During the past time, accurate detection of haptens are widely required because of their use in clinical in vitro diagnostics and environmental or food monitoring. However, all the conventional methods have their own drawbacks. To overcome these demerits, open-sandwich immunoassay (OS-IA) was established for the noncompetitive detection of small molecules. The principle of OS-IA is based on the association of the antibody variable region (VH and VL) in the presence of the corresponding antigen. In order to obtain highly specific and sensitive binders to haptens, here we introduced a novel selection method Open-Sandwich (OS) selection. In this study we performed OS selection to select better binder for 17β-estradiol (E2) from a randomized library (Chapter 2) and binders for analogous steroids progesterone, estriol, and cortisol from a binary code library (Chapter 3). These targets are the most commonly measured steroid compounds.

On the first chapter of PhD thesis, I overviewed this area and made introduction to the following two topics: 1) Affinity and specificity maturation of anti-E2 antibody by OS selection from a randomized library, and 2) Selection of anti-steroid antibodies by OS selection from a binary code library.

In Chapter two, the affinity and specificity maturation of anti-E2 antibody by OS selection from a randomized library is described. The first step was to construct a randomized library. The VH domain was diversified using in vitro mutagenesis called error-prone PCR, and the randomized fragments were ligated to a phagemid vector and transformed to E. coli. After VH-displaying phages were produced from the bacteria, three rounds of OS biopanning were performed, and a mutant that showed improved signal was obtained. To estimate the affinity and specificity of the mutant, OS phage ELISA with immobilized MBP-VL was performed. According to the dose-dependency for E2, the mutant exhibited increased E2-dependent signal compared with the wild type, suggesting its higher affinity and suitability to OS-IA. However, for other E2 analogs of testosterone, estriol, cortisol and progesterone, the mutant showed much lower cross-reactivity than the wild type.

To further confirm the affinity and specificity of the clones, chemiluminescence-based (CL) OS- and competitive ELISAs were performed. For CL OS-ELISA, the result was consistent with the result of phage OS ELISA, suggesting the mutant's increased affinity for E2 and selectivity against the analogs. Due to higher sensitivity of the CL assay than the chromogenic assay, the detection sensitivity for E2 was improved compared with that of phage OS ELISA, as judged by their EC50s. In the case of competitive ELISA with single-chain Fv (scFv) fragments, free E2 exhibited stronger inhibition to the binding of the mutant scFv to E2-BSA, than the wild-type scFv. However, the rest of analogs exhibited much weaker inhibition for the mutant than the wild-type, also confirming the increased affinity and specificity of the mutant. The calculated cross reactivity suggested that the mutant was specific enough for clinical diagnostics for E2.

After nucleotide sequencing of the mutant and wild-type, we found that the mutant had one mutation at site 85 compared with the wild type, changing from E (glutamine) to K (lysine). Assuming the similar property of amino acid R (arginine) to K (lysine), the residue was substituted by amino acid R (arg) using site directed mutagenesis by SOE PCR to construct the mutant E85R. The result of OS phage ELISA showed that the affinity and specificity of E85R are both decreased compared with the original mutant, demonstrating the importance of the original mutation (E to K).

In the third chapter of this thesis, an attempt to select novel anti-steroid antibodies by OS selection from a binary code library was described. The recognition of antigen by antibody variable region is mainly mediated by the complementarity determining regions (CDRs), because each CDR region gives a large surface to contact with the corresponding antigen. Several studies revealed that the CDR regions of the known antibodies are rich in the amino acids Y (tyrosine) and S (serine), maybe because these two kinds of amino acids are suitable for molecular recognition. As the most contacts with antigen were mediated by Y and S, it has been shown possible to develop a library that can be used as an universal source of antigen binders. However, it has not been clear that the above strategy is also applicable to anti-hapten antibodies. Hence, here I chose estrone, estriol, progesterone, cortisol, and testosterone as the target haptens. To begin with the construction of the library, the CDR region of VH domain was introduced by a binary Y/S diversity through SOE PCR. Using the diversified VH fragments, a VH-displaying phage library with a size of about 107 was constructed. For the selection, MBP-fused VL was immobilized, and VH-phage and target haptens were incubated to select for specific antigen binders. For efficient selection, increasing selection stringency was adopted through reducing the amount of the corresponding antigen from one round to another. After three rounds of biopanning, polyclonal OS phage ELISA with the same concentration of phage was performed to estimate the efficacy of the selection. In case of estriol, the signal was increased from round 1 to round 3, but in cases of the rest of four targets, round two showed the highest signal.

In the fourth chapter, I used the combinatorial mutagenesis to try to improve the affinity and specificity of the anti-estradiol antibody. Because the mutant M2 we obtained in the previous study showed a much higher signal than the wild type and the rest mutants, and the mutant m6 also showed increased signal compared with other mutants, we decided the mutation position 63 at M2 and mutation position 85 at m6 as the mutation positions in this combinatorial mutagenesis library construction. In this combinatorial mutation chapter, three libraries were constructed, namely, H63, H85, and H63+85 respectively. H63 is originally Phe in the wild type, which was mutated to Tyr in M2, and this position is randomized to 8 mainly hydrophobic residues F/Y/L/H/I/N/V/D. Also, H85 is randomized to 12 hydrophilic residues using codon (CAG)(CAG)N. After one round of OS selection, 96 colonies were picked up for monoclonal ELISA. The result demonstrates that the affinity and specificity of the mutant E85D are improved at the same time, however, the specificity improvement is more obvious.

In the fifth chapter of the thesis, I described the conclusion and perspective. Compared with the conventional antibody selection methods, antigen conjugation is not necessary in OS selection, and the selected VH mutants could be directly applicable for OS-IAs. This system will be applied to develop sensitive detection systems for many small molecules.

本論文は、抗体工学的手法を用いて低分子化合物、特に疾患診断ならびに環境保全に重要なステロイドホルモンであるエストラジオール(E2、 Mw:272.4)の免疫化学的検出法の性能向上を目指した研究について述べたものであり、5章より構成されている。

第1章は序論であり、本研究の意義を明確にするために、抗体工学、ELISA法に代表される免疫測定に関する既往の研究を中心に、研究の背景が述べられている。

第2章では、本論文の主題となる抗原濃度に依存した抗体の2個の可変領域VH - VL間の相互作用変化を利用した免疫測定法、オープンサンドイッチ法(OS法)の原理に基づくE2検出系の感度・特異性向上に関して述べられている。従来、E2のような分子量1,000以下の低分子ハプテン認識抗体の選択・取得法としては、生体内外を問わずハプテンキャリア複合体の調製が選択において不可欠であり、このため極めてしばしばハプテンでなく複合体を認識する抗体が取得・濃縮されてしまう問題があった。

この問題を原理的に避けうる選択法として、片方の可変領域断片(VL)を固相に固定化、もう一方(VH)をランダム化しM13ファージ粒子上に提示して、低濃度の抗原存在下でVL に結合するVH提示ファージを選択するオープンサンドイッチ選択法を用い、抗E2抗体の親和性向上ならびに測定系の検出感度向上を試みている。具体的には、複製エラーの起こりやすいPCR法により構築したVHライブラリより作製したファージライブラリを用いて3ラウンドのOS選択を行い、OS-ELISAでスクリーニングした結果、数個の候補クローンの取得に成功している。これらを検討した結果、うち一つ(Glu85Lys変異体)が野生型より高い検出感度を示したことから、このクローンについて詳細な解析を行っている。

興味深いことに、この変異体はファージOS-ELISAにおいてE2の検出能が向上するのみならず、E2の3位のみが異なる異性体である男性ホルモンtestosterone (TS)をはじめ他のステロイドに対する交差反応性が顕著に低下したクローンであることを見出している。この結果は大腸菌マルトース結合蛋白MBP融合VH/VLを用いたOS-ELISAにおいても再現され、さらに同様の結果が一本鎖抗体(scFv)提示ファージならびにMBP-scFvを用いた競合ELISAでも得られたと述べている。この原因を探るため、類似の電荷変化を伴う変異体Glu85Argを用いた解析を行ったが、若干の交差反応性向上は見られたものの検出感度はむしろ低下するという結果を得ている。これらの結果を総合して、Glu85Lys変異体の感度と特異性向上は、変異に伴う電荷変化が原因ではなく、VH断片が変異によって安定化したため実現した可能性があると考察している。

第3章では、前章で得られたOS選択法をより発展させ、より多様性を高めたライブラリから各種ステロイドに結合する抗体断片を得る試みについて述べられている。具体的には抗E2抗体VHの3つの相補性決定領域CDRのアミノ酸配列をSidhuらにより報告されている(Tyr/Ser)残基によりランダム化し、これをファージ提示したライブラリから5種類のステロイドを抗原として前章と同様3ラウンドのOS選択を行い、結合能が向上したクローンの濃縮に成功したと述べている。

第4章では、第2章の結果の考察を深め、さらに検出能を向上させるための網羅的なアミノ酸変異導入とその結果について述べられている。具体的には、第2章で見られた85位に加え、別の変異体で見られた63位にも網羅的にアミノ酸変異を導入し、それぞれの変異の影響を検討している。さらに、両者を同時に変異させたライブラリをOS選択に用いることで、更なる検出感度向上のための変異体の取得を試みた実験結果について述べられている。

第5章では研究全体の総括と、今後の展望が述べられている。

以上、本論文は、抗原キャリア複合体を用いない新規ファージ選択法による、ハプテン認識抗体の性能向上に成功し、さらに安定化した抗体断片を選択することで結果的にこれらの抗体の特異性を向上できる可能性を示したものであり、その成果はケミカルバイオエンジニアリング分野、特に汎用分子認識素子である抗体の蛋白質工学、ならびにこれを用いた分析化学の進展に寄与するところ大である。

よって本論文は博士(工学)の学位請求論文として合格と認められる。