Article pubs.acs.org/molecularpharmaceutics
Evidence for Metabolic Cleavage of a PEGylated Protein in Vivo Using Multiple Analytical Methodologies Victoria L. Elliott,⊥,† George T. Edge,⊥,† Marie M. Phelan,‡ Lu-Yun Lian,‡ Rob Webster,§ Rory F. Finn,∥ B. Kevin Park,† and Neil R. Kitteringham*,† †
MRC Centre for Drug Safety Science, Department of Pharmacology & Therapeutics, University of Liverpool, Liverpool, United Kingdom ‡ Liverpool NMR Centre for Structural Biology, University of Liverpool, Liverpool, United Kingdom § Pharmacokinetics, Dynamics and Metabolism, Pfizer Global Research and Development, Sandwich, United Kingdom ∥ BioTech, Pfizer Global Research and Development, St. Louis, Missouri, United States S Supporting Information *
ABSTRACT: PEGylation of therapeutic proteins is commonly used to extend half-lives and to reduce immunogenicity. However, reports of antibodies toward PEGylated proteins and of poly(ethylene glycol) (PEG) accumulation suggest that efficacy and safety concerns may arise. To understand the relationship among the pharmacology, immunogenicity, and toxicology of PEGylated proteins, we require knowledge of the disposition and metabolic fate of both the drug and the polymer moieties. The analysis of PEG by standard spectrophotometric or mass spectrometric techniques is problematic. Consequently, we have examined and compared two independent analytical approaches, based on gel electrophoresis and nuclear magnetic resonance (NMR) spectroscopy, to determine the biological fate of a model PEGylated protein, 40KPEGinsulin, within a rat model. Both immunoblotting with an antibody to PEG and NMR analyses (LOD 0.5 μg/mL for both assays) indicated that the PEG moiety remained detectable for several weeks in both serum and urine following intravenous administration of 40KPEG-insulin (4 mg/kg). In contrast, Western blotting with anti-insulin IgG indicated that the terminal halflife of the insulin moiety was far shorter than that of the PEG, providing clear evidence of conjugate cleavage. The application of combined analytical techniques in this way thus allows simultaneous independent monitoring of both protein and polymer elements of a PEGylated molecule. These methodologies also provide direct evidence for cleavage and definition of the chemical species present in biological fluids which may have toxicological consequences due to unconjugated PEG accumulation or immunogenic recognition of the uncoupled protein. KEYWORDS: PEGylation, insulin, pharmacokinetics, NMR, Western blotting
1. INTRODUCTION PEGylation, the conjugation of poly(ethylene glycol) (PEG) with therapeutic proteins, peptides, or other biologicals, has been widely adopted as a method to improve the pharmacological properties of biopharmaceuticals.1−4 In addition to enhancing the pharmacokinetic profile of PEGylated proteins, through the extension of plasma half-lives, there is strong evidence to support a reduction in immunogenicity of biopharmaceuticals due to the shielding effect of the large inert PEG polymer.2 Surprisingly, however, recent reports of the detection of neutralizing antibodies in animals5,6 and patients7,8 treated with some PEGylated biopharmaceuticals indicate that the addition of a PEG moiety does not entirely preclude the interaction between a conjugated protein and the immune system, though the mechanism for this is currently unclear. Although many compounds currently in clinical usage, for example, interferon α-2a (PEGASYS) and granulocyte colony stimulating factor (Neulasta), are PEGylated, there exists a paucity of information on the biological fate of the PEG associated with PEGylated proteins: in particular, whether the © 2012 American Chemical Society
PEG may dissociate from the protein, opening up the potential for polymer accumulation or an immune reaction against the liberated protein. With more stringent regulatory requirements comes the need for better understanding of the disposition, metabolism, and toxicological potential of PEGylated proteins and of the PEG moiety itself. Toxicological studies of PEGylated biologicals are typically performed before clinical studies are commenced in humans and have not revealed any significant PEG-specific toxicity with therapeutic doses.9,10 When carried out on small organic molecules, these toxicology safety studies are usually conducted alongside metabolism studies, which explore the biological fate of the molecule. Such metabolism studies have proved difficult to conduct on PEGylated biologicals as the PEG moiety is difficult to analyze: it has no UV chromophore, there are few Received: Revised: Accepted: Published: 1291
November 18, 2011 February 24, 2012 April 5, 2012 April 5, 2012 dx.doi.org/10.1021/mp200587m | Mol. Pharmaceutics 2012, 9, 1291−1301
Molecular Pharmaceutics
Article
of PEG is relatively straightforward to interpret as the majority of protons along the PEG backbone are in the same coupling environment and hence produce a single peak at a chemical shift of ∼3.65 ppm. NMR has also been utilized to characterize PEGylated proteins19−21 and could prove to be a useful tool in providing a quantitative assay for measuring the metabolism of PEG since it is possible to integrate all PEG moieties. Thus, there is currently no clear consensus on the optimal analytical route to apply to pharmacokinetic studies of PEGylated proteins since no single method provides sufficient information on all of the components of the original conjugate. Consequently, we have applied a combination of NMR and gelbased analytical procedures to monitor the disposition of a model PEGylated protein, 40KPEG-insulin, and assessed the potential of this approach to provide a suitable platform for pharmacokinetic studies of PEGylated proteins at therapeutic concentrations. Through the use of such combined technologies, which allow the monitoring of the metabolic fate of both PEG and protein moieties, we provide clear evidence of conjugate cleavage, providing a potential mechanism for immune recognition of the free protein.
high affinity antibodies, and high molecular weight PEGs are recalcitrant to MS analysis due to their polydispersed nature, without the application of sophisticated gas phase-based charge stripping MS approaches.11 Generally, studies which have sought to determine the metabolic fate of PEG conjugates have utilized radiolabeled compounds. A study by Modi et al.12 indicated that the clearance of 14C-PEG(40K)-interferon α-2a was essentially hepatic, with little evidence that the PEG moiety was severed from the parent protein. Some metabolic products were also revealed to be excreted by the kidney, with radio-labeled material discovered in urine. The use of radio-isotopes however, is costly and not readily available and cannot be applied to large scale clinical studies. The location of the radiolabel within the PEG-conjugate is also crucial. The easiest incorporation of a radiolabel is within the protein; however, if the PEG-conjugate is cleaved metabolically, any information on the disposition of the PEG moiety will be lost. Alternative options are to locate the radiolabel within the PEG, which is costly, or within the linker between the PEG and the protein. A study by Parton et al.13 utilized proton nuclear resonance spectroscopy (1H NMR) to measure the PEG component of certolizumab pegol in tissues, urine and feces. The authors found that PEG did not accumulate in any one specific organ and that PEG was predominately excreted in urine. In addition, sodium dodecyl sulfate−polyacrylamide gel electrophoresis (SDS-PAGE) indicated that the PEG component was completely cleaved from the protein prior to excretion and that the PEG moiety remained intact, that is, no debranching of the PEG. These two studies indicate that the fate of the PEGylated conjugates may vary dependent on the protein, the type of PEG used, and the linkage used between the conjugates. Hence the uncertainty still remains concerning the fate of the PEG component of such compounds and consequently their safety profile. As mentioned previously, the limiting factor in determining the disposition and metabolism of PEG is the measurement of PEG in vivo. SDS-PAGE has been utilized for the characterization of PEGylated proteins, although, due to steric hindrance by PEG, their migration is hindered through the gel. Hence, this gives a higher than anticipated MW for the conjugate when compared to protein MW markers. A traditional Coomassie stain for proteins can be utilized to visualize PEGylated proteins; however, it is unsuitable for PEG itself, as the complex formed is not long-lasting. Alternatively, a unique barium iodide (BaI2) colorimetric stain, which forms a complex with the PEG molecule, has been demonstrated.14,15 This offers alternative measurement possibilities to determine both PEGylated proteins and also the PEG component itself. Although PEG is considered to be immunogenically inert, antibody-based methods have recently been reported for the measurement of PEGylated biologicals, based upon Western blotting and enzyme-linked immunosorbent assay (ELISA).16,17 These studies report antibodies raised against PEG, two of which were raised against the backbone of PEG16 and one against the terminal methoxy group.17 However, in our experience these antibodies are not suitable for detecting nonconjugated PEG (see below), although they are capable of detecting a variety of PEGylated proteins. Nuclear magnetic resonance (NMR) polymer analysis can be used to determine the molecular weight, polymer chain branching, and molecular end-groupings.18 A NMR spectrum
2. MATERIALS AND METHODS 2.1. Materials. All chemicals, unless specified, were purchased from Sigma-Aldrich (Gillingham, Dorset, U.K.). Sinapinic acid matrices were purchased from LaserBioLaboratories (Sophia-Antipolis, France). Antibodies for Western blot analysis were generated in-house or obtained from commercial sources, as follows: rabbit anti-PEG IgG polyclonal was generated in-house (Pfizer, Sandwich, U.K.). Anti-insulin IgG polyclonal antibody and rabbit polyclonal antiguinea-pig IgG-HRP were supplied by Abcam (Cambridge, U.K.) and goat polyclonal antirabbit IgG-HRP by Dako (Cambridge, U.K.). Activated 40 kDa branched PEG (NOF Sunbright GL2400AL3) was purchased from NOF Corporation (Tokyo, Japan). 2.2. Synthesis of 40KPEG-Insulin. To ensure that the bioanalytical methodologies utilized in this study would be of practical value for the analysis of therapeutic PEG-protein conjugates, insulin was PEGylated using a branched 40 kDa PEG (two 20 kDa “arms” plus a short linker region; Figure 1)
Figure 1. Synthesis of 40KPEG-insulin. Insulin was coupled to a branched 40KPEG aldehyde by a two stage reaction involving a Schiff base formation and subsequent reduction with cyanoborohydride. The resulting conjugate was purified by cation exchange.
since this is a conformation frequently utilized to modify biopharmaceuticals: for example, PEG-interferon α2a (pegasys15) and PEG-antiVEGF-aptamer (pegaptanib22). Insulin solution (1 mg mL−1, human recombinant from yeast) was prepared in 0.02 M C2H3O2Na containing 0.20 M NaCl (pH 4.5). The 40 kDa branched PEG was added to 50 mL of the 1292
dx.doi.org/10.1021/mp200587m | Mol. Pharmaceutics 2012, 9, 1291−1301
Molecular Pharmaceutics
Article
samples were left on ice for 10 min and then centrifuged at 3000 rpm for 10 min. The resulting supernatant was removed and dried using vacuum centrifugation to complete dryness. 2.8. 40KPEG-Insulin Recovery Following Dialysis and/ or Deproteination. To correct for any signal loss during sample processing, analyses were conducted to determine the percent recovery over a range of concentrations. 40K PEG-insulin, at the stated concentrations, was spiked into ddH2O, control rat urine, or control rat plasma. Urine samples were dialyzed and deproteinated as described in sections 2.6 and 2.7, while plasma samples were only deproteinated. ddH2O samples were left untreated. Samples were analyzed by SDSPAGE followed by barium iodide staining (section 2.9). The densitometry of the resulting bands was performed with Quantity One software. The ratios of band intensity for unextracted/extracted urine and plasma samples were used as correction factors for subsequent estimates of pharmacokinetic values. 2.9. SDS-PAGE and Gel Staining Procedures. SDSPAGE experiments were carried out using mini-gels (8 or 12% polyacrylamide) on a Hoefer SE250 mini-gel system. Prior to loading on the gel, the deproteinated plasma and urine samples were reconstituted with Laemmli sample buffer (0.1 M TrisHCl, 3% SDS, 15% glycerol, 0.2% phenol blue, 5% mercaptoethanol) and boiled for 10 min. The gels were run at a uniform temperature of 4 °C and a constant current of 80 mA for ∼35 min. After electrophoresis, the gel was stained for PEG as described by Kurfürst.14 First, the gel was soaked in 5% glutaraldehyde solution for 30 min at room temperature for fixation. After which the gel was placed in 20 mL of 0.1 M perchloric acid for 30 min. A portion of 5 mL of a 5% barium chloride solution (Riedel de Haen) was then added, followed by 2 mL of 0.1 M iodine solution. The stained PEG bands appear within a few seconds. After a few minutes in the staining solution the gel was placed into water to briefly destain. 2.10. Western Immunoblotting. Samples were electrophoresed as described above and transferred to nitrocellulose membrane at 250 mA and 300 V. Blots were blocked overnight with 5% skim milk (BioRad) and incubated for 2 h at room temperature with rabbit anti-PEG IgG polyclonal antibody for 2 h. If probing for insulin the blots were incubated for 2 h at room temperature with guinea-pig anti-insulin IgG polyclonal antibody (Abcam). The blots were washed with TST (0.05% Tween 20 in 0.01 M Tris-HCl, pH 8; 0.15 M NaCl) before incubation with corresponding secondary antibodies: goat polyclonal antirabbit IgG-HRP (Dako) or rabbit polyclonal antiguinea-pig IgG-HRP (Abcam) for 1 h at room temperature. Blots were then washed with TST before specific bands were visualized by enhanced chemiluminescence (ECL) detection according to the manufacturer's instructions (Perkin-Elmer). 2.11. NMR. Following dehydration samples were reconstituted in 600 μL of methanol-d4 and centrifuged at 13 300 rpm for 8 min. Supernatants were transferred into 535-PP NMR tubes and the stem insert, containing 12.5 mM nicotinamide (in D2O), placed within the tube. The analysis occurred at 300 K on a Bruker 600 MHz Avance III with CryoProbe. 1H NMR spectra were acquired with a relaxation delay of 3.0 s, 32 scans, and a constant receiver gain of 32. The 90° pulse was calibrated for each sample (8.57−9.77 μs) as was the 1H offset (2919−2924 Hz) to maximize suppression of the residual solvent peak (CD3OH). Following acquisition and Fourier transform all spectra were processed using an exponential window function plus phase correction.
insulin solution at a 2:1 PEG to protein/molar ratio. NaBH3CN (1 M stock in dH2O) was mixed with the insulin/PEG solution to a final concentration of 3 mM. The reaction was left at room temperature in the dark for 20−40 h. The reaction was monitored by size exclusion chromatography (SEC) highperformance liquid chromatography (HPLC) with UV/RI/LS (Shodex KW 803 column in PBS at pH 7.4, 1 mL/min for 20 min). The reaction mixture was then purified by cation exchange (CEX) chromatography. 2.3. Purification of 40KPEG-Insulin. The reaction mixture buffer was exchanged with 20 mM C2H3O2Na, pH 4.0 by dialysis (Spectra, MWCO 5000). Following buffer exchange, the sample was diluted 3-fold to approximately 0.33 mg/mL 40K PEG-insulin. Approximately 1 mg of 40KPEG-insulin in 3 mL was loaded at 0.1 mL/min onto a 5 mL SP Hitrap HP column (GE Healthcare, Amersham) equilibrated in 20 mM C2H3O2Na, pH 4.0. Following loading, the column was washed with three column volumes of starting buffer at 1 mL min−1, and the sample was then eluted with a 0−0.25 M NaCl linear gradient over 30 min with 1 mL fractions collected over this period. The final protein concentration of the purified 40KPEGinsulin was determined by the bicinchoninic acid (BCA) assay (Pierce/Fisher Scientific, U.K.). 2.4. Molecular Weight Determination by MALDI-TOFMS. The final molecular weight of the PEGylated insulin was determined by matrix-assisted laser desorption ionization timeof-flight mass spectrometry (MALDI-TOF-MS) performed on a Voyager DE Pro (Applied Biosystems) in linear positive ion mode. 40KPEG-insulin was diluted in 0.1% TFA and mixed 1:8 (v/v) with sinapinic acid dissolved in 50% ACN/0.05% TFA (v/v) prior to spotting onto the MALDI target plate (bovine serum albumin was used as a molecular weight calibrant). The 40 kDa branched PEG utilized comprised two PEG arms (approximately 21 kDa) linked by a spacer of approximately 2 kDa between the glycerol backbone and the aldehyde reactive group; therefore, the average molecular weight of the PEG is ∼44 kDa. MALDI-TOF-MS confirmed the addition of a single 40K PEG with the molecular weight of the final conjugate determined to be 50.5 kDa. 2.5. Animals. Male Wistar rats were maintained and all experiments undertaken in accordance with criteria outlined in a license granted under the Animals (Scientific Procedures) Act of 1986. Rats were administered 40KPEG-insulin intravenously at a dose of 4 mg/kg (total conjugate concentration) by continuous infusion over 15 min with a dosing volume of 4 mL/kg. Following 40KPEG-insulin administration, animals were housed in metabolism cages for collection of urine/feces for a maximum of 14 days. Blood sampling via venopuncture (tail vein) occurred at days 1, 3, 7, and 14 in one cohort of four rats, and at days 7, 14, 21, and 28 in a second cohort of three animals. Urine and feces were collected over 24 h in the first cohort from 0 to 5 days and then on days 7, 13 and 14; in the second cohort urine and feces were collected over 24 h on days 17, 21, 24 ,and 28. 2.6. Dialysis of Urine Samples. All urine samples were desalted by dialysis (Spectra, MWCO 12 000−14 000) against several volumes of buffer with decreasing salt content: 0.1% saline for 2 h, 0.01% saline overnight, and finally H2O for 2 h. 2.7. Deproteination of Plasma and Urine Samples. Icecold acetone was added to plasma (10 μL) and urine (100−300 μL) samples at a ratio of 4:1 (v/v) for gel electrophoresis analysis. A portion of 1 mL of dialyzed urine was taken for NMR analysis and was mixed 1:1 (v/v) with acetone. All 1293
dx.doi.org/10.1021/mp200587m | Mol. Pharmaceutics 2012, 9, 1291−1301
Molecular Pharmaceutics
Article
2.12. 40KPEG-Insulin Stability in Plasma. 40KPEG-insulin, at different concentrations, was added to control rat plasma and incubated at 37 °C. Aliquots were taken at T = 0, 24, 48, 72, and 168 h. Plasma samples were deproteinated and Westernblotted using anti-PEG and anti-insulin IgG, as described in sections 2.7, 2.9, and 2.10.
show no evidence of nonspecific binding to plasma proteins. Furthermore, they have potential utility in determining whether the conjugate remains intact after disposition, as a much stronger signal is produced when the PEG is coupled to a protein. In our hands, these antibodies were unsuitable for immunoprecipitation methodologies, and although the antibodies have been successfully used for the determination of PEGylated proteins by ELISA methodologies,24 they are not amenable to the determination of nonconjugated 40KPEG in this way. Development of a Quantitative NMR Method for 40KPEG. In addition to assessing the utility of gel-based methods, we also evaluated NMR as an alternative analytical “tool” for PEGylated proteins since this offers the possibility of absolute quantification. 1H NMR is particularly suited to analysis of PEG since it produces a single proton peak at 3.65 ppm due to the fact that all hydrogens are in the same chemical environment. To produce a fully quantitative assay, an internal standard was required. Nicotinamide is a useful compound for use as an internal standard25 as it is nonvolatile and nonlabile and resonates in a different region to peaks from the urine, producing four proton signals between 8.22 ppm and 6.87 ppm (labeled A−D in Figure 2). Theoretically, each of the four
3. RESULTS Determination 40KPEG-Insulin in Vivo. To determine the disposition of 40KPEG-insulin, methods for the measurement of both 40KPEG-insulin and 40KPEG were required. Initial methods focused upon the use of MALDI-TOF-MS, and although 40K PEG-insulin is amenable to this form of detection, 40KPEG itself proved recalcitrant to being determined in this way. It was found that low molecular weight PEGs (
