Relative to other extrinsic factors, the effects of hydrodynamic flow fields on protein stability and conformation remain poorly understood. Flow-induced protein remodeling and/or aggregation is observed both in Nature and during the large-scale industrial manufacture of proteins. Despite its ubiquity, the relationships between the type and magnitude of hydrodynamic flow, a protein's structure and stability, and the resultant aggregation propensity are unclear. Here, we assess the effects of a defined and quantified flow field dominated by extensional flow on the aggregation of BSA, β 2 -microglobulin (β 2 m), granulocyte colony stimulating factor (G-CSF), and three monoclonal antibodies (mAbs). We show that the device induces protein aggregation after exposure to an extensional flow field for 0.36-1.8 ms, at concentrations as low as 0.5 mg mL −1 . In addition, we reveal that the extent of aggregation depends on the applied strain rate and the concentration, structural scaffold, and sequence of the protein. Finally we demonstrate the in situ labeling of a buried cysteine residue in BSA during extensional stress. Together, these data indicate that an extensional flow readily unfolds thermodynamically and kinetically stable proteins, exposing previously sequestered sequences whose aggregation propensity determines the probability and extent of aggregation.extensional flow | aggregation | unfolding | bioprocessing | antibody P roteins are dynamic and metastable and consequently have conformations that are highly sensitive to the environment (1). Over the last 50 y the effect of changes in temperature, pH, and the concentration of kosmatropic/chaotropic agents on the conformational energy landscape of proteins has become well understood (1). This, in turn, has allowed a link to be established between the partial or full unfolding of proteins and their propensity to aggregate (2). The force applied onto a protein as a consequence of hydrodynamic flow has also been observed to trigger protein aggregation and has fundamental (3), medical (4), and industrial relevance, especially in the manufacture of biopharmaceuticals (5-8). Although a wealth of studies have been performed (7,(9)(10)(11)(12)(13), no consensus has emerged on the ability of hydrodynamic flow to induce protein aggregation (7,14,15). This is due to the wide variety of proteins used (ranging from lysozyme, BSA, and alcohol dehydrogenase to IgGs), differences in the type of flow field generated (e.g., shear, extensional, or mixtures of these), and to the presence or absence of an interface (16). A shearing flow field (Fig. 1A, Top) is caused by a gradient in velocity perpendicular to the direction of travel and is characterized by the shear rate (s −1 ). This results in a weak rotating motion of a protein alongside translation in the direction of the flow. An extensional flow field (Fig. 1A, Bottom) is generated by a gradient in velocity in the direction of travel and is characterized by the strain rate (s −1 ). A protein in this type of flow would experien...

No abstract

No abstract

Monoclonal antibodies (mAbs) currently dominate the biopharmaceutical sector due to their potency and efficacy against a range of disease targets. These proteinaceous therapeutics are, however, susceptible to unfolding, mis‐folding, and aggregation by environmental perturbations. Aggregation thus poses an enormous challenge to biopharmaceutical development, production, formulation, and storage. Hydrodynamic forces have also been linked to aggregation, but the ability of different flow fields (e.g., shear and extensional flow) to trigger aggregation has remained unclear. To address this question, we previously developed a device that allows the degree of extensional flow to be controlled. Using this device we demonstrated that mAbs are particularly sensitive to the force exerted as a result of this flow‐field. Here, to investigate the utility of this device to bio‐process/biopharmaceutical development, we quantify the effects of the flow field and protein concentration on the aggregation of three mAbs. We show that the response surface of mAbs is distinct from that of bovine serum albumin (BSA) and also that mAbs of similar sequence display diverse sensitivity to hydrodynamic flow. Finally, we show that flow‐induced aggregation of each mAb is ameliorated by different buffers, opening up the possibility of using the device as a formulation tool. Perturbation of the native state by extensional flow may thus allow identification of aggregation‐resistant mAb candidates, their bio‐process parameters and formulation to be optimized earlier in the drug‐discovery pipeline using sub‐milligram quantities of material.

Large, multi-dimensional spatio-temporal datasets are omnipresent in modern science and engineering. An effective framework for handling such data are Gaussian process deep generative models (GP-DGMs), which employ GP priors over the latent variables of DGMs. Existing approaches for performing inference in GP-DGMs do not support sparse GP approximations based on inducing points, which are essential for the computational efficiency of GPs, nor do they handle missing data -a natural occurrence in many spatio-temporal datasets -in a principled manner. We address these shortcomings with the development of the sparse Gaussian process variational autoencoder (SGP-VAE), characterised by the use of partial inference networks for parameterising sparse GP approximations. Leveraging the benefits of amortised variational inference, the SGP-VAE enables inference in multi-output sparse GPs on previously unobserved data with no additional training. The SGP-VAE is evaluated in a variety of experiments where it outperforms alternative approaches including multi-output GPs and structured VAEs.