Pharmacokinetic, Biodistribution, and Biophysical Proﬁles of TNF Nanobodies Conjugated to Linear or Branched Poly(ethylene glycol) Yulia Vugmeyster,*,† Cliﬀord A. Entrican,‡ Alison P. Joyce,† Rosemary F. Lawrence-Henderson,† Beth A. Leary,† Christopher S. Mahoney,‡ Himakshi K. Patel,‡ Stephen W. Raso,‡ Stephane H. Olland,∥ Martin Hegen,§ and Xin Xu†,# †
Department of Pharmacokinetics, Dynamics, and Metabolism and ‡Department of Analytical Research & Development, Pﬁzer Inc., Andover, Massachusetts, United States § Department of Immunology and Autoimmunity and ∥Department of Global Biotherapeutics Technologies, Pﬁzer Inc., Cambridge, Massachusetts, United States ABSTRACT: Covalent attachment of poly(ethylene glycol) (PEG) to therapeutic proteins has been used to prolong in vivo exposure of therapeutic proteins. We have examined pharmacokinetic, biodistribution, and biophysical proﬁles of three diﬀerent tumor necrosis factor alpha (TNF) Nanobody−40 kDa PEG conjugates: linear 1 × 40 KDa, branched 2 × 20 kDa, and 4 × 10 kDa conjugates. In accord with earlier reports, the superior PK proﬁle was observed for the branched versus linear PEG conjugates, while all three conjugates had similar potency in a cell-based assay. Our results also indicate that (i) a superior PK proﬁle of branched versus linear PEGs is likely to hold across species, (ii) for a given PEG size, the extent of PEG branching aﬀects the PK proﬁle, and (iii) tissue penetration may diﬀer between linear and branched PEG conjugates in a tissue-speciﬁc manner. Biophysical analysis (Rg/Rh ratio) demonstrated that among the three protein− PEG conjugates the linear PEG conjugate had the most extended time-average conformation and the most exposed surface charges. We hypothesized that these biophysical characteristics of the linear PEG conjugate accounts for relatively less optimal masking of sites involved in elimination of the PEGylated Nanobodies (e.g., intracellular uptake and proteolysis), leading to lower in vivo exposure compared to the branched PEG conjugates. However, additional studies are needed to test this hypothesis.
INTRODUCTION Covalent attachment of poly(ethylene glycol) (PEG) to therapeutic proteins has been used to extend the half-life and prolong in vivo exposure of therapeutic proteins.1 The pharmacokinetic (PK) and pharmacodynamic (PD) proﬁles of a PEGylated protein can be aﬀected by a number of factors, such as the site of PEG attachment, the molecular weight (MW) of the PEG used, the number of PEG molecules attached to a protein, and the stability of the protein−PEG bond (reviewed in refs 1−4). While more comprehensive analyses on the eﬀect of bound PEG conjugate on protein structure remain to be conducted, most of the current data indicate that the secondary protein structure is unchanged upon PEG attachment.5−7 In addition, it appears that the structural properties, such as hydrodynamic volume, of the protein−PEG conjugate in aqueous solutions are dominated by the structural properties of the PEG.8 For both the protein−PEG conjugate and the free PEG, MW of PEG molecule and/or the combined MW of the protein− PEG conjugate have been shown to have a signiﬁcant impact on the PK proﬁle and the major route of elimination.9,10 Several mechanisms are likely to contribute to the eﬀect of PEGylation on PK proﬁles of therapeutic proteins: reduction in renal clearance through increased overall MW, masking of the © 2012 American Chemical Society
proteolytic sites on the therapeutic protein, and reduction of intracellular uptake (and subsequent intracellular proteolysis/ catabolism) by masking recognition sites for speciﬁc and nonspeciﬁc processes (reviewed in refs 2−4). The 125I-labeled PEG molecules with MW below 20 kDa (with no attached protein moiety) have been shown to be cleared primarily through the kidney.9 For 125I-labeled PEG molecules with MW above 20 kDa, the contribution of renal ﬁltration decreases and the contribution of biliary excretion increases, and above a MW of ∼50 kDa, hepatobiliary clearance is the major route of elimination, with increase in 125I-PEG uptake by liver Kupﬀer cells.9,10 Both the linear and the branched PEGs have been used to improve circulating half-lives of therapeutic proteins (reviewed in refs 3,4). Several case studies describe improved exposure (AUC) and circulating half-lives for a branched protein−PEG conjugate compared to a linear PEG conjugate of the same or similar total PEG MW in single dose mouse or rabbit PK studies.11,12 It remains to be established whether the diﬀerences in PK proﬁles for branched versus linear protein−PEG Received: February 9, 2012 Revised: May 29, 2012 Published: June 25, 2012 1452
dx.doi.org/10.1021/bc300066a | Bioconjugate Chem. 2012, 23, 1452−1462
Figure 1. Structures for TNF Nanobody PEG conjugates. The C-terminal cysteine of the TNF Nanobody was conjugated to linear 1 × 40 kDa (A), branched 2 × 20 kDa (B), or branched 4 × 10 kDa (C) methoxy-poly(ethylene glycol) (mPEG)-malemide.
at the dose levels examined and, thus, PK optimization eﬀorts were focused on the PEG moiety.
conjugates is a consistent phenomenon that holds across the diﬀerent types of proteins and diﬀerent PEG sizes/types and degree of branching, as well as across species including primates. In addition, diﬀerences in biodistribution for a linear versus branched protein−PEG conjugate remains to be examined. The possible cause for these diﬀerences in PK proﬁles between the linear and branched PEG conjugates of the same MW is disputed. Proposed explanations include diﬀerences in hydrodynamic volume (for the protein−PEGs conjugates of MW both below and well above the postulated glomerular ﬁltration cutoﬀ); diﬀerences in conformational ﬂexibility leading to diﬀerence in diﬀusion rates through the membrane pore; masking of the protein surface resulting in a reduced proteolysis rate; and diﬀerences in clearance by the immune system.12−14 It is likely that the size/type of PEG conjugate, as well as protein moiety of the conjugate and the site of conjugation, would contribute to a certain set of characteristics that could result in diﬀerent PK proﬁles for branched versus linear PEGs. Likewise, the optimal degree of branching for diﬀerent types/sizes of PEGs remains to be examined and also likely varies for diﬀerent therapeutic proteins. In this study, we compare PK proﬁles of 1 × 40 kDa linear, 2 × 20 kDa branched, and 4 × 10 kDa branched protein−PEG conjugates in mice, rats, and cynomolgus monkeys and the biophysical characteristics of these protein−PEG conjugates. We also assess biodistribution of the linear versus branched PEG conjugates in mice. The protein moiety used for this study is a bivalent antihuman tumor necrosis factor alpha (TNF) Nanobody (protein molar mass of ∼27 kDa) that cross-reacts with cynomolgus monkey but not mouse or rat TNF.15 The TNF Nanobody−PEG conjugates are being considered as possible agents to treat autoimmune disorders, such as rheumatoid arthritis (RA).16 The target contribution to the disposition of these biotherapeutics in normal subjects is likely to be minimal
EXPERIMENTAL PROCEDURES Test Articles. TNF Nanobody, a 27 kDa protein comprising of two anti-human TNF speciﬁc humanized llama VHH domains, was produced at Pﬁzer, Inc. using stable Chinese hamster ovary (CHO) cell lines.15 The engineered C-terminal cysteine of the TNF Nanobody was conjugated to the branched 2 × 20 kDa, branched 4 × 10 kDa, or linear 1 × 40 kDa methoxy-poly(ethylene glycol) (mPEG)-malemide (abbreviated as “PEG”; obtained from NOF American Corporation (White Plains, New York); Figure 1). Speciﬁcally, after protein A aﬃnity chromatography the engineered cysteine of the puriﬁed protein was uncapped with a 10 mM DTT treatment. A buﬀer exchange by G-25 Sephadex gel ﬁltration removed the reducing agent, and the protein was then immediately PEGylated with a 2−10-fold molar excess of maleimide activated PEG for 1 h at 25 °C. The pH of the crude reaction mix was adjusted down below the pI of the protein and the mixture was loaded on a cation exchange column. After developing a linear NaCl gradient and eluting the PEGylated Nanobody, it was further puriﬁed by size exclusion chromatography in PBS. The fractions of over 99% purity were pooled and concentrated via Amicon Stirred Cell (10 kD MWCO, PES). After PEGylation, all materials predominantly consisted of the TNF Nanobody covalently linked to a single 40 kDa PEG. Depending on the preparation, the main peak (representing the PEGylated Nanobody) ranged from 86.8% to 99%, determined by size exclusion HPLC; 83.3−96.8% determined by capillary electrophoresis-sodium dodecyl sulfate; and 77.0−98.3% determined by reverse-phase HPLC with low level of high and low molecular weight species (data not shown). Under physiological conditions (37 °C), no cleavage of PEG moiety from the conjugate was detected. 1453
dx.doi.org/10.1021/bc300066a | Bioconjugate Chem. 2012, 23, 1452−1462
Bioactivity. U937 cells (a human monocyte cell, ATCC, 1.875 × 105 cells/mL in 96-well plate resuspended in the assay media) were incubated for 2 h at 37 °C, 5%CO2 with 1 ng/mL TNFα (R&D Systems) in the presence of a TNF inhibitor (in a dilution series from 9 to 0.3 ng/mL). The assay media used for the bioassay was RPMI (no phenol red) supplemented with 10% fetal bovine serum, 2% penicillin/streptomycin, 2 mM Lglutamine, 25 mM Hepes buﬀer, 1 mM sodium pyruvate, and 0.25% glucose. The apoptotic activity was measured by the Caspase-Glo 3/7 luminescence assay, with modiﬁcations to the manufacturer’s instructions (Promega). This assay measures caspase-3/7 activities which occur early in apoptosis. The luminescent data were ﬁtted with a four-parameter logistic equation (Microsoft Excel software) to estimate IC50, and the bioactivity (IC50 relative to that of the PEGylated reference material) was calculated for each protein−PEG conjugate and unPEGylated protein. Iodination Procedure and Preparation of 125I-Labeled Dosing Solutions. Iodination was performed by the IODOBEADS method according to manufacturer’s instructions (Pierce, Rockford, IL), using ∼0.2 mg of PEGylated proteins and 2 mCi of 125-iodine (Perkin-Elmer; Waltham, MA). A dosing solution was prepared by mixing unlabeled test article, a trace amount of 125I-labeled test article, and the formulation buﬀer. Dosing solutions were characterized by gamma-counting of trichloroacetic acid-precipitable radioactivity and gel electrophoresis, as previously described.17 In rat PK studies, the speciﬁc activity of radiolabeled proteins ranged from 37 to 62 μCi/mg (with