Article pubs.acs.org/molecularpharmaceutics
Contribution of Antibody Hydrodynamic Size to Vitreal Clearance Revealed through Rabbit Studies Using a Species-Matched Fab Whitney Shatz,† Philip E. Hass,† Mary Mathieu,‡ Hok Seon Kim,‡ Kim Leach,§ Michelle Zhou,§ Yongping Crawford,§ Amy Shen,§ Kathryn Wang,# Debby P. Chang,# Mauricio Maia,∥ Susan R. Crowell,⊥ Leslie Dickmann,⊥ Justin M. Scheer,† and Robert F. Kelley*,# †
Departments of Protein Chemistry, ‡Antibody Engineering, §Early Stage Cell Culture, ∥Bioanalytical Assay Services, ⊥Preclinical and Translational Pharmacokinetics, and #Drug Delivery, Genentech Inc., 1 DNA Way, South San Francisco, California 94080, United States S Supporting Information *
ABSTRACT: We have developed a tool Fab fragment of a rabbit monoclonal antibody that is useful for early evaluation in rabbit models of technologies for long acting delivery (LAD) of proteins to the eye. Using this Fab we show that vitreal clearance can be slowed through increased hydrodynamic size. Fab (G10rabFab) and Fab′ (G10rabFab′) fragments of a rabbit monoclonal antibody (G10rabIgG) were expressed in Chinese hamster ovary (CHO) cells and purified using antigen-based affinity chromatography. G10rabFab retains antigen-binding upon thermal stress (37 °C) for 8 weeks in phosphate-buffered saline (PBS) and can be detected in rabbit tissues using an antigen-based ELISA. Hydrodynamic radius, measured using quasi-elastic light scattering (QELS), was increased through site-specific modification of the G10rabFab′ free cysteine with linear methoxy-polyethylene glycol(PEG)-maleimide of 20000 or 40000 molecular weight. Pharmacokinetic studies upon intravitreal dosing in New Zealand white rabbits were conducted on the G10rabFab and PEGylated G10rabFab′. Results of single and multidose pharmacokinetic experiments yield reproducible results and a vitreal halflife for G10rabFab of 3.2 days. Clearance from the eye is slowed through increased hydrodynamic size, with vitreal half-life showing a linear dependence on hydrodynamic radius (RH). A linear dependence of vitreal half-life on RH suggests that molecule diffusivity makes an important contribution to vitreal clearance. A method for prediction of vitreal half-life from RH measurements is proposed. KEYWORDS: long acting delivery, ocular pharmacokinetics, hydrodynamic radius, PEGylation
■
INTRODUCTION The anti-VEGF antibody Fab fragment ranibizumab has proven to be effective in the treatment of the wet form of age-related macular degeneration (AMD),1 retinal vein occlusion (RVO),2 and diabetic macular edema (DME).3 Since the site of action for ranibizumab is the back of the eye at the retina and because of a relatively short residence time of antibody Fab fragments in the eye,4 maximum benefit to the patient is obtained through frequent intravitreal injections of drug product. Increased patient convenience and compliance, as well as a decreased risk of inflammation, could be obtained through less frequent dosing. A variety of technologies to achieve less frequent dosing through long acting delivery (LAD) are being considered.5 Many of these approaches will require preclinical testing in an animal model to assess feasibility for sustained delivery. Currently, the standard paradigm includes preliminary in vivo testing using intravitreal (ITV) dosing in rabbits.6 However, a limitation of studies in rabbits, and as observed for ranibizumab,7 is that human or humanized antibodies can be immunogenic in rabbits. Antibodies against ranibizumab in the © XXXX American Chemical Society
vitreous humor are observed as early as 14 days post-ITV injection and can result in exclusion of animals from pharmacokinetic (PK) analysis.7 This can result in inconclusive results, a problem that could be exacerbated upon extended duration of exposure that would be expected for a LAD system. In order to mitigate the effects of immunogenicity on early stage evaluations of delivery technologies using a rabbit ocular model, we have developed a tool Fab fragment (G10rabFab) derived from a rabbit monoclonal antibody. The parental rabbit monoclonal antibody (G10rabIgG) is specific for a phosphorylated peptide derived from the intracellular domain of the human cMet receptor. Since the antibody binds an intracellular target, it is unlikely to show target-mediated clearance from the vitreous body. Although Fab fragments can sometimes be Special Issue: Ocular Therapeutics: Drug Delivery and Pharmacology Received: April 18, 2016 Revised: May 27, 2016 Accepted: May 31, 2016
A
DOI: 10.1021/acs.molpharmaceut.6b00345 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics generated through proteolytic cleavage of rabbit IgGs,8 we have chosen to develop recombinant methods for production of rabbit antibody Fab fragments. Stability studies and ocular pharmacokinetic studies in rabbits have been performed to assess suitability of G10rabFab for testing long acting delivery technologies. Therapeutic agents administered via intravitreal injection can be eliminated from the vitreous body through anterior and posterior pathways.7,9 The anterior route involves passage to the anterior segment followed by removal through normal aqueous humor outflow. Posterior elimination requires passage across a blood−ocular barrier and into the systemic blood circulation. Not surprisingly, macromolecules show low permeability to the blood−ocular barrier and tend to be cleared through the anterior pathway, whereas the posterior pathway may contribute significantly to the clearance of small molecules.9a,10 Diffusivity of macromolecules within the vitreous chamber could be a rate-determining factor for clearance via the anterior pathway. To test this hypothesis we have performed systematic pharmacokinetic studies using G10rabFab constructs of varied hydrodynamic size.
column volumes (CV) of (1) 25 mM Tris, 150 mM sodium chloride, 5 mM EDTA, 2 mM sodium azide, pH 7.5, 5 CV of (2) +0.1% TX114, and 2 CV of 400 mM potassium phosphate, 5 mM EDTA, 0.2% polysorbate 20, pH 7.0 with 5 CV washes of (1) after each of the above steps. The column was then eluted with 50 mM sodium citrate, 150 mM sodium chloride, pH 3.0, and immediately neutralized to pH 5.0 with a 16% volume of 1 M arginine, 400 mM succinic acid, pH 9.0. This eluted pool was diluted to a conductivity of 2 mS with water for irrigation (WFI) then loaded at room temperature on an SPHP (GE Healthcare) column washed with 10 CV of Solvent A, which consists of 20 mM sodium acetate at pH 5.0, then washed with 20 CV of Solvent A + 0.05% polysorbate 20 and 0.05% triton X-114, followed by a wash to baseline with buffer A, and finally eluted with a 30 CV of Solvent B (Solvent A + 1 M sodium chloride) at 0−30% gradient. The pyro-glu nterminal Fab peak (>90% of total Fab) was pooled, concentrated, and loaded onto an S200 (GE Healthcare) column equilibrated in 200 mM arginine, 137 mM succinic acid, pH 5.0. The eluted monomeric peak was pooled, 30 kDa concentrated to 25 mg/mL, dialyzed into phosphate-buffered saline (PBS), and 0.2 μm filtered. The expressed protein was >98% pure with 1.8% aggregate and 0.011 endotoxin units per milligram protein. Identity was confirmed as the pyro-glu heavy chain n-terminal Fab by mass spectrometry. No carbohydrate additions were detected; the light chain has a consensus sequence N-link site, and the heavy chain has an O-link site (T214) that is partially (9%) occupied on the full length CHO expressed parent antibody. A similar protocol was utilized for purification of G10rabFab′ from conditioned CHO media. PEGylation of G10rabFab′. G10rabFab′ was dialyzed against PBS, pH 7.4, followed by addition of 5 mM EDTA. In order to remove thiol adducts from G10rabFab′ c-terminus, fresh DTT was added at a molar ratio of 1:1.2 (Fab′/DTT), and the sample was allowed to sit at room temperature overnight. Following confirmation of thiol adduct removal using LC−MS, Fab′ through the reactive c-terminal cysteine was conjugated to polytheylene glycol maleimide (PEG-mal) from NOF America Corporation having either a size of 20kD (Sunbright ME-200MA) or 40kD (Sunbright ME-400MA). The PEGs were diluted in water and added to the Fab′ pool at a molar ratio of 1:3 (Fab′/PEG). The reaction was gently rotated overnight and progress monitored by LC−MS. Removal of contaminants was performed by cation exchange using a 5 mL GE Healthcare SP HP column. The column was washed with 5 CVs of 25 mM sodium acetate pH 5.0 then eluted with 1 M NaCl over 30 CVs. Fractions (0.5 mL) were collected and peak fractions were separated by 4−20% TrisGlycine SDS-PAGE to analyze purity and pooled accordingly. Aggregate levels were determined by size exclusion chromatography (SEC) using a GE Superdex 200 10/300 GL run under isocratic conditions using phosphate buffered saline (PBS) pH 7.2, 150 mM NaCl. Finally, for in vivo studies endotoxin levels were measured using a Charles River Endosafe-PTS. Light Scattering Measurements. To characterize PEGylated and unPEGylated G10rabFab′ molecules, molar mass and hydrodynamic radius (RH) were determined by light scattering (LS)12 in tandem with SEC.13 All measurements were done in PBS pH 7.2, 150 mM NaCl at 25 °C using a GE Superdex 200 10/300 GL in-line with two Wyatt Technology detectors: a DAWN HELEOS-II multiangle laser light scattering (MALS) photometer and an Optilab T-rEX differential refractive index (dRI) detector. Diffusion coef-
■
MATERIALS AND METHODS Stable Cell Lines for G10rabFab and G10rabFab′. For transient production of Fab fragment in mammalian cells, oligonucleotide-directed mutagenesis was used to delete the portion of the DNA encoding the Fc and a stop codon was introduced in place of Cys-227. A Fab′ version was constructed by retaining Cys-227 and changing residue 228 to a stop codon. Transient cotransfection of the truncated heavy chain and light chain plasmids in CHO cells was performed as described for full length IgG (Supporting Information) and shown to give acceptable Fab titers of 5−10 mg/L. Stable cell lines were established using the targeted integration (TI) host and process developed by Crawford et al.11 Briefly, genes encoding the G10rabFab and G10rabFab′ were cloned into an expression vector containing two LoxP sites (L3 and 2L) under the control of a CMV promoter, respectively. The expression vector was then electroporated into the CHO K1 TI host (S15) together with another plasmid expressing the Cre recombinase using the BioRad gene pulser Xcell electroporation system. Upon Cre expression, Cre recombinase mediate cassette exchange (RMCE) will result in the stable integration of the G10rabFab or G10rabFab′ at a predetermined genomic location. Cells that have undergone correct RMCE were selected using 2 μM Ganciclovir and 10 μg/mL puromycin double-selection. All stable cells were cultured in serum-free proprietary medium at 37 °C and 5% CO2. Typical expression titers for the Fab and Fab′ at the pool stage were ∼0.15 and
