- Email: [email protected]
pH dependent effect of glycosylation on protein stability
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/ejps
pH dependent effect of glycosylation on protein stability Wei Wang b , Kris Antonsen c , Y. John Wang d , D.Q. Wang a,∗ a
Biotechnology, Bayer HealthCare, Berkeley, CA 94701, United States Pfizer Inc., Global Biologics, 700 Chesterfield Parkway West, Chesterfield, MO 63017, USA c BioMarin Pharmaceutical Inc., 105 Digital Drive, Novato, CA 94949, USA d Genentech, 1 DNA Way, South San Francisco, CA 94080, USA b
a r t i c l e
i n f o
a b s t r a c t
Article history:
The effect of glycosylation state on the thermal and storage stability of interleukin-2 mutein
Received 25 June 2007
(IL-2 mutein) was investigated. The thermal stability of IL-2 mutein was studied by DSC and
Received in revised form
UV. An accelerated storage stability study was conducted at 40 ◦ C in the dark and analyzed
1 October 2007
by UV, SDS-PAGE, and RP-HPLC. The unfolding temperatures (Tu ) of both glycosylated and
Accepted 27 October 2007
unglycosylated forms of IL-2 mutein are similar (within ±1 ◦ C) at pH 5.5 and 7.5. At pH 4.0, the
Published on line 19 November 2007
Tu of glycosylated IL-2 mutein was 4 ◦ C lower than that of the unglycosylated form. The precipitation temperature of glycosylated IL-2 mutein is similar to that of the unglycosylated
Keywords:
form at pH 5.5 but 4 ◦ C higher at pH 7.5. The precipitation temperature is not detectable
Protein formulation
for both forms at pH 4.0. During storage, both glycosylated and unglycosylated IL-2 mutein
Thermal/storage stability
form aggregates (soluble and insoluble) and other degradation products. The aggregates are
Unfolding
formed by both physical and chemical mechanisms. The major pathway of chemical aggre-
Aggregation/precipitation
gation appears to be disulfide bond formation/exchange. The glycosylated form is much less stable than the unglycosylated form at pH 4.0 and both forms are most stable at pH 5.5 in terms of thermal stability, precipitation rate and total degradation rate. This study clearly demonstrates that the effect of glycosylation on the stability of a protein is pH-dependent. © 2007 Elsevier B.V. All rights reserved.
1.
Introduction
The glycosylation state can potentially affect many biochemical properties of proteins, including stability, solubility, intracellular trafficking, activity, pharmacokinetics, and antigenicity (Liu, 1992). In many cases, glycosylation increases the thermal stability of proteins. For example, native RNase A at 0.3 mg/mL in 10 mM phosphate buffer (pH 8.0) loses more than 90% of its original activity after incubation at 90 ◦ C for 15 min while the glycosylated form maintains about 75% of its activity under the same treatment (Baek and Vijayalakshmi, 1997). Glycosylated ␣1 -antitrypsin has been found to be more resis-
∗
Corresponding author. Tel.: +1 510 705 4910. E-mail address: [email protected] (D.Q. Wang). 0928-0987/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ejps.2007.10.008
tant to urea-induced unfolding and thermal aggregation than the unglycosylated form (Kwon and Yu, 1997). Several factors may contribute to carbohydrate-induced protein stabilization. These include introduction of hydrophilic residues, immobilization of the polypeptide backbone by formation of hydrogen bonds with the backbone and/or surface hydrophilic amino acids, and steric interaction with adjacent peptide residues (Baek and Vijayalakshmi, 1997). In other cases, glycosylation has been shown to have no effect on the thermal stability of proteins. Glycosylated horseradish peroxidase (HRP) has eight N-linked glycans, and its thermal stability and activity are the same as those of the deglycosylated form (Tams and Welinder,
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
1998). Deglycosylation of crude ␣-amylase in culture filtrate does not affect its activity and stability upon heating at 60 ◦ C or under acidic (pH 3.5) or proteolytic treatment (Eriksen et al., 1998). Deglycosylation of saposin B (one O-linked site at Thr) does not affect its hydrolytic activity and its stability to proteolytic digestion (Hiraiwa et al., 1993). Interleukin-2 mutein (IL-2 mutein) is a recombinant N88R mutein of human IL-2. The purified IL-2 mutein, like the wide-type IL-2, is a mixture of O-glycosylated (Thr3 ) and unglycosylated IL-2 mutein. One question is whether there is any difference in storage stability between the glycosylated and unglycosylated IL-2 mutein. If there is, the fermentation and purification conditions may be tailored such that the preferred form of IL-2 mutein is produced. If a mixture of the two forms has to be used in the final formulation, identification of any difference in storage stability may help in selection of a formulation composition that favors the storage stability of all forms. Therefore, this study was designed to compare the relative thermal and storage stability of the glycosylated and unglycosylated IL-2 mutein in solution at different pH’s. A storage stability study was conducted at 40 ◦ C and samples were analyzed by UV, SDS-PAGE, RP-HPLC, and DSC.
2.
Materials and methods
2.1.
Materials
IL-2 mutein was expressed in Chinese hamster ovary (CHO) cells. As expressed, the mutein consists of a mixture of three major species. One is unglycosylated. The other two species are glycosylated at the threonine residue in position 3. One species contains a single sialic acid residue, and the other two residues. Mass spectrometry reveals that the first two N-terminal amino acids (Ala-Pro) are partly missing in the glycosylated species but completely missing in the unglycosylated species. The IL-2 mutein was purified in a sequence of five chromatography steps. These were, in order, cation exchange, ceramic hydroxyapatite adsorption, hydrophobic interaction, anion exchange, and finally, a second cation exchange step. The anion exchange step was used to fractionate the IL-2 mutein by charge into two pools, the first of which is predominantly unglycosylated (88%), and the other which contains predominantly the two major glycosylated species (87%). The isoelectric pH of the unglycosylated species is 8.3, and for the glycosylated species it is 7.3–7.9. USP-grade chemicals, including NaCl (J.T. Baker, Phillipsburg, NJ), sodium citrate, citric acid, and sodium phosphate (Spectrum, Gardena, CA) were used as received. High purity guanidine hydrochloride (GdnHCl, 8 M) was purchased from Pierce (Rockford, IL).
pHs 4.0 and 5.5) or sodium phosphate (for pH 7.5) were prepared and sterile filtered. IL-2 mutein stability samples at pH 4.0, 5.5, and 7.5 were prepared in duplicate by gently mixing equal volumes of the diluted IL-2 mutein and the buffer solutions. The final protein concentration was 1 mg/mL and the buffer concentration was 20 mM. One and half milliliter of the mixed samples was aliquoted in 2-mL O-ring, screw-capped sterile polypropylene vials. All protein solutions were clear and colorless after preparation. Storage stability study was conducted at 40 ◦ C and samples were pulled on days 0, 3, 7, and 14. All the samples were frozen immediately and stored at −70 ◦ C before analysis. There was no evidence of freeze-thaw induced instability for IL-2 mutein.
2.3. Determination of IL-2 mutein thermal unfolding and precipitation temperatures The thermal unfolding temperature of IL-2 mutein was determined in duplicate on a differential scanning microcalorimeter (VP-DSC, MicroCal, Northampton, MA). Buffers were tested several times to obtain a stable baseline before scanning the sample. The thermal scan rate was set at 1.5 ◦ C/min. The peak temperature in DSC thermograms was defined as the IL-2 mutein unfolding temperature (Tu ). In most cases, the Tu of a duplicate sample showed ±0.5 ◦ C repeatability. The thermal precipitation temperature of IL-2 mutein was determined in duplicate on a UV–vis spectrophotometer (DU® 650, Beckman, Fullerton, CA). The spectrophotometer is equipped with a cell holder for six 300 L cuvettes and a high performance Peltier temperature controller. The first cuvette was loaded with 300 L buffer for correction of its UV contribution at different temperatures. The remaining 5 cuvettes were loaded with the same volume of IL-2 mutein samples. The temperature of these samples was ramped from 20 to 95 ◦ C at 1 ◦ C/min. IL-2 mutein precipitation was monitored at 350 nm. The temperature, at which optical density at 350 nm reaches 2, is defined as the IL-2 mutein precipitation temperature (Tp ). In most cases, the Tp of a duplicate sample showed ±0.5 ◦ C repeatability.
2.4. Pretreatment of storage stability samples before analysis All the frozen storage stability samples were thawed in a water bath at 20 ◦ C and centrifuged at 15,000 × g for 15 min. The supernatant was then analyzed by SDS-PAGE and RPHPLC. The protein precipitates separated by centrifugation were washed with respective formulation buffers three times by repeated centrifugation and decantation. The washed precipitates were dissolved in 6 M GdnHCl and measured by UV (see below).
2.5. 2.2.
121
Quantitation of IL-2 mutein by UV
Preparation of stability samples
Purified frozen bulk of glycosylated and unglycosylated IL-2 mutein at 5 mg/mL was thawed in a water bath at 20 ◦ C, dialyzed overnight into a solution containing 150 mM NaCl at 5 ◦ C, diluted to 2 mg/mL, and sterile filtered. Separately, buffer solutions containing 150 mM NaCl and 40 mM sodium citrate (for
UV determination of purified IL-2 mutein was made at 280 nm on a UV–vis spectrophotometer (DU® 650, Beckman, Fullerton, CA). The relative amounts of IL-2 mutein precipitates generated during storage were also estimated by this method after the precipitates were dissolved in 6 M GdnHCl. The extinction coefficient for human IL-2 at 280 nm was
122
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
reported to be 0.624 in 6 M GdnHCl in 20 mM phosphate buffer at pH 6.5 (Appel et al., 1994). This extinction coefficient was used in the calculation of the IL-2 mutein concentrations.
rate of IL-2 mutein was estimated by linear regression of A280–vs–time curve.
3. 2.6.
This study compared the thermal and storage stability of glycosylated and unglycosylated IL-2 mutein. In order to select a proper storage temperature for accelerated stability studies, the thermal stability of IL-SA was first determined.
All the materials used in running SDS-PAGE were from Novex (San Diego, CA). SDS-PAGE was performed using 4–12% Bis–Tris NuPAGE gels. IL-2 mutein stability samples were first treated in the sample buffer at 95 ◦ C for 5–10 min. About 10–15 g of the treated protein sample was loaded in the gel for analysis. Mark 12TM was used as the protein standard without further treatment. Gels were run in MES buffer at 200 V and stained with Colloidal Blue Stain Kit. The air-dried gels were pictured on a Kodak Imaging Station (New Haven, CT).
2.7.
Results
SDS-PAGE
3.1.
Thermal stability of IL-2 mutein
The thermal stability of IL-2 mutein was determined by DSC and UV. There are two purposes in these experiments: (1) to compare the relative thermal stability of the two IL-2 mutein forms, and (2) to select a storage temperature for accelerated stability studies, at which IL-2 mutein remains folded and does not precipitate instantaneously. The representative DSC results are shown in Figs. 1 and 2, respectively, for the glycosylated and unglycosylated IL-2 mutein. All the thermograms showed an endothermic peak representing unfolding of IL2 mutein. After unfolding, the baselines of thermograms for samples at pH 5.5 and 7.5 shifted significantly below the preunfolding level. The downward shift after unfolding reflected an exothermic event, possibly due to precipitation. If the exothermic event took place before complete unfolding, the unfolding temperatures at pH 5.5 and 7.5 could be potentially underestimated. The average unfolding temperatures for both forms of IL-2 mutein at different pHs are listed in Table 1, which showed the following apparent order: pH 4.0 < pH 7.5 < pH 5.5. Only at pH 4.0, the Tu of glycosylated IL-2 mutein was significantly different (lower) from that of the unglycosylated form. Since the DSC thermograms suggest precipitation of IL-2 mutein immediately following thermal unfolding at pHs 5.5 and 7.5, samples were also analyzed for thermal precipitation by UV (Fig. 3). IL-2 mutein precipitation did occur for samples at pHs 5.5 and 7.5 as evidenced by a sharp increase in optical density at 350 nm. At temperatures above 80 ◦ C, large precipitates started to settle and caused a sharp drop in UV reading. After thermal scanning, the sample cuvettes were inspected and large flake-like precipitates were clearly visible, confirming IL-2 mutein precipitation. At pH 4.0, however, glycosylated IL-2 mutein did not form precipitates until the temperature was raised above 80 ◦ C and precipitation was not detectable for unglycosylated IL-2 mutein during the entire heating process.
Reversed phase HPLC (RP-HPLC)
Reversed phase HPLC analysis was used to assess purity and integrity of the IL-2 mutein. Analysis was performed on a HP 1100 system (Hewlett Packard, Pleasanton, CA). A reversed ˚ Vydac, phase C-18 column (4.6 mm × 250 mm, 5 m, 300 A, Hesperia, CA) was used and its temperature was controlled at 40 ◦ C during analysis. The mobile phase contain part A (0.1% TFA in water) and part B (0.1% TFA in a mixture of 70% acetonitrile and 30% water). Gradient elution was used and the program was set as follows: 50–100% B from 0 to 15 min; 100–50% B from 15 to 16 min; and 50% B from 16 to 25 min. The flow rate of the mobile phase was set at 1.0 mL/min and the eluate was monitored at 280 nm. Under this condition, the IL-2 mutein was eluted at 15.6 min.
2.8. Estimation of IL-2 mutein degradation and precipitation rates during storage During stability studies, degradation products of IL-2 mutein were observed by RP-HPLC as well as formation of visible IL-2 mutein precipitates. Therefore, the concentration of soluble IL-2 mutein in the stability samples dropped with time. In order to compare the relative stability of IL-2 mutein during storage under different pH’s, the rate of content drop, i.e. the total degradation rate of IL-2 mutein, was estimated, assuming a linear relationship of IL-2 mutein content drop with time within the duration of the stability study. Similarly, assuming the precipitation process is linear within the duration of the stability study, the precipitation
Table 1 – Unfolding and precipitation temperatures (◦ C) of IL-2 mutein (1 mg/mL) Unfolding temperaturea pH 4.0 Glycosylated Unglycosylated a b c
56 60
pH 5.5 68 69
Precipitation temperatureb pH 7.5
pH 4.0
66 66
These are the average unfolding temperatures (◦ C) of two samples determined by VP-DSC. These are the average precipitation temperatures (◦ C) of two samples determined by UV. The precipitation temperature was not reached.
c
N/A N/A
pH 5.5
pH 7.5
69 68
69 65
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
Fig. 1 – Effect of temperature on the unfolding/aggregation of glycosylatd IL-2 mutein at 1 mg/mL at pH 4.0 (A), 5.5 (B) and 7.5 (C) by differential scanning microcalorimeter. These are representative single determinations. The thermal scan rate was at 1.5 ◦ C/min.
123
Fig. 2 – Effect of temperature on the unfolding/aggregation of unglycosylated IL-2 mutein at 1 mg/mL at pH 4.0 (A), 5.5 (B) and 7.5 (C) by differential scanning microcalorimeter. These are representative single determinations. The thermal scan rate was at 1.5 ◦ C/min.
The average precipitation temperatures (Tp ) of IL-2 mutein at pHs 5.5 and 7.5 are listed along with the Tu in Table 1. The Tp corresponded to the protein’s Tu very well (within ±1 ◦ C) except at pH 7.5 for the glycosylated IL-2 mutein, whose Tp is a few degrees higher than its Tu . The higher precipitation temperature suggests a delay in precipitation after protein unfolding at this pH or an underestimation of the Tu , as mentioned before.
3.2.
Storage stability of IL-2 mutein
The thermal unfolding study suggests that IL-2 mutein may unfold at temperatures above 40 ◦ C, while thermal precipitation study indicates a precipitation temperature above 50 ◦ C. Therefore, the accelerated stability study on IL-2 mutein was conducted at 40 ◦ C, which is well below the unfolding and precipitating temperatures. Even at this storage temperature, significant IL-2 mutein precipitation occurred during the course of the stability study. To quantitate the amount of precipitates, all the stability samples in duplicates were centrifuged to separate the precipitates (see Section 2). Table 2 shows the average rates of precipitation estimated by linear regression of A280–vs–time
Fig. 3 – Effect of temperature on the unfolding/aggregation of glycosylated (Gly) and unglycosylated (Ungly) IL-2 mutein at 1 mg/mL at different pHs by UV–vis spectrophotometer. These are representative single determinations. The thermal scan rate was at 1.0 ◦ C/min.
124
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
Table 2 – Precipitation and degradation rates of IL-2 mutein (1 mg/mL) Precipitation ratea (A280/day) pH 4.0 Glycosylated Unglycosylated a b c
d
b
0.20 0.0063 ± 0.0019
Degradation ratec (%/day)
pH 5.5
pH 7.5
pH 4.0
pH 5.5
pH 7.5
0.0023 ± 0.0009 0.0021 ± 0.0007
0.0071 ± 0.0021 0.016 ± 0.002
d
0.5 ± 0.2 0.9 ± 0.1
2.8 ± 0.3 3.4 ± 0.5
32.5 3.2 ± 0.2
These are the estimated average precipitation rates in A280/day at 95% confidence interval. This is the precipitate rate based only on the day-3 UV reading. These are the degradation rate of IL-2 mutein in %/day at 95% confidence interval. These results were based on the IL-2 mutein recoveries determined by RP-HPLC. This is the degradation rate based only on the day-3 recovery.
curve for glycosylated and non-glycosylated IL-2 mutein during storage at pHs 4.0, 5.5 and 7.5. At pH 4.0, glycosylated IL-2 mutein formed precipitates much faster than the unglycosylated form. The amount of precipitates of glycosylated IL-2 mutein on day 3 was equivalent to the original amount of IL-2 mutein based UV analysis, suggesting complete precipitation. At pH 5.5, both forms of IL-2 mutein showed similar precipitation rates, while at pH 7.5, the unglycosylated formed precipitate faster than the glycosylated from. The faster rate of precipitation at pH 7.5 corresponded to a lower precipitation temperature as measured by UV. It is clear that the formulation pH strongly influenced the rate of precipitation for both forms. The rate of precipitation for glycosylated IL-2 mutein during storage has the apparent order of pH 4.0 pH 7.5 > pH 5.5, while that for unglycosylated IL-2 mutein is pH 7.5 > pH 4.0 > pH 5.5. The apparent order of precipitation rate for glycosylated IL-2 mutein corresponded to its order of thermal instability in terms of Tu , whereas that for the unglycosylated form did not. Stability samples on day 0, 7 and 14 were analyzed by SDSPAGE under both non-reduced (Fig. 4) and reduced (Fig. 5) conditions. The starting material for both forms contained a tiny amount of dimer and an unknown component having a molecular weight between the monomer and the dimer (lanes 2–4). Additionally, the glycosylated form at pH 4.0 showed two more bands below the monomer, which were possibly hydrolytic products formed during the sample preparation process since minimal bands were observed at other pHs. As stated previously, glycosylated IL-2 mutein precipitated significantly at pH 4.0 during storage at 40 ◦ C. This is supported by a significant drop in band intensity on day 7 (lane 5). The residual IL-2 mutein band both on days 7 and 14 (lanes 5 and 8) indicates incomplete precipitation of glycosylated IL-2 mutein at pH 4.0. This seems to contradict with what was found by UV determination of the dissolved precipitates (see above). The discrepancy is likely due to a possibility that the IL-2 mutein precipitates were not completely dissolved in 6 M GdnHCl before UV measurement, and presence of any particles would cause light scattering, leading to an overestimation of the amount of dissolved protein. If IL-2 mutein forms physical aggregates during storage, another possibility for this discrepancy would be the dissolution of some aggregates in the SDS-PAGE buffer during sample preparation process, leading to detection of certain amount of monomers in the gel.
The intensity of the unglycosylated IL-2 mutein band at pH 4.0 also dropped with time due to formation of a series of protein aggregates and a cleavage product. This cleavage product of unknown identity had a molecular weight of about half that of IL-2 mutein. Under other pH conditions, the intensity of IL-2 mutein bands did not change significantly. All the IL-2 mutein stability samples on days 7 and 14 contain visible precipitates. However, only the unglycosylated IL-2 mutein stability samples at pH 4.0 showed a significant number of aggregate bands by SDS-PAGE. The lack of protein aggregate bands under these conditions suggests that precipitates were formed physically but dissolved in SDS sample buffer completely except for unglycosylated IL-2 mutein at pH 4.0. Under reduced conditions (Fig. 5), almost all the aggregate bands at pH 4.0 disappeared, suggesting that these aggregates were mainly disulfide bonded. The presence of
Fig. 4 – Non-reduced SDS-PAGE of liquid stability samples of glycosylated (panel A) and unglycosylated (panel B) IL-2 mutein at 1 mg/mL at pHs 4.0, 5.5, and 7.5 at 40 ◦ C. Lane 1 was the molecular weight standard. Lanes 2–4, 5–7, and 8–10 were IL-2 mutein stability samples on days 0, 7, and 14.
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
125
Fig. 6 – Degradation rates of IL-2 mutein at 1 mg/mL during storage at 40 ◦ C as a function of pH. Samples were analyzed by RP-HPLC. Degradation rates were estimated by linear regression of the % recovery–vs–time curves. Error bars are the standard error of degradation rates. Fig. 5 – Reduced SDS-PAGE of liquid stability samples of glycosylated (panel A) and unglycosylated (panel B) IL-2 mutein at 1 mg/mL at pHs 4.0, 5.5, and 7.5 at 40 ◦ C. Lane 1 was the molecular weight standard. Lanes 2–4, 5–7, and 8–10 were IL-2 mutein stability samples on days 0, 7, and 14.
residual dimers under reduced conditions suggests either presence of non-disulfide cross-linking of IL-2 mutein or incomplete reduction during the sample preparation process. Since increasing the amount of reducing agent in the sample buffer did not eliminate the dimer band completely (data not shown), it is likely that non-disulfide cross-linking reactions occurred. The supernatant of all the stability samples was also analyzed by RP-HPLC. The method and a representative RPHPLC chromatogram have been reported (Ha et al., 2002). A number of peaks appeared during storage, especially for unglycosylated IL-2 mutein at pH 4.0 but only two peaks have been identified – the parent IL-2 mutein with a retention time of 16 min and the oxidized IL-2 mutein with a retention time of 15 min. Due to the formation of both degradation products and precipitates, the content of soluble IL-2 mutein dropped. Based on the IL-2 mutein recovery–vs–time curves, the rate of total degradation was estimated. These average rates of total degradation are listed in Table 2 and shown in Fig. 6. The total degradation rate of glycosylated IL2 mutein at pH 4.0 (about 32.5% per day) was significantly higher than that of the unglycosylated form. The greater instability of the glycosylated form corresponded to its lower thermal stability and higher precipitation rate; and was supported by the SDS-PAGE result, as discussed above. At pHs 5.5 and 7.5, the total degradation rates of the unglycosylated form were similar to those of the glycosylated form. The most stable pH was 5.5 for both forms of IL-2 mutein with about 90% recovery at the end of a 14-day incubation period.
4.
Discussion
The results from this study indicate that the effect of glycosylation on IL-2 mutein stability is pH-dependent. While the unfolding temperatures of both forms of IL-2 mutein are similar (within ±1 ◦ C) at pH 5.5 and 7.5, the Tu of glycosylated IL-2 mutein at pH 4.0 was 4 ◦ C lower than that of the unglycosylated form, suggesting that the unglycosylated form is thermally more stable than the glycosylated form. Both forms have the highest thermal stability at pH 5.5. The Tp of glycosylated IL-2 mutein is similar (within ±1 ◦ C) to that of the unglycosylated form at pH 5.5 and their Tp s are similar to their Tu s, suggesting protein unfolding caused immediate aggregation/precipitation at this pH. However, the Tp of glycosylated IL-2 mutein was 4 ◦ C higher than that of the unglycosylated form at pH 7.5, suggesting that glycosylation increased the thermal stability of IL-2 mutein at this pH. Theoretically, the glycosylated form would aggregate/precipitate more easily than the unglycosylated form, as its pI (7.3–7.9) is closer to the experimental pH. Therefore, the increased resistance to aggregation/precipitation of the glycosylated form is apparently not attributable to its pI difference. The increased stability of the glycosylated form at pH 7.5 appears to be kinetic rather than thermodynamic in nature. This contention was supported by the fact that the Tp of the glycosylated form at pH 7.5 was 3 ◦ C higher than its Tu , indicating a delay between protein unfolding and precipitation (if the determined Tu reflects the true unfolding temperature). In contrast, precipitation of the unglycosylated form (65 ◦ C) started immediately (if not earlier) upon protein unfolding (66 ◦ C). In a similar case, the unfolding of glycosylated horseradish peroxidase (HRP) in the presence of GdnHCl was 2–3 times slower than the deglycosylated form, although both forms had the same thermal transition temperature of 57 ◦ C, suggesting glycosylation made
126
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
the protein kinetically more stable (Tams and Welinder, 1998). An interesting finding in this study is that IL-2 mutein precipitation did not occur to a significant level during thermal scanning at pH 4.0 for both forms. Therefore, the Tp at this pH could not be determined by UV. One obvious explanation is that the aggregates/precipitates are soluble in the acidic environment. However, this is unlikely to be the case since both glycosylated and unglycosylated IL-2 mutein formed precipitates at pH 4.0 during storage at 40 ◦ C and in fact, more precipitates were formed at pH 4.0 than at pH 5.5 or 7.5. It has been shown that protein aggregation starts from partially unfolded intermediates rather than the completely unfolded random coils (Clark, 1998; Fink, 1998; Gupta et al., 1998). We believe that at 40 ◦ C, both forms of IL-2 mutein have higher percentages of partially unfolded conformations at pH 4.0 than at pH 5.5 or 7.5 because their Tu s are lowest at pH 4.0. Thus, formation of aggregates/precipitates would be faster at pH 4.0 during storage if partially-unfolded IL-2 mutein molecules were the aggregation-initiating species. On the other hand, when the temperature was increased quickly above the IL-2 mutein unfolding temperature, the transient and partially unfolded IL-2 mutein intermediates could not form aggregates/precipitates within the experimental time scale before they became completely unfolded. If the unfolded states were aggregationincompetent at pH 4.0, immediate aggregation/precipitation would not be observed during the short period of thermal scanning. During storage at 40 ◦ C, both glycosylated and unglycosylated IL-2 mutein formed soluble aggregates and precipitates within days at all pH’s tested. The formation of aggregates/precipitates apparently involves both physical and chemical processes. The majority of chemically-formed aggregates was disulfide bonded and very little, if at all, amounts were non-disulfide bonded. Such a behavior is expected, as the IL-2 mutein contains three cysteines (Cys58 , Cys105 and Cys205 ); two being disulfide-linked (Cys58 , Cys105 ). This aggregation pattern was also observed previously for unfractionated IL-2 mutein (data not shown). The relatively low rate of IL-2 mutein precipitation at pH 5.5 corresponded to the lowest degradation rate estimated by RP-HPLC and the highest Tu by DSC for both forms. The SDS-PAGE results reveal a difference in the relative distribution of all visible aggregate bands among different pHs, suggesting that the aggregation and/or precipitation mechanisms may be slightly different at different pH’s. It is surprising to find that at pH 4.0, the glycosylated IL2 mutein was much less stable than the unglycosylated form in terms of Tu , precipitation rate, and degradation rate. Due to the presence of N-acetylneuraminic acid, the sugar chain is expected to be unionized (or partially ionized) at pH 4.0. The unionized sugar chain can only interact with the protein structure through H-bonding or van der Waals forces, which are generally weaker than ionic interaction. Therefore, it is difficult to speculate how such weaker forces could alter the 3-D structure and destabilize the protein. The apparent destabilization of the sugar chain on IL-2 mutein at pH 4.0 needs further investigation.
A negative effect of glycosylation on protein stability has rarely been reported. Nevertheless, glycosylation has been shown to destabilize a few proteins. Langlois et al. (1992) demonstrated that the temperature of inactivating 50% activity of glycosylated serum creatine kinase was 35 ◦ C but 45 ◦ C for the unglycosylated protein, indicating a lower thermal stability for the glycosylated form. Rose et al. (1984) demonstrated that glycosylated ovine submaxillary mucin aggregated into polymeric rope-like strands at 50–100 g/mL but assumed a globular and monomeric conformation upon enzymatic removal of the saccharide units, suggesting that the carbohydrate residues play a role in mucin aggregation. However, neither of the two studies explained how the glycosylation state affected the protein structure. The Tu of a protein is an indication of its thermal stability. It has been used to compare the relative stability of proteins and to screen protein formulation excipients (Dill et al., 1989; Remmele et al., 1998). In this study, the Tu of IL-2 mutein was examined to determine whether this temperature could correlate with IL-2 mutein storage stability at different pH’s. For the glycosylated form, a lower Tu did correspond to a higher precipitation rate. However, there was no correlation for the unglycosylated IL-2 mutein. The deviation seems to occur at pH 4.0, where unfolding of the protein was not followed by any detectable precipitation as measured by UV. We speculate that removal of the sugar moieties on the protein somehow altered the sequence interaction, which is more resistant to aggregation. Therefore, use of Tu as a measurement of tendency of storage-induced protein aggregation/precipitation should be cautioned, as both pH and glycosylation states may have a strong effect on protein aggregation/precipitation.
Acknowledgements We sincerely thank our summer intern Joe Elzweig for analyzing some of the samples by DSC and SDS-PAGE. Finally, we are indebted to Drs. Shian-Jiun Shih, Jun Ouyang, and Sherry Martin-Moe for their critical review of this manuscript.
references
Appel, R.D., Bairoch, A., Hochstrasser, D.F., 1994. A new generation of information retrieval tools for biologists: the example of the ExPASy WWW server. Trends Biochem. Sci. 19, 258–260. Baek, W.O., Vijayalakshmi, M.A., 1997. Effect of chemical glycosylation of RNase A on the protein stability and surface histidines accessibility in immobilized metal ion affinity electrophoresis (IMAGE) system. Biochim. Biophys. Acta 1336, 394–402. Clark, E.D.B., 1998. Refolding of recombinant proteins. Curr. Opin. Biotechnol. 9, 157–163. Dill, K.A., Alonso, D.O., Hutchinson, K., 1989. Thermal stabilities of globular proteins. Biochemistry 28, 5439–5449. Eriksen, S.H., Jensen, B., Olsen, J., 1998. Effect of N-linked glycosylation on secretion, activity, and stability of alpha-amylase from Aspergillus oryzae. Curr. Microbiol. 37, 117–122. Fink, A.L., 1998. Protein aggregation: folding aggregates, inclusion bodies and amyloid. Fold Des. 3, R9–R23.
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
Gupta, P., Hall, C.K., Voegler, A.C., 1998. Effect of denaturant and protein concentrations upon protein refolding and aggregation: a simple lattice model. Protein Sci. 7, 2642–2652. Ha, E., Wang, W., Wang, Y.J., 2002. Peroxide formation in polysorbate 80 and protein stability. J. Pharm. Sci. 91, 2252–2264. Hiraiwa, M., Soeda, S., Martin, B.M., Fluharty, A.L., Hirabayashi, Y., O’Brien, J.S., Kishimoto, Y., 1993. The effect of carbohydrate removal on stability and activity of saposin B. Arch. Biochem. Biophys. 303, 326–331. Kwon, K.S., Yu, M.H., 1997. Effect of glycosylation on the stability of alpha1-antitrypsin toward urea denaturation and thermal deactivation. Biochim. Biophys. Acta 1335, 265–272. Langlois, M.R., Delanghe, J.R., De Buyzere, M.L., Leroux-Roels, G.G., 1992. Glycation of human tissue and serum creatine kinase. Clin. Chim. Acta 211, 83–92.
127
Liu, D.T., 1992. Glycoprotein pharmaceuticals: scientific and regulatory considerations, and the US Orphan Drug Act. Trends Biotechnol. 10, 114–120. Remmele Jr., R.L., Nightlinger, N.S., Srinivasan, S., Gombotz, W.R., 1998. Interleukin-1 receptor (IL-1R) liquid formulation development using differential scanning calorimetry. Pharm. Res. 15, 200–208. Rose, M.C., Voter, W.A., Sage, H., Brown, C.F., Kaufman, B., 1984. Effects of deglycosylation on the architecture of ovine submaxillary mucin glycoprotein. J. Biol. Chem. 259, 3167–3172. Tams, J.W., Welinder, K.G., 1998. Glycosylation and thermodynamic versus kinetic stability of horseradish peroxidase. FEBS Lett. 421, 234–236.
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/ejps
pH dependent effect of glycosylation on protein stability Wei Wang b , Kris Antonsen c , Y. John Wang d , D.Q. Wang a,∗ a
Biotechnology, Bayer HealthCare, Berkeley, CA 94701, United States Pfizer Inc., Global Biologics, 700 Chesterfield Parkway West, Chesterfield, MO 63017, USA c BioMarin Pharmaceutical Inc., 105 Digital Drive, Novato, CA 94949, USA d Genentech, 1 DNA Way, South San Francisco, CA 94080, USA b
a r t i c l e
i n f o
a b s t r a c t
Article history:
The effect of glycosylation state on the thermal and storage stability of interleukin-2 mutein
Received 25 June 2007
(IL-2 mutein) was investigated. The thermal stability of IL-2 mutein was studied by DSC and
Received in revised form
UV. An accelerated storage stability study was conducted at 40 ◦ C in the dark and analyzed
1 October 2007
by UV, SDS-PAGE, and RP-HPLC. The unfolding temperatures (Tu ) of both glycosylated and
Accepted 27 October 2007
unglycosylated forms of IL-2 mutein are similar (within ±1 ◦ C) at pH 5.5 and 7.5. At pH 4.0, the
Published on line 19 November 2007
Tu of glycosylated IL-2 mutein was 4 ◦ C lower than that of the unglycosylated form. The precipitation temperature of glycosylated IL-2 mutein is similar to that of the unglycosylated
Keywords:
form at pH 5.5 but 4 ◦ C higher at pH 7.5. The precipitation temperature is not detectable
Protein formulation
for both forms at pH 4.0. During storage, both glycosylated and unglycosylated IL-2 mutein
Thermal/storage stability
form aggregates (soluble and insoluble) and other degradation products. The aggregates are
Unfolding
formed by both physical and chemical mechanisms. The major pathway of chemical aggre-
Aggregation/precipitation
gation appears to be disulfide bond formation/exchange. The glycosylated form is much less stable than the unglycosylated form at pH 4.0 and both forms are most stable at pH 5.5 in terms of thermal stability, precipitation rate and total degradation rate. This study clearly demonstrates that the effect of glycosylation on the stability of a protein is pH-dependent. © 2007 Elsevier B.V. All rights reserved.
1.
Introduction
The glycosylation state can potentially affect many biochemical properties of proteins, including stability, solubility, intracellular trafficking, activity, pharmacokinetics, and antigenicity (Liu, 1992). In many cases, glycosylation increases the thermal stability of proteins. For example, native RNase A at 0.3 mg/mL in 10 mM phosphate buffer (pH 8.0) loses more than 90% of its original activity after incubation at 90 ◦ C for 15 min while the glycosylated form maintains about 75% of its activity under the same treatment (Baek and Vijayalakshmi, 1997). Glycosylated ␣1 -antitrypsin has been found to be more resis-
∗
Corresponding author. Tel.: +1 510 705 4910. E-mail address: [email protected] (D.Q. Wang). 0928-0987/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ejps.2007.10.008
tant to urea-induced unfolding and thermal aggregation than the unglycosylated form (Kwon and Yu, 1997). Several factors may contribute to carbohydrate-induced protein stabilization. These include introduction of hydrophilic residues, immobilization of the polypeptide backbone by formation of hydrogen bonds with the backbone and/or surface hydrophilic amino acids, and steric interaction with adjacent peptide residues (Baek and Vijayalakshmi, 1997). In other cases, glycosylation has been shown to have no effect on the thermal stability of proteins. Glycosylated horseradish peroxidase (HRP) has eight N-linked glycans, and its thermal stability and activity are the same as those of the deglycosylated form (Tams and Welinder,
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
1998). Deglycosylation of crude ␣-amylase in culture filtrate does not affect its activity and stability upon heating at 60 ◦ C or under acidic (pH 3.5) or proteolytic treatment (Eriksen et al., 1998). Deglycosylation of saposin B (one O-linked site at Thr) does not affect its hydrolytic activity and its stability to proteolytic digestion (Hiraiwa et al., 1993). Interleukin-2 mutein (IL-2 mutein) is a recombinant N88R mutein of human IL-2. The purified IL-2 mutein, like the wide-type IL-2, is a mixture of O-glycosylated (Thr3 ) and unglycosylated IL-2 mutein. One question is whether there is any difference in storage stability between the glycosylated and unglycosylated IL-2 mutein. If there is, the fermentation and purification conditions may be tailored such that the preferred form of IL-2 mutein is produced. If a mixture of the two forms has to be used in the final formulation, identification of any difference in storage stability may help in selection of a formulation composition that favors the storage stability of all forms. Therefore, this study was designed to compare the relative thermal and storage stability of the glycosylated and unglycosylated IL-2 mutein in solution at different pH’s. A storage stability study was conducted at 40 ◦ C and samples were analyzed by UV, SDS-PAGE, RP-HPLC, and DSC.
2.
Materials and methods
2.1.
Materials
IL-2 mutein was expressed in Chinese hamster ovary (CHO) cells. As expressed, the mutein consists of a mixture of three major species. One is unglycosylated. The other two species are glycosylated at the threonine residue in position 3. One species contains a single sialic acid residue, and the other two residues. Mass spectrometry reveals that the first two N-terminal amino acids (Ala-Pro) are partly missing in the glycosylated species but completely missing in the unglycosylated species. The IL-2 mutein was purified in a sequence of five chromatography steps. These were, in order, cation exchange, ceramic hydroxyapatite adsorption, hydrophobic interaction, anion exchange, and finally, a second cation exchange step. The anion exchange step was used to fractionate the IL-2 mutein by charge into two pools, the first of which is predominantly unglycosylated (88%), and the other which contains predominantly the two major glycosylated species (87%). The isoelectric pH of the unglycosylated species is 8.3, and for the glycosylated species it is 7.3–7.9. USP-grade chemicals, including NaCl (J.T. Baker, Phillipsburg, NJ), sodium citrate, citric acid, and sodium phosphate (Spectrum, Gardena, CA) were used as received. High purity guanidine hydrochloride (GdnHCl, 8 M) was purchased from Pierce (Rockford, IL).
pHs 4.0 and 5.5) or sodium phosphate (for pH 7.5) were prepared and sterile filtered. IL-2 mutein stability samples at pH 4.0, 5.5, and 7.5 were prepared in duplicate by gently mixing equal volumes of the diluted IL-2 mutein and the buffer solutions. The final protein concentration was 1 mg/mL and the buffer concentration was 20 mM. One and half milliliter of the mixed samples was aliquoted in 2-mL O-ring, screw-capped sterile polypropylene vials. All protein solutions were clear and colorless after preparation. Storage stability study was conducted at 40 ◦ C and samples were pulled on days 0, 3, 7, and 14. All the samples were frozen immediately and stored at −70 ◦ C before analysis. There was no evidence of freeze-thaw induced instability for IL-2 mutein.
2.3. Determination of IL-2 mutein thermal unfolding and precipitation temperatures The thermal unfolding temperature of IL-2 mutein was determined in duplicate on a differential scanning microcalorimeter (VP-DSC, MicroCal, Northampton, MA). Buffers were tested several times to obtain a stable baseline before scanning the sample. The thermal scan rate was set at 1.5 ◦ C/min. The peak temperature in DSC thermograms was defined as the IL-2 mutein unfolding temperature (Tu ). In most cases, the Tu of a duplicate sample showed ±0.5 ◦ C repeatability. The thermal precipitation temperature of IL-2 mutein was determined in duplicate on a UV–vis spectrophotometer (DU® 650, Beckman, Fullerton, CA). The spectrophotometer is equipped with a cell holder for six 300 L cuvettes and a high performance Peltier temperature controller. The first cuvette was loaded with 300 L buffer for correction of its UV contribution at different temperatures. The remaining 5 cuvettes were loaded with the same volume of IL-2 mutein samples. The temperature of these samples was ramped from 20 to 95 ◦ C at 1 ◦ C/min. IL-2 mutein precipitation was monitored at 350 nm. The temperature, at which optical density at 350 nm reaches 2, is defined as the IL-2 mutein precipitation temperature (Tp ). In most cases, the Tp of a duplicate sample showed ±0.5 ◦ C repeatability.
2.4. Pretreatment of storage stability samples before analysis All the frozen storage stability samples were thawed in a water bath at 20 ◦ C and centrifuged at 15,000 × g for 15 min. The supernatant was then analyzed by SDS-PAGE and RPHPLC. The protein precipitates separated by centrifugation were washed with respective formulation buffers three times by repeated centrifugation and decantation. The washed precipitates were dissolved in 6 M GdnHCl and measured by UV (see below).
2.5. 2.2.
121
Quantitation of IL-2 mutein by UV
Preparation of stability samples
Purified frozen bulk of glycosylated and unglycosylated IL-2 mutein at 5 mg/mL was thawed in a water bath at 20 ◦ C, dialyzed overnight into a solution containing 150 mM NaCl at 5 ◦ C, diluted to 2 mg/mL, and sterile filtered. Separately, buffer solutions containing 150 mM NaCl and 40 mM sodium citrate (for
UV determination of purified IL-2 mutein was made at 280 nm on a UV–vis spectrophotometer (DU® 650, Beckman, Fullerton, CA). The relative amounts of IL-2 mutein precipitates generated during storage were also estimated by this method after the precipitates were dissolved in 6 M GdnHCl. The extinction coefficient for human IL-2 at 280 nm was
122
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
reported to be 0.624 in 6 M GdnHCl in 20 mM phosphate buffer at pH 6.5 (Appel et al., 1994). This extinction coefficient was used in the calculation of the IL-2 mutein concentrations.
rate of IL-2 mutein was estimated by linear regression of A280–vs–time curve.
3. 2.6.
This study compared the thermal and storage stability of glycosylated and unglycosylated IL-2 mutein. In order to select a proper storage temperature for accelerated stability studies, the thermal stability of IL-SA was first determined.
All the materials used in running SDS-PAGE were from Novex (San Diego, CA). SDS-PAGE was performed using 4–12% Bis–Tris NuPAGE gels. IL-2 mutein stability samples were first treated in the sample buffer at 95 ◦ C for 5–10 min. About 10–15 g of the treated protein sample was loaded in the gel for analysis. Mark 12TM was used as the protein standard without further treatment. Gels were run in MES buffer at 200 V and stained with Colloidal Blue Stain Kit. The air-dried gels were pictured on a Kodak Imaging Station (New Haven, CT).
2.7.
Results
SDS-PAGE
3.1.
Thermal stability of IL-2 mutein
The thermal stability of IL-2 mutein was determined by DSC and UV. There are two purposes in these experiments: (1) to compare the relative thermal stability of the two IL-2 mutein forms, and (2) to select a storage temperature for accelerated stability studies, at which IL-2 mutein remains folded and does not precipitate instantaneously. The representative DSC results are shown in Figs. 1 and 2, respectively, for the glycosylated and unglycosylated IL-2 mutein. All the thermograms showed an endothermic peak representing unfolding of IL2 mutein. After unfolding, the baselines of thermograms for samples at pH 5.5 and 7.5 shifted significantly below the preunfolding level. The downward shift after unfolding reflected an exothermic event, possibly due to precipitation. If the exothermic event took place before complete unfolding, the unfolding temperatures at pH 5.5 and 7.5 could be potentially underestimated. The average unfolding temperatures for both forms of IL-2 mutein at different pHs are listed in Table 1, which showed the following apparent order: pH 4.0 < pH 7.5 < pH 5.5. Only at pH 4.0, the Tu of glycosylated IL-2 mutein was significantly different (lower) from that of the unglycosylated form. Since the DSC thermograms suggest precipitation of IL-2 mutein immediately following thermal unfolding at pHs 5.5 and 7.5, samples were also analyzed for thermal precipitation by UV (Fig. 3). IL-2 mutein precipitation did occur for samples at pHs 5.5 and 7.5 as evidenced by a sharp increase in optical density at 350 nm. At temperatures above 80 ◦ C, large precipitates started to settle and caused a sharp drop in UV reading. After thermal scanning, the sample cuvettes were inspected and large flake-like precipitates were clearly visible, confirming IL-2 mutein precipitation. At pH 4.0, however, glycosylated IL-2 mutein did not form precipitates until the temperature was raised above 80 ◦ C and precipitation was not detectable for unglycosylated IL-2 mutein during the entire heating process.
Reversed phase HPLC (RP-HPLC)
Reversed phase HPLC analysis was used to assess purity and integrity of the IL-2 mutein. Analysis was performed on a HP 1100 system (Hewlett Packard, Pleasanton, CA). A reversed ˚ Vydac, phase C-18 column (4.6 mm × 250 mm, 5 m, 300 A, Hesperia, CA) was used and its temperature was controlled at 40 ◦ C during analysis. The mobile phase contain part A (0.1% TFA in water) and part B (0.1% TFA in a mixture of 70% acetonitrile and 30% water). Gradient elution was used and the program was set as follows: 50–100% B from 0 to 15 min; 100–50% B from 15 to 16 min; and 50% B from 16 to 25 min. The flow rate of the mobile phase was set at 1.0 mL/min and the eluate was monitored at 280 nm. Under this condition, the IL-2 mutein was eluted at 15.6 min.
2.8. Estimation of IL-2 mutein degradation and precipitation rates during storage During stability studies, degradation products of IL-2 mutein were observed by RP-HPLC as well as formation of visible IL-2 mutein precipitates. Therefore, the concentration of soluble IL-2 mutein in the stability samples dropped with time. In order to compare the relative stability of IL-2 mutein during storage under different pH’s, the rate of content drop, i.e. the total degradation rate of IL-2 mutein, was estimated, assuming a linear relationship of IL-2 mutein content drop with time within the duration of the stability study. Similarly, assuming the precipitation process is linear within the duration of the stability study, the precipitation
Table 1 – Unfolding and precipitation temperatures (◦ C) of IL-2 mutein (1 mg/mL) Unfolding temperaturea pH 4.0 Glycosylated Unglycosylated a b c
56 60
pH 5.5 68 69
Precipitation temperatureb pH 7.5
pH 4.0
66 66
These are the average unfolding temperatures (◦ C) of two samples determined by VP-DSC. These are the average precipitation temperatures (◦ C) of two samples determined by UV. The precipitation temperature was not reached.
c
N/A N/A
pH 5.5
pH 7.5
69 68
69 65
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
Fig. 1 – Effect of temperature on the unfolding/aggregation of glycosylatd IL-2 mutein at 1 mg/mL at pH 4.0 (A), 5.5 (B) and 7.5 (C) by differential scanning microcalorimeter. These are representative single determinations. The thermal scan rate was at 1.5 ◦ C/min.
123
Fig. 2 – Effect of temperature on the unfolding/aggregation of unglycosylated IL-2 mutein at 1 mg/mL at pH 4.0 (A), 5.5 (B) and 7.5 (C) by differential scanning microcalorimeter. These are representative single determinations. The thermal scan rate was at 1.5 ◦ C/min.
The average precipitation temperatures (Tp ) of IL-2 mutein at pHs 5.5 and 7.5 are listed along with the Tu in Table 1. The Tp corresponded to the protein’s Tu very well (within ±1 ◦ C) except at pH 7.5 for the glycosylated IL-2 mutein, whose Tp is a few degrees higher than its Tu . The higher precipitation temperature suggests a delay in precipitation after protein unfolding at this pH or an underestimation of the Tu , as mentioned before.
3.2.
Storage stability of IL-2 mutein
The thermal unfolding study suggests that IL-2 mutein may unfold at temperatures above 40 ◦ C, while thermal precipitation study indicates a precipitation temperature above 50 ◦ C. Therefore, the accelerated stability study on IL-2 mutein was conducted at 40 ◦ C, which is well below the unfolding and precipitating temperatures. Even at this storage temperature, significant IL-2 mutein precipitation occurred during the course of the stability study. To quantitate the amount of precipitates, all the stability samples in duplicates were centrifuged to separate the precipitates (see Section 2). Table 2 shows the average rates of precipitation estimated by linear regression of A280–vs–time
Fig. 3 – Effect of temperature on the unfolding/aggregation of glycosylated (Gly) and unglycosylated (Ungly) IL-2 mutein at 1 mg/mL at different pHs by UV–vis spectrophotometer. These are representative single determinations. The thermal scan rate was at 1.0 ◦ C/min.
124
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
Table 2 – Precipitation and degradation rates of IL-2 mutein (1 mg/mL) Precipitation ratea (A280/day) pH 4.0 Glycosylated Unglycosylated a b c
d
b
0.20 0.0063 ± 0.0019
Degradation ratec (%/day)
pH 5.5
pH 7.5
pH 4.0
pH 5.5
pH 7.5
0.0023 ± 0.0009 0.0021 ± 0.0007
0.0071 ± 0.0021 0.016 ± 0.002
d
0.5 ± 0.2 0.9 ± 0.1
2.8 ± 0.3 3.4 ± 0.5
32.5 3.2 ± 0.2
These are the estimated average precipitation rates in A280/day at 95% confidence interval. This is the precipitate rate based only on the day-3 UV reading. These are the degradation rate of IL-2 mutein in %/day at 95% confidence interval. These results were based on the IL-2 mutein recoveries determined by RP-HPLC. This is the degradation rate based only on the day-3 recovery.
curve for glycosylated and non-glycosylated IL-2 mutein during storage at pHs 4.0, 5.5 and 7.5. At pH 4.0, glycosylated IL-2 mutein formed precipitates much faster than the unglycosylated form. The amount of precipitates of glycosylated IL-2 mutein on day 3 was equivalent to the original amount of IL-2 mutein based UV analysis, suggesting complete precipitation. At pH 5.5, both forms of IL-2 mutein showed similar precipitation rates, while at pH 7.5, the unglycosylated formed precipitate faster than the glycosylated from. The faster rate of precipitation at pH 7.5 corresponded to a lower precipitation temperature as measured by UV. It is clear that the formulation pH strongly influenced the rate of precipitation for both forms. The rate of precipitation for glycosylated IL-2 mutein during storage has the apparent order of pH 4.0 pH 7.5 > pH 5.5, while that for unglycosylated IL-2 mutein is pH 7.5 > pH 4.0 > pH 5.5. The apparent order of precipitation rate for glycosylated IL-2 mutein corresponded to its order of thermal instability in terms of Tu , whereas that for the unglycosylated form did not. Stability samples on day 0, 7 and 14 were analyzed by SDSPAGE under both non-reduced (Fig. 4) and reduced (Fig. 5) conditions. The starting material for both forms contained a tiny amount of dimer and an unknown component having a molecular weight between the monomer and the dimer (lanes 2–4). Additionally, the glycosylated form at pH 4.0 showed two more bands below the monomer, which were possibly hydrolytic products formed during the sample preparation process since minimal bands were observed at other pHs. As stated previously, glycosylated IL-2 mutein precipitated significantly at pH 4.0 during storage at 40 ◦ C. This is supported by a significant drop in band intensity on day 7 (lane 5). The residual IL-2 mutein band both on days 7 and 14 (lanes 5 and 8) indicates incomplete precipitation of glycosylated IL-2 mutein at pH 4.0. This seems to contradict with what was found by UV determination of the dissolved precipitates (see above). The discrepancy is likely due to a possibility that the IL-2 mutein precipitates were not completely dissolved in 6 M GdnHCl before UV measurement, and presence of any particles would cause light scattering, leading to an overestimation of the amount of dissolved protein. If IL-2 mutein forms physical aggregates during storage, another possibility for this discrepancy would be the dissolution of some aggregates in the SDS-PAGE buffer during sample preparation process, leading to detection of certain amount of monomers in the gel.
The intensity of the unglycosylated IL-2 mutein band at pH 4.0 also dropped with time due to formation of a series of protein aggregates and a cleavage product. This cleavage product of unknown identity had a molecular weight of about half that of IL-2 mutein. Under other pH conditions, the intensity of IL-2 mutein bands did not change significantly. All the IL-2 mutein stability samples on days 7 and 14 contain visible precipitates. However, only the unglycosylated IL-2 mutein stability samples at pH 4.0 showed a significant number of aggregate bands by SDS-PAGE. The lack of protein aggregate bands under these conditions suggests that precipitates were formed physically but dissolved in SDS sample buffer completely except for unglycosylated IL-2 mutein at pH 4.0. Under reduced conditions (Fig. 5), almost all the aggregate bands at pH 4.0 disappeared, suggesting that these aggregates were mainly disulfide bonded. The presence of
Fig. 4 – Non-reduced SDS-PAGE of liquid stability samples of glycosylated (panel A) and unglycosylated (panel B) IL-2 mutein at 1 mg/mL at pHs 4.0, 5.5, and 7.5 at 40 ◦ C. Lane 1 was the molecular weight standard. Lanes 2–4, 5–7, and 8–10 were IL-2 mutein stability samples on days 0, 7, and 14.
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
125
Fig. 6 – Degradation rates of IL-2 mutein at 1 mg/mL during storage at 40 ◦ C as a function of pH. Samples were analyzed by RP-HPLC. Degradation rates were estimated by linear regression of the % recovery–vs–time curves. Error bars are the standard error of degradation rates. Fig. 5 – Reduced SDS-PAGE of liquid stability samples of glycosylated (panel A) and unglycosylated (panel B) IL-2 mutein at 1 mg/mL at pHs 4.0, 5.5, and 7.5 at 40 ◦ C. Lane 1 was the molecular weight standard. Lanes 2–4, 5–7, and 8–10 were IL-2 mutein stability samples on days 0, 7, and 14.
residual dimers under reduced conditions suggests either presence of non-disulfide cross-linking of IL-2 mutein or incomplete reduction during the sample preparation process. Since increasing the amount of reducing agent in the sample buffer did not eliminate the dimer band completely (data not shown), it is likely that non-disulfide cross-linking reactions occurred. The supernatant of all the stability samples was also analyzed by RP-HPLC. The method and a representative RPHPLC chromatogram have been reported (Ha et al., 2002). A number of peaks appeared during storage, especially for unglycosylated IL-2 mutein at pH 4.0 but only two peaks have been identified – the parent IL-2 mutein with a retention time of 16 min and the oxidized IL-2 mutein with a retention time of 15 min. Due to the formation of both degradation products and precipitates, the content of soluble IL-2 mutein dropped. Based on the IL-2 mutein recovery–vs–time curves, the rate of total degradation was estimated. These average rates of total degradation are listed in Table 2 and shown in Fig. 6. The total degradation rate of glycosylated IL2 mutein at pH 4.0 (about 32.5% per day) was significantly higher than that of the unglycosylated form. The greater instability of the glycosylated form corresponded to its lower thermal stability and higher precipitation rate; and was supported by the SDS-PAGE result, as discussed above. At pHs 5.5 and 7.5, the total degradation rates of the unglycosylated form were similar to those of the glycosylated form. The most stable pH was 5.5 for both forms of IL-2 mutein with about 90% recovery at the end of a 14-day incubation period.
4.
Discussion
The results from this study indicate that the effect of glycosylation on IL-2 mutein stability is pH-dependent. While the unfolding temperatures of both forms of IL-2 mutein are similar (within ±1 ◦ C) at pH 5.5 and 7.5, the Tu of glycosylated IL-2 mutein at pH 4.0 was 4 ◦ C lower than that of the unglycosylated form, suggesting that the unglycosylated form is thermally more stable than the glycosylated form. Both forms have the highest thermal stability at pH 5.5. The Tp of glycosylated IL-2 mutein is similar (within ±1 ◦ C) to that of the unglycosylated form at pH 5.5 and their Tp s are similar to their Tu s, suggesting protein unfolding caused immediate aggregation/precipitation at this pH. However, the Tp of glycosylated IL-2 mutein was 4 ◦ C higher than that of the unglycosylated form at pH 7.5, suggesting that glycosylation increased the thermal stability of IL-2 mutein at this pH. Theoretically, the glycosylated form would aggregate/precipitate more easily than the unglycosylated form, as its pI (7.3–7.9) is closer to the experimental pH. Therefore, the increased resistance to aggregation/precipitation of the glycosylated form is apparently not attributable to its pI difference. The increased stability of the glycosylated form at pH 7.5 appears to be kinetic rather than thermodynamic in nature. This contention was supported by the fact that the Tp of the glycosylated form at pH 7.5 was 3 ◦ C higher than its Tu , indicating a delay between protein unfolding and precipitation (if the determined Tu reflects the true unfolding temperature). In contrast, precipitation of the unglycosylated form (65 ◦ C) started immediately (if not earlier) upon protein unfolding (66 ◦ C). In a similar case, the unfolding of glycosylated horseradish peroxidase (HRP) in the presence of GdnHCl was 2–3 times slower than the deglycosylated form, although both forms had the same thermal transition temperature of 57 ◦ C, suggesting glycosylation made
126
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
the protein kinetically more stable (Tams and Welinder, 1998). An interesting finding in this study is that IL-2 mutein precipitation did not occur to a significant level during thermal scanning at pH 4.0 for both forms. Therefore, the Tp at this pH could not be determined by UV. One obvious explanation is that the aggregates/precipitates are soluble in the acidic environment. However, this is unlikely to be the case since both glycosylated and unglycosylated IL-2 mutein formed precipitates at pH 4.0 during storage at 40 ◦ C and in fact, more precipitates were formed at pH 4.0 than at pH 5.5 or 7.5. It has been shown that protein aggregation starts from partially unfolded intermediates rather than the completely unfolded random coils (Clark, 1998; Fink, 1998; Gupta et al., 1998). We believe that at 40 ◦ C, both forms of IL-2 mutein have higher percentages of partially unfolded conformations at pH 4.0 than at pH 5.5 or 7.5 because their Tu s are lowest at pH 4.0. Thus, formation of aggregates/precipitates would be faster at pH 4.0 during storage if partially-unfolded IL-2 mutein molecules were the aggregation-initiating species. On the other hand, when the temperature was increased quickly above the IL-2 mutein unfolding temperature, the transient and partially unfolded IL-2 mutein intermediates could not form aggregates/precipitates within the experimental time scale before they became completely unfolded. If the unfolded states were aggregationincompetent at pH 4.0, immediate aggregation/precipitation would not be observed during the short period of thermal scanning. During storage at 40 ◦ C, both glycosylated and unglycosylated IL-2 mutein formed soluble aggregates and precipitates within days at all pH’s tested. The formation of aggregates/precipitates apparently involves both physical and chemical processes. The majority of chemically-formed aggregates was disulfide bonded and very little, if at all, amounts were non-disulfide bonded. Such a behavior is expected, as the IL-2 mutein contains three cysteines (Cys58 , Cys105 and Cys205 ); two being disulfide-linked (Cys58 , Cys105 ). This aggregation pattern was also observed previously for unfractionated IL-2 mutein (data not shown). The relatively low rate of IL-2 mutein precipitation at pH 5.5 corresponded to the lowest degradation rate estimated by RP-HPLC and the highest Tu by DSC for both forms. The SDS-PAGE results reveal a difference in the relative distribution of all visible aggregate bands among different pHs, suggesting that the aggregation and/or precipitation mechanisms may be slightly different at different pH’s. It is surprising to find that at pH 4.0, the glycosylated IL2 mutein was much less stable than the unglycosylated form in terms of Tu , precipitation rate, and degradation rate. Due to the presence of N-acetylneuraminic acid, the sugar chain is expected to be unionized (or partially ionized) at pH 4.0. The unionized sugar chain can only interact with the protein structure through H-bonding or van der Waals forces, which are generally weaker than ionic interaction. Therefore, it is difficult to speculate how such weaker forces could alter the 3-D structure and destabilize the protein. The apparent destabilization of the sugar chain on IL-2 mutein at pH 4.0 needs further investigation.
A negative effect of glycosylation on protein stability has rarely been reported. Nevertheless, glycosylation has been shown to destabilize a few proteins. Langlois et al. (1992) demonstrated that the temperature of inactivating 50% activity of glycosylated serum creatine kinase was 35 ◦ C but 45 ◦ C for the unglycosylated protein, indicating a lower thermal stability for the glycosylated form. Rose et al. (1984) demonstrated that glycosylated ovine submaxillary mucin aggregated into polymeric rope-like strands at 50–100 g/mL but assumed a globular and monomeric conformation upon enzymatic removal of the saccharide units, suggesting that the carbohydrate residues play a role in mucin aggregation. However, neither of the two studies explained how the glycosylation state affected the protein structure. The Tu of a protein is an indication of its thermal stability. It has been used to compare the relative stability of proteins and to screen protein formulation excipients (Dill et al., 1989; Remmele et al., 1998). In this study, the Tu of IL-2 mutein was examined to determine whether this temperature could correlate with IL-2 mutein storage stability at different pH’s. For the glycosylated form, a lower Tu did correspond to a higher precipitation rate. However, there was no correlation for the unglycosylated IL-2 mutein. The deviation seems to occur at pH 4.0, where unfolding of the protein was not followed by any detectable precipitation as measured by UV. We speculate that removal of the sugar moieties on the protein somehow altered the sequence interaction, which is more resistant to aggregation. Therefore, use of Tu as a measurement of tendency of storage-induced protein aggregation/precipitation should be cautioned, as both pH and glycosylation states may have a strong effect on protein aggregation/precipitation.
Acknowledgements We sincerely thank our summer intern Joe Elzweig for analyzing some of the samples by DSC and SDS-PAGE. Finally, we are indebted to Drs. Shian-Jiun Shih, Jun Ouyang, and Sherry Martin-Moe for their critical review of this manuscript.
references
Appel, R.D., Bairoch, A., Hochstrasser, D.F., 1994. A new generation of information retrieval tools for biologists: the example of the ExPASy WWW server. Trends Biochem. Sci. 19, 258–260. Baek, W.O., Vijayalakshmi, M.A., 1997. Effect of chemical glycosylation of RNase A on the protein stability and surface histidines accessibility in immobilized metal ion affinity electrophoresis (IMAGE) system. Biochim. Biophys. Acta 1336, 394–402. Clark, E.D.B., 1998. Refolding of recombinant proteins. Curr. Opin. Biotechnol. 9, 157–163. Dill, K.A., Alonso, D.O., Hutchinson, K., 1989. Thermal stabilities of globular proteins. Biochemistry 28, 5439–5449. Eriksen, S.H., Jensen, B., Olsen, J., 1998. Effect of N-linked glycosylation on secretion, activity, and stability of alpha-amylase from Aspergillus oryzae. Curr. Microbiol. 37, 117–122. Fink, A.L., 1998. Protein aggregation: folding aggregates, inclusion bodies and amyloid. Fold Des. 3, R9–R23.
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 3 ( 2 0 0 8 ) 120–127
Gupta, P., Hall, C.K., Voegler, A.C., 1998. Effect of denaturant and protein concentrations upon protein refolding and aggregation: a simple lattice model. Protein Sci. 7, 2642–2652. Ha, E., Wang, W., Wang, Y.J., 2002. Peroxide formation in polysorbate 80 and protein stability. J. Pharm. Sci. 91, 2252–2264. Hiraiwa, M., Soeda, S., Martin, B.M., Fluharty, A.L., Hirabayashi, Y., O’Brien, J.S., Kishimoto, Y., 1993. The effect of carbohydrate removal on stability and activity of saposin B. Arch. Biochem. Biophys. 303, 326–331. Kwon, K.S., Yu, M.H., 1997. Effect of glycosylation on the stability of alpha1-antitrypsin toward urea denaturation and thermal deactivation. Biochim. Biophys. Acta 1335, 265–272. Langlois, M.R., Delanghe, J.R., De Buyzere, M.L., Leroux-Roels, G.G., 1992. Glycation of human tissue and serum creatine kinase. Clin. Chim. Acta 211, 83–92.
127
Liu, D.T., 1992. Glycoprotein pharmaceuticals: scientific and regulatory considerations, and the US Orphan Drug Act. Trends Biotechnol. 10, 114–120. Remmele Jr., R.L., Nightlinger, N.S., Srinivasan, S., Gombotz, W.R., 1998. Interleukin-1 receptor (IL-1R) liquid formulation development using differential scanning calorimetry. Pharm. Res. 15, 200–208. Rose, M.C., Voter, W.A., Sage, H., Brown, C.F., Kaufman, B., 1984. Effects of deglycosylation on the architecture of ovine submaxillary mucin glycoprotein. J. Biol. Chem. 259, 3167–3172. Tams, J.W., Welinder, K.G., 1998. Glycosylation and thermodynamic versus kinetic stability of horseradish peroxidase. FEBS Lett. 421, 234–236.