The application of aqueous two phase systems to the analysis of protein isoforms of importance in clinical biochemistry and biopharmaceutical production

Abstract

Aqueous Phase Partitioning has a long history of applications to the analytical characterisation of biomolecules. However process applications have attracted the most interest in biotechnology where it has become widely recognized as a cost-effective technique. The main aim of this work was to explore the proposition that partition in Aqueous Two Phase Systems (ATPS) can be used as an analytical tool to detect protein isoforms and to assess the applicability of the method in clinical assays and for quality control in bioprocessing through examination of several analytical problems. The work also examined the development of automated methods of system preparation and sampling techniques to determine the partition coefficient in ATPS. The study demonstrated that the geometrical form of the phase diagram co-existence curve was of crucial importance since this directly affected the accuracy with which systems of defined Tie Line Length and Mass Ratio could be constructed. The TLL %Bias (accuracy) of a theoretical system range in the PEG1000-(NH4)2SO4 system at shorter TLL (12.2) was in the range +80.6% to -100% while at a longer TLL (53.1) the %Bias (accuracy) was reduced to +0.1% to -1.9%. At the same time the MR %Bias (accuracy) at shorter TLL (12.2) was in the range +59.5% to -21.3% while at the longer TLL (53.1) this was reduced to +2.7% to -2.6%. By contrast in the PEG8000-Dextran500 system the TLL %Bias (accuracy) at shorter TLL (13.1) was in the range +3.7% to -4.12%, while at a longer TLL (31.1) the range was +0.74% to -0.67%. The MR %Bias (accuracy) at the shorter TLL (13.1) was in the range +3.6% to -3% while at the longer TLL (31.1) the range was +1.1% to -1.4%. This illustrated that it is more difficult to work with a high degree of accuracy (e.g. %Bias <5%) close to the critical point in PEG-salt systems than in PEG-dextran systems. Two different approaches were taken to examine analytical phase partitioning. In the first approach the structure of the isoforms of a model protein (ovalbumin) were altered enzymatically. Analytical methods involving Strong Anion-Exchange chromatography were developed and applied to the separation of the ovalbumin isoforms. Removal of the phosphorylated groups (dephosphorylation of ovalbumin) was undertaken using alkaline phosphatase and de-glycosylation was attempted using neuraminidase and Endo-glycosidase F. However, both enzymatic approaches to deglycosylation were unsuccessful. Dephosphorylated isoforms were successfully produced and characterised. After partitioning in ATPS a clear difference was demonstrated between the behaviour of the native and dephosphorylated forms of ovalbumin. The mean % recovery in a PEG-salt ATPS was 99.8% (± 3.59) for the naive protein and 75.6% (± 4.03) for the dephosphorylated form. On the other hand, in a PEG3350-Dextran500 system, where solubility was maintained, a significant difference in the partition coefficient (K) of native and dephosphorylated ovalbumin was found. K for native ovalbumin was 0.85 while the partition coefficient of the dephosphorylated ovalbumin was 0.61. Analysis of covariance (ANCOVA) indicated that the regression coefficients of the respective partition isotherms were significantly different (p value < 0.05). In a second approach to examine analytical phase partitioning, chemical modification of a specific target surface amino acid of another model protein (serum albumin) was used to determine the degree of conjugation of the protein and also to determine its oxidative state. The method examined the reactivity of a free surface thiol to a wide range of labels ( (a) 2-methylsulfonyl-5-phenyl -1,3,4 oxidiazole reagent, (b) N-Ethylmaleimide (NEM) reagent, (c) 5, 5’-dithiobis (2-nitrobenzoate)(DTNB) (Ellman’s reagent), (d) N-pyrenylmaleimide (NPM) reagent, (e) Fluorescein-5-maleimide (F-5-M) Reagent). Only DTNB was found to modify the surface free thiol of serum albumin in a highly specific and quantitative manner. In the course of the development of a partitioning assay for surface free thiols of serum albumin significant oxidative properties were found to be associated with poly(ethylene glycol) PEG solutions and several attempts were made to find an oxidatively safe partitioning system by including antioxidants and by removal of contaminants by freeze drying. PEG3350-Dextran500 was found to provide an oxidatively safe environment for the development of a partitioning assay for the determination of albumin free thiols. A phase partitioning assay system capable of quantitatively resolving protein associated free thiols and low molecular weight thiols from a mixture of the two was developed. Correlation coefficients (R2) for the regression of experimentally determined protein free thiols in the presence of different levels of added LMW free thiol on the known addition of protein ranged from 0.77 to 0.83. The results demonstrated that the assay could quantify and distinguish both types of thiol in a simple two-step procedure.