Differential scanning calorimetry measures the thermal transition temperature(s) and total heat energy required to denature a protein. Results obtained are used to assess the thermal stability of protein antigens in vaccine formulations.
Differential scanning calorimetry (DSC) is an analytical technique that measures the molar heat capacity of samples as a function of temperature. In the case of protein samples, DSC profiles provide information about thermal stability, and to some extent serves as a structural “fingerprint” that can be used to assess structural conformation. It is performed using a differential scanning calorimeter that measures the thermal transition temperature (melting temperature; Tm) and the energy required to disrupt the interactions stabilizing the tertiary structure (enthalpy; ∆H) of proteins. Comparisons are made between formulations as well as production lots, and differences in derived values indicate differences in thermal stability and structural conformation. Data illustrating the use of DSC in an industrial setting for stability studies as well as monitoring key manufacturing steps are provided as proof of the effectiveness of this protocol. In comparison to other methods for assessing the thermal stability of protein conformations, DSC is cost-effective, requires few sample preparation steps, and also provides a complete thermodynamic profile of the protein unfolding process.
Differential scanning calorimetry (DSC) is an experimental method that directly measures the difference in heat energy uptake taking place in a sample relative to a reference during a regulated temperature change 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12. Carried out in a differential scanning calorimeter, the method involves introducing heat energy into a sample cell and a reference cell simultaneously while identically increasing the temperature of both cells over time 2, 13, 14. Due to difference in the composition of the sample and the reference, different amount of energy will be required to raise the temperature of the cells 2, 12, 13. Thus, the excess amount of energy required to compensate for the temperature difference between the cells is measured and directly correlated to specific thermodynamic properties of the sample 1, 3.
In the 1960s, M.J. O'Neil and E. Watson of Perkin Elmer developed the first differential scanning calorimeter to measure the heat flow of solid materials 2, 3, 4. In parallel, P.L. Privalov and D.R. Monaseldze E.L. of the Institute of Physics, Republic of Georgia (former USSR) created a unique differential adiabatic calorimeter that can be used for biochemical research 5, 6. Subsequently, Andronikashvili's team at the Institute of Physics, Republic of Georgia, reported the heat capacity of biomolecules such as fibrous and globular proteins, DNA, and RNA using DSC 7, 8, 9. Several teams led by Sturtevant 10, 11, 12, Brandts 13, and Privalov 14, 15, 16focused on the development of the theory and practical applications of DSC to investigate the thermodynamic details of protein unfolding. The value of DSC in studying large supramolecular structures such as phages, chloroplast, phospholipid liquid crystals, and meat proteins have also been reported 17, 18, 19, 20.
DSC has now become commonplace in pharmaceutical research and development for the assessment of the thermal stability of biomolecules, especially proteins 1, 21, 22. This is mostly due to advancements in terms of sensitivity and automation of the instrumentation used to carry out the experiment 23, 24. Here, the final result of the DSC experiment, namely, molar heat capacity as a function of temperature, is used to estimate the following thermodynamic parameters (change in heat capacity (∆Cp), enthalpy (∆H), entropy (∆S), and Gibbs free energy (∆G)) using the equation below:
(1)
(2)
(3)
(4)
(5)
Where Cp is measured heat capacity; q is the heat flow into the test material; T0 and T are the initial and final temperatures of the transition respectively 22, 25. It is also worth noting that the equations above apply to single-domain proteins that can undergo two-state transition and reversible thermal unfolding 22. Analysis of more complex proteins (e.g., non-two-state proteins, and oligomers) have been reported by Friere et al. 26; Johnson et al. 27; and Kasimova et al. 28.
To determine whether a protein undergoes two-state transition or forms intermediates during thermal denaturation, the experimentally derived enthalpy (ΔH; also referred to as calorimetric enthalpy ΔHCal) is compared to the enthalpy derived using the van't Hoff equation given below (also referred to as van't Hoff enthalpy; ΔHVH):
(6)
Where Tm is the midpoint temperature of the transition, R is the ideal gas constant (1.987 cal mol-1 K-1) and Y is the fraction of the protein population in the unfolded state 16, 29. If ΔHVH is equal to ΔHCal; or ΔHVH/ΔHCal is equal to 1, then the protein undergoes an "all-or-none" transition (i.e. two-state transition) 16, 25, 29. However, if ΔHVH is less than ΔHCal; or ΔHVH/ΔHCal is less than 1, the protein undergoes a non-two-state transition 16, 25, 29. The ratio of ΔHVH/ΔHCal also corresponds to the proportion of the protein structure that melts as a thermodynamic cooperative unit or domain 26.
The thermodynamic parameters mentioned above such as ΔG and ΔH provide useful information about the thermal stability of proteins, including biologics 30. However, emphasis will be laid on Tm and ΔH in this publication, as they are the reported values for this protocol. Tm is the mid-point temperature of the transition, where the folded and the unfolded states of the protein are at equilibrium (i.e., ΔG = 0) 25, 31. The higher the Tm of a protein, the higher its thermal stability 31. ΔH corresponds to the area under the peak(s) of the Heat Capacity versus Temperature graph (also known as thermogram) generated at the end of the DSC experiment 16, 25. It is the energy required to denature proteins and can be used to estimate the active fraction (Fa) in a protein formulation (i.e., the proportion of proteins with active conformation in a sample) using the following equation:
(7)
Where ΔH is the experimentally derived enthalpy of the protein sample and Q is the enthalpy determined for a well characterized reference or standardized protein 22. The estimation of Fa is significant for monitoring the real-time stability of products as well as conducting stability studies under stress conditions as required by ICH guidelines 32. Comparison of ΔH also provides information about the compactness of the tertiary structure conformation of a protein 31.
This protocol details a procedure for assessing the thermal stability of proteins in an industrial setting and has been extensively used for the formulation of vaccines. It was developed using an automated differential scanning calorimeter that generates reproducible results for protein concentrations as low as 300 µg/mL.
1. Instrument Start-up
2. Sample Preparation
3. Experimental Parameter Setup
Note: Depending on the instrumentation, samples can be loaded into the cell either manually using a syringe, or automatically using an autosampler. In this case (i.e. an industrial setting), an autosampler is used to save time.
4. Data Analysis
The raw data from most DSC experiments are presented as a heat flux versus temperature graph, as the calorimeter actually measures the difference in the rate of heat flow into the sample solution and buffer 35. Therefore, if both cells (i.e. sample and reference cells) contain identical solutions during an experiment, the raw data from the scan should be a flat line with no observable peaks. Any peak observed can be attributed to instrumentation error (e.g. damaged or contaminated cells), which is why running buffer scans prior to sample analysis is an adequate system suitability test. Figure 1 illustrates the result of a typical buffer scan indicating that the calorimeter was in good working condition prior to sample analysis.
Figure 2 shows the raw data for a DSC experiment carried out on different lots of two protein samples. As implied earlier, the observed peaks are the differences in the heat flux of the samples and their respective buffers. Differences in sample concentration can cause variations in heat capacity recorded by the calorimeter; however, these variations are normalized during sample analysis as per section 4.2 of the procedure. Higher concentrations can also reveal additional thermodynamic domains not contributing to the transition at lower concentrations. In addition, each transition represents a thermodynamic domain that may include one or more structural domains of the protein 36. In this case, the Protein 1 has three structural domains that melt cooperatively.
Figure 3 shows the results generated from the analysis of the raw data for protein 1 and 2 presented in Figure 2, i.e., after baseline subtraction and iterative curve fitting. The resulting thermograms have been normalized for scanning rate (automatically carried out by a pre-set algorithm in the analysis software) and concentration; thus, presenting the results of the experiment in comparable heat capacity versus temperature graphs. The analysis software uses data from the Heat Capacity versus Temperature graphs, such as Tm and ΔCp, to derive other thermodynamic parameters using variations of the equations given above depending on the cooperativity of protein unfolding.
When testing unknown samples, setting the appropriate temperature range is crucial. Otherwise, incomplete thermograms may result, as illustrated in Figure 4. Although Tm from such profiles can be derived, ΔH cannot be accurately determined. Therefore, the sample must be retested with a larger temperature range to completely capture the thermal transition. Some proteins also readily form aggregates after complete denaturation, resulting in an increasing post-transition heat capacity: this often appears as an incomplete thermogram as illustrated in Figure 2B. However, retesting with a higher final temperature can help confirm whether there is an occurrence of a conformational transition at that region of the thermogram or it is merely the heat absorbing effect of protein aggregates.
Thermal stability is one of the most significant physical properties of proteins and protein-based products in the industry 37. In pharmaceuticals, it is used to determine the stability of biologics under different conditions, including formulation buffers and environmental factors such as humidity and temperature. It is also used to monitor key manufacturing step (e.g., purification and detoxification) to ensure conformational consistency between production lots. Figures 5 and 6 illustrate the use of DSC to examine the effects of chemical detoxification and storage conditions respectively on the stability and structural conformation of two different proteins. The significant differences in Tm and ∆H indicate conformational changes and protein degradation respectively. In addition, the loss of the third transition in Figure 6 further illustrates the degradation of a domain which was confirmed by a decrease in molecular weight when the samples were analyzed using Size-Exclusion Chromatography with multi-angle light scattering (SEC-MALS) (data not shown).
Figure 1: Buffer Scans. The similarity in gradient of each scan with no observable peaks indicates that the instrument is in good working condition and generated reproducible results. Please click here to view a larger version of this figure.
Figure 2: Raw Data Collected from DSC Experiments. These graphs are good representations of the unanalyzed (raw) data acquired after experimental runs (i.e. prior to baseline subtraction and curve fitting). Each line represents a production lot. Protein 2 tends to aggregate more readily upon heating, resulting in an increase in heat capacity above 100 °C in the post-transition region of the thermogram. Please click here to view a larger version of this figure.
Figure 3: Analyzed DSC Data. These graphs are good representations of analyzed DSC data (i.e. after baseline subtraction and curve fitting). The blue line represents the thermogram after baseline subtraction, while the red line represents the curve with the best fit to the thermogram. (A) The Tm and ΔH for Protein 1 Sample # 12 are 80.16 °C and 1.69 x 106 cal/mol respectively. (B) The Tm and ΔH for Protein 1 Sample# 13 are 80.15 °C and 1.71 x 106 cal/mol respectively. (C) The Tm and ΔH for Protein 2 Sample # 21 are 75.01 °C and 4.08 x 106 cal/mol respectively. (D) The Tm and ΔH for Protein 2 Sample # 22 are 75.67 °C and 4.22 x 106 cal/mol respectively. Please click here to view a larger version of this figure.
Figure 4: An Incomplete Thermogram. Raw data collect for Protein 1 analyzed at an inadequate temperature range. The final temperature of experiment was set to 90 °C which did not accommodate the entire transition profile of the protein as compared to the experiment for Figure 2A which was set to 120 °C. Please click here to view a larger version of this figure.
Figure 5: Analyzed Data Showing the Effect of Chemical Detoxification on the Tertiary Structure of Protein 1. (A) Protein 1 is a toxin in its native conformation and has its Tm at 56.84 °C and ΔH at 2.57 x 105 cal/mol. (B) The detoxified form of Protein 1 (i.e. toxoid) has Tm and ΔH values of 81.01 °C and 1.89 x 106 cal/mol respectively. Thus, it can be concluded that the detoxification step introduced some form of variation to the structural conformation of Protein 1 which confers greater stability (higher Tm) to its detoxified form. Please click here to view a larger version of this figure.
Figure 6: Analyzed Data Showing the Effect of Storage Conditions on the Conformation of Protein 3. These graphs illustrate the effect of storage temperature (2 – 8 °C) on the stability and tertiary structure of Protein 3 over 30 weeks. The Tm and ΔH values for Protein 3 at the 8th (A) and 38th week (B) of storage are given in Table 1 below. Please click here to view a larger version of this figure.
Sample | Tm 1 (°C) | ∆H 1 (cal/mol) | Tm 2 (°C) | ∆H 2 (cal/mol) | Tm 3 (°C) | ∆H 3 (cal/mol) |
Protein 3 stored at 2-8 °C for 8 weeks | 61.91 | 1.71 x 105 | 75.54 | 2.17 x 105 | 90.29 | 4.17 x 105 |
Protein 3 stored at 2-8 °C for 38 weeks | 61.87 | 1.66 x 105 | 75.18 | 1.46 x 105 | 90.22 | 5.76 x 105 |
Table 1: Tm and ΔH Values for Protein 3 at the 8th and 38th Week of storage at 2 – 8 °C. Although the Tm values at both time points are similar, the difference in the ΔH values indicates that the tertiary structure of Protein 3 has degraded over 30 weeks under the specified storage condition.
This procedure has been successfully incorporated in various characterization test packages, including stability and product comparability studies 21. In real-time stability studies, DSC is used to monitor the Tm, as well as estimate the Fa of biologics over time to determine their shelf-life. With regards to product comparability, it is used to assess the impact of process and facility change as well as the effect of key manufacturing steps on the structural conformation of produced lots. This typically involves the direct comparison of the ΔH of produced lot to a reference product that has been designated as the ideal product. In addition, DSC has proven to be a useful analytical tool for product formulation studies 37. The Tm of a protein in different buffers and at different concentrations can be used to determine the formulation that proffers the most stability to the protein.
To ensure the reliability of this method and objectivity of its results, it is important to keep testing parameters consistent from run to run within the same study (e.g., vaccine formulation study). However, the procedure can be modified to accommodate differences in the physical properties of various proteins. An example of a modification that can be made is altering the scanning rate of the experiment 38, 39. Proteins that were prone to forming aggregates when heated were examined at a faster scanning rate (e.g., 120 °C/h) to avoid the contribution of aggregates to the thermal transition profile as well as clogging the capillaries of the calorimeter. It is worth noting that scanning rate can influence the outcome of a DSC experiment 38. Broadening of the thermal transition peak has been observed with increasing scanning rates in some proteins; however, Tm remained fairly constant 38. Furthermore, the dialysis and degassing steps for sample preparation are also very crucial for accurate results 31. Dialysis ensures that the only difference in the composition of the sample and the buffer is the protein; thus, all the excess heat absorbed by the sample can be attributed to the heat capacity of the protein. Degassing ensures precise volume analysis, as the extrapolation of the thermodynamic parameter assumes that the unfolding event is occurring under constant volume and pressure 31. The constant pressure portion of the assumption is accounted for by nitrogen pressurization of the system as per section 1.1 of the procedure.
In comparison to other methods of determining the stability of protein conformations such as Circular Dichroism (CD) and Fluorescence Spectroscopies, DSC offers a number of advantages in a commercial setting including cost and time savings. Firstly, the adiabatic design of a differential scanning calorimeter allows for the measurement of thermal stability with better temperature precision as compared to measurements with instrumentation for CD and Fluorescence Spectroscopies 6. Secondly, unlike CD, the accuracy of DSC data is not dependent on the helicity of the protein 39, 40; however, CD provides additional information about the unfolding of the secondary structure, which would be complimentary to the DSC 41. Additionally, the pressurization of the DSC system allows for testing with a wide temperature range without boiling the sample; thus, a wide range of proteins can be tested by DSC.
While DSC is a relatively fast and straightforward approach to determine the thermal stability of biologics, it is not without limitations. First, the baseline subtraction step introduces some form of human inconsistency into the raw data analysis; thus, variations in results may be observed among different users. Second, differential scanning calorimeters have minimum concentration limits which might be difficult to achieve at bulk manufacturing scale. Third, the ΔH of irreversible thermal denaturation is not absolute; which implies that derived ΔG (an indicator of protein stability) in similar scenarios can be misleading. Furthermore, the method works best for purified samples. Presence of impurities may either cause a shift in the Tm if there is an interaction with the protein under investigation, or appearance of new thermal transitions if there is no interaction. In any case these extra features on the thermograms can be wrongly attributed to samples, thus impacting the interpretation of results. Despite these limitations, DSC remains a reliable method that can provide detailed thermodynamic information about the protein unfolding process if implemented properly 42.
In conclusion, DSC offers considerable advantage as a conformational readout tool for vaccine products and their intermediates. The two parameters, Tm and ΔH, collected for an array of lots of the same product can become an empirical baseline that can be used to examine the impact of process changes, formulation, and storage conditions on the tertiary structure and stability of protein and viral antigens 21, 43.
The authors have nothing to disclose.
The authors are very thankful to Joseph Mancini (formerly with GE Healthcare), Pawel Czudec, Thomas Cage (Malvern Instruments limited) for their role in the installation and training on the differential scanning calorimeter, Sasmit Deshmukh and Webster Magcalas for their discussions.
Differential Scanning Calorimeter | Malvern Instruments Ltd | 28428948 (Via GE Healthcare) | Has an autosampler for automated dispensing of samples into the cell to reduce human effort and errors. |
Contrad 100 | Decon Laboratories Inc | 1504 | Dilute with water to 20% before use |
500µL Polypropylene round bottom 96 well plate | Canadian Life Science | ML072100 | Equivalent plates from other suppliers (e.g., VW) can also be used |
MicroCal ThermoVac | Malvern Instruments Ltd | N/Ap | provided with the Cap VP DSC |
Biosafety cabinet | Labconco | Logic+ – A2 | biocontainment laminar flow cabinet for sample preparation |
Slide-A-Lyzer dialysis cassette | Thermo Scientific | 66810 or 66380 | to equilibrate the sample and buffer |