- Email: [email protected]
Isolation, structural characterization, and antiviral activity of positional isomers of monopegylated interferon α-2a (PEGASYS)
Protein Expression and Purification 30 (2003) 78–87 www.elsevier.com/locate/yprep
Isolation, structural characterization, and antiviral activity of positional isomers of monopegylated interferon a-2a (PEGASYS) Stefan Foser,a Alfred Schacher,a Karl A. Weyer,a,* Doris Brugger,b Elke Dietel,c Stefan Marti,c and Thomas Schreitm€ ullerc a
Department of Pharma Technical Operations Biotechnology, Biotech Products, Biotech Development and Production, Hoffmann-La Roche Ltd., Building 066/508, Ch-4070 Basel, Switzerland b Department of Pharma Research Discovery Structure, Biotech Products, Hoffmann-La Roche Ltd., Basel, Switzerland c Department of Analytical Development and Quality Control, Biotech Products, Hoffmann-La Roche Ltd., Basel, Switzerland Received 12 December 2002, and in revised form 29 January 2003
Abstract Interferon a-2a plays an essential role in the treatment of chronic hepatitis C, but it is limited in its efficacy by the short in vivo half-life. To improve the half-life and efficacy, interferon a-2a is conjugated with a 40-kDa branched polyethylene glycol moiety (PEG-IFN, PEGASYS). From this preparation the positional PEG-IFN isomers were isolated and characterized by different analytical methods and antiviral assay. Two chromatographic steps were used to separate and purify nine isomers. The analytical methods IE-HPLC, RP-HPLC, SE-HPLC, SDS–PAGE, and MALDI-TOF MS indicated that each of these nine isomers is conjugated to the branched polyethylene glycol chain at a specific lysine. No isomer with a modification at the amino terminus was observed. All positional isomers induced viral protection of MDBK cells in the antiviral assay. When comparing the quantitative potency of the individual isomers with the whole mixture of PEG-IFN, significant differences in the specific activities were observed: PEG-Lys(31) and PEG-Lys(134) showed higher activities than the mixture, PEG-Lys(164) was equal to the mixture, whereas the activities of PEG-Lys(49), PEG-Lys(70), PEG-Lys(83), PEG-Lys(112), PEG-Lys(121), and PEG-Lys(131) were lower. Ó 2003 Elsevier Science (USA). All rights reserved. Keywords: Interferon a-2a; Pegylation; Forty kilodaltons of branched polyethylene glycol; Separation of positional isomers; Antiviral effect
Interferon a-2a (IFN) belongs to the class of type I interferons which are a family of homologous cytokines (interferon-a, -b, and --). These interferons have a wide variety of biological effects that include antiviral, antiproliferative, and immunomodulatory activities [1]. Since 1986 recombinant human IFN (ROFERON-A) has become an important therapeutic agent to treat patients with viral and oncological diseases. Worldwide, approximately 170 million people are infected with hepatitis C virus [19]. Most viral infections are primarily due to blood transfusions with blood products contaminated with virus. An essential component of the treatment of chronic hepatitis C infection has been interferon [2]. * Corresponding author. Fax: +41-61-688-2420. E-mail address: [email protected] (K.A. Weyer).
The frequency of the administration of interferon is dependent on the disease. For example, in a viral indication of hepatitis C the recommended administration is three times per week over 24–48 months, depending on the genotype of the virus [2,3]. With a subcutaneous or an intravenous administration the reported terminal half-life of IFN is in a range 4–8 h. After 24 h the serum concentration of IFN already decreases below its detection limit [4,5]. Since the early 1990s a major effort focussed on improving the half-life in blood, in order to decrease the frequency of administration of IFN [6]. One promising way considered was pegylation of IFN. The type of pegylation selected in this case was the modification of the lysine residues on the surface of the protein with a large, branched polyethylene glycol (PEG) through covalent linkage [7].
1046-5928/03/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S1046-5928(03)00055-X
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Pegylation changes physicochemical and biological properties of the protein. One effect, is the decrease of the proteolytic degradation and the renal clearance. This increases the half-life of the pegylated protein in blood. Another effect is the altered distribution in the body, depending on the size of the PEG moiety of the protein. This was shown in an animal study with pegylated interferons: increasing PEG molecular weight led to a decrease of renal clearance with a simultaneous increase of hepatic clearance [8]. The goal was to develop a pegylated IFN with an improved biological activity and the reduction of the administration frequency to one injection per week. An earlier developed molecule was IFN modified with a linear 5-kDa PEG via an urea linkage. This pegylated IFN consisted of a mixture of 11 monopegylated species as demonstrated by analytical separation using IE-HPLC. All isomers showed the same range of biological activity in the cell assays as the parent monopegylated IFN [9]. However, with this 5-kDa monopegylated IFN an improvement of the dosing regiment towards less injections per week could not be achieved [10]. Since then, several further studies have been done to analyze the influence of the various PEG sizes and PEG types to in vitro and in vivo bioactivity. These results showed the important correlation of the PEG characteristics and the biological activity. Based on this knowledge, coupled with the latest advancements in pegylation procedures, a new generation PEG-interferon a-2a was developed [7]. The new generation PEG-interferon a-2a (40 kDa), an interferon a-2a conjugated via an amide linkage with a branched 40-kDa PEG (PEG-IFN), exhibits sustained adsorption and reduced renal clearance, resulting in a strong antiviral pressure throughout a once-weekly dosing schedule [11–13]. This report describes the isolation of the positional isomers of PEG-IFN based on their local charge differences. In addition, the results obtained from structural analyses and an antiviral assay of these chemically modified IFN species are reported here.
Materials and methods The PEG-IFN is produced at Hoffmann-La Roche. This pegylated interferon a-2a is prepared by the conjugation of lysine e-amino groups at the surface of the interferon molecule with an activated branched polyethylene glycol moiety of molecular weight 40 kDa. Pegylation reaction mixture was comprised of IFN and 40 kDa PEG-NHS reagent at 1:3 molar ratio in 50 mM sodium borate buffer, pH 9.0. The reaction was performed for 2 h at 4 °C [10,18]. Based on the average molecular weight of 44 kDa for the used PEG moiety
79
and a molecular weight of 19.2 kDa for IFN, the calculated average molecular weight for PEG-IFN is around 63 kDa. All other chemicals and solvents for this investigation were of the highest purity available. Endoproteinase Lys-C was purchased from Wako Biochemicals. Separation of positional isomers of PEG-IFN by ion exchange chromatography An analytical strong-cation exchange column was used to check the purity of the separated positional isomers (see below) from each purification step (TOSOH-BIOSEP, SP-5PW, 10 lm particle size, 7.5 mm diameter, 7.5 cm length). The column was pre-equilibrated with 3.4 mM sodium acetate, 10% ethanol, and 1% diethylene glycol, adjusted to pH 4.4 (buffer A). After loading the PEG-IFN samples, the column was washed with buffer A, followed by an ascending linear gradient to 10 mM dibasic potassium phosphate, 10% ethanol, and 1% diethylene glycol, adjusted to pH 6.6 (buffer B). The flow rate was 1.0 mL/min and the detection was at 215 nm. A two-step purification scheme was used to prepare the monopegylated isoforms of PEG-IFN. The first step is a separation of the positional isomers on a preparative low pressure liquid chromatography column with a weak-cation exchange matrix (TOSOH-BIOSEP, Toyopearl CM-650S, 16 mm diameter, 120 cm length, chromatogram not shown). A linear pH-gradient of increasing sodium acetate concentration (25 mM, pH 4.0 up 75 mM to pH 7.8) was applied at a flow rate of 0.7 mL/min. Detection was at 280 nm. With this chromatographic step, species 1, 2, 5, 6, and a mixture of 3, 4, 4a, 7, and 8 could be collected. The second preparation step was performed to further separate and increase the purity of the fractions from the first step. A preparative column with the same matrix as the analytical strong-cation exchange column but larger dimensions (30 mm i.d. and 70 mm length), further a higher flow rate and an extended run time was used. The resulting protein samples were collected manually and stored at 4 °C to be analyzed by a variety of protein chemical methods and an antiviral assay. The protein concentration of the PEG-IFN isomer was determined by spectrophotometry, based on the 1 mg=mL 280 nm absorption ðE280 ¼ 1:05Þ of the protein moiety of the PEG-IFN [10]. Mass spectrometry peptide mapping Mass spectra were recorded on a MALDI-TOF MS instrument (PerSeptive Biosystems Voyager-DE STR with delayed extraction). Each IEC fraction was desalted by dialysis, reduced with 0.02 M of 1,4-dithio-D L -
80
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
threitol (DTT), and alkylated with 0.2 M of 4-vinyl pyridine. Then the proteins were digested with endoproteinase Lys-C in 0.25 M Tris at pH 8.5 with an approximate enzyme to protein ratio of 1:30. The reaction was carried out overnight at 37 °C. A solution of 20 mg/mL a-cyano-4-hydroxycinnamic acid and 12 mg/mL nitrocellulose in acetone/isopropanol 40/60 (v/v) was used as matrix (thick-layer application). First, 0:5 lL matrix was placed on the target and allowed to dry. Then, 1:0 lL sample was added. The spectra were obtained in linear positive ionization mode with an accelerating voltage of 20,000 V and a grid voltage of 95%. At least 190 laser shots covering the complete spot were accumulated for each spectrum. Des-Arg1 -bradykinin and bovine insulin were used for internal calibration. Reverse-phase high-performance liquid chromatography peptide mapping The IEC fractions were reduced, alkylated, and digested with endoproteinase Lys-C as described for the MALDI-TOF MS peptide mapping. The analysis of the digested isomers was carried out on a Waters Alliance HPLC system with a Vydac RP-C18 analytical column (5 lm; 2:1 250 mm) and a precolumn with the same packing material. Elution was performed with an acetonitrile gradient from 1 to 95% for 105 min in water with a flow rate of 0.2 mL/min. Both solvents contained 0.1% (v/v) TFA. One hundred microliters of each digested sample was injected and monitored at 215 nm. MALDI-TOF spectra of undigested protein An 18 mg/mL solution of trans-3-indoleacrylic acid in acetonitrile/0.1% trifluoroacetic acid 70/30 (v/v) was premixed with the same volume of sample solution. Then 1:0 lL mixture was applied to the target surface. Typically 150–200 laser shots were averaged in linear positive ionization mode. The accelerating voltage was set to 25,000 V and the grid voltage to 90%. Bovine albumin Mþ and M2þ were used for external calibration. Size exclusion HPLC Size exculsion HPLC (SE-HPLC) was performed with a Waters Alliance 2690 HPLC system equipped with a TosoHaas TSK gel G 4000 SWXL column ð7:8 300 mmÞ. Proteins were eluted using a mobile phase containing 0.02 M NaH2 PO4 , 0.15 M NaCl, 1% (v/v) diethylene glycol, and 10% (v/v) ethanol (pH 6.8) at a flow rate of 0.4 mL/min and detected at 210 nm. The injection amounts were 20 lg of each isomer.
SDS–PAGE SDS–PAGE was carried out using Tris–glycine gels of 16% polyacrylamide (1.5 mm, 10 well). Novex Mark 12 molecular weight markers with a mass range from 2.5 to 200 kDa were used for calibration, bovine serum albumin (BSA) was used as sensitivity standard (2 ng). Approximately 1 lg of all the samples and 0:5 lg of standard were applied to the gel. The running conditions were 125 V and 6 W for 120 min. The proteins were fixed and stained using the silver staining kit SilverXpress from Novex. In vitro antiviral activity of PEG-IFN species The antiviral activity was estimated by its protective effect on Madin–Darby bovine kidney (MDBK) cells against the infection by vesicular stomatitis virus (VSV) and compared with a PEG-IFN standard. Samples and reference standard were diluted in EagleÕs minimum essential medium (MEM) containing 10% fetal bovine serum to a final concentration of 10 ng/mL (assay starting concentration). Each sample was assayed in quadruplicate.
Results and discussion IEC chromatography to separate the monopegylated isoforms of PEG-IFN A strong-cation high pressure liquid chromatography method was developed for the separation of monopegylated IFN isomers based on their local charge differences. An analytical elution profile of 180 lg PEG-IFN is shown in Fig. 1. The result of this method is a separation into 8 peaks, 2 peaks with baseline separation, and 6 with partial separation. The decrease of the baseline absorption towards the end of the chromatogram suggests that there were no other monopegylated species of IFN eluting at higher retention time. In addition, looking carefully at the IEC-chromatogram, a further peak close to the detection limit is visible between peaks 2 and 3, indicating the presence of additional positional isomers that should also contribute to the specific activity of the PEG-IFN mixture. Additional species are expected as the interferon a-2a molecule exhibits 12 sites for pegylation (11 lysines and the N-terminus). However, given the obvious low abundance of these species, they could not be isolated and characterized. One reason for the different yields of the individual pegylated species or even their absence could be the steric hindrance between the protein structure in the environment of individual lysines and the large size of the PEG tails.
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 1. Analytical IEC-HPLC of 180 lg PEG-IFN. An analytical strong-cation exchange column was used to check the purity of the separated positional isomers from each purification step (TOSOHBIOSEP, SP-5PW, 10 lm particle size, 7.5 mm diameter, 7.5 cm length).
It was shown that pegylation with a low molecular weight PEG-molecule, a 5-kDa, was less selective for the individual lysines and modification of all 11 lysines was found [9], but also no species with a modification on the amino terminus was identified. The stability of the isolated positional isomers from PEG-IFN, starting from the time of their isolation until the biochemical characterization, was a concern at the beginning of the investigations as the eluted samples in the elution buffer had been adjusted to pH 4 by acetic acid. These storage conditions differ from the established storage conditions for the drug substance, which has to be stored in a buffer of pH 6 and in a solid state at )80 °C. Therefore, isomer samples derived from IEC optimization runs were investigated directly after the isola-
81
tion ðt ¼ 0Þ and after 2 weeks of storage at 5 °C (data not shown). No significant differences were observed for the protein derived from IEC-peaks with regard to the protein content as determined by spectrometric methods; nor were there any changes to be detected in the monopegylation site, the content of oligo-PEG-IFN, the amount of aggregates, and the bioassay activity. Taking into account the relative abundance of the individual isomers—as determined by the IEC method—as well as the specific activities—as determined in the antiviral assay—almost the total specific bioactivity of the PEGIFN mixture used for their isolation is recovered (approximately 93%). It is likely that adsorption of the protein in highly diluted solutions to sample tubes, pipette tips, etc. is the main factor contributing to the gap in bioactivity of approximately 7%. The analytical IE-HPLC was used to check the purity of the individual isomers with respect to contamination with other positional isomers in the IEC fractions. The isomers 2, 3, 4, 4a, 5, and 7 had more than 98%, the isomers 1 and 8 had 93%, and isomer 6 had 88% purity. Size exclusion HPLC and SDS–PAGE were used to determine the amount of oligo-PEG-IFN forms and aggregates in the different IEC fractions. The reference material contains 2.3% aggregates and 2.2% oligomers (Fig. 4). Isomers 1, 4, 4a, 5, 6, and 8 contain <0.7% of the oligopegylated IFN forms, whereas in isomers 2, 3, and 7 the percentage of the oligopegylated IFN forms are under the detection limit (<0.2%). In the case of the aggregates a different trend could be seen. In all isomers the amount of aggregates is below 0.9%. SDS–PAGE was carried out both under non-reducing and under reducing conditions. The application of a non-reduced SDS gel in addition to the standard reduced gel was to demonstrate that no significant amount of the protein is crosslinked by disulfide bridges. The gels that were recorded under non-reducing conditions for the IEC fractions 1–8 (Fig. 2) show a
Fig. 2. (A/B) SDS–PAGE analysis with Tris–glycine (16%), the samples were electrophoresed under non-reduced conditions. The gels were stained for protein with silver stain. Lanes: M, molecular weight marker proteins; 1, Peak 1; 2, Peak 2; 3, Peak 3; 4, Peak 4; 5, Peak 4a; 6, Peak 5; 8, Peak 6; 9, Peak 7; 10, Peak 8; 7 and 11, 1 PEG-IFN standard; 12, 1.5 PEG-IFN standard; and C1 , IFN standard.
82
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 3. (A/B) SDS–PAGE analysis with Tris–glycine (16%), the samples were electrophoresed under reduced conditions. The gels were stained for protein with silver stain. Lanes: M, molecular weight marker proteins; 1, Peak 1; 2, Peak 2; 3, Peak 3; 4, Peak 4; 5, Peak 4a; 6, Peak 5; 8, Peak 6; 9, Peak 7; 10, Peak 8; 7 and 11, 1 PEG-IFN standard; 12, 1.5 PEG-IFN standard; and C1 , IFN standard.
major band of PEG-IFN around 97 kDa that is comparable to that of the PEG-IFN reference standard. Under reducing conditions (Fig. 3), the gels show the major bands of PEG-IFN at about 120 kDa. The explanation for the observed higher apparent molecular weight of the PEG-IFN compared to the calculated average molecular weight of around 63 kDa is the larger hydrodynamic volume of the PEG due to the binding of water. This effects that the electrophoretic mobility of pegylated IFN is considerably slowed [10]. Thus, PEGIFN appears to have a higher molecular weight than the calculated molecular weight when compared to unpegylated molecular weight marker protein. Between 6 and 10 kDa protein fragments appear for isomers 6, 7, and 8 (in Fig. 3 lanes 8, 9, and 10). These bands correspond to approximately 1% of clipped material. In the lanes of isomers 1, 6, 7, and 8 (in Fig. 3 lanes 1, 8, 9, and 10) additional bands with an apparent molecular weight of more than 200 kDa can be seen which are also present in the standard. These can be assigned to oligomers. Thus, SDS–PAGE confirms the results of the SE-HPLC analysis. Overall, RP-HPLC and SDS–PAGE experiments indicate that the purity of the IEC fractions can be considered comparable to the PEG-IFN reference standard. Structural analysis of the monopegylated isoforms of PEG-IFN The structure of the PEG-IFN species derived from the nine IEC-fractions were identified using the following strategy: First, the molecular weight of each isomer was determined by MALDI-TOF spectrometry in order to ensure that the PEG-IFN molecules were still intact after IEC chromatography and to confirm the monopegylation. Each IEC peak was measured without further modification. The spectra of all molecules show the ex-
pected broad Mþ peaks with maxima at 63 kDa and the corresponding M2þ peaks at 32 kDa and M3þ peaks at 21 kDa (Fig. 5). Second, each isomer was proteolytically digested using endo-Lys-C protease and the resulting MALDITOF peptide maps were compared with the one derived from the PEG-IFN reference standard. Interpretation of the spectra and structural identification of the positional isomers is based on the following considerations: 1. Dipegylation of the isomers can be ruled out because of the molecular weight determination of the entire molecule (see above). 2. The single lysine of a specific isomer having the PEG polymer group attached is not recognized as lysine by the endo-Lys-C protease [2] and, therefore, the polypeptide chain is not cleaved at that specific position. 3. It is therefore expected that the peptide map of a specific isomer is lacking the peptides (and only those peptides) which are related to its single pegylated lysine. 4. It is not expected to detect the mass peak of the peptides having the PEG residue attached to the MALDI-TOF peptide maps as the mass range chosen for the most accurate detection of the non-pegylated peptides ranges from 850 to 6000 Da. The PEG-moiety itself has already an average molecular weight of 44 kDa. However, the pegylated peptides have also been detected using the same digest and trans-3-indoleacrylic acid (IAA) as matrix. For each Lys-C digested isomer a broad peak at 46–47 kDa was observed, confirming the presence of the monopegylated peptides. Due to the broad mass distribution induced by the PEG-residue, no direct identification of the attached peptides could be made in these experiments (data not shown). The resulting peptide maps are shown in Fig. 6. Peaks that are missing in comparison to the standard are indicated by arrows.
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
83
Fig. 4. Size exclusion HPLC with different IEC fractions. Size exclusion (SE-) HPLC was used to determine the amount of oligo PEG-IFN forms and aggregates in the different IEC fractions. SE-HPLC was performed with a TosoHaas TSK gel G 4000 SWXL column ð7:8 300 mmÞ.
Regarding the spectra of the two references of interferon a-2a and PEG-IFN a-2a, no significant differences can be seen. Due to the fact that PEG-IFN is a mixture of different pegylation isomers, all peptide peaks detected for interferon are detected for PEGIFN, too. In the spectrum of the endo-Lys-C digested protein derived from IEC fraction 1 the peptides comprising amino acids 24–31 and 32–49 are missing in the region between 850 and 6000 Da, all other peaks are present. Therefore, the PEG residue must be attached to Lys 31.
The other fractions were identified in the same way. In each case the pegylated peptides are missing in comparison to the reference standard spectrum. For fractions 3 and 4a only one peptide peak is missing, for the second peptide 132–133 the mass is too small to be detected in the defined mass window. Only fraction 4a could not be identified with this method, no conclusions could be made. In order to identify isomer 4a, an endo-Lys-C peptide mapping method with RP-HPLC/UV detection has been developed. The protein was digested with endoproteinase Lys-C as described for the MALDI-TOF MS pep-
84
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 5. MALDI-TOF spectrometry of the different IEC fractions. MALDI-TOF spectrometry was used to determine the molecular weight of each isomer in order to ensure that the PEG-IFN molecules were still intact after IEC chromatography and to confirm the monopegylation.
tide mapping. The peptides were separated by means of a water/acetonitrile/TFA gradient. With the PEG-IFN reference standard, 13 peaks were observed. All fractions were collected manually and identified by MALDI-TOF mass spectrometry. The assignment of the pegylation site of IEC fraction 4a was again done by comparing the chromatogram of the sample to the one obtained for the reference material. The peak containing the two peptides 134–164 and 134–165 is clearly missing in the sample chromatogram and, therefore, IEC fraction 4a can be assigned to the isomer containing the PEG at Lys 164. The chromatograms of the PEG-IFN reference standard ð46 lg=mLÞ and the one of fraction 4a are shown in Fig. 7.
In conclusion, the identity of each of the 9 isomers isolated by IEC could be determined unambiguously. A graphical representation of the 9 PEG-IFN positional isomers isolated and characterized is given in Fig. 8. In vitro antiviral activity of PEG-IFN species The antiviral protection by PEG-IFN isomers of Madin–Darby bovine kidney cells (MDBK) infected with vesicular stomatitis virus was tested [20]. All isomers induced an activity in the antiviral assay as presented in Table 1. The activities range between 1061 and 339 U=lg, indicating that the difference in the specific
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
85
Fig. 6. MALDI-TOF Lys-C peptide maps of the PEG-IFN reference standard and the isomers 1, 2, 3, 4, 4a, 5, 6, 7, and 8. Missing peaks compared to the standard are indicated by arrows.
activities of the individual positional isomers is significant. A possible reason for the observed differences in the antiviral activities of the various PEG-IFN positional isomers could be that the chemical attachment of the
large PEG moiety at different sites to the IFN may lead to changes of the receptor–ligand interactions, specific for each isomer. This could also lead to different affinities which would explain the varying antiviral activities observed in our experiments.
86
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 7. RP-HPLC chromatograms of the Lys-C digests of the PEG-IFN reference and peak 4a.
Fig. 8. Ribbon structure of interferon a-2a showing the pegylation sites. This is the high resolution structure of human interferon a-2a determined with NMR spectroscopy [17]. The pegylation sites of pegylated interferon a-2a are colored black and labeled with residue type and residue number.
Furthermore, the amino acid domains Cys29 -Asp35 and Phe123 -Trp140 in IFN were defined as responsible for the receptor binding interaction [14–17]. These two domains are located on the surface of IFN and are in close spatial proximity. Interestingly, the four major pegylation sites in PEG-IFN (K31, K121, K134, and K131) also fall within these domains. All these points led us to think that the position of pegylation may have an influence on the interaction of the PEG-IFN with the receptor. The know-how and the results generated so far will allow the initiation of further investigations to establish this structure-function relationship between the positional isomers and the IFN a receptors. In summary, the separation of the positional isomers of PEG-IFN mixture was optimized to a stable and reproducible IEC method. The PEG-IFN isomers that were isolated comprised of an interferon a-2a being pegylated at Lys(31), Lys(134), Lys(70), Lys(83), Lys(121), Lys(131), Lys(49), Lys(112), and Lys(164), respectively. Lys(23), Lys(133), and the N-terminal PEG
Table 1 Identified PEG-peptides and the antiviral activities of individual positional isomers Antiviral activity
Identified PEG sites in peptide map
PEG-IFN Peak 1 Peak 2 Peak 3 Peak 4 Peak 4a Peak 5 Peak 6 Peak 7 Peak 8
PEG site
Missing peaks
Sequence
U=lg
K31 K134 K131 K121 K164 K70 K83 K49 K112
A, E I, I0 C B, C
24–49 134–164 122–131a 113–131 134–164a;b 50–83 71–112 32–70 84–121
1061 50 1818 127 1358 46 761 97 339 33 966 107 600 27 463 25 513 20 468 23
b
D, F D, H E, F B, H
The antiviral activity was determined in MDBK cells infected with vesicular stomatitis virus. The results present the averages of three assays performed independently. The antiviral activity of PEG-IFN is given for comparison. a 132–133 too small to detect. b RP-HPLC.
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
isomer could not be observed and isolated, most likely because the amino groups at this positions do not react due to the steric hindrance. As expected, individual isomers exhibit significantly distinct specific bioactivities but none are inactive. It is anticipated that the observed activity differences are related to differences in the isomer interferon receptor binding kinetics and complex stability.
Acknowledgments We thank Dr. Siegfried Jegge and Tania Ruchty for the SDS–PAGE, Joelle Bisch for performing antiviral bioassay. We are grateful to Dr. Salima Matthews for critical review of the manuscript.
References [1] G.R. Stark, How cells respond to interferons, Annu. Rev. Biochem. 67 (1998) 227–264. [2] S. Zeuzem, S.V. Feinman, J. Rasenack, J. Heathcote, L. MingYang, E. Gane, J. OÕGrady, J. Reichen, M. Diago, A. Lin, J. Hoffman, M.J. Brunda, Peginterferon alfa-2a in patients with chronic hepatitis C, N. Eng. J. Med. 343 (2000) 1666–1672. [3] E.J. Heathcote, M.L. Shiffman, W.G. Cooksley, G.M. Dusheiko, S.S. Lee, L. Balart, R. Reindollar, R.K. Reddy, T.L. Wrigth, A. Lin, J. Hoffman, J. De Pamphilis, Peginterferon alfa-2a in patients with chronic hepatitis C and cirrhosis, N. Eng. J. Med. 343 (2000) 673–1680. [4] R.J. Wills, Clinical pharmacokinetics of interferons, Clin. Pharmacokinet. 19 (1990) 390–399. [5] F.M. Barouki, F.R. Witter, D.E. Griffin, P.I. Nadler, A. Woods, D.L. Wood, P.S. Lietman, Time course of virus for 24 weeks following the completion of treatment of interferon levels, antiviral state, 20 ; 50 -oligoadenylate synthetase and side effects in healthy men, J. Interferon Res. 7 (1987) 29–39. [6] S. Zeuzem, J. Heathcote, N. Martin, K. Nieforth, M. Modi, Peginterferon alfa-2a (40 kDa) monotherapy: a novel agent for chronic hepatitis C therapy, Expert Opin. Invest. Drugs 10 (2001) 2201–2213. [7] J.M. Harris, N.E. Martin, M. Modi, Pegylation: A novel process for modifying pharmacokinetics, Clin. Pharmacokinet. 40 (2001) 539–551.
87
[8] T. Yamaoka, Y. Tabata, Y. Ikada, Distribution and tissue uptake of poly(ethylene glycol) with different molecular weights after intravenous administration to mice, J. Pharm. Sci. 4 (1994) 601– 606. [9] S.P. Monkarsh, Y. Ma, A. Aglione, P. Bailon, D. Ciolek, B. De Barbieri, M.C. Graves, K. Hollfelder, H. Michel, A. Palleroni, J.E. Porter, E. Russoman, S. Roy, Y.C. Pan, Positional isomers of monopegylated interferon a-2a: isolation, characterization, and biological activity, Anal. Biochem. 247 (1997) 434–440. [10] P. Bailon, A. Palleroni, C.A. Schaffer, C. Spence, F. Wen-Jian, J.E. Porter, G.K. Ehrlich, W.P. Zhi-Xin Xu, M.W. Modi, A. Farid, W. Berthold, Rational design of a potent, long-lasting form of interferon: A 40 kDa branched polyethylene glycol-conjugated interferon-2a for the treatment of hepatitis C, Bioconjugate Chem. 12 (2001) 195–202. [11] C.Z.-X. Xu, A. Rakhit, P. von Brummelen, PK/PD modelling approach to support clinical development of a long acting interferon (RO25-3036) for the treatment of hepatitis C, Hepatology 28 (1998) 702A. [12] M.C. Perry, B. Jarvis, Peginterferon-a-2a (40 kD): A review of its use in the management of chronic hepatitis C, Drugs 15 (2001) 2263–2288. [13] M.W. Lamb, E.N. Martin, Weight-based versus fixed dosing of peginterferon (40 kDa) alfa-2a, Ann. Pharmacother. 36 (2002) 933–938. [14] J. Piehler, L.C. Roisman, G. Schreiber, New structural and functional aspects of the type I interferon-receptor interaction revealed by comprehensive mutational analysis of the binding interface, J. Biol. Chem. 51 (2000) 40425–40433. [15] C.H. Jordan, R. Nivasch, R. Levy, S. Albeck, G. Schreiber, J. Anglister, The human interferon receptor: NMR-based modeling, mapping of the IFN-a2 binding site and observed ligand-induced tightening, Biochemistry 41 (2002) 3575–3585. [16] K. Morita, K. Tanaka, S. Saito, T. Kitamura, T. Kiba, T. Fujii, K. Numata, H. Sekihara, Expression of interferon receptor genes in the liver as a predictor of interferon response in patients with chronic hepatitis C, J. of Medical Virology 58 (1999) 359– 365. [17] W. Klaus, B. Gsell, A.M. Labhardt, B. Wipf, H. Senn, The threedimensional high resolution structure of human interferon a-2a determined by heteronuclear NMR spectroscopy in solution, J. Mol. Biol. 274 (1997) 661–675. [18] P.S. Ballon, A.V. Palleroni, European Patent No. 809996, 1997. [19] WHO Consultation, Global surveillance and control of hepatitis C, Report of a WHO Consultation organised in collaboration with the Viral Hepatitis Prevention Board, Antwerp, Belgium J. Viral Hepat. 6 (1999) 35–47. [20] S. Rubinstein, P. Familletti, S. Peska, Convenient assay for interferon, J. Virol. 37 (1981) 755–758.
Isolation, structural characterization, and antiviral activity of positional isomers of monopegylated interferon a-2a (PEGASYS) Stefan Foser,a Alfred Schacher,a Karl A. Weyer,a,* Doris Brugger,b Elke Dietel,c Stefan Marti,c and Thomas Schreitm€ ullerc a
Department of Pharma Technical Operations Biotechnology, Biotech Products, Biotech Development and Production, Hoffmann-La Roche Ltd., Building 066/508, Ch-4070 Basel, Switzerland b Department of Pharma Research Discovery Structure, Biotech Products, Hoffmann-La Roche Ltd., Basel, Switzerland c Department of Analytical Development and Quality Control, Biotech Products, Hoffmann-La Roche Ltd., Basel, Switzerland Received 12 December 2002, and in revised form 29 January 2003
Abstract Interferon a-2a plays an essential role in the treatment of chronic hepatitis C, but it is limited in its efficacy by the short in vivo half-life. To improve the half-life and efficacy, interferon a-2a is conjugated with a 40-kDa branched polyethylene glycol moiety (PEG-IFN, PEGASYS). From this preparation the positional PEG-IFN isomers were isolated and characterized by different analytical methods and antiviral assay. Two chromatographic steps were used to separate and purify nine isomers. The analytical methods IE-HPLC, RP-HPLC, SE-HPLC, SDS–PAGE, and MALDI-TOF MS indicated that each of these nine isomers is conjugated to the branched polyethylene glycol chain at a specific lysine. No isomer with a modification at the amino terminus was observed. All positional isomers induced viral protection of MDBK cells in the antiviral assay. When comparing the quantitative potency of the individual isomers with the whole mixture of PEG-IFN, significant differences in the specific activities were observed: PEG-Lys(31) and PEG-Lys(134) showed higher activities than the mixture, PEG-Lys(164) was equal to the mixture, whereas the activities of PEG-Lys(49), PEG-Lys(70), PEG-Lys(83), PEG-Lys(112), PEG-Lys(121), and PEG-Lys(131) were lower. Ó 2003 Elsevier Science (USA). All rights reserved. Keywords: Interferon a-2a; Pegylation; Forty kilodaltons of branched polyethylene glycol; Separation of positional isomers; Antiviral effect
Interferon a-2a (IFN) belongs to the class of type I interferons which are a family of homologous cytokines (interferon-a, -b, and --). These interferons have a wide variety of biological effects that include antiviral, antiproliferative, and immunomodulatory activities [1]. Since 1986 recombinant human IFN (ROFERON-A) has become an important therapeutic agent to treat patients with viral and oncological diseases. Worldwide, approximately 170 million people are infected with hepatitis C virus [19]. Most viral infections are primarily due to blood transfusions with blood products contaminated with virus. An essential component of the treatment of chronic hepatitis C infection has been interferon [2]. * Corresponding author. Fax: +41-61-688-2420. E-mail address: [email protected] (K.A. Weyer).
The frequency of the administration of interferon is dependent on the disease. For example, in a viral indication of hepatitis C the recommended administration is three times per week over 24–48 months, depending on the genotype of the virus [2,3]. With a subcutaneous or an intravenous administration the reported terminal half-life of IFN is in a range 4–8 h. After 24 h the serum concentration of IFN already decreases below its detection limit [4,5]. Since the early 1990s a major effort focussed on improving the half-life in blood, in order to decrease the frequency of administration of IFN [6]. One promising way considered was pegylation of IFN. The type of pegylation selected in this case was the modification of the lysine residues on the surface of the protein with a large, branched polyethylene glycol (PEG) through covalent linkage [7].
1046-5928/03/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S1046-5928(03)00055-X
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Pegylation changes physicochemical and biological properties of the protein. One effect, is the decrease of the proteolytic degradation and the renal clearance. This increases the half-life of the pegylated protein in blood. Another effect is the altered distribution in the body, depending on the size of the PEG moiety of the protein. This was shown in an animal study with pegylated interferons: increasing PEG molecular weight led to a decrease of renal clearance with a simultaneous increase of hepatic clearance [8]. The goal was to develop a pegylated IFN with an improved biological activity and the reduction of the administration frequency to one injection per week. An earlier developed molecule was IFN modified with a linear 5-kDa PEG via an urea linkage. This pegylated IFN consisted of a mixture of 11 monopegylated species as demonstrated by analytical separation using IE-HPLC. All isomers showed the same range of biological activity in the cell assays as the parent monopegylated IFN [9]. However, with this 5-kDa monopegylated IFN an improvement of the dosing regiment towards less injections per week could not be achieved [10]. Since then, several further studies have been done to analyze the influence of the various PEG sizes and PEG types to in vitro and in vivo bioactivity. These results showed the important correlation of the PEG characteristics and the biological activity. Based on this knowledge, coupled with the latest advancements in pegylation procedures, a new generation PEG-interferon a-2a was developed [7]. The new generation PEG-interferon a-2a (40 kDa), an interferon a-2a conjugated via an amide linkage with a branched 40-kDa PEG (PEG-IFN), exhibits sustained adsorption and reduced renal clearance, resulting in a strong antiviral pressure throughout a once-weekly dosing schedule [11–13]. This report describes the isolation of the positional isomers of PEG-IFN based on their local charge differences. In addition, the results obtained from structural analyses and an antiviral assay of these chemically modified IFN species are reported here.
Materials and methods The PEG-IFN is produced at Hoffmann-La Roche. This pegylated interferon a-2a is prepared by the conjugation of lysine e-amino groups at the surface of the interferon molecule with an activated branched polyethylene glycol moiety of molecular weight 40 kDa. Pegylation reaction mixture was comprised of IFN and 40 kDa PEG-NHS reagent at 1:3 molar ratio in 50 mM sodium borate buffer, pH 9.0. The reaction was performed for 2 h at 4 °C [10,18]. Based on the average molecular weight of 44 kDa for the used PEG moiety
79
and a molecular weight of 19.2 kDa for IFN, the calculated average molecular weight for PEG-IFN is around 63 kDa. All other chemicals and solvents for this investigation were of the highest purity available. Endoproteinase Lys-C was purchased from Wako Biochemicals. Separation of positional isomers of PEG-IFN by ion exchange chromatography An analytical strong-cation exchange column was used to check the purity of the separated positional isomers (see below) from each purification step (TOSOH-BIOSEP, SP-5PW, 10 lm particle size, 7.5 mm diameter, 7.5 cm length). The column was pre-equilibrated with 3.4 mM sodium acetate, 10% ethanol, and 1% diethylene glycol, adjusted to pH 4.4 (buffer A). After loading the PEG-IFN samples, the column was washed with buffer A, followed by an ascending linear gradient to 10 mM dibasic potassium phosphate, 10% ethanol, and 1% diethylene glycol, adjusted to pH 6.6 (buffer B). The flow rate was 1.0 mL/min and the detection was at 215 nm. A two-step purification scheme was used to prepare the monopegylated isoforms of PEG-IFN. The first step is a separation of the positional isomers on a preparative low pressure liquid chromatography column with a weak-cation exchange matrix (TOSOH-BIOSEP, Toyopearl CM-650S, 16 mm diameter, 120 cm length, chromatogram not shown). A linear pH-gradient of increasing sodium acetate concentration (25 mM, pH 4.0 up 75 mM to pH 7.8) was applied at a flow rate of 0.7 mL/min. Detection was at 280 nm. With this chromatographic step, species 1, 2, 5, 6, and a mixture of 3, 4, 4a, 7, and 8 could be collected. The second preparation step was performed to further separate and increase the purity of the fractions from the first step. A preparative column with the same matrix as the analytical strong-cation exchange column but larger dimensions (30 mm i.d. and 70 mm length), further a higher flow rate and an extended run time was used. The resulting protein samples were collected manually and stored at 4 °C to be analyzed by a variety of protein chemical methods and an antiviral assay. The protein concentration of the PEG-IFN isomer was determined by spectrophotometry, based on the 1 mg=mL 280 nm absorption ðE280 ¼ 1:05Þ of the protein moiety of the PEG-IFN [10]. Mass spectrometry peptide mapping Mass spectra were recorded on a MALDI-TOF MS instrument (PerSeptive Biosystems Voyager-DE STR with delayed extraction). Each IEC fraction was desalted by dialysis, reduced with 0.02 M of 1,4-dithio-D L -
80
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
threitol (DTT), and alkylated with 0.2 M of 4-vinyl pyridine. Then the proteins were digested with endoproteinase Lys-C in 0.25 M Tris at pH 8.5 with an approximate enzyme to protein ratio of 1:30. The reaction was carried out overnight at 37 °C. A solution of 20 mg/mL a-cyano-4-hydroxycinnamic acid and 12 mg/mL nitrocellulose in acetone/isopropanol 40/60 (v/v) was used as matrix (thick-layer application). First, 0:5 lL matrix was placed on the target and allowed to dry. Then, 1:0 lL sample was added. The spectra were obtained in linear positive ionization mode with an accelerating voltage of 20,000 V and a grid voltage of 95%. At least 190 laser shots covering the complete spot were accumulated for each spectrum. Des-Arg1 -bradykinin and bovine insulin were used for internal calibration. Reverse-phase high-performance liquid chromatography peptide mapping The IEC fractions were reduced, alkylated, and digested with endoproteinase Lys-C as described for the MALDI-TOF MS peptide mapping. The analysis of the digested isomers was carried out on a Waters Alliance HPLC system with a Vydac RP-C18 analytical column (5 lm; 2:1 250 mm) and a precolumn with the same packing material. Elution was performed with an acetonitrile gradient from 1 to 95% for 105 min in water with a flow rate of 0.2 mL/min. Both solvents contained 0.1% (v/v) TFA. One hundred microliters of each digested sample was injected and monitored at 215 nm. MALDI-TOF spectra of undigested protein An 18 mg/mL solution of trans-3-indoleacrylic acid in acetonitrile/0.1% trifluoroacetic acid 70/30 (v/v) was premixed with the same volume of sample solution. Then 1:0 lL mixture was applied to the target surface. Typically 150–200 laser shots were averaged in linear positive ionization mode. The accelerating voltage was set to 25,000 V and the grid voltage to 90%. Bovine albumin Mþ and M2þ were used for external calibration. Size exclusion HPLC Size exculsion HPLC (SE-HPLC) was performed with a Waters Alliance 2690 HPLC system equipped with a TosoHaas TSK gel G 4000 SWXL column ð7:8 300 mmÞ. Proteins were eluted using a mobile phase containing 0.02 M NaH2 PO4 , 0.15 M NaCl, 1% (v/v) diethylene glycol, and 10% (v/v) ethanol (pH 6.8) at a flow rate of 0.4 mL/min and detected at 210 nm. The injection amounts were 20 lg of each isomer.
SDS–PAGE SDS–PAGE was carried out using Tris–glycine gels of 16% polyacrylamide (1.5 mm, 10 well). Novex Mark 12 molecular weight markers with a mass range from 2.5 to 200 kDa were used for calibration, bovine serum albumin (BSA) was used as sensitivity standard (2 ng). Approximately 1 lg of all the samples and 0:5 lg of standard were applied to the gel. The running conditions were 125 V and 6 W for 120 min. The proteins were fixed and stained using the silver staining kit SilverXpress from Novex. In vitro antiviral activity of PEG-IFN species The antiviral activity was estimated by its protective effect on Madin–Darby bovine kidney (MDBK) cells against the infection by vesicular stomatitis virus (VSV) and compared with a PEG-IFN standard. Samples and reference standard were diluted in EagleÕs minimum essential medium (MEM) containing 10% fetal bovine serum to a final concentration of 10 ng/mL (assay starting concentration). Each sample was assayed in quadruplicate.
Results and discussion IEC chromatography to separate the monopegylated isoforms of PEG-IFN A strong-cation high pressure liquid chromatography method was developed for the separation of monopegylated IFN isomers based on their local charge differences. An analytical elution profile of 180 lg PEG-IFN is shown in Fig. 1. The result of this method is a separation into 8 peaks, 2 peaks with baseline separation, and 6 with partial separation. The decrease of the baseline absorption towards the end of the chromatogram suggests that there were no other monopegylated species of IFN eluting at higher retention time. In addition, looking carefully at the IEC-chromatogram, a further peak close to the detection limit is visible between peaks 2 and 3, indicating the presence of additional positional isomers that should also contribute to the specific activity of the PEG-IFN mixture. Additional species are expected as the interferon a-2a molecule exhibits 12 sites for pegylation (11 lysines and the N-terminus). However, given the obvious low abundance of these species, they could not be isolated and characterized. One reason for the different yields of the individual pegylated species or even their absence could be the steric hindrance between the protein structure in the environment of individual lysines and the large size of the PEG tails.
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 1. Analytical IEC-HPLC of 180 lg PEG-IFN. An analytical strong-cation exchange column was used to check the purity of the separated positional isomers from each purification step (TOSOHBIOSEP, SP-5PW, 10 lm particle size, 7.5 mm diameter, 7.5 cm length).
It was shown that pegylation with a low molecular weight PEG-molecule, a 5-kDa, was less selective for the individual lysines and modification of all 11 lysines was found [9], but also no species with a modification on the amino terminus was identified. The stability of the isolated positional isomers from PEG-IFN, starting from the time of their isolation until the biochemical characterization, was a concern at the beginning of the investigations as the eluted samples in the elution buffer had been adjusted to pH 4 by acetic acid. These storage conditions differ from the established storage conditions for the drug substance, which has to be stored in a buffer of pH 6 and in a solid state at )80 °C. Therefore, isomer samples derived from IEC optimization runs were investigated directly after the isola-
81
tion ðt ¼ 0Þ and after 2 weeks of storage at 5 °C (data not shown). No significant differences were observed for the protein derived from IEC-peaks with regard to the protein content as determined by spectrometric methods; nor were there any changes to be detected in the monopegylation site, the content of oligo-PEG-IFN, the amount of aggregates, and the bioassay activity. Taking into account the relative abundance of the individual isomers—as determined by the IEC method—as well as the specific activities—as determined in the antiviral assay—almost the total specific bioactivity of the PEGIFN mixture used for their isolation is recovered (approximately 93%). It is likely that adsorption of the protein in highly diluted solutions to sample tubes, pipette tips, etc. is the main factor contributing to the gap in bioactivity of approximately 7%. The analytical IE-HPLC was used to check the purity of the individual isomers with respect to contamination with other positional isomers in the IEC fractions. The isomers 2, 3, 4, 4a, 5, and 7 had more than 98%, the isomers 1 and 8 had 93%, and isomer 6 had 88% purity. Size exclusion HPLC and SDS–PAGE were used to determine the amount of oligo-PEG-IFN forms and aggregates in the different IEC fractions. The reference material contains 2.3% aggregates and 2.2% oligomers (Fig. 4). Isomers 1, 4, 4a, 5, 6, and 8 contain <0.7% of the oligopegylated IFN forms, whereas in isomers 2, 3, and 7 the percentage of the oligopegylated IFN forms are under the detection limit (<0.2%). In the case of the aggregates a different trend could be seen. In all isomers the amount of aggregates is below 0.9%. SDS–PAGE was carried out both under non-reducing and under reducing conditions. The application of a non-reduced SDS gel in addition to the standard reduced gel was to demonstrate that no significant amount of the protein is crosslinked by disulfide bridges. The gels that were recorded under non-reducing conditions for the IEC fractions 1–8 (Fig. 2) show a
Fig. 2. (A/B) SDS–PAGE analysis with Tris–glycine (16%), the samples were electrophoresed under non-reduced conditions. The gels were stained for protein with silver stain. Lanes: M, molecular weight marker proteins; 1, Peak 1; 2, Peak 2; 3, Peak 3; 4, Peak 4; 5, Peak 4a; 6, Peak 5; 8, Peak 6; 9, Peak 7; 10, Peak 8; 7 and 11, 1 PEG-IFN standard; 12, 1.5 PEG-IFN standard; and C1 , IFN standard.
82
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 3. (A/B) SDS–PAGE analysis with Tris–glycine (16%), the samples were electrophoresed under reduced conditions. The gels were stained for protein with silver stain. Lanes: M, molecular weight marker proteins; 1, Peak 1; 2, Peak 2; 3, Peak 3; 4, Peak 4; 5, Peak 4a; 6, Peak 5; 8, Peak 6; 9, Peak 7; 10, Peak 8; 7 and 11, 1 PEG-IFN standard; 12, 1.5 PEG-IFN standard; and C1 , IFN standard.
major band of PEG-IFN around 97 kDa that is comparable to that of the PEG-IFN reference standard. Under reducing conditions (Fig. 3), the gels show the major bands of PEG-IFN at about 120 kDa. The explanation for the observed higher apparent molecular weight of the PEG-IFN compared to the calculated average molecular weight of around 63 kDa is the larger hydrodynamic volume of the PEG due to the binding of water. This effects that the electrophoretic mobility of pegylated IFN is considerably slowed [10]. Thus, PEGIFN appears to have a higher molecular weight than the calculated molecular weight when compared to unpegylated molecular weight marker protein. Between 6 and 10 kDa protein fragments appear for isomers 6, 7, and 8 (in Fig. 3 lanes 8, 9, and 10). These bands correspond to approximately 1% of clipped material. In the lanes of isomers 1, 6, 7, and 8 (in Fig. 3 lanes 1, 8, 9, and 10) additional bands with an apparent molecular weight of more than 200 kDa can be seen which are also present in the standard. These can be assigned to oligomers. Thus, SDS–PAGE confirms the results of the SE-HPLC analysis. Overall, RP-HPLC and SDS–PAGE experiments indicate that the purity of the IEC fractions can be considered comparable to the PEG-IFN reference standard. Structural analysis of the monopegylated isoforms of PEG-IFN The structure of the PEG-IFN species derived from the nine IEC-fractions were identified using the following strategy: First, the molecular weight of each isomer was determined by MALDI-TOF spectrometry in order to ensure that the PEG-IFN molecules were still intact after IEC chromatography and to confirm the monopegylation. Each IEC peak was measured without further modification. The spectra of all molecules show the ex-
pected broad Mþ peaks with maxima at 63 kDa and the corresponding M2þ peaks at 32 kDa and M3þ peaks at 21 kDa (Fig. 5). Second, each isomer was proteolytically digested using endo-Lys-C protease and the resulting MALDITOF peptide maps were compared with the one derived from the PEG-IFN reference standard. Interpretation of the spectra and structural identification of the positional isomers is based on the following considerations: 1. Dipegylation of the isomers can be ruled out because of the molecular weight determination of the entire molecule (see above). 2. The single lysine of a specific isomer having the PEG polymer group attached is not recognized as lysine by the endo-Lys-C protease [2] and, therefore, the polypeptide chain is not cleaved at that specific position. 3. It is therefore expected that the peptide map of a specific isomer is lacking the peptides (and only those peptides) which are related to its single pegylated lysine. 4. It is not expected to detect the mass peak of the peptides having the PEG residue attached to the MALDI-TOF peptide maps as the mass range chosen for the most accurate detection of the non-pegylated peptides ranges from 850 to 6000 Da. The PEG-moiety itself has already an average molecular weight of 44 kDa. However, the pegylated peptides have also been detected using the same digest and trans-3-indoleacrylic acid (IAA) as matrix. For each Lys-C digested isomer a broad peak at 46–47 kDa was observed, confirming the presence of the monopegylated peptides. Due to the broad mass distribution induced by the PEG-residue, no direct identification of the attached peptides could be made in these experiments (data not shown). The resulting peptide maps are shown in Fig. 6. Peaks that are missing in comparison to the standard are indicated by arrows.
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
83
Fig. 4. Size exclusion HPLC with different IEC fractions. Size exclusion (SE-) HPLC was used to determine the amount of oligo PEG-IFN forms and aggregates in the different IEC fractions. SE-HPLC was performed with a TosoHaas TSK gel G 4000 SWXL column ð7:8 300 mmÞ.
Regarding the spectra of the two references of interferon a-2a and PEG-IFN a-2a, no significant differences can be seen. Due to the fact that PEG-IFN is a mixture of different pegylation isomers, all peptide peaks detected for interferon are detected for PEGIFN, too. In the spectrum of the endo-Lys-C digested protein derived from IEC fraction 1 the peptides comprising amino acids 24–31 and 32–49 are missing in the region between 850 and 6000 Da, all other peaks are present. Therefore, the PEG residue must be attached to Lys 31.
The other fractions were identified in the same way. In each case the pegylated peptides are missing in comparison to the reference standard spectrum. For fractions 3 and 4a only one peptide peak is missing, for the second peptide 132–133 the mass is too small to be detected in the defined mass window. Only fraction 4a could not be identified with this method, no conclusions could be made. In order to identify isomer 4a, an endo-Lys-C peptide mapping method with RP-HPLC/UV detection has been developed. The protein was digested with endoproteinase Lys-C as described for the MALDI-TOF MS pep-
84
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 5. MALDI-TOF spectrometry of the different IEC fractions. MALDI-TOF spectrometry was used to determine the molecular weight of each isomer in order to ensure that the PEG-IFN molecules were still intact after IEC chromatography and to confirm the monopegylation.
tide mapping. The peptides were separated by means of a water/acetonitrile/TFA gradient. With the PEG-IFN reference standard, 13 peaks were observed. All fractions were collected manually and identified by MALDI-TOF mass spectrometry. The assignment of the pegylation site of IEC fraction 4a was again done by comparing the chromatogram of the sample to the one obtained for the reference material. The peak containing the two peptides 134–164 and 134–165 is clearly missing in the sample chromatogram and, therefore, IEC fraction 4a can be assigned to the isomer containing the PEG at Lys 164. The chromatograms of the PEG-IFN reference standard ð46 lg=mLÞ and the one of fraction 4a are shown in Fig. 7.
In conclusion, the identity of each of the 9 isomers isolated by IEC could be determined unambiguously. A graphical representation of the 9 PEG-IFN positional isomers isolated and characterized is given in Fig. 8. In vitro antiviral activity of PEG-IFN species The antiviral protection by PEG-IFN isomers of Madin–Darby bovine kidney cells (MDBK) infected with vesicular stomatitis virus was tested [20]. All isomers induced an activity in the antiviral assay as presented in Table 1. The activities range between 1061 and 339 U=lg, indicating that the difference in the specific
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
85
Fig. 6. MALDI-TOF Lys-C peptide maps of the PEG-IFN reference standard and the isomers 1, 2, 3, 4, 4a, 5, 6, 7, and 8. Missing peaks compared to the standard are indicated by arrows.
activities of the individual positional isomers is significant. A possible reason for the observed differences in the antiviral activities of the various PEG-IFN positional isomers could be that the chemical attachment of the
large PEG moiety at different sites to the IFN may lead to changes of the receptor–ligand interactions, specific for each isomer. This could also lead to different affinities which would explain the varying antiviral activities observed in our experiments.
86
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
Fig. 7. RP-HPLC chromatograms of the Lys-C digests of the PEG-IFN reference and peak 4a.
Fig. 8. Ribbon structure of interferon a-2a showing the pegylation sites. This is the high resolution structure of human interferon a-2a determined with NMR spectroscopy [17]. The pegylation sites of pegylated interferon a-2a are colored black and labeled with residue type and residue number.
Furthermore, the amino acid domains Cys29 -Asp35 and Phe123 -Trp140 in IFN were defined as responsible for the receptor binding interaction [14–17]. These two domains are located on the surface of IFN and are in close spatial proximity. Interestingly, the four major pegylation sites in PEG-IFN (K31, K121, K134, and K131) also fall within these domains. All these points led us to think that the position of pegylation may have an influence on the interaction of the PEG-IFN with the receptor. The know-how and the results generated so far will allow the initiation of further investigations to establish this structure-function relationship between the positional isomers and the IFN a receptors. In summary, the separation of the positional isomers of PEG-IFN mixture was optimized to a stable and reproducible IEC method. The PEG-IFN isomers that were isolated comprised of an interferon a-2a being pegylated at Lys(31), Lys(134), Lys(70), Lys(83), Lys(121), Lys(131), Lys(49), Lys(112), and Lys(164), respectively. Lys(23), Lys(133), and the N-terminal PEG
Table 1 Identified PEG-peptides and the antiviral activities of individual positional isomers Antiviral activity
Identified PEG sites in peptide map
PEG-IFN Peak 1 Peak 2 Peak 3 Peak 4 Peak 4a Peak 5 Peak 6 Peak 7 Peak 8
PEG site
Missing peaks
Sequence
U=lg
K31 K134 K131 K121 K164 K70 K83 K49 K112
A, E I, I0 C B, C
24–49 134–164 122–131a 113–131 134–164a;b 50–83 71–112 32–70 84–121
1061 50 1818 127 1358 46 761 97 339 33 966 107 600 27 463 25 513 20 468 23
b
D, F D, H E, F B, H
The antiviral activity was determined in MDBK cells infected with vesicular stomatitis virus. The results present the averages of three assays performed independently. The antiviral activity of PEG-IFN is given for comparison. a 132–133 too small to detect. b RP-HPLC.
S. Foser et al. / Protein Expression and Purification 30 (2003) 78–87
isomer could not be observed and isolated, most likely because the amino groups at this positions do not react due to the steric hindrance. As expected, individual isomers exhibit significantly distinct specific bioactivities but none are inactive. It is anticipated that the observed activity differences are related to differences in the isomer interferon receptor binding kinetics and complex stability.
Acknowledgments We thank Dr. Siegfried Jegge and Tania Ruchty for the SDS–PAGE, Joelle Bisch for performing antiviral bioassay. We are grateful to Dr. Salima Matthews for critical review of the manuscript.
References [1] G.R. Stark, How cells respond to interferons, Annu. Rev. Biochem. 67 (1998) 227–264. [2] S. Zeuzem, S.V. Feinman, J. Rasenack, J. Heathcote, L. MingYang, E. Gane, J. OÕGrady, J. Reichen, M. Diago, A. Lin, J. Hoffman, M.J. Brunda, Peginterferon alfa-2a in patients with chronic hepatitis C, N. Eng. J. Med. 343 (2000) 1666–1672. [3] E.J. Heathcote, M.L. Shiffman, W.G. Cooksley, G.M. Dusheiko, S.S. Lee, L. Balart, R. Reindollar, R.K. Reddy, T.L. Wrigth, A. Lin, J. Hoffman, J. De Pamphilis, Peginterferon alfa-2a in patients with chronic hepatitis C and cirrhosis, N. Eng. J. Med. 343 (2000) 673–1680. [4] R.J. Wills, Clinical pharmacokinetics of interferons, Clin. Pharmacokinet. 19 (1990) 390–399. [5] F.M. Barouki, F.R. Witter, D.E. Griffin, P.I. Nadler, A. Woods, D.L. Wood, P.S. Lietman, Time course of virus for 24 weeks following the completion of treatment of interferon levels, antiviral state, 20 ; 50 -oligoadenylate synthetase and side effects in healthy men, J. Interferon Res. 7 (1987) 29–39. [6] S. Zeuzem, J. Heathcote, N. Martin, K. Nieforth, M. Modi, Peginterferon alfa-2a (40 kDa) monotherapy: a novel agent for chronic hepatitis C therapy, Expert Opin. Invest. Drugs 10 (2001) 2201–2213. [7] J.M. Harris, N.E. Martin, M. Modi, Pegylation: A novel process for modifying pharmacokinetics, Clin. Pharmacokinet. 40 (2001) 539–551.
87
[8] T. Yamaoka, Y. Tabata, Y. Ikada, Distribution and tissue uptake of poly(ethylene glycol) with different molecular weights after intravenous administration to mice, J. Pharm. Sci. 4 (1994) 601– 606. [9] S.P. Monkarsh, Y. Ma, A. Aglione, P. Bailon, D. Ciolek, B. De Barbieri, M.C. Graves, K. Hollfelder, H. Michel, A. Palleroni, J.E. Porter, E. Russoman, S. Roy, Y.C. Pan, Positional isomers of monopegylated interferon a-2a: isolation, characterization, and biological activity, Anal. Biochem. 247 (1997) 434–440. [10] P. Bailon, A. Palleroni, C.A. Schaffer, C. Spence, F. Wen-Jian, J.E. Porter, G.K. Ehrlich, W.P. Zhi-Xin Xu, M.W. Modi, A. Farid, W. Berthold, Rational design of a potent, long-lasting form of interferon: A 40 kDa branched polyethylene glycol-conjugated interferon-2a for the treatment of hepatitis C, Bioconjugate Chem. 12 (2001) 195–202. [11] C.Z.-X. Xu, A. Rakhit, P. von Brummelen, PK/PD modelling approach to support clinical development of a long acting interferon (RO25-3036) for the treatment of hepatitis C, Hepatology 28 (1998) 702A. [12] M.C. Perry, B. Jarvis, Peginterferon-a-2a (40 kD): A review of its use in the management of chronic hepatitis C, Drugs 15 (2001) 2263–2288. [13] M.W. Lamb, E.N. Martin, Weight-based versus fixed dosing of peginterferon (40 kDa) alfa-2a, Ann. Pharmacother. 36 (2002) 933–938. [14] J. Piehler, L.C. Roisman, G. Schreiber, New structural and functional aspects of the type I interferon-receptor interaction revealed by comprehensive mutational analysis of the binding interface, J. Biol. Chem. 51 (2000) 40425–40433. [15] C.H. Jordan, R. Nivasch, R. Levy, S. Albeck, G. Schreiber, J. Anglister, The human interferon receptor: NMR-based modeling, mapping of the IFN-a2 binding site and observed ligand-induced tightening, Biochemistry 41 (2002) 3575–3585. [16] K. Morita, K. Tanaka, S. Saito, T. Kitamura, T. Kiba, T. Fujii, K. Numata, H. Sekihara, Expression of interferon receptor genes in the liver as a predictor of interferon response in patients with chronic hepatitis C, J. of Medical Virology 58 (1999) 359– 365. [17] W. Klaus, B. Gsell, A.M. Labhardt, B. Wipf, H. Senn, The threedimensional high resolution structure of human interferon a-2a determined by heteronuclear NMR spectroscopy in solution, J. Mol. Biol. 274 (1997) 661–675. [18] P.S. Ballon, A.V. Palleroni, European Patent No. 809996, 1997. [19] WHO Consultation, Global surveillance and control of hepatitis C, Report of a WHO Consultation organised in collaboration with the Viral Hepatitis Prevention Board, Antwerp, Belgium J. Viral Hepat. 6 (1999) 35–47. [20] S. Rubinstein, P. Familletti, S. Peska, Convenient assay for interferon, J. Virol. 37 (1981) 755–758.