A General Method To Select Antibody Fragments Suitable for

Here, we devised a phage-based “split-Fv system” to rapidly evaluate and select ... using primers OmpXbaRV and OmpSalFR, with pFLAG-ATS (Sigma-Ald...
0 downloads 0 Views 197KB Size
Anal. Chem. 2003, 75, 4057-4064

A General Method To Select Antibody Fragments Suitable for Noncompetitive Detection of Monovalent Antigens Takahide Aburatani,† Kenzo Sakamoto,† Kenji Masuda,‡ Kosuke Nishi,§ Hideo Ohkawa,§ Teruyuki Nagamune,† and Hiroshi Ueda*,†,‡

Department of Chemistry and Biotechnology, School of Engineering, University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-8656, Japan, Department of Integrated Biosciences, Graduate School of Frontier Sciences, University of Tokyo, Building FSB-401, 5-1-5 Kashiwanoha Kashiwa Chiba 277-8562, Japan, and Research Center for Environmental Genomics, Kobe University, 1-1 Rokkodai-cho, Nada-ku, Kobe 657-8501, Japan

Previously, immunological detection of a small hapten was only possible in competitive format, which needed a competitor antigen either labeled by a reporter or attached to a carrier protein. Recently, we proposed the open sandwich (OS) immunoassay, a simple immunoassay that can noncompetitively determine monovalent antigen concentration by measuring the antigen-dependent change in a heavy-chain variable region (VH)/light-chain variable region (VL) interaction of an antibody. However, there was a limitation in the assay that the antibody used should have a suitable property such that the VH/VL interaction would become fairly strong along with the addition of antigen. Here, we devised a phage-based “split-Fv system” to rapidly evaluate and select antibody variable region (Fv) fragments that are suitable to OS immunoassay. When three antibodies raised against endocrine disruptor bisphenol A were tested with this system, all were more or less suitable to OS-ELISA. Among them, the best Fv selected was used to construct fusion proteins of VH tethered to an alkaline phosphatase and a tagged VL that can be site-specifically biotinylated to perform direct OSELISA. The results showed that the OS-ELISA detects bisphenol A with higher sensitivity than the corresponding competitive assay, also implying that many antibodies to small haptens have suitable properties for OS-ELISA. Investigating protein-protein interaction is considered to be one of most important research targets in the era of proteomics. However, many protein-protein interactions are known to be modulated by their cofactor(s) or interacting molecule(s). For example, the interaction between calmodulin and M13 fragment of myosin light chain kinase is controlled by the Ca2+ concentration,1 and that of FKBP and FRAP by an antibiotic rapamycin.2 * Corresponding author. Phone: +81-3-5841-7362. Fax: +81-3-5841-7362. E-mail: [email protected]. † School of Engineering, University of Tokyo. ‡ Graduate School of Frontier Sciences, University of Tokyo. § Kobe University. (1) Rhoads, A. R.; Friedberg, F. FASEB J. 1997, 11, 331-340. (2) Chen, J.; Zheng, X. F.; Brown, E. J.; Schreiber, S. L. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 4947-4951. 10.1021/ac034280n CCC: $25.00 Published on Web 07/11/2003

© 2003 American Chemical Society

Figure 1. Schematic diagram of open-sandwich ELISA. Antigendriven stabilization of an antibody Fv is monitored by the amount of surface-bound enzyme-VH conjugate on the VL-immobilized surface.

On the other hand, the interaction detected by the existing screening methods for protein-protein interaction, such as yeast two-hybrid system or various display systems, is essentially limited to binary interaction, and the binding of and its modulation by effector(s) or ligand(s) is not easily identified. As a practical application of cofactor-dependent protein-protein interaction in nature, recently we proposed open sandwich (OS) ELISA, an assay based on a phenomenon that in some antibodies, the association of separated VH and VL chains from a variable domain of antibody is strongly favored in the presence of antigen (Figure 1).3,4 When compared with conventional sandwich immunoassays employing two Abs, antigen-driven association of VH and VL chains has proven to have several advantages, such as a wider dynamic range, a shorter measurement time, and the applicability to homogeneous assays.5-7 Among them, a significant advantage of OS-ELISA over conventional ELISA is that it can measure the concentration of monovalent antigen such as hapten in noncompetitive format with the need of a single variable domain to conduct the assay. Generally, a noncompetitive immunoassay shows sensitivity that is superior to a competitive one,8 which needs preparation of competitor (3) Ueda, H.; Tsumoto, K.; Kubota, K.; Suzuki, E.; Nagamune, T.; Nishimura, H.; Schueler, P. A.; Winter, G.; Kumagai, I.; Mohoney, W. C. Nat. Biotechnol. 1996, 14, 1714-1718. (4) Schneider, R. J. Anal. Bioanal. Chem. 2003, 375, 44-46. (5) Arai, R.; Ueda, H.; Tsumoto, K.; Mahoney, W. C.; Kumagai, I.; Nagamune, T. Protein Eng. 2000, 13, 369-376. (6) Arai, R.; Nakagawa, H.; Tsumoto, K.; Mahoney, W.; Kumagai, I.; Ueda, H.; Nagamune, T. Anal. Biochem. 2001, 289, 77-81. (7) Yokozeki, T.; Ueda, H.; Arai, R.; Mahoney, W.; Nagamune, T. Anal. Chem. 2002, 74, 2500-2504.

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003 4057

antigens and careful tuning of assay conditions. So far, a high affinity mutant of anti-NP Ab, N1G9 ,is the only antihapten Ab proven to be applicable to OS-ELISA.9 Although it is theoretically possible to test each variable domain one by one, elucidation of a rapid method to evaluate and select Ab’s suitable to OS-ELISA is considered to be of prime importance for wider use of this method. Over the past few years, a variety of environmental contaminants have been reported to adversely affect humans and wildlife through interactions with the endocrine system. These compounds have been broadly defined as environmental endocrine disrupting chemicals (EDC). As an EDC, bisphenol A (BPA) is a widely used small monomer in the manufacture of polycarbonate plastics and epoxy resins and has been reported to have estrogenic effects.10 BPA is routinely measured by gas chromatography/mass spectrometry (GC/MS) or high performance liquid chromatography (HPLC).10 Although GC/MS and HPLC boast high sensitivity and specificity, they cannot process multiple samples rapidly. Moreover, the preparation of samples for gradient separation and purification is time-consuming. Therefore, the development of more rapid and high-throughput detection system for EDC has been undertaken. For example, the detection of BPA using competitive ELISA has been reported,11 first with rather modest detection limit (5 µg/L). Later, with the use of an automated system using an immunoaffinity column12 or bacterial magnetic particle,13 higher sensitivity was attained despite competitive detection. Given the sensitivity of these methods is sufficient in most cases, they could be costly due to the use of sophisticated instruments. To cultivate the possibility of sensitive noncompetitive detection of EDC and other small chemicals by OS-ELISA, here we developed a “split Fv (spFv) system”, a phage display-based screening system that can rapidly evaluate and select proteinprotein interaction of a heterodimeric receptor (Fv) and its modulation by the ligand (antigen). Using this system, we tested a panel of anti-BPA antibodies derived from established hybridomas to select suitable clones for OS-ELISA. The selected VH fused to Escherichia coli alkaline phosphatase showed good performance in one-step OS-ELISA, also showing a better sensitivity than that of corresponding competitive ELISA. EXPERIMENTAL SECTION Construction of the spFv Phagemid for HyHEL-10. SpFv expression vector was constructed as follows. The gene fragment encoding OmpA signal peptide was amplified by PCR using primers OmpXbaRV and OmpSalFR, with pFLAG-ATS (SigmaAldrich, Tokyo, Japan) as a template (The sequences of the primers are summarized in Table 1). The amplified fragment was digested with XbaI and SalI, and inserted to pHSG397 (TakaraBio, Kyoto, Japan) digested with the same. The plasmid was confirmed for the inserted sequence with DNA sequencer SQ-5500 (Hitachi, (8) Pradelles, P.; Grassi, J.; Creminon, C.; Boutten, B.; Mamas, S. Anal. Chem. 1994, 66, 16-22. (9) Suzuki, C.; Ueda, H.; Mahoney, W.; Nagamune, T. Anal. Biochem. 2000, 286, 238-246. (10) Shin, H.-S.; Park, C.-h.; Park, S.-J.; Pyo, H. J. Chromatogr., A 2001, 912, 119-125. (11) Goda, Y.; Kobayashi, A.; Fukuda, K.; Fujimoto, S.; Ike, M.; Fujita, M. Water Sci. Technol. 2000, 42, 81-88. (12) Ohmura, N.; Lackie, S. J.; Saiki, H. Anal. Chem. 2001, 73, 3392-3399. (13) Matsunaga, T.; Ueki, F.; Obata, K.; Tajima, H.; Tanaka, T.; Takeyama, H.; Goda, Y.; Fujimoto, S. Anal. Chim. Acta 2003, 475, 75-83.

4058

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

Tokyo, Japan) and designated pHSG397/Omp. The gene for the minor coat protein p9 of filamentous phage M13 (gene 9) was amplified by PCR using primers Flg9XhoRV and g9XbaFR with ssDNA for helper phage M13KO7 as a template. The amplified fragment was digested by XhoI and XbaI, subcloned to pHSG397/ Omp digested with the same, and confirmed for the sequence. The joined Flag-g9-ompA fragment was amplified by PCR with the primers LinkBackX and LinkFor, and was used as a spFv linker. Similarly, the gene 7 of filamentous phage M13 was amplified by PCR using primers g7KpnRV and g7EcoFR with ssDNA for M13KO7 as a template. The fragment was digested with KpnI and EcoRI, subcloned to pBluescript II(KS+) (Stratagene), digested with the same, and confirmed for the nucleotide sequence (pBS-g7). Phagemid pKS1 was constructed as follows: A DNA fragment encoding anti-bovine serum albumin (BSA) single-chain Ab (scFv) followed by a His6-Myc tag was PCR-amplified using primers M13RV (TakaraBio) and MycAKpnFor, with pIT2(13CG2)14 as a template. The fragment was digested with SfiI, blunted with T4 DNA polymerase, and digested with KpnI. The digested fragment was ligated with phagemid pK1,15 which had been digested with PstI near the 5′ of mature gene 3, blunted with T4 DNA polymerase, and digested with KpnI. The resultant vector encoding RBSA scFv-His6-Myc (pScFv3) was digested with KpnI and EcoRI, and the gene 7 fragment obtained by digestion of pBS-g7 by the same enzymes was inserted, denoted pScFv/g7. The VH and VL fragments of anti-hen egg lysozyme HyHEL-10 were separately amplified by PCR with a plasmid pCantab5E(HyHEL-10) as a template. The primers used for the amplification were M13RV and VH1For-2X for the amplification of VH, and Vκ3Back and reverseSEQ′ for the amplification of VL. The gelpurified VH, VL, and spFv linker fragments were assembled by overlap extension PCR with primers M13RV and reverseSEQ′. The assembled spFv fragment was digested with NcoI and NotI and incorporated to the vector pScFv/g7 digested with the same to obtain an spFv expression vector phagemid pKS1(HyHEL-10) (Figure. 2C). To avoid spontaneous deletion of the spFv fragment inserted to pKS1, the NcoI-EcoRI fragment of pKS1(HyHEL-10) was transferred to pCantab5E (AP Biotech), and two tandem repeats of glutamine permease terminators (tHP) were incorporated upstream and downstream of the spFv coding sequence, resulting in a new spFv expression vector pKST2, the detail of which will be described elsewhere. Construction of Anti-Bisphenol A spFv Phagemids. The spFv expression phagemids for the anti-BPA antibodies were constructed by PCR amplification of the VH and VL fragments with the plasmids encoding anti-BPA scFv (kindly provided by Horiba Biotech. Inc., Kyoto) as templates.16 The primers used were MH1back17 and VH1FOR-2X for the amplification of VH, Vk3Back and MJK1FONX18 for BBA2187/BBA2617 VLs, and Vk3Back and MJK4FONX18 for BTE3456 VL. Each VH-VL pair was assembled (14) Aburatani, T.; Ueda, H.; Nagamune, T. J. Biochem. (Tokyo) 2002, 132, 775782. (15) Kristensen, P.; Winter, G. Fold. Des. 1998, 3, 321-328. (16) Nishi, K.; Takai, M.; Morimune, K.; Ohkawa, H. Biosci. Biotechnol. Biochem. 2003, 67, 1358-1367. (17) Wang, Z.; Raifu, M.; Howard, M.; Smith, L.; Hansen, D.; Goldsby, R.; Ratner, D. J. Immunol. Methods 2000, 233, 167-177.

Table 1. The Nucleotide Sequences of the Primers Used

Underline shows the recognition sites for restriction enzymes.

with the spFv linker by overlap extension PCR. The amplified spFv fragments were reamplified by PCR to incorporate restriction sites, with primers MH1backSfi and JK1NOT10 for BBA2187 and BBA2617, or MH1backSfi and JK4NOT10 for BTE3456. The amplified products were digested with NcoI and NotI, and ligated with pKST2 digested with the same. Phage Display of spFv Fragment. TG1 (supE, hsd∆5, thi, ∆ (lac-proAB), /F′ [traD36, proAB+, lacIq, lacZ∆M15]) or HB2151 (ara, ∆ (lac-proAB), thi/F′ proAB+, lacIq, lacZ∆M15) cells (AP Biotech, Tokyo, Japan) carrying each spFv-encoding phagemid were incubated with shaking at 37 °C in 2× TY containing 100 µg/mL ampicillin and 1% glucose until the OD600 reached 0.5, then helper phage M13KO7 was added with m.o.i. of 20. The cells were incubated at 37 °C without shaking for 30 min and then centrifuged at 3300g for 10 min. The cells were resuspended in 2× TY containing 100 µg/mL ampicillin and 25 µg/mL kanamycin and were incubated with shaking at 30 °C overnight. When TG1 was used as a host strain, the overnight culture was centrifuged at 10800g for 10 min, and then PEG/NaCl (20% PEG 6000, 2.5 M NaCl) was added to the supernatant. Left at 4 °C for at least 1 h, (18) Pope, A. R.; Embleton, M. J.; Mernaugh, R. In Antibody Engineering; McCafferty, J., Hoogenboom, H. R., Chiswell, D. J. Eds.; IRL Press: Oxford, 1996; Vol. 169, 1-40.

the supernatant was centrifuged at 3300g for 30 min. The pellet was resuspended in TE (pH 8). The phage solution was centrifuged at 16000g for 10 min, and the supernatant containing the phage particle displaying spFv was recovered and stored at 4 °C. When HB2151 was used as a host strain, the overnight culture was centrifuged at 10800g for 10 min, and the supernatant containing VH-displaying phage and soluble VL was recovered and stored at 4 °C. Phage ELISA. The antigen-Fv interaction was tested with spFv-phages prepared from TG1 as a host strain. The Falcon 3912 microplates (Becton Dickinson, Oxnard, CA) were coated overnight with 100 µL per well of antigen solution in 10 mM PBS. Blocked at room temperature for 2 h with 25% Block Ace (Dainippon Pharmaceutical, Osaka, Japan) in PBS, washed three times with PBS containing 0.1% Tween-20 (PBST), the plates were incubated at room temperature for 1 h with 100 µL per well of spFv-phage solution in 10% Block Ace in PBS (BPBS). The plates were washed three times with PBST and incubated at room temperature for 1 h with 100 µL per well of 5000-fold diluted HRPconjugated mouse anti-M13 (AP Biotech) in BPBS. The plate was washed six times with PBST and developed with 100 µL/well substrate solution (100 µg/mL 3,3′,5,5′-tetramethylbenzidine (TMB, Sigma) and 0.04 µL/mL H2O2, in 100 mM NaOAc, pH 6.0). Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4059

After incubation for 5-30 min, the reaction was stopped with 50 µL per well of 1 M sulfuric acid, and the absorbance was read at 450 nm. OS-ELISA with spFv System. The VH/VL interaction was tested by OS-ELISA with spFv-phages prepared from HB2151 as a host strain. One hundred microliters per well of 1 µg/mL protein L (Actigen, Cambridge, U.K.), which specifically binds VL, was coated overnight in PBS. Blocked at room temperature for 2 h with 25% Block Ace in PBS and washed three times with PBST, the plate was incubated at room temperature for 1 h with 100 µL per well of 1011 cfu/mL spFv-phage as well as various concentrations of antigen, which were mixed 1 h before and preincubated. Because of its limited water solubility, BPA was dissolved in methanol and diluted in PBS containing a final 5% methanol. The plate was washed three times with PBST and incubated at room temperature for 1 h with 100 µL per well of 2000-fold diluted rabbit anti-fd bacteriophage-biotin conjugate (Sigma) in BPBS. The plate was washed three times with PBST and incubated at 25 °C for 1 h with 100 µL per well of 2000-fold diluted avidin-HRP (EY Laboratories, San Mateo, CA) in BPBS. The plate was washed six times with PBST and developed as above. OS-ELISA with Alkaline Phosphatase Fusion Proteins. The VH (BBA2187)-alkaline phosphatase fusion expression plasmid, pVH(BBA2187)-PhoA, was constructed on the basis of pVH(NP)-PhoA(D101S).19 The VH fragment of BBA-2187 was amplified by PCR using primers 2187HERV and 2187HHnd. The amplified fragment was digested with EcoRV and HindIII and inserted into pVH(NP)-PhoA(D101S) digested with the same to replace the VH(NP) fragment. To express VL with an N-terminal biotinylation substrate peptide (Avi-tag), the sequence encoding Avi-tag was inserted between the pelB leader and NcoI site of pET20b(+) (Novagen, Madison, WI) by PCR using primers T7p and pelbioFR. The amplified fragment was digested with XbaI and NcoI and ligated with pET20b(+) digested with the same, designated pET20bio. The BBA2187 VL fragment was amplified by PCR using primers 2187VLNcoRV and 2187VLNotFR. The amplified fragment was digested with NcoI and NotI and ligated with pET20bio digested with the same, designated pVL(BBA2187)bio. VH-PhoA and tagged VL proteins were prepared as described previously.19 Purified tagged VL was biotinylated in vitro by biotin protein ligase (Avidity, Denver, CO) at 30 °C for 30 min according to the manufacturer’s instructions. One hundred microliters per well of 100 µg/L bioVL was immobilized on a streptavidin-coated microplate at 25°C for 1 h and washed four times with TBST. A mixture of 50 µg/L VH-PhoA and varied concentrations of BPA in a total of 100 µL TBS containing 5% methanol was added to the VL-immobilized wells, and the plate was incubated for 1 h. After washing four times with TBST and two times with 200 mM TrisHCl (pH 9.8) and 10 mM MgCl2, bound VH-PhoA was detected with 80 µL of CDP-star substrate solution (Applied Biosystems, Foster City, CA). After incubation for 30 min, light emission was measured with MicroLumat plus (Perkin-Elmer, Shelton, CT). Competitive ELISA. The scFv(BBA2187) fragment was amplified with primers 2187HERV and 2187LHnd. The amplified fragment was digested with EcoRV and HindIII and ligated with

pVH(NP)-PhoA(D101S) digested with the same. The resulting plasmid was used for the expression and subsequent purification of scFv-PhoA as above. A 100-µL portion per well of 3 mg/L BPA-RSA conjugate was coated overnight in TBS. Blocked at room temperature for 2 h with 25% Block Ace in PBS and washed three times with TBST, the plates were incubated at room temperature for 90 min with 50 µL per well of 100 µg/L scFvPhoA solution and 50 µL of varied concentrations of BPA in TBS containing 5% methanol.

(19) Suzuki, C.; Ueda, H.; Tsumoto, K.; Mahoney, W. C.; Kumagai, I.; Nagamune, T. J. Immunol. Methods 1999, 224, 171-184.

(20) Gao, C.; Mao, S.; Lo, C. H.; Wirsching, P.; Lerner, R. A.; Janda, K. D. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 6025-6030.

4060

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

RESULTS AND DISCUSSION Development of the spFv System. Until now, OS-ELISA has been only possible after recombining individual cDNA fragments for VH and VL to appropriate expression vectors, expressing them, and purifying them to perform each assay. To obviate these steps, we decided to develop a more robust screening method, termed the spFv-phage display system (Figure 2). Phage display is a powerful method for screening functional antibody fragments retaining a high affinity to the antigen. For the screening of the antigen binding ability of VH/VL by phage display, a simultaneous display of both fragments in close proximity on the same phage is necessary. On the other hand, to perform OS-ELISA, that is, measurement of VH/VL interaction, separate expression of VH/ VL, for example, phage display of a VH fragment as well as production of a soluble VL fragment is desired. To enable the facile switch of these two display formats, we adopted a filamentous phage p7-p9 display system 20 to individually display VH and VL fragments as a functional Fv on the tip of the phage. However, we put an important and crucial modification into the reported system that an amber codon was placed between VL (tethered with His-myc tag) and gene 7, which was expected to enable/ disable display of VL by changing the host sup phenotype. When sup+ strains, such as TG-1 or XL1-blue, are used for phage production, VL-(His-myc)-p7 will be expressed; thus, functional Fv will be displayed on the phage (Figure 2A). On the other hand, when sup- strains, such as HB2151, are used for the phage production, soluble VL-(His-myc) along with a VH-displaying phage will be expressed; thus, OS-ELISA can be performed in a plate in which VL is immobilized by anti-myc Ab or VL-specific ligand protein L coated on the plate (Figure 2B). To perform these assays, a phagemid pKS1, which encodes tagged p7 instead of p3 in a normal phagemid for phage display, was constructed. To this vector, a DNA fragment encoding VH-p9 followed by a secretion signal and a VL (spFv fragment) was inserted to express Fv either on the phage or in the split form (Figure 2C). To show the proof of principle, first we cloned an Fv fragment of anti-hen egg lysozyme (HEL) HyHEL-10, which is known to be well-suited to OS-ELISA, to pKS1. By producing phages using suppressor strain TG-1, the antigen-binding ability of the spFv-phage was confirmed by performing the phage ELISA. In a plate with immobilized HEL, the spFv-phage showed strong binding, which was comparable to that of the scFv-phage to HEL (Figure 2D). On the other hand, the spFv-phage did not bind to BSA-immobilized wells or wells with no ligands; neither did the scFv-phage. These results show that in addition to the scFv-phage, spFv-phage retains specificity and affinity comparable to the original Fv.

Figure 2. The split Fv (spFv) system. (A) Schematic diagram of the system when the spFv-phage is produced by an amber suppressor host strain. (B) The same when the VH-displaying phage and soluble VL were produced by a nonsuppressing host. (C) The structure of the spFvcoding region of the phagemid. An amber codon is placed between the His6-myc-tagged VL and g7 sequences to switch two display modes. (D) Phage-ELISA to assay HEL-binding activity of spFv (HyHEL-10)-phage derived from amber suppressor TG1. 1 × 1010 cfu/mL of spFv-phage was reacted with the HEL-coated microplate. (E) OS-ELISA with spFv (HyHEL-10) culture supernatant prepared with nonsuppressing HB2151. Several concentrations of HEL together with the supernatant containing 1 × 1011 cfu/mL spFv-phage were reacted with the 9E10-coated microplate. Detection was performed with HRP-conjugated mouse anti-M13. The plots for HEL are the average of three independent measurements, with error bars showing 1 SD.

OS-ELISA with spFv (HyHEL-10)-Phage. To test whether the system can be used to evaluate VH/VL interaction of the Fv fragment, a nonsuppressing strain HB-2151 was transformed with the same phagemid and infected with helper phage, and culture supernatant containing the mixture of VH-displaying phage and soluble VL was prepared. First, the binding of the VH-phage to HEL-coated plate in the presence of VL was tested (Figure 2D). The supernatant showed a binding pattern similar to that of spFvor scFv-phage, showing cooperative binding of VH/VL fragments to the specific antigen. Then the culture supernatant was poured into microplate wells immobilized with anti-myc Ab 9E10. In this step, the VL fragment with C-terminal myc-tag was expected to be captured by the immobilized Ab, and VH-phage and antigen HEL, if present, to make a complex with the immobilized VL. After incubation, conventional ELISA using HRP-labeled anti-M13 was performed to determine the amount of immobilized VH-phage. As shown in Figure 2E, a clear increase in signal was observed as the concentration of HEL increased, indicating that with the use of this spFv system, OS-ELISA, to determine VH/VL interaction and its dependency to antigen concentration could be easily performed, as well as normal ELISA for antigens. Model Panning of Specific spFv-Phage. To show that biopanning of spFv-displaying phages is also possible, as with Fabor scFv-displaying phages, a model selection of two spFv with different specificities was performed. A minor fraction of HyHEL10 spFv-displaying phages diluted in an irrelevant antifluorescein 31IJ3 spFv-phage (1011 cfu/mL) was poured in the wells precoated with HEL, washed, and eluted, and the ratio of HyHEL-10 spFv-

Figure 3. Model panning of HyHEL-10 spFv-displaying phage. A minor fraction of the HyHEL-10 spFv-phage produced by TG-1 was diluted in an irrelevant antifluorescein 31IJ3 spFv-phage in BPBS and was reacted in the microplate wells precoated with HEL for 2 h at 25°C, washed twice each with PBST and PBS, and eluted with 0.2M glycine (pH 2.2). The recovered phages were infected into TG-1, panned again, and the number of HyHEL-10 spFv-phages recovered out of 48 colonies for each round was obtained by ELISA.

phage recovered was obtained by ELISA and colony PCR. As shown in Figure 3, specific enrichment of HyHEL-10 spFvdisplaying phage to 38 and 83% was attained from one in 106 mixture after one and two rounds of biopanning, respectively. The result shows that the system can be used for the selective enrichment of a specific antigen binder. Phage ELISA of Anti-BPA spFv-Phages. Since the spFv system was shown to work for the model Fv, the system was Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4061

Figure 4. (A) Dose dependency in antigen binding of the three anti-BPA spFv-phages produced by TG-1(sup+). BPA-RSA conjugate (1 µg/mL) was incubated overnight and immobilized on the microplate. Serially diluted phages were reacted with the wells coated with BPA-RSA conjugate and detected with HRP-conjugated mouse anti-M13. (B) OS-ELISA for BPA-RSA conjugate with spFv supernatant produced by HB2151(sup-). Several concentrations of the conjugate and spFv culture supernatant containing 1 × 1011 cfu/mL of VH-phages were reacted with protein L-coated microplate. Detection was performed with biotinylated rabbit anti-fd and avidin-HRP. (C) OS-ELISA for BPA. Several concentrations of BPA and BBA2187 spFv culture supernatant containing 1 × 1011 cfu/mL of VH-phage were assayed with protein L-coated (PL+) or uncoated (PL-) microplates, as above. The plots are the average of two independent measurements, with error bars showing 1 SD.

applied to evaluate the suitability of established antihapten mAbs to OS-ELISA. We chose BPA as a model hapten because of its importance as an EDC, and focused on three anti-BPA mAbs, BBA2187, BBA-2617, and BTE-3456.16 The mAbs were derived from hybridoma cells from mice immunized with BPA conjugated to bovine serum albumin BSA, and their cDNA as well as scFv genes were isolated. We subcloned the genes to pKST2, a modified version of pKS1, which was inserted with two transcription terminator sequences both upstream and downstream of the spFv operon, which significantly stabilized the inserted sequence after prolonged culture (data not shown). To confirm the antigenbinding abilities of these clones, TG-1 was transformed to prepare the spFvs displayed on phages, and their binding affinities to BPA conjugated to rabbit serum albumin (BPA-RSA) immobilized on a microplate were compared by phage-ELISA with serially diluted phages (Figure 4A). The result showed that all the three antiBPA spFv-phages showed dose-dependent binding toward BPARSA. In contrast, all the phages at 1010 cfu/mL did not show detectable binding to BSA, nor M13KO7 at 1010 pfu/mL bound to BPA-RSA (data not shown). This indicates that the spFvs of anti-BPA antibodies displayed on filamentous phages are essentially functional. Among the three clones, BBA-2187 showed binding at its lowest titer, which coincided with the results obtained with whole mAbs, scFvs, and Fvs.16 OS-ELISA with the spFv-System. To perform OS-ELISA with these clones, a nonsuppressing host HB2151 was transformed with the phagemids encoding spFv genes, and the phages were rescued with M13KO7. The culture supernatant containing VH-displaying phages and soluble VL was prepared, and each was diluted to 1 × 1011 cfu/mL in phage titer. This was mixed with varied concentrations of BPA-RSA conjugate, and the VH/VL complex formed was captured on a microplate with immobilized protein L. When the amount of immobilized phage was measured, a clear relationship between the signal generated and the BPA-RSA conjugate concentration was observed (Figure 4B). An increase in signal at the antigen concentration of more than ∼1 µg/L was observed for all three clones. Among these, BBA-2187 showed the strongest signal in response to BPA-RSA concentration, while the clone also showed highest background, presumably as a result of intrinsically stronger VH/VL interaction. Considering its largest 4062 Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

increase in signal, thereafter we focused on this clone for further examinations. Next we performed OS-ELISA for free BPA with the same culture supernatant. After varied concentrations of BPA was added to the supernatant, the VH/VL complex formed was captured on a microplate through protein L, as above (Figure 4C). In the presence of protein L, the signal increased as the antigen concentration increased. On the other hand, no signal increase was observed in the wells without protein L, ruling out the possibility of detecting nonspecifically bound BPA on microplate wells. The lowest measurable BPA concentration was between 0.1 and 1 µg/L, which was equivalent to or less than the detection limit attained with the competitive assay conducted with scFvphage.16 OS-ELISA with VH-PhoA Protein. To confirm that the spFv system-selected clone was really suitable to OS-ELISA in a practical assay format, a direct OS-ELISA utilizing purified VH enzyme conjugate together with isolated VL fragment was performed. For this purpose, a fusion protein of VH fragment and a high-activity mutant (D101S) of Escherichia coli alkaline phosphatase (VH-PhoA), and VL that has an N-terminal tag for in vitro (and possibly in vivo) biotinylation were prepared. To perform OS-ELISA at optimal conditions, several sets of VH-PhoA and biotinylated VL concentrations were tested and determined for their best concentrations. This time, biotinylated VL was first immobilized onto streptavidin-coated plate wells, and VH-PhoA mixed with several concentrations of BPA was added to the VLimmobilized wells. As a result, a significant signal increase according to the increase in BPA concentration was observed (Figure 5A). The measurable lower limit of antigen concentration was between 0.1 and 1 µg/L, which was similar to the value obtained with the spFv system. Competitive ELISA. To compare OS-ELISA with a conventional competitive assay, indirect competitive ELISA was performed with scFv (BBA2187)-PhoA prepared similarly. Several sets of scFv-PhoA and immobilized BPA-RSA conjugate concentrations were tested, and a pair of 50 µg/L scFv-PhoA and 3 µg/L BPA-RSA was chosen since this gave almost the same maximum chemiluminescence intensity as in the OS-ELISA. As a result, the detection range obtained was ∼10 to 100 µg/L (Figure 5B). The

Figure 5. (A) OS-ELISA with VH(BBA-2187)-PhoA and biotinylated VL for the quantitation of BPA. Biotinylated VL was immobilized on a streptavidin-coated microplate, and several concentrations of BPA and VH-PhoA were reacted. Bound VH-PhoA was detected by CDP-star. The plots are the average of three independent measurements, with error bars showing 1 SD. (B) Indirect competitive ELISA with scFv (BBA-2187)PhoA protein. BPA-RSA conjugate was immobilized on a microplate and competed with serially diluted BPA samples to bind scFv-PhoA. Bound scFv-PhoA was detected as above.

competitive ELISA gave a narrower detection range and ∼100 times higher detection limit than OS-ELISA in which the same Ab at the same concentration was used for detection. Though the optimization of reaction condition might reduce this remarkable difference in sensitivity to some extent, a characteristic of OSELISA that a positive signal was generated in a wider detection range was clearly shown in this comparison. CONCLUSIONS Here we showed that the split Fv system is a method for fast and effective evaluation/selection of Ab Fv fragment suitable to OS-ELISA. Among three anti-BPA mAbs, BBA-2187 was identified to be most suitable to OS-ELISA for the spFv system, probably because of its highest affinity to the antigen. In our previous attempt to perform OS-ELISA with anti-NP Ab, the increased affinity toward its antigen was critical to attain a good response in OS-ELISA.9 In addition, the sensitivity obtained here (∼0.2 µg/L = 1 nM) was far better than that with anti-NP OS-ELISA (∼1 µM). The difference between the antigen-binding affinities of the two antibodies (N1G9 mutant for NP, 3.4 × 107/M; BBA-2187 for BPA estimated from IC50 in competitive ELISA, 1.0 × 109/M) may at least in part account for this difference and also suggests the importance of high affinity to an antigen in OS-ELISA, as well as in conventional ELISA. In this sense, our two-step selection employed here was reasonable, because the first selection of highest affinity binders to antigen will give us a higher possibility of spotting most suitable candidates to OS-ELISA. Although OS-ELISA in a practical format was performed here by genetically linking the VH to alkaline phosphatase, the format for practical OS immunoassay is not limited to ELISA. For example, with the use of β-galactosidase ∆R/∆ω complementation, 1000-fold higher sensitivity was attained than OS-ELISA in a homogeneous format (OS-ECIA), probably owing to the need for reduced reagent (i.e., VH and VL) concentrations that leads to lower background VH/VL association.4,7 Probably, we can expect not only higher sensitivity by the combination of the selected Fv and OSECIA but also even higher sensitivity by future technical developments of more sensitive interaction assays. Pellequer et al. categorized the changes in compactness of the VH/VL interface between bound and unbound antibodies on the

size of the antigen and found that small antigens or haptens cause a closure of the interface, whereas larger protein antigens have little effect of the compactness of the VH/VL interface.21 This is also in accordance with previous observations that antihapten Abs recognize their antigens between the VH/VL interface, whereas antiprotein Abs do it on the upper surface. In our previous attempt for applying Abs to OS-ELISA, four anti-BSA Abs14 were not suitable to OS-ELISA, mostly becaues of their strong VH/VL interaction in the absence of the antigen. On the contrary, all three anti-BPA Abs examined here showed increased signal along with the increased BPA-RSA concentration, while we are also aware of an antidigoxin antibody whose VH/VL interaction is exceptionally strong (HU, unpublished observation). These led us to a hypothesis that Abs against small antigens or haptens are more suitable to OS-ELISA than those against larger antigens. Further studies with natural and engineered libraries using spFv system will verify this hypothesis and allow detailed analysis on the molecular basis of variable VH/VL interaction strength and its antigen dependency (Masuda et al., in preparation). Aside from selecting OS-compatible Abs, the spFv system will also be useful in library-vs-library screening of various natural heterodimeric protein pairs recognizing specific ligands, which will include T-cell receptors, MHC class II, cytokine receptors, and transcription factors, such as c-fos and c-jun. In the future, enrichment of specific ligand binder pairs from a combinatorial phage library of either natural or synthetic origin may enable isolation of these proteins with a desired specificity. As alternative recognition units to Fv for the OS-immunoassay and for the antigen-mediated genetically modified cell amplification system (AMEGA),22 such protein pairs will become useful tools in molecular biology, diagnostics, and medicine. Abbreviations Used. Ab, antibody; BPA, bisphenol A; BSA, bovine serum albumin; ELISA, enzyme-linked immunosorbent assay; FKBP, FK506-, and rapamycin-binding protein; FRAP, FKBP-rapamycin-associated protein; Fv, Ab variable region; HEL, hen egg lysozyme; HRP, horse raddish peroxidase; NP, 4-hydroxy(21) Pellequer, J. L.; Chen, S.; Roberts, V. A.; Tainer, J. A.; Getzoff, E. D. J. Mol. Recognit. 1999, 12, 267-275. (22) Kawahara, M.; Ueda, H.; Morita, S.; Tsumoto, K.; Kumagai, I.; Nagamune, T. Nucleic Acids Res. 2003, 31, e32.

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4063

3-nitrophenacetyl; OS-ELISA, open sandwich ELISA; PBS, phosphate-buffered saline; PCR, polymerase chain reactions; RSA, rabbit serum albumin; scFv, single chain Fv; spFv, split Fv; VH, Ab heavy chain variable region; VL, Ab light chain variable region. ACKNOWLEDGMENT We are grateful to Drs. P. Kristensen, I. Tomlinson, and P. Holliger, respectively, in the Laboratory of Molecular Biology, Medical Research Council, Cambridge, U.K. for kindly providing phagemids pK1, pIT2(31IJ3), and pCantab5E(HyHEL-10). We also

4064

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

thank Drs. W. Mahoney, G. Winter, and T. Ueda for helpful discussions and support. T.A. was supported by research fellowships of the Japan Society for the Promotion of Science (JSPS) for Young Scientists. This work was supported in part by Grantsin-Aid for Scientific Research (C12650782 and B140350430), from JSPS, Japan. Received for review March 19, 2003. Accepted May 27, 2003. AC034280N