Article pubs.acs.org/ac
Biosensing Probe for Quality Control Monitoring of the Structural Integrity of Therapeutic Antibodies Hideki Watanabe,† Seiki Yageta,†,‡ Hiroshi Imamura,† and Shinya Honda*,†,‡ †
Biomedical Research Institute, the National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba, Ibaraki 305-8566, Japan ‡ Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, the University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8562, Japan S Supporting Information *
ABSTRACT: Ideal quality control of therapeutic antibodies involves analytical techniques with high-sensitivity, highresolution, and high-throughput performance. Few technologies that assess the physicochemical heterogeneity of antibodies, however, meet all the required demands. We developed a biosensing method for the quality control of therapeutic antibodies based on an artificial protein, AF.2A1, which discriminates between the native and the non-native threedimensional structures of immunoglobulin G (IgG). AF.2A1 specifically recognized non-native IgG spiked into a solution of native IgG, thereby making it possible to detect contamination by a small amount of non-native IgG, which is difficult using conventional size-based separation or spectroscopic techniques. Using AF.2A1 as an analytical probe, we determined the concentration of non-native IgG formed under various pH conditions. The probe was also applicable to accelerated tests of the long-term stability of a therapeutic antibody, allowing monitoring of the formation of non-native IgG at elevated temperatures and extended periods of storage. AF.2A1, a proteinous probe, can be combined with established methods or devices to achieve high-throughput assays and provides the potential for probe-based biosensing for the quality control of therapeutic antibodies.
T
Ideal quality control involves analytical techniques with highsensitivity, high-resolution, and high-throughput performance. Conventional physicochemical characterization techniques for therapeutic proteins include spectroscopy (e.g., nuclear magnetic resonance (NMR), circular dichroism (CD), fluorescence and infrared (IR)) and size-based separation (e.g., size exclusion chromatography (SEC), dynamic light scattering (DLS), field-flow-fractionation (FFF), and analytical ultracentrifugation (AUC)).9 Although all these methods give useful information based on their measurement principles, each has methodological limitations. Spectroscopy provides information on higher-order structure, but with relatively lowsensitivity and low-throughput performance for routinely performed quality control. Size-based separation detects product heterogeneity resulting from multimerization or degradation, but provides little or no information on structural differences in molecules of similar size. Few techniques can detect structural changes with high-sensitivity and highthroughput performance, but such a technique can be realized by employing an analytical probe that specifically recognizes non-native structures of a protein. To date, chemical probes for
herapeutic antibodies have increasing biopharmaceutical application for treating challenging diseases such as cancer, autoimmune diseases, and infectious diseases, as evidenced by their growing market.1 The high potency of an antibody is attributed to the higher-order structure and complex function of the protein: target specificity is determined by hypervariable loops with defined three-dimensional structures positioned on variable domains, and efficacy is further enhanced by effector functions such as antibodydependent, cell-mediated cytotoxicity (ADCC) or complement-dependent cytotoxicity (CDC) mediated by the Fc region. Due to their complex structures, therapeutic antibodies require strict quality control of both the intermediates and the final drug products to ensure both efficacy and safety. Proteins are generally physicochemically unstable and prone to adopt denatured and aggregated states upon exposure to chemical or physical stress,2−4 and the resulting heterogeneity of the drug could cause not only lower efficacy, but also adverse effects such as undesirable immunogenic response.5,6 Consequently, the structural heterogeneity of therapeutic proteins significantly impacts their efficacy and safety. Regulatory authorities have identified several product-related factors that affect the immunogenicity of therapeutic proteins, such as post-translational modifications, higher-order structural changes, and protein aggregation.7,8 © XXXX American Chemical Society
Received: July 3, 2016 Accepted: September 27, 2016
A
DOI: 10.1021/acs.analchem.6b02526 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
approach over conventional technologies, we discuss the usefulness of probe-based biosensing for the quality control of therapeutic antibodies.
detecting protein aggregates have been developed such as an extrinsic fluorescent dye that binds noncovalently to exposed hydrophobic surfaces of proteins.10−13 These probes are useful for high-throughput assays for manufacturing and formulation development of therapeutic antibodies. Biosensing is a powerful analytical technique that is highly superior to other techniques in both sensitivity, resolution, and throughput performance.14 High affinity molecular interactions enable sensitive detection of a small amount of a target in crude samples. Furthermore, some protein binders specifically recognize the conformation, or higher-order structure, of their targets.15,16 Indeed, several proteins and synthetic peptides show binding to acid-induced aggregation-prone IgG.17,18 This potential prompted an idea for the quality control of therapeutic antibodies: an analytical approach based on a protein probe that discriminates between native and nonnative three-dimensional structures. Non-native structures in an antibody are induced by physical or chemical stress and often result in alternatively folded states (AFSs). AFSs form ordered secondary structures that are stable but different from the native structure.19−22 The characteristic structure of a nonnatively folded antibody can be a promising target for specific biosensing. We previously generated a 25-residue artificial protein named AF.2A1 that exhibits nanomolar affinity for immunoglobulin G (IgG).23 AF.2A1 exhibits exquisite conformational specificity and binds to non-native IgG generated by chemical or physical stress (here, non-native IgG is defined as one that loses the native tertiary structure regardless of its aggregation state, i.e., monomer or multimer), but not to the native IgG.23 Non-native IgG is recognized by AF.2A1 mainly through non-native structure of the Fc region (Supporting Information, Figure S1). The specificity of AF.2A1 to nonnative IgG was verified by a pulldown assay23 and surface plasmon resonance assays (Supporting Information, Figure S2). We focused on this specificity to apply AF.2A1 as a biosensing probe for detecting non-native IgG. In this paper, we report an analytical approach based on AF.2A1 for the quality control of therapeutic antibodies (Figure 1). We demonstrate that AF.2A1 is a useful probe for the sensitive detection of small amounts of non-native IgG currently undetectable by conventional methods. Based on the technological advantages of this
■
EXPERIMENTAL PROCEDURES Preparation of Non-Native IgG. Non-native IgG was generated by acidification as follows. Humanized IgG1 was dissolved in 50 mM Tris-HCl, 150 mM NaCl, pH 7.4, at a concentration of 10 μM, and dialyzed against 10 mM glycineHCl, 150 mM NaCl, pH 2.0, for 14 h. The acid-treated IgG was neutralized by dialysis against 50 mM Tris-HCl, 150 mM NaCl, pH 7.4. The acid-treated IgG was prepared just before analysis due to its instability and propensity to aggregate during longterm storage. Surface Plasmon Resonance. Surface plasmon resonance (SPR) assays were performed with a Biacore T100 (GE Healthcare). Chemically synthesized AF.2A1 was purchased from Bio-Synthesis, Inc. 3000 RU of AF.2A1 was immobilized on a Series S CM5 sensor chip (GE Healthcare) via the primary amine groups. Another flow cell was blocked with 1 M ethanolamine to provide a reference sensor chip. Binding assays were performed in HBS-T (10 mM HEPES, 150 mM NaCl, 0.05% (w/v) Tween20, pH 7.4). Dilution series of IgG samples spiked with non-native IgG were prepared by spiking native IgG with acid-treated IgG. To avoid overestimating nonspecific adsorption of non-native IgG, nonspecific responses to the reference sensor chip were subtracted from binding responses to the AF.2A1-immobilized sensor chip. Native IgG solution contained a small amount of non-native IgG. To distinguish between this intrinsically contained non-native IgG and the acid-treated IgG, the sensorgram for native IgG solution (the black dotted curve in Figure 2) was subtracted from each observed sensorgram so as to not overestimate the bulk effect and intrinsic non-native IgG. The obtained sensorgrams were processed using Biacore T100 evaluation software (GE Healthcare). Size Exclusion Chromatography for the Detection of IgG Aggregates. SEC was performed in HBS-T buffer using a Superdex 200 (5/150) column equipped with an AKTA purifier (GE Healthcare). The obtained chromatogram at 280 nm was processed using UNICORN version 5.2 (GE Healthcare) to calculate peak areas. The running buffer and test samples were identical to those used for the SPR assay described above. Quantitative Measurement of Non-Native IgG Monomer by Calibration-Free Concentration Analysis. Calibration-free concentration analysis (CFCA) was performed with a Biacore T100. 3000 RU of chemically synthesized AF.2A1_Q5R was immobilized on a CM5 sensor chip. CFCA was performed using each of the following buffers: HBS-T (10 mM HEPES, 150 mM NaCl, 0.05% Tween20, pH 7.4), MBS-T (10 mM MES, 150 mM NaCl, 0.05% Tween20, pH 6.0), ABST (10 mM acetate, 150 mM NaCl, 0.05% Tween20, pH 5.0), ABS-T (10 mM acetate, 500 mM NaCl, 0.05% Tween20, pH 4.5), and ABS-T (10 mM acetate, 500 mM NaCl, 0.05% Tween20, pH 4.0). IgG monomer dissolved in these buffers was fractionated using a Superdex 200 (10/300) column (GE Healthcare) prior to CFCA. Streptococcal protein G B1 domain used for control experiments was prepared as previously described.24 The obtained data were processed using Biacore T100 Evaluation software (GE Healthcare) to determine the active concentration to the ligand. The percent concentration of the active form was calculated by dividing the active
Figure 1. Schematic view of biosensing for the quality control of therapeutic antibodies. The biosensing involves the use of a specific probe, AF.2A1, which exhibits a binding response to non-native IgG (red curve), but not to native IgG (green curve). B
DOI: 10.1021/acs.analchem.6b02526 Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
■
Article
RESULTS
Detection of Non-Native IgG Spiked into a Solution of Native IgG. An ideal probe must specifically detect its target in crude samples, so it was necessary to verify that AF.2A1 can specifically detect a small amount of non-native IgG present in a solution of native IgG. We evaluated the sensitivity of AF.2A1 to detect non-native IgG spiked into a solution of native IgG. Non-native IgG is often generated upon acidification, a condition commonly used for elution during purification or for virus clearance during antibody manufacturing. Acid-treated IgG was used as a model for non-native IgG and was prepared by dialyzing a humanized monoclonal IgG1 against an acidic buffer (pH 2.0), followed by neutralization. A dilution series of spiked IgG solutions was then prepared that contained both native and non-native IgG. A surface plasmon resonance (SPR) assay was performed using an AF.2A1-immobilized sensor chip to detect the non-native IgG. The SPR results demonstrated a binding response that dose-dependently increased with an increase in the acid-treated IgG (Figure 2a). The limit of detection (LOD), defined here as the amount providing
