Investigation of the Role of FcγR and FcRn in mAb Distribution to the

Jul 27, 2012 - ACS eBooks; C&EN Global Enterprise .... To evaluate the role of Fc receptors (FcR) on IgG distribution to the brain, the disposition of...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/molecularpharmaceutics

Investigation of the Role of FcγR and FcRn in mAb Distribution to the Brain Lubna Abuqayyas and Joseph P. Balthasar* Department of Pharmaceutical Sciences, School of Pharmacy and Pharmaceutical Sciences, University at Buffalo, the State University of New York, Buffalo, New York 14260, United States ABSTRACT: To evaluate the role of Fc receptors (FcR) on IgG distribution to the brain, the disposition of 8C2, a murine monoclonal IgG1 antibody, was investigated after intravenous administration in FcRn α-chain knockout mice, FcγRIIb knockout mice, FcγRI/RIII knockout mice, and C57BL/6 control mice. 125I-8C2 was co-administered with 51Cr-labeled red blood cells to allow accurate assessment of residual blood content in brain samples. Blood and brain tissues were harvested from subgroups of three mice at several timepoints up to 10 days, and radioactivity was counted. The blood and brain areas under 8C2 concentration vs time curves (AUCs) were calculated using the linear trapezoidal rule, and the associated standard deviations (SD) were assessed using a modified Bailer method. Concentration data were also analyzed with a semiphysiological population pharmacokinetic model. The brain/blood AUC ratios were comparable across all strains of mice (ratios ± SD): 0.00774 ± 0.000452, 0.00841 ± 0.000535, 0.00636 ± 0.000548, and 0.00917 ± 0.000478 for C57BL/6 control mice, FcγRI/RIII knockouts, FcγRIIb knockouts, and FcRn α-chain knockout mice (p > 0.05). Statistically significant improvement in model fitting of the data was shown with incorporation of a strain-specific parameter for antibody clearance for FcRn knockout mice; however, no significant improvements in model fitting were found for strain effects on any other parameter, including the brain uptake clearance or efflux clearances for 8C2. The predicted 8C2 brain efflux clearance was found to be ∼135-fold faster than the brain uptake clearance, consistent with the observed low ratio of brain−blood exposure. The experimental results and modeling results indicate that, in mice, FcRn and FcγR do not contribute to the “blood−brain barrier” that limits mAb uptake into the brain. KEYWORDS: antibody, blood−brain barrier, FcγR, FcRn, pharmacokinetics, brain to blood ratio

■

the CNS. Prior work has shown that β2 microglubulin (β2m) knockout (KO) mice, which lack expression of functional FcRn, exhibit similar brain to plasma exposure ratios for mAb as observed in control mice (with expression of FcRn).1 However, the interpretation of these data are somewhat complicated by possible confounding influences associated with β2m knockout, as β2m is a subunit of several proteins (in addition to FcRn). Additionally, a thorough investigation of the effects of Fcγ receptors on mAb disposition in the brain has not been reported. It is anticipated that an improved understanding the role of Fc receptors as determinants of IgG exposure in the brain will help to guide efforts to engineer mAbs with improved brain penetration. In this work, in vivo disposition studies have been conducted in control mice and in KO mice lacking expression of FcγRI, FcγRIIb, FcγRIII, and FcRn (i.e., FcRn α-chain knockouts). Pharmacokinetic analyses, using noncompartmental methods

INTRODUCTION Therapeutic monoclonal antibodies (mAbs) and endogenous IgG antibodies show limited uptake into the brain. Central nervous system (CNS) to plasma or serum ratios for mAbs are typically reported to be within the range 0.1−1%.1−5 Several hypotheses have been proposed to explain the poor exposure of mAb in the brain, including inefficient paracellular transport through the brain capillary endothelium, rapid convective clearance of IgG from the brain via relatively rapid turnover of CNS fluids, saturable IgG uptake into the brain, and efficient receptor-mediated efflux of IgG across brain capillaries.6−9 Efflux may be mediated by transport proteins that bind to conserved regions within Fc-domains of IgG antibodies (i.e., Fc receptors).1,10−12 In particular, the Fc receptor of the neonate, FcRn, has received significant attention as a potential mediator of efflux of IgG from the brain.10,11,13 FcRn is highly expressed in the vascular endothelium within the brain,12 and this receptor has been demonstrated to facilitate IgG transport across cell monolayers in culture.14 Additionally, Fc-gamma receptors have been considered as possible contributors to the IgG “outflow pathway” from the CNS.15 To date, few reports have been published to test the hypotheses that Fc-receptors function to limit IgG exposure in © 2012 American Chemical Society

Special Issue: Drug Delivery across the Blood-Brain Barrier Received: Revised: Accepted: Published: 1505

April 17, 2012 July 18, 2012 July 27, 2012 July 27, 2012 dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

2, 6, and 12 h and 1, 2, 4, 7, and 10 days for wild-type and FcγR KO mice and 1, 2, 6, and 12 h and 1, 2, 3, and 4 days for FcRn α-chain KO mice. Blood and brain tissues were harvested. Brain tissues were blotted to remove any excess fluid and weighed. Radioactivity associated with 51Cr and 125I was quantified using a gamma counter. 8C2 Concentration in Blood and Brain. Radioactive counts of collected blood and tissues were corrected for background and for decay. The counts were converted to concentrations using eq 1:

and using a semiphysiological population pharmacokinetic model, were then employed to allow quantitative comparison of brain and blood concentration vs time data obtained from control and KO mice.

■

EXPERIMENTAL SECTION Materials. 8C2, a murine antitopotecan monoclonal IgG1 antibody,16 was used as a model mAb in this work. 8C2 was produced and purified from the culture of hybridoma cells, as previously reported.16 Sodium iodide (Na−125I) and chromium (51Cr) were obtained from Perkin-Elmer Inc. (Waltham, MA). Fresh C57BL/6 murine blood was obtained from Hilltop Lab Animals, Inc. (Scottdale, PA). Animals and Animal Handling. Four different strains of mice were used: (a) B6.129P2-Fcer1gtm1Rav N12, a gamma chain knockout strain, which lacks expression of FcγRI and FcγRIII (FcγRI/RIII KO), (b) B6.129S4-Fcgr2btm1TtK N12, which is a knockout strain lacking expression of FcγRIIb, (c) B6.129X1Fcgrttm1DCR/DCRJ, which is a knockout for the FcRn α-chain, and (d) C57BL/6 “wild-type” mice (as a control reference strain). Of note, the C57BL/6 strain is the genetic background strain for each of the knockout strains used in this work. The wild-type and the FcγR knockout strains were obtained from Taconic (Hudson, NY), and the FcRn α-chain KO mouse model was obtained from the Jackson Laboratory (Bar Harbor, ME). Mice were housed and monitored according to the State University of New York at Buffalo Laboratory Animal Facility regulations. Mice were kept on a 12:12 h light/dark cycle with free access to food and water. All animal studies were conducted according to the Institutional Animal Care and Use Committee of the State University of New York at Buffalo. Radiolabeling of 8C2 with Iodine-125. To enable the determination of 8C2 in blood and brain tissues, 8C2 was radiolabeled with 125I using a modified chloramine-T method, as detailed in prior work.17 The purity of the iodinated IgG was >99%, as assessed via instant thin layer chromatography (Pall Corporation, East Hills, NY). Radiolabeling of RBC with Chromium-51. 51Cr-red blood cells (RBCs) were employed to allow accurate determination of residual blood in excised brain samples. Labeling was performed according to the method recommended by the International Committee for Standardization in Hematology.1,18 Briefly, 1 mL of blood was centrifuged at 150g for 5 min to separate the RBCs. The supernatant was discarded and the RBC pellet was washed three times with isotonic sodium chloride. The cells were then reconstituted with 0.5 mL of normal saline, and 1 mCi of Na2−51CrO4 was added. Following incubation for 45 min at room temperature, with occasional stirring, the cell suspension was centrifuged at 150g for 5 min and supernatant was discarded. Excess free 51Cr was removed by successive washing with saline. Labeled cells were then suspended in saline and radioactivity was measured by gamma counting. Quantitative Assessment of 8C2 Disposition to the Brain. Two days prior to mAb injection, potassium iodide (KI 0.2 g/L) was added to drinking water supplied to mice to block thyroidal uptake of radiolabeled iodine.125I-labeled 8C2 (8 mg/ kg, 400 μCi/kg 125I activity, ∼10 μCi/mouse) and 51Cr-RBC (400 μCi/kg 51Cr activity, ∼10 μCi/mouse) were coadministered via penile vein injection to groups of FcRn αchain KO mice, FcγRIIb KO mice, FcγRI/RIII KO mice, and C57BL/6 control mice. At predetermined time points, subgroups of 3 mice were sacrificed. Time points included 1,

conc (nM) =

CCPM × dose (mg) × 1000 injected CPM × tissue weight (g) × 0.15 (mg/nmol) (1)

where CCPM refers to the decay and background corrected counts per minute. Correction for Residual Blood in Brain Tissues. The residual blood volume in each brain sample was calculated based on the 51Cr-RBC radioactivity in blood and brain tissues. The residual blood content was calculated as the ratio between 51 Cr counts per minute per gram of brain and 51Cr counts per minute per gram of blood and was converted to milliliters of blood in 100 g of tissues (v/w %).1,19 The brain density was assumed to be 1 g/mL.20 Brain concentrations of 8C2 were corrected for residual blood contents by subtracting the amount of mAb in the residual blood, as described previously.1 Data Analysis. Blood and brain areas under the 8C2 concentration vs time curves (AUC0→t‑last) were calculated using the linear trapezoidal rule, and the standard deviations (SD) of AUC values were calculated using a modified Bailer method,21 as implemented within the noncompartmental analysis module in WinNonlin 6.1 (Phoenix, Pharsight Corporation, Palo Alto, CA). Brain to blood 8C2 AUC ratios were estimated, and variance was calculated using the method of error propagation for ratios.22 Brain to blood AUC ratios were compared across the strains using one way analysis of variance (ANOVA) with Bonferroni’s correction for multiple comparisons (Graph Pad Prism 5, Graph Pad, San Diego, CA). Pharmacokinetic Model of 8C2 Disposition in Mice. A semiphysiological pharmacokinetic model was developed to characterize 8C2 disposition in blood and brain tissue for control, FcγRIIb KO, FcγRI/III KO, and FcRn KO mice. A schematic diagram of the model structure is presented in Figure 1.The model consists of a blood (central) compartment, a peripheral distribution compartment, and a compartment representing the brain. 8C2 elimination is assumed to occur from the central compartment (exclusively). Distribution to and from the peripheral distribution compartment and the brain compartment is assumed to proceed via linear kinetics. The data from all strains were fit simultaneously using eqs 2−4. Vc

dcc = CLeff Cc + CLdCt − (CLup + CLd + CLc )Cc dt

(2)

Vb

dC b = −CLeff Cb + CLupCc dt

(3)

Vt

dCt = CLdCc − CLdCt dt

(4)

The blood, brain, and tissue concentrations of 8C2 are Cc, Cb, and Ct, respectively. CLc represents 8C2 elimination 1506

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

distributed with a mean of 0 and variance of σ2prop. The same residual error model was used to describe blood and brain data together. The precision of the parameter estimates was expressed as the standard error of the mean (SEM). Model selection criteria used to discriminate between candidate models included the following: (i) evaluation of individual and population mean parameter estimates and their precision, (ii) graphical examination of standard goodness-of-fit plots and plots of the observed versus individual predicted concentrations, (iii) reduction in both residual variability and interanimal variability, and (iv) comparison of the minimum value of the objective function (MVOF) for nested models. To distinguish two nested models, a 3.84 drop in the MVOF was required. This value corresponds to a p-value of 0.05, for the addition of a single parameter, based on a χ2-distribution. Pharmacokinetic Covariate Analyses. Covariate analyses were performed on parameters with well-estimated interanimal variability. The effects of strain on such parameters were estimated. A simple additive shift was used to describe the relationship between strain and the relevant PK model parameters. Covariates contributing at least a 3.84 unit change in the MVOF (α = 0.05, one degree of freedom) were considered significant. After the initial univariable analyses were completed, the covariate contributing the most significant change in the MVOF was included in the new base covariate model. This process was repeated until there were no further covariates that produced significant changes in the MVOF. The resulting model was considered as the full multivariable model. Following the identification of statistically significant strain effects and formation of the full multivariable model, the interanimal variability and residual variability models were then evaluated as part of model refinement. Following model refinement, the final model was identified. To test for the presence of statistically significant differences in parameters among pairs of strain groups, the data set was reduced to subsets of data from each combination of pairs. Pairwise testing was performed for model parameters significantly influenced by strain. For each of these subsets, the base structural model was fit to data encompassing both strains in the pair. In a separate run for the same pair, a shift in the model parameter was estimated for one strain. The MVOF was reported, and the p-value was calculated. Differences in MVOF of 3.48 units or higher were considered statistically significant at α = 0.05 .The p-value based on a χ2-distribution was reported.

Figure 1. Schematic representation of the semiphysiologic pharmacokinetic model for 8C2 disposition. Antibody distributes to the peripheral and brain tissues via linear kinetics, and all elimination is assumed to occur from the central compartment. CLc represents clearance from the central compartment, CLd is the distribution clearance between the central and peripheral compartments, and CLup and CLeff are 8C2 brain uptake and efflux clearances. Vc, Vb, and Vt are the volumes of distribution of the central, brain, and peripheral tissue compartments. The brain volume was fixed to the physiological brain volume in mice.

clearance, and CLd represents the distribution clearance between the central and peripheral compartments. CLup and CLeff represent clearance values for brain uptake and efflux. Vc, Vb, and Vt are the volumes of distribution for the central, brain, and peripheral tissue compartments. Brain volume was fixed at 17.5 mL/kg, which was the mean weight of collected brain tissues from all strains. Pharmacokinetic Analysis Methods. Model fitting was performed using a population nonlinear mixed effect modeling approach, as implemented in version 7 of NONMEM (ICON Development Solutions Ellicott City, Maryland). The first order conditional estimation method was used for all modeling, and the ADVAN13 subroutine was used to solve differential equations. The structural model parameters, the magnitude of interanimal variability in these parameters, and the magnitude of residual variability were estimated. The need to incorporate interanimal variability was investigated for all model parameters. Interanimal variability was described using an exponential variance model:

Pj = Ppop exp(ηj)

■

RESULTS 8C2 PK Study. 8C2 concentration vs time profiles in brain and blood samples are shown in Figure 2. The mean 8C2 blood and brain AUCs for all strains are summarized in Table 1, and the mean brain to blood AUC ratios are presented in Table 1 and Figure 3. The brain to blood ratio of 0.00774 ± 0.000452 in the wild type mice was similar to the brain to blood ratios in FcγRI/RIII KO (0.00841 ± 0.000535), FcγRIIb KO (0.00636 ± 0.000548), and FcRn α-chain KO mice (0.00917 ± 0.000478). All ratios in knockout strains were within ±20% of the brain to blood ratio in the wild type mice. The low distribution of IgG to the brain of wild type and α-chain FcRn KO mice is consistent with previous findings for IgG in control and β2m KO mice.1 No significant differences in the AUC ratio between each KO strain and the wild type were observed (P > 0.05). The time courses of brain/blood exposure ratios, for each mouse strain, are presented in Figure 4. As shown in the

(5) th

Pj and Ppop represent parameters for the j animal and the typical animal value. ηj is the interanimal variability in the jth animal, with a normal distribution around 0 and variance of ω2. The exponential model assumes a log−normal distribution of the parameters. Residual variability was described using a constant coefficient of variation (proportional) error model: Cij = Ciĵ (1 + εij)

(6)

In this error model, Cij is the measured 8C2 blood or brain concentration at the ith time-point in the jth animal. Ĉ ij is the model predicted 8C2 blood or brain concentration at the ith time-point in the jth animal, and εij is a random variable representing discrepancy between the ith measurement in the jth animal and the predicted value. εij is assumed to be normally 1507

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

Figure 3. 8C2 brain to blood AUC0→t‑last ratios in wild type and FcR KO mice. Vertical bars represent the brain to blood AUC0→t‑last ratios. Mean AUC0→t‑last was calculated using a linear trapezoidal method, and the associated standard deviations were estimated using a modified Bailer method. The standard deviations associated with the AUC ratios were calculated using the propagation of error method for ratios. No significant differences were determined by ANOVA (p > 0.05).

Figure 2. Blood and brain disposition of 8C2 in wild type and FcR KO mice. Blood (solid lines) and brain (dashed lines) concentration vs time profiles for 8C2 in wild type, FcγRI/RIII, FcγRIIb, and FcRn αchain knockout mice. 8C2 was administered intravenously at a dose of 8 mg/kg. Brain concentrations are corrected for residual blood content. Error bars indicate standard deviation associated with each mean concentration (n = 3/time point/strain).

figure, the exposure ratios show a similar time-course for all strains, reaching a plateau approximately 1 day after dosing. The mean (±SD) percentages of residual blood content in the brain tissues (v/v%) were as follows: 0.796 ± 0.188 (n = 27), 0.761 ± 0.243 × 10−1 (n = 27), and 0.885 ± 0.142 × 10−1 (n = 24) for wild type, FcγRIIb, and FcRn α-chain KO mice. The blood content in the brain tissues was consistent with previous findings.1,23 In the FcγRI/RIII KO strain, 51Cr-RBC showed rapid elimination, consistent with use of a damaged preparation of RBC. As such, to correct for residual blood content in this strain, the mean value of residual blood in all of the other strains (0.814%) was used. PK Model Characterization of 8C2 Data. 8C2 blood and brain data were characterized using the base model structure shown in Figure 1. The parameter estimates and associated standard errors are reported in Table 2. Model parameters were estimated with moderate to high precision. A high degree of interanimal variability in systemic clearance was observed with initial fitting of the model (147% CV). Residual variability was less than 15%. A summary of the strain effect tested in the forward selection covariate analysis, the resultant MVOF, changes from the comparator model, and measures of statistical significance are provided in Table 3. Only the effect of strain on systemic clearance was statistically significant (p < 0.00001). Population analysis showed no statistical differences in CLeff across the strains. Model refinement resulted in the successful estimation of interanimal variability on Vc (ωVc) and covariance between ωCL and ωVc. The parameter estimates and associated

Figure 4. Time courses of 8C2 brain to blood AUC0‑t ratios in wild type and FcR KO mice. Vertical bars represent the brain to blood AUC0‑t ratios in wild type, FcγRI/RIII KO, FcγRIIb KO, and FcRn αchain KO mice. AUC0‑t was calculated using the linear trapezoidal method, and the associated standard deviations were estimated using a modified Bailer method. Error bars represent the standard deviations of the AUC0‑t ratios, calculated using the propagation of error method for ratios.

standard errors from the final model are reported in Table 4. All parameters were estimated with moderate to high precision, except for ωCL and ωVc. However, including an effect of strain on CL explained more than 100% of the interanimal variability (19.5 vs 147%) for ωCL. The presence of significant differences in CL between strains was tested through analysis of subsets of data, for each pair of strains. Results from the pairwise tested

Table 1. Blood and Brain AUC0→t‑last and AUC0→t‑last Ratios of 8C2 for Wild Type and FcR-Deficient Micea group WT FcγRI/RIII KO FcγRIIb KO FcRn α-chain KO

blood AUC (nM × day) 2.15 2.36 2.87 6.25

× × × ×

103 103 103 102

± ± ± ±

3.69 6.65 51.3 1.41

× × × ×

brain AUC (nM × day)

101 101 101 101

1.66 1.99 1.83 5.64

× 101 ± 9.27 × 10−1 × 101 ± 1.13 × 101 ± 1.54 ± 2.71 × 10−1

brain/blood AUC ratio 0.00774 0.00841 0.00636 0.00907

± ± ± ±

0.000452 0.000535 0.000548 0.000478

Values are listed as mean ± standard deviation. Standard deviation of the AUC was calculated by a modified Bailer method in WinNonlin; n = 3 was assumed. a

1508

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

Table 2. Parameter Estimates from the Base Population Pharmacokinetic Model Fitting to Pooled Data from All Dose Groups and Strainsa param Vb Vc Vt CLd CLeff CLup CLc interanimal variability ω2CLc ω2CLeff prop resid variability σprop a

L/kg L/kg L/kg L/(day L/(day L/(day L/(day

%SEM

model param: covariate

MVOF

0.0175 0.0973 0.108 0.0879 0.563 0.00893 0.00669

FIXED 4.2 7.6 11.4 46.1 44.8 27.5

468.51 468.55 482.32 484.00 430.15 388.27 434.60 434.06

2.17 (147.30%CV) 0.0456 (21.40%CV)

31.9 33.1

0.0167 (12.7% CV)

32.1

WT + FcγRI/RIII KO WT vs FcγRI/RIII KO WT + FcγRIIb KO WT vs FcγRIIb KO WT + FcRn KO WT vs FcRn KO FcγRI/RIII KO + FcγRIIb KO FcγRI/RIII KO vs FcγRIIb KO FcγRI/RIII KO + FcRn KO FcγRI/RIII KO vs FcRn KO FcγRIIb KO + FcRn KO FcγRIIb KO vs FcRn KO

final param est

unit

kg) kg) kg) kg)

Table 5. Results from Testing Pairwise Differences in CL between the Different Strains

Minimum value of the objective function = 940.83.

a

Table 3. Results from Forward Selection to Identify Significant Covariate Relationships model param: covariate Step 1 base model CLc: strain CLeff: strain Step 2 base model + strain on CLc CLeff: strain

MVOF

change in MVOF

p-value

940.83 878.52 937.92

62.31 2.91

2.12 × 10−12 0.088

878.52 875.66

2.86

Vb Vc Vt CLd CLeff CLup CLc wild-type shift for FcγRI/RIII KO shift for FcγRIIb KO shift for FcRn KO interanimal variability ω2CLc Cov ω2CLc and ω2Vc ω2Vc ω2CLeff prop resid variability σprop a

unit

0.0908

L/kg L/kg L/kg L/(day L/(day L/(day L/(day L/(day L/(day L/(day

estimate

kg) kg) kg) kg) kg) kg) kg)

0.0175 0.0988 0.115 0.0617 0.895 0.00663 0.00828 −0.00319 −0.00815 0.0604

0.0092 (9.59%CV)

36.7

−1.68

NA

41.88

9.72 × 10−11

0.54

0.46

54.25

1.77 × 10−13

88.058

6.36 × 10−21

NA: not applicable due to random but minimal increase in MVOF.

DISCUSSION Antibodies and other macromolecules show poor distribution to the brain.1,2,4,5 This limited exposure of macromolecules in the brain has been primarily explained by low rates of convective transport into the brain, due to the existence of “tight junctions” between endothelial cells in brain capillaries. However, in addition to the inefficient convective uptake of antibodies into the brain, low concentrations of IgG in the brain may be partly explained by (i) rapid catabolism of IgG in the brain, (ii) rapid convective elimination of IgG from the brain, or (iii) active efflux of IgG from brain tissue. The latter mechanism has received substantial attention following reports from Pardridge and co-workers, which showed high-level expression of FcRn in the brain vascular endothelium,10 and also demonstrated an apparent Fc-dependent elimination of IgG from the brain after direct intracranial injection in rats.10,12 However, the significance of Fc-dependent efflux of mAb from the brain remains uncertain, as there has been no convincing demonstration of efflux transport mediated by known Fc receptors. In prior work conducted by our group, we found that the brain to plasma AUC ratio of a murine IgG1 mAb, 7E3, was virtually identical in control mice and in β2m KO mice.1 This work suggests that FcRn is not responsible for limiting brain exposure to IgG, as β2m KO mice are unable to express functional FcRn.15 However, the β2m KO model has some disadvantages, as β2m is expressed in all MHC class I molecules, and knocking out β2m may affect the function of other proteins,25 potentially confounding the interpretation of our prior results. To further evaluate the role of FcRn in the

fixed 4.90 15.7 27.2 21.8 0.0014 29.4 43.6 13.9 9.5 75.9 32.5 72.4 29.8

NAa

■

%SEM

0.0382 (19.5%CV) 0.021 0.0117 (10.8%CV) 0.0598 (24.5%CV)

p-value

−0.04

The final multivariate population model showed no strain effects on the uptake clearance and the efflux clearance to and from the brain. Furthermore, the modeling identified no differences in the volume of distribution in the blood and peripheral compartments or in the distribution clearance to and from the peripheral compartment. Based on the model generated values of 8C2 brain CLup and CLeff, the efflux clearance of 8C2 from the brain was ∼135-fold faster than 8C2 uptake into the brain. The mean volumes of the central and peripheral compartments were similar (98.8 and 115 mL/kg). The estimated blood volume, 98.8 mL/kg, is similar to the reported blood volume of 2.3 ± 0.1 mL in 25−30 g mice.24

Table 4. Parameter Estimates for the Final Pharmacokinetic Modela param

412.14 357.89 468.51 380.45

change in MVOF

Minimum value of the objective function = 826.08.

models are reported in Table 5. There were no statistically significant differences in CL between the wild-type, FcγRI/RIII KO, and FcγRIIb KO mice. However, a highly statistically significant difference in CL was evident when comparing FcRn KO mice to all the other strains. Mean 8C2 clearance was approximately ∼8-fold faster in FcRn KO mice than in WT mice [68.7 vs 8.28 mL/(d kg)]. Goodness-of-fit plots for 8C2 blood and brain data are shown in Figures 5 and 6. 1509

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

Figure 5. Goodness-of-fit plots for 8C2 concentration in blood and brain.S catter plots and regression lines (solid) of 8C2 observed and predicted concentrations in blood and brain. Left panels are individual predictions vs observed values in blood (r2 = 0.975) and brain (r2 = 0.99). Right panels are population predictions vs observed values in blood (r2 = 0.902) and brain (r2 = 0.628). The dotted lines represent the line of identity.

distribution of IgG to the brain and to investigate the possible role played by FcγR, we evaluated the disposition of a model murine IgG1 mAb, 8C2, in FcRn α-chain KO mice, FcγRIIb KO mice,26 and FcγRI/RIII KO mice. Of note, the FcγRI/RIII model is a γ-chain KO that is known to be deficient in expression of FcγRI and FcγRIII, and it may also lack expression of FcγRIV.27,28 8C2 Fab domains exhibit high affinity for topotecan,16 an anticancer drug, but no known affinity for any endogenous substance present in mice. Consequently, the interpretation of our data is simplified by an expected lack of influence of specific binding of 8C2 to target cells or proteins. The blood and the brain pharmacokinetics of 8C2 after acute dosing in control, FcγR KO, and FcRn α-chain KO mice were examined, and brain/blood AUC ratios were calculated with correction for the quantity of residual blood present within the brain tissue samples. The residual blood volumes, which were determined using 51Cr-RBC, are in the range of values reported in the literature.1,23 The measured 8C2 brain/blood exposure ratios were similar in each group of mice, with mean values ranging from 0.0064 to 0.0091. No statistically significant differences were found between values obtained from knockout mice and the exposure ratio found for control mice. Of note, given the variability observed in brain/blood exposure ratios, retrospective analyses indicate that the study design provided a

Figure 6. Comparison of observed and model predicted 8C2 brain and blood pharmacokinetics in mice. Blood and brain data from the wild type, FcγR, and FcRn knockout mice were fitted simultaneously to the final model (parameters presented in Table 4; model structure is shown in Figure 1). The dashed lines represent population predicted 8C2 blood and brain concentrations. Symbols represent the observed 8C2 concentrations in blood and brain in (○) wild type, (□) FcγRI/ RIII KO, (◇) FcγRIIb KO, and (▽) FcRn α-chain KO mice.

1510

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

FcRn, FcγRI, FcγRIIb, and FcγRIII) are not among the main determinants of IgG exposure in the brain. The mechanism of clearance of IgG from the brain has not yet been conclusively identified. Investigations by Bergman and co-workers, using intrathecal administration of murine monoclonal IgM and IgG3 in rats and monkeys, indicated that the elimination of the IgM mAb from the cerebrospinal fluid (CSF) was explained by the bulk flow rate of CSF. The elimination of the IgG3 mAb from the CSF exceeded the rate of CSF flow (by ∼2.5-fold); however, CSF clearance was sensitive to treatments (e.g., furosemide) that decrease the CSF flow rate.34 For the present investigation, where whole brain mAb exposure was evaluated, convective elimination is likely to be a function of the bulk flow rate of brain interstitial fluid (ISF). Based on the reported flow rate for brain ISF in rats [0.15−0.29 μL/(min g) brain14] and on the mean brain weight for mice (17.5 mg/kg), brain ISF flow may be estimated to be ∼0.1 μL/ min. This value and the reported flow rate of cerebrospinal fluid (CSF) in mice (∼0.3 μL/min)35 are well below the estimate for CLeff, 895 mL/(day kg) (∼16.0 μL/min), determined from pharmacokinetic modeling. Fixing CLeff to 0.1 μL/min or to 0.3 μL/min did not allow good fitting to the data (results not shown). As such, the apparent rapid efflux of IgG from brain may not be entirely explained by bulk flow. In the present work, determinants of mAb distribution to the brain were investigated following intravenous mAb administration. Direct administration of antibody into the brain, via intracerebral or intraventrical injection, was not evaluated. Direct injection approaches are widely used and accepted for small molecule drugs; however, direct injection of antibodies into the brain may lead to substantial risk for inaccurate or biased estimates of brain clearance. Direct administration of antibody into the brain space is likely to affect hydrostatic pressure gradients, thus altering a main determinant of the convective transport of mAb. Of note, for small molecules, rates of transport by diffusion are expected to be much greater than rates of transport by convection, and consequently, alterations in pressure gradients and/or other determinants of convective transport may have little impact on measured rates of brain clearance for small molecules. The approach that we have taken, with intravenous mAb administration and assessment of brain and blood drug exposure, is less likely to lead to confounding alterations in determinants of convective transport, and this general approach has been well established for investigation of the role of transporters on drug distribution to the brain (e.g., investigations of brain distribution in pglycoprotein KO mice).36 It is important to stress that our work has employed KO mouse models, and we did not investigate the possible influence of compensatory mechanisms. Additionally, our results may be specific to mice. Nonetheless, the present results clearly do not support the hypothesis that low penetration of IgG mAb into the brain is resultant from efficient efflux mediated by FcRn, FcγRI, FcγRIIb, or FcγRIII.

statistical power of 0.8 to detect a 20.2% difference from the mean exposure ratio found in the control group. The observed brain/blood AUC ratios were greater than the brain to plasma AUC ratio (∼0.0022) found in our prior work with 7E3 (which, like 8C2, is a murine IgG1 mAb).1 Adjusting for the hematocrit expected for mice (∼55%) explains much of this difference (e.g., based on a hematocrit of 55%, the mean 8C2 brain/plasma AUC ratios may be estimated to be ∼0.0029−0.0041). The remaining differences in the exposure ratios are likely related to differences between the two IgG molecules (e.g., mAb to mAb differences in surface charge or glycosylation) that may contribute, in part, to the nonspecific uptake of IgG to the brain.29 In addition to the noncompartmental analysis of the data, which revealed no significant differences in brain to blood exposure ratios between the control and KO strains of mice investigated in this study, the blood and brain concentration vs time data were also analyzed with a semiphysiological pharmacokinetic model. The model structure included parameters that describe the uptake and efflux of 8C2 to and from the brain (Figure 1). To characterize the collected data from different strains of mice, a population analysis approach was conducted. Initial modeling was conducted with the most simple form of the model, where all mice were assumed to belong to the same population. In this initial analysis, extremely high interanimal variability (>140% CV) was estimated for the mAb elimination clearance, CLc. This finding was not unexpected, as FcRn KO animals are well-known to exhibit much more rapid IgG elimination relative to that observed in FcRn-competent mice. At the end of covariate analysis, only the effect of strain on systemic clearance in FcRn KO mice was found to be statistically significant. Most of the interanimal variability in the clearance (>100% CV) was explained by the effect of strain. The pairwise comparison of systemic clearance between strains showed that only FcRn knockout clearance was significantly different from each of the other strains (p < 0.000001). This is in agreement with the observed data, where the separation in the elimination phase is only clear for FcRn knockout mice, when compared to other strains. Interestingly, the population estimate for clearance in the FcγRIIb knockout strain was extremely low. This estimate is in agreement with the shallow terminal disposition phase observed for FcγRIIb knockout mice, as compared to wild type and FcγRI/RIII knockout mice. However, given the long terminal half-life of 8C2 in this work, incorporation of a more protracted sampling scheme may be needed to allow confident conclusions regarding a possible slower rate of mAb clearance in FcγRIIb KO mice. Overall, our blood concentration vs time data were similar to data collected in prior studies, and model predicted clearance values were in the range of values reported for murine IgG clearance in control and FcRn α-chain knockout mice.17,30−33 The model predicted similar rates of uptake and efflux clearance for all mouse strains, and variability in CLeff was not explained by the addition of a strain effect. The estimated high efflux clearance of IgG from brain [895 mL/(day kg)] is suggestive of an efficient elimination pathway for IgG from the brain (e.g., via convective transport or metabolic elimination), and our findings are in agreement with literature reports of rapid efflux of IgG from the brain of wild-type mice.3,10−12,34 The inability to identify strain differences in brain distribution parameters suggests that the FcR evaluated in this work (i.e.,

■

AUTHOR INFORMATION

Corresponding Author

*Address: 452 Kapoor Hall, Department of Pharmaceutical Sciences, School of Pharmacy and Pharmaceutical Sciences, University at Buffalo, the State University of New York, Buffalo, New York 14260, USA. Phone: (716) 645-4807. Fax: (716) 645-369. E-mail: jb@buffalo.edu. 1511

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

Notes

(12) Schlachetzki, F.; Zhu, C.; Pardridge, W. M. Expression of the neonatal Fc receptor (FcRn) at the blood-brain barrier. J. Neurochem. 2002, 81, 203−206. (13) Boado, R. J.; Zhang, Y.; Xia, C. F.; Pardridge, W. M. Fusion antibody for Alzheimer’s disease with bidirectional transport across the blood-brain barrier and abeta fibril disaggregation. Bioconjugate Chem. 2007, 18, 447−455. (14) Abbott, N. J. Evidence for bulk flow of brain interstitial fluid: significance for physiology and pathology. Neurochem. Int. 2004, 45, 545−552. (15) Siegelman, J.; Fleit, H. B.; Peress, N. S. Characterization of immunoglobulin G-Fc receptor activity in the outflow system of the cerebrospinal fluid. Cell Tissue Res. 1987, 248, 599−605. (16) Chen, J.; Balthasar, J. P. Development and characterization of high affinity anti-topotecan IgG and Fab fragments. In Handbook of pharmaceutical biotechnology; John Wiley and Sons: Hoboken, NJ, 2007; pp 835−850. (17) Garg, A.; Balthasar, J. P. Physiologically-based pharmacokinetic (PBPK) model to predict IgG tissue kinetics in wild-type and FcRnknockout mice. J. Pharmacokinet. Pharmacodyn. 2007, 34, 687−709. (18) Standard techniques for the measurement of red-cell and plasma volume. A report by the International Committee for Standardization in Hematology (ICSH): Panel on Diagnostic Applications of Radioisotopes in Haematology. Br. J. Hamaetol. 1973, 25, 801−814. (19) Bernareggi, A.; Rowland, M. Physiologic modeling of cyclosporin kinetics in rat and man. J. Pharmacokinet. Biopharm. 1991, 19, 21−50. (20) Brown, R. P.; Delp, M. D.; Lindstedt, S. L.; Rhomberg, L. R.; Beliles, R. P. Physiological parameter values for physiologically based pharmacokinetic models. Toxicol. Ind. Health 1997, 13, 407−484. (21) Nedelman, J. R.; Gibiansky, E.; Lau, D. T. Applying Bailer’s method for AUC confidence intervals to sparse sampling. Pharm. Res. 1995, 12, 124−128. (22) Motulsky, H. Intuitive biostatistics, 1st ed.; Oxford University Press: New York, 1995. (23) Boswell, C. A.; Ferl, G. Z.; Mundo, E. E.; Bumbaca, D.; Schweiger, M. G.; Theil, F. P.; Fielder, P. J.; Khawli, L. A. Effects of anti-VEGF on predicted antibody biodistribution: roles of vascular volume, interstitial volume, and blood flow. PLoS One 2011, 6, e17874. (24) Barbee, R. W.; Perry, B. D.; Re, R. N.; Murgo, J. P. Microsphere and dilution techniques for the determination of blood flows and volumes in conscious mice. Am. J. Physiol. 1992, 263, R728−733. (25) Simister, N. E.; Mostov, K. E. An Fc receptor structurally related to MHC class I antigens. Nature 1989, 337, 184−187. (26) Takai, T.; Ono, M.; Hikida, M.; Ohmori, H.; Ravetch, J. V. Augmented humoral and anaphylactic responses in Fc gamma RIIdeficient mice. Nature 1996, 379, 346−349. (27) Takai, T.; Li, M.; Sylvestre, D.; Clynes, R.; Ravetch, J. V. FcR gamma chain deletion results in pleiotrophic effector cell defects. Cell 1994, 76, 519−529. (28) Takai, T. Multiple loss of effector cell functions in FcR gammadeficient mice. Int. Rev. Immunol. 1996, 13, 369−381. (29) Herve, F.; Ghinea, N.; Scherrmann, J. M. CNS delivery via adsorptive transcytosis. AAPS J. 2008, 10, 455−472. (30) Urva, S. R.; Yang, V. C.; Balthasar, J. P. Physiologically based pharmacokinetic model for T84.66: a monoclonal anti-CEA antibody. J. Pharm. Sci. 2010, 99, 1582−1600. (31) Hansen, R. J.; Balthasar, J. P. Pharmacokinetics, pharmacodynamics, and platelet binding of an anti-glycoprotein IIb/IIIa monoclonal antibody (7E3) in the rat: a quantitative rat model of immune thrombocytopenic purpura. J. Pharmacol. Exp. Ther. 2001, 298, 165−171. (32) Junghans, R. P.; Anderson, C. L. The protection receptor for IgG catabolism is the beta2-microglobulin-containing neonatal intestinal transport receptor. Proc. Natl. Acad. Sci. U. S. A. 1996, 93, 5512−5516. (33) Akilesh, S.; Petkova, S.; Sproule, T. J.; Shaffer, D. J.; Christianson, G. J.; Roopenian, D. The MHC class I-like Fc receptor

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to thank Dr. Murad Melhem for his contribution to modeling discussion and population analysis. This work was supported by funding from Center for Protein Therapeutics, University at Buffalo, the State University of New York.

■

ABBREVIATIONS IgG, immunoglobulin G; PK, pharmacokinetics; mAbs, monoclonal antibodies; FcR, Fc receptors; FcRn, neonatal Fc receptor; FcγR, Fc receptors for IgG; AUC, area under the concentration vs time profiles; Vss, volume of distribution at steady state; NCA, noncompartmental pharmacokinetic analysis

■

REFERENCES

(1) Garg, A.; Balthasar, J. P. Investigation of the influence of FcRn on the distribution of IgG to the brain. AAPS J. 2009, 11, 553−557. (2) Bard, F.; Cannon, C.; Barbour, R.; Burke, R. L.; Games, D.; Grajeda, H.; Guido, T.; Hu, K.; Huang, J.; Johnson-Wood, K.; Khan, K.; Kholodenko, D.; Lee, M.; Lieberburg, I.; Motter, R.; Nguyen, M.; Soriano, F.; Vasquez, N.; Weiss, K.; Welch, B.; Seubert, P.; Schenk, D.; Yednock, T. Peripherally administered antibodies against amyloid beta-peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease. Nat. Med. 2000, 6, 916−919. (3) Rubenstein, J. L.; Combs, D.; Rosenberg, J.; Levy, A.; McDermott, M.; Damon, L.; Ignoffo, R.; Aldape, K.; Shen, A.; Lee, D.; Grillo-Lopez, A.; Shuman, M. A. Rituximab therapy for CNS lymphomas: targeting the leptomeningeal compartment. Blood 2003, 101, 466−468. (4) Kaschka, W. P.; Theilkaes, L.; Eickhoff, K.; Skvaril, F. Disproportionate elevation of the immunoglobulin G1 concentration in cerebrospinal fluids of patients with multiple sclerosis. Infect. Immun. 1979, 26, 933−941. (5) Rubenstein, J. L.; Fridlyand, J.; Abrey, L.; Shen, A.; Karch, J.; Wang, E.; Issa, S.; Damon, L.; Prados, M.; McDermott, M.; O’Brien, J.; Haqq, C.; Shuman, M. Phase I study of intraventricular administration of rituximab in patients with recurrent CNS and intraocular lymphoma. J. Clin. Oncol. 2007, 25, 1350−1356. (6) Friden, P. M.; Walus, L. R.; Musso, G. F.; Taylor, M. A.; Malfroy, B.; Starzyk, R. M. Anti-transferrin receptor antibody and antibody-drug conjugates cross the blood-brain barrier. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 4771−4775. (7) Pardridge, W. M.; Buciak, J. L.; Friden, P. M. Selective transport of an anti-transferrin receptor antibody through the blood-brain barrier in vivo. J. Pharmacol. Exp. Ther. 1991, 259, 66−70. (8) Kumagai, A. K.; Eisenberg, J. B.; Pardridge, W. M. Absorptivemediated endocytosis of cationized albumin and a beta-endorphincationized albumin chimeric peptide by isolated brain capillaries. Model system of blood-brain barrier transport. J. Biol. Chem. 1987, 262, 15214−15219. (9) Zlokovic, B. V.; Skundric, D. S.; Segal, M. B.; Lipovac, M. N.; Mackic, J. B.; Davson, H. A saturable mechanism for transport of immunoglobulin G across the blood-brain barrier of the guinea pig. Exp. Neurol. 1990, 107, 263−270. (10) Zhang, Y.; Pardridge, W. M. Mediated efflux of IgG molecules from brain to blood across the blood-brain barrier. J. Neuroimmunol. 2001, 114, 168−172. (11) Deane, R.; Sagare, A.; Hamm, K.; Parisi, M.; LaRue, B.; Guo, H.; Wu, Z.; Holtzman, D. M.; Zlokovic, B. V. IgG-assisted age-dependent clearance of Alzheimer’s amyloid beta peptide by the blood-brain barrier neonatal Fc receptor. J. Neurosci. 2005, 25, 11495−11503. 1512

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513

Molecular Pharmaceutics

Article

promotes humorally mediated autoimmune disease. J. Clin. Invest. 2004, 113, 1328−1333. (34) Bergman, I.; Burckart, G. J.; Pohl, C. R.; Venkataramanan, R.; Barmada, M. A.; Griffin, J. A.; Cheung, N. K. Pharmacokinetics of IgG and IgM anti-ganglioside antibodies in rats and monkeys after intrathecal administration. J. Pharmacol. Exp. Ther. 1998, 284, 111− 115. (35) Rudick, R. A.; Zirretta, D. K.; Herndon, R. M. Clearance of albumin from mouse subarachnoid space: a measure of CSF bulk flow. J. Neurosci. Methods 1982, 6, 253−259. (36) Kim, R. B.; Fromm, M. F.; Wandel, C.; Leake, B.; Wood, A. J.; Roden, D. M.; Wilkinson, G. R. The drug transporter P-glycoprotein limits oral absorption and brain entry of HIV-1 protease inhibitors. J. Clin. Invest. 1998, 101, 289−294.

1513

dx.doi.org/10.1021/mp300214k | Mol. Pharmaceutics 2013, 10, 1505−1513