OR WAIT 15 SECS
Volume 29, Issue 11
This article demonstrates that a modified SDS–PAGE can be easily used as a tool for quantifying the degree of protein degradation.
The degradation of proteins during cleaning is of great interest to biologics manufacturers as it can support the validation of the cleaning processes used by these manufacturers. Degradation of protein products during cleaning potentially provides a strong argument that the risk to patients of product carryover into another product is minimal. Degradation cannot be assumed, however, and must be demonstrated for this phenomenon to be used for defending cleaning validation programs (1).
This article discusses how to quantify protein degradation; what the effects of sodium hydroxide, sodium hypochlorite, and hydrogen peroxide are on protein degradation, including analysis of a design of experiment on the effects of concentration, temperature, and time of exposure; and how this information can be used to develop a customized, cost-efficient, and science-based cleaning process to support cleaning validation.
Since the publication of the International Society for Pharmaceutical Engineering’s (ISPE) Risk-Mapp Guide in 2010 (2), many companies have adopted the use of the permitted daily exposure (PDE) or acceptable daily exposure (ADE) for setting cleaning validation limits, and the European Medicines Agency (EMA) is now requiring their use (3). In some cases, particularly for some proteins, the ADE-based acceptance limits calculated are below the detection limits (DL) of the current methods. In response, some biologics manufacturers have argued that the PDE/ADE does not apply to them, because their protein molecules are denatured and degraded by their cleaning processes, and that alternate approaches to the ADE should be used. A previous article in this series explored why alternative approaches to using the PDE/ADE for determining cleaning validation limits cannot be justified and proposed an approach based on the ADE that could be used to demonstrate that the risk is acceptable in these cases (1). As explained in this article, protein degradation studies can provide valuable information that can support the acceptance of cleaning validation for such proteins while still using the ADE-based limits. Consequently, a reliable and valid methodology for demonstrating protein degradation under the conditions of cleaning is needed.
Although there have been a few studies published on protein degradation concerning pharmaceuticals (4, 5), these articles did not provide detailed descriptions of the methodology used. Both of these publications demonstrated that degradation may be complete, partial, or not at all. Both of these studies used Sodium Dodecyl Sulphate Polyacrylamide Gel Electrophoresis (SDS-PAGE) for the analysis.
SDS–PAGE is a repeatable and reproducible protein analysis method with good detectability. However, it partially destroys protein molecules during its process by treating proteins with a strong detergent (SDS) as well as a 90 °C-95 °C heating step, which can confuse the interpretation of the results of the degradation study. The secondary and tertiary structure is disrupted during the SDS–PAGE analysis procedure, even without heating, due to the presence of SDS molecules, and the use of reducing agents (such as beta-mercaptoethanol) will also break disulfide bonds. Native PAGE preserves the original protein molecular structure, but it is a relatively more challenging and unstable method. This method is mostly obsolete in both academic and industry laboratories after the introduction of its successor, the SDS–PAGE itself. In protein degradation studies, however, a non-degrading analysis method will provide more valuable information. For several reasons, a combination of the two methods is more scientifically solid and allows a deeper and more holistic understanding of what is actually happening in the reacting tube.
In Native Page (see Figure 1), the relative distance of protein bands depends on their molecular weights and isoelectric points, which makes it hard to perform meaningful calculations due to the inherent differences in isoelectric points among proteins. However, introducing the SDS molecule, which uses a detergent with a strong electrical charge, negates the isoelectric point issue and allows a quantitative method to calculate both the molecular weight and protein volume. In degradation studies, a combination of the two methods was used by using SDS–PAGE as a quantitative method to calculate the molecular weight and volume of residues, and by using Native PAGE to ascertain that the degradation of the protein was by the combination treatments rather than the SDS molecule.
Figure 1. One Native PAGE example. The monoclonal antibody proteins (mAb1
and mAb2) stayed in a single band of high molecular weight, showing that the
Native PAGE method protects the protein integrity. Lanes 1-5, molecular marker;
lane 6, mAb1 (original concentration); lane 7, mAb1 (1/10 original concentration);
lane 9, mAb2 (original concentration); lane 10, mAb2 (1/10 original concentration). [All figures courtesy of the authors]
The 90 °C-95 °C heating step in SDS-PAGE is designed to unfold the higher-order structure of protein molecules and help the SDS to cover the entire molecule. This heating also results in the protein molecule breaking into its characteristic smaller sub units, which typically results in more bands in the gel yielding a “fingerprint” for the protein. Because degradation studies of cleaning parameters are conducted to determine if degradation has occurred and the degree of the degradation, results of the degradation study may be confounded by this heating step. To eliminate this potential interference and also yield better linearity, the heating step in the standard SDS-PAGE method was eliminated from the procedure.
Figure 2 shows an SDS–PAGE of a monoclonal antibody (mAb) and Bovine Serum Albumin (BSA). Notice that there are dramatic changes in the mAb gels with many additional bands after one minute but little change in the BSA even after five minutes. The effect of heating on protein molecules largely varies depending on their characteristics. The heating step in the sample preparation can cause additional degradation, in particular when the sample contains cleaning agents, which would make it difficult to tell the degradation effect of the cleaning agent during the cleaning process from the one attributable to the sample preparation. The problem can be resolved by excluding the heating step from SDS–PAGE sample preparation.
Figure 2. Analysis of the effects of Sodium Dodecyl Sulphate Polyacrylamide Gel
Electrophoresis (SDS–PAGE) heating step (95 °C). Lanes: 1, molecular marker;
2, mAb (unheated); 3, mAb after 1 min; 4, mAb after 3 min; 5, mAb after 5 min; 6,
empty; 7, Bovine serum (unheated); 8, Bovine serum after 1 min; 9, Bovine serum
after 2 min; 10, Bovine serum after 5 min.
All Native-PAGE and SDS–PAGE were conducted with the BioRad Mini-PROTEAN Tetra System. The BioRad Mini-PROTEAN Tetra System is a standard gel construction and electrophoresis system in molecular and cellular biology laboratories. It provides good detection of protein residues, and the method is repeatable and reproducible with properly trained analysts.
For more accurate results, gels using a gel preparation kit (Sigma Aldrich) were constructed rather than using pre-cast gels. Due to the high molecular weight of antibodies, the gels were constructed of 7.5% polyacrylamide (PAA) concentration. This proved to be the best working concentration for antibody analysis because gels of lower concentrations break apart more easily, and higher concentration of PAA brings too much restriction on the protein molecules to move and yields poor linearity. To determine the ideal gel concentration, a molecular marker (Sigma-Aldrich) was run with the initial gels.
For aesthetic purposes, most PAGE studies will dilute proteins with a certain amount of distilled deionized water (ddH2O) to yield clear bands without any overlapping. This, however, leaves room for the possible misinterpretation of degradation study results. In some cases, a ½ diluted protein will clearly show residues on the gel after caustic treatment while a 1/10 diluted protein will show no residues at all (see Figure 3). However, the control group of 1/2 and 1/10 diluted proteins look almost identical-due to the good detectability of the PAGE methods. Thus, a universal principle must be established-that the proteins shall not be excessively diluted to a point that will alter the interpretation of results.
Figure 3. Demonstration of the effect of dilution. Lanes: 1: MW Marker; 2:
mAb1 (original concentration), not treated; 3: mAb1 (original concentration),
caustic treatment*; 4: mAb1 (1/5 original concentration), not treated; 5: mAb1 (1/5
original concentration), caustic treatment; 7: mAb2 (original concentration), not
treated; 8: mAb2 (original concentration), caustic treatment; 9: mAb2 (1/5 original
concentration), not treated; 10: mAb2 (1/5 original concentration), caustic treatment
caustic treatment*: treated with 0.1M NaOH solution, room temperature, 10 minutes.
The gels were analyzed using the open source software Gelanalyzer 2010. This software provides automatic protein volume and molecular weight calculations based on given standards. By adjusting the baseline curve, accurate results can be obtained by eliminating background noises and other possible variations. The selection of gel lanes and bands, however, relies on manual operation, because the automatic process usually ignores insignificant yet solid bands and may recognize neighboring bands as a single one.
One major flaw with the current SDS–PAGE method is that it provides good qualitative but poor quantitative results. By abandoning the heating process during the SDS–PAGE protocol, however, any aggregation and degradation of proteins was reduced, significantly limiting the variability. Additionally, the negative charge imposed on the protein by the SDS molecule allowed the molecular weight of the proteins to be quantitatively determined using the Gelanalyzer software.
The gels in Figure 4 showed that the current method has good linearity from 5X concentration to 1/80x concentration for the mAb 1. In lane 8, one can see that a 1/160 working concentration of mAb1 is barely showing any bands, meaning the limit of detection of mAb1 using SDS-PAGE is between 1/80 to 1/160 of its working concentration. Similar methods can be applied to different mAbs using the SDS–PAGE method.
Figure 4. Linearity tests. Lanes: 1: MW Marker; 2: mAb1 (5x working
concentration); 3: mAb1 (working concentration); 4: mAb1 (1/2 working
concentration); 5: mAb1 (1/10 working concentration); 5: mAb1 (1/20 working
concentration); 6: mAb1 (1/40 working concentration) 7: mAb1 (1/80 working
concentration) 8: mAb 1 (1/160 working concentration).
The analysis of the linearity tests showed promising results (Figure 5). Without heating, any variation brought by additional aggregation and degradation was limited, and the R-square of the linearity improved to 0.996. Using this method, the volume of protein residues in future degradation studies was quantified with good accuracy.
Figure 5. Linearity result.
Currently, many biological manufacturers heat cleaning solutions to approximately 90 °C to help clean protein products that can lead to degradation. The effect of heating on proteins, however, has not been fully researched. Two monoclonal antibodies (mAb1 and mAb2) were used in the following experiments to study the degrading effects on these similar proteins. Both mAb1 and mAb2 were exposed to 65 °C, 75 °C, and 85 °C heat for five minutes, and the protein volumes were acquired based on the previously described method with adjusted linearity calculation.
The result revealed that both mAb1 and mAb2 are subject to complete degradation within one minute when exposed to 85 °C. However, mAb2 was more heat-resistant at 65 °C and 75 °C. With a treatment of 75 °C for five minutes, mAb1 will be nearly completely degraded while mAb2 will still be approximately 50% intact.
All proteins are constructed of roughly 20 different amino acids, and the stability of a protein will largely depend on its constituent building blocks. Due to the fact that the chemical structure of different amino acids varies considerably, there will be inherent chemical and physical differences between proteins, including stability and resistance against oxidizing and alkaline agents.
To theoretically estimate the stability of an organic substance, the standard enthalpy of formation (ΔƒHÏ´) is frequently used to quantify the change of enthalpy to form one mole of the substance from its forming elements. The more energy that is released to form a substance, the more stable it is. It may not be possible to accurately calculate the stability of a specific protein as there will be additional hydrogen bonds, ionic bonds, and disulfide bridges in three-dimensional (3D) protein structures, but one can certainly expect reasonable physical and chemical difference between proteins by demonstrating different ΔƒHÏ´ among amino acids.
To determine the effect of an oxidizer on degradation, the two proteins (mAb1 and mAb2) were exposed to 500 parts per million (ppm) sodium hypochlorite (NaOCl)-an oxidizer solution commonly used in the biological manufacturing industry. The results demonstrated that both mAb1 and mAb2 are subject to significant degradation upon two minutes of exposure to 500 ppm NaOCl. These data indicate that cleaning process could be designed to quickly and sufficiently degrade proteins without prolonged exposure of oxidizer solution to manufacturing equipment.
A design of experiment (DoE) was performed based on the previous results to determine the design space for the three common cleaning factors (time, temperature, and concentration) for each mAb and for each cleaning agent. The three parameters of protein degradation play different roles and should be prioritized differently in cleaning processes according to their effects on protein degradation as supported by data. By probing the proper design spaces using upper and lower values of each factor in all possible combinations, the DoE will be able to reveal and measure the individual effects of these factors and their interactions within an actual cleaning process.
From earlier data, both mAb1 and mAb2 start to degrade at 75 °C without any presence of chemical degradants. Thus, the upper level of temperature for the DoEs was set at 70 °C, and the lower temperatures were set at 70 °C and 20 °C to allow the detection of the degrading effects of time and concentration. Also based on previous results, the upper and lower times of exposure were set at 10 minutes and 2 minutes. Pareto charts were used to show the magnitude of each factor and interaction and their significance as compared to an error term.
Based on common industrial practices, in the first DoE on NaOCl, the upper and lower levels of concentration were set at 500 ppm and 1500 ppm (see Table I).
|High||70 °C||10 minutes||1500 ppm|
|Low||25 °C||2 minutes||500 ppm|
The result of the mAb2 DoE demonstrated that concentration was a key factor at lower temperatures and time. However, mAb1 is much more susceptible to degradants than mAb2 because all lanes under 500 ppm NaOCl were degraded as shown in Figure 6. In theory, there should be something left from lane 3 to lane 10. Figure 6 shows the misleading result, and Figure 7 shows why the result is misleading. In Figure 7 for mAb1, none of the factors matter. However, at least one factor should, but the concentration setting was too high and the result was masked. This also tells us mAb1 and mAb2 have significantly different resistance to degradation.
Figure 6. mAb1 DoE Gel. Lane 1-2: Linearity Calibration. Lane 3-10: DoE
treatments. Factor 1: Temperature. Factor 2: Concentration. Factor 3: Time. Lane
3: LLL; Lane 4: HLL; Lane 5: LHL; Lane 6: HHL; Lane 7: LLH; Lane 8: HLH; Lane
9: LHH; Lane 10: HHH. All bands have disappeared due to mAb1’s high sensitivity
towards NaOCl solution.
To explore this further, two new DoEs with lower concentration settings were done. The upper level of concentration was reduced to 500 ppm and 300 ppm, while the lower was reduced to 300 ppm and 100 ppm.
Figure 7. Pareto charts of effects for NaOCl–Run 1. ppm is parts per million.
The results of the second DoE on mAb1 at the new NaOCl concentrations showed that between 300-500 ppm NaOCl solution, the temperature played a more important role, while the effect on NaOCl concentration was more significant between 100-300ppm. This revealed that NaOCl starts to degrade mAb1 at 100 ppm, and the degrading effect rapidly increases to around 300 ppm and is more significant than temperature at 70 °C. However, the degrading effect of NaOCl at 300-500ppm reached a plateau, with the 70 °C temperature playing a relatively more important role (see Figure 8).
Figure 8. Pareto charts of effects for NaOCl–Run 2. ppm is parts per million.
For these two mAbs, these data demonstate that NaOCl concentrations higher than 300 ppm do not have any significant increase in degradation. Based on this, a cleaning process could be designed using 300 ppm NaOCl and lower temperatures to prevent exposure of manufacturing equipment to high concentrations of NaOCl solution (reducing possible corrosion effects) and still provide excellent degradation.
These studies showed that NaOH solutions at 0.1-0.5 mol/L do not have any more significant degrading effect for these two mAbs than simply applying heat at 70 °C. Because a 100-300ppm NaOCl solution at 70 °C is significantly more degrading than 0.1-0.5 mol NaOH (4000-20,000 ppm) at 70 °C, one can see that NaOCl is a much stronger degradant than NaOH (see Figure 9).
Figure 9. Pareto charts of effects for NaOH.
The previous results showed reasonable protein degradation with the combined 70 °C heating treatment with oxidizer/alkaline exposure, with the understanding that heating at 70 °C alone may not have a sufficient degrading effect. Surprisingly, the DoE of 0.5-1.5% hydrogen peroxide (H2O2) solution showed little degradation with insignificant differences between treatments. In summary, H2O2 does not appear to be as good a degradant as compared to NaOCl or even NaOH (see Figures 10 and 11).
Figure 10. H2O2 DoE treatment. Factor 1: Temperature. Factor 2: Concentration.
Factor 3: Time. Gel on the left: Lane 1-4: DoE treatment. Lane 1: HHH; Lane 2: HLH;
Lane 3: HLL; Lane 4: HHL. Lane 5-10: Linearity Calibration. Gel on the right: Lane
1-6: Linearity Calibration. Lane 7: LHH; Lane 8: LHL; Lane 9: LLH; Lane 10: LLL.
Figure 11. Pareto charts of effects for H2O2.
A previous article demonstrated how the risk from cleaning processes must be evaluated against the ADE/PDE for biopharmaceuticals (4). The article also discussed how risk could be evaluated through combining data from multiple sources, including protein degradation studies, rather than simply relying on passing swab or rinse sample results. Taking this approach requires a quantitative analysis for protein degradation studies, and this article proposes one possible method. This method can also be used to demonstrate the effects of oxidizers such as sodium hypochlorite or hydrogen peroxide on the biopharmaceuticals of interest. These bench-scale studies can be used to increase understanding of the probability that a cleaning process has reduced residues to safe levels with data rather than assumptions.
It cannot be simply assumed that proteins are denatured or destroyed by temperature or alkaline conditions; as in this article, differences in degradation could be shown for two closely related proteins. Different proteins will have different degradation behavior, which should be known and demonstrated. For example, a study in this laboratory showed that Bovine Serum Albumin possesses a remarkable heat-resistance quality (Figure 12). Under the treatment of 75 °C for 10 minutes, both mAb1 and mAb2 degrade to less than 10%, while the bovine serum slowly drops to 30%. Furthermore, as no significant degradation could be observed after five minutes, one can conclude it may require a temperature higher than 75 °C to sufficiently degrade bovine serum.
Figure 12. Bovine Serum Heating Test. Lane 1-5: Linearity Calibration. Lane 6:
75 °C, 1min. Lane 7: 75 °C, 2min. Lane 8: 75 °C, 3min. Lane 9: 75 °C, 5min. Lane
10: 75 °C, 10min.
Moreover, there must also be evidence provided that the cleaning process will be effective under all circumstances and for all product-contact parts of the equipment. Often, examples of protein degradation are presented using bioreactors or other vessels that are more easily cleaned, while finished product filling equipment may present a more difficult cleaning and degradation situation as they may be cleaned with different cleaning procedures. In addition, biopharmaceuticals are subsequently formulated to be stabilized, and these finished formulations may present a higher risk than the bulk API and should be evaluated as well.
This article demonstrates that a modified SDS–PAGE can be easily used as a tool for quantifying the degree of protein degradation of biopharmaceutical products under various cleaning conditions using the Gelanalyzer 2010 open source software. It was shown that an oxidizing agent such as sodium hypochlorite at low concentrations can be used to significantly degrade the mAbs in this study, thereby effectively reducing the potential patient exposure to these proteins to extremely low levels. While Hydrogen peroxide was not found to be as effective as Sodium hypochlorite for these two particular mAbs, the potential of this oxidizing agent should always be considered for degradation studies for hazardous biologics. Protein degradation data as generated in this article can be effectively used for risk evaluation in cleaning validation programs (4) to demonstrate the degree and completeness of degradation under various cleaning conditions.
The authors wish to thank Alfredo J. Canhoto, PhD and Andreas Flueckiger, PhD for reviewing this article and for their insightful comments and helpful suggestions.
1. A. Walsh, Biopharm International 28 (9) (September 2015).
2. ISPE, ISPE Baseline Guide: Risk-Based Manufacture of Pharmaceutical Products (Risk-MaPP), First Edition (ISPE September 2010).
3. EMA, Guideline on Setting Health Based Exposure Limits for Use in Risk Identification in the Manufacture of Different Medicinal Products in Shared Facilities, www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2014/11/WC500177735.pdf.
4. K. Kendrick, A. J. Canhoto, and M. Kreuze, Journal of Validation Technology (Summer 2009).
5. N. Rathore, et al., Biopharm International 22 (3) (March 2009).
Vol. 29, No. 11
Pages: 38–44, 53
When referring to this article, please site it as X. Wang et al., "Development of a Technique for Quantifying Protein Degradation," BioPharm International 29 (11) (November 2016).