Rapid quantification of protein–polyethylene glycol conjugates by multivariate evaluation of chromatographic data

Abstract

Size exclusion chromatography (SEC) is often applied for characterization of protein–polyethylene glycol (PEG) conjugates regarding the number of attached PEG chains (PEGamers). SEC analysis is advantageous as it is precise, robust, and straightforward to establish. However, most SEC based assays have a maximal throughput of a few samples per hour. We present a strategy to increase analytical throughput based on combining a short column with a fast flow rate, and finally multivariate calibration in order to compensate for the resolution lost in the trade off for speed. Different multivariate approaches were compared and multilinear regression was shown to result in the most precise calibrations. Further, a dynamic calibration approach was developed in order to account for changes in column performance over time. In this way, it was possible to establish a highly precise assay for protein PEGamer quantification with a throughput of 30 samples per hour.

Introduction

Attaching polyethylene glycol (PEG) polymer chains to proteins (PEGylation) has been shown to improve their pharmacokinetics. The increase in size reduces renal clearance of the therapeutic and there is proof of immunogenicity and antigenicity being reduced by PEGylation [1], [2]. Further, solubility of hydrophobic proteins can be increased by PEGylation [3]. The PEGylation process determines the PEGamer distribution and positions of the PEG molecules attached to the protein. If random PEGylation is performed, the product of the reaction will be very heterogeneous. This poses a problem in a regulative environment which demands defined products of extremely high homogeneity. In general, there are two ways to reach a homogeneously PEGylated product. Either the product mixture of the random PEGylation can be purified to contain only the desired PEGylated species or site specific PEGylation can be performed.

In our lab, small scale PEGylation performed automated on liquid handling stations has been established. This enables high throughput screenings of PEGylation process parameters and is a part of a general trend where automated high throughput experimentation (HTE) is applied for high throughput process development (HTPD) of biologicals [4]. For the evaluation of the performed experiments to stand in relation to experimental speed when performing HTE, the analytical throughput must match the experimental throughput. In some cases this comes easily, for instance if the evaluation of screenings performed in HTE mode can be based merely on univariate spectroscopic measurements such as total protein quantification via UV absorption measurements [5], [6], [7]. If selective or specific quantification is necessary, other methods have to be considered. In the case of quantitative separation of protein PEGamers, size exclusion chromatography (SEC) is widely applied. However, for SEC based assays to match the speed of HTE, the analysis time per sample must be reduced to merely a few minutes.

SEC is a standard method for selective and specific quantification of proteins. If exact quantification is the primary objective, most analysts will seek to achieve high resolution (R > 1.5) of the components. If a short analysis time is the primary objective, faster flow rates and/or shorter columns can be applied, however, at the expense of resolution. One approach to increase analytical speed without losing resolution is the interlacing of injections and/or parallel operation of two columns on a chromatographic system [8], [9]. When interlaced injections are performed in SEC analysis, the result is elimination of initial lag time between sample injection and start of elution. If interlaced injections are combined with parallel operation of two columns, the waiting time post elution of the smallest sample molecule of interest can be reduced or even eliminated. Thus, in favorable situations, analysis time per sample can be reduced to the time span in which the molecules of interest elute.

The separation of protein PEGamers using SEC poses a very challenging task, as the relative increase in molecular size decreases for each additional PEG molecule attached to the protein [10]. A size based chromatographic separation of native protein and mono-PEGylated species might be achieved effortlessly, however a resolution of the higher protein PEGamers will become increasingly difficult.

Our proposition is that for analytical purposes the degree of separation necessary for correct quantification can be decreased significantly by applying alternative methods for the evaluation of chromatographic data. This idea is based on the assumption that overlapping elution of different species will result in a chromatogram which is a linear superposition of the signals of each single analyte. Therefore, a linear multivariate correlation between elution profile (chromatogram) and amounts of the different components in the respective samples can be expected. This, of course, presumes that the amount of the analytes does not influence the elution profile. In the linear range of adsorption, this assumption would apply as peak shape is a function of the chromatography system (dead volume, column packing etc.) and protein characteristics (adsorption and diffusion) but not of protein concentration. Further, the separation mechanism of SEC does not depend on direct interaction with the column material and thus it is not expected that different load concentrations should cause a non-linear change of the elution profile. If the linear correlation between analyte concentration and elution profile is given, it should be possible to calibrate a multivariate regression model based on one chromatogram of each pure component and one defined mixture of all components. Such a multivariate regression model could then give precise determinations of sample composition despite low chromatographic resolution of the sample components.

The presented work describes how multivariate calibration can be used to gain quantitative results of high quality from low resolution chromatograms. By doing so, faster analysis times are achievable through application of short columns and high flow rates that would otherwise be avoided due to inferior separation performance.

In general, a balanced ratio of analytical and experimental time allows a more efficient use of the HTE platform. The goal was to achieve an assay time of maximum 2 min. This would facilitate a 24 h experiment-analysis cycle for PEGylation screenings in 96 well HTE format, where screening experiments which are performed during day time can be analyzed over night. Further, fast assays can be useful for PEGylation reaction monitoring and subsequent purification process monitoring.