Capillary Electrophoresis–Sodium Dodecyl Sulfate of a Heavily Glycosylated Protein: Feasibility of Stability Monitoring

In the biotechnology industry, traditional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) has been widely used to monitor the integrity and purity of therapeutic proteins during formulation and process development, and for lot release and stability testing. In this technique, the proteins are bound to SDS at a ratio of 1.4 g of SDS to 1 g of protein.1 This constant binding leads to similar free solution mobilities for all protein molecules2 and, therefore, their separation based on hydrodynamic size. Recently, there has been a growing interest in using capillary electrophoresis (CE)-SDS to replace traditional SDS-PAGE in the therapeutic protein area. The applications of CE-SDS in recombinant GB virus-C protein,3 recombinant carboxypeptidase B,4 collagen, 5 a protective antigen protein (the major component of human anthrax vaccine),6 and monoclonal antibodies7–10 have been demonstrated. The potential advantages of CE-SDS over SDS-PAGE include faster analysis, the ability for quantification, full automation, low sample consumption, and potentially better resolution. In this paper, the authors will present the development of a CE-SDS method and will evaluate its feasibility for the stability testing of a heavily glycosylated protein: Protein X. Over 40% of the total mass in Protein X is composed of carbohydrates. The high heterogeneity of the molecule poses extra challenges for the development of the CE-SDS method, since the main peak of the protein is expected to be broad, and therefore may compromise the sensitivity of the assay.

The method development discussed in this paper is focused primarily on optimizing electrophoresis conditions and improving sensitivity by a variety of means, including the use of extended light path capillaries. A homemade sample buffer was successfully implemented to replace the commercial sample buffer in order to minimize sample dilution, which further improved sensitivity. In addition, accelerated stability studies were performed by protease cleavage of Protein X, or by incubation of the protein at elevated temperatures in buffers at various pH levels. The results demonstrated that the described CE-SDS method is capable of monitoring the degraded species quantitatively.

Experimental

Reagents and materials

Protein X was produced in transfected Chinese hamster ovary cells (Amgen Inc.,Thousand Oaks, CA). The CE-SDS protein kit including CE-SDS sample buffer, CE-SDS protein run buffer, 0.1 N NaOH, 0.1 N HCl, and internal standard (IS) benzoic acid (1 mg/mL) was acquired from Bio-Rad Laboratories (Hercules, CA). Unless otherwise stated, all other chemicals were purchased from Sigma (St. Louis, MO). Both the extended light path capillary (50 μm i.d. × 33 cm, bubble factor 3) and standard capillary (50 μm i.d. × 33 cm) were obtained from Agilent Technologies (Palo Alto, CA).

Sample preparation using Bio-Rad sample buffer: Protein X was mixed with sample buffer at a ratio of 1:8 (v/v) followed by the addition of benzoic acid at 5% (v/v). β-Mercaptoethanol (BME) was further added at 2.5% (v/v) to the mixture to reduce the protein. The mixture was incubated at 95 °C for 10 min.

Sample preparation using homemade sample buffer: Protein X was incubated with 20% or 10% SDS (w/v, in H2O) at a mass ratio of 1:1.7 (protein:SDS) with benzoic acid added as an IS at 5% (v/v). BME was further added to the mixture at 2.5% (v/v) to reduce the protein. The sample mixture was heated at 95 °C for 10 min.

CE-SDS conditions: CE-SDS was performed on a G1600 HPCE instrument (Agilent Technologies) equipped with a UV detector. Unless stated separately, for electrophoretic injection, the sample was introduced at 20 kV (negative polarity) for 20 sec. For pressure injection, the sample was introduced at 3 bar for 24 sec. A negative voltage of up to 25 kV was applied across the capillary and the sample was electrophoresed for 30 min. Signal collection was conducted at 220 nm.

Results and discussion

Selection of capillary

Figure 1 - Comparison of the bubble-cell capillary and standard capillary. CE-SDS conditions: electrophoretic injection at –10 kV for 20 sec; run voltage at –15 kV.

Method development was started by using the CE-SDS protein kit. The lower trace shown in Figure 1 is a typical electropherogram for Protein X at 0.5 mg/mL, using a standard bare silica capillary with an internal diameter of 50 μm. The signal intensity obtained was low, and such low sensitivity would be a problem for stability monitoring since any degraded species are expected to grow at a much lower concentration relative to the main peak. In order to improve the sensitivity, extended light path capillaries (also called bubble-cell capillaries) were used instead of standard capillaries. The unique feature of the bubble-cell capillary is that its detection window is blown to the shape of a bubble to increase the optical light path (Figure 1, inset). It can be seen from the figure that the signal intensity of Protein X obtained by using the bubble-cell capillary is almost 3× higher than that obtained by using the standard capillary. The increased intensity is a direct result of the increased light path. Another advantage of the bubble-cell capillary is that the running current stays comparable with the standard capillary because the inner diameter of the capillary is only increased at its detection window while the rest of the capillary remains unchanged. Therefore, the Joule heating effect is minimal. However, the bubble-cell capillary resulted in a slight decrease in the resolution of the isoforms under the main peak, which could be microheterogeneity contributed from the carbohydrate structures.

Optimizing CE-SDS running conditions

Next, the effect of run voltage at –15, –20, and –25 kV on the separation was investigated. It was found that higher voltages offered faster separation and sharper peak shape. However, as the run voltage increased, the baseline became less stable. Under the consideration of both separation time and baseline, –20 kV was chosen as the optimum run voltage.

For electrophoretic injection, injection time and voltage both have an impact on the signal intensity. As far as the injection time is concerned, the longer the injection time, the higher the signal intensity. It was found that at a fixed injection voltage of –20 kV, the peak became wider as the injection time was increased from 5 to 20 sec due to the larger sample plug introduced. In the end, an optimum injection time of 20 sec was chosen. An injection time longer than 20 sec was not considered due to the wider peak width.

A similar effect was found for injection voltage. The signal intensity was increased as the injection voltage was increased from –10 to –20 kV. Based on the same consideration for the peak width, –20 kV was chosen as the optimum injection voltage.

Optimizing sample preparation

Sample preparation is an important aspect to consider for the CE-SDS development. One crucial component for the sample preparation is the sample buffer, which contains SDS for protein binding. The original sample preparation was performed by mixing Protein X with the Bio-Rad sample buffer at a ratio of 1:8, which resulted in at least a ninefold dilution of the protein. If the protein concentration is low, such dilution could be a concern for monitoring species in the stability samples. The authors have developed a homemade sample buffer consisting of 20% SDS (w/v, in H2O), and its binding capability with Protein X was investigated. The concentration of SDS was made high so that a very minimal volume of SDS solution could be used in order to minimize the dilution from the sample buffer. Another reason to use homemade buffer is that the buffer composition is known so that it is possible to manipulate the binding ratio to improve the method sensitivity.

Figure 2 - Comparison of the Bio-Rad sample buffer with the homemade sample buffer. The data have been offset for better visual comparison. The numbers represent the mass binding ratio of SDS to protein. CE-SDS conditions: electrophoretic injection at –20 kV for 20 sec; run voltage at –20 kV.

Shown in Figure 2 are the electropherograms comparing the Bio-Rad sample buffer with the homemade SDS buffer. The binding ratio of SDS to Protein X based on mass, as indicated by the numbers in Figure 2, ranges from 0.9 to 20.7. These binding ratios correspond to the dilution factor from 1.05 to 1.6, compared to 9 for the Bio-Rad sample buffer. One notable point is that the final concentration for the protein sample was made the same (0.2 mg/mL) by adding an appropriate amount of H2O to the mixture of the protein and sample buffer in order to compare the signal intensity. It can be seen that the signal intensity is the lowest for the Bio-Rad sample buffer. For the homemade buffer, the intensity increases as the binding ratio decreases from 20.7 to 1.7 and changes very little thereafter. One possible reason for the increased signal intensity could be that the excess SDS in the sample may have an adverse effect on the electrophoretic injection of the protein (i.e., it lowers the loading efficiency, which results in the lower signal). For later method development, a binding ratio of 1.7 was used. This ratio is similar to that used in SDS-PAGE, in which a binding ratio of 1.4 is typically used.

It was also found that a 10% SDS solution provided sufficient signal intensity while maintaining minimum dilution of the sample. Therefore, a 10% SDS solution was used as the sample buffer for the subsequent method development.

Effect of salt concentration

Figure 3 - Effect of buffer salt concentration on electrophoretic injection and pressure injection. a) Buffer salt effect on electrophoretic injection; b) comparison of pressure and electrophoretic injection under the same salt matrix.

During the experiments, it was noticed that the salt concentration in the sample had an effect on the signal intensity when the sample was injected electrophoretically. Figure 3a shows an example in which the injection was performed at –20 kV for 20 sec. The blue trace represents the signal for Protein X at 5.8 mg/mL in the buffer of 20 mM NaPi and 140 mM NaCl. A higher signal was obtained when the sample was diluted fivefold, as shown by the red trace under the sample injection conditions. This comparison clearly demonstrates that high salt content in the sample reduces the loading efficiency of the protein. In the red trace, although the protein was more dilute compared to that in the blue trace, the salt concentration was 5× lower than that in the blue trace. Therefore, more protein was loaded in the capillary, which resulted in the higher signal in the red trace.

In contrast to electrophoretic injection, pressure injection has no bias toward the sample in terms of loading efficiency. Shown as the red trace in Figure 3b is the electropherogram for Protein X at 1.0 mg/mL in the buffer of 20 mM NaCl and 140 mM NaPi, where the sample was injected using pressure at 3 bar for 24 sec. It can be seen that, although the salt concentration was as high as that in the blue trace (Figure 3b), much higher signal intensity was obtained.

The recommendation for the sample preparation is to desalt if the salt concentration is higher than 50 mM for the electrophoretic injection. However, pressure is a preferred injection mode in order to minimize sample manipulation. In the authors’ subsequent experiments, the injection was performed using pressure at 3 bar for 24 sec.

CE-SDS for stability testing

Figure 4 - Electropherograms of Protein X control and Protein X incubated with a protease at 37 °C for 24 hr. CE-SDS conditions: pressure injection at 3 bar for 24 sec; run voltage at –20 kV.

To generate degraded protein species for accelerated stability testing, Protein X was incubated with a protease that the protein is known to be sensitive to. As shown in Figure 4, clipped species were identified by comparing the electrophoretic profile to that of the control sample. The level of clips was determined to be 4.8% relative to the total protein peak area. Additional protein cleavage was induced by incubating Protein X at elevated temperatures (37 and 55 °C) in 20 mM buffers at pH 5, 6, and 8 over time. Table 1 summarizes the effect of temperature and pH on the generation of the clipped species, with the protein sampled at intervals of 5, 7, and 9 weeks. It is obvious that the protein is most stable in the pH 6 buffer, and least stable in the pH 8 buffer. Higher temperature appears to accelerate the protein degradation.

Conclusion

A CE-SDS method has been developed using the Agilent HPCE instrument for Protein X. In order to improve the method sensitivity for this highly heterogeneous glycoprotein, a bubble-cell capillary was used. Compared to the standard capillary, the signal intensity was increased threefold. For the sample preparation, a homemade sample buffer was employed to minimize the sample dilution, which allowed further improvement to sensitivity. It was also found that pressure injection was able to withstand higher salt content in the sample than electrophoretic injection, and also required very minimal sample treatment. Finally, in order to qualify the method for stability testing, clipped species were induced by incubation of the protein with a protease, or at elevated temperatures in various pH buffers for up to 9 weeks. The results demonstrated that the CE-SDS method was stability-indicating and that the method may be used in formulation development. In the future, the feasibility of using a fluorescent tag7 will be explored to further improve the method sensitivity. It is believed that the method will have a wider range of application in addition to stability testing, such as impurity detection for process development, with the increased sensitivity.

References

  1. Reynolds, J.A.; Tanford, C. The gross conformation of protein-sodium dodecyl sulfate complexes. J. Biol. Chem. 1970, 245, 5161–5.
  2. See, Y.P.; Jackowski, G. In Creighton, E.D., ed. Protein Structure: a Practical Approach. URL Press: Oxford, U.K., 1989; pp 1–21.
  3. Kundu, S.; Fenters, C.; Lopez, M.; et al. Capillary electrophoresis for purity estimation and in-process testing of recombinant GB virus-C proteins. J. Cap. Elec. 1997, 4, 7–13.
  4. Good, D.L.; Cummins-Bitz, S.; Fields, R.M.; Nunnally, B.K. Capillary electrophoresis of proteins in a quality control environment. Meth. Mol. Biol. 2004, 276, 121–36.
  5. Chang, P.; Kuan, S.; Eberlein, G.; Burke, D.; Jones, R. Characterization of bovine collagens using capillary electrophoresis—an alternative to slab gel electrophoresis. J. Pharm. Biomed. Anal.2000, 22, 957–66.
  6. Farchaus, J.W.; Ribot, W.J.; Jendrek, S.; Little, S.F. Fermentation, purification, and characterization of protective antigen from a recombinant, avirulent strain of Bacillus anthracis. Appl. Environ. Microbiol.1998, 64, 982–91.
  7. Hunt, G.; Nashabeh, W. Capillary electrophoresis sodium dodecyl sulfate nongel sieving analysis of a therapeutic recombinant monoclonal antibody: a biotechnology perspective. Anal. Chem. 1999, 71, 2390–7.
  8. Lee, H.G. High-performance sodium dodecyl sulfate-capillary gel electrophoresis of antibodies and antibody fragments. J. Immunol. Meth.2000, 234, 71–81.
  9. Ma, S.; Nashebeh, W. Analysis of protein therapeutics by capillary electrophoresis. Chromatographia2001, 53, S59–65.
  10. Schenerman, M.A.; Bowen, S.H. Optimization, validation, and use of capillary gel electrophoresis for quality control testing of synagis®, a monoclonal antibody. Chromatographia2001, 53, S66–72.

    The authors are with the Department of Process and Analytical Sciences Operations, Amgen Inc.,One Amgen Center Dr., Thousand Oaks, CA 91320, U.S.A.; tel.: 805-447-1000; fax: 805-499-0126; e-mail: [email protected].