Quantitation of soluble aggregates by sedimentation velocity analytical ultracentrifugation using an optical alignment system - Aspects of method validation.

Sedimentation velocity analytical ultracentrifugation (SV-AUC) is routinely used for quantitation of soluble aggregates as an orthogonal technique to size-exclusion chromatography (SEC). SV-AUC presents many advantages over the SEC, yet lower precision of aggregate quantitation by SV-AUC often complicates comparison between aggregate values generated by these techniques and subsequent decision making. In an earlier report, we described the development of an optical alignment (OA) system and evaluated the intermediate precision of aggregate quantitation offered by the OA. Here, we determine the limit of detection (LOD) and limit of quantitation (LOQ) which can be achieved with the OA. For a common setup using three cells, the improvement lent by the OA system is almost 2.5-fold compared to the earlier reported limits. In addition, we estimate the contribution of the fitting variability and compare options to further increase the precision of aggregate quantitation by SV-AUC.

An optical alignment system improves precision of soluble aggregate quantitation by sedimentation velocity analytical ultracentrifugation.

Appropriate characterization of soluble aggregates is an important aspect of biologics development and manufacturing, and sedimentation velocity analytical ultracentrifugation (SV-AUC) is often used an orthogonal technique to size-exclusion chromatography (SEC) for this purpose. Precise quantification of low levels of soluble aggregates by SV-AUC can be adversely impacted by improper cell alignment. This report describes the development of an optical system capable of quantifying cell alignment that affords a substantial improvement compared to historical approaches.

Estimating analytical variability in two-dimensional data.

Throughout the course of drug development there are many instances in which a variability assessment within a set of analytical data is required, which may be challenging for techniques that produce two-dimensional data. This note describes an interval-based approach to variability assessment and demonstrates its applicability for analysis of near-UV circular dichroism (CD) spectra. The approach is generalizable and could be applied to two-dimensional data from other analytical techniques as well.

High-throughput analysis of concentration-dependent antibody self-association.

Monoclonal antibodies are typically monomeric and nonviscous at low concentrations, yet they display highly variable associative and viscous behavior at elevated concentrations. Although measurements of antibody self-association are critical for understanding this complex behavior, traditional biophysical methods are not capable of characterizing such concentration-dependent self-association in a high-throughput manner. Here we describe a nanoparticle-based method, termed self-interaction nanoparticle spectroscopy, that is capable of rapidly measuring concentration-dependent self-interactions for three human monoclonal antibodies with unique solution behaviors. We demonstrate that gold nanoparticles conjugated with antibodies at low protein concentrations (<40 µg/mL) display self-association behavior (as measured by the interparticle distance-dependent plasmon wavelength) that is well correlated with static light-scattering measurements obtained at three orders of magnitude higher antibody concentrations. Using this methodology, we find that the antibodies display a complex pH-dependent self-association behavior that is strongly influenced by the solution ionic strength. Importantly, we find that a polyclonal human antibody is nonassociative for all solution conditions evaluated in this work, suggesting that antibody self-association is more specific than previously realized. We expect that our findings will guide rational manipulation of antibody phase behavior, and enable studies that elucidate sequence and structural determinants of antibody self-association.

Computational design and biophysical characterization of aggregation-resistant point mutations for γD crystallin illustrate a balance of conformational stability and intrinsic aggregation propensity.

γD crystallin is a natively monomeric eye-lens protein that is associated with hereditary juvenile cataract formation. It is an attractive model system as a multidomain Greek-key protein that aggregates through partially folded intermediates. Point mutations M69Q and S130P were used to test (1) whether the protein-design algorithm RosettaDesign would successfully predict mutants that are resistant to aggregation when combined with informatic sequence-based predictors of peptide aggregation propensity and (2) how the mutations affected relative unfolding free energies (ΔΔG(un)) and intrinsic aggregation propensity (IAP). M69Q was predicted to have ΔΔG(un) ≫ 0, without significantly affecting IAP. S130P was predicted to have ΔΔG(un) ∼ 0 but with reduced IAP. The stability, conformation, and aggregation kinetics in acidic solution were experimentally characterized and compared for the variants and wild-type (WT) protein using circular dichroism and intrinsic fluorescence spectroscopy, calorimetric and chemical unfolding, thioflavin-T binding, chromatography, static laser light scattering, and kinetic modeling. Monomer secondary and tertiary structures of both variants were indistinguishable from WT, while ΔΔG(un) > 0 for M69Q and ΔΔG(un) < 0 for S130P. Surprisingly, despite being the least conformationally stable, S130P was the most resistant to aggregation, indicating a significant decrease of its IAP compared to WT and M69Q.

Molecular level insights into thermally induced α-chymotrypsinogen A amyloid aggregation mechanism and semiflexible protofibril morphology.

Understanding nonnative protein aggregation is critical not only to a number of amyloidosis disorders but also for the development of effective and safe biopharmaceuticals. In a series of previous studies [Weiss et al. (2007) Biophys. J. 93, 4392-4403; Andrews et al. (2007) Biochemistry 46, 7558-7571; Andrews et al. (2008) Biochemistry 47, 2397-2403], α-chymotrypsinogen A (aCgn) and bovine granulocyte colony stimulating factor (bG-CSF) have been shown to exhibit the kinetic and morphological features of other nonnative aggregating proteins at low pH and ionic strength. In this study, we investigated the structural mechanism of aCgn aggregation. The resultant aCgn aggregates were found to be soluble and exhibited semiflexible filamentous aggregate morphology under transmission electron microscopy. In addition, the filamentous aggregates were demonstrated to possess amyloid characteristics by both Congo red binding and X-ray diffraction. Peptide level hydrogen exchange (HX) analysis suggested that a buried native ß-sheet comprised of three peptide segments (39-46, 51-64, and 106-114) reorganizes into the cross-ß amyloid core of aCgn aggregates and that at least ∼50% of the sequence adopts a disordered structure in the aggregates. Furthermore, the equimolar, bimodal HX labeling distribution observed for three reported peptides (65-102, 160-180, and 229-245) suggested a heterogeneous assembly of two molecular conformations in aCgn aggregates. This demonstrates that extended ß-sheet interactions typical of the amyloid are sufficiently strong that a relatively small fraction of polypeptide sequence can drive formation of filamentous aggregates even under conditions favoring colloidal stability.

Understanding the Increased Aggregation Propensity of a Light-Exposed IgG1 Monoclonal Antibody Using Hydrogen Exchange Mass Spectrometry, Biophysical Characterization, and Structural Analysis.

This work compares the conformational stability, backbone flexibility, and aggregation propensity of monomer and dimer fractions of an IgG1 monoclonal antibody (mAb) generated on UVA light exposure for up to 72 h collected by preparative size-exclusion chromatography, compared with unstressed control. UVA light exposure induced covalent aggregation, and fragmentation as measured by size-exclusion chromatography, sodium dodecyl sulfate polyacrylamide gel electrophoresis, and extensive oxidation of specific methionine residues (Met 257, Met 433, and Met 109) in both size fractions identified by reverse phase chromatography coupled to mass spectrometry. Compared with unstressed mAb, both the monomer and dimer fractionated from 72 h UVA light-exposed mAb had decreased thermal melting temperatures (Tm1) by 1.4°C as measured by differential scanning calorimetry, minor changes in tertiary structure as measured by near-UV CD, increased monomer loss, and aggregation on accelerated storage at 35°C. Hydrogen/deuterium exchange mass spectrometry identified local segments with increased flexibility in CH2 and CH3 domains of both size fractions, and decreased flexibility in few segments of Fab and CH1 domains in the dimer fraction. Segment 247-256 in heavy chain, an established aggregation hotspot in IgG1 mAbs had large increase in flexibility in both size fractions compared with unstressed mAb.

Nonnative protein polymers: structure, morphology, and relation to nucleation and growth.

Thermally induced aggregates of alpha-chymotrypsinogen A and bovine granulocyte-colony stimulating factor in acidic solutions were characterized by a combination of static and dynamic light scattering, spectroscopy, transmission electron microscopy, and monomer loss kinetics. The resulting soluble, high-molecular weight aggregates (approximately 10(3)-10(5) kDa) are linear, semiflexible polymer chains that do not appreciably associate with one another under the conditions at which they were formed, with classic power-law scaling of the radius of gyration and hydrodynamic radius with weight-average molecular weight (M(w)). Aggregates in both systems are composed of nonnative monomers with elevated levels of beta-sheet secondary structure, and bind thioflavine T. In general, the aggregate size distributions showed low polydispersity by light scattering. Together with the inverse scaling of M(w) with protein concentration, the results clearly indicate that aggregation proceeds via nucleated (chain) polymerization. For alpha-chymotrypsinogen A, the scaling behavior is combined with the kinetics of aggregation to deduce separate values for the characteristic timescales for nucleation (tau(n)) and growth (tau(g)), as well as the stoichiometry of the nucleus (x). The analysis illustrates a general procedure to noninvasively and quantitatively determine tau(n), tau(g), and x for soluble (chain polymer) aggregates, as well as the relationship between tau(n)/tau(g) and aggregate M(w).

Variability in Flow-Imaging Microscopy Measurements and Considerations for Biopharmaceutical Development.

Flow-imaging microscopy is widely used in the biopharmaceutical industry to characterize populations of subvisible (1-100 µm) particles due to high sensitivity and the ability to discriminate different particle morphologies. The present work provides a comprehensive assessment of the capabilities of flow-imaging microscopy by exploring the impacts of a variety of factors on the observed variability of these measurements. A novel graphical presentation is proposed to facilitate both determination of expected levels and detection of potential atypical results. Data collected across different products and container-closure systems illustrate that a substantial amount of historical experience is typically required to adequately define the expected levels of subvisible particles for any specific system. It is also shown, however, that an appropriate level of control can be demonstrated without the need to pool large numbers of containers or perform replicate measurements.

Technical Decision Making With Higher Order Structure Data: Perspectives on Higher Order Structure Characterization From the Biopharmaceutical Industry.

Characterization of the higher order structure (HOS) of protein-based biopharmaceutical products is an important aspect of their development. Opinions vary about how best to apply biophysical methods, in which contexts to use these methods, and how to use the resulting data to make technical decisions as drug candidates are commercialized [Gabrielson JP, Weiss WF IV. J Pharm Sci. 2015;104(4):1240-1245]. The aim of this commentary is to provide guidance for the development and implementation of a robust and comprehensive HOS characterization strategy. We first consider important concepts involved in developing a strategy that is appropriately suited to a particular biologic, and then discuss ways industry can partner with academia, technology companies, government laboratories, and regulatory agencies to improve the consistency with which HOS characterization is applied across the biopharmaceutical industry.

Technical decision-making with higher order structure data: starting a new dialogue.

Characterization of the higher order structure (HOS) of biological products has been growing in importance in recent years. Scientists in the biopharmaceutical industry, academic researchers, and regulators are all increasingly aware of the critical role that HOS plays in maintaining the stability and intended biological function of biopharmaceutical products. We organized a consortium of scientists and researchers from industry and academic institutions to address how HOS data can be used most effectively to drive decisions during product development. In this commentary, we introduce the purpose, objectives, and scope of the consortium and then provide some brief points to consider in the context of characterizing HOS of biopharmaceutical products. Scientific advances in HOS analysis, as well as continued dialogue among academia, industry, and regulatory agencies will ensure that appropriate methodologies are used to inform technical decision-making during biopharmaceutical development.

Technical decision-making with higher order structure data: specific binding of a nonionic detergent perturbs higher order structure of a therapeutic monoclonal antibody.

Robust higher order structure (HOS) characterization capability and strategy are critical throughout biopharmaceutical development from initial candidate selection and formulation screening to process optimization and manufacturing. This case study describes the utility of several orthogonal HOS methods as investigational tools during purification process development. An atypically high level of residual detergent in a development drug substance batch of a therapeutic monoclonal antibody triggered a root cause investigation. Several orthogonal biophysical techniques were used to uncover and characterize a specific interaction between the detergent and the antibody. Isothermal titration calorimetry (ITC) was used to quantify the molar ratio and affinity of the binding event, and circular dichroism (CD) spectroscopy and differential scanning calorimetry (DSC) were used to evaluate corresponding impacts on secondary/tertiary structure and thermal stability, respectively. As detergents are used routinely in biopharmaceutical processing, this case study highlights the value and power of HOS data in informing technical investigations and underlines the importance of HOS characterization as a component of overall biopharmaceutical analytical control strategy.

Arrhenius time-scaled least squares: a simple, robust approach to accelerated stability data analysis for bioproducts.

Defining a suitable product presentation with an acceptable stability profile over its intended shelf-life is one of the principal challenges in bioproduct development. Accelerated stability studies are routinely used as a tool to better understand long-term stability. Data analysis often employs an overall mass action kinetics description for the degradation and the Arrhenius relationship to capture the temperature dependence of the observed rate constant. To improve predictive accuracy and precision, the current work proposes a least-squares estimation approach with a single nonlinear covariate and uses a polynomial to describe the change in a product attribute with respect to time. The approach, which will be referred to as Arrhenius time-scaled (ATS) least squares, enables accurate, precise predictions to be achieved for degradation profiles commonly encountered during bioproduct development. A Monte Carlo study is conducted to compare the proposed approach with the common method of least-squares estimation on the logarithmic form of the Arrhenius equation and nonlinear estimation of a first-order model. The ATS least squares method accommodates a range of degradation profiles, provides a simple and intuitive approach for data presentation, and can be implemented with ease.

Reduction of the C191-C220 disulfide of α-chymotrypsinogen A reduces nucleation barriers for aggregation.

Proper disulfide formation can be essential for the conformational stability of natively folded proteins. For proteins that must unfold in order to aggregate, disruption of native disulfides may therefore promote aggregation. This study characterizes differences in the aggregation process for wild-type (WT) α-chymostrypsinogen A (aCgn) and the same molecule with one of its native disulfides (C191-C220) reduced to free thiols (aCgnSH) at acidic pH, where WT aCgn forms semi-flexible amyloid polymers. Loss of the disulfide leads to no discernable differences in folded monomer secondary or tertiary structure based on circular dichroism (CD) or intrinsic fluorescence (FL), and causes a small decrease in the free energy change upon unfolding. After unfolding-mediated aggregation, the resulting amyloid morphology and structure are similar or indistinguishable for aCgn and aCgnSH by CD, FL, ThT binding, multi-angle laser light scattering, and transmission electron microscopy. Aggregates of aCgn and aCgnSH are also able to cross-seed with monomers of the other species. However, aggregates of aCgnSH are more resistive than aCgn aggregates to urea-mediated dissociation, suggesting some degree of structural differences in the aggregated species that was not resolvable in detail without higher resolution methods. Mechanistic analyses of aggregation kinetics indicate that the initiation or nucleation of new aggregates from aCgnSH involves a mono-molecular rate limiting step, possibly the unfolding step. In contrast, that for aCgn involves an oligomeric intermediate, suggesting native disulfide linkages help to hinder non-native protein aggregation by providing conformational barriers to key nucleation event(s).

Aggregation and pH-temperature phase behavior for aggregates of an IgG2 antibody.

Monomer unfolding and thermally accelerated aggregation kinetics to produce soluble oligomers or insoluble macroscopic aggregates were characterized as a function of pH for an IgG2 antibody using differential scanning calorimetry (DSC) and size-exclusion chromatography (SEC). Aggregate size was quantified via laser light scattering, and aggregate solubility via turbidity and visual inspection. Interestingly, nonnative oligomers were soluble at pH 5.5 above approximately 15°C, but converted reversibly to visible/insoluble particles at lower temperatures. Lower pH values yielded only soluble aggregates, whereas higher pH resulted in insoluble aggregates, regardless of the solution temperature. Unlike the growing body of literature that supports the three-endotherm model of IgG1 unfolding in DSC, the results here also illustrate limitations of that model for other monoclonal antibodies. Comparison of DSC with monomer loss (via SEC) from samples during thermal scanning indicates that the least conformationally stable domain is not the most aggregation prone, and that a number of the domains remain intact within the constituent monomers of the resulting aggregates. This highlights continued challenges with predicting a priori which domain(s) or thermal transition(s) is(are) most relevant for product stability with respect to aggregation.

Characterization of high-molecular-weight nonnative aggregates and aggregation kinetics by size exclusion chromatography with inline multi-angle laser light scattering.

Size exclusion chromatography with an inline multi-angle light scattering detector (SEC-MALS) was assessed as a means to characterize and monitor the formation of soluble, high-molecular-weight (HMW) protein aggregates so as to better quantify and model nonnative aggregation kinetics. Assay configuration and robustness were tested with respect to sample preparation, column type, and assay variability. Independent comparison of SEC-MALS with batch light scattering analysis indicates good agreement between the two methods. Weight-average molecular weight (M(w)), radius of gyration (R(g)), apparent polydispersity, and mass fraction monomer (m) together are shown to provide qualitative and quantitative experimental signatures to distinguish high-MW aggregate growth via chain polymerization versus that via aggregate-aggregate condensation. Mechanistic treatment of aggregation kinetics monitored by SEC-MALS is illustrated by data regression using a recently developed Lumry-Eyring Nucleated Polymerization model that explicitly accounts for aggregate nucleation and competing growth via chain- and condensation-polymerization. The combination of time-dependent M(w) and m data are shown to provide a convenient and robust means to separate and quantify characteristic time scales or rate coefficients for concurrent stages of irreversible aggregation. In addition, the scaling of R(g) with M(w) for HMW aggregates provides useful insights into aggregate morphology and mechanisms of aggregate growth.

Principles, approaches, and challenges for predicting protein aggregation rates and shelf life.

Control and prevention of unwanted aggregation for therapeutic proteins is a ubiquitous hurdle during biopharmaceutical product manufacture, storage, shipping, and administration. Methods to predict the relative or absolute rates of aggregation are therefore of great practical interest in biopharmaceutical research and development. Aggregation is often well-described as a multi-stage process involving unfolding or misfolding of free monomers, along with one or more assembly steps to form soluble or insoluble oligomers or higher-molecular-weight species. This report reviews the current state of the art in experimental and practical theoretical approaches that attempt to predict in vitro protein aggregation rates or propensities relevant to pharmaceutical proteins. Most available approaches fall within four primary categories. The principles and assumptions underlying each category are reviewed, along with advantages and limitations in each case. The importance of appropriate experimental techniques and models to probe and quantify the thermodynamics and/or dynamics of multiple steps or stages within the overall aggregation process is stressed. The primary focus is on aggregation in solution, relevant to parenteral dosage forms. Additional challenges are briefly reviewed.

Nucleation, growth, and activation energies for seeded and unseeded aggregation of alpha-chymotrypsinogen A.

The intrinsic time scales for nonnative aggregate nucleation (tau0(n)) and chain growth (tau0(g)) were determined for alpha-chymotrypsinogen A as a function of temperature under acidic conditions where the resulting aggregates do not appreciably condense. Previous results (Andrews and Roberts (2007) Biochemistry 46, 7558) indicated that the product tau0(n)tau0(g) increases with increasing temperature but could not distinguish tau0(n) and tau0(g). Separate experimental values of tau0(n) and tau0(g) are reported here from two approaches based on either (i) combining unseeded monomer loss kinetics with static light scattering of the resulting aggregates or (ii) seeded monomer loss kinetics as a function of number concentration of seed. Values of tau0(n) and tau0(g) from (i) and (ii) agree quantitatively, and indicate that nucleation has a large, negative effective activation energy (ca. -76 kcal/mol) while growth has at most a weak dependence on temperature. The results are consistent with a model in which nucleation requires significant conformational changes within a nonnative oligomer, beyond those for monomer unfolding. The results more generally illustrate the potential utility of approaches (i) and (ii) for quantitatively determining in vitro tau0(n) and tau0(g) values, as well as how the effects of seeding can be predicted purely from unseeded kinetics and static light scattering measurements prior to significant aggregate condensation.