Demonstrating the value of DSC as tool to predict protein stability at: lower temperatures, screen buffers and excipients, screen therapeutic candidates and to predict protein aggregation.
During its lifetime, a therapeutic protein will be subjected to a range of different conditions throughout processing and storage. These often involve low pH, different buffer components and ionic strengths, and temperature fluctuations during storage and transportation. In addition, the final formulation must be stable for several years, often at very high protein concentrations. The final drug product delivered to the patient must retain its native conformation and activity, with minimal self-association and aggregation. Any analytical tool capable of screening for the effect of these conditions on the conformation and self-association of the protein, will aid selection of which biotherapeutic candidates to move forward in development, and the process and formulation conditions to use.
Differential Scanning Calorimetry (DSC) is commonly employed to assess the thermal and conformational stability of a protein under different buffer conditions (1-6). The melting temperature of a protein, or of individual domains, can be obtained from the DSC profile (6). If the reaction is reversible, the thermodynamic parameters of the unfolding can also be determined. In addition, unfolding is often accompanied by an exotherm that corresponds to the aggregation and precipitation of the unfolded protein.
DSC has been used to characterize the thermally-induced unfolding of antibodies and Fc fragments, with the transitions of the individual domains identified (6-9). The CH2 domain usually unfolds first (7) followed by the Fab and then the CH3 domains.
Several controlled experiments (Figure 1) indicate that the CH2 and CH3 domains of the Fc fragment unfold at 71.0°C and 83.1°C, respectively, under near physiological conditions in PBS. The thermal transition of the Fab domain of a monoclonal antibody (MAb) usually occurs between the transitions of the CH2 and CH3 domains or overlaps with one of these two transitions.
|
SEC or gel filtration (GF) is commonly used as a stability-indicating assay, allowing quantification of the monomer and other higher molecular weight species that are generated either by conditions to which the protein is exposed or during storage.
In the work described here we compare the stability of different antibodies and Fc-conjugated proteins stored under different conditions at different temperatures, as determined by SEC-HPLC analysis, with the stability predicted by DSC in the same buffer. The results demonstrate that DSC can be used to select conditions under which the proteins are the most stable for long-term storage, and also to screen different related proteins, such as analogs, for their relative long-term stability.
The samples were prepared in 20 mM sodium citrate with 140 mM NaCl at the pH indicated unless otherwise specified. Two sets of identical samples of each protein were prepared: one set for the DSC analysis and the other set for HPLC analysis. All experiments were performed at a protein concentration of 0.5 mg/ml.
The default storage temperature was 4°C for all the samples. However, it can take months or years to observe changes in aggregation at this temperature for most of the proteins used in this study, reflecting the long-term shelf-life desired for a successful therapeutic protein. Therefore, 37°C was also used as storage temperature in order to complete this project within a limited time (two months).
A commercial SEC-HPLC system with on-line UV, light scattering and refractive index detectors (SEC-UV/LS/RI) was used for the study. The Tosoh TSKgel™ G3000SWXL (7.8 × 300 mm) SEC column had a flow rate of 0.5 ml/min and samples were injected at the indicated times. The UV chromatograms were monitored at 280 nm.
The DSC experiments were performed using a Malvern MicroCal VP-DSC system. All samples were degassed for five minutes before analysis. The reference cell was filled with a buffer corresponding to the sample buffer. The samples were heated from 4°C to 110°C at a heating rate of 60°C/h. The pre-scan was 15 minutes, the filtering period was 10 s, and the feedback mode/gain was set to passive. The midpoint of a thermal transition temperature (Tm, or thermal transition temperature) was obtained by analyzing the data using Origin™ 7 software.
DSC scans of a Fc-conjugated protein X at different pH values are shown in Figure 2. The thermal transition temperatures of the scans are also included. At pH 7 there are two thermal transitions at 65.4°C and 78.9°C, corresponding to the unfolding of CH2 and CH3 domains. The thermal transition temperature decreases with decreasing pH, and the stability order predicted by DSC is pH 7 > pH 5 > pH 4. The actual protein stability was studied using SEC-HPLC with on-line UV, light scattering and refractive index detectors. The advantage of using the light scattering detector is that the monomer peak can be easily confirmed (10). The SEC chromatograms of Fc-conjugated protein X stored at 4°C, at pH 7 and pH 4, are shown in Figures 3 and 4 respectively. The comparison of the monomer peak percentages at different time points for pH 7, pH 5, and pH 4 samples stored at 4°C are shown in Figure 5. All percentage values of the SEC monomer data in this application note are normalized against the value at T = 0.
|
|
|
|
Since it is very hard to differentiate the stability of pH 7 and pH 5 samples within weeks at 4°C, accelerated studies were performed at 37°C. The comparison of the monomer peak percentage from SEC obtained at different time points for the pH 7 and pH 5 samples are shown in Figure 6. With storage at 37°C, the stability of Fc-conjugated protein X at pH 7 and pH 5 is clearly differentiated. The increase of the monomer peak percentage to over 100% for the samples stored at 37°C was due to evaporation, which was observed for all the 37°C samples, and resulted in increased protein concentration in all the samples.
|
The SEC results from the above two sets of experiments (4°C and 37°C) clearly suggest that the actual stability of Fc-conjugated protein X is pH 7 > pH 5 > pH 4, confirming the DSC prediction.
The DSC scans of Fc-conjugated protein X at pH 6 and pH 5 are shown in Figure 7. The comparison of monomer peak percentage at different time points after storage at 37°C, obtained from SEC for pH 6 and pH 5, are shown in Figure 8. Once again, the results suggest that the stability order predicted by DSC, pH 6 > pH 5, correlates very well with the actual stability data obtained from SEC.
|
One other factor contributing to the apparent thermal stability is the solubility of the unfolded protein. Using the Malvern MicroCal VP-DSC, the final exotherm (sharp down curve in the negative direction) after the transition reflects the solubility of the unfolded protein. This is another factor, in addition to the transition temperature that may have a significant impact on protein stability.
Aggregation/precipitation is an irreversible reaction. As it occurs, it shifts the reversible equilibrium of the unfolding reaction in favor of the unfolded form of the protein. Thus, the less soluble the unfolded intermediate, the more unfolding and aggregation will occur with time. If aggregation is simultaneous with the first thermal transition, it will occur before the entire reaction. This will further complicate the analysis.
After studying the Fc-conjugated protein X, the same method and procedure were used for a MAb. The DSC scans of a MAb Y at different pH values are shown in Figure 9. The thermal transition temperatures are labeled on the figure. At pH 7 there is only one thermal transition, at 73.2°C, followed almost immediately by the aggregation exotherm. When the pH is decreased, the thermal stability of CH2 domain decreases, and the Tm values are 66.1°C and 47.9°C for the pH 5 and pH 4 samples, respectively. The stability order predicted by DSC is pH 7 > pH 5 > pH 4.
|
|
The actual protein stability was again studied using SEC-HPLC. The monomer peak was confirmed by the light scattering detection method (10). The monomer peak percentages at different time points for the pH 7 and pH 4 samples stored at 4°C were compared (results not shown). Since it was not feasible to undertake extended experiments to differentiate stability of MAb Y at pH 7 versus pH 4 with storage at 4°C for up to several months, accelerated degradation studies were performed at 37°C.
The comparison of the monomer peak percentage obtained from SEC for the pH 7, pH 5, and pH 4 samples stored at 37°C is shown in Figure 10. The results confirm the stability order predicted by DSC: pH 7 (slightly) > pH 5 (significantly) > pH 4.
|
As described earlier, we studied the correlation between thermal stability and protein stability at different pH values for the same protein. We also wanted to investigate whether we could use the same approach to compare the stability of different proteins under the same buffer conditions. In general, when comparing different proteins, the correlation between the relative thermal stability and
relative long-term stability decreases as the structural similarity of the proteins decreases. DSC has been used successfully to assess the stability of related proteins, including screening analogs of the same protein.
Comparisons of the thermal transition temperatures from the DSC scans of MAb Y and Fc-conjugated protein X at pH 4, pH 5, and pH 7 are shown in Figure 11. The comparisons of the monomer peak percentages of MAb Y and Fc-conjugated protein X at pH 4, pH 5, and pH 7 are shown in Figures 12 to 14. At pH 4 and pH 5, the DSC prediction of the stability order (MAb Y > Fc-conjugate protein X) is clearly confirmed by the SEC stability data. The pH 7 data are more complicated. The DSC stability data show that MAb Y is more stable than Fc-conjugate protein X, but the difference in the SEC data is not obvious.
There are several possible contributing factors. One is that both proteins are very stable at pH 7 and much longer storage time is needed in order to see differences in the amount of monomer using SEC. Another possible reason is that the unfolded Fc-conjugated protein X is more soluble than MAb Y (Figure 2 and 9) and that this partially compensates for the difference in the thermal transition temperatures, and makes the actual stability of both proteins more similar than would be predicted from the thermal transition temperature alone. This suggests that the aggregation exotherm is also an important factor to consider.
The approach described in this article was also used to study several different proteins, both antibodies and Fc conjugates. The relative order of thermal stability obtained from the DSC data did indeed reflect not only the actual storage stability and aggregation of the same protein in different buffers seen with SEC-HPLC analysis, but also the actual storage stability and aggregation of the different related proteins in general.
After establishing the correlation between the thermal stability and storage stability for a particular protein, DSC can be used to quickly assay different mutants, different constructs, Fc related proteins, and MAbs. As shown above, other factors, such as the solubility of the unfolded protein intermediates, and similarities and differences in the protein structure, need to be considered in addition to the thermal transition temperature.
|
|
|
|
DSC was used to study the thermal stability of monoclonal antibodies and Fc-conjugated proteins at different pH values. SEC-HPLC was used to study the storage stability of the same set of proteins at corresponding pH values. The SEC-HPLC stability data correspond well with the stability predicted by DSC, suggesting that the thermal stability data obtained from DSC correlates with protein stability at lower temperatures. In addition, the relative thermal stability of the related proteins also reflects differences in their actual long-term stability. Therefore, DSC is a useful tool to predict the protein stability at lower temperatures, screen buffers and excipients, screen therapeutic candidates and to predict protein aggregation.
This application note was kindly provided by Jie Wen, Yijia Jiang, Kathryn Hymes*, Ke Gong*, and Linda Narhi, Amgen Inc., Thousand Oaks, CA.
*summer interns
