Effect of lysozyme solid-phase PEGylation on reaction kinetics and isoform distribution.

The combination of PEG-protein conjugation and chromatographic separation is generally known as solid-phase or on-column PEGylation and can provide advantages compared to commonly applied batch PEGylation. Even though the concept was already applied by several authors, changes in the isoform distribution compared to liquid-phase PEGylation due to sterically hindered PEGylation sites could not be confirmed. In this manuscript, a method for solid-phase PEGylation experiments in a 96-well plate format, using an automated liquid handling station is described. Applying size exclusion chromatography (SEC) and highly sensitive isoform analytics for mono-PEGylated lysozyme, we were able to investigate the differences in reaction kinetics and isoform distribution between adsorber-based PEGylation and modifications in free solution. Accordingly, solid-phase PEGylation with SP Sepharose FF and XL generally showed a reduced PEGylation reaction. In contrast to the predominant N-terminal PEGylation of lysozyme in liquid phase, a main modification of lys 97 and lys 116 was found for solid-phase experiments, which could be explained by binding orientations on corresponding adsorbent materials. Further experiments with varying amounts of bound protein additionally showed an influence on the isoform distribution of mono-PEGylated lysozyme.

Molecular dynamics simulations approach for the characterization of peptides with respect to hydrophobicity.

It has been shown that molecular dynamics (MD) simulations are a powerful tool to generate knowledge about complex interactions in the field of bioprocess technologies at the atomic level. In this field, one of the most important nonspecific interactions is the hydrophobic interaction, which is still not fully understood after nearly 30 years of research. To date established hydrophobicity scales, which base mostly on proteins' primary structure, are used to estimate the overall hydrophobicity. The structural complexity and the influence of the protein's environment cannot be accommodated with these scales. In this work, free solution molecular dynamics simulations were used to investigate the hydrophobic character of low molecular weight peptides. Therefore, local densities of a small hydrophobic tracer molecule and unprotonated triethylamin (TEA) in particular were used to localize and quantify hydrophobic patches among the peptide surface. Comparisons between local densities and the retention behavior in reversed phase chromatography showed significant correlations. Moreover, neighbor effects caused by charges could be identified. We were able to show that the developed in silico method is applicable to characterize peptides in respect to hydrophobicity in agreement with experimental data. We are confident to apply this method to larger protein structures.

Optimization of random PEGylation reactions by means of high throughput screening.

Since the first FDA approval of a PEGylated product in 1990, so called random PEGylation reactions are still used to increase the efficacy of biopharmaceuticals and represent the major technology of all approved PEG-modified drugs. However, the great influence of process parameters on PEGylation degree and the PEG-binding site results in a lack of reaction specificity which can have severe impact on the product profile. Consequently, reproducible and well characterized processes are essential to meet increasing regulative requirements resulting from the quality-by-design (QbD) initiative, especially for this kind of modification type. In this study we present a general approach which combines the simple chemistry of random PEGylation reactions with high throughput experimentation (HTE) to achieve a well-defined process. Robotic based batch experiments have been established in a 96-well plate format and were analyzed to investigate the influence of different PEGylation conditions for lysozyme as model protein. With common SEC analytics highly reproducible reaction kinetics were measured and a significant influence of PEG-excess, buffer pH, and reaction time could be investigated. Additional mono-PEG-lysozyme analytics showed the impact of varying buffer pH on the isoform distribution, which allowed us to identify optimal process parameters to get a maximum concentration of each isoform. Employing Micrococcus lysodeikticus based activity assays, PEG-lysozyme33 was identified to be the isoform with the highest residual activity, followed by PEG-lysozyme1 . Based on these results, a control space for a PEGylation reaction was defined with respect to an optimal overall volumetric activity of mono-PEG-lysozyme isoform mixtures.

Self-interaction chromatography in pre-packed columns: a critical evaluation of self-interaction chromatography methodology to determine the second virial coefficient.

The characterization of protein-protein interactions is commonly conducted via self-interaction chromatography to describe magnitude and direction of the interactions with the resulting osmotic second virial coefficient (B22). However, the method is invasive and protein immobilization on the adsorber surface can influence the results obtained. In order to replace batch immobilization procedures followed by a column packing, direct on-column preparation was optimized in terms of protein immobilization under a continuous flow. Surface load was measured applying a novel method based on partial least squares analysis of spectral scans to reduce analytical error when determining the amount of immobilized protein. Subsequently influencing parameters such as the effects of absolute surface load, injected protein concentration and distribution of protein orientation were analyzed and system performance evaluated. The results disprove the consistency of the SIC method regarding the non-random orientation of proteins on adsorber particles. Thus the determined B22-values differ quantitatively from those determined with static light scattering. Furthermore, variations in immobilization conditions influence the results obtained. These results make clear that SIC does not fulfill the theoretical framework of B22-analysis. It is rather a qualitative measure of protein-protein interactions in the respective system used for experimentation.

Accurate retention time determination of co-eluting proteins in analytical chromatography by means of spectral data.

Chromatography is the method of choice for the separation of proteins, at both analytical and preparative scale. Orthogonal purification strategies for industrial use can easily be implemented by combining different modes of adsorption. Nevertheless, with flexibility comes the freedom of choice and optimal conditions for consecutive steps need to be identified in a robust and reproducible fashion. One way to address this issue is the use of mathematical models that allow for an in silico process optimization. Although this has been shown to work, model parameter estimation for complex feedstocks becomes the bottleneck in process development. An integral part of parameter assessment is the accurate measurement of retention times in a series of isocratic or gradient elution experiments. As high-resolution analytics that can differentiate between proteins are often not readily available, pure protein is mandatory for parameter determination. In this work, we present an approach that has the potential to solve this problem. Based on the uniqueness of UV absorption spectra of proteins, we were able to accurately measure retention times in systems of up to four co-eluting compounds. The presented approach is calibration-free, meaning that prior knowledge of pure component absorption spectra is not required. Actually, pure protein spectra can be determined from co-eluting proteins as part of the methodology. The approach was tested for size-exclusion chromatograms of 38 mixtures of co-eluting proteins. Retention times were determined with an average error of 0.6 s (1.6% of average peak width), approximated and measured pure component spectra showed an average coefficient of correlation of 0.992.

Isoform separation and binding site determination of mono-PEGylated lysozyme with pH gradient chromatography.

Covalent attachment of PEG to proteins, known as PEGylation, is currently one of the main approaches for improving the pharmacokinetics of biopharmaceuticals. However, the separation and characterization especially of positional isoforms of PEGylated proteins are still challenging tasks. A common purification strategy uses ion exchange chromatography with increasing ionic strength by shallow salt gradients. This paper presents a method which applies a linear pH gradient chromatography to separate five of six possible isoforms of mono-PEGylated lysozyme, modified with 5 kDa and 10 kDa mPEG-aldehyde. To identify the corresponding PEGylation sites a comparison of elution pH values and calculated isoelectric points of each isoform, was used. The resulting correlation showed an R(2)>0.99. Fractionation, tryptic digestion and subsequent MALDI-MS analysis of each peak, verified the predicted elution order. Based on UV areas the N-terminal amine at lysine 1 exhibited the highest reactivity, followed by the lysine 33 residue.

Molecular dynamics simulations on aqueous two-phase systems - Single PEG-molecules in solution.

BACKGROUND: Molecular Dynamics (MD) simulations are a promising tool to generate molecular understanding of processes related to the purification of proteins. Polyethylene glycols (PEG) of various length are commonly used in the production and purification of proteins. The molecular mechanisms behind PEG driven precipitation, aqueous two-phase formation or the effects of PEGylation are however still poorly understood. RESULTS: In this paper, we ran MD simulations of single PEG molecules of variable length in explicitly simulated water. The resulting structures are in good agreement with experimentally determined 3D structures of PEG. The increase in surface hydrophobicity of PEG of longer chain length could be explained on an atomic scale. PEG-water interactions as well as aqueous two-phase formation in the presence of PO4 were found to be correlated to PEG surface hydrophobicity. CONCLUSIONS: We were able to show that the taken MD simulation approach is capable of generating both structural data as well as molecule descriptors in agreement with experimental data. Thus, we are confident of having a good in silico representation of PEG.

High-throughput screening-based selection and scale-up of aqueous two-phase systems for pDNA purification.

Plasmid DNA (pDNA) is among the promising gene delivery vehicles currently evaluated for gene therapy. The large-scale production of pDNA for pharmaceutical application necessitates purification steps with a high capacity and good separation of RNA from pDNA. Most commonly used process step in the production of biopharmaceutical, namely the divers modes of chromatography, fail as they offer too limited a capacity for the considerably larger pDNA molecules. Alternative separation steps might thus be beneficial. One such separation step, aqueous two-phase extraction (ATPE) has previously been shown to work well for the purification of pDNA. The application of such a process step is however hampered by the large amount of material and time that goes into its development. In this publication, we demonstrate the use of an automatic, miniaturized ATPE screening system to the separation of pDNA from RNA. Two optimization strategies are presented: response surface methodology and genetic algorithms. Using a fully automated optimization strategy, we derived promising conditions that were scale-up tenfold. The resulting purity and recovery surpassed previously published results demonstrating that a complex optimization task such as ATPE demands an appropriately complex optimization routine.

Development and characterization of an automated high throughput screening method for optimization of protein refolding processes.

Optimization of protein refolding parameters by automated, miniaturized, and parallelized high throughput screening is a powerful approach to meet the demand for fast process development with low material consumption. In this study, we validated methods applicable on a standard liquid handling robot for screening of refolding process parameters by dilution of denatured lysozyme in refolding buffer systems. Different approaches for the estimation of protein solubility and folding were validated concerning resolution and compatibility with the robotic system and with the complex buffer and protein structure composition. We established an indirect method to assess soluble lysozyme concentration independent of matrix effects and protein structure varieties by automated separation of aggregated protein, resolubilization, and measurement of absorption at 280 nm. Using this nonspecific solubility assays the correlation between favorable parameters for high active and soluble lysozyme yields were evaluated. An overlap of good refolding buffer compositions was found provided that the redox environment was controlled with redox reagents. In addition, the need to control unfolding conditions like time, temperature, lysozyme, and dithiothreitol concentration was pointed out as different feedstocks resulted in different refolding yields.

Examination of a genetic algorithm for the application in high-throughput downstream process development.

Compared to traditional strategies, application of high-throughput experiments combined with optimization methods can potentially speed up downstream process development and increase our understanding of processes. In contrast to the method of Design of Experiments in combination with response surface analysis (RSA), optimization approaches like genetic algorithms (GAs) can be applied to identify optimal parameter settings in multidimensional optimizations tasks. In this article the performance of a GA was investigated applying parameters applicable in high-throughput downstream process development. The influence of population size, the design of the initial generation and selection pressure on the optimization results was studied. To mimic typical experimental data, four mathematical functions were used for an in silico evaluation. The influence of GA parameters was minor on landscapes with only one optimum. On landscapes with several optima, parameters had a significant impact on GA performance and success in finding the global optimum. Premature convergence increased as the number of parameters and noise increased. RSA was shown to be comparable or superior for simple systems and low to moderate noise. For complex systems or high noise levels, RSA failed, while GA optimization represented a robust tool for process optimization. Finally, the effect of different objective functions is shown exemplarily for a refolding optimization of lysozyme.

Application of an aqueous two-phase systems high-throughput screening method to evaluate mAb HCP separation.

Aqueous two-phase systems (ATPSs) as separation technique have regained substantial interest from the biotech industry. Biopharmaceutical companies faced with increasing product titers and stiffening economic competition reconsider ATPS as an alternative to chromatography. As the implementation of an ATPS is material, time, and labor intensive, a miniaturized and automated screening process would be beneficial. In this article such a method, its statistical evaluation, and its application to a biopharmaceutical separation task are shown. To speed up early stage ATPS profiling an automated application of the cloud-point method for binodal determination was developed. PEG4000-PO(4) binodals were measured automatically and manually and were found to be identical within the experimental error. The ATPS screening procedure was applied to a model system and an industrial separation task. PEG4000-PO(4) systems at a protein concentration of 0.75 mg/mL were used. The influence of pH, NaCl addition, and tie line length was investigated. Lysozyme as model protein, two monoclonal antibodies, and a host cell protein pool were used. The method was found to yield partition coefficients identical to manually determined values for lysozyme. The monoclonal antibodies were shifted from the bottom into the upper phase by addition of NaCl. This shift occurred at lower NaCl concentration when the pH of the system was closer to the pI of the distributed protein. Addition of NaCl, increase in PEG4000 concentration and pH led to significant loss of the mAb due to precipitation. Capacity limitations of these systems were thus demonstrated. The chosen model systems allowed a reduction of up to 50% HCP with a recovery of greater than 95% of the target proteins. As these values might not be industrially relevant when compared to current chromatographic procedures, the developed screening procedure allows a fast evaluation of more suitable and optimized ATPS system for a given task.

3D structure-based protein retention prediction for ion-exchange chromatography.

The interest in understanding fundamental mechanisms underlying chromatography drastically increased over the past decades resulting in a whole variety of mostly semi-empirical models describing protein retention. Experimental data about the molecular adsorption mechanisms of lysozyme on different chromatographic ion-exchange materials were used to develop a mechanistical model for the adsorption of lysozyme onto a SP Sepharose FF surface based on molecular dynamic simulations (temperature controlled NVT simulations) with the Amber software package using a force-field based approach with a continuum solvent model. The ligand spacing of the adsorbent surface was varied between 10 and 20A. With a 10A spacing it was possible to predict the elution order of lysozyme at different pH and to confirm in silico the pH-dependent orientation of lysozyme towards the surface that was reported earlier. The energies of adsorption at different pH values were correlated with isocratic and linear gradient elution experiments and this correlation was used to predict the retention volume of ribonuclease A in the same experimental setup only based on its 3D structure properties. The study presents a strong indication for the validity of the assumption, that the ligand density of the surface is one of the key parameters with regard to the selectivity of the adsorbent, suggesting that a high ligand density leads to a specific interaction with certain binding sites on the protein surface, while at low ligand densities the net charge of the protein is more important than the actual charge distribution.

Effects of ionic strength and mobile phase pH on the binding orientation of lysozyme on different ion-exchange adsorbents.

Chromatography is the most widely used technique for the purification of biopharmaceuticals in the biotech industry. Surprisingly, process development is often still based on empirical studies or experience; recently high-throughput screening stations are employed to minimize development time and to improve screening quality. Still, experimental effort remains high and a more detailed understanding of adsorption mechanisms on a molecular level underlying chromatographic separation could help in the future to select and design chromatography steps in silico. In this study, we focused on the elucidation of protein orientation upon adsorption onto a chromatographic resin. We identified two characteristic binding sites of lysozyme on SP Sepharose Fast Flow and one multipoint interaction of lysozyme with SP Sepharose XL. Increasing ionic strength did not significantly influence the binding, whereas changes in the mobile phase pH led to a re-orientation on SP Sepharose FF. This phenomenon agrees well with theoretical considerations, including a detailed description of the surface charge distribution with changing pH and linear elution experiments, giving an idea why proteins are often retained on ion-exchange materials beyond their isoelectric point.

A novel two-zone protein uptake model for affinity chromatography and its application to the description of elution band profiles of proteins fused to a family 9 cellulose binding module affinity tag.

A novel two-zone model (TZM) is presented to describe the rate of solute uptake by the stationary phase of a sorption-type chromatography column. The TZM divides the porous stationary-phase particle into an inner protein-free core and an outer protein-containing zone where intraparticle transport is limited by pore diffusion and binding follows Langmuir theory. The TZM and the classic pore-diffusion model (PDM) of chromatography are applied to the prediction of stationary-phase uptake and elution bands within a cellulose-based affinity chromatography column designed to selectively purify proteins genetically labelled with a CBM9 (family 9 cellulose binding module) affinity tag. Under both linear and nonlinear loading conditions, the TZM closely matches rates of protein uptake within the stationary phase particles as measured by confocal laser scanning microscopy, while the PDM deviates from experiment in the linear-binding region. As a result, the TZM is shown to provide improved predictions of product breakthrough, including elution behavior from a bacterial lysate feed.

A novel approach to characterize the binding orientation of lysozyme on ion-exchange resins.

Much work has been done to qualify and quantify chromatographic adsorption and transportation mechanisms in different adsorber materials. An important aspect in all studies is the understanding of the binding mechanism between protein and resin on a molecular level in order to optimize processes on the level of adsorber design. We established a method to determine the binding orientation of lysozyme for different materials under various experimental conditions enabling us to observe changes in the mode of adsorption. We varied the protein load of two different adsorber types, Source 15S, a conventional cation exchange resin and EMD Fractogel SO(3), a tentacle-type cation exchanger. We found similar preferential binding sites for the interaction between lysozyme and the surface of these adsorbers at low surface coverage, however, the tentacle adsorber exhibited multi-point binding whereas the binding on Source was limited to one binding site only. With increasing protein density on the surface, lysozyme rotates from a space-consuming side-on to a space-saving end-on orientation on Fractogel, explaining a higher maximum binding capacity for Fractogel. This re-orientation could not be observed for Source.