Nonnative protein aggregation is a common challenge during biopharmaceutical product development. Protein aggregates are often observed during protein expression, purification, and fill-finish operations, as well as during product storage prior to use. Aggregates may be soluble or insoluble, and the former are often further categorized as low or high molecular weight (MW). Nonnative aggregates of all types are potentially problematic with respect to regulatory approval of a product, due to purity and safety requirements. To better predict and control aggregation under a variety of conditions it would be beneficial to have a general experimental and modeling approach to efficiently and accurately assess the underlying kinetics and their dependence on a range of experimental variables, such as pH, temperature, protein concentration, and excipient type / concentration. This challenge is made more difficult by limitations of many standard experimental techniques to detect and monitor one or more key intermediates in multi-stage aggregation pathways, and outstanding questions regarding which techniques, if any, are best suited for monitoring aggregation.

The work here uses alpha-chymotrypsinogen A (aCgn) as a model protein that readily undergoes nonnative aggregation under accelerated conditions [1]. aCgn is shown here to display a range of different aggregate morphologies and kinetic behaviors that depend primarily on sample pH and electrolyte concentration. Extensive, multivariate data sets for aggregate formation kinetics, structure, morphology, and (soluble) aggregate size distributions were obtained as function of temperature, pH, [NaCl], [sucrose], and protein concentration. The approach includes: (1) statistical design of experiments and multivariate data analysis; (2) quantitative mechanistic kinetic modeling with a Lumry-Eyring Nucleated Polymerization treatment that was extended here to explicitly account for aggregate-aggregate condensation steps; (3) semi-quantitative characterization of aggregate size, structure, and morphology. Together, (1) to (3) provide means to efficiently interpolate and predict qualitative and quantitative differences in aggregate characteristics and changes in the underlying mechanism of aggregation across a wide range of experimental variables. The results also clearly illustrate the potential power of combining in-line multi-angle static laser light scattering with size-exclusion chromatography for systems with high-MW aggregates. The experimental and modeling framework further facilitates categorizing and discriminating among a variety of existing aggregation models.

[1] Andrews JM, Roberts CJ. Biochemistry 46 7558-7571 (2007); Weiss WF IV, Hodgdon TK, Kaler EW, Lenhoff AM, Roberts CJ. Biophys. J. 93 4392-4403 (2007); Andrews JM, Weiss WF IV, Roberts CJ. Biochemistry 47 2397-2403 (2008).