Abstract
Interferon therapies suffer from a relatively short half-life of the products in circulation. To address this issue we investigated the effects of polyethylene glycol modification (PEGylation) on the pharmacokinetic properties of human interferon (IFN)-β-1a. PEGylation with a linear 20-kDa PEG targeted at a single site on the N-terminal amine had no deleterious effect on its specific activity in an in vitro antiviral assay. In monkeys, PEG IFN-β-1a treatment induced neopterin and β2-microglobulin expression (pharmacodynamic markers of activity). Systemic clearance values in monkeys, rats, and mice decreased, respectively, from 232, 261, and 247 ml/h/kg for the unmodified IFN-β-1a to 30.5, 19.2, and 18.7 ml/h/kg for the PEGylated form, while volume of distribution values decreased from 427, 280, and 328 ml/kg to 284, 173, and 150 ml/kg. The decreased clearance and volume of distribution resulted in higher serum antiviral activity in the PEG IFN-β-1a-treated animals. In the rat, a more extensive set of dosing routes was investigated, including intraperitoneal, intratracheal, and oral administration. Bioavailability for the PEG IFN-β-1a was similar to the unmodified protein for each of the extravascular routes examined. For the intraperitoneal route, bioavailability was almost 100%, whereas for the oral and intratracheal routes absorption was low (<5%). In rats, subcutaneous bioavailability was moderate (28%), whereas in monkeys it was approximately 100%. In all instances an improved pharmacokinetic profile for the PEGylated IFN-β-1a was observed. These findings demonstrate that PEGylation greatly alters the pharmacokinetic properties of IFN-β-1a, resulting in an increase in systemic exposure following diverse routes of administration.
Type 1 interferons (IFN-α, -β, -κ, -τ, and -ω) are a family of structurally and functionally related proteins that bind to a common multimeric cell surface receptor. Productive receptor binding results in a cascade of intracellular events leading to the expression of interferon-inducible genes and proteins (Uze et al., 1995; Peters, 1996), which in turn leads to diverse effects that can be classified according to their antiviral, antiproliferative, immunomodulatory, and other activities (Tyring, 1995; Weinstock-Guttman et al., 1995; Peters, 1996). These activities form the basis for the clinical benefits that have been observed with interferon therapy (Baron et al., 1992; Weinstock-Guttman et al., 1995). Both IFN-α and IFN-β therapies have been successful in the clinic, leading further to the interest in this family of cytokines. Human IFN-β is a 166 amino acid glycoprotein that can be produced by most cells in the body in response to viral infection or exposure to other biologics (Stewart, 1981; Sen and Lengyel, 1992). Two recombinant IFN-β therapeutics, AVONEX (IFN-β-1a) and Betaseron (IFN-β-1b), are approved for treatment of multiple sclerosis in the United States (Jacobs et al., 1996; The IFNβ Multiple Sclerosis Study Group, 1993). Recombinant IFN-β-1a is expressed in Chinese hamster ovary cells to produce a glycosylated form analogous to natural IFN-β (Kagawa et al., 1988). Recombinant IFN-β-1b is expressed as a nonglycosylated protein inEscherichia coli (Derynck et al., 1980; Mark et al., 1984). Recombinant IFN-α is a 165 amino acid protein that is expressed inE. coli. Recombinant IFN-α2 is approved for the treatment of hepatitis B and C, and various oncological indications.
Like many low molecular weight protein drugs, interferon therapies suffer from a relatively short serum half-life of the products. Consequently, if vascular retention is considered to be desired for enhanced efficacy, strategies that can improve a drug's pharmacokinetic and pharmacodynamic properties might improve its therapeutic benefits. PEGylation is one of several strategies that has been successfully applied to proteins to alter their pharmacokinetic and pharmacodynamic properties (for reviews, see Francis et al., 1996,1998; Delgado et al., 1992, 1997). The beneficial effects of PEGylation can be achieved by any of a number of mechanisms, including increased solubility of the PEGylated product, reduced renal clearance and clearance by antibodies, reduced proteolysis, and/or steric hindrance to receptor-mediated clearance. Some effects are product specific and, importantly, can vary depending on the site of attachment, the chemistry used for generating the conjugate, and the characteristics of the PEG itself (size, branching, etc.). In generating PEG conjugates a significant effort goes into the design of the conjugate. Recent positive clinical data from Schering-Plough for a PEGylated form of human IFN-α-2b (PEG-INTRON) have validated this strategy for cytokine targets for the treatment of hepatitis C (Glue et al., 1999, 2000). For PEG-INTRON, a nonselective, succinimidyl carbonate chemistry was used and consequently the product is a mixture of PEGylated forms (Wang et al., 2000). Although the in vitro potency of PEG-INTRON was reduced by about 75%, its serum half-life was increased by about 6-fold, which enabled less frequent administration of the drug for efficacy (Glue et al., 1999).
The crystal structure of IFN-β-1a has provided a model for understanding the properties of the protein at a molecular level (Karpusas et al., 1997; Runkel et al., 1998). Based on the structure and structure-function data (Runkel et al., 2000), which allowed us to identify functionally important regions in the cytokine, we identified surface residues that were distinct from the receptor binding site and therefore potentially could be modified without loss of function. Of these sites, the N terminus was particularly attractive because chemistries exist to selectively target the N-terminal amine over other amines and thereby provide a site-specific modification. Indeed, selective modification of the N terminus of IFN-β-1a with polyethylene glycol had no deleterious effect on its antiviral activity in vitro (see below). Here we have used PEGylated IFN-β-1a in animals to study the effects of PEGylation on clearance following intravenous administration, and on bioavailability and clearance following subcutaneous, intraperitoneal, intratracheal, and oral administration. The data highlight the effects of PEGylation and of route of administration on systemic IFN-β-1a exposure.
Materials and Methods
Preparation and Characterization of PEGylated IFN-β-1a.
Recombinant human IFN-β-1a (Biogen, Inc., Cambridge, MA) at 125 μg/ml in 50 mM sodium phosphate, 50 mM 2-[N-morpholino]ethanesulfonic acid, pH 5.0, 100 mM NaCl, was loaded onto an SP-Sepharose column (Amersham Pharmacia Biotech, Piscataway, NJ) at 6 mg protein/ml resin. The column was washed with 5 mM sodium phosphate pH 5.5, 75 mM NaCl, and the product was eluted with 30 mM sodium phosphate pH 6.0, 600 mM NaCl. Fractions containing protein were analyzed for their absorbance at 280 nm and the concentration of interferon was calculated from the absorbance using an extinction coefficient of 1.51 for a 1-mg/ml solution. IFN-β-1a in the SP eluate was diluted to 1 mg/ml and treated with 20-kDa PEG aldehyde (Shearwater Polymers, Huntsville, AL). The final reaction mix contained 50 mM sodium phosphate pH 6.0, 5 mM sodium cyanoborohydride (Aldrich, Milwaukee, WI), and 5 mg/ml PEG aldehyde. The sample was incubated at room temperature for 20 h. The PEGylated interferon was purified from reaction products by sequential chromatography steps on a Superose 6 FPLC sizing column (Amersham Pharmacia Biotech), and an SP-Sepharose column. The sizing column was run using 5 mM sodium phosphate pH 5.5, 150 mM NaCl as the mobile phase and resulted in baseline separation of the unmodified and PEGylated IFN-β-1a. The PEG IFN-β-1a-containing elution pool from the gel filtration step was diluted 1:1 with water and loaded at 2 mg of PEG IFN-β-1a/ml resin onto the SP-Sepharose column. The column was washed with 5 mM sodium phosphate pH 5.5, 75 mM NaCl, and the PEGylated IFN-β-1a was eluted from the column with 5 mM sodium phosphate pH 5.5, 800 mM NaCl. Elution fractions were analyzed for protein content by absorbance at 280 nm. The concentration of PEGylated interferon is reported in interferon equivalents as the PEG moiety did not contribute to absorbance at 280 nm.
Samples were analyzed for extent of reaction and purity by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on a 10 to 20% gradient gel (Owl Separation Systems, Portsmouth, NH). Proteins were detected by staining with Coomassie brilliant blue R-250. Prior to electrophoresis, samples were heated at 65°C for 5 min in reducing electrophoresis sample buffer (50 mM Tris-HCl pH 6.8, 2% sodium dodecyl sulfate, 12.5% glycerol, 0.1% bromophenol blue, 2% mercaptoethanol).
The specificity of the PEGylation reaction was evaluated by peptide mapping. Aliquots (20 μg) of PEGylated, and unmodified IFN-β-1a as a control, in 240 μl of 200 mM Tris-HCl pH 9.0, 1 mM EDTA were digested with 1.5 μg of lysyl endoproteinase from Achromobacter (Wako Bioproducts, Richmond, VA) for 3 to 4 h at 27°C. Guanidine HCl (200 mg) was added to each sample and the cleavage products were fractionated by reverse phase high-pressure liquid chromatography on a Vydac C4 column (0.21 cm internal diameter × 25 cm). The column was developed with a 70-min gradient from 0.1% trifluoroacetic acid in water to 0.1% trifluoroacetic acid in 70% acetonitrile at a flow rate of 1.4 ml/min. The column effluent was monitored for absorbance at 215 nm.
The PEGylated IFN-β-1a was stored at −70°C at 75 μg/ml in formulation buffer (50 mM sodium phosphate pH 7.2, 100 mM NaCl, 14 mg/ml human serum albumin). Immediately prior to use, a sample was thawed and diluted further with formulation buffer to the desired concentration.
Antiviral Activity Assays.
Serum levels of human IFN-β-1a were determined by testing serum samples for antiviral activity using a cytopathic effect assay. In this assay, human lung carcinoma A549 cells (CCL-185; American Type Culture Collection, Rockville, MD) are exposed to encephalomyocarditis (EMC) virus. Two versions of the assay were used: one described in detail in the article by Alam et al. (1997) in which viable cells were detected using crystal violet staining and samples were analyzed in quadruplicate, and the other in which viable cells were detected using the metabolic dye 2,3-bis[2- methoxy-4-nitro-5-sulfo-phenyl]-2H-tetrazolium-5-carboxyanilide (XTT) and samples were tested in triplicate. For the latter assay, A549 cells (1.5 × 104 cells/100 μl/well) were plated into 96-well plates. The following day, serial dilutions of IFN-β-1a test samples were added, and 24 h later the cells were challenged with EMC virus. Dilutions of an IFN-β-1a standard (10, 6.7, 4.4, 2.9, 1.3, 0.9, and 0.6 U/ml) were assayed on every plate. Viable cells were quantified 2 days after viral challenge with XTT (Jost et al., 1992). Cell viability was assessed from absorbance at 450 nm (OD450). Standard curves were generated for each plate and used to determine the amount of IFN-β activity in each test sample. One laboratory unit is defined as the IFN-β concentration conferring 50% protection from viral killing (50% maximum OD450).
Assessing Pharmacokinetics of Human PEGylated IFN-β-1a in Mice, Rats, and Monkeys.
C57Bl/6 mice were injected i.v. through the tail vein with 2 MU/kg of IFN-β-1a, PEGylated IFN-β-1a, or an equal volume of phosphate buffer given as a control. Blood from these mice was obtained via retro-orbital bleeds immediately after injection and 0.25, 1, 4, 24, and 48 h postdosing. At least three mice were bled at each time point. Whole blood was collected into tubes containing anticoagulant, cells were removed, and the resulting plasma frozen until the time of assay. Plasma samples were diluted 1:10 into Dulbecco's modified Eagle's medium, passed through a 0.2-μm syringe filter, and assayed for antiviral activity.
Female Lewis rats (two per each route of administration and formulation) received 16 MU/kg IFN-β-1a or 2.4 MU/kg PEGylated IFN-β-1a intravenously via a jugular catheter, or subcutaneously, intraperitoneally, intratracheally, or orally. Blood samples were obtained from the jugular vein at the specified time points after protein administration: i.v. IFN-β-1a at 0, 0.083, 0.25, 0.5, 1.25, 3, and 5 h; i.v. PEG IFN-β-1a at 0, 0.083, 0.25, 0.5, 1.25, 3, 24, 48, and 72 h; s.c. IFN-β-1a at 0, 0.5, 1, 1.5, 3, 5, 7, and 24 h; s.c. PEG IFN-β-1a at 0, 0.5, 1, 1.5, 4, 7, 24, 48, and 72 h; i.p. IFN-β-1a at 0, 0.083, 0.25, 0.5, 1.25, 3, 5, and 24 h; i.p. PEG IFN-β-1a at 0, 0.083, 0.25, 0.5, 1.25, 3, 24, 48, and 72 h; i.t. IFN-β-1a at 0, 0.5, 1, 1.5, 3, 5, and 7 h; i.t. PEG IFN-β-1a at 0, 0.5, 1, 1.5, 4, 7, 24, 48, and 72 h; p.o. IFN-β-1a at 0, 0.5, 1, 1.5, 3, 5, and 7 h; and p.o. PEG IFN-β-1a at 0, 0.5, 1, 1.5, 4, 7, 24, 48, and 72 h. Plasma samples were analyzed for IFN-β-1a levels using the antiviral assay.
Rhesus monkeys (three per route) received 1 MU/kg IFN-β-1a or 1 MU/kg PEG IFN-β-1a by either s.c. or i.v. administration. The i.v. dose was administered as a slow bolus injection into a cephalic or saphenous vein. The s.c. dose was administered under the skin on the back after shaving the injection site. Blood was collected via the femoral vein at 0, 0.083, 0.25, 0.5, 1, 2, 4, 6, 8, 24, 48, 96, 168, and 336 h postdose, and allowed to clot to obtain serum. Serum was analyzed for levels of functional drug substances using the antiviral cytopathic effect method. Because rhesus monkeys are biologically responsive to human IFN-β-1a, serum neopterin and β2-microglobulin levels were measured as pharmacodynamic markers of activity using commercially available assays (Goldstein et al., 1989). Neopterin, a product of interferon-induced GTP cyclohydrolase, is elevated in concentration in serum postinjection reflecting a downstream response to receptor activation.
Evaluation of PEGylated Murine IFN-β in Mice.
Mouse IFN-β that had been produced in Chinese hamster ovary cells was PEGylated through its N terminus with 20-kDa PEG aldehyde as described for modification of the human IFN-β-1a. Unlike with the human protein, the specific activity of the PEGylated mouse IFN-β was affected by the conjugation and was only 20% of that of the unmodified IFN-β (the specific activity of unmodified murine IFN-β is 185 MU/mg and of PEGylated murine IFN-β is 34 MU/mg). For assessing the pharmacokinetic properties of the murine IFN-β in mice, animals were dosed i.v. or s.c. with 150 ng of either the unmodified (25,000 U) or PEGylated (5,000 U) protein essentially as described above. However, the pharmacokinetic samples for mouse IFN-β were evaluated by enzyme-linked immunosorbent assay.
Data Analysis.
Pharmacokinetic parameters were calculated using WinNonlin version 2.0 software (Scientific Consulting Inc., Apex, NC). The concentration data were analyzed by noncompartmental analysis to obtain pharmacokinetic parameters, which includeCmax (maximum serum concentration),tmax (time to achieve maximum serum concentration), CL (systemic clearance), Vss (total volume of distribution), t1/2 (terminal phase half-life), and bioavailability. Area under the curve (AUC) was calculated using the trapezoidal rule. Concentration values reported as below limits of quantitation were not used in the pharmacokinetic analysis. Percentage of bioavailability was calculated from the following equation: (AUCextravascular/AUCIV) × (DoseIV/Doseextravascular) × 100. Statistical analyses, including arithmetic mean and S.D., were performed using Microsoft Excel version 5.0 software (Microsoft Corp., Redmond, WA).
Results
Biochemical Characterization of PEG IFN-β-1a.
IFN-β-1a was PEGylated at a single site on the N-terminal amine using PEG aldehyde as the reactant. Addition of a single PEG resulted in a shift in the apparent mass of the interferon by SDS-PAGE from 20 kDa (Fig. 1, lane b) to 55 kDa (Fig. 1, lanes c and d). In the PEGylated sample there was no evidence of unmodified IFN-β-1a nor of higher mass forms resulting from the presence of additional PEG groups. The presence of a single PEG was verified by matrix-assisted laser desorption ionization mass spectrometry. By size exclusion chromatography, the PEGylated product eluted as a single sharp peak with an apparent mass of approximately 200 kDa and there was no evidence of aggregated material (data not shown). The specificity of the PEGylation reaction was evaluated by peptide mapping (Fig.2). All of the predicted peptides from the endoproteinase Lys-C digest of IFN-β-1a have been identified by N-terminal sequencing and mass spectrometry, and of these only the peptide that contains the N terminus of interferon (AP8) was altered by the modification as evident by its disappearance from the map. The mapping data therefore indicate that the PEG moiety is specifically attached to this peptide. The data further indicate that the PEG modification is targeted at the N terminus of the protein since only the N-terminal modification would result in the specific loss of this peptide. Additional evidence for this conclusion was obtained by isolating the PEGylated N-terminal peptide from the endoproteinase Lys-C digest, digesting the peptide further with cyanogen bromide and subjecting this sample to matrix-assisted laser desorption ionization postsource decay sequence analysis. Cyanogen bromide digestion of the N-terminal peptide will further cleave this peptide into two fragments, the N-terminal methionine (residue 1) containing the PEG moiety and SYNLLGFLQR (residues 2–11 in the mature IFN-β sequence). Sequence analysis identified the unmodified peptide SYNLLGFLQR, which was the predicted outcome of this treatment.
The PEGylated IFN-β-1a was tested for its ability to inhibit the cytopathic effect of EMC virus on A549 cells. Results from the analysis are shown in Fig. 3. The concentration of IFN-β-1a that protected the cells from viral killing by 50% (50% cytopathic effect) was approximately 11 pg/ml for both the unmodified and PEGylated forms. Thus, targeted PEGylation through the N terminus with 20-kDa PEG aldehyde did not alter the antiviral activity of IFN-β-1a.
Pharmacokinetics of IFN-β-1a and PEG IFN-β-1a in Monkeys, Rats, and Mice.
The effects of PEGylation on the pharmacokinetic properties of human IFN-β-1a were evaluated in monkeys, rats, and mice following i.v. and s.c. administration. Blood plasma curves for unmodified and PEGylated IFN-β-1a in rats and monkeys are shown in Figs. 4 and5, respectively. Total CL,t1/2, and Vss were calculated from the i.v. data and are summarized in Table 1. In all three animal species, the addition of the PEG moiety increased the serum half-life and decreased the clearance of the IFN-β-1a. On average, serum half-life increased by about 5-fold and clearance decreased by about 10-fold. The volume of distribution of the unmodified protein was greater than that of the PEGylated protein, suggesting that the addition of the PEG moiety reduces distribution outside of the vascular compartment. In monkeys, where the unmodified and PEGylated proteins were dosed equally at 1 MU/kg, both theCmax and the AUC were greater for PEG IFN-β-1a than for unmodified IFN-β-1a with either i.v. or s.c. administration (Table 2). In rats, where the unmodified protein was dosed at about 7 times that of the PEGylated protein, the AUC for PEG IFN-β-1a was still greater than for the unmodified protein, while not unexpectedly theCmax for the unmodified IFN-β-1a was greater than that for the PEGylated protein (Table 2). The bioavailability of PEG IFN-β-1a following s.c. administration was 100% in monkeys, and 28% in rats (Table 2). In rats the bioavailability of IFN-β-1a and PEG IFN-β-1a was also evaluated following i.p., i.t., and p.o. administration. Bioavailability following i.p. administration was >90%, whereas absorption from the i.t. route was <5% of the total dose (Fig. 4; Table 2). The serum half-life of PEGylated IFN-β-1a following i.p. and i.t. administration was also longer and consequently larger AUC was observed for the PEGylated product. Circulating IFN-β-1a levels following oral administration were too low to calculate meaningful pharmacokinetic parameters. PEG IFN-β-1a levels immediately following p.o. administration and after 0.25 h were 40 U/ml, but then were below the limits of detection of the assay at all other time points. Unmodified IFN-β-1a levels of 20 U/ml were detected immediately after administration and 10 U/ml at 0.25 h.
Comparative Pharmacodynamics of IFN-β-1a and PEG IFN-β-1a in Monkeys.
Rhesus monkeys are pharmacologically responsive to human IFN-β-1a and therefore provide a system in which the effects of PEGylation on in vivo biological activity can be evaluated. Administration of both IFN-β-1a and PEG IFN-β-1a produced an increase in serum neopterin concentrations. Serum neopterin concentrations prior to protein administration were 0.58 ± 0.21 ng/ml, and peak serum concentrations after either protein was administered were 5.51 ± 1.38 ng/ml. β2-Microglobulin serum concentrations were elevated from 0.38 ± 0.05 μg/ml before treatment to 0.54 ± 0.07 μg/ml after treatment with either protein. For both pharmacodynamic markers, peak induction occurred at about 24 h postdosing. Although the serum half-life of the PEG IFN-β-1a molecule was greater than that of the unmodified protein, and the serum half-life following s.c. administration was greater than following i.v. administration, the pharmacodynamic responses were similar with both molecules and for both routes of administration (Fig.5). Body temperatures in monkeys were elevated by 1.1 ± 0.9°F 4 h after administration of either IFN-β-1a or PEG IFN-β-1a but were then back to normal by 12 h postdose. Administration of PEG IFN-β-1a at a dose of 1 MU/kg administered twice over a 29-day period was not associated with any overt signs of toxicity. Monkeys produced neutralizing antibodies to both human IFN-β-1a and PEG IFN-β-1a administered s.c. (incidence: two of three and three of three, respectively). Following i.v. administration, neutralizing antibodies developed only to PEG IFN-β-1a (two of three), presumably because of its decreased clearance rate compared with the unmodified protein.
Pharmacokinetics of Unmodified and PEGylated Murine IFN-β in Mice.
The effects of PEGylation on the pharmacokinetic properties of murine IFN-β were evaluated in mice following i.v. and s.c. administration. Blood plasma curves for unmodified and PEGylated IFN-β are shown in Fig. 6. Murine IFN-β has an i.v. half-life of 0.18 h and a s.c. half-life of 2 h. PEGylated murine IFN-β has an i.v. half-life of 2.77 h and a s.c. half-life of 4.93 h. The AUC for murine IFN-β increased from 316 for the unmodified protein to 8816 pg × h/ml for the PEGylated form following the s.c. route of administration.
Discussion
We have demonstrated that PEGylation has a marked effect on the pharmacokinetic properties of human IFN-β-1a. When tested in mice, rats, and monkeys following s.c. administration, PEGylation resulted in approximately a 10-fold increase in AUC. The increase in AUC was the result of a decrease in clearance and volume of distribution. While various studies have characterized the pharmacokinetics of PEGylated proteins following i.v. administration, very little is published on its effects for other routes of administration. This study has demonstrated that the effects on s.c. administration are particularly striking. With a single 20-kDa PEG attached, a relatively constant exposure of the PEGylated IFN-β-1a can be produced for 72 h postdosing, signifying a sustained absorption process. A second interesting feature of the s.c. route of administration was the high bioavailability of the PEGylated product in monkeys (∼100%). In rats, the PEGylated IFN-β-1a was 28% bioavailable following s.c. administration. The reason for the differences in s.c. bioavailability across species is not known. In addition to directly comparing i.v. and s.c. administration across these species, three additional routes of administration, i.p., i.t., and p.o., were also tested in rats. Bioavailability of IFN-β-1a and PEG IFN-β-1a following i.p. administration was >90%, whereas from the i.t. and p.o. routes, it was <5% of the total dose. The oral route of administration was evaluated because some studies have shown that oral or nasogastric administration of type I IFNs can influence responses in animal disease models (Nelson et al., 1996; Satoh et al., 1999). If systemic exposure to IFN-β is required for efficacy in animals models of multiple sclerosis then these pharmacokinetic studies do not support the oral route of administration as an alternative to systemic administration.
PEGylation has been successfully applied to many proteins to improve their pharmacological properties (Knauf et al., 1988; Delgado et al., 1992; Clark et al., 1996; Francis et al., 1996; Glue et al., 2000;Koumenis et al., 2000). While the mechanisms leading to these improvements are product specific, in many instances the benefits have resulted from an increase in solubility, reduced renal and immunoclearance, reduced proteolytic susceptibility, and/or steric hindrance to receptor-mediated clearance. Consequently, the benefits that can be obtained from PEGylation vary markedly from product to product and more significantly can vary depending on the site of attachment, the chemistry used for generating the conjugate, and the physical properties of the PEG itself (size, branching, etc.) (Delgado et al., 1997; Francis et al., 1998). In selecting the 20-kDa PEG aldehyde adduct, we first screened a variety of sized PEGs for their effects on pharmacokinetics and activity in the antiviral assay. Lower molecular weight PEG aldehyde IFN-β-1a conjugates were fully active, but failed to produce the desired enhancement in pharmacokinetic properties. Higher molecular weight forms, in contrast, which should further improve the pharmacokinetic properties of the molecule, compromised activity. When an N-terminal 30-kDa PEG aldehyde adduct was tested in the antiviral assay we observed a 6-fold loss in potency, and when an N-terminal 40-kDa PEG aldehyde adduct was tested the protein was inactive (data not shown). Consequently, the 20-kDa form was selected for detailed evaluation in animals.
Human IFN-β-1a is highly immunogenic in monkeys. Because of the potential for reducing immunogenicity following PEGylation, serum samples from monkeys treated with IFN-β-1a and PEGylated IFN-β-1a were analyzed for the presence of anti-human IFN-β-1a antibody on days 14 and 21 following administration. PEGylation at the N terminus did not diminish immunogenicity of the product in monkeys as measured at these time points, rather PEG IFN-β-1a appeared to be somewhat more immunogenic. This observation may be the result of the higher exposure levels to the PEGylated protein. It is not possible to extrapolate from these data to effects in humans because administration of human IFN-β-1a (AVONEX) to humans is associated with a low incidence of immunogenicity and therefore immunogenicity in the rhesus monkey is not predictive of immunogenicity in humans.
This study has clearly shown that the addition of a 20-kDa PEG to human IFN-β-1a increases the serum half-life and systemic exposure of the protein. In monkeys, the biological response to IFN-β-1a administration was evaluated by measuring the induction of neopterin and β2-microglobulin, two markers of IFN-β-receptor activation. Neopterin levels increased by about 9-fold following IFN-β-1a administration compared with the levels seen prior to treatment. Induction of β2-microglobulin was poor in the rhesus monkey and therefore this marker was not considered to be a reliable indicator for biological activity. One surprising finding in this study was that although the serum half-life of IFN-β-1a was greatly extended by PEGylation, the neopterin response was not affected. Both the magnitude and the duration of the neopterin response were similar for the unmodified and the modified proteins, and for both the s.c. and the i.v. routes of administration. The lack of an increase in the magnitude of the response may indicate that maximal neopterin production had already been reached with the unmodified IFN-β-1a. With the primary goal of the study to generate accurate pharmacokinetic measurements, the doses of IFN-β-1a used in the monkey studies were 10- to 20-fold higher than therapeutic doses in humans. If neopterin production were already saturated with the unmodified IFN-β-1a, then the PEGylated form would not be expected to produce an increased response. The reason for the lack of a more prolonged biological response is more difficult to rationalize, but may reflect a down-regulation of the interferon receptor or downstream signaling components due to the high doses of IFN-β-1a that were used. The fact that the neopterin response curves following i.v. and s.c. administration were the same regardless of whether the sample was PEGylated or not supports this hypothesis. Even for unmodified IFN-β-1a we observed a disconnect between the pharmacokinetic profiles and the neopterin response. An alternative explanation for the incongruity in the pharmacokinetic and pharmacodynamic data is that systemic exposure of the PEG IFN-β-1a is not coupled to the neopterin response perhaps because of the reducedVss of the PEG product. If neopterin-producing cells were less accessible to the PEG IFN-β-1a than to the unmodified IFN-β-1a, then this should result in an apparent decrease in its potency. However, the similar shapes of the response curves argues against such a mechanism, since the duration of the response would have been prolonged in the PEG IFN-β-1a-treated animals even if the magnitude of the effect were smaller. More extensive studies are needed to sort through these alternatives. While neopterin and β2-microglobulin are widely used as surrogate markers for type 1 interferon function, it is not known whether their induction has any relevance to duration of therapeutic efficacy in multiple sclerosis or other diseases in which interferons are efficacious. These are questions that still remain to be resolved in clinical testing.
The administration of both IFN-β-1a and PEG IFN-β-1a to monkeys was associated with a slight transient increase in body temperature. This is a known response to IFN-β-1a administration in both monkeys and in humans. Administration of IFN-β-1a and PEG IFN-β-1a was not associated with any overt signs of toxicity.
In conclusion, we have determined that PEGylation has a large effect on the pharmacokinetic properties of IFN-β-1a, which results in an increase in systemic exposure following diverse routes of administration. The improvements in exposure may translate into clinical benefits and therefore represent an attractive opportunity for further development.
Acknowledgments
We thank Kazumi Kobayashi, Xiaoping Hronowski, Wen-Li Chung, Ling Ling Chen, Christine Chutkowski, Robert Arduini, Konrad Miatkowski, Margot Brickelmaier, Zhifang Li, Donna Hess, and the Bioassay group at Biogen for assisting in various studies. Special thanks to Joe Rosa and Susan Goelz for their contributions to the project.
Footnotes
-
Send reprint requests to: R. Blake Pepinsky, Department of Protein Chemistry, Biogen, Inc., 14 Cambridge Ctr., Cambridge, MA 02142. E-mail: Blake_Pepinsky{at}biogen.com
- Abbreviations:
- IFN
- interferon
- PEG
- polyethylene glycol
- PEGylated
- polyethylene glycol-modified
- EMC
- encephalomyocarditis
- XTT
- 2,3-bis[2-methoxy-4-nitro-5-sulfo-phenyl]-2H-tetrazolium-5-carboxyanilide
- i.t.
- intratracheal
- AUC
- area under the curve
- CL
- systemic clearance
- PAGE
- polyacrylamide gel electrophoresis
- Received January 9, 2001.
- Accepted February 18, 2001.
- The American Society for Pharmacology and Experimental Therapeutics