LINGO-1 (leucine-rich repeat and Ig domain containing NOGO receptor interacting protein-1) is a negative regulator of myelination and repair of damaged axons in the central nervous system (CNS). Blocking LINGO-1 function leads to robust remyelination. The anti-LINGO-1 Li81 antibody is currently being evaluated in clinical trials for multiple sclerosis (MS) and is the first MS therapy that directly targets myelin repair. LINGO-1 is selectively expressed in brain and spinal cord but not in peripheral tissues. Perhaps the greatest concern for Li81 therapy is the limited access of the drug to the CNS. Here, we measured Li81 concentrations in brain, spinal cord, and cerebral spinal fluid in rats after systemic administration and correlated them with dose-efficacy responses in rat lysolecithin and experimental autoimmune encephalomyelitis spinal cord models of remyelination. Remyelination was dose-dependent, and levels of Li81 in spinal cord that promoted myelination correlated well with affinity measurements for the binding of Li81 to LINGO-1. Observed Li81 concentrations in the CNS of 0.1 to 0.4% of blood levels are consistent with values reported for other antibodies. To understand the features of the antibody that affect CNS penetration, we also evaluated the pharmacokinetics of Li81 Fab2, Fab, and poly(ethylene glycol)-modified Fab. The reagents all showed similar CNS exposure despite large differences in their sizes, serum half-lives, and volumes of distribution, and area under the curve (AUC) measurements in the CNS directly correlated with AUC measurements in serum. These studies demonstrate that exposure levels achieved by passive diffusion of the Li81 monoclonal antibody into the CNS are sufficient and lead to robust remyelination.
Multiple sclerosis (MS) is a chronic, debilitating neurological disease of young adults that affects more than 1 million people worldwide. The disease is caused by an autoimmune response to proteins in myelin that leads to the formation of CNS lesions and resulting disabilities. Current MS therapies target the immunological response. We have discovered that LINGO-1 antagonists promote myelin repair, and we are investigating antagonist antibodies as a new paradigm for the treatment of MS that can target the repair of damaged myelin to restore neurological function (this study; Mi et al., 2007, 2009a).
LINGO-1 is a leucine-rich repeat- and Ig domain-containing transmembrane protein that is selectively expressed in the CNS where it negatively regulates oligodendrocyte differentiation leading to axon myelination (Mi et al., 2005, 2007, 2009a). LINGO-1 is highly conserved evolutionarily with human and rat orthologues sharing 99.5% identity (Mi et al., 2004). LINGO-1 expression regulates the timing of CNS-onset myelination during development, and up-regulation of its expression in disease suggests a deleterious role for the endogenous protein (Mi et al., 2004, 2005; Inoue et al., 2007; Fu et al., 2008). Consistent with this hypothesis, blocking LINGO-1 function leads to robust myelination in animal models of demyelination (Mi et al., 2007, 2009a). The biological effects of LINGO-1 antagonists have been validated by using a panel of different reagents after local and systemic administration (Mi et al., 2005, 2007, 2009a; Ji et al., 2006; Pepinsky et al., 2011). Of these, antibodies were preferred because of their potency, selectivity, and long serum half-life.
Antibodies are a common platform for biopharmaceuticals, and with more than 20 antibodies approved and several hundred others in clinical trials they account for approximately 30% of biological proteins in clinical trials (Aires da Silva et al., 2008; Labrijn et al., 2008). Most target diseases with an immunological basis or cancer. The use of antibodies as drugs that target antigens expressed in the CNS is challenging because of poor exposure to the CNS after systemic administration (Shen et al., 2004; Levites et al., 2006; Garg and Balthasar, 2009). Available estimates are that only approximately 0.1% of protein in blood gains access to the CNS through passive diffusion, although published values range from as low as 0.01% to as high as 0.4% (Bergman et al., 1998; Shen et al., 2004; Levites et al., 2006; Garg and Balthasar, 2009; Braen et al., 2010). However, there is a paucity of pharmacokinetic data with most limited to measurements in cerebrospinal fluid (CSF) (Bergman et al., 1998; Braen et al., 2010). Here, we establish pharmacokinetic attributes of anti-LINGO-1 Li81 mAb, Fab2, Fab, and poly(ethylene glycol) (PEG)-Fab in brain, spinal cord, and CSF of rats after systemic administration and use dose-response efficacy studies to correlate exposure levels in spinal cord with activity. The study verifies the low-level exposure of antibodies to the CNS reported by others, but more importantly shows that these levels are sufficient for promoting a biological response. The ability to achieve dose-dependent efficacy that is saturable indicates that systemic administration of Li81 antibody may be a viable approach for targeting diseases of the CNS. These studies validate the use of the anti-LINGO-1 Li81 mAb for the treatment of CNS diseases such as MS that are caused by demyelination.
Materials and Methods
Production of the Li81 mAb, Fab2, Fab, and PEG-Fab.
The Li81 mAb, in a human IgG1 aglycosyl framework, was expressed in Chinese hamster ovary (CHO) cells (Mi et al., 2009b). The mAb was purified from clarified and filtered culture medium on recombinant Protein A Sepharose Fast Flow (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK) at 10 mg of mAb/ml of resin. The column was washed at room temperature with 3 column volumes (CV) of 20 mM NaH2PO4, pH 7.0, 150 mM NaCl (PBS), 5 × 0.5 CV of 25 mM NaH2PO4, pH 5.5, 100 mM NaCl, and 1 CV of 50 mM NaCl with no buffer, and the mAb was eluted from the resin with 20 mM sodium citrate, pH 3.3, and 50 mM NaCl collecting 8 × 0.25 CV fractions. The protein content of the eluted samples was estimated from the absorbance at 280 nm using an extinction coefficient of 1.4 for a 1 mg/ml solution. The sample was neutralized by adding 1/10th volume of 100 mM sodium citrate, pH 4.7, and NaOH to 2.5 mM and filtered through a 0.2-μm membrane. The Protein A eluate was diluted 1:1 with water and loaded by gravity onto a Fractogel EMD SO4 (M) (40–90 μm; Merck, Darmstadt, Germany) cation-exchange chromatography column equilibrated with 10 mM sodium citrate, pH 4.7 at room temperature at 25 mg of mAb/ml of resin. The column was washed with 2 CV of 10 mM Na2HPO4, pH 6.0, 50 mM NaCl, and Li81 eluted with 9 × 0.5 CV fractions of 10 mM Na2HPO4, pH 6.0, and 200 mM NaCl. The absorbance of fractions at 280 nm was monitored and peak fractions were pooled, 0.2-μm filtered, aliquoted, and stored at −70°C.
The Li81 Fab2 fragment was generated from the mAb by digestion with pepsin at an enzyme/protein ratio of 1:500. The mAb in 10 mM sodium acetate, pH 3.6, was incubated at 37°C for 4 h for complete conversion of the mAb to Fab2. The reaction was quenched by adding Tris-HCl to a final concentration of 300 mM, and the sample was dialyzed against 50 mM Tris-HCl, pH 8.9, and 50 mM NaCl. The sample was subjected to anion exchange chromatography on a Fractogel EMD TMAE (M) (40–90 μm; Merck) column equilibrated in 50 mM Tris-HCl, pH 8.9, and 50 mM NaCl. The column was washed with 50 mM Tris-HCl, pH 8.9, and 50 mM NaCl, and Fab2 was eluted with 20 mM Na2HPO4, pH 7.5, and 150 mM NaCl (PBS).
The Li81 Fab fragment was generated from Fab2 by digestion with papain at an enzyme/Fab2 ratio of 1:300. Fab2 in 10 mM Na2HPO4, pH 7.5, 5 mM EDTA, pH 7.5, 20 mM cysteine-HCl, and 20 mM NaOH was incubated with papain at 37°C for 2.5 h and quenched with the thiol protease inhibitor E64. The sample was dialyzed against PBS.
For N-terminal PEGylation, Li81 Fab at 2 mg/ml in 10 mM sodium citrate, pH 6.0 was incubated with 5 mg/ml of 20-kDa methoxy-PEG propionaldehyde and 5 mM sodium cyanoborohydride at room temperature for 24 h. The pH was then adjusted to 4.0. The sample was concentrated to 10 mg of Fab/ml and subjected to cation exchange chromatography at room temperature on a Fractogel EMD SO4 column at 10 mg of Fab/ml of resin. The column was washed with 2.5 CV of 10 mM sodium citrate, pH 4.7, followed by 1 CV of 10 mM sodium citrate, pH 4.7, and 15 mM NaCl. PEG-Fab was eluted with 10 mM sodium citrate, pH 4.7, and 50 mM NaC1, and 0.25 CV fractions were collected. Fractions were analyzed by SDS-PAGE, and peak fractions containing mono-PEGylated Fab were pooled, filtered, aliquoted, and stored at −70°C. Protein concentrations were estimated from absorbance measurements at 280 nm using the theoretical extinction coefficient for Fab because the PEG moiety did not contribute to the absorbance at 280 nm, and they are reported as Fab equivalents.
SDS-Polyacrylamide Gel Electrophoresis.
Samples were subjected to SDS-PAGE on 4 to 20% Tris-glycine gradient gels (Invitrogen, Carlsbad, CA). Nonreduced samples were treated with 5 mM N-ethyl maleimide for 5 min at room temperature, diluted with Laemmli nonreducing sample buffer, and heated at 95°C for 2 min before analysis. Reduced samples were treated with sample buffer containing 2% 2-mercaptoethanol and heated as above.
Characterization by Size Exclusion Chromatography.
Samples (100 μg in 300 μl of column buffer) were subjected to size exclusion chromatography (SEC) at room temperature on a Superdex 200 HR10/30 column (GE Healthcare) using PBS as the mobile phase. The column was run at 0.3 ml/min. The column effluent was monitored by UV detection at 280 nm.
Analysis of Function Using a Competition Enzyme-Linked Immunosorbent Assay.
LINGO-1 was expressed as an Fc-fusion protein containing the human LINGO-1 ectodomain (residues 1–532) linked to the hinge and Fc of human IgG1 (Mi et al., 2004). Costar 96-well easy wash plates (Corning Life Sciences, Lowell, MA) were coated with 5 μg/ml LINGO-1 (50 μl/well), in 50 mM sodium carbonate, pH 9.5, overnight at 4°C. All subsequent steps were carried out at room temperature and included four washes with PBS plus 0.05% Tween 20 between steps. Plates were blocked for 1 h with 1% bovine serum albumin, 0.1% ovalbumin, and 0.1% nonfat dry milk in Hank's balanced salt buffer plus 25 mM HEPES, pH 7.0, incubated for l h with 3-fold serial dilutions of test samples in 0.1% bovine serum albumin, 0.1% ovalbumin, and 0.1% nonfat dry milk in Hank's balanced salt buffer plus 25 mM HEPES, pH 7.0, then incubated for 1 h with donkey anti-human Fab-alkaline phosphatase (Jackson ImmunoResearch Laboratories Inc., West Grove, PA) in sample dilution buffer. Wells were then treated with 10 mg/ml 4-nitrophenylphosphate alkaline phosphatase substrate in 100 mM glycine, pH 10.5, 1 mM MgCl2, and 1 mM ZnCl2. Plates were read at 405 nm using a Molecular Devices (Sunnyvale, CA) plate reader.
Oligodendrocyte Differentiation Assay.
Enriched populations of oligodendrocytes from female Long Evans postnatal day 2 rats were grown in culture. In brief, the forebrain was dissected and placed in Hank's buffered salt solution (Invitrogen). The tissue was cut into 1-mm fragments and incubated at 37°C for 15 min in 0.01% trypsin and 10 μg/ml DNase. Dissociated cells were plated on poly-l-lysine-coated T75 tissue culture flasks and grown at 37°C for 10 days in Dulbecco's modified Eagle's medium containing 20% fetal calf serum (Invitrogen). Oligodendrocyte precursors (A2B5+) were collected by shaking the flask overnight at 200 rpm at 37°C, resulting in a 95% pure population. Cultures were maintained in high-glucose Dulbecco's modified Eagle's medium with fibroblast growth factor/platelet-derived growth factor (PeproTech, Rocky Hill, NJ) (10 ng/ml) for 1 week. For assessing the differentiation of rat A2B5+ progenitor cells into mature myelin basic protein positive (MBP+)-myelinating oligodendrocytes, A2B5+ cells were plated into 24-well culture plates in fibroblast growth factor/platelet-derived growth factor-free growth medium supplemented with N2, treated for 72 h with Li81 antibody, and analyzed for MBP expression by ELISA.
Animal Treatment and Handling.
All animal procedures were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, 1996) and approved by the Biogenidec institutional animal care and use committee.
Li81 mAb, Fab2, Fab, and PEG-Fab Pharmacokinetic Studies.
Jugular vein-cannulated male Sprague Dawley rats used in this study, 8 to 12 weeks old and weighing 250 to 300 g, were from Taconic Farms (Germantown, NY). Dosing solutions were prepared in PBS at 3 mg/ml and administered to animals at 1 ml/kg via intraperitoneal injection (four rats/group). Blood samples (200 μl) were drawn at 1, 6, 24, 48, 96, 168, 216, 264, and 336 h postdose for the mAb and PEG-Fab and at 0.25, 1, 4, 7, 24, 48, and 72 h for Fab2 and Fab. Blood samples were allowed to clot. Serum was isolated and stored at −70°C until all samples were obtained. For assessing CNS uptake, the Li81 mAb, Fab2, Fab, and PEG-Fab were administered by intraperitoneal injection into Brown Norway rats (150 g; three rats per time point for each group) at 30 mg/kg for the mAb and PEG-Fab, 60 mg/kg for Fab2, and 250 mg/kg for Fab. Serum, CSF, spinal cord, and brain were collected at 6, 24, 48, 96, 144, 196, 360, and 720 h postdose. Extracts from brain and spinal cord were prepared by isolating the corresponding tissues from animals that had been perfused for 10 min at 20 ml/min with 1.5 mM KH2PO4, 8 mM NaH2PO4, pH 7.2, 132 mM NaCl, and 2.7 mM KCl. The tissues were weighed, frozen on dry ice, and pulverized using a Covaris (Woburn, MA) Cryoprep apparatus. Samples were then homogenized in lysis buffer (50 mM Tris, pH 7.2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 10 mM MgCl2, and 5% glycerol) and extracted overnight at 4°C with rocking, and particulates were removed by centrifugation. Antibody levels were measured by ELISA. ELISA plates (Nalge Nunc International, Rochester, NY) were coated at 4°C with 10 μg/ml of donkey anti-human IgG-Fc (Jackson ImmunoResearch Laboratories Inc.) for analysis of the Li81 mAb or 10 μg/ml of LINGO-1 for Fab2, Fab, or PEG-Fab (50 μl/well) in PBS, blocked with 250 μl/well of Superblock (Thermo Fisher Scientific, Waltham, MA) at room temperature for 2 h, then incubated with Li81 mAb-, Fab2-, Fab-, and PEG-Fab-containing samples (serum, CSF, brain lysates, or spinal cord lysates) at room temperature for 2 h. The plates were washed four times with PBS plus 0.05% Tween 20 after each step. Bound Li81 signal was detected after incubation with 50 μl/well of alkaline phosphate-conjugated goat anti-human IgG Fab′ (Jackson ImmunoResearch Laboratories Inc.) at room temperature for 1 h, using 100 μl/well of 1-step p-nitrophenyl phosphate substrate (Thermo Fisher Scientific) and reading absorption at 405 nm. Li81 mAb, Fab2, Fab, and PEG-Fab concentrations in test samples were interpolated from appropriate standard curves prepared for each test article in serum, CSF, brain lysate, and spinal cord lysate. Tissue concentrations are reported as amount/volume assuming the densities of brain tissue and spinal cord are 1. Pharmacokinetic parameters were calculated by noncompartmental analysis using WinNonlin Software (WinNonlin professional version 5.0.1; Pharsight, Mountain View, CA).
Rat MOG-Experimental Autoimmune Encephalomyelitis Model.
Nine-week-old Brown Norway rats (150 g) were anesthetized by intraperitoneal injection of ketamine (80 mg/kg; Bionichepharma, Lake Forest, IL) and xylazine (8 mg/kg; Alfasan, Woerden, The Netherlands), followed by intradermal injection at the base of the tail with a 200-μl cocktail solution containing 100 μl of complete Freund's adjuvant (Chondrex Inc., Redmond, WA), 200 μg of heat-inactivated Mycobacterium tuberculosis (strain H 37 RA; Difco, Detroit, MI), and 100 μl of 100 μg of recombinant rat MOG corresponding to the N-terminal sequence of rat MOG (amino acids 1–125) in saline (1:1). Animals developed signs of EAE after 10 to 15 days. The animals were treated by intraperitoneal injection with anti-Li81 mAb or control antibody weekly starting at day 14 after MOG injection and sacrificed on day 29 (24 h after the third injection) for the evaluation of exposure levels in serum, brain, and spinal cord. EAE clinical scores were measured daily to monitor functional recovery after Li81 or isotype control treatment by using the following scoring scale: grade 0.5, distal paresis of the tail; grade 1, complete tail paralysis; grade 1.5, paresis of tail and mild hind leg paresis; grade 2, unilateral severe hind limb paresis; grade 2.5, bilateral severe hind limb paresis; grade 3, complete bilateral hind limb paresis; grade 3.5, complete bilateral hind limb paresis and paresis of one front limb; and grade 4, complete paresis, moribund state, or death. Moribund animals losing more than 20% of their body weight (measured daily) and/or exhibiting a score of 3.5 for more than 48 h were euthanized. The investigator reading EAE scores was blinded as to the treatment. Myelination in EAE lesions was assessed by immunohistochemistry using anti-myelin basic protein antibody from Covance Research Products (Princeton, NJ) and Alexa-labeled secondary antibody (Mi et al., 2009a). Antidrug antibodies were also assessed at the end of each study. There were no detectable antidrug antibodies in any of the treatment groups.
Lysolecithin-Induced Demyelination in Rat Spinal Cord.
Female Sprague Dawley rats (230 g, 9 weeks old) were anesthetized by intraperitoneal injection with a cocktail consisting of ketamine (35 mg/kg), xylazine (6 mg/kg), and acepromazine (1 mg/kg). Buprenorphine (0.05 mg/kg s.c.) was administered once preoperatively and then twice a day for 2 days postoperatively for analgesia. The animals were immobilized in a stereotaxic frame and placed on a heating pad to maintain body temperature during the surgical procedure. To induce demyelination in spinal cord, 1.5 μl of a 1% solution of lysolecithin (LPC) was injected into the dorsal column on day 0. Li81 mAb was administered intraperitoneally on day 2, and Fab2 and PEG-Fab were administered intraperitoneally on days 2, 4, 6, and 8. The animals were sacrificed and perfused with 4% paraformaldehyde and 2.5% glutaraldehyde on day 9, and the region of the spinal cord encompassing the lesion was excised, postfixed with 4% paraformaldehyde, sectioned, and stained with toluidine blue. Myelinated axons within the lesions were quantified by light microscopy. For these studies, pictures (100× magnification) of toluidine blue-stained, 0.1-μm transverse thin sections were taken from five fields of the lesion that were mid-dorsal, midlateral (left and right), midventral, and at the center of the lesion. Myelinated axons from five fields per animal and three animals per group were counted. The investigator reading sections was blinded as to the treatment.
Li81 Product Attributes.
The Li81 antibody was isolated using Fab phage display technology and engineered into a human IgG1 aglycosyl framework. The Li81 mAb was expressed in CHO cells and purified by sequential steps on Protein A Sepharose and Fractogel EMD SO4. Figure 1A shows an analysis of the samples by SDS-PAGE under reducing and nonreducing conditions. Li81 heavy chain and light chain bands at 50 and 25 kDa, respectively, were observed under reducing conditions, and the characteristic 150-kDa tetrameric two heavy chain-two light chain complex was observed under nonreducing conditions. Fab2 and Fab fragments of the IgG1 mAb were generated enzymatically with pepsin and papain, respectively. Fab2 and Fab had apparent masses by SDS-PAGE of 110 and 45 kDa under nonreducing conditions, and both collapsed into 25-kDa bands upon reduction. Li81 PEG-Fab was generated by targeted PEGylation at the N terminus of the Fab with 20-kDa methoxy-PEG propionaldehyde. PEG-Fab had an apparent molecular mass of 75 kDa under nonreducing conditions and dissociated into bands of 60 kDa (PEGylated) and 25 kDa (non-PEGylated) upon reduction.
The Li81 mAb, Fab2, Fab, and PEG-Fab all produced expected SEC elution profiles (Fig. 1B). The mAb, Fab2, Fab, and PEG-Fab eluted as single prominent peaks with molecular masses of 150, 100, 50, and 250 kDa, respectively. The observed size of PEG-Fab is consistent with predictions based on the large hydrodynamic volume of PEG relative to mass (Leong et al., 2001; Pepinsky et al., 2011). The elongated structure of PEG typically leads to approximately a 10-fold difference in mass versus apparent size (Leong et al., 2001; Pepinsky et al., 2011). Levels of aggregate detected in the mAb, Fab2, and PEG-Fab samples by SEC were less than 1%. Residual Fab2 in the Fab preparation detected by SEC (sample 3), accounting for ∼3% of the total protein, was caused by incomplete digestion with papain.
Apparent binding affinities of the Li81 mAb, Fab2, Fab, and PEG-Fab for human LINGO-1 were measured by ELISA. Binding curves are shown in Fig. 2A, and EC50 values for binding are summarized in Table 1. An EC50 value of 0.02 nM was measured for both the Li81 mAb and Fab2. A 2-fold drop in EC50 from 0.02 to 0.04 nM resulted from the conversion of the bivalent sample to a monovalent Fab. PEGylation produced approximately a 4-fold drop in EC50 for binding of Fab for LINGO-1. EC50 values for binding of the samples to rat LINGO-1 (rat and human LINGO-1 share 99.5% identity) were also determined by using a direct binding ELISA and were indistinguishable from the data generated for binding to human LINGO-1 (data not shown).
Functional attributes of the Li81 mAb, Fab2, Fab, and PEG-Fab were also evaluated in an in vitro bioassay that measured the differentiation of A2B5+ progenitor cells into mature MBP+-myelinating oligodendrocytes (Mi et al., 2005). In this assay, primary A2B5+ progenitor cells were cultured for 72 h in the presence or absence of the test samples, and expression of MBP that was measured by ELISA was used as a marker for maturation. Treatment with the Li81 mAb, Fab2, Fab, and PEG-Fab resulted in dose-dependent increases in MBP expression (Fig. 2B). The lowest concentration that promoted myelination was 0.1 μg/ml for all samples, which corresponds to levels of 0.7, 1, 2, and 2 nM for the Li81 mAb, Fab2, Fab, and PEG-Fab, respectively.
Assessing Serum Pharmacokinetics of the Li81 mAb, Fab2, Fab, and PEG-Fab in Rats.
Serum pharmacokinetic properties for the Li81 mAb, Fab2, Fab, and PEG-Fab were evaluated in Sprague Dawley rats after 3 mg/kg i.p. administration (Fig. 3; Table 1). The pharmacokinetic parameters for the Li81 mAb, Fab2, Fab, and PEG-Fab (Table 1) agree with published parameters reported for similar biologics constructs (Koumenis et al., 2000; Leong et al., 2006; Kontermann, 2009; Pepinsky et al., 2011). Cmax levels of 27, 13, 3.7, and 27 μg/ml were observed for the Li81 mAb, Fab2, Fab, and PEG-Fab, respectively. The minimum concentration in serum that led to measurable Li81 levels in the CNS was 3 μg/ml (discussed below). A serum concentration of 3 μg/ml or more was maintained for 336 h for the mAb, but lasted only 72 h for PEG-Fab and 48 h for Fab2 and was observed only at the 6-h time point for Fab. Apparent elimination half-lives (t1/2) of 120, 12, 2.7, and 26 h and area under the serum concentration-AUC values of 3600, 300, 27, and 1300 μg · h/ml were calculated from the data for the mAb, Fab2, Fab, and PEG-Fab. Elimination half-lives were 10- and 40-fold lower for Fab2 and Fab than for the mAb and 10-fold higher for PEG-Fab than for Fab. AUC values were 10- and 130-fold lower for Fab2 and Fab than for the mAb and 50-fold higher for PEG-Fab than for Fab. Vz/F values of 140, 180, 430, and 90 ml/kg for the mAb, Fab2, Fab, and PEG-Fab, respectively, showed a dependence of the size of the reagents on their distribution, where PEG-Fab and the mAb were limited largely to the vasculature, whereas the smaller Fab was more widely distributed.
Assessing Pharmacokinetics in Rat Brain, Spinal Cord, and CSF.
Because animals had to be sacrificed to measure Li81 concentrations in the CNS, a modified protocol was used where each time point was generated in a separate animal. In addition, doses were increased to ensure adequate measurements of test articles in brain, spinal cord, and CSF, and Fab2 and Fab doses were further increased to compensate for the lower Cmax values observed in the Sprague Dawley rat study (Table 1). Serum pharmacokinetic profiles in Brown Norway rats at the higher doses were proportional to the serum pharmacokinetic profiles in Sprague Dawley rats at 3 mg/kg. Table 2 shows mean concentrations of the Li81 mAb, Fab2, Fab, and PEG-Fab in serum, brain, spinal cord, and CSF at each time point. In all instances, there was approximately a 1000-fold difference between serum levels and those in the three CNS compartments, but only small 1- to 10-fold differences between levels found in brain, spinal cord, and CSF. The larger variability in drug level measurements in CSF than in brain or spinal cord that we and others observed is reported to be caused by uneven distribution of compounds in the CSF compartment (Shen et al., 2004).
To visualize the net influx or efflux of compounds in serum, brain, spinal cord, and CSF over time, Li81 mAb, Fab2, Fab, and PEG-Fab concentrations were plotted as a percentage of the maximum value observed for each sample in each compartment (Fig. 4). Levels in brain, spinal cord, and CSF tracked with serum concentrations but were offset in time, taking longer to be absorbed and cleared from these compartments. The mAb showed the greatest shift, followed by PEG-Fab and Fab2, and Fab showed the smallest shift. For example, serum levels for the mAb peaked after 6 h but there was a 24- to 48-h lag to achieve maximum levels in the CNS and a long delay in its elimination, whereas peak concentrations for Fab2 in the CNS were achieved at the earliest time point tested and elimination was also faster. Other pharmacokinetic attributes were more readily apparent from logarithmic plots of the data (Fig. 5, A and C). As evident from the semilog plot of concentration-time curves in Fig. 5A, elimination of the Li81 mAb from brain and spinal cord followed the same exponential disposition seen in serum. This probably occurs because the rapid turnover rate of brain interstitial fluid and CSF (2 h in rats) prevents accumulation in the CNS (Bergman et al., 1998; Cirrito et al., 2003) and, thus to a first approximation, serum half-life determines the half-life in CNS tissues. Fab2, Fab, and PEG-Fab also showed linear relationships when plotted on a logarithmic scale, although the slopes of the lines were steeper because of their higher clearance rates (Fab2 data discussed below).
Pharmacokinetic parameters calculated from the study are summarized in Table 3. Overall, the parameters for each compound in brain, spinal cord, and CSF were remarkably similar. tmax values revealed a rapid influx of the test articles from serum into the three CNS compartments. In fact, tmax values for Fab2 and Fab were achieved at 6 h, the earliest time point tested. Cmax and AUC measurements revealed the small amount of the original dose that reached the CNS. Cmax values for the four product modalities ranged from 119 to 1000 ng/ml in the three CNS compartments versus 300,000 to 1,000,000 ng/ml in serum, and AUC values ranged from 3 to 111 μg · h/ml in the three CNS compartments versus 7100 to 71,000 μg · h/ml in the serum. Elimination half-life t1/2 values in serum and the three CNS compartments were similar for each test article. Serum and mean t1/2 values in the CNS were 80 and 111 h for the mAb, 14 and 23 h for Fab2, 3 and 7 h for Fab, and 19 and 33 h for PEG-Fab. These similarities further demonstrate the contributions of the fast equilibrium between serum and CNS compartments and the elimination of free Li81 from the CNS caused by rapid turnover of brain interstitial fluid and CSF on the t1/2 values in brain, spinal cord, and CSF. AUC and Cmax values both provided measurements that could be used to estimate bioavailability from serum into the CNS, but the limited data set with respect to the number of time points evaluated favors the calculations based on AUC. Perhaps the most striking feature of the data set was that exposure levels for the mAb, Fab2, Fab, and PEG-Fab in brain, spinal cord, and CSF as measured from AUC values ([AUCCNS compartment]/[AUCserum] × 100) was ∼0.1% for all samples despite the differences in their sizes and serum kinetics. We had hypothesized that the larger Vz/F value for Fab would lead to better penetration into the CNS, but it was not borne out by the data. These findings further support the notion that the test articles reached the CNS by passive diffusion. One notable difference in the data sets was that the percentage of serum AUC values for the mAb were 2-fold lower in CSF than in brain and spinal cord, whereas levels for Fab2 and PEG-Fab were 6- to 10-fold higher in CSF than in brain and spinal cord.
Figure 5B shows Li81 mAb brain and spinal cord data plotted as a percentage of serum levels as a function of time. A 2- to 10-fold increase in brain and spinal cord levels occurred in the later versus earlier time points, indicating retention; however, the absolute amount of the dose retained at the later time points was small. This increase is presumably caused in part by target binding to LINGO-1 in the brain and spinal cord. Similar increases were observed for Fab2, Fab, and PEG-Fab, but the levels in the brain and spinal cord fell to below the detection limits after 144 h for Fab2 and PEG-Fab and after 48 h for Fab. Fab2 data (Fig. 5, C and D) are an important comparator for results with the mAb because Fab2 maintains the same bivalent functionality as the mAb but has a shorter half-life. The elimination rate of Fab2 is faster from serum than from brain and spinal cord as evident by the steeper slope of the line (Fig. 5C). Increases in Fab2 levels in spinal cord and brain over time as a percentage of serum levels were similar to findings for the mAb (compare Fig. 5, B and D). Li81 mAb and Fab2 levels in CSF follow the same exponential disposition seen in serum (Fig. 5, A and C), but unlike the results observed with Fab2, Fab, and PEG-Fab, there was no increase in mAb levels in CSF as a percentage of serum concentration over time (Fig. 5B). Together, these studies reveal a direct dependence of CNS exposure after systemic administration on circulating serum levels of the Li81 antibody and that pharmacokinetic parameters in the CNS can be approximated from serum parameters.
Li81 Treatment Promotes Function Recovery in the Rat EAE Model.
MOG-induced EAE is routinely used to study autoimmune-mediated demyelination. In this model, the inflammatory response and demyelination/remyelination occur concurrently (Mi et al., 2007). To evaluate dose-efficacy responses in the MOG-EAE model, the Li81 mAb was administered weekly at 0.3, 3, and 10 mg/kg by intraperitoneal injection starting at the first sign of disease on day 14 (Fig. 6A). Rats treated with 3 and 10 mg/kg Li81 mAb had lower EAE scores with improved hind limb and tail movement versus isotype control-treated animals. Statistically significant improvements (p < 0.05) were measured on days 21 to 28 with average EAE scores of 1.1 and 1.2 for the 3 and 10 mg/kg Li81 treatment groups and 1.5 for the control group. Treatment with 0.3 mg/kg Li81 was not efficacious. Immunohistochemical analysis of lesions from 3 mg/kg control- and Li81-treated rats (Fig. 6C) revealed robust remyelination after Li81 treatment versus extensive demyelination in lesions from control-treated rats.
Li81 mAb concentrations in serum, spinal cord, and brain from the same animals were measured at the end of the study on day 29, 24 h after the final dose. A linear relationship was observed between dose and Li81 mAb levels detected in serum, brain, and spinal cord (Fig. 6B). mAb levels in brain were 0.2% of the serum levels, and mAb levels in spinal cord were 0.4% of the serum levels at all doses, which were similar to the levels detected in healthy rats (Fig. 5). Additional rats were evaluated on day 22 when EAE animals were at their peak disease score. Li81 levels detected in brain and spinal cord on day 22 were indistinguishable from those observed on day 29.
To determine whether functional recovery required short-term or continuous exposure to the Li81 antibody, Li81 Fab2 was tested after dosing for once a week and three times a week. The long half-life of the mAb relative to the duration of the study made it unacceptable for this analysis. Fab2 was used because, along with its shorter half-life, it retains the same bivalent functionality of the mAb. As shown in Fig. 6D, Li81 Fab2 promoted functional recovery (p < 0.05 on days 21–23) when administered three times a week at 6 mg/kg, but was not efficacious after once-a-week dosing at 6 mg/kg. Figure 6E shows a simulation for the pharmacokinetics of Fab2 in serum after dosing at once a week and three times a week. Both treatments led to similar Cmax values, but differed in AUC. The dosing regimen of three times a week approximated the levels observed for the mAb after a single 3 mg/kg treatment. This study suggests that functional recovery requires continuous exposure because similar blood concentrations were achieved with both dosing regimens, but once-a-week dosing led to a transient exposure of the reagent, whereas the three-times-a-week dosing led to continuous exposure throughout the duration of the experiment.
Dose-Efficacy Responses in the Rat Lysolecithin Model.
The LPC model was used to provide a direct and quantitative biochemical readout for remyelination. In the LPC model, the lesions are more readily delineated than in EAE because of local administration of the LPC and there is a temporal separation between demyelination and remyelination events with minimal inflammation allowing for a more accurate assessment of the impact of Li81 treatment on remyelination (Mi et al., 2009a). LPC was stereotactically injected into the dorsal column of adult rats on day 0, which led to a rapid, local demyelination within a few hours of the LPC injection. Li81 mAb at doses of 0.1 0.3, 1, 2, 10, 30, and 100 mg/kg and isotype control at 10 mg/kg were administered by intraperitoneal injection on day 2. The animals were sacrificed on day 9, and the region of the spinal cord encompassing the lesion was excised and sectioned. Myelinated axons in LPC-induced lesions were quantified by light microscopy from toluidine blue-stained thin sections by directly counting the number of myelinated axons (Mi et al., 2007). Figure 7A shows a clear dose response in the extent of remyelination. The lowest concentration that achieved statistical significance was 1 mg/kg, although there was a slight increase in the number of myelinated axons after 0.3 mg/kg treatment. Treatment with 10 mg/kg produced a maximum response. LPC treatment led to ∼90% reduction in the number of myelinated axons at the site of the lesion. With 10 mg/kg Li81 treatment, 70% of the axons in the lesion were remyelinated after 7 days versus only 20% myelinated axons in the isotype control-treated group. Serum trough levels of Li81 mAb for the 0.3, 1, and 2 mg/kg treatment groups, spanning the minimal efficacious dose, were measured on day 9 at the end of in-life phase of the LPC study. Observed concentrations of 0.5, 1.5, and 4.9 μg/ml from the efficacy study agreed well with levels of 0.5, 2.0, and 4.8 μg/ml measured at the corresponding 168-h time point from pharmacokinetic studies in healthy Sprague Dawley rats dosed at 0.3, 1, and 2 mg/kg.
To investigate the relationship between the concentrations of Li81 that promoted remyelination in the LPC model (Fig. 7A) and receptor occupancy caused by LINGO-1 binding (Fig. 7B), the milligram/kilogram doses were converted to estimated levels of Li81 mAb in spinal cord, and the dose-response data were directly compared with LINGO-1 binding data (Fig. 7C). The overlay of the binding and efficacy data sets (Fig. 7C) revealed a remarkable correlation between binding and extent of remyelination both in the shape and dose dependence of the curves. EC10, EC50, and EC90 values for binding of Li81 to LINGO-1-expressing cells occurred at 10 ng/ml (60 pM), 30 ng/ml (200 pM), and 100 ng/ml (600 pM), respectively. Over this range of concentrations, there was a direct dependence of dose on the extent of remyelination in the LPC model. These findings provide clear evidence that remyelination after Li81 treatment is directly linked to binding and consequently the blocking of LINGO-1 function caused by Li81 binding.
The Li81 mAb is a fully human antibody that was engineered into an aglycosyl IgG1 framework for reduced effector function. The Li81 mAb was selected from a large panel of LINGO-1 antagonists based on its potency, selectivity, biochemical and pharmacokinetic attributes, manufacturability, and efficacy profile in animal disease models (Mi et al., 2005, 2007, 2009; Ji et al., 2006; Pepinsky et al., 2010, 2011). The Li81 mAb is the first drug candidate in clinical trials for MS that targets CNS myelin repair. The dose-response efficacy data in the rat MOG-EAE and lysolecithin models verify that efficacious doses in the CNS can be achieved after systemic administration. Lysolecithin lesions were quantified because the local treatment produces small, easily identifiable lesions with well delineated borders that undergo reproducible demyelination and repair events, thus allowing for more reliable measurements. The efficacy response was saturable, and concentration estimates in the CNS from the dose response assuming 0.14% bioavailability correlate well with data for binding of Li81 mAb to LINGO-1-expressing cells obtained by fluorescence-activated cell sorting.
To understand antibody attributes that affect their exposure to CNS, we produced a panel of product modalities including the mAb, Fab2, Fab, and PEG-Fab and evaluated the biochemical attributes and pharmacokinetics in blood, brain, spinal cord, and CSF after systemic intraperitoneal administration. All of the modalities exhibited expected blood pharmacokinetic parameters of half-life, AUC, and volume of distribution that have been observed previously by us and others (Koumenis et al., 2000; Leong et al., 2006; Kontermann, 2009; Pepinsky et al., 2011). The pharmacokinetic parameters for the CNS compartments paralleled serum pharmacokinetic measurements except that only a small fraction of the systemically administered compounds reached the CNS. In all cases, exposure in the CNS as measured from AUC was ∼0.1% of levels in blood despite the differences in size, half-life, and volume of distribution of the mAb, Fab2, Fab, and PEG-Fab. These findings indicate that pharmacokinetics in the CNS after systemic administration are largely governed by the pharmacokinetics in blood. By calculating levels in the CNS as a percentage of the corresponding blood levels as a function of time, there is a modest increase in values in later time points, which could reflect the binding of the reagent to target CNS tissue. This increase is most apparent when concentrations in the blood are low; thus the percentage of the dose reaching the CNS that is accounted for by these values is small. Another difference between the data sets for the mAb, Fab2, and PEG-Fab is that the percentage of serum AUC levels in the CSF are 2-fold lower for the mAb than in brain and spinal cord, whereas levels in the CSF for Fab2 and PEG-Fab are 6- to 10-fold higher in the CSF than in the brain and spinal cord. The differences are exaggerated in Fig. 5, B and D in the plots of CSF levels as a percentage of serum levels where there are no changes in mAb levels over time but a 20-fold change in Fab2 levels. These findings suggest that the Li81 mAb is being retained in the brain and spinal cord by product attributes specific to its Fc region, perhaps through FcRn binding, which leads to the long half-life of mAbs in blood. Alternatively, the values could result from differences in the elimination rates of the mAb, Fab2, and PEG-Fab from CSF. Previously, Bergman et al. (1998) compared the elimination rates of IgG and IgM from CSF and concluded that IgM was eliminated at a rate consistent with bulk flow, but IgG was eliminated at a rate that was two times faster than bulk flow, suggesting a facilitated elimination mechanism.
The greatest challenge in developing neuropharmaceuticals is the inability of drugs to cross the blood-brain barrier (BBB). This network of specialized endothelial cells inhibits transcellular passage by low pinocytotic activity and inhibits paracellular diffusion by an elaborate network of tight junctions (Kniesel and Wolburg, 2000). Most attempts have focused on developing small-molecule drugs that can be optimized to facilitate their transport into the CNS (Hitchcock, 2008). Despite the many successes, many targets have been intractable to small-molecule development. Many successful biopharmaceuticals that affect CNS diseases, such as immune modulators, circumvent this issue through their effects on peripheral targets. The use of anti-Aβ antibodies for the treatment of Alzheimer's disease with potential CNS and peripheral targets is a significant focus of many companies, and several antibodies are in clinical trials or clinical development (Levites et al., 2006). LINGO-1 is a new therapeutic target for CNS repair. Because LINGO-1 is expressed only in the CNS, we can conclude that all efficacy responses result from Li81 antibody that has reached its target in the CNS and that the small percentage of Li81 that reaches the CNS is sufficient to elicit myelin repair. Although there are many strategies currently under evaluation that increase uptake of proteins into the CNS by active, receptor-mediated transport or partial disruption of the BBB, and they represent important discovery opportunities, these approaches have significant technological and/or safety challenges (Bullard et al., 1984; Abulrob et al., 2005; Boado et al., 2010; Neuwelt et al., 2011; Zhou et al., 2011). In some instances, local impairment of the BBB in disease settings such as stroke, Parkinson's disease, Alzheimer's disease, and MS could provide for enhanced delivery, but the extent and duration of the leakage is highly variable (Neuwelt et al., 2011). Previously, Jorgensen et al. (2007) reported a modest increase in the levels of an antibody in the CNS of MBP-induced EAE versus control rats after systemic administration, but similar to our findings, only 0.2% of the injected dose was detected in the brain and spinal cord of the diseased animals. The analysis of the pharmacokinetics of biologics in CNS compartments is difficult and not well documented. The lessons we learn in the clinical development of the Li81 mAb will have significant impact on the directions we and others take in the development of biopharmaceuticals for CNS-specific targets.
Participated in research design: Pepinsky, Ji, Wang, Meier, and Mi.
Conducted experiments: Pepinsky, Shao, Ji, Meng, Walus, Lee, and Mi.
Contributed new reagents or analytic tools: Pepinsky, Walus, Graff, and Garber.
Performed data analysis: Pepinsky, Shao, Ji, Wang, Meng, Walus, Lee, Hu, and Mi.
Wrote or contributed to the writing of the manuscript: Pepinsky and Mi.
We thank Joe Amatucci, Courtney Kirouac, Shelly Martin, and Gisela Chiang for cell line selection and generating conditioned medium; Sheng Gu, Dingyi Wen, Craig Wildes, and Monika Vecchi for analytical biochemistry; and Darren Baker for critical reading of the manuscript.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- multiple sclerosis
- central nervous system
- cerebrospinal fluid
- column volumes
- poly(ethylene glycol)
- polyacrylamide gel electrophoresis
- size exclusion chromatography
- enzyme-linked immunosorbent assay
- experimental autoimmune encephalomyelitis
- myelin basic protein
- myelin oligodendrocyte glycoprotein
- area under the curve
- blood-brain barrier
- monoclonal antibody
- leucine-rich repeat and Ig domain containing NOGO receptor interacting protein-1
- Chinese hamster ovary
- phosphate-buffered saline
- oligodendrocyte progenitor cell.
- Received May 5, 2011.
- Accepted July 29, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics