratories, Victoria) m 50 mM sodium carbonate buffer, pH 9.6, and incubated overnight at 4°C to facilitate attachment. The wells were washed

252 twice with buffer containing 0.1 M sodium phosphate, pH 7.4, and 0.1% BSA. Non-specific binding sites were blocked by incubating for 1 h at 3 7 ° C with 300/.tl of buffer containing 1% BSA. In selected experiments BSA was replaced with insulin at the same molar concentration both in the wash buffer and in the blocking buffer. Mouse serum, hybridoma supernatant or purified ascites fluid was added (50 p.l) at appropriate dilutions and incubated for 2 h at 37 ° C. The wells were washed and t25I-ligand (100,000 cpm in 50 /.tl) was added prior to incubation overnight at 4 o C. Unbound radioactivity was removed r~y suction, and the wells were again wash,xl with buffer. Bound radioactivity was quantified by counting individual wells in a y-counter (Wa!lae, Decem GTL 300-500). In addition to the assay described, the antigen binding capacity of the monoclonal lgM was also detected using a numbei" of other systems including a sandwich enzyme-linked immunosorbant assay, liquid-phase radioimmunoassay using 12-~I-PDGF-AI94-211, and a solid-phase radioimmunoassay using ~25I-labeled monoclonal antibody and immobilized peptide (data not shown).

Results

Characterization of antipeptide monoclor.al antibody reacticity with the synthetic peptide 'To determine the optimal titer of antibody for interaction with its cognate peptide, a standard titration curve was constructed. At optimal dilution of antibody ( I / 1 0 , 0.11 /xM final concentration), 10.5% of the ~25I-PDGF-A194-211 was bound (Fig. 1). Significant binding of J2sl-PDGFA194-211 to antibody was detected with antibody dilutions un to 1/1000. Competition experiments were designed to establish the specificity of the interaction between antibody and peptide. An excess of unlabeled peptide (0.12 # M ) was sufficient to inhibit completely the binding of ~251-PDGF-AI94-211 to antibody (Fig. 2). In contrast, even a 1.2 tzM concentration of an unrelated peptide, MR-361380, had no effect. In separate experiments equilibrium binding of ~z~I-PDGF-AI94-211 to antibody was deter-

120t m

g.; ,~ ,~

100 ~ . . . . , ~

6o 40 0 . 0

.

1

.

"-

.

.

2

.

.

3

.

.

4

.

.

5

6

LOG ANTIBODY DILUTION

Fig. 1. Titration of antipeptide antibody with 12sl-PDGF-

A194-211. Different dilutions of purified monoclonal antibody were coated onto polystyrenewells and incubated with a tracer amount (100,000 cpm) of ]251-PDGF-AI94-211. Bound radioactivitywas quantified as described in the materials and methods section.

mined over a range of peptide concentrations (data not shown), and Scatchard analysis of the binding data (Collignon et al., 1988) revealed a dissociation constant of 2.8 × 10 -7 M.

Characterization of antipeptide monoclonal antibody with patent proteins To determine whether the antibody could recognise an identical peptide sequence within an intact protein, its reactivity with *251-recombinant PDGF-AAL was characterized. As noted before, P D G F - A A L is a composite molecule containing not only the P D G F - A A s sequence, but also an

140~ 1201

-'

-I--

°

lOOt_T_

~=

,olim

A MR A -'-O.12uM 1.2uM Fig. 2. Specificbindingof antipeptide antibody to z251-PDGFA194-211. L~51-PDGF-AI94-211 was added alone or with either unlabeled PDGF-AI94-211 (0.12 #M or 1.2 /zM)or MR-361-380 (0.12 p.M or 1.2 /zM) to wells precoated with purified monoclonal antibody. Bound radioactivitywas determined as described in the materials and methods section.

253 octadecapeptide carboxyterminal extension which is identical to the synthetic peptide. In addition to binding the synthetic octadecapeptide, the antipeptide monoclonal antibody bound '2SI-PDGFA A L and binding was inhibited by excess unlabeled P D G F - A A L (Fig. 3A). As a control, the interaction betwe~:, P D G F - A A s and the antibody was evaluated. Because P D G F - A A s lacks the carboxyterminal octadecapeptide extension of P D G F - A A L and because it has no internal regions in which the primary sequence is identical to the octadecapeptide fragment, it was not expected to react with the antipeptide antibody. Suprisingly, the antibody was capable of binding tzSI-PDGF-AAs and binding was inhibited in a dose-dependent fashion by excess unlabeled P D G F - A A s (Fig. 3B). However, total binding of antibody was approximately 4-fold lower for P D G F - A A s than for P D G F - A A L despite similar specific activities for each radiolabeled protein. To confirm this unexpected crossreactivity of the antibody, its reactivity towards ~-"~I-PDGF-AA L was evaluat~:d in the presence of excess unlabeled P D G F - A A s (Fig. 3C), and its reactivity towards IzSI-PDGF-AAs was assessed in the presence of excess unlabeled PDGF-AAL (Fig. 3D). Binding of each 125I-iabeled antigen was inhibited by an 1000 !l

4ooo

!

i

tooo oI . 2

.

1

.

o

.

.

1

2

800 i

i

,oo I

x,,.

,ooi -2

-1

1000r I

lOOOt

0

1

....

2

3

~

800 ~ , ~ , , , ~

.x

I

• ",;

o,~.. . . . . . -2

-1

0

1 2 LOG [COMPETITOR, nM]

Fig. 3. Reactivity o f a n t i p e p t i d e antibody with ~z~I-PD(;F-

AA L and 1'51-PDGF-AAs. tzSI-PDGF-AAI was added in the presence of increasingconcentrationsof either unlabeled PDGF-AAI (A) or PDGF-AAs (C) to wells prccoatcd with purified monoclonalantibody. Similarly. ~-'51-PDGF-AAs was added in the presence of selected concentrationsof PDGFAAs (B) or PDGF-AAL (D). Bound radioactivitywas determined as describedin the materials and methodssection.

120j

=g

40

2o

r

I

Q.o~.. 40

~" ~

2g -1

o

I

.

0

1

2

3

4

5

6

LOG [COMPETITOR,nM] Fig. 4. Reactivity of antipeptide antibody with I"--~I-PDGFAAs, t-'51-PDGF-AALand t'-sl-PDGF-AI94-211in the presence of selected competitors. 1"51-PDGF-AAL(e, 500,000 cpm) and I-'SI-PDGF-AAs(©. 500,000 cpm) were added separately in the presence of various concentrationsof unlabeled PDGF-AI94-211 to wells coated with purified monoclonal antibody ( A ). I-'sl-PDGF-AI94-211(200,000 epm) was treated similarly with either PDGF-A194-211 (11)or lysylIvsine ([3) (B). Bound radioactivity was quantitated as described in the materialsand methodssection.

excess of the other unlabeled antigen as competitor. Furthermore, the binding of 125I-PDGF-AAs and IzSI-PDGF-AAL could be competed in a dose-dependent manner with unlabeled PDGFA194-211 (Fig. 4A). The concentrations of peptide required to inhibit by 50% the binding of each radiolabeled isoform were 60 nM and 250 nM respectively (Fig. 4A). Hence, the longer A chain was bound with greater affinity than the short. These data suggest that the antibody recognised only a part of the octadeeapeptide sequence or conformation as an epitope, and that this smaller epitope was also present in one or more regions of the P D G F - A A s molecule.

Effects of unrelated prote#ts on 125I-PDGF-AAL binding to antipeptide antibody To elucidate further the basis of the unexpected crossreaetivity of the antibody and to identify the epitope responsible, a number of proteins and polypeptides were used as potential competitors of 1251-PDGF-AAL binding (Table i). Of the panel of proteins and peptides tested, the rank of effective competitors in order of decreasing molar affinity for the antibody was

254 TABLE I EFFECT OF PROTEINS AND PEPTIDES AS COMPETITORS OF tz'~I-PDGF-AAt. BINDING TO ANTIPEPTIDE MONOCLONAL ANTIBODY Competitor. I

- Log % Inhibition of [1 (M)]~, 1251-PDGF-AAL binding at [I]m~,~" Po y- -Ivsne 6.93 81.3 /3-thromboglobulin 6.68 72.0 PDGF-AI94-211 6.50 89.3 Myoglobin b 25.3 Insulin > 7.34 0 PDGF-AI84-200 > 7.81 0 Leupeptin > 8.44 1) Glyc~'l-tyrosinc > 8.74 0 MR-361-380 > 7.75 0

LysylIvs ne motif + + + + -

Maximum competitor concentrations used were 0.84 .aM poly-t.-lysine. 0.56 .aM /3-thromboglobulin.5.6 .aM PDGFA194-211. 0.69 .aM myoglobin. 2.2 .aM insulin, 6.5 .aM PDGF-At84-200. 27.4 .aM leupeptin. 54.6 .aM glycyl-tyrosine and 5.6 .aM MR-361-380. h Sufficient concentrations of myoglobin were not used to determine [i]50.

poly-L-lysinc, /3-thromboglobulin, A194-211 (the synthetic octadecapeptide identical to the carboxyterminal extension on PDGF-AAL), and myoglobin. A large molar excess of BSA similarly proved to inhibit t25I-PDGF-AAL binding to antibody. However, a complete dose-response curve was not determined and hence its inhibitory capacity could not be quantified. Proteins and peptides that failed to compete with 1251-PDGF-AA L binding to the antibody included leupeptin, insulin, and the synthetic peptides MR-361-380, PDGF-AI84-200, and glycyl-tyrosine. To determine whether the presence of BSA in the washing and blocking buffers had attenuated total binding, insulin was used instead of BSA in the buffers. These conditions resulted in augmented binding of ~2Sl-labeled ligands but no change in the rank order of selected competitors of binding (not shown). In separate experiments, selected t251-1abeled proteins were tested directly for binding to the antipeptide antibody in order to demonstrate that any previously observed inhibition of ~251-PDGFA A L binding resulted from a specific interaction

between competitor and antibody. Results of such experiments (data not shown) confirmed that 125IBSA bound specifically to the antibody whereas ~25I-recombinant h u m a n growth hormone did not. Because poly-L-lysine was the most potent inhibitor of 125I-PDGF-AAL binding and because lysine moieties are the only amino acid residues present in this competitor, it seemed likely that the antigenic site primarily involved iysine residues. Indeed, examination of the primary sequence of each protein or peptide that competed with ~2sI-PDGF-AAL for binding to antibody revealed only one shared structural motif: regions of two adjacent lysine residues. Non-competing proteins and peptides did not contain this motif'. Investigations were carried out to determine whether antibody crossreactivity with poly-L-lysine could be reduced by including in the wash buffer a range of concentrations of sodium chloride between 0-1.0 M. Poly-L-lysine (250 nM) was able to inhibit completely ~251-PDGF-A194-211 binding to antibody even in the presence of 0.5 M NaC! (data not shown). 1 M NaCi, as expected, abolished total binding. These data indicate that non-specific chemoattractive forces are not responsible for crossreactivity.

Effects of small peptide competitors on 1251-PDGFA194-211 binding to antipeptide antibody To determine whether the lysylysine sequence constituted the primary determinant recognised

TABLE I1 EFFECT OF SMALL PEPTIDES AS COMPETITORS OF 12"~I-PDGF-AI94-211 BINDING TO ANTIPEPTIDE MONOCLONAL ANTIBODY Competitor, 1

PDGF-AI94-211 KKGE KK KL

% Inhibition of t25I-PDGF-AI94-211 binding at [lira~x" 100 100 96.5 o

Lysyl-lysine motif + + + -

a Maximum competitor concentrations used were 0.12 .aM PDGF-AI94-211, 5.0 .aM KKGE, 298.5 .aM KK, and 298.5 .aM KL

255 by the antibody, selected small synthetic peptides containing this or alternative sequences were evaluated as competitors of 125I-PDGF-A194-211 binding to antibody (Table II). Compared with unlabeled PDGF-A194-211, the tetrapeptide KKGE proved to be an equipotent competitor, as did the dipeptide lysyl-lysine. However. replacement of the terminal lysine residue with leucine to generate a lysyi-leucine dipeptide resulted in complete loss of activity as a competitor (Table II). The concentration of PDGF-AI94-211 and lysyl-lysine required to inhibit the binding of 12sI-PDGF-A194-211 to antibody were 20 nM and 310 nM respectively (Fig. 4B). This indicated that the antibody bound to PDGF-AI94-211 with an affinity greater than that for the dipeptide.

Discussion It is widely believed that a synthetic peptide corresponding to virtually any region of a protein can be used to produce antibodies that react with the native protein, provided that certain selection guidelines are followed (Niman et al., 1983; Shinnick et al., 1983). The results presented in this study suggest that monoclonal antibodies with preseleeted specificity might not be obtained predictably when synthetic peptides selected according to standard guidelines are used as immunogens. In addition our findings indicate that a remarkably small epitope, namely a specific locus of only two adjacent amino acid residues, can potentially function as a discrete antigenic determinant and thus can constitute the structural basis for antibody crossreactivity with multiple target molecules. Our results substantiate earlier reports that, in the design of monospecific monoclonal antibodies using peptides as immunogens, crossreactivity of a resultant antibody with unrelated antigens can occur (Tanaka et al., 1985: Weiss et al., 1987). Moreover, our results are consistent with the notion that antipeptide monoclonal antibodies appear to recognize not the entire sequence of a protein or peptide but smaller sets of residues (Schools et al., 1988). Unpredictable crossreactivity may be observed despite painstaking precautions to limit antibody specificity solely to the

antigen used as immunogen. Although generation of monospecific monoclonal antibodies is in part serendipitous, well-substantiated strategies have been developed both to increase the likelihood of obtaining antibodies with predetermined specificity and to minimize the chance of crossreactive antibodies being produced (Shinnick et al., 1983). When antibodies directed against a specific peptide sequence of a native protein are desired, the choice of corresponding peptide immunogen is particularly critical (Geysen et al., 1987). In general synthetic peptides corresponding exactly to sequences exposed on the surface of a native protein are superior to interior sequences as antigens. Although there is a lack of consensus regarding optimal peptide length, peptides of 10-20 amino acid residues are usually employed. Other important criteria include high hydrophilicity and the presence of proline residues to favor specific peptide conformations able to mimic regional conformations surrounding proline-associated fl turns in the parent molecule. The design of the octadecapeptide PDGF-AI94-211 used in our study conforms precisely to these guidelines, its sequence is identical to the corresponding region of P D G F - A A L. Because this sequence corresponds to the carboxyterminus, it is expected to occupy a surface location on the native protein (Thornton et al.. 1983; Getzoff et ai., 1988; Tanaka et al., 1988). The peptide length is within the recommended range. The peptide also corresponds to the region of P D G F - A A L with highest hydrophilicity predicted from a hydropathic prow file using the method of Kyte and Doolittle (1982). In addition it contains two non-adjacent internal proline residues. Collectively these properties supported the prediction a priori that our synthetic octadecapeptide would be an ideal candidate immunogen for eliciting monoclonal antibodies of predetermined specificity. However, despite its favorable profile, thi~ immunogen resulted in a widely crossreactivc antibody. Precisely, why such an antibody was obtained is not entirely clear. Although the carrier protein and coupling agent can influence the way in which an antigen is presented to the immune system, we used a common carrier molecule and followed a standard coupling protocol. Nonetheless, the coupled octade-

2.~6 capeptide may not have been exposed sufficiently or have been presented optimally to elicit a highly specific antibody response. The immunization and harvesting schedule, the cell fusion technique, and the antibody screening method for clone selection all followed well-established standard protocols and were not likely to have contributed to the generation or selection of a crossreactive antibody. Furthermore, despite the known experimental limitations associated with the use of IgM antibodies (Rodwell et al., 1983), this isotype is not recognized as having a high propensity for crossreactivity. Approaches that have been used by other investigators to identify the side chains characteristic of the binding site on an antigenic protein or peptide, include proteolysis of the antigen-antibody complex (Jemmerson and Paterson, 1986), structural specificity studies (Mehra et al., 1986), peptide mapping (Houghten, 1985), and chemical modification of bound and free antigen (Cooper et al., 1987). We explored the structural basis for the crossreactivity of anti-PDGF-A194-211 with a panoply of target molecules using competitive radioimmunoassay. Not only did the antibody react with its cognate ~ynthetic peptide PDGFA194-211 (Fig. 1), but also with the parent PDGF-AA L protein which contained an identical sequence (Fig. 2). Suprisingly, crossreactivity with PDGF-AA s, which lacks a sequence corresponding to synthetic peptide A194-211, was also observed (Fig. 3). Based upon the results of experiments with selected protein or peptide competitors of antibody binding to 12sI-PDGF-AAL (Table I) or ~I-PDGF-A194-211 (Table II), the only identifiable common structural motif that could account for the interaction of each moiety with the antibody was a pair of adjacent lysine residues. The antibody could recognize the lysyllysine determinant whether it occurred as an isolated dipeptide or was flanked by other amino acids. These findings suggested strongly that two adjacent lysine residues were sufficient to constitute the primary antigenic determinant recognized by the antibody. That lysine residues are found to constitute the epitope is not entirely unexpected considering that this amino acid has been assigned a relatively high 'propensity factor'. This factor predicts residues in synthetic peptides

that could elicit antibodies reactive with the native proteins (reviewed in Getzoff et al., 1988). We cannot determine unambiguously whether the antibody requires the lysine residues to be adjacent in the primary peptide sequence of the antigen, or whether they could be separated in the primary sequence but in close spatial proximity as a result of peptide folding. There is no consensus regarding the smallest moiety that can function as a complete antigenic determinant. If only antipeptide and antiprotein antibodies are considered, most estimates for the minimum number of amino acids which constitute a sequential epitope is between five and eight ((3eysen et al., 1987). However, for selected antibodies as few as three amino acids appear to be sufficient to supply both the intrinsic energy and all the determinants of specificity of the antigen-antibody interaction (Weiss et al., 1987). Using a direct enzyme-linked immunosorbent assay, Geysen (1985) was able to show that antibodies directed towards lysozyme, myoglobin and foot-and-mouth disease virus, also interacted weakly with dipeptides and even single amino acids. Geysen surmized that antibody specificity for a given antigen is dependent not merely on one or two residues but also on the immediate environment of the contact residues (three or four amino acids). The present study, however, indicates that anti-PDGF-A194-211 is capable of interacting with proteins and peptides differing markedly in length and conformation as long as the lysyl-lysine motif is present. Thus, antibody specificity can be determined by as few as two residues and this specificity need not be restricted by the surrounding environment and three-dimensional structure of the antigenic site. In summary, our results contrast with most previous estimates of the minimum number of amino acids that can serve as a functional antigenic determinant. For selected antibodies two amino acids clearly are sufficient to constitute a complete antigenic determinant. In addition, our results provide evidence that the accepted criteria governing choice of synthetic peptide may be unreliable. Although these criteria provide a useful guideline, the practical approach for generating monospecific monoclonal antibodies of defined specificity remains empirical.

257

Acknowledgements This work was supported by a Queen Elizab e t h II S i l v e r J u b i l e e T r u s t A w a r d , a p r o j e c t grant from the National Health and Medical Research Council of Australia, and equipment grants from the Clive and Vera Ramaciotti Foundations and Utah Foundation. We thank Dr. C.-H. Heldin (Ludwig Institute o f C a n c e r R e s e a r c h , U p p s a l a , S w e d e n ) for p r o viding recombinant PDGF-AA isoforms, Dr. M.C. Stuart (Garvan Institute of Medical Research, Sydney) for recombinant eSI-human growth hormone, and Dr. A. Tseng (Pacific-Biotechnology, Sydney) for assistance with the amino acid analysis o f p e p t i d e s .

References Altman. A.. Cardenas. J.M., Houghten, R.A., Dixon, F.J. and Theofilopoulos, A.N. (1984) Antibodies of predetermined specificity against chemically synthesized peptides of human interleukin 2. Proc. Natl. Acad. Sci. U.S.A. 81. 2176. Bassiri, R.M. and Utinger. R.D. (1972) The preparation and specificity of antibody to thyrotropin releasing hormone. Endocrinology 90, 722. Begg, G.S., Pepper, D.S., Chesterman, C.N. and Morgan, F.J. (1978) Complete covalent structure of human ,8-thromboglobulin. Biochemistry, 17, 1739. Betsholtz, C., Johnsson, A., J-ieldin, C.-H., Westermark. B., Lind, P.. Urdea, M.S., Eddy, R.. Shows. T.B.. Philpott, K., Mellor, A.L., Knott. T.J. and Scott. J. (1986) eDNA sequence and chromosomal localization of human plateletderived growth factor A chain and its expression in tumor cell lines. Nature 320, 695. Bonthron. D.T., Morton, C.C.. Orkin. S.H. and Collins. T. (1988) Platelet-derived growth factor A chain: Gene structure, chromosomal location and basis for alternative splicing. Proc. Natl. Acad. Sci. U.S.A. 85, 1492. Borrebaeck, C.A.K. (1984) Dependence on T-cell-replacing factor and immunogenic endose for the production of monoclonal antibodies using the in vitro immunization technique. Mol. Immunol. 21. 841. Bywater, M., Rorsman. F., Bongcam-Rudloff, E., Mark, G., Hammacher. A., Heldin, C.-II., Westermark. B. and C. Betsholtz (1988) Expression of recombinant platelet-derived growth facto( A- and B-chain homodimcrs in rat-I cells and human fibroblasts reveals differences in protein processing and autocrine effects. Mol. Cell. Biol. 8, 2753. Chuug, H.K, and Rhee, S.G. (1984) Separation of glutamine synthetase species with different states of adenylation by chromatography on monoclonal anti-AMP antibody affinity columns. Proc. Natl. Acad. Sci. U.S.A. 81.4t~77.

Collingnon, A.. Geniteu-Legendre. M., Sandre. C., Quero, A. -M. and Labarre. C. (1988) Specific binding characteristics of high affinity monoclonal antidigitoxin ar.tibodies. Hybridoma 7, 355. Collins. T., Bonthron. D.T. and Orkin. S.H. (1987) Alternative RNA splicing affects function of encoded platelet-derived growth factor A chain. Nature 328. 621. Cooper. H.M.. Jemmerson.. Hunt. D.F., Griffin. P.R.. Yates. J.R.- Shabanowitz, J.. Zhu, N. and Paterson, Y. (1987) Site-directed chemical modification of horse cytochrome c results in changes in antigenicity due to local and long range conformational perturbations. J. Biol. Chem. 262. 11591. -. Getzoff. E.D., Tainer. J.A., Lerner. R.A. and Geysen. H.M. (1988) The chemistry and mechanism of antibody binding to protein antigens. Adv. lmmunol. 43, I. Geysen, H.M. (1985) Antigen-antibody interactions at the molecular level: adventures in peptide synthesis. Immunol. Today 6. 364. Geysen, H.M.. Rodda. S.J., Mason. TJ.. Tribbick. G. and Schools, P.G. (1987) Strategies for epitope analysis using peptide synthesis. J. lmmunol. Methods 102. 259, Greenwood. F.C., Hunter. W.M. and Glover. J.S. (1963) The preparation of (311-labeled human growth hormone of high specific radioactivity. Biochem. J. 89, 114. Heldin, C.-H.. Hammacher. A.. Nister. M. and Westermark. B. (1988) Structural and functional aspects of platelet-derived growth factor. Br. J. Cancer 57. 591. Houghten. R.A. (1985) General method for the rapid solidphase synthesis of large numbers of peptides: Specificity of antigen-antibody interaction at the level of individual amino acids. Proc. Natl. Aead. Set. U.S.A. 82. 5131. Jemmerson. R. and Patterson, Y. (1986) Mapping epitopes on a protein antigen by the proteolysis of antigen-antibody complexes. Science 232. 1001. K/ihler. G. and Milstein. C. (1975) Continuous cultures of fused cells secreting antibody of predetermined specificity. Nature 256. 495. Kyte. J. and Doolittle, R.F. (1982) A simple method for displaying the hydropa:thic character of a protein. J. Mol. Biol, 157. 11)5. Lerner. R.A. (1982) Tapping the immunological repertoire to produce antibodies of predetermined specificity. Nature 299, 592. Maher. D.W.. Lee. B.A. and Donoghue, D.J. (1989) The alternatively spliced exon of the platelet-derived growth factor A-chain encodes a nuclear targeting signal. Mol. Cell. Biol. 9, 2251. Matoskova, B., Rorsman. F., Svensson, V. and Betsholtz, C. (1989) Alternative splicing of the platelet-derived growth factor A-chain transcript occurs in normal as well as tumor cells and is conserved among mammalian species. Mol. Cell. Biol. 9. 3148. Mehra. V.. Sweetsen, D. and Yound, R.A. (1986) Efficient mapping of protein antigenic determinants. Proc. Natl. Acad. Sci. U,S.A. 83. 7013. Niman, H.L, Houghten. R.A.. Walker. L.E.. Reisfeld, R.A.. Wilson. I.A., Hogle. J.M. and Lerner. R.A. (1983) Genera-

tion of protein reactive antibodies by short peptides is an event of high frequency: Implications for the structural basis of immune recognition. Proc. Natl. Acad. Sci. U.S.A. 80, 4949. Rathjen, D. and Underwood, P.A. (1985) Optimization of conditions for in vitro antigenic stimulation of dissociated mouse spleen cells for the production of monoclonal antibodies against peptide hormones. J. Immunol. Methods 78, 227. Rodwell, J.D., Gearhart, P.J. and Karush, F. (1983) Restriction in lgM expression. J. Immunol. 130, 313. Rorsman, F., Bywater, M., Knott, T.J., Scott, J. and Betsholtz, C. {1988) Structural characterization of the human platelet-derived growth factor A chain cDNA and gene: Alternative exon usage predicts two different precursor proteins. Mol. Cell. Biol. 8, 571. Schoofs, P.G.. Geysen, H.M., Jackson, D.C., Brown, L.E., Tang, X.-L. and White, D.O. (1988) Epitopes of an influenza viral peptide recognized by antibody at single amino acid resolution. J. lmmunol. 140, 611. Sen, S., Houghten. R.A.. Sherr, C.J. and Sen, A. (1983) Antibodies of predetermined specificity detect two retroviral oncogene products and inhibit their kinase activities. Proc. Natl. Acad. Sci. U.S.A. 80, 1246. Shelton, E.R., Cohn, R., Fish, L., Obernolte. R., Tahilramani, R., Nestor, J.J. and Chan, H.W. (1990) Characterization of

/3-amyloid precursor proteins with or without the protease-inhibitor domain using antipeptide antibodies. J. Neurochem. 55, 60. Shinnick, T.M., Sutcliffe, J.G., Green, N. and Lerner, R.A. (1983) Synthetic peptide immunogens as vaccines. Ann. Rev. Microbiol. 37, 425. Sturani, E, Brambilla, R., Morello, L., Cattaeo, M.G., Zippel, R. and Alberghina, L. (1989) Effect of different forms of the platelet-derived growth factor on cellular responses in mouse Swiss 3T3 fibroblasts. FEBS Lett. 255, 191. Tanaka, T., Slamon, D.J. and Cline. M.J. (1985) Efficient generation of antibodies to oncoproteins by using synthetic peptide antigens. Proc. Natl. Acad. Sci. U.S.A. 82, 3400. Thornton, J.M. and Sibanda, B.L. (1983) Amino- and carboxy-terminal regions in globular proteins. Mol. Biol. 167, 443. Tong, B.D., Auer, D.E., Jaye, M., Kaplow, J.M., Ricca, G., McConathy, E., Drohan, W, and Deuel, T.F. (1987) cDNA clones reveal differences between human glial and endothelial cell platelet-derived growth factor A-chains. Nature 328, 619. Weiss, E.R., Hadcock, J.R., Johnson, G.L. and Malbon, C.C. (1987) Antipeptide antibodies directed against cytoplasmic rhodopsin sequences recognise the fl-andrenergic receptor. J. Biol. Chem. 262, 4319.