Correcting Ultraviolet-Visible Spectra for Baseline Artifacts.

Rayleigh and Mie light scattering from particulates, soluble protein aggregates, or large proteins can lead to inaccuracy of concentration measurements using ultraviolet (UV) spectroscopy and Beer's Law. While a number of light scattering correction equations have been proposed in the literature, they can also lead to incorrect values if samples vary in particulate and/or soluble aggregate levels or depart in other ways from which the equations were developed. We propose a curve-fitting baseline subtraction approach based on fundamental Rayleigh and Mie scattering equations which also factors in instrument baseline artifacts. We validated this Rayleigh-Mie correction against a wide variety of positive and negative controls, including protein size standards, protein aggregates induced by forced degradation, lentivirus and polystyrene nanospheres.

Universal Qualification of Analytical Procedures for Characterization and Control of Biologics.

A diverse set of analytical tools is required to characterize the complex structural properties of biopharmaceutical products and to ensure their quality, stability, safety, and efficacy. It is generally necessary to demonstrate that such tools are capable of measuring one or more intended attribute(s) of the product with a desired degree of precision, accuracy, linearity, specificity and sensitivity. Here we present a general framework upon which experiments may be designed to establish analytical procedure performance, predicated on the hypothesis that many analytical procedures have universal performance characteristics - that is, the validity of the measured result is a function of the measurement system and data characteristics and is not a function of the specific analyte being measured. Using simulated data, we demonstrate that the generalized approach improves the scientific validity of resulting descriptions of procedure performance by reducing the incidence of false failures and missed faults during future use of the procedure. Broad adoption of these principles will facilitate an improved understanding of procedure performance characteristics while requiring fewer human resources for procedure qualification studies.

Determining Spectroscopic Quantitation Limits for Misfolded Structures.

Protein secondary structures are frequently assessed using infrared and circular dichroism spectroscopies during drug development (e.g., during product comparability and biosimilarity studies, reference standard characterization, etc.) However, there is little information on the lower limits of quantitation of structural misfolds and impurities for these methods. A model system using a monoclonal antibody reference material was spiked at various levels with a protein that had a significantly different secondary structure to represent the presence of a stable and discreet structural misfold. The ability of circular dichroism, transmission Fourier transform infrared spectroscopy and microfluidic modulation spectroscopy, along with various spectral comparison algorithms, were assessed for their ability to detect the presence and quantify the amount of the misfolded structure.

Guidance to Achieve Accurate Aggregate Quantitation in Biopharmaceuticals by SV-AUC.

The levels and types of aggregates present in protein biopharmaceuticals must be assessed during all stages of product development, manufacturing, and storage of the finished product. Routine monitoring of aggregate levels in biopharmaceuticals is typically achieved by size exclusion chromatography (SEC) due to its high precision, speed, robustness, and simplicity to operate. However, SEC is error prone and requires careful method development to ensure accuracy of reported aggregate levels. Sedimentation velocity analytical ultracentrifugation (SV-AUC) is an orthogonal technique that can be used to measure protein aggregation without many of the potential inaccuracies of SEC. In this chapter, we discuss applications of SV-AUC during biopharmaceutical development and how characteristics of the technique make it better suited for some applications than others. We then discuss the elements of a comprehensive analytical control strategy for SV-AUC. Successful implementation of these analytical control elements ensures that SV-AUC provides continued value over the long time frames necessary to bring biopharmaceuticals to market.

Precision of protein aggregation measurements by sedimentation velocity analytical ultracentrifugation in biopharmaceutical applications.

Sedimentation velocity analytical ultracentrifugation (SV-AUC) is routinely applied in biopharmaceutical development to measure levels of protein aggregation in protein products. SV-AUC is free from many limitations intrinsic to size exclusion chromatography (SEC) such as mobile phase and column interaction effects on protein self-association. Despite these clear advantages, SV-AUC exhibits lower precision measurements than corresponding measurements by SEC. The precision of SV-AUC is influenced by numerous factors, including sample characteristics, cell alignment, centerpiece quality, and data analysis approaches. In this study, we evaluate the precision of SV-AUC in its current practice utilizing a multilaboratory, multiproduct intermediate precision study. We then explore experimental approaches to improve SV-AUC measurement precision, with emphasis on utilization of high quality centerpieces.

Detection of protein aggregates by sedimentation velocity analytical ultracentrifugation (SV-AUC): sources of variability and their relative importance.

Sedimentation velocity analytical ultracentrifugation (SV-AUC) has found application in the biopharmaceutical industry as a method of detecting and quantifying protein aggregates. While the technique offers several advantages (i.e., matrix-free separation and minimal sample handling), its results exhibit a high degree of variability relative to orthogonal size-sensitive separation techniques such as size exclusion chromatography (SEC). The goal of this work is to characterize and quantify the sources of variability that affect SV-AUC results, particularly size distributions for a monoclonal antibody monomer/dimer system. Contributions of individual factors to the overall variability are examined. Results demonstrate that alignment of sample cells to the center of rotation is the most significant contributing factor to overall variability. The relative importance of other factors (e.g., temperature equilibration, time-invariant noise, meniscus misplacement, etc.) are quantified and discussed.

Common excipients impair detection of protein aggregates during sedimentation velocity analytical ultracentrifugation.

The final formulations of modern pharmaceutical protein products typically contain sugars or sugar alcohols as stabilizers. Migration of these sugars under the influence of an applied gravitational field during sedimentation velocity analytical ultracentrifugation (SV-AUC) produces dynamic density and viscosity gradients. If the formation of such gradients is not taken into account during data analysis, the capability of the SV-AUC technique to detect protein oligomers/aggregates may be dramatically impacted. In the example described here, the limit of quantitation (LOQ) of a simulated monoclonal antibody (mAb) dimer increases from 0.8% to 2.4% upon addition of 5% sorbitol to the formulation. This study uses simulated and experimental SV-AUC data to demonstrate the detrimental effect of dynamic gradients; it further explores how sophisticated data analysis techniques, including SEDFIT's inhomogeneous solvent options, may be used to mitigate the detection problems caused by the sedimentation of excipients.

Measurement of the second osmotic virial coefficient for protein solutions exhibiting monomer-dimer equilibrium.

The second osmotic virial coefficient (B) is a measure of solution nonideality that is useful for predicting conditions favorable for protein crystallization and for inhibition of aggregation. Static light scattering is the technique most commonly used to determine B values, typically using protein concentrations less than 5 mg/mL. During static light scattering experiments at low protein concentrations, frequently the protein is assumed to exist either as a single nonassociating species or as a combination of assembly states independent of protein concentration. In the work described here, we examined the limit for ignoring weak reversible dimerization (Kd > or =1 mM) by comparing B values calculated with and without accounting for self-association. Light scattering effects for equilibrium dimer systems with Kd <20 mM and Kd <1 mM will significantly affect apparent B values measured for 20 and 150-kDa proteins, respectively. To interpret correctly light scattering data for monomer-dimer equilibrium systems, we use an expanded coefficient model to account for separate monomer-monomer (B(22)), monomer-dimer (B(23)), and dimer-dimer (B(33)) interactions.

High concentration formulations of recombinant human interleukin-1 receptor antagonist: I. Physical characterization.

At relatively high protein concentrations (i.e., up to 100 mg/mL), recombinant human interleukin-1 receptor antagonist (rhIL-1ra) was found to exist in a monomer-dimer equilibrium controlled by solution ionic strength. Sedimentation equilibrium at 25 degrees C was used to measure the increase in the dimer dissociation constant (K(d)) as a function of ionic strength. K(d) increased from 2.0 to 12.6 mM as the solution ionic strength was increased from 0.011 to 0.184 molal. These K(d) values were used with both static light scattering and membrane osmometry data collected over a protein concentration range of 1-100 mg/mL to determine second osmotic virial coefficients. Expanding the second osmotic virial coefficient model to account for separate monomer-monomer (B(22)), monomer-dimer (B(23)), and dimer-dimer (B(33)) interactions reveals net monomer-dimer interactions are attractive, whereas the others are repulsive. Lastly, isothermal titration calorimetry dilution experiments showed that rhIL-1ra dimerization is enthalpically driven (DeltaH(dimerization) << 0), which is consistent with intermolecular cation-pi interactions previously proposed as the monomer-monomer contact sites in dimers.

High concentration formulations of recombinant human interleukin-1 receptor antagonist: II. Aggregation kinetics.

At high protein concentrations (i.e., 50-100 mg/mL) and 37 degrees C, low solution ionic strength accelerates aggregation of recombinant human interleukin-1 receptor antagonist (rhIL-1ra). We have used a variety of physical and spectroscopic techniques to explain this observation. A population balance model was applied to a continuous mixed suspension, mixed product removal (MSMPR) reactor at steady-state to determine aggregate nucleation and growth rates. Nucleation rates increase at low ionic strength, while growth rates are unaffected. At low rhIL-1ra concentrations (i.e., <1 mg/mL), no conformational changes or differences in free energies of unfolding (DeltaG(unf)) were observed at 37 degrees C over the solution ionic strength range of 0.025-0.184 molal used for aggregation studies. However, increasing the protein concentration to 100 mg/mL shifts the rhIL-1ra monomer-dimer equilibrium significantly at low ionic strength to favor dimerization, which is reflected in subtle conformational changes in the circular dichroism and second-derivative FTIR spectra. In addition to a reversible dimer, an irreversible dimer forms by second-order kinetics during incubation at 37 degrees C. This noncovalent dimer does not significantly participate in further aggregation. The loss of native protein due to aggregation at 37 degrees C was third order in protein thermodynamic activity due to the rate-limiting formation of an aggregation-prone trimer. This trimer forms from irreversible attractive monomer-reversible dimer interactions, which were quantified using second osmotic cross virial coefficients. Lastly, the activity coefficient of rhIL-1ra estimated from aggregation rates is 50% higher at 100 mg/mL protein concentration than at 50 mg/mL, in close agreement with predictions from a hard-sphere model for activity coefficients.

Protein misfolding and aggregation.

Interest in the problem of protein misfolding and aggregation has exploded in recent years for two reasons: (1) the sharp rise in the number and volume of therapeutic proteins produced commercially and (2) the recognition of the central role of protein aggregates in degenerative diseases. The systematic study of protein aggregation presents major challenges to both the experimentalist and the theoretician. Much of the work retains an empirical flavor due to the experimental complexities; the sensitivity of protein aggregation to the slightest change in protein amino acid composition, solvent properties, or protein concentration; and the lack of robust theoretical models of misfolding and aggregation. Novel experimental and computational approaches are being developed, and we anticipate substantial progress will be made in the near future. Several presentations describing the latest advances in protein misfolding and aggregation were given at the American Chemical Society meeting (BIOT division) held in September, 2006 in San Francisco.

Sedimentation velocity analytical ultracentrifugation and SEDFIT/c(s): limits of quantitation for a monoclonal antibody system.

Sedimentation velocity analytical ultracentrifugation (SV-AUC) has emerged in the biopharmaceutical industry as a technique to detect small quantities of protein aggregates. However, the limits of detection and quantitation of these aggregates are not yet well understood. Although diverse factors (molecule, instrument, technique, and software dependent) preclude an all-encompassing measurement of these limits for the complete system, it is possible to use simulated data to determine the quantitation limits of the data analysis software aspect. The current study examines the performance of the SEDFIT/c(s) data analysis tool with simulated antibody monomer/dimer and monomer/aggregate systems. Under completely ideal conditions (zero noise, known meniscus, and shape factor homogeneity), the software limit of quantitation was 0.01% for the monomer/aggregate system and 0.03% for the less well-resolved monomer/dimer system. Under more realistic conditions (0.005 OD root mean square [RMS] noise, shape factor variability, and long solution column), the software limits of quantitation were 0.2 and 0.6% (0.002 and 0.006 OD) for the monomer/aggregate and monomer/dimer systems, respectively. Interestingly, diminished quantitation accuracy at very low levels of oligomer was not accompanied by deterioration of fit quality (as measured by root mean square deviation [RMSD] and residuals bitmap images).

Population balance modeling of aggregation kinetics of recombinant human interleukin-1 receptor antagonist.

The kinetics of benzyl alcohol-induced nonnative aggregation of recombinant human interleukin-1 receptor antagonist (rhIL-1ra) were investigated using a population balance model. Steady-state size distributions of rhIL-1ra aggregates formed in a continuous mixed suspension, mixed product removal (MSMPR) reactor were measured and used to extrapolate aggregate nucleation and growth rates parameters. Aggregate growth rate was size-dependent and a linear growth rate model was used to derive a population density function. Addition of 0.9 wt/v% benzyl alcohol increased the nucleation rate by approximately four orders of magnitude. The growth rate for aggregates, however, changed little as a function of benzyl alcohol concentration in the range of 0-0.9%. The addition of sucrose to buffer containing 0.9% benzyl alcohol decreased rhIL1-ra nucleation rate by orders of magnitude and had little impact on growth rate kinetics. The simplicity of the population balance model and the physical relevance of the information obtained from this model render it a useful tool to study protein aggregation kinetics and the effects of excipients on this process.

Protein-solute interactions affect the outcome of ultrafiltration/diafiltration operations.

Protein production operations often involve a final diafiltration of the protein into formulation buffer. For several Amgen product proteins, post-diafiltration assays revealed a significant difference in molar excipient concentrations on the retentate and the permeate side of the membrane. For example, post-diafiltration assays of formulated 200 mg/mL human interleukin-1 receptor antagonist showed molar chloride concentrations up to 30% lower than those of the diafiltration buffer. Deviations from expected results were also observed in cases where a fusion conjugate protein (AMG-719) was formulated by dialysis in 10 mM acetate and where PEGylated soluble tumor necrosis factor receptor (PEG-sTNF-RI) was formulated in 270 mM glycine and 10 mM histidine. Classical thermodynamic theory describing intermolecular interactions predicts that the partitioning of small solutes during dialysis will be dependent on the protein concentration, charge, and surface area. This study illustrates methods to approximate these effects using readily available protein data (theoretical titration curves based on protein sequence, density information, etc.). Additionally, guidelines are provided to determine when intermolecular interactions are likely to significantly impact the outcome of dialysis/diafiltration operations.

Effects of sucrose on conformational equilibria and fluctuations within the native-state ensemble of proteins.

Osmolytes increase the thermodynamic conformational stability of proteins, shifting the equilibrium between native and denatured states to favor the native state. However, their effects on conformational equilibria within native-state ensembles of proteins remain controversial. We investigated the effects of sucrose, a model osmolyte, on conformational equilibria and fluctuations within the native-state ensembles of bovine pancreatic ribonuclease A and S and horse heart cytochrome c. In the presence of sucrose, the far- and near-UV circular dichroism spectra of all three native proteins were slightly altered and indicated that the sugar shifted the native-state ensemble toward species with more ordered, compact conformations, without detectable changes in secondary structural contents. Thermodynamic stability of the proteins, as measured by guanidine HCl-induced unfolding, increased in proportion to sucrose concentration. Native-state hydrogen exchange (HX) studies monitored by infrared spectroscopy showed that addition of 1 M sucrose reduced average HX rate constants at all degrees of exchange of the proteins, for which comparison could be made in the presence and absence of sucrose. Sucrose also increased the exchange-resistant core regions of the proteins. A coupling factor analysis relating the free energy of HX to the free energy of unfolding showed that sucrose had greater effects on large-scale than on small-scale fluctuations. These results indicate that the presence of sucrose shifts the conformational equilibria toward the most compact protein species within native-state ensembles, which can be explained by preferential exclusion of sucrose from the protein surface.

Roles of conformational stability and colloidal stability in the aggregation of recombinant human granulocyte colony-stimulating factor.

We studied the non-native aggregation of recombinant human granulocyte stimulating factor (rhGCSF) in solution conditions where native rhGCSF is both conformationally stable compared to its unfolded state and at concentrations well below its solubility limit. Aggregation of rhGCSF first involves the perturbation of its native structure to form a structurally expanded transition state, followed by assembly process to form an irreversible aggregate. The energy barriers of the two steps are reflected in the experimentally measured values of free energy of unfolding (DeltaG(unf)) and osmotic second virial coefficient (B(22)), respectively. Under solution conditions where rhGCSF conformational stability dominates (i.e., large DeltaG(unf) and negative B(22)), the first step is rate-limiting, and increasing DeltaG(unf) (e.g., by the addition of sucrose) decreases aggregation. In solutions where colloidal stability is high (i.e., large and positive B(22) values) the second step is rate-limiting, and solution conditions (e.g., low pH and low ionic strength) that increase repulsive interactions between protein molecules are effective at reducing aggregation. rhGCSF aggregation is thus controlled by both conformational stability and colloidal stability, and depending on the solution conditions, either could be rate-limiting.

Oxidative dimer formation is the critical rate-limiting step for Parkinson's disease alpha-synuclein fibrillogenesis.

Intraneuronal deposition of alpha-synuclein as fibrils and oxidative stress are both implicated in the pathogenesis of Parkinson's disease. We found that the critical rate-limiting step in nucleation of alpha-synuclein fibrils under physiological conditions is the oxidative formation and accumulation of a dimeric, dityrosine cross-linked prenucleus. Dimer formation is accelerated for the pathogenic A30P and A53T mutant alpha-synucleins, because of their greater propensity to self-interact, which is reflected in the smaller values of the osmotic second virial coefficient compared to that of wild-type synuclein. Our finding that oxidation is an essential step in alpha-synuclein aggregation supports a mechanism of Parkinson's disease pathogenesis in which the separately studied pathogenic factors of oxidative stress and alpha-synuclein aggregation converge at the critical step of alpha-synuclein dimer formation.