Effects of pH and Buffer Concentration on the Thermal Stability of Etanercept Using DSC and DLS

Abstract

The protein size, electrical interaction, and conformational stability of etanercept (marketed as Enbrel®) were examined by thermodynamic and light scattering methods with changing pH and buffer concentration. As pH of etanercept increased from pH 6.6 to 8.6, electrical repulsion in the solution increased, inducing a decrease in protein size. However, the size changed less in high buffer concentration and irreversible aggregation issues were not observed; in contrast, aggregates of about 1000 nm were observed in low buffer concentration at the pH range. Three significant unfolding transitions (T_m) were observed by differential scanning calorimetry (DSC). Unlikely to T_m1, T_m2 and T_m3 were increased as the pH increased. Higher T_m at high buffer concentration was observed, indicating increased conformational stability. The apparent activation energy of unfolding was further investigated since continuous increase of T_m2 and T_m3 was not sufficient to determine optimal conditions. A higher energy barrier was calculated at T_m2 than at T_m3. In addition, the energy barriers were the highest at pH from 7.4 to 7.8 where higher T_m1 was also observed. Therefore, the conformational stability of protein solution significantly changed with pH dependent steric repulsion of neighboring protein molecules. An optimized pH range was obtained that satisfied the stability of all three domains. Electrostatic circumstances and structural interactions resulted in irreversible aggregation at low buffer concentrations and were suppressed by increasing the concentration. Therefore, increased buffer concentration is recommended during protein formulation development, even in the earlier stages of investigation, to avoid protein instability issues.

Advances in biotechnology in the last decades have made it possible to produce a number of proteins for therapeutic uses. Protein pharmaceuticals have drawn a lot of attention by virtue of their effectiveness and patient needs. Consequently, therapeutic proteins have overtaken low molecular weight compounds as the objects of novel drug development in the last decade.^1,2) However, these proteins have complicated structures, marginal stability, and high sensitivity to degradation. Moreover, the proteins must be stabilized in solution to maintain their efficacy and benefits.^3–6)

Protein aggregates are problematic because of possible activity loss and immunogenicity issues.^7,8) Therefore, it is important to determine the causes of protein aggregation in order to minimize it to an acceptable level. Mutual interactions between pH and various buffers with lysozyme were investigated using a robust design (RD) method in a previous study.⁹⁾ Acceptable pH range and buffer were selected based on thermodynamic properties from differential scanning calorimetry (DSC). In addition, selection of suitable buffer system of human epithelial growth factor based on DSC and dynamic light scattering (DLS) was investigated recently.¹⁰⁾ In order to obtain rapid and comprehensive physicochemical characterizations of biopharmaceuticals, the results from various biophysical methods during the biopharmaceutical development should be combined for the proper understanding of the characteristics of the target protein and its aggregation.^11–14)

Proteins are stable against aggregation over a narrow pH range, and this stable pH range for proteins must be elucidated. Otherwise, the protein will aggregate or unfold rapidly in a solution with a pH outside the acceptable range. Several studies have evaluated the influence of pH on proteins.^15–17) It is important to understand the destabilization of therapeutic proteins: why they are stable over a narrow pH and how to find the acceptable pH range. In addition, buffer concentration is an important factor to consider since the buffer strength can also affect the physical stability of proteins. This preludes the improved physicochemical stability of biopharmaceutical formulations in solution.

In the early stages, the development of protein formulations faces many difficulties since each protein has its own unique physicochemical properties and hence displays unique stability behavior.¹⁴⁾ The purpose of the present study is to investigate the effects of pH and buffer concentration as formulation factors on etanercept stability using DSC and DLS. Etanercept was chosen as a model protein to understand the mechanism of biophysical stability according to medium conditions and to look for an acceptable protein solution.

MATERIALS AND METHODS

Materials

Etanercept, a fusion protein prepared by recombinant DNA engineering, has two soluble human 75-kDa tumor necrosis factor receptors (TNFRs) linked to the F_c portion of immunoglobulin G1 (IgG1). The commercial product, Enbrel® (25 mg lyophilized Etanercept with mannitol, sucrose, and Tris), was used as a model drug. Its molecular weight is 51234.9 g/mol and it is comprised of 934 amino acids. After the etanercept was dissolved in water for injection, it was dialyzed for 24 h at 4°C in a Cellu Sep® H1 cellulose membrane with a MW cut-off of 10000 Da (Membrane Filtration Products, Seguin, TX, U.S.A.); 10 mM and 30 mM Tris buffers with pH range from 6.6 to 8.6 were selected for this study. Each buffer’s pH was monitored by pH meter (Metrohm 827 pH Lab, Zurich, Swiss) before and after dialysis (pH ±0.02). Protein concentration in solution was adjusted to 5 mg/mL before dialysis and was determined by UV absorption measurement (Optizen Pop, Mecasys Co., Ltd., Korea) at 280 nm after dialysis.

Dynamic Light Scattering (DLS) and Zeta Potential

Electrostatic interaction of etanercept at various concentrations was evaluated using a Zetasizer Nano ZS90 apparatus (Malvern Instruments, Worcestershire, U.K.). The apparatus was equilibrated at a working temperature of 10°C and 1 mL of each prepared sample was measured using a disposable sizing cuvette (Sarstedt, Numbrecht, Germany) for hydrodynamic size and a disposable capillary cell (Malvern Instruments) for zeta potential. Each sample was measured five times with an intervening interval of 30 s. Zeta average size, polydispersity index (PDI), and zeta potential were calculated from the auto-correlated function using Zetasizer software version 6.32 (Malvern Instruments).

Differential Scanning Calorimetry (DSC)

DSC measurements were conducted using a VP-DSC Microcalorimeter (Microcal, Northampton, MA, U.S.A.) having 0.51471 cm³ twin cells for the reference and sample solutions. Prior to the DSC measurements, the sample and the reference were degassed under vacuum while being stirred. Tris buffers prepared for dialysis were used as references to obtain baselines. Measurements were repeated three times at a scan rate of 0.75 K/min, 1 K/min, 1.25 K/min, and 1.5 K/min for every sample. The temperature ranged from 15 to 120°C and the final thermogram was processed by subtracting the baseline from the sample thermogram. The thermo compensation curves were evaluated using the Microcal LLC DSC plug-in for the Origin 7.0 software package provided with the equipment. These results were fit to a multistate model with three transitions to calculate parameters including transition melting point (T_m) values and enthalpies (∆H).

Apparent Activation Energy (E_app)

From the Lumry and Eyring model IV in situation C, several useful relationships were adopted to find the scanning rate effect on T_m.^9,18,19) At constant the following equation was used.

  

In this equation, v is the scanning rate or heating rate of the protein solution, E_app is the apparent activation energy, and R is the gas constant. The plot of 1/T_m against ln(1/T_m²) gives a straight line whose slope can be used to determine E_app. It is assumed that the equilibrium between the native and the unfolded states is always established and the amount in the unfolded state is always very low. Each prepared protein solution was scanned at four different heating rates.

RESULTS

Effect of pH and Buffer Concentration on Zeta Size and Zeta Potential

DLS was used to investigate the effects of pH and buffer concentration on biophysical properties of etanercept. It detects the electrostatic interaction between protein particles through zeta potential and formation of aggregates through light scattering in the nano-size range. Table 1 lists average size, zeta potential, and PDI of etanercept at selected pH range with different buffer concentrations (10, 30 mM). The size range of monomer in this study was 3.1–6.7 nm and was similar to the previously reported 7.1 nm hydrodynamic radius of etanercept.²⁰⁾ The average size of etanercept in 10 mM Tris buffer (pH range from 6.6 to 8.6) was 5.86 to 3.14 nm. The size decreased continuously with increasing absolute value of zeta potential and PDI. Similarly, average size of etanercept in 30 mM Tris buffer at the pH range decreased slightly with increasing absolute value of zeta potential. At two concentrations, zeta potential was doubled when pH was increased by 2 digits.

Table 1. Average Size, Zeta Potential, and PDI of Etanercept in Various pH and Buffer Concentrations Measured by DLS

10 mM Tris bufferpH	Average size		Zeta potential		Polydispersity index
	r.nm	±S.D.	mV	±S.D.	Average	±S.D.
6.6	5.86	0.36	−11.23	0.67	0.31	0.00
7.0	4.64	0.23	−11.03	0.75	0.27	0.05
7.4	4.08	0.11	−14.37	0.95	0.39	0.06
7.8	3.79	0.22	−16.93	0.55	0.53	0.02
8.2	3.29	0.35	−18.07	0.75	0.63	0.09
8.6	3.14	0.10	−20.07	1.21	0.57	0.05
30 mM Tris bufferpH	Average size		Zeta potential		Polydispersity index
	r.nm	±S.D.	mV	±S.D.	Average	±S.D.
6.6	6.72	0.10	−7.39	0.18	0.96	0.07
7.0	6.61	0.03	−7.19	0.73	0.26	0.06
7.4	6.40	0.13	−8.85	0.31	0.36	0.02
7.8	6.20	0.18	−10.48	1.03	0.33	0.02
8.2	6.00	0.18	−13.84	0.18	0.47	0.11
8.6	5.63	0.11	−14.92	0.66	0.58	0.06

r.nm, radius of etanercept in nanometer. S.D., standard deviation.

Zeta potential may represent the electrical repulsion of proteins affecting neighboring proteins and it increased in absolute value with increasing pH (Table 1). Figure 1(a) shows zeta potential plotted against pH and an isoelectric point (pI ) was estimated when the straight line was extrapolated to the y axis. The pI where the net charge of etanercept was negligible and least stable was 4.44 in 10 mM and 5.13 in 30 mM Tris buffer. It seemed that pI shifted from 4.44 to 5.13 due to buffer concentration. As the pH of etanercept in 10 mM Tris buffer was far from the pI, higher absolute zeta potential was observed than in 30 mM at the same pH. This might cause a significant decrease in protein size.

Fig. 1. (a) Plot of Zeta Potential against Various pH Values and Straight Line Extrapolated Connecting Six Coordinates in Order to Calculate x-Intercept That Represents pI and (b) Size of Aggregates in Etanercept Solutions at Different pH and Buffer Concentrations with Size Varying Less than 1000 nm

Error bars represent standard deviation with five measurements.

There were extra size-distributed particles in every protein solution, which might indicate protein aggregation since all the excipients that could suppress the aggregation were salted out during dialysis. The distribution of aggregates (<1000 nm) plotted against pH is presented in Fig. 1(b). The aggregates in 10 mM Tris buffer were larger than those in 30 mM Tris buffer but they decreased as pH increased. However, aggregates in 30 mM Tris buffer exhibited continuous size increases from pH 7 to 8.2. Figures 2(a) and (b) show DLS data of size distribution by intensity in 30 mM Tris buffer at pH 7.4 and 10 mM Tris buffer at pH 7.4, respectively. Significant peaks indicated monomer and aggregation of etanercept in both conditions. Another large aggregation peak was observed in 10 mM Tris buffer at pH 7.4 (Fig. 2(b)). The aggregate was about 1480 nm and existed in every 10 mM Tris buffer solution, but not in 30 mM Tris buffer (data not shown).

Fig. 2. DLS Data of Etanercept Distributed by Light Intensity in (a) 30 mM Tris Buffer and (b) 10 mM Tris Buffer at pH 7.4

On the other hand, PDI is dimensionless and scaled such that values smaller than 0.05 are rarely seen other than with highly monodisperse standards. However, values greater than 0.7 indicate that the sample has a very polydispersed distribution and is not suitable for the DLS technique. PDI values of the samples were less than 0.7 except etanercept in 30 mM Tris buffer at pH 6.6, which indicates that this solution is not suitable for measurement (Table 1). Etanercept in 10 mM Tris buffer exhibited relatively higher PDI value because of an additional aggregation peaks (Fig. 2(b)). This may indicate more aggregation issues of etanercept in 10 mM buffer.

Thermodynamic Properties of Etanercept

A typical DSC curve of 5 mg/mL etanercept in 10 mM Tris buffer at pH 7.4 with a scan rate of 1.25 K/min is presented in Fig. 3. The result shows the unfolding transition temperatures (T_m) and enthalpies (∆H) of etanercept in solution. Three thermal unfolding states with transition temperatures of T_m1, T_m2, and T_m3 were observed, suggesting the existence of two domains originating from the F_c component of etanercept containing the domains of C_H2 (T_m2), C_H3 (T_m3), and extra-cellular ligand binding portion, tumor necrosis factor receptor (TNFR; T_m1).²¹⁾ These transition temperatures of etanercept in various pH and buffer concentrations are summarized in Table 2. The three peaks shifted depending on the pH and buffer concentration (Table 2, Fig. 4(a)).

Fig. 3. DSC Thermogram of Etanercept in 10 mM Tris Buffer at pH 7.4 Obtained after Baseline Subtraction and Dotted Lines Indicate Curve Fitting into Three Significant Peaks

Table 2. Transition Temperatures (T_m1, T_m2, and T_m3) of Etanercept with Various pH and Two Different Buffer Concentrations Measured by DSC at a Scan Rate of 1 K/min

10 mM Tris bufferpH	Tm1		Tm2		Tm3
	°C	±S.D.	°C	±S.D.	°C	±S.D.
6.6	53.58	0.39	65.23	0.03	82.25	0.04
7.0	53.98	0.42	66.36	0.05	81.74	0.05
7.4	54.50	0.37	67.18	0.05	81.61	0.06
7.8	55.19	0.34	67.79	0.04	81.99	0.05
8.2	53.23	0.26	67.74	0.04	82.26	0.04
8.6	53.25	0.20	68.20	0.04	82.54	0.04
30 mM Tris bufferpH	Tm1		Tm2		Tm3
	°C	±S.D.	°C	±S.D.	°C	±S.D.
6.6	54.47	0.35	66.12	0.03	82.43	0.04
7.0	53.73	0.27	67.07	0.03	82.85	0.14
7.4	55.32	0.27	67.90	0.05	82.90	0.06
7.8	54.84	0.21	68.70	0.05	82.87	0.06
8.2	54.14	0.14	68.89	0.04	82.78	0.04
8.6	51.20	0.13	68.88	0.03	82.53	0.03

Fig. 4. Overlaid DSC Thermograms of Etanercept (a) in 10 mM Tris Buffer at Various pH and (b) in 10 mM Tris Buffer at pH 7.0 with Four Different Scan Rates (0.75 K/min, 1 K/min, 1.25 K/min, and 1.5 K/min)

As the pH increased, T_m1 shifted while T_m2 and T_m3 were increased. T_m1, representing the first unfolding transition step of etanercept, was the highest at pH 7.4 in 30 mM Tris buffer and at pH 7.8 in 10 mM Tris buffer, 55.32°C and 55.19°C, respectively. In addition, the second highest T_m1 was at pH 7.8 in 30 mM Tris buffer and at pH 7.4 in 10 mM Tris buffer, 54.84°C and 54.50°C, respectively. These might indicate that unfolding of TNFR structure (T_m1) in etanercept would be relatively stable from pH 7.4 to 7.8. However, T_m2 and T_m3 also had unique values depending on pH (Fig. 5). The results showed that the C_H2 domain was stable as pH continuously increased. Its optimum range might be from pH 7.8 to 8.6 (Table 2, Fig. 5a), especially for the 30 mM buffer. In addition, the C_H3 domain was highly stable at pH from pH 7.0 to 7.8 in 30 mM Tris buffer (Table 2, Fig. 5b). Therefore, conformational stability (T_m) was highly dependent on pH and buffer concentration. The changes might be small, but showed significant change in a pH-dependent manner. Further study was necessary as T_m2 and T_m3 values were not sufficient to determine optimal conditions.

Fig. 5. Transition Temperatures (T_m) Plotted against Various pH Values with Highlighted Region Exhibiting Relative T_m: (a) T_m2 and (b) T_m3

Apparent Activation Energy (E_app) of T_m2 and T_m3

To better understand the unfolding mechanism, apparent activation energy (E_app) was evaluated with four different scan rates using Lumry and Eyring model IV in situation C equation. The plot of 1/T_m against ln(1/T_m²) yielded a straight line whose slope was used to determine the E_app (Fig. 6). E_app of C_H2 domain at pH 7.4 in 30 mM and 10 mM buffers were 1312.56 kJ/mol and 1089.69 kJ/mol, respectively (Fig. 6(a)). These results indicate the energy barrier against protein unfolding. The C_H2 domain of etanercept had higher E_app in 30 mM Tris buffer. Similarly, E_app of C_H3 domain at pH 7.4 in 30 mM and 10 mM buffers were 413.47 kJ/mol and 378.55 kJ/mol, respectively (Fig. 6(b)). Both domains in etanercept exhibited higher E_app when the buffer concentration was increased with a good linearity (i.e., R²>0.96). Every E_app of C_H2 and C_H3 in various pH and buffer concentrations was calculated and the result was plotted (Table 3, Fig. 7). Etanercept at pH from 7.4 to 7.8 exhibited relatively higher energy barrier against unfolding of C_H2 (Fig. 7(a)) and C_H3 (Fig. 7(b)). On the other hand, E_app of C_H2 in 10 mM and 30 mM of Tris buffer at pH 6.6 were 339.20 kJ/mol and 345.27 kJ/mol, respectively. It exhibited the lowest energy barrier against protein unfolding, but gradually increased in a pH dependent manner until pH 7.8.

Fig. 6. The Plot of 1/T_m against ln(1/T_m2) According to DSC Results

The straight lines were used to obtain the activation energy by using Lumry–Eyring equation. (a) represents C_H2 domain at pH 7.4 in 30 mM and 10 mM Tris buffer and (b) represents C_H3 domain at pH 7.4 in 30 mM and 10 mM Tris buffer with good linearity (i.e., R²>0.96).

Table 3. Apparent Activation Energy (E_app) of Etanercept with Various pH and Two Different Buffer Concentrations Measured by DSC

pH value	Apparent activation energy (Eapp)
	10 mM Tris buffer		30 mM Tris buffer
	CH2 (kJ/mol)	CH3 (kJ/mol)	CH2 (kJ/mol)	CH3 (kJ/mol)
6.6	339.20	380.49	345.27	402.38
7.0	763.95	298.81	1068.58	422.63
7.4	1089.69	378.55	1312.56	413.47
7.8	944.51	396.55	1353.37	477.83
8.2	1032.85	435.93	946.43	378.08
8.6	859.67	351.86	1571.01	400.15

Fig. 7. E_app Plotted against Various pH Values

(a) E_app of C_H2 domain and (b) E_app of C_H3 domain at two different buffer concentrations.

DISCUSSION

Two biophysical methods were compared for their ability to detect biophysical changes in the model protein with pH and buffer concentration. The protein size was dependent on pH which influenced electrical repulsion of neighboring proteins. Higher electrical repulsion caused a dramatic decrease in size and increase in aggregation (Fig. 2). Aggregates of around 100 nm were detected in every solution since no additives were added to suppress aggregation. Moreover, aggregation of etanercept was evaluated by various analytical methods.²²⁾ Larger aggregates around 1000 nm were observed in 10 mM Tris buffer over the pH range. This might indicate that irreversible aggregation did not occur in higher buffer concentration. Thus, etanercept might be more biophysically stable in higher buffer concentration that suppresses irreversible aggregation.

According to the Lumry–Eyring model for protein aggregation, the equation N⇆TS→A was established and applied to many protein studies.^18,22,23) Native proteins (N) undergo reversible conformational change into a transition state (TS) that functions as a trigger for non-native aggregation or irreversible aggregation (A) because of increased exposure of hydrophobic patches that were previously embedded in the core of the native protein.^23–25) Aggregates (A) will subsequently form and can further develop to subvisible and visible particles. Since aggregates were found in every solution, the samples have the potential to undergo non-native aggregation, but this only occurred in 10 mM Tris buffer. Buffer concentration could be an important factor for stabilizing etanercept in solution and could enhance the ability to suppress exposure of hydrophobic core of protein. Further studies were performed to investigate in more detail by thermal analysis.

DSC analysis to detect unfolding intermediates (TS) was useful for the analysis of protein instability. With changing pH, buffer concentration, and scan rate, unfolding transition temperatures shifted significantly (Fig. 4). Etanercept had three significant endothermic peaks representing the extra-cellular ligand binding portion (TNFR) and the F_c portion of IgG1 (C_H2 and C_H3), whereas the F_ab domain was absent. Since the stability of IgG1 F_c portion (C_H2 and C_H3) depends mainly on environmental stresses like pH and temperature, DSC was used to evaluate the conformational and thermodynamic stability of etanercept.^26,27) The result from a reference of IgG1 was similar to this study.^28–30) They exhibited two significant endothermic peaks representing C_H2 and C_H3 from F_c portion of IgG1 with similar transition temperatures. Furthermore, the first transition temperature representing conformational stability of TNFR was the highest (55.32°C) at pH 7.4 in 30 mM Tris buffer. However, the first transition temperature of etanercept at pH 7.4 and 7.8 in 10 mM Tris buffer was also relatively high (>54°C) (Table 2). This finding led to the question: why did they aggressively aggregate even though T_m1 was high?

The second unfolding transition temperatures exhibited continuous increase with pH in both concentrations. Similarly, the third unfolding transition temperatures also increased up to pH 7.8 in 30 mM Tris buffer, but not in 10 mM Tris buffer (Table 2, Fig. 5). The third unfolding transition temperatures in 10 mM Tris buffer decreased down to pH 7.4 and then increased up to pH 8.6. In contrast, the size of aggregates in 10 mM Tris buffer decreased when T_m2 and T_m3 were relatively lower than in 30 mM buffer, suggesting conversion to irreversible aggregation or larger aggregates (>1000 nm). Aggregates in 30 mM Tris buffer exhibited sizes varying from 100 to 200 nm, but were largest at pH 8.2 where the apparent activation energy was relatively low. The results suggest that aggregation of etanercept might be dependent on T_m2 and T_m3 from the F_c portion of IgG1. Further investigation was performed to evaluate T_m2 and T_m3 by calculating apparent activation energy (E_app).

T_m2 and T_m3 shifted depending on four scan rates and the plotted straight lines were linear (R²>0.96) (Figs. 4(b), 6). E_app of C_H2 and C_H3 domains in 30 mM Tris buffer were higher than in 10 mM Tris buffer over the entire pH range except at pH 8.2 (Table 3, Fig. 7). E_app in this study represents the least energy required to start unfolding reaction from native to unfolded protein. The advantage of this method is that it is not dependent on T_m itself since it is calculated from the amount of changes in T_m induced by scan rate. Higher E_app was observed in 30 mM than 10 mM Tris buffer, suggesting higher resistance to unfolding (Fig. 7). Consequently, pH 7.8 exhibited highest E_app in 30 mM Tris buffer and, with respect to the T_m, size of monomer and aggregation, zeta potential, and PDI obtained earlier, pH range from 7.4 to 7.8 in higher buffer concentration displayed optimal stability. As the pH and buffer concentration increased, there was relative increase in T_m and E_app. This might have decreased the potential for non-native aggregation since exposure of hydrophobic patch of the protein was prevented.

CONCLUSION

Many factors affect the biophysical stability of protein formulations and some of the issues remain to be elucidated. In this study, the influences of pH and buffer concentration were investigated using established and validated biophysical analytical methods—DSC and DLS. DLS revealed increased conformational stability of etanercept by pH and buffer concentration that mediated electrostatic interactions from the medium. In addition, the study provided key factors to prevent non-native aggregation mechanism. Non-native aggregation of etanercept seems dependent on the Fc portion of IgG1. Therefore, antibodies are recommended for formulations in higher buffer concentration and at suitable pH range in order to avoid undesirable results.

Acknowledgment

This study was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2012002399) and a Grant of the Korean Healthcare technology R&D Project, Ministry of Health & Welfare, Republic of Korea (Grant No.: A103017).

REFERENCES

• 1) Crommelin DJ, Storm G, Verrijk R, de Leede L, Jiskoot W, Hennink WE. Shifting paradigms: biopharmaceuticals versus low molecular weight drugs. Int. J. Pharm., 266, 3–16 (2003).
• 2) Walsh G. Biopharmaceutical benchmarks 2010. Nat. Biotechnol., 28, 917–924 (2010).
• 3) Brange J. Physical stability of proteins. Pharmaceutical formulation and development of peptides and proteins. (Frokjaer S, Hovgaards L eds.) Taylor and Francis, London, pp. 89–112 (2000).
• 4) Khossravi M, Borchardt RT. Chemical pathways of peptide degradation. X: effect of metal-catalyzed oxidation on the solution structure of a histidine-containing peptide fragment of human relaxin. Pharm. Res., 17, 851–858 (2000).
• 5) Manning MC, Patel K, Borchardt RT. Stability of protein pharmaceuticals. Pharm. Res., 6, 903–918 (1989).
• 6) Volkin DB, Middaugh CR. Part A: Chemical and physical pathways of protein degradation. Stability of protein pharmaceuticals. (Ahern TJ, Manning MC eds.) Plenum Press, New York, p. 215 (1992).
• 7) Schellekens H. Immunogenicity of therapeutic proteins: clinical implications and future prospects. Clin. Ther., 24, 1720–1740, discussion, 1719 (2002).
• 8) Hermeling S, Crommelin DJ, Schellekens H, Jiskoot W. Structure–immunogenicity relationships of therapeutic proteins. Pharm. Res., 21, 897–903 (2004).
• 9) Kim NA, An IB, Lee SY, Park ES, Jeong SH. Optimization of protein solution by a novel experimental design method using thermodynamic properties. Arch. Pharm. Res., 35, 1609–1619 (2012).
• 10) Kim NA, Lim DG, Lim JY, Kim KH, Jeong SH. Fundamental analysis of recombinant human epidermal growth factor in solution with biophysical methods. Drug Dev. Ind. Pharm. (2014), in press.
• 11) Hawe A, Friess W, Sutter M, Jiskoot W. Online fluorescent dye detection method for the characterization of immunoglobulin G aggregation by size exclusion chromatography and asymmetrical flow field flow fractionation. Anal. Biochem., 378, 115–122 (2008).
• 12) Hawe A, Kasper JC, Friess W, Jiskoot W. Structural properties of monoclonal antibody aggregates induced by freeze-thawing and thermal stress. Eur. J. Pharm. Sci., 38, 79–87 (2009).
• 13) Mahler H-C, Friess W, Grauschopf U, Kiese S. Protein aggregation: Pathways, induction factors and analysis. J. Pharm. Sci., 98, 2909–2934 (2009).
• 14) Jeong SH. Analytical methods and formulation factors to enhance protein stability in solution. Arch. Pharm. Res., 35, 1871–1886 (2012).
• 15) Gu LC, Erdos EA, Chiang HS, Calderwood T, Tsai K, Visor GC, Duffy J, Hsu WC, Foster LC. Stability of interleukin 1 beta (IL-1 beta) in aqueous solution: analytical methods, kinetics, products, and solution formulation implications. Pharm. Res., 8, 485–490 (1991).
• 16) Nielsen L, Khurana R, Coats A, Frokjaer S, Brange J, Vyas S, Uversky VN, Fink AL. Effect of environmental factors on the kinetics of insulin fibril formation: elucidation of the molecular mechanism. Biochemistry, 40, 6036–6046 (2001).
• 17) Krishnan S, Chi EY, Webb JN, Chang BS, Shan D, Goldenberg M, Manning MC, Randolph TW, Carpenter JF. Aggregation of granulocyte colony stimulating factor under physiological conditions: characterization and thermodynamic inhibition. Biochemistry, 41, 6422–6431 (2002).
• 18) Sanchez-Ruiz JM. Theoretical analysis of Lumry–Eyring models in differential scanning calorimetry. Biophys. J., 61, 921–935 (1992).
• 19) Ishiwatari N, Fukuoka M, Sakai N. Effect of protein denaturation degree on texture and water state of cooked meat. J. Food Eng., 117, 361–369 (2013).
• 20) Hawe A, Hulse WL, Jiskoot W, Forbes RT. Taylor dispersion analysis compared to dynamic light scattering for the size analysis of therapeutic peptides and proteins and their aggregates. Pharm. Res., 28, 2302–2310 (2011).
• 21) Tunga BS, Sharma S, Dua R, Dutta S, Mallubhotla H, Pandey V, Chhatbar C. Liquid formulation of polypeptides containing an F_c domain of an immunoglobulin. WO Patent 2,011,141,926 (2011).
• 22) Maarschalkerweerd A, Wolbink GJ, Stapel SO, Jiskoot W, Hawe A. Comparison of analytical methods to detect instability of etanercept during thermal stress testing. Eur. J. Pharm. Biopharm., 78, 213–221 (2011).
• 23) Chi EY, Krishnan S, Randolph TW, Carpenter JF. Physical stability of proteins in aqueous solution: mechanism and driving forces in nonnative protein aggregation. Pharm. Res., 20, 1325–1336 (2003).
• 24) Philo JS, Arakawa T. Mechanisms of protein aggregation. Curr. Pharm. Biotechnol., 10, 348–351 (2009).
• 25) Wang W, Nema S, Teagarden D. Protein aggregation-pathways and influencing factors. Int. J. Pharm., 390, 89–99 (2010).
• 26) Welfle K, Misselwitz R, Hausdorf G, Hohne W, Welfle H. Conformation, pH-induced conformational changes, and thermal unfolding of anti-p24 (HIV-1) monoclonal antibody CB4-1 and its F_ab and F_c fragments. Biochim. Biophys. Acta, 1431, 120–131 (1999).
• 27) Ishikawa T, Ito T, Endo R, Nakagawa K, Sawa E, Wakamatsu K. Influence of pH on heat-induced aggregation and degradation of therapeutic monoclonal antibodies. Biol. Pharm. Bull., 33, 1413–1417 (2010).
• 28) Ionescu RM, Vlasak J, Price C, Kirchmeier M. Contribution of variable domains to the stability of humanized IgG1 monoclonal antibodies. J. Pharm. Sci., 97, 1414–1426 (2008).
• 29) Chennamsetty N, Voynov V, Kayser V, Helk B, Trout BL. Design of therapeutic proteins with enhanced stability. Proc. Natl. Acad. Sci. U.S.A., 106, 11937–11942 (2009).
• 30) Manikwar P, Majumdar R, Hickey JM, Thakkar SV, Samra HS, Sathish HA, Bishop SM, Middaugh CR, Weis DD, Volkin DB. Correlating excipient effects on conformational and storage stability of an IgG1 monoclonal antibody with local dynamics as measured by hydrogen/deuterium-exchange mass spectrometry. J. Pharm. Sci., 102, 2136–2151 (2013).