Selection of peptides inhibiting a b-lactamase-like activity Anne-Sophie Yribarren, Daniel Thomas, Alain Friboulet and Be ´ range ` re Avalle Ge ´ nie enzymatique et cellulaire, UMR 6022 CNRS, Universite ´ de Technologie de Compie ` gne, France A library of random peptide sequences was used to select peptides that inhibit an anti-idiotypic catalytic Ig, immuno- globulin (IgG) 9G4H9, with a b-lactamase-like activity. This library displays cyclic heptapeptides on the surface of bac- teriophagesandrepresentsacollectionofupto4.5· 10 9 peptides. The first selection step aimed at enriching the lib- rary in species that bind to the whole Ig molecule. The sec- ond step was to discriminate peptides that bind to part of the molecule other than the active site. Selected peptides were then screened by surface plasmon resonance analysis. Those displaying measurable K d values were assayed for their ability to inhibit the catalytic Ig. Keywords: catalytic antibodies; anti-idiotypy; b-lactamase; phage display library; constrained peptide. The elucidation of the rules governing the relationships between the structure of proteins and their function is currently one of the major challenges. In this field, combinatorial approaches offer tremendous possibilities. The main requirement is to be able to sort, from the engineered diversity, mutants or new compounds according to the required function. Random peptide libraries are inexhaustible sources of small compounds likely to bind targets such as enzymes, antibodies, receptors or even DNA (recently reviewed in [1]). Combinatorial libraries can be chemically synthesized or displayed on engineered bacteriophages, as proposed by Smith’s pioneer studies from the 1980s [2,3]. The use of phages allows one to work on huge libraries. It allows rapid isolation of the sequence of the selected compounds, leading to the identification of residues involved in the interaction [4]. We have developed previously an original approach to generate catalytic proteins, other than enzymes, yielding valuable information about structure–function relationships [5–8]. According to Jerne’s concept [9], some of the anti- idiotypic Igs generated by immunization by antibodies may mimic some of the antigenic properties and functions. These antibodies are said to be the Ôinternal imagesÕ of the original antigens. We applied this theory to the elicitation of a new catalyst, displaying a b-lactamase-like activity. Starting from b-lactamase, we have produced an inhibitory active site-specific Ig (Ab1), capable of inducing anti-idiotypic Igs (Ab2) displaying the model-like catalytic activity [10]. We have also demonstrated that the b-lactamase-like Ig (IgG 9G4H9) is a true anti-idiotypic Ig by generating polyclonal Ab3, which recognizes b-lactamase [7]. The Ab2 catalytic Ig would then be of valuable help to elucidate on the one hand some structure–function relationships associated with b-lactamase activity and, on the other hand, the mecha- nisms governing the transfer of catalytic information throughout the idiotypic network. In this context, we proposed to achieve the selection of peptide inhibitors, specific for the activity observed in the b-lactamase Ig. We have carried out a two-step selection procedure from a random library of constrained peptides displayed on bacteriophages, based on exclusive affinity for theIgactivesite. Materials and methods Production and purification of IgG 9G4H9 Hybridomas were grown in Dulbecco’s Modified Eagle’s Medium (Eurobio) containing 10% (v/v) fetal bovine serum, 1% (w/v) L -glutamine 200 m M , 1% (w/v) pyruvate 100 m M ,50UÆmL )1 penicillin and 50 lgÆmL )1 strepto- mycin. Harvested cells were expanded by injection into pristane (2,6,10,14 tetramethyl pentadecan)-treated BALB/ C mice. Resulting ascitic fluids were collected 10 days later, centrifuged at 350 g for 5 min at 4 °C and stored at )20 °C. Purification of monoclonal Igs was carried out on a HiTrap rProtein A column (Amersham Pharmacia Biotech) that presents a high affinity for the constant region of IgGs of isotype 2b, such as 9G4H9. The binding buffer contained 50 m M Tris/HCl,pH9,and0.2 M NaCl. The elution buffer was obtained from Biorad and was used according to the manufacturer’s instructions. The IgG-containing fractions were dialysed overnight against a 0.1 M sodium phosphate buffer, pH 7.4. After concentration by centrifugation at 7000 g for 45 min on 30 kDa Microsep TM filters, IgG concentration was determined by the bicinchoninic acid assay (BCA, Sigma). The purity of the preparation was estimated by silver-stained 8% (w/v) polyacrylamide gel electrophoresis. Amplification and purification of phage peptide libraries The random cyclic heptapeptides library (named CX 7 C) was supplied by E. Koivunen, Dept of Biosciences, Division Correspondence to B. Avalle, Ge ´ nie enzymatique et cellulaire, UMR 6022 CNRS, Universite ´ de Technologie de Compie ` gne, BP 20529, 60205 Compie ` gne Cedex, France. Fax: + 33 3 44 20 39 10, Tel.: + 33 3 44 23 44 12, E-mail: avalle@utc.fr Abbreviations: TU, transforming units; SPR, surface plasmon resonance; BLIP, b-lactamase inhibitory protein. (Received 6 March 2003, revised 15 April 2003, accepted 6 May 2003) Eur. J. Biochem. 270, 2789–2795 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03651.x of Biochemistry, University of Helsinki, Finland [11]. The library was amplified by infection of mid-log phase Escherichia coli cells by 3.10 10 of phage transforming units (TU) for 30 min, without shaking, at 37 °C. The infected bacteria were then grown overnight, shaken at 37 °Cin1L Luria-Bertani medium containing 20 lgÆmL )1 tetracycline. The amplified phages present in the culture supernatant were precipitated overnight at 4 °C by adding 20% (v/v) of 20% polyethylene glycol 8000, 2.5 M NaCl (PEG/NaCl). Phages were then pelleted by centrifugation at 10 000 g for 45 min at 4 °C, suspended in 10 mL deionized water and filtered through a 0.45 lm filter unit (Millipore). The phages were reprecipitated in PEG/NaCl for 1 h at 4 °C. The library was finally recovered in 1 mL Tris-buffered saline (TBS), pH 7.5, containing 50 m M Tris/HCl and 150 m M NaCl. The library was stored at )20 °C. Titration of the library was performed by infection of mid-log phage E. coli bacteria by dilutions of phages for 30 min at 37 °C without shaking. One hundred microlitres were then plated on solid Luria–Bertani agar plates containing 20 lgÆmL )1 tetracycline. The plates were incu- bated overnight at 37 °C and visible clones counted. Solid phase selection procedures Experimental concentration conditions are summarized in Table 1. Positive selection. Maxisorb plates (96 wells) were pur- chased from Nunc. The panning method was derived from Parmley and Smith [3]. Unless otherwise stated, each step of the procedure was followed by three washings of the plates in TBS containing 0.1% Tween 20 (TTBS) and plates were incubated1hat37°C without shaking. The microplates were coated with 100 lLof2.5lgÆmL )1 rat anti-(mouse IgG) in 0.1 M carbonate buffer pH 8.6. Wells were then saturated with 100 lL of 2% milk in NaCl/ Tris (MTBS) (w/v), and 100 lL of 1–10 lgÆmL )1 IgG 9G4H9 in NaCl/Tris were added to the saturated wells. Between 5 · 10 10 and 2.5 · 10 11 TU were first incubated for 10 min at 37 °C in MTBS, to avoid the binding of displayed peptides to milk proteins. This solution was then transferred to the plate and incubated for 2 h at 25 °C. This step was followed by 10 TTBS washings, each washing lasting at least 5 min. The bound bacteriophages were eluted from the plates by a 10 min incubation at room temperature with a solution containing 0.1 M HCl adjusted at pH 2.2 by the addition of solid glycine. The pH of the collected phage solution was neutralized by 11% of 2 M Tris (v/v). The enrichment of the starting library in phage specific for the IgG was followed by titration of the solutions before and after each round of selection. This is expressed as the following ratio: TU at round (n)/TU at round (n)1). After each round of selection, eluted phage were ampli- fied by infection of 20 mL mid-log phase bacteria as described below, then used for the subsequent round. Negative selection. This selection procedure was derived from the positive one. The steps and washings were the same, but the IgG 9G4H9 was exposed to contact with penicillin G under conditions described previously [12]. The covalent complex thus formed was added to saturated wells. The bacteriophage issued from the positive selection that were not retained on this complex, i.e. that are not specific for the active site, were directly recovered and amplified for further rounds of selection. Phage ELISA Unless otherwise stated, wells were filled with 100 lLofthe indicated solutions, plates incubated for 1 h at 37 °Cand each step followed by three washings in TTBS. One microgram IgG 9G4H9 in 0.1 M carbonate buffer, pH 8.6, was coated on a Maxisorb microplate. Wells were saturated by MTBS. Variable amounts of bacteriophage, from 5 · 10 7 to 5 · 10 10 TU per well, were then added and incubated for 2 h in MTBS at room temperature. After 10 washes in TTBS, bound phage were detected by 1 lganti- M13 horseradish peroxidase-conjugated monoclonal anti- bodies (mAbs) (Amersham Pharmacia Biotech). The fixation of labelled antibodies was measured by the addition of 0.3 mgÆmL )1 azino ethylbenzthiazoline sulfonic acid (Sigma), a substrate of peroxidase, in 50 m M citrate buffer, pH 5, with 0.05% H 2 O 2 . Absorbance was read at 405 nm. Sequencing of phage DNA Oligonucleotides were purchased from Oligo Express and DNA sequencing was performed by Genome Express (both Paris, France). The DNA peptide inserts of individual plated clones issued from rounds of selection were amplified by PCR performed on an Eppendorf thermocycler. A precycle at 94 °C for 10 min and 35 cycles of denaturation at 94 °Cfor 1 min, AB 348 and AB 349 primers annealing at 54 °Cfor 1 min and elongation at 72 °Cfor1.5min.Taq DNA polymerase was purchased from New England Biolabs. The forward primer AB 348, 5¢-TTAGCAAAACCTC ATACAGAA-3¢, and the backward primer AB 349, 5¢-GATGCTGTCTTTCGCTGCTGAG-3¢,wereusedfor DNA amplification. The M13 forward primer, 5¢-ATTCACCTCGAAAGCAAGCTG-3¢,wasusedfor sequencing. Table 1. Experimental conditions of selection. Concentrations of rat anti-(mouse IgG), IgG 9G4H9 and phages from the library CX 7 Care indicated from the first to the sixth round of positive selection on IgG 9G4H9 and for the two rounds of negative selection on IgG 9G4H9 complexed to its suicide substrate penicillin G. The terms in brackets indicate the name of the round used in the text. Catalytic IgG 9G4H9 (lgÆmL )1 ) CX7C library ( M ) 1 (R 1) 2.5 10 8.3 · 10 )8 2 (R 2) 2.5 10 1.66 · 10 )8 3 (R 3) 2.5 1 1.66 · 10 )8 4 (R 4) 2.5 1 1.66 · 10 )8 5 (R 5) 2.5 1 1.66 · 10 )8 6 (R 6) 2.5 1 1.66 · 10 )8 1 (NS 1) 2.5 1 1.66 · 10 )8 2 (NS 2) 2.5 1 1.66 · 10 )8 2790 A S. Yribarren et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Synthetic peptides Several selected peptides displayed on the surface of bacteriophages were synthesized in their soluble form. The cyclic and biotinylated peptides containing 18 amino acids were purchased from Eurogentec (Angers, France) and designed with the two N-terminal residues belonging to the pIII protein of the bacteriophages. These were followed by two cysteines flanking the peptide sequence. The C-terminal sequence corresponds to the biotin moiety (linked to the lysine of Gly-Ala-Ala-Gly-Ala-Glu-Lys). The sequence of free peptides is thus: NH 2 GACX 1 X 2 X 3 X 4 X 5 X 6 X 7 CGAA GAEKCONH 2 ,withX n corresponding to the random residues of the displayed peptide. Synthesis was performed under oxidative conditions, to preferentially form the cyclizing disulfide bond. Between 1 and 3 m M of peptide stock solutions were stored at )20 °C and prepared in water or in 50 m M acetic acid solution according to their hydrophobic features. Surface plasmon resonance (SPR) analysis SPR analysis is performed with a BIAcore X (Biacore AB, Sweden). Streptavidin coated chips and running buffer HBS pH 7.4 containing 10 m M Hepes, 0.15 M NaCl, 3.4 m M EDTA and 0.005% surfactant P20, were from Pharmacia (Uppsala, Sweden). Immobilization of the biotinylated ligands (40 lgÆmL )1 in HBS) on the surface of SA chips was done by injecting 40 lL at a flow rate of 5 lLÆmin )1 .TheK d values were determined by injecting 40 lL of Ig solutions ranging from 100 to 600 n M in HBS at a flow rate of 7 lLÆmin )1 .The regeneration of the surface was carried out by injecting 6 lLof50m M HCl. Curve fitting and kinetic data analyses were carried out using BIAEVALUTION 3.0.1 software. Kinetic assays The catalytic activity of 2 l M IgG 9G4H9 was measured spectrophotometrically by following the hydrolysis of ampicillin at 232 nm in 0.1 M phosphate buffer (pH 7.4). IC 50 values were determined on the basis of the residual activity of 9G4H9 after a 20 min incubation with different amounts of peptides or phage. Results Selection of peptides binding to the Ig The library (CX 7 C) of constrained heptapeptides displayed on the surface of bacteriophages was kindly provided by E. Koivunen [11]. Its diversity extends to 4.5 · 10 9 different combinations. During the amplification and propagation of phage, we always ensured that the whole diversity was expressed by titrating the purified phage. The first step of the selection procedure was performed on plates coated with monoclonal antibodies directed against the constant region of mouse IgGs, as 9G4H9. The active site within the variable region of 9G4H9 should thus be favorably exposed to the library. We also confirmed that 9G4H9 activity was fully maintained in TBS. From the third round, the selection pressure was increased by reducing the concentration of 9G4H9 from 10 to 1 lg (Table 1). The peptides selected from this third round should therefore display improved affinities for the Ig. After five rounds, the yield of specific peptides retained on 9G4H9 was 1000-fold higher than in the original library. As shown in Fig. 1 (abscissa R 1 to R 5), a significant enrichment of specific peptides was obtained from the second round. The signals measured in phage ELISA from the first to the sixth round (Fig. 2, abscissa R 1 to R 6) indicate that the affinity of the pool of selected peptides progressively increases. We randomly chose 27 isolated clones from the fifth round, which were amplified and sequenced. These peptides could then be gathered in 10 families according to their amino acid content (Table 2). Within each family, the peptides share the same sequence. The analysis of the Fig. 1. Enrichment of the CX 7 C library in phages recognizing the target. The yield of recovered TU corresponds to the ratio of phages collected at the end of each round over phages inserted in this round. Open bar: Phages recovered during the positive steps of the selection (rounds R1 to R5) carried out on IgG 9G4H9. (Grey bar): Phages recovered after the negative selection (rounds NS 1 and NS 2), i.e. with IgG 9G4H9- penicillin G covalent complex as a target. Fig. 2. Phage ELISA on enriched CX 7 C library at each round of selection. Values of absorbance at 405 nm correspond to the fixation of 5 · 10 10 M TU to 1 lg of immobilized IgG 9G4H9, revealed by anti- M13 horseradish peroxidase-conjugated mAbs in the presence of azino ethylbenzthiazoline sulfonic acid. Open bar represents the phages of the positive step of selection (R 1 to R 6) (grey bar) those of the negative step (NS 1 and NS 2), and black bar the original library. The background signal corresponds to the fixation of CX 7 Ctothe microplate not coated with IgG 9G4H9. Standard deviations are cal- culated from three independent experiments. Ó FEBS 2003 Peptides inhibiting a catalytic antibody (Eur. J. Biochem. 270) 2791 aligned sequences, notably the frequency of amino acids at each position, provided two pieces of information. First, the three central residues X 3 X 4 X 5 (notation as in Synthetic peptides section) are mostly hydrophobic (more than 59% of the considered residues). Trp is present in 19% of these three residues, Leu in 11% and Met and Phe in 7%. Second, the Pro is present in 18.5% of the peptide sequences. Binding of the peptides displayed on phages to IgG 9G4H9 was measured by a monoclonal phage ELISA. Clones from the fifth round of positive selection were assayed in addition to the original library CX 7 C (Fig. 3). They all showed a significant capability to bind the Ig. The library gave a weaker ELISA signal than isolated clones that have encountered five rounds of selection because it contains many phage that are not specific for the Ig. This first positive step of selection is consequently efficient, because it allows one to obtain peptides that significantly bind to the target. Selection of peptides binding to the Ig active site The second step of the procedure, called negative selection, was carried out on phage from the fifth round of positive selection. This step was applied using IgG 9G4H9 whose catalytic site was covalently blocked by penicillin G. Previous studies on 9G4H9 mechanism have shown that hydrolysis of penicillin G by the Ig behaves as a suicide substrate mechanism by rapidly forming a covalent irre- versible complex [12]. This property was thus exploited to select peptides with inhibitory features. Peptides that both (a) bind to the Ig and (b) do not bind to the Ig-suicide substrate complex should be specific for the Ig active site. They were then amplified prior to another round of negative selection. Figure 1 (abscissa NS 1 and NS 2) shows that the two rounds of negative selection were efficient in enriching the yield of peptides not retained to the Ig inactivated by penicillin G, i.e. specific for the active site. The phage ELISA carried out on the free IgG 9G4H9 (Fig. 2, abscissa NS 1 and NS 2) gave significant signals, meaning that some of the peptides isolated from the two negative rounds remain specific for the Ig. The results were compared with those of the sixth round of positive selection (R 6): no gain of affinity for this round of selection was observed whereas the affinity was improved for the two rounds of negative selection. However, no significant difference could be observed between the first and the second round of negative selection in our experimental conditions. On the one hand, there is an increase in the yield of specific peptides recovered, but on the other hand the affinity remains the same. Nevertheless, increasing the selection pressure by decreasing the amount of Ig linked to its substrate should not be efficient in improving affinities because, for negative selec- tion, only peptides that are not retained are collected. We isolated and sequenced clones from the two rounds of negative selection. Peptides issued from the negative selec- tion display a high percentage of Trp residues (13%) (Table 3). Three families of peptides previously sequenced at the fifth round of positive selection were retrieved in the first round of negative selection. The three families of peptides have a Pro at their C-terminal (X 6 or X 7 ), and they contain a Trp. More than 38% of the residues are hydrophobic. Two families (2 and 5) were also retrieved in the second round. The peptides of these two families Table 2. Amino acid sequences of the 10 families of the 27 selected clones. Clones isolated from the five first rounds of positive selection were isolated and sequenced. Family Amino acid sequence 1 WVGIARD 2 WDTGDPE 3 IVNGIGF 4 TQNMRSA 5 YHFLWGP 6 EGWLMLM 7 TGWREMF 8 HVWFPES 9 LPWLGPR 10 GWVGQLG Fig. 3. Phage ELISA on clones isolated from the fifth round of positive selection. The name of the clone (F x ) corresponds to the family reported on Table 2 to which belongs the clone. Open bar corresponds to the signal given by isolated clones and black bar corresponds to the signal obtained with the library CX 7 C. Absorbance values at 405 nm are the means of three identical experiments carried out independently. Error bars indicate standard deviations. Table 3. Sequences of clones issued from the negative selection. Family (positive selection) First round Second round Name of the free corresponding peptide Family 2 WDTGDPE a,b WDTGDPE a,b Pep 96 Family 5 YHFLWGP a,b YHFLWGP a,b Pep 90 Family 9 LPWLGPR a Pep 91 GPHWWGY Pep 92 WGHYWWQ Pep 93 ERDKHEG Pep 94 ERLSPSF Pep 95 GWVGQLG SAKYALW RSQGTLE PLDLLPL a Sequences that were already identified after the fifth round of positive selection. b Sequences that are retrieved in both rounds of negative selection. 2792 A S. Yribarren et al.(Eur. J. Biochem. 270) Ó FEBS 2003 appeared clearly as good candidates for the inhibition of the catalytic activity shown by IgG 9G4H9. Analysis of binding properties by surface plasmon resonance Phage ELISA appeared to be an efficient qualitative assay. However, it cannot be used to measure accurate K d values because the affinity of the Ig for phage particles probably takes part of the signal. We thus worked on synthetic free cyclic peptides. The free structures corresponding to the sequences of seven negatively selected peptides (Table 3), displaying the best signals in phage ELISA, were synthesized according to the following feature: NH 2 GACX n CGAAGAEKCONH 2 , with X n corresponding to the residues of the displayed peptide. The synthesis was performed under oxidative conditions in order to induce the formation of the disulfide bond by association of the two bordering cysteines, thus leading to a cyclic structure. This cyclic structure is flanked at the N-terminal end by two residues belonging to the pIII phage coat protein and at the C-terminal by a biotin moiety. For SPR analysis, peptides were immobilized on strept- avidin-coated chips via their biotin moiety. The selected peptide was thus correctly exposed to the circulating analytes. The signal correponding to the immobilization reaches 150–180 response units for each of the seven peptides. Binding of the rat anti-mouse IgG that was coated in the selection procedure was assayed as the background control. Kinetic parameters and K d values were determined from sensorgrams and are reported in Table 4. IgGs being dimeric proteins, the calculations were extrapolated from a bivalent analyte model. Among the seven assayed peptides, three of them (Pep 91, Pep 93, Pep 95) do not significantly bind to IgG 9G4H9 under SPR experimental conditions. Among the four other peptides, three (Pep 92, Pep 94 and Pep 96) displayed a weak binding ability not easily meas- urable in SPR conditions, and one of them (Pep 90) revealed a high affinity for the Ig, with a K d value of 7.4 · 10 )8 M (Fig. 4). As shown on the figure, the experimental data fit the theoretical plot. We verified the absence of mass transfer interference by assaying several analysis conditions, notably by varying the flow rate of the circulating analyte. Inhibition of the 9G4H9 b-lactamase activity Pep 90 was tested for its inhibition properties of 9G4H9 catalytic activity. Kinetics were carried out using peptide concentrations from 0 to 180 l M . Pep 95 was assayed as a negative control because it presents the same biochemical properties as Pep 90, notably its pI, but has a low affinity for 9G4H9. The inhibitory capabilities of Pep 90 were investigated by measuring the residual b-lactamase activity of the IgG 9G4H9 after 20 min of incubation with various amounts of peptide (Fig. 5). A half inhibitory concentration (IC 50 ) can be calculated from this inhibition plot. For 9G4H9, the IC 50 ¼ 9.10 )5 M . No inhibition effect could be measured with Pep 95. This indicates that the inhibitory effect is undoubtedly due to the selected sequence and not to biotin or phage pIII moieties. Strynadka et al. have previously reported the character- ization of a 165-residue b-lactamase-inhibitor protein, BLIP (K i ¼ 0.3 n M ) [13,14]. We tested the inhibition of the b-lactamase activity of the anti-idiotypic Ig using a bacteriophage displaying BLIP on its surface, engineered and provided by T. Palzkill, Dept of Molecular Virology and Microbiology, Baylor College of Medicine, Houston, TX, USA [15]. BLIP displayed on phage were found to inhibit the IgG with an IC 50 around 10 )5 M . We then carried out a competitive ELISA assay, derived from the previously described phage ELISA, where IgG 9G4H9 was coated on Table 4. Kinetic values and dissociation constants for seven selected peptides. Measurements were performed by SPR analysis. BT, below threshold; ND, not determined. Peptide k on ( M )1 Æs )1 ) k off (s )1 ) K d ( M ) Pep 90 3.1 · 10 4 ± 4.8 · 10 2 2.3 · 10 )3 ±10 )4 7.4 · 10 )8 Pep 91 BT BT ND Pep 92 6.9 · 10 3 ± 6.8 · 10 1 4.1 · 10 )2 ± 4.8 · 10 )3 5.9 · 10 )6 Pep 93 BT BT ND Pep 94 4.6 · 10 3 ± 4.5 · 10 2 5.2 · 10 )3 ± 5.5 · 10 )4 1.1 · 10 )6 Pep 95 BT BT ND Pep 96 4.8 · 10 3 ± 7.1 · 10 1 1.2 · 10 )3 ± 5.1 · 10 )5 2.4 · 10 )7 Fig. 5. Inhibition of IgG 9G4H9 b-lactamase activity by Pep 90. 9G4H9 (3.2 l M ) was incubated with various concentrations of Pep 90 (0–180 l M ) for 20 min at room temperature. The activity was revealed with 1.8 m M ampicillin in phosphate buffer 0.1 M pH 7.4 (grey bar). Pep 95 (open bar) was used as a control from 0–100 l M . Fig. 4. Sensorgrams corresponding to SPR analysis. Sensograms of SPR analysis of the binding of Pep 90 (40 lgÆmL )1 in HBS) immo- bilized on SA chips, to various concentrations of IgG 9G4H9 from 156 n M to 468 n M , in HBS buffer, circulating at 7 lLÆmin )1 .The experimental data fit the theoretical plot. Ó FEBS 2003 Peptides inhibiting a catalytic antibody (Eur. J. Biochem. 270) 2793 plates and incubated with BLIP on phage. The binding of BLIP was measured by an anti-M13 peroxidase-conjugated Ig, that gives a chromogenic signal in the presence of its substrate. We added various amounts of free Pep 90, from 0 to 100 lgÆmL )1 . The residual binding of BLIP was then estimated (Fig. 6). Pep 90 was found to expel BLIP from 9G4H9 active site. This clearly indicates that BLIP and Pep 90 share the same interaction site. Discussion The selection of inhibitory compounds is a real challenge because biocatalysts are generally elaborate macromole- cules. We have developed a procedure to obtain peptides that on the one hand display a tight affinity for the biocatalyst and, on the other hand, valuably inhibit a b-lactamase-like activity shown by an anti-idiotypic cata- lytic Ig. By this approach, we reduced the diversity of the original library from more than one billion peptides to only four, with one displaying the required function. Our method isallthemoreefficientinthatthetarget,anIgG,isa macromolecule of high molecular mass (approximately 150 kDa) thus displaying many potential interacting sites. We have previously suggested that only a few residues (two serines, one lysine and one glutamic acid) were involved in the catalytic activity of the Ig; they are located on the light chain and gathered together [12,16]. It was previously demonstrated that cyclic peptides displayed on phage could favourably lead to ligands presenting better affinities than linear ones [17]. However, the main restriction in the development of procedures based on phage display technology to obtain worthwhile com- pounds is that the selected molecules are interesting only if they retain their activity in their soluble form. Some authors indicate that cyclic peptides are more likely to retain their function once in free form than their linear counterparts [18]. In our work, four peptides out of seven have retained binding features once in soluble form. The sequence of the inhibitory peptide Pep 90 Tyr-His- Phe-Leu-Trp-Gly-Pro is mainly composed of central hydro- phobic residues and contains a Pro in the C-terminal, similar to most of the peptides selected against IgG 9G4H9. This result allows the definition of a basic feature of structures able to bind to this IgG active site. BLIP was described to be an efficient inhibitory protein of b-lactamase [13] and it was found to also inhibit IgG 9G4H9 with a weaker IC 50 . The residues of BLIP that are involved in the inhibition function of b-lactamase were identified [19,20] as Ala-Ala-Gly-Asp-Tyr-Tyr, which form a hairpin. The primary structures of this part of BLIP and Pep 90 are different, and their tri-dimensional schemes cannot be overlaid. Consequently, the selection procedures that we have developed have led us to find a completely new solution for the inhibition of a catalyst. The ability of BLIP to inhibit IgG 9G4H9 is a crucial point because it proves that the active site IgG 9G4H9 is the real internal image of b-lactamase. The b-lactamase-like catalytic information was then successfully transferred from the enzyme to the abzyme throughout the idiotypic pathway. The catalytic features of a biocatalyst, such as activity and inhibition by elaborated molecules, can thus be preserved by the immune system within the idiotypic network. This opens routes towards the idea that auto- immune diseases involving catalytic antibodies [21,22] might have appeared because of the presence of abnormal amounts of nonimmunogen molecules stimulating the apparition of catalytic Igs. The procedure developed is efficient for the selection of inhibitory peptides even in the absence of any information about the structure of the target. It is much less time- consuming and less expensive than rationally designed synthetic compounds. Indeed, b-lactamase is widespread throughout the world and is considered to be a plague for the treatment of infections by antibiotics [23]. Designing new antibiotic molecules is a flourishing activity of medical research nowadays. The in vitro selection of peptides able to inhibit b-lactamase activity appears, consequently, as a good alternative route to the rational design of small molecules based on the knowledge of the targeted catalytic mechanisms. Acknowledgements We are grateful to Drs Barbee and Koivunen for kindly providing the peptide library and Dr Palzkill for giving its BLIP construction for the expression on phage. We gratefully acknowledge the Picardie region and ALTERNATECH for supporting A.S. Yribarren. References 1. Azzazy, H.M.E. & Highsmith, W.E. Jr (2002) Phage display technology: clinical applications and recent innovations. Clin. Biochem. 35, 425–445. 2. Smith, G.P. (1985) Filamentous fusion phage: novel expression vectors that displayed cloned antigens on the virion surface. Science 228, 1315–1317. 3. Parmley, S.F. & Smith, G.P. (1988) Antibody-selectable fila- mentous fd phage vectors: affinity purification of target genes. Gene 73, 305–318. 4. Devlin, J.J., Panganiban, L.C. & Devlin, P.E. (1990) Random peptide libraries: a source of specific binding molecules. Science 249, 404–406. 5. Avalle, B., Friboulet, A. & Thomas, D. (2000) Catalysis by anti- idiotypic antibodies. Chem. Immunol. 77, 80–88. 6. Kolesnikov, A.V., Kozyr, A.V., Alexandrova, E.S., Koralewski, F., Demin, A.V., Titov, M.I., Avalle, B., Tramontano, A., Paul, S., Fig. 6. Competitive ELISA for the binding to IgG 9G4H9. Phage BLIP (10 11 TU) were incubated on 1 lg coated IgG 9G4H9. After 1 h incubation at 37 °C, various amounts of Pep 90 (0–100 l M )were added to wells. Absorbance values are related to the residual binding of BLIP to IgG 9G4H9. The CX 7 C library and M13 were assayed to evaluate the background binding signal. The standard deviations were calculated from two separate experiments. 2794 A S. Yribarren et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Thomas, D., Gabibov, A.G. & Friboulet, A. (2000) Enzyme mimicry by the anti-idiotypic antibody approach. Proc.NatlAcad. Sci. USA 97, 13526–13531. 7. Avalle, B., Thomas, D. & Friboulet, A. (1998) Functional mimicry: elicitation of a monoclonal anti-idiotypic antibody hydrolysing b-lactams. FASEB J. 11, 1055–1060. 8. Pillet, D., Paon, M., Vorobiev, I.I., Gabibov, A.G., Thomas, D. & Friboulet, A. (2002) Idiotypic network mimicry and antibody catalysis: lessons for the elicitation of efficient anti-idiotypic pro- tease antibodies. J. Immunol. Methods 269, 5–12. 9. Jerne, N.K. (1974) Toward a network theory of the immune system. Ann. Immunol. 125c, 373–389. 10. Avalle, B., Mistro, D., Thomas, D. & Friboulet, A. (1996) Poly- clonal catalytic anti-idiotypic antibodies with b-lactamase activity. Ann. NY Acad. Sci. 799, 172–175. 11. Koivunen, E., Wang, B. & Ruoslahti, E. (1994) Isolation of a highly specific ligand for the alpha-5, beta-1 integrin from a phage display library. J. Cell. Biol. 124, 373–380. 12. Lefe ` vre, S., De ´ bat, H., Thomas, D., Friboulet, A. & Avalle, B. (2001) A suicide substrate mechanism for hydrolysis of b-lactams by an anti-idiotypic catalytic antibody. FEBS Lett. 489, 25–28. 13. Strynadka, N.C., Jensen, S.E., Johns, K., Blanchard, H., Page, M., Matagne, A., Frere, J.M. & James, M.N. (1994) Structural and kinetic characterization of a beta-lactamase-inhibitor protein. Nature 368, 657–660. 14. Strynadka, N.C., Jensen, S.E., Alzari, P.M. & James, M.N. (1996) A potent new mode of beta-lactamase inhibition revealed by the 1.7 A X-ray crystallographic structure of the TEM-1-BLIP com- plex. Nat. Struct. Biol. 3, 290–297. 15. Huang, W., Zhang, Z. & Palzkill, T. (2000) Design of potent beta- lactamase inhibitors by phage display of beta-lactamase inhibitory protein. J. Biol. Chem. 275, 14964–14968. 16. De ´ bat, H., Avalle, B., Friboulet, A. & Thomas, D. (2000) Structure-function studies of a new generation of catalytic protein: an abzyme with a b-lactamase activity. Int. J. Biochromat. 5, 91–96. 17. Giebel, L.B., Cas, R.T., Milligan, D.L., Young, D.C., Arze, R. & Johnson, C.R. (1995) Screening of cyclic peptide phage libraries identifies ligands that bind streptavidin with high affinities. Biochemistry 34, 15430–15435. 18. McLafferty, M.A., Kent, R.B., Ladner, R.C. & Markland, W. (1993) M13 bacteriophage displaying disulfide constrained microproteins. Gene 128, 29–36. 19. Rudgers, G.W., Huang, W. & Palzkill, T. (2001) Binding properties of a peptide derived from beta-lactamase inhibitory protein. Antimicrob. Agents Chemother. 45, 3279–3286. 20. Rudgers, G.W. & Palzkill, T. (2001) Protein minimization by random fragmentation and selection. Protein Eng. 14, 487–492. 21. De ´ bat, H., Avalle, B., Chose, O., Sarde, C.O., Friboulet, A. & Thomas, D. (2001) Overpassing an aberrant Vk gene to sequence an anti-idiotypic abzyme with b-lactamase like activity that could have a possible linkage with autoimmune disease. FASEB J. 15, 815–822. 22. Friboulet, A., Avalle, B., De ´ bat, H. & Thomas, D. (1999) A possible role of catalytic antibodies in metabolism. Immunol. Today 20, 474–475. 23. Helfand, M.S. & Bonomo, R.A. (2003) Beta-lactamases: a survey of protein diversity. Curr. Drug. Targets. Infect. Disord. 3, 9–23. Ó FEBS 2003 Peptides inhibiting a catalytic antibody (Eur. J. Biochem. 270) 2795 . forward primer AB 348, 5¢-TTAGCAAAACCTC ATACAGAA-3¢, and the backward primer AB 349, 5¢-GATGCTGTCTTTCGCTGCTGAG-3¢,wereusedfor DNA amplification. The M13 forward primer, 5¢-ATTCACCTCGAAAGCAAGCTG-3¢,wasusedfor sequencing. Table. Designing new antibiotic molecules is a flourishing activity of medical research nowadays. The in vitro selection of peptides able to inhibit b-lactamase activity appears, consequently, as a good alternative. Selection of peptides inhibiting a b-lactamase-like activity Anne-Sophie Yribarren, Daniel Thomas, Alain Friboulet and Be ´ range ` re Avalle Ge ´ nie enzymatique et cellulaire, UMR