Towards rational design of aggregation-resistant protein variants
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Proteins have been used for therapeutic purposes since the late 19th century, and as of 2017, hundreds of therapeutic peptides and proteins had been approved by US-FDA for clinical use. This major class of therapeutics has some shortcomings, however, and addressing these shortcomings can reduce costs and timeline for drug development and commercialization and aid the sector reach its full potential. Protein instability, in vitro and in vivo, is one of these shortcomings, and one major form of protein instability is non-native protein aggregation. This phenomenon can jeopardize drug efficacy and patient safety; therefore, aggregation has to be strictly controlled, as also mandated by regulatory administrations. ☐ One method of controlling and precluding aggregation is disrupting the aggregation-prone regions (APRs) on a peptide or a protein. One goal of this dissertation was to generate protein variants with higher aggregation resistance, by eliminating APR/s and decreasing intrinsic aggregation propensity (IAP). IAP is aggregation propensity of a protein when its APRs are surface exposed. In connection with this first goal, another goal was to investigate APR/s, IAP, conformational stability (or preservation of the native state conformation, when otherwise was not feasible) and the relation of this entity and properties with aggregation rates. To these goals, a rational design strategy was tested on two proteins from different families and with different secondary structures: human granulocyte colony-stimulating factor (hG-CSF) and 4-4-20 single-chain variable fragment (scFv). As part of this strategy, a combination of an APR consensus prediction approach and a relative conformational stability predictor, PyRosetta, was implemented. Accordingly, a subset of the variants that were predicted to decrease IAP without significantly affecting conformational stability were chosen for biophysical characterization. Variants with lower IAP than and comparable conformational stability to wild-type (WT) were expected to demonstrate lower aggregation rates. ☐ Selected hG-CSF variants, L83G, Y84A and L83A were initially studied with static light scattering (SLS) temperature ramp experiment, whose result showed that the aggregation Tonset ranking was WT ≈ L83A > L83G ≈ Y84A. Therefore, these selected variants did not increase aggregation resistance. In pursuit of an explanation for this result, isothermal guanidine hydrochloride induced unfolding of these proteins was monitored via intrinsic fluorescence. Based on this experiment, the apparent conformational stability of the variants were lower than that of WT, a result that is at least part of the reason that the variants did not increase aggregation resistance of rhG-CSF. In an attempt to compare WT and L83A IAP values, aggregation kinetics of these two proteins were also examined under partially denaturing conditions. This experiment showed that at 3 M GdnHCl, at which both proteins were unfolded to great extents, probably exposing their APRs, they aggregated at comparable rates. Based on this, L83A and WT had comparable IAP values. This experiment laid a foundation for a potential method of IAP isolation and comparison. This experiment also showed that relatively large irreversible WT and L83A aggregates formed under partially denaturing conditions over day to week time scales. ☐ Based on the same rational design strategy, variants that were predicted to lower (or not affect) IAP without affecting conformational stability were also chosen for 4-4-20 scFv: F94N, F94A, F94T, F94L, F94V and Y93F. All these mutations, except Y93F, were predicted to decrease aggregation rates; whereas, Y93F was predicted not to affect it. Unfortunately, the masses and concentrations of these selected variants were too low to study aggregation rates and conformational stability with conventional techniques (denaturant or pressure unfolding monitored via intrinsic protein fluorescence). Instead, WT and the selected variants were studied and compared in terms of their FL binding capacity as a function of pressure. Pressurization experiments at 1.3:1 protein-to-FL ratio showed that WT 4-4-20 scFv that could quench ~ 0.6 of FL fluorescence at 0 kbar could only quench ~ 0.1 of this fluorescence at 3 kbar. This unquenching might result from 4-4-20 scFv unfolding (followed by FL unbinding) or dissociation of the scFv-FL, when scFv is still folded. 0 kbar data of this experiment demonstrated that the selected mutations affected FL binding and thus the native state conformation of 4-4-20 scFv negatively. This is not necessarily enlightening regarding the relative conformational stability of the variants. However, it still shows that the rational design strategy needs improvement, and a predictor for the effects of mutations on ligand binding affinity might be incorporated to this rational design strategy. In addition to WT and variant comparison and limited evaluation of the rational design strategy, thermodynamics of WT 4-4-20 scFv-FL binding was also investigated between 20 ℃ and 37 ℃. This binding was found to be both enthalpically and entropically favorable (ΔH_binding^0 and TΔS_binding^0 were -10 ± 1 kcal/mol and 2 ± 1 kcal/mol, respectively) and enthalpy-driven in this temperature range. Overall, this dissertation demonstrates general challenges in aggregation-resistant protein design, as both APRs/relative IAP and relative conformational stability have to be accurately predicted.
Description
Keywords
Aggregation hot spots/aggregation-prone regions, Conformational stability, Intrinsic aggregation propensity, Non-native protein aggregation, Protein engineering, Rational design