Comparison of GPC and Mass Spectrometry
Possibilities and Limits for the Control of Protein Stability
Gel permeation chromatography (GPC) is often used for the chromatographic characterisation of protein formulations. GPC is simple to use and detects high molecular weight contaminants (e.g. aggregates) and low molecular weight degradation products. Here, GPC is used for the stability examination of haemoglobin preparations. This article compares two common GPC columns and considers their efficiency by means of mass spectrometry investigations.
Up to now, a validated method based on the Tosoh TSKgel SuperSW2000 GPC column has been used for these measurements. The present measurements are intended to investigate whether the method can be transferred to the Agilent Bio SEC-3 GPC column and whether the measurements can be continued with this new column. According to information from the manufacturer, the stationary phase features a coated silica gel, for which a different stability can be expected.
The investigations present a detailed comparison with regard to the linearity, detection limit, reproducibility and selectivity of the two GPC columns.
In addition, the results are compared with MS examinations, which have a considerably greater separation performance / selectivity. With this, the importance (correctness) of GPC for the investigation of protein stability is to be evaluated.
Stability of the Elution
Statistical evaluation of the retention showed good reproducibility for both columns (Srel < 0.1%, see Table 1). However, it was established that retention of the main component increases significantly if the concentration is lower (from 15.56 min with 1000 μg/ml to 16.15 min with 5 μg/ml). At present this cannot be explained. Because of this, samples with a constant, low concentration have to be used.
For a detailed comparison, the retention times or peak heights of the individual measurements are compared graphically in Figure 1. Examination of the peak areas of the main components showed no significant trend for the two columns and this is therefore not included in the graphic evaluation.
The Bio SEC-3 column also required several measurements in order to achieve a constant retention or peak height.
However, the trend is not so pronounced as with the TSKgel SuperSW2000 column. An outlier at the 9th measurement (15.97 min.) with the Bio SEC-3 column was not taken into account in the comparison.
The GPC is to detect the formation of higher molecular weight oligomers or low molecular weight degradation products. For the evaluation of the two systems, the same samples were analyzed and compared with the two columns. The Bio SEC-3 column showed a different differentiation of the high molecular weight component. The low molecular weight contamination can also not be detected in the expected way.
Fig. 2 shows the separation of the same haemoglobin batch for various storage conditions. This was also measured with the two columns. With the Bio SEC-3 column, the high molecular weight component can only be differentiated from the main peak with storage at 35 °C. With the other two samples, the high molecular weight component cannot be definitely differentiated from the main peak and therefore cannot at present be calculated as a percentage of the main component using the TSKgel column. In contrast to this (Fig. 2b) the high molecular weight component can be explicitly determined for all 3 samples using the TSKgel Super SW2000 column.
Essential statistical parameters were calculated for a series of 12 measurements (Table 1).
The results for the reproducibility do not show a significant difference between the two columns. The only conspicuous feature is the proof of the retention time trend of the TSKgel SuperSW2000, which has already been described above, occurs with the first measurements before a constant elution is achieved in the course of subsequent measurements.
By dilution of a standard haemoglobin preparation (1000 μg/ml) 5 further calibrators were obtained (5, 10, 50, 100, 500 μg/ml). In addition, the blank buffer injections were used as a 0 μg/ml calibrator.
After linear regression verified, by variance analysis, the straight line was rejected as the calibration model for the Bio SEC column; the 2nd degree polynomial was accepted for the entire calibration range. In the lower concentration range from 0-10 μg/ml a straight line was accepted.
For the Super SW column, both the straight line and the 2nd degree polynomial model were rejected. A straight calibration line could only be used in the range 10 – 1000 μg/ml (without the 0 μg/ml calibrator).
From the calibration data follows that in the range from 0 to 100% the peak area is not based on a straight correlation with the concentration of the components; however in the range from 0-1%, so that the detection limit can be calculated with a straight line model.
From the registration of the 1st calibrator, the detection limit of the Bio SEC column can be estimated as 0.5 μg/ml (signal/noise = 3:1, confirmed linearity in the range < 10 μg/ml). For the Super SW column this is 1μg/ml.
Recovery was determined by means of 2-dimensional HPLC. In the GPC measurement of 2 different proteins, the main peak of the sample was passed to a reversed phase cartridge (heartcut) during elution of the main signal and quantified via a reversed phase chromatography (RPC) column by means of UV detection. Direct RPC measurement of the protein sample was used as a 100% reference. The two columns do not show any serious differences in terms of recovery. The Bio SEC column appears to have a slightly better recovery due to its coating (Table 2).
Separation was illustrated on the basis of chromatograms of two proteins (ribonuclease A and haemoglobin) with new columns, and is depicted in Table 3.
The equivalent theoretical plate heights of the two columns are comparable in spite of different particle sizes (Bio SEC-3: 3 μm, TSKgel: 4 μm).
Alternatively to GPC, mass spectrometry was used to check the protein stability (ESI-TOFMS with online desalination without HPLC separation). In the mass spectrum (Fig. 3), in addition to the two subunits of haemoglobin (m/z = 15037.68 and 16033.56) a further mass was detected in the relevant range (m/z = 16195.61) The concentration of this component increases with longer incubation periods, especially at higher temperatures. It is assumed that in this case there is an adduct with a low molecular weight component of the pharmaceutical formulation.
This component could not be detected with GPC. In addition, several masses within the range 2500-5000 m/z (see mass list Table 4) are not differentiated in GPC.
The correctness of GPC analyses of protein stability must therefore be regarded critically.
In general, it must be taken into consideration that the signal area percentages of the GPC measurement do not allow any conclusions to the present concentrations due to the (usually unknown) absorption coefficients. The GPC analysis with the Bio SEC-3 column did not show any clear difference with regard to the detection limit and the reproducibility. There are also no significant differences according to linearity and recovery.
The selectivity of the elution is a different matter. In contrast to the TSKgel column, with the Bio SEC column shortly before and after the main peak no or only poorly separated components (high or low molecular weight) elute, which could be assessed as indicators of protein degradation. Therefore, the TSKgel column is preferable for this task.
Regardless of this, it must not be ignored that the separating performance of GPC is very low (peak capacity ~10). Therefore it cannot be expected that all of the protein components will be detected after storage. This is impressively demonstrated by the MS measurements. An assessment of protein stability merely on the basis of GPC examinations must therefore be rejected.
N. Thiessen, M. Müller, E. Reh