Article pubs.acs.org/ac
One-Shot In Vitro Evolution Generated an Antibody Fragment for Testing Urinary Cotinine with More Than 40-Fold Enhanced Affinity Hiroyuki Oyama, Izumi Morita, Yuki Kiguchi, Erika Banzono, Kasumi Ishii, Satoshi Kubo, Yoshiro Watanabe, Anna Hirai, Chiaki Kaede, Mitsuhiro Ohta, and Norihiro Kobayashi* Kobe Pharmaceutical University, 4-19-1, Motoyama-Kitamachi, Higashinada-ku, Kobe 658-8558, Japan S Supporting Information *
ABSTRACT: Immunoassays for cotinine, a major nicotine metabolite, in the urine are useful for monitoring the degree of tobacco smoke exposure. However, hybridoma-based anti-cotinine antibodies lack sufficient binding affinity to perform practically sensitive measurements, and thus most cotinine assays still rely on polyclonal antibodies. Here, we describe the generation of a mutant single-chain Fv fragment (scFv) that was used in an enzyme-linked immunosorbent assay (ELISA) to determine urinary cotinine levels in passive smokers. A “wild-type” scFv (scFv-wt) with a Ka value of 2.7 × 107 M−1 (at 4 °C) was prepared by linking the VH and VL domains in a mouse anti-cotinine antibody. “Oneshot” random mutagenesis on the scFv-wt gene by error-prone PCR generated mutant scFv genes, which were expressed on phage particles. Repeated panning directed toward mutants with slower off-rates selected scFv clones that showed improved sensitivity in an ELISA system. One of these mutants (scFv#m1-54) with five amino acid substitutions showed more than a 40-fold enhanced Ka (1.2 × 109 M−1 at 4 °C) and, thus, was used to monitor human urinary cotinine. A limited amount of soluble scFv was reacted with urine specimens (or cotinine standards) at 4 °C for 120 min in microwells on which cotinine residues had been immobilized. The midpoint of the dose−response curves under optimized conditions (0.27 ng/assay) was more than 100-fold lower than the ELISA results obtained using scFv-wt. The limit of detection (8.4 pg/assay) corresponded to 0.17 ng/mL urinary cotinine, which was satisfactorily low for testing the threshold levels for passive smoke exposure. The assay values for volunteers correlated with the values determined using a commercial assay kit. This study evidently showed the potential of a molecular breeding approach, in which simple in vitro evolution might generate superior antibody reagents as cloned proteins, overcoming the limited molecular diversity inherent to conventional immunization-based antibodies.
T
produced continuously as reagent-grade proteins with welldefined biochemical and binding properties. Generating monoclonal antibodies against cotinine with satisfactory performance, however, was only reported once 30 years ago.24 Indeed, a patent report highlighted the difficulty in generating anti-cotinine monoclonal antibodies by indicating that they had no success even after repeated attempts over the course of 15 years.25 Insufficient affinity for practical use should mainly be caused by the extremely low molecular mass of cotinine (Mr 176.2)26,27 and the limited number of polar functional groups. Over the last several years, we have studied in vitro affinity maturation of antibody fragments, particularly those targeting small biomarkers,28−30 and have succeeded in generating a single-chain Fv fragment (scFv) against estradiol-17β with an over 150-fold improved equilibrium affinity constant (Ka) via 3 iterative cycles of mutagenesis and selection.29,30 Therefore,
he association between tobacco smoke exposure and an increased risk of lung cancer,1,2 cardiovascular diseases,3−5 and respiratory diseases6 is often recognized. Consequently, it is important to develop feasible and reliable methods to estimate the degree of tobacco smoke exposure, not only due to active smoking but also to passive smoke exposure. The urinary levels of cotinine [(S)-(−)-cotinine], a major nicotine metabolite (Figure. 1),7,8 serve as a suitable index for tobacco smoke exposure.9,10 This is because the elimination half-life (∼20 h) of cotinine is much longer than that of nicotine (∼2 h),8,10,11 causing urinary cotinine levels to be much higher than the levels of nicotine, and to be less fluctuating than cotinine levels in blood.12,13 Thus, immunoassays based on specific anti-cotinine antibodies should be useful for sensitive, reliable, and noninvasive monitoring of tobacco smoke exposure.14,15 In fact, several immunoassay systems that measure the cotinine levels have been developed to date, 14−23 and some of them are commercially available. However, most of these assays19−21,23 still rely upon rabbit polyclonal anti-cotinine antibodies despite evident advantages of cloned antibodies, which can be © XXXX American Chemical Society
Received: November 4, 2016 Accepted: December 1, 2016
A
DOI: 10.1021/acs.analchem.6b04332 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
into the pEXmide 5 vector,29 and transformed into Escherichia coli (E. coli) XL1-Blue cells (Stratagene). Transformants were grown, and induced with isopropyl β-D-thiogalactopyranoside and sucrose.29−32 Periplasmic extracts containing scFv-wt proteins were prepared by the osmotic shock method29−32 and used for ELISA without further purification. Mutagenesis of the scFv Gene and Phage Display. Error-prone PCR33,34 was performed as previously described,29−32 with or without 0.50 mM MnCl2. The pEXmide 5 containing the scFv-wt gene (1 ng) was mixed in the buffer solution33,34 (100 μL) with the CT#45VH-Rev and CT#45VHFor (or CT#45VL-Rev and CT#45VL-For) primers (0.10 nmol each), AmpliTaq DNA polymerase (Applied Biosystems; 5 U), and 0.10 μmol of each dNTP except dATP (0.020 μmol). This mixture was amplified for 25 cycles at 94 (1 min), 50 (1 min), and 70 °C (3 min), followed by a 10 min extension at 70 °C. The mutated VH and VL genes (the products obtained with and without 0.50 mM MnCl2 were combined) were randomly spliced to produce scFv genes, assembled in the order 5′-VHlinker-VL-FLAG-3′, which were then ligated into pEXmide 5 and transformed into XL1-Blue cells. The resulting bacterial library was used to rescue phage particles.29−32 Phages were partially purified by polyethylene glycol-precipitation and examined as described below. Selection of the Cotinine-Binding Phage Clones. MaxiSorp polystyrene tubes (12 × 75 mm; Nunc) were coated by overnight incubation at room temperature with a 100 μg/ mL CT−BSA solution in CB (1.0 mL) and blocked with MPBS-2.31 In each round of panning, phages [∼1 × 1012 colony forming unit (cfu)] in M-PBS-2 (1 mL) were added to the coated tubes, which were incubated at 37 °C for 1 h with continuous tumbling. Tubes were then washed three times with T-PBS-2 (4 mL), and bound phages were eluted by the addition of cotinine solution (1.0 mL) with different conditions according to the round of panning (described in the Supporting Information). The recovered phage suspension was added to log-phase XL1-Blue cells in a 2 × YT medium containing 10 μg/mL tetracycline (9 mL) and then incubated at 37 °C for 30 min. After centrifugation (1800 × g for 20 min), the pellet was suspended in 2 × YT medium (0.1 mL) and spread on a 90-φ plate containing 2 × YT agar supplemented with 100 μg/mL ampicillin, 10 μg/mL tetracycline, and 10 g/L glucose, and incubated overnight at 37 °C. Colonies were then scraped into 2 × YT medium containing 15% glycerol (1.5 mL) and a small aliquot was used for phage rescue. The resulting phages were used in the next round of selection. Preparation, Purification, and Analysis of Binding Parameters of Soluble (Non-Phage-Linked) scFvs. Recombinant plasmids were extracted from phage-infected bacterial clones and scFv genes therein were PCR amplified with primers CT#45VH-Rev and CT#45VL-For-stop (Table S1) to add stop codons to the 3′-termini.29,30 The products were gel-purified, ligated into pEXmide 5, and transformed into XL1-Blue cells. Transformants were grown, induced as described above, and soluble scFvs in the periplasmic extracts were used in ELISA without further purifications. scFv-wt and scFv#m1-54 were purified by affinity chromatography using anti-FLAG-M2 agarose (Sigma) and then HiTrap Protein L (GE Healthcare).35 The affinity and dissociation rate constants (ka and kd) of these scFvs against the immobilized cotinine residues were determined at 4 and 37 °C using Biacore T-200 surface plasmon resonance (SPR) sensor (GE Healthcare), equipped with the CM5 sensor chip that had been immobilized
Figure 1. Schematic representation of the metabolic pathway of nicotine in humans.
improving the affinity of anti-cotinine antibodies, which has long been difficult with hybridoma-based methods, was an attractive and challenging subject for our group. Success in this endeavor would highlight the potential of the “antibodybreeding” approach30 in overcoming the limitations of producing animal-derived antibodies. Here, we describe the generation of a desirable scFv clone via “one-shot” in vitro evolution. One of these scFvs showed more than 40-fold enhanced Ka against cotinine compared with the prototype (wild-type) scFv and enabled generation of a sensitive enzyme-linked immunosorbent assay (ELISA) for testing urinary cotinine levels due to passive tobacco smoke exposure.
■
MATERIALS AND METHODS Cotinine and Related Materials. Cotinine, related compounds, and conjugate of cotinine and bovine serum albumin (CT−BSA; cotinine/albumin coupling molar ratio: 24) used were those described previously.15 Primers. Single-stranded oligonucleotides used as PCR primers were synthesized and purified by Tsukuba Oligo Service. The sequences of these primers are shown in Table S1 in the Supporting Information. Buffers. The buffers used in this study, abbreviated as PB, PBS, G-PBS, T-PBS, M-PBS, PVG-PBS, PBS-2, M-PBS-2, TPBS-2, and CB were previously described29 and are reviewed in detail in the Supporting Information. Preparation of the Wild-Type scFv. The VH and VL genes of the mouse anti-cotinine antibody Ab-CT#45 (IgG1κ) were separately amplified using first-strand cDNAs.15,29 The primers used were CT#45VH-Rev and CT#45VH-For (for amplifying VH) and CT#45VL-Rev and CT#45VL-For (for amplifying VL) (Table S1 in the Supporting Information). The amplified products were gel-purified and spliced by overlap-extension PCR to provide a gene encoding wild-type scFv (scFv-wt), in which the VH and VL genes were combined via a linker sequence encoding (GlyGlyGlyGlySer)3, and a FLAG sequence was added at the 3′-end.29−32 The scFv-wt gene was then subcloned B
DOI: 10.1021/acs.analchem.6b04332 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
step was directed at the complementarity-determining regions (CDRs),29 while the latter two mutagenesis steps were targeted the entire variable regions.29,30 The results suggested that a few point mutations in the framework regions (FRs) could dramatically enhance the affinity. Thus, we expected that desirable anti-cotinine mutant scFvs could be isolated after random mutagenesis of the entire variable regions. A single error-prone PCR was performed to amplify the VH and VL regions of the scFv-wt gene separately: the template gene, scFv-wt, was prepared by joining the VH and VL genes for Ab-CT#45.15 The mutated VH and VL genes, once obtained, were spliced in a shuffling manner to generate an scFv gene library,29,30 which was then transformed into XL1-Blue cells. We obtained a bacterial library involving ∼2 × 107 scFv-positive transformants. Phage particles were rescued from this library, and subjected to three rounds of panning, strategy of which was intended not to miss specific phages with extremely slower kd as well as practically fast ka (procedure and outcome are described in the Supporting Information). After the final round, 176 phage clones were screened for their cotinine-binding abilities in the competitive ELISA, which was probed with an enzyme-labeled anti-M13 phage antibody.34 Three phage clones (#m1-17, #m1-54, and #m1-106) showed higher sensitivities (data not shown) and, thus, relevant soluble scFvs (scFv#m1-17, #m1-54, and #m1-106) were prepared from them. Structure of Improved scFvs. The primary structures of scFv-wt and the mutant scFvs are shown in Figure 2A. The VH and VL sequences of these scFvs were classified as belonging to subgroup IIA and V, respectively, based on the Kabat definition.36 In scFv#m1-17, scFv#m1-54, and scFv#m1-106, 4, 5, and 5 amino acids were substituted, respectively. The YH50N substitution [meaning that tyrosine (Y) that had been present at the position 50 in the VH of the scFv-wt was substituted with asparagine (N)] at the N-terminus of VHCDR2 (numbering of amino acids and CDR positions were defined based on the Kabat system36) occurred commonly in these mutants. Notably, asparagine occasionally appears at the position H50 of mouse VH, which belongs to subgroup IIA.36 Furthermore, this substitution obeys the propensity observed for in vivo affinity maturation: namely, at the antigen-binding interface in paratopes, tyrosine is one of the residues that tend to decrease during the somatic hypermutation process, while asparagine is one of the residues that tend to be introduced.37,38 Substitutions were also found in VH-CDR1 (scFv#m1-17 and scFv#m1-106), VH-CDR2 (scFv#m1-54), and VL-CDR1 (scFv#m1-106); however, they were not found in VH- and VL-CDR3s, which usually play a particularly important role in antigen binding. scFv#m1-17 and scFv#m1-106 had substitution(s) in their linker sequences. On sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels, affinity-purified soluble scFv-wt and scFv#m1-54, the latter of which was selected as the best mutant (see below), migrated as single bands with nearly the expected relative molecular masses (Mr) of 27208 and 27255, respectively (Figure 2B). Minor bands due to scFv-dimers or oligomers were not observed, suggesting that these scFvs have little tendency to aggregate. Based on the amino acid sequences, protein ribbon structures for scFv#m1-54 (docked to cotinine) were constructed in silico (Figure 3).39,40 These modeling views suggested that less than 10 amino acids in the paratope might directly contact cotinine, which is markedly lower than the average number of antigen-contacting amino
with CT−BSA. scFvs to be analyzed were run on the sensor chip (30 μL/min) as solutions (0.25−8.0 μM) in G-PBS to prevent aggregation or adsorption of the proteins. ELISA for Screening scFv-Displaying Phages and scFv Characterization. The 96-well microplates (#3590, Costar) were coated by overnight incubation at room temperature with 1.0 μg/mL CT−BSA in CB (100 μL/well) and blocked with Block Ace (Dainippon Pharmaceutical) as previously described.29 After washing the wells three times with T-PBS, the soluble scFv proteins (used as crude periplasmic extract) (100 μL) and cotinine standard (cotinine monoperchlorate was used)15 or related compounds (50.0 μL), both of which were dissolved in PVG-PBS, were added, mixed, and incubated at 4 or 37 °C for 60, 120, or 240 min. The solutions were aspirated and the wells were washed similarly, and then a 5.0 μg/mL solution of anti-FLAG M2 antibody labeled with peroxidase (POD) (Sigma) in PVG-PBS was added (100 μL/well) and incubated at 37 °C for 30 min. The wells were washed and the bound POD activity was determined colorimetrically using ophenylenediamine as a hydrogen donor.15,32 Measurement of Human Urinary Cotinine Levels with the Optimized ELISA. Blank urine (cotinine was not detectable, that is,
