Polyethylene glycol (PEG) is among the most commonly used chemicals in protein purification processes. Its use ranges from precipitating agent
], phase forming component in liquid-liquid extraction
], displacer in hydrophobic interaction chromatography
] to potential additive in protein formulation
] and drug modifying agent
]. Despite its frequent application, the exact nature of the polymer solvent interactions as well as the structural dynamics in solution governing its role in protein purification processes remain unclear.
One approach to understand the nature of molecules on an atomic scale are molecular dynamics (MD) simulations. MD simulations have proven a useful tool to understand the binding and elution behavior of proteins in ion exchange chromatography
]. Its results could be correlated to experimental data and thus used predictively. Besides HTS process development methodologies, MD simulations represent another promising approach to speed up process development in protein purification.
Several molecular dynamics studies have been performed focusing on the tertiary structure of polyethylene glycol or polyethylene oxide in aqueous solutions. Lee et al.
] used a revision of the CHARMM ether force field (C35)
] to establish a molecular understanding of the relation between hydrodynamic radius and radius of gyration of PEG, which differs in behavior from polymer theory under certain conditions. They extended their study in 2009
] by using a coarse grained model and were thus able to extend the simulated timespan from 20 to 800 ns. These studies were again focused on investigating the hydrodynamic properties of the polymer molecules, their hydrodynamic radius in relation to their radius of gyration and the coherence of the simulation results with polymer theory. One major goal was to understand - on a molecular scale - the influence of the three dimensional structure of polymer molecules in solution on macroscopic, hydrodynamic properties of these solutions e.g. viscosity. Other studies, such as the one conducted by Borodin et al.
] are geared towards polymer dynamics of pure polymers and the effect of added water. One main goal of this study was again to understand the physicochemical properties of polymer solutions such as viscosity.
Tasaki et al.
performed MD simulation of a single PEG molecule of 722 Da in 1996
] focused more on the secondary structure and polymer-water interaction. While their study yielded interesting insights into the 3D structure of single PEG molecules in solution, the advance in computational power since their publication now allows for longer simulations of larger PEG molecules. Another, more recent, MD simulation study using a PEG-derivate focused the importance of choosing a proper force field
] to obtain secondary structure results in agreement with experimental data. One force field, not tested in their study, but commonly used for protein simulations, is the amber03 force field
]. While this force field is designated to proteins, a self parameterizing algorithm can be applied to adjust for non proteinaceous molecules
]. The advantage of using a force field designated to proteins is that combined MD simulations of proteins and PEG could be readily run, once it is established that the self parameterizing algorithm yield reasonable results for PEG. In order to verify the structural data resulting from such MD simulations, experimentally determined structures are needed. There have been several reports of 3D structural data of PEG
] with the recent addition of a high quality crystallographic record
As stated above, MD simulations focused on the secondary structure of single PEG molecules of fixed molecular weight as well as studies concentrating on the three dimensional structure and its implications for macroscopic solution properties have been conducted. Simulations of larger PEG molecules and an investigation of the effects of polymer molecular weight on properties of PEG such as secondary structure, surface hydrophobicity and H-bond network with the solvent are lacking. These properties most likely govern the characteristics of PEG molecules in protein purification, thus their deeper understanding on a molecular scale would be beneficial. PEG polymer molecular weight influences the effect PEG has on protein solubility or aqueous two-phase formation. To understand the influence of molecular weight of PEG molecules on their structural and surface characteristics on an atomic scale would thus be valuable.
This study’s first aim is to confirm the validity of all-atom molecular dynamics simulations of PEG in water based on the amber03 force field by comparing the resulting structure to the recently published crystallographic data and previous MD results. Results from 10-30 ns long MD simulations of PEG molecules ranging in molecular weight from 300 Da to 3500 Da are shown. This study puts a focus on the effect PEG chain length has on geometric parameters, secondary structure, polymer surface characteristics and polymer solvent interactions. We aim to understand the influence on PEG molecular weight on polymer properties important for its use in protein purification. Correlations to experimental data on the phase forming behavior of 4 different PEGs in the presence of phosphate are given. Correlation between solvent polarity determined using a solvatochromic dye to hydrophobicity determined via MD simulations are shown.