Protein Engineering vol.10 no.11 pp.

13111318, 1997

Novel selection method for engineered antibodies using the mechanism of Fv fragment stabilization in the presence of antigen

Kouhei Tsumoto1, Yoshiyuki Nishimiya1, Nobuhiro Kasai1, Hiroshi Ueda2, Teruyuki Nagamune2, Kyoko Ogasahara3, Katsuhide Yutani3, Kenji Tokuhisa4, Masaaki Matsushima4 and Izumi Kumagai1,5
1Department

of Biochemistry and Engineering, Graduate School of Engineering, Tohoku University, Sendai 980-77, 2Department of Chemistry and Biotechnology, Graduate School of Engineering, University of Tokyo, Bunkyo-ku 113, 3Institute for Protein Research, Osaka University, Suita, Osaka 565, and 4Rational Drug Design Laboratories, Matsukawa, Fukushima 960-12, Japan whom correspondence should be addressed

5To

Although the heavy and light chain domains of some antibody variable region fragments (Fvs) readily dissociate under physiological conditions, the Fvs are stable in the presence of antigen. This antigen-driven Fv stabilization mechanism was applied to the selection of clones with specicity toward target antigens. The results can be summarized as follows. (i) Some of the residues in the heavy chain complementarity determining region 2 (HCDR2) of anti-hen egg white lysozyme (HEL) monoclonal antibody HyHEL10 heavy chain variable region (VH) were randomized. (ii) The randomized VH fragments of HyHEL10 were displayed on a lamentous bacteriophage and mixed with the target antigen, before being applied to a light chain variable region (VL) which was immobilized on microtiter plates and subjected to selection by panning. (iii) After four rounds of panning, four clones that showed signicant binding to human lysozyme (hL), which HyHEL10 recognized poorly, were selected from the HCDR2 library. (iv) The soluble Fv fragments selected were expressed in Escherichia coli, puried, and subjected to an inhibition assay of lysozyme enzymatic activities and an isothermal titration calorimetry. These Fv fragments had increased afnity toward hL, and thermodynamic analysis suggested that the reduced entropy loss due to binding by the replacement of residues in HCDR2 resulted in the higher hL binding activity. Keywords: antibody engineering/phage display/complementarity determining regions/Fv fragment/isothermal titration calorimetry

Introduction One of the major goals of antibody engineering is the development of a procedure to graft specic function into an antibody and to enhance the afnity toward the target antigen (Winter and Milstein, 1991). Antibodies may be regarded as a protein engineering system perfected by nature for the generation of a virtually unlimited repertoire of complementary molecular surfaces (Mariuzza and Poljak, 1993), and the structural variation of the antigen binding site is mainly conned to the complementarity determining regions (CDRs). The antibody binding site is formed by a small number of protein segments of variable structure (the CDRs) grafted onto a scaffolding of
Oxford University Press

essentially invariant architecture (framework regions) (Padlan, 1994), and this characteristic has been critical to the success of articial functional conversion of antibodies (Verhoeyen et al., 1988). The display of antibody fragments on the surface of lamentous bacteriophages by fusing them to a minor coat protein of the phage (pIII) (Smith, 1985; Scott and Smith 1990), and subsequent selection of the phage with antigen (by panning) has provided a powerful means of making antibodies of predened binding specicity from V gene repertoires (Winter et al., 1994). Many approaches have been tried for obtaining a novel antibody (Marks et al., 1991; Kruif et al., 1995) or a functionally improved antibody (Hawkins et al., 1992; Riechman and Weill, 1993; Deng et al., 1995; Yang et al., 1995). In addition, the chain shufing method (Kang et al., 1991; Marks et al., 1992) has been proposed; a new library is constructed and is subjected to selection by xing one variable chain obtained from the naive library and shufing the other chain from V genes repertoires. Most Fv fragments of immunoglobulins readily dissociate under physiological conditions. To overcome the instability of Fv fragments, linking domains with peptide linkers (single chain Fv, scFv) (Glockshuber et al., 1990) or disulde bond (disulde stabilized Fv, dsFv) (Brinkmann et al., 1993) has been attempted; however, Fv fragments are usually stable in the presence of their antigens (Glockshuber et al., 1990, Ueda et al., 1996). On the basis of the antigen driven stabilization of Fv, we have recently proposed the novel open sandwich method (Ueda et al., 1996). This method is an alternative sandwich method for determining antigen concentration. Its sensitivity is relatively high, and the background is signicantly lower than for the usual sandwich method. Thus, open sandwich is expected to be an alternative and powerful method for the functional conversion and improvement of antibodies by chain shufing. We have focused on the interaction between hen egg white lysozyme (HEL) and its monoclonal antibody, HyHEL10. The interaction is structurally well studied (Padlan et al., 1989) and therefore is most suitable for the precise analysis of antigenantibody interactions. Previously we reported a thermodynamical analysis that combined bacterial expression system and site-directed mutagenesis (Tsumoto et al., 1994a,b, 1995, 1996). For convenient analysis of antigenantibody interactions, a novel phage display system has been developed (Maenaka et al., 1996), and using this system, the HyHEL10 single chain variable region (scFv) fragment and the variable region of the heavy chain (VH) can be stably displayed on a lamentous phage. Although HyHEL10 recognizes several avian lysozymes (Smith-Gill et al., 1984a,b), the Fv fragment poorly recognizes human lysozyme (hL). As shown in Figure 1, the most signicant difference in amino acid residues between the HEL epitope region recognized by HyHEL10 and the corresponding one of hL lies in the loop at site 98102 of HEL, and the local 1311

K.Tsumoto et al.

Fig. 1. Comparison of the hL and HEL local structures recognized by HyHEL10. (A) Only residues recognized by HyHEL10 are shown. Numbering of amino acid residues is based on that of HEL. Since one residue is inserted between the 46th and 47th positions in hL, the correct numbering of Trp63 to Pro102 in hL is Trp64 to Pro103. L and H in the bottom column represent the sites recognized by the light and heavy chains respectively. (B) X-Ray crystallographic analysis of the HyHEL10 Fv fragment-HEL complex (Padlan et al., 1989). The drawing is from the Protein Data Bank (3HFM). Red lines and zones represent the main chain and side chain of HEL, and blue lines and zones those of HyHEL10 Fv. HEWL appeared in the gure represents HEL. (C) HEL of (B) is replaced with hL, and the main chain and side chain of hL are shown by white lines and zones. HLZ appeared in the gure represents hL.

1312

Novel selection method for antibody engineering

antigenic structure has been found to be rigid in comparison of hL with HEL (McKenzie and White, 1991; Song et al., 1994). Preliminary modeling indicated that Pro103 of hL interfered with the close tting of HyHEL10, especially to HTyr53, Ser54, Ser56 and Tyr58 in the CDR2 of the VH (HCDR2). To graft specicity toward hL into the Fv fragment of HyHEL10, these residues in HCDR2 were randomized, displayed on a lamentous bacteriophage and subjected to the selection of desired clones. In this paper, we report the novel method for engineering of an antibody using the mechanism of Fv stabilization in the presence of antigen. Increased afnity toward hL into the antiHEL monoclonal antibody, HyHEL10, the Fv fragment could be introduced by the method described here. Materials and methods Materials All enzymes for genetic engineering were purchased from Takara shuzo, Toyobo or Boehringer. Biotin-NHS was from Boehringer. Streptavidin was purchased from Sigma. HEL was obtained from Seikagaku-kogyo (Tokyo). Horseradish peroxidase (HRP)-conjugated anti-M13 mouse antibody was from Pharmacia Biotec Inc. (Tokyo). 2,2 -azinobis(3ethlbenzthiazoline-sulforic acid (ABTS) was from Wako Junyaku Inc. (Osaka). The variable region of the light chain (VL) fragment of HyHEL10 was prepared and biotinylated as described previously (Ueda et al., 1996). DNA oligonucleotides for randomization of HyHEL10 VH and polymerase chain reaction (PCR) were synthesized using a 380A DNA synthesizer (Perkin Elmer). All other reagents were of biochemical research grade. Construction of HyHEL10 randomized VH fragments The HCDR2 of HyHEL10 was randomized by PCR. The plasmid, pTZPsFv2, which is the modied phagemid vector of pLFv constructed for the phage display of HyHEL10 scFv (Maenaka et al., 1996), was used as a template. Prior to use for PCR, the back primer HCDR2RANBACK (5 -CGCCTCGAGTACATGGGCTACGTCAGCNNKNNKGGCNNKACCTHCTACAACCCC-3 ) and forward primer HyHELFOR (5 ACCCGCGGAGACGGTGACGAGGGTGCC-3 ) were biotinylated using a 5 -biotinylation kit (Amersham), and mixed to a nal concentration of 25 pmol/ml (N, K and H represent the mixtures of AGCT, GT and ACT respectively). The PCR reaction was carried out in 100 l of 250 nM dNTP, with 25 pmol of each primer, 10 ng pTZPsFv2 as a template and 0.5 U Taq DNA polymerase (Perkin Elmer) in the manufacturers recommended buffer. The PCR reaction was cycled 35 times (94C for 1 min, 55C for 1.5 min and 72C for 2 min) using a thermal cycler MP3000 (Takara). The PCR amplied product was subjected to agarose gel electrophoresis, recovered from the gel with Geneclean II (Bio101), and digested with XhoI and SacII. The randomized VH fragment was puried using Streptavidin Beads (Promega), followed by ligation with the vector, pTZPsVH2 at the XhoI/SacII site using T4 DNA ligase (Boehringer), and was designated pTZPsVH2-ran. The Escherichia coli strain JM109 was transformed with pTZPsVH2-ran, infected with helper phage, M13KO7 (Promega), and incubated overnight at 37C. Phage VH was recovered by PEG precipitation (Maenaka et al., 1996), and dissolved in 500 l phosphate-buffered saline (PBS) per 20 ml culture.

Panning One hundred microliters of 10 g/ml streptavidin (Sigma) was added to each well of a microtiter plate (Falcon) and incubated overnight at 4C. After washing the wells with PBS containing 0.05% Tween-20 (PBST), 100 l of 10 g/ml biotinylated VL was added to the wells, incubated for 1 h at 37C. After removing the solution from the wells, phage VH premixed with soluble hL at room temperature for 30 min was applied to the VL-immobilized microtiter plate wells and incubated at room temperature for 1 h. The plate was washed six times with PBST, and eluted with 100 mM Glycine (pH 2.0). Eluate was rapidly neutralized with 100 mM TrisHCl (pH 8.5) containing 500 mM NaCl, and reinfected into early log-phase E.coli JM109. After 1 h, 1000 l of the culture was plated and incubated at 37C overnight. Colonies were transferred into an LB culture containing 100 mg/l ampicillin, infected with helper phage, M13KO7, and then incubated overnight at 37C. The phage was prepared by centrifugation and PEG precipitation and was subjected to the next selection stage. After four stages, clones on plates were cultured and phage VH was prepared according to the method described previously (Maenaka et al., 1996). The number of clones present on the plates were counted at each stage. Detection of binding activity of phage VH by enzyme-linked immunosorbent assay (ELISA) ELISA was performed in principle according to the manual of RPAS detection module (Pharmacia). One hundred microliters of 10 g/ml HyHEL10 VL fragment in PBS was applied to each well of the microtiter plates and incubated for 1 h at room temperature. After removing the solution, 200 l SuperBlock (Piearce) was added to each well and incubated for 1 h at room temperature. After discarding the buffer, 100 l of VH phage, which had been mixed with hL or HEL and prediluted with 1 volume of binding buffer 30 min before, was added to each well and incubated for 1 h at room temperature. After washing twice with PBST, 100 l of 5000 times diluted horseradish peroxidase (HRP)-anti M13 (Pharmacia) was added and incubated for 1 h. After three washes with PBST, 200 l of 50 mM sodium succinate buffer containing 10 mg/ml ABTS and 0.01% H2O2 was added and incubated for 1 h. Absorbance was measured at 410 nm using a microplate reader (Bio-Rad, type 550), using 630 nm as the control. DNA sequencing Di-deoxy reactions for DNA sequencing were performed using an Auto read Sequencing Kit (Pharmacia) according to the recommendations of the manufacturer. Analysis of DNA sequences was performed using an ALF express auto-read sequencer (Pharmacia). Expression of soluble proteins For more convenient preparation of the selected clones, a His 6 peptide tag was added to the C-terminus of the VH chain, and expressed as a scFv fragment (Tsumoto et al., 1994a). The transformant E.coli strain BL21(DE3) harboring a plasmid clone was grown at 28C in 2 YT (Sambrook et al., 1989) supplemented with 200 g/ml ampicillin, until the early stationary phase was reached. To induce the expression of soluble scFv, isopropyl -D-thiogalactopyranoside (IPTG) was added to 1 mM and the culture was grown overnight at 28C. From 200 ml of culture, two fractions (bacterial supernatant and periplasmic fractions) were separated as follows (Tsumoto 1313

K.Tsumoto et al.

Fig. 2. Schematic representation of open sandwich selection method. Detail procedures are described in Materials and methods.

Table 1. Binding of phage antibodies using open sandwich selection methoda Backgroundc Rounds of panning 1 2 3 4
aExperimental bThe cThe

Against hLd Number of clones recoveredb 1.2 4.8 5.6 2.4 106 104 103 103 Input number 1.2 6.4 2.4 5.5 1010 109 109 109 number of clones recovered 1.2 6.4 2.4 2.2 106 104 105 106

Input numberb 1.2 4.8 3.6 2.4 1010 109 109 109

procedures are described in Materials and methods. number of phage is shown colony forming unit (c.f.u.). The procedure for estimating values is described in Materials and methods. number of c.f.u. by panning without hL are shown. dThe number of c.f.u. by panning with hL (10 g/mL) are shown.

et al., 1994b). After the removal of the culture supernatant by centrifugation, the cell pellet was resuspended in 10 ml of 20 mM TrisHCl (pH 7.5), 0.5 M sucrose, 0.1 mM EDTA and incubated for 5 min at room temperature. Then, 40 ml water was added to give an osmotic shock and the cells were left on ice for 30 min. The cells were spun down and the supernatant was saved as the periplasmic fraction. SDSPAGE To examine the secretory expression level, each fraction mentioned in the previous section was taken for protein analysis. The total protein in the supernatant was precipitated with 6% trichloroacetic acid (TCA) and 0.083% sodium 1314

deoxycholate (DOC), and subjected to protein analysis by SDSPAGE in the buffer system described by Laemmli (1970). Purication of soluble proteins Each fraction prepared was salted out with ammonium sulfate at 80% saturation, and the precipitates were collected by centrifugation at 15 000 r.p.m. for 10 min. The protein precipitates were dissolved in 50 mM phosphate buffer (pH 7.2) containing 0.2 M NaCl, and dialysed against the same buffer for two days. The precipitates formed during dialysis were removed by centrifugation at 10 000 r.p.m. for 15 min. The supernatant was further concentrated by ultraltration, then loaded onto a Ni-NTA column (Qiagen), previously equilib-

Novel selection method for antibody engineering

Table 2. Amino acid sequences of selected clones deduced from DNA sequencing(1)

Table 4. Binding activities of HyHEL10 mutant Fv fragment toward HEL or hLa Mutants Wild-type 23 26 Wild-type 23 26 Lysozyme HEL HEL HEL hL hL hL Inhibition activity (mol/mol)b 0.8 2.0 100 500 100 300 Relative to wild typec 1.0 0.4 0.006 1.0 5.0 1.7

1The

aExperimental conditions bThe values of the molar

phagemid of the selected clone was sequenced and only the deduced amino acid sequence is shown. The shaded boxes indicate mutated residues in HCDR2. WT, wild -type HyHEL10; VH, variable region of the heavy chain. 2After four rounds of panning, 30 clones on plates were randomly picked up and sequenced. Numbers of clones having the same sequences are shown as Frequences.

are given in the Materials and methods. ratio of Fv fragment to HEL which showed the 50% reduced enzymatic activity were given. cThe molar ratio relative to the wild type is shown. HEL, hen egg-white lysozyme; hL, human lysozyme; N.D., not determined.

Table 3. Binding of selected clones to lysozymes Clone number ELISAa HEL 23 26 2 18 Wild Type
aAbsorbance

hL 0.15 0.05 0.06 0.05 0.25 0.30 0.20 0.15 0.10 0.05 0.10 0.05 0.06 0.05 0.05

0.75 0.10 0.15 0.15 1.50

at 410 nm obtained in ELISA are shown. ELISA procedures are described in the Material and methods section. Uncertainties are the average of at least three experiments S.D.

rated with the same buffer. The column was washed with the same buffer. The adsorbed protein was eluted with the same buffer containing 1~1000 mM imidazole. The eluate was rapidly subjected to dialysis against PBS. Measurement of inhibitory activity toward lysozymes Inhibition of lysozyme enzymatic activity was measured as described previously (Ueda et al., 1993). The Fv fragment at various concentrations was mixed with 1.5 M lysozyme and incubated at 25C for 1 h in 30 l of 50 mM phosphate buffer (pH 7.2). Each mixture was then added to 970 l of 50 mM Na2HPO4 (pH 6.2 adjusted with NaOH) containing 340 g of Micrococcus lysodeikticus cells. The initial rate of the decrease in A540nm was monitored at 25C. Isothermal titration calorimetry Thermodynamic parameters of the interactions between lysozymes and the mutant Fv fragments were determined by microtitration calorimetry using an OMEGA titration microcalorimeter (Wiseman et al., 1989) from MicroCal Inc. (Northampton, MA). The experimental conditions were the same as those reported previously (Tsumoto et al., 1995); the lysozyme at a concentration of 3 M in 50 mM phosphate buffer (pH 7.2) containing 200 mM NaCl in a calorimeter cell was titrated with a 75 M solution of the mutant Fv fragments in the same buffer at 25C. The ligand solution was injected 16 times in portions of 7 l during 15 s. Thermogram data were analysed using a computer program (Origin) supplied by MicroCal Inc. (Wiseman et al., 1989). Estimation of protein concentration The concentration of HEL was estimated using A280 of 26.5 (Imoto et al., 1972) and that of hL or mutant HyHEL10 Fv fragments was estimated using quantitative amino acid analysis. Results and discussion Construction of VH-CDR2 libraries and panning As mentioned above, preliminary modeling indicated that Pro103 of hL interfered with the close binding of HyHEL10, especially to HTyr53, Ser54, Ser56 and Tyr58 in HCDR2. Then, we chose the randomized site (53, 54, 56 and 58), and constructed VH-CDR2 libraries which consisted of only 2.4 104 (20 20 20 3) kinds of mutant. Mutation was introduced by a general PCR method (Figure 2), and the amplied fragments were ligated into the phagemid vector. 1.2 105 identical transformants were obtained, indicating that full libraries for randomization of selected sites could be 1315

Fig. 3. SDSPAGE analysis of expressed and puried scFv clone selected. The case of purication of clone 23 was shown. Samples for PAGE were prepared as described in the Materials and methods and then followed by 12% SDSPAGE and stained with Coomassie Brilliant Blue C-250. Lane 1, molecular size marker (from top to bottom: 200, 116, 97, 66, 45, 30, 21 and 14 kDa, respectively); lane 2, wild-type HyHEL10 scFv; lane 3, total proteins in culture supernatant; lane 4, pass through fraction; lane 5, fraction of wash with PBS; lanes 611, eluted with PBS containing imidazole of 1, 5, 10, 50, 100 and 1000 mM, respectively.

K.Tsumoto et al.

Table 5. Thermodynamic parameters of the interactions between HyHEL10 mutants and HEL or hLa Mutants Wild-type 23 26 Wild-type 23 26
aExperimental

Lysozyme HEL HEL HEL hL hL hL

Ka 106M1 250 15.5 1.22 0.83 9.36 2.39

G 48.3 42.0 34.9 34.0 39.9 36.5

G kJ mol1 0 6.3 13.4 0 5.9 2.5

H 80.6 78.5 31.2 58.0 56.7 30.2

H 0 2.1 49.4 0 1.3 28.3

TS 32.3 36.5 3.7 24.0 16.8 6.3

TS 0 4.2 36.0 0 7.2 30.3

conditions are given Material and methods. Abbreviations used: n, stoichiometry, Ka, binding constant, G, H and TS are change in Gibbs energy, binding enthalpy and entropy, respectively. G, H and TS are the differences in each of the values from those of wild-type Fv; HEL, hen egg-white lysozyme; hL, human lysozyme.

constructed. Before subjection to panning, several clones on plates were sequenced and it was conrmed that certain clones are not biased. Here the VH-CDR2 libraries were subjected to the novel selection method (Figure 2). As described above, Fv fragments are usually stable in the presence of their antigens (Glockshuber et al., 1990; Ueda et al., 1996). We applied this mechanism to the selection of desired clones. Phage VH was mixed with soluble hL, applied to the HyHEL10 VL-immobilized microtiter plate and incubated at room temperature. The plate was extensively washed, eluted and reinfected into early logphase E.coli cells. The phage prepared was subjected to the next selection stage. After four rounds of panning, several clones seemed to be enriched (Table 1). Colonies on the culture plates were randomly picked, followed by preparation of phagemid from the cells. As discussed below, the deletion of genes encoding antibody fragments of randomly mutated CDRs has often been observed in phage-displaying scFv fragments (Tsumoto,K. and Nishimiya,Y., unpublished results); however, few deletions of randomized genes were observed in phages displaying only VH fragments (data not shown). Thirty clones were randomly picked up, and the DNA sequences of the clones were determined. Twenty-four clones were found to have the same sequence (Table 2, designated as clone 23), and four clones were identical (Table 2, designated as clone 26). Open sandwich ELISA analysis using the phagedisplayed VH and soluble VL was performed (Ueda et al., 1996). As shown in Table 3, these clones had enhanced binding activity toward hL in comparison with wild-type Fv. Expression and purication of mutant soluble proteins To characterize the selected clones, two clones (23 and 26) were chosen for further examination. scFv fragments were used for precise analysis because of their convenient and efcient purication. The rst step of the usual purication procedure for Fv fragments is afnity chromatography (Skerra and Pluckthun, 1988; Ward et al., 1990; Tsumoto et al., 1994b); however, in the case of mutants with a low afnity constant toward the target antigen, only a poor yield of Fv fragments is obtained from afnity chromatography (Tsumoto et al., 1995). Metal chelating chromatography would overcome the difculty in this purication step. The genes encoding these mutants were inserted into the expression vector, pKTN-NSX, which was the modied vector of high expression for the HyHEL10 Fv fragment, pKTN2 (Tsumoto et al., 1994b). The vector was constructed for expression of the gene by fusing six histidines to its Cterminus. Soluble scFv fragments of these clones were highly 1316

expressed in E.coli, which was almost the same as the case of wild-type scFv, and was puried with only one step of metalchelating chromatography (Figure 3). Inhibition assay of lysozyme enzymatic activity To investigate the binding of selected clones to HEL and hL, inhibition of the enzymatic activity of HEL and hL by these clones was tested. As described previously (Ueda et al., 1993), the HyHEL10 Fv fragment completely inhibited the activity of its antigen, HEL, in a 1:1 molar ratio. The inhibitory activities of the selected clones toward both HEL and hL were measured (Table 4). The inhibitory activity of clone 23 scFv toward HEL was only slightly reduced, and that of clone 26 scFv was signicantly reduced. On the other hand, the inhibition of clone 23 scFv toward hL was signicantly increased in comparison with wild-type Fv, indicating that clone 23 scFv has enhanced afnity toward hL. Clone 26 scFv fragment was also found to inhibit hL enzymatic activity (Table 4). Thermodynamic study of the interaction between selected mutants and lysozymes The interactions between mutant HyHEL10 scFv fragments and lysozymes were investigated by titration calorimetry. The enthalpy change (H) and binding constant (K) upon antigen antibody interaction were directly obtainable from the experimental titration curve (Wiseman et al., 1989; Tsumoto et al., 1996). The Gibbs energy change (G RT lnK) of binding was calculated from the binding constant, and the entropy change (TS G H ) upon the association could also be estimated. The thermodynamic parameters of the interactions between selected clones and HEL or hL are shown in Table 5. Although the negative enthalpy (H) of the interaction between clone 23 scFv and hL was almost the same as that of wild-type Fv, Gibbs energy (G) for the interaction between scFv and hL was increased, indicating that the negative entropy change (TS) was decreased by the mutation. While the negative value of the enthalpy change of the interaction between clone 26 scFv and hL was 28 kJ mol1 less than for the wild-type. The Gibbs energy was calculated to be 37 kJ mol1, indicating that the entropy change made a positive contribution to the interaction. The negative enthalpy of the interaction between clone 23 scFv and HEL was observed to be slightly reduced in comparison with the case of wild-type scFv. Gibbs energy was also slightly reduced, indicating that the negative entropy of the interaction between the mutant and HEL was increased by mutation. On the other hand, the entropy change made a

Novel selection method for antibody engineering

positive contribution to the interaction between clone 26 scFv and HEL. These results clearly indicate that the negative entropy changes of the interactions between selected clones and hL are decreased (i.e. entropic loss is reduced), resulting in enhanced afnity toward hL. This may originate mainly from two factors; solvent displacement (Sigurskjold et al., 1991) and conformational change (Spolar and Record, 1994; Schwarz et al., 1995; Tsumoto et al., 1996). Energetic advantages by reducing negative entropic loss on solvent displacement and/ or conformational change may have an advantage over reducing enthalpy change by decrease in hydrogen bond formation and van der Waals interactions. Structural investigations of the interactions between lysozymes and selected clones are now under progress, and it seems that the hydrophobic effect of the Trp residue and the burial of the electrostatic charge of Arg are important for the enhanced afnity toward hL of the clone 23, and that decreasing conformational changes in HyHEL10 HCDR2 upon binding result in a slight increase in afnity toward hL of clone 26 (Tsumoto and Kumagai, unpublished). Recent report of the crystal structures of hapten-germline Fab and hapten-afnity maturated Fab complexes has proposed that afnity maturation stabilize the conguration with optimal hapten complementarity of the antibody (Wedemayer et al., 1997). Since the selection method reported here can be regarded as an in vitro mimicry of afnity maturation, the mutant Fv fragments selected in our experiment may stabilize the conformation complementary to local antigenic structure, causing decrease in entropic loss. Application of open sandwich selection method for functional conversion of Fv fragments As described above, the HyHEL10 Fv fragment is stabilized in the presence of its antigen (Ueda et al., 1996), indicating that only Fv fragments which have a high afnity toward target antigens can be stabilized if random mutagenesis is introduced into the CDRs of Fv fragments. Here it was shown that stabilization is applicable to the functional alteration of Fv fragments. Although only one chain can be engineered by this protocol, this method is more effective than the usual panning procedures using single-chain Fv fragment from three points. (i) Few deletions of randomized genes were observed in phages displaying only VH fragments. The deletion of genes encoding antibody fragments of randomly mutated CDRs has often been observed in phage-displaying scFv fragments and therefore, displaying only one partner of Fv fragments may overcome the instability of randomized genes during the selection procedure. (ii) Background clones (i.e. non-specic binding clones) were highly reduced from the libraries, and clones which have signicantly enhanced binding activity toward hL were enriched during selection. Thus, the novel method will save laborious and time consuming screening steps in the usual panning procedures. (iii) Since variable region fragments are not combined with linkers, the method reported here can diminish the effect of introducing linkers on afnity (Chaudhary et al., 1990) and expression level in E.coli (Anand et al., 1991; Tsumoto et al., 1994a). Since this method is based on a weak VHVL interaction and a lot of Fv fragments have weak interchain associations, we think that this method will be a powerful tool for functional conversion of other Fv fragments. Open sandwich selection may not be applicable to Fv fragments with strong interchain

associations; however, adjusting the reaction pH and ion strength to lower the interchain association, or further engineering of the framework region seem to be effective (Ueda et al., 1996). Acknowledgements
We thank Dr Katsumi Maenaka (Oxford University) and Dr Yuichiro Nakaoki (University of Tokyo) for constructing the phage display system for the HyHEL10 VH fragment. This work was supported in part by the Original Industrial Technology R&D Promotion Program from the New Energy and Industrial Technology Development Organization (NEDO) of Japan, and by a Grants-in-aid for Scientic Research in Priority Areas (No.05258207), for Scientic Research (No.06454688), and for the Encouragement of Young Scientists (No.08780641) from the Ministry of Education, Science, Sports and Culture of Japan. This work was also supported in part by Takeda Science Foundation.

References
Anand,N.N., Mandel,S., MacKenzie,C.R., Sadowska,J., Sigurskjold,B., Young,N.M., Bundle,D. and Narang,S.A. (1991) J. Biol. Chem., 266, 2187421879. Brinkmann,U., Reiter,Y., Jung,S.H., Lee,B. and Pastan,I. (1993) Proc. Natl Acad. Sci. USA, 90, 75387542. Chaudhary,V.K., Gallo,M.G., Fitzgerald,D.J. and Pastan,I. (1990) Proc. Natl. Acad. Sci. USA, 87, 94919494. Deng,S.J., Mackenzie,C.R., Hirama,T., Brousseau,R., Lowary,T.L., Young,N.M., Bundle,D.R. and Narang,S.A. (1995) Proc. Natl Acad. Sci. USA, 92, 49924996. Glockshuber,R., Malia,M., Pftizinger,I. and Pluckthun,A. (1990) Biochemistry, 29, 13621367. Hawkins,R.E., Russell,S.J. and Winter,G. (1992) J. Mol. Biol., 226, 889896. Imoto,T., Johnson,L.N., North,A.C.T., Phillips,D.C. and Rupley,J.A. (1972) The Enzymes, Vol. 7, 665868. Kang,A.S., Jones,T.M. and Burton,D.R. (1991) Proc. Natl Acad. Sci. USA, 88, 1112011123. Kruif,J., Boel,E. and Logtenberg,T. (1995) J. Mol. Biol., 248, 97105. Laemmli,U.K. (1970) Nature, 227, 680684. Maenaka,K., Furuta,M., Tsumoto,K., Watanabe,K., Ueda,Y. and Kumagai,I. (1996) Biochem. Biophy. Res. Commun., 218, 682687. Marks,J.D., Hoogenboom,H.R., Bonnert,T.P., McCafferty,J., Grifths,A.D. and Winter,G. (1991) J. Mol. Biol., 222, 581597. Marks,J.D., Ouwehand,W.H., Bye,J.M., Finnern,R., Gorick,B.D., Voak,D., Thorpe,S., Hughes-Jones,N.C. and Winter,G. (1992) Bio/Technology, 10, 779783. Mariuzza,R.A. and Poljak,R.J. (1993) Curr. Opin. Immunol., 5, 5055. McKenzie,H.A. and While,F.H. (1991) Adv. Protein Chem., 41, 173315. Padlan,E.A. (1994) Mol. Immunol., 31, 169217. Padlan,E.A., Silverton,E.W., Sheriff,S., Cohen,G.H., Smith-Gill,S.J. and Davies,D.R. (1989) Proc. Natl Acad. Sci. USA, 86, 59385942. Riechmann,L. and Weill,M. (1993) Biochemistry, 32, 88488855. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning: A Laboratory Manual, 2nd Edn. Cold Spring Harbor Laboratory Press. Schwarz,F.P., Tello,D., Goldbaum,F.A., Mariuzza,R.A and Poljak,R.J. (1995) Eur. J. Biochem., 2289, 388394. Sigurskjold,B.W., Altman,E. and Bundle,D.R. (1991) Eur. J. Biochem., 197, 239246. Spolar,R.S. and Record,M.T. Jr (1994) Science, 263, 777782. Scott,J.K. and Smith,G.P. (1990) Science, 249, 386390. Skerra,A. and Pluckthun,A. (1988) Science, 240, 10381041. Smith-Gill,S.J., Lavoie,T.B. and Mainhart,C.R. (1984a) J. Immunol., 133, 384393. Smith-Gill,S.J., Mainhart,C.R., Lavoie,T.B., Rudikoff,S. and Potter,M. (1984b) J. Immunol., 132, 963967. Smith,G.P. (1985) Science, 228, 13151317. Song,H., Inaka,K., Maenaka,K. and Matsusima,M. (1994) J. Mol. Biol., 244, 522540. Tsumoto,K., Nakaoki,Y., Ueda,Y., Ogasahara,K., Yutani,K., Watanabe,K. and Kumagai,I. (1994a) Biochem. Biophys. Res. Commun., 201, 546551. Tsumoto,K., Ueda,Y., Maenaka,K., Ogasahara,K., Yutani,K., Watanabe,K. and Kumagai,I. (1994b) J. Biol. Chem., 269, 2877728782. Tsumoto,K., Ogasahara,K., Ueda,Y., Watanabe,K., Yutani,K. and Kumagai,I. (1995) J. Biol. Chem., 270, 1857118577. Tsumoto,K., Ogasahara,K., Ueda,Y., Watanabe,K., Yutani,K. and Kumagai,I. (1996) J. Biol. Chem., 271, 3261232616.

1317

K.Tsumoto et al. Ueda,H., Tsumoto,K., Kubota,K., Suzuki,E., Nagamune,T., Nishimura,H., Schueler,P.A., Winter,G., Kumagai,I. and Mahoney,W.C. (1996) Nature Biotech., 14, 17141718. Ueda,Y., Tsumoto,K., Watanabe,K. and Kumagai,I. (1993) Gene, 129, 129134. Verhoeyen,M., Milstein,C. and Winter,G. (1988) Science, 239, 15341536. Ward,E.S., Gussow,D., Grifths,A.D., Jones,P.T. and Winter,G. (1990) Nature, 341, 544546. Wedemayer,G.J., Patten,P.A., Wang,L.H., Schultz,P.G. and Stevens,R.C. (1997) Science, 276, 16651669. Winter,G., Grifths,A.D., Hawkins,R.E. and Hoogenboom,H.R. (1994) Annu. Rev. Immunol., 12, 433455. Winter,G. and Milstein,C. (1991) Nature, 349, 293299. Wiseman,T., Williston,S., Brandts,J.F. and Lin,L.-N. (1989) Anal. Biochem., 179, 131137. Yang,W.P., Green,K., Pinz-Sweeney,P., Briones,A.T., Burton,D.R. and Barbas III,C.F. (1995) J. Mol. Biol., 254, 392403. Received June 9, 1997; revised July 22, 1997; accepted July 23, 1997

1318