Efflux of Monoclonal Antibodies from Rat Brain by Neonatal Fc Receptor

brain research 1534 (2013) 13–21

Available online at www.sciencedirect.com

www.elsevier.com/locate/brainres

Research Report Efﬂux of monoclonal antibodies from rat brain by neonatal Fc receptor, FcRn

n Philip R. Coopera, ,1, Gary J. Ciambroneb,1, Connie M. Kliwinskic, Eva Mazeb, Lowell Johnsonb, Qianqiu Lid, Yiqing Fenga, Pamela J. Hornbya aAntibody Drug Discovery, USA bDrug Product Development, USA cBiologics Pharmacology and Toxicology, USA dBio-Stats PA, USA article info abstract

Article history: Monoclonal antibody (mAb) engineering that optimizes binding to receptors present on Accepted 17 August 2013 brain vascular endothelial cells has enabled them to cross through the blood–brain barrier Available online 23 August 2013 (BBB) and access the brain parenchyma to treat neurological diseases. However, once in the brain the extent to which receptor-mediated reverse transcytosis clears mAb from the Keywords: brain is unknown. The aim of this study was to determine the contribution of the neonatal Blood–brain-barrier Fc-receptor (FcRn) in rat brain efflux employing two different in vivo drug delivery models. IgG Two mAb variants with substantially different affinities to FcRn, and no known neuronal FcRn targets, (IgG1 N434A and H435A) were administered to rats via intranasal-to-central Efflux nervous system (CNS) and intra-cranial dosing techniques. Levels of full-length IgG were Intracranial quantified in serum and brain hemispheres by a sensitive enzyme-linked immunosorbent Intranasal assay (ELISA). Following intra-nasal delivery, low cerebral hemisphere levels of variants were obtained at 20 min, with a trend towards faster clearance of the high FcRn binder (N434A); however, the relatively higher serum levels confounded analysis of brain FcRn contribution to efflux. Using stereotaxic coordinates, we optimized the timing and dosing regimen for injection of mAb into the cortex. Levels of N434A, but not H435A, decreased in the cerebral hemispheres following bilateral injection into the rat cortex and higher levels of N434A were detected in serum compared to H435A after 24 h. Immunohistochemical staining of human IgG1 in sections of cortex was consistent with these results, illustrating relatively less intense immunostaining in N434A than H435A dosed animals. Using two in vivo methods with direct cranial administration, we conclude that FcRn plays an important role in efflux of IgG from the rat brain. & 2013 Elsevier B.V. All rights reserved.

Abbreviations: FcRn, neonatal Fc receptor; IgG, immunoglobulin G; WT, wild type n Corresponding author at: Biologics Research, Janssen R&D, Pharmaceutical Companies of Johnson & Johnson, Welsh & McKean Roads, PO Box 776, Spring House, PA 19477, USA. E-mail address: [email protected] (P.R. Cooper). 1Philip R. Cooper and Gary J. Ciambrone are joint-ﬁrst authors.

et al., 2001; Yeung et al., 2009). The monoclonal antibodies 1. Introduction were generated to target respiratory syncytial virus (RSV) and would not be expected to bind to targets in the brain. A The blood brain barrier (BBB) is the protective layer separating human mAb was used to avoid potentially faster clearance of fl the systemic circulation from the brain extracellular uid. mouse mAb dosed to rats, and enable detection of the human It is made up of tight junctions between endothelial cells on Fc in rat tissues. Studies were 24 h or less to avoid differences cranial capillaries, a thick basement membrane and astro- in serum levels due to the relationship of FcRn binding cytic end-feet. The BBB serves to restrict bacteria and other affinity and circulating half-life. The two variants have been large hydrophilic molecules from entering the brain parench- shown to have rat FcRn binding affinities, of 77 nM for N434A yma, while allowing small hydrophobic molecules and nutri- and 41000 nM for H435A at pH 6.0 (Kliwinski et al., 2013). ents to enter. Pharmacologic treatment of neurologic diseases has relied on brain penetration of small lipophilic molecules. However, 2. Results where high selectivity and potency is desirable, an alternative therapeutic approach could be the use of monoclonal anti- 2.1. Physicochemical characteristics of the FcRn binding bodies (mAbs). Immunoglobulin-Gs (IgGs), along with other variants plasma proteins, are large hydrophilic molecules that are unable to pass through the BBB in sufficient quantity to be Both variants had identical pI values of 7.2. The circular efficacious when systemically administered (Poduslo et al., dichroism (CD) spectra for both the near and far ultra-violet 1994). Researchers are currently experimenting with receptor- ranges showed very similar secondary and tertiary protein based brain endothelial transcytosis, such as using the trans- structure for both of the variants. They had the same Size ferrin receptor (Bickel et al., 1994; Pardridge et al., 1991; Yu Exclusion Chromatography (SEC) profiles with no covalent et al., 2011) or insulin receptor (Boado et al., 2007; Pardridge aggregates, and were stable at 25 1C for 4 d. There was no et al., 1995) for IgGs to enter the brain parenchyma. However, interaction with mucins, which would confound their deliv- once mAbs enter the brain, the extent to which they are ery by intranasal route (data not shown). cleared by receptor-mediated reverse transcytosis is not well- known. 2.2. Pharmacokinetics of FcRn binding variants after Evidence of the involvement of an Fc-receptor in the intranasal-to-CNS administration clearance of IgG from the central nervous system (CNS) has been shown by a shorter half-life of IgG, compared to IgM FcRn binding variants (H435A and N434A) were administered (antibody that lacks Fc region), in both rat and monkey intranasally into each nostril of rats (40 nmol/rat) and plasma fl fl cerebrospinal uid (Bergman et al., 1998). Moreover, ef ux of was collected after 20, 40, and 90 min post-dose. The levels of IgG though the BBB is competitively inhibited by the addition the FcRn binding variant increased to levels that reached of Fc fragments (Boado et al., 2007; Zhang and Pardridge, 2001). 200 ng/mL in the circulation at a greater rate than the non- β fl Indeed, the Fc-receptor mediated A -IgG ef ux mechanism FcRn binding variant (Fig. 1A). Rat brain hemispheres were has been shown to facilitate the clearance of IgG complexes collected after brain perfusion, at 20, 40, and 90 min post- from brains (Deane et al., 2005). dose from different rats. FcRn binding variants delivered into There are data to both support and refute the role of the the brain (ng/g) were detected by an ELISA-based MSD assay fl neonatal Fc-receptor (FcRn) in IgG ef ux from the brain. Using that detects full-length mAb (Fig. 1B). N434A entered the non-compartment mathematical modeling in mice which lack brain at a faster rate than H435A and peaked at a higher level fl FcRn functionally, there was no apparent difference in ef ux at 20 min. Despite the greater degree of uptake of N434A, compared to wild-type mice based on labeled IgG and residual levels of this variant dropped to very low levels within the blood volume (Abuqayyas and Balthasar, 2013; Garg and same 90 min timeframe as H435A. Statistical comparison of Balthasar, 2009). FcRn is visualized by confocal microscopy in the AUC values generated for each variant showed a statis- brain microvasculature endothelial cells (Schlachetzki et al., tically significant difference (N434A AUC 1637 ng min/g vs. fl 2002), but whether the receptor is involved in ef ux in addition H435A AUC 827 ng min/g, Po0.05), representing an approxi- to its role in recycling IgG is unknown. In vascular endothelial mately two-fold faster rate of efflux for N434A compared fi fl cells, IgG is taken up from the circulation by non-speci c uid- to H435A. phase pinocytosis where it binds to FcRn in the acidic endo- To monitor that test article was correctly deposited with some. It is recycled to the capillary lumen where it has a long the tube insertion technique; olfactory epithelia from both half-life (Roopenian and Akilesh, 2007). It is therefore postu- nostrils were collected at 20, 40, and 90 min post-dose and lated that expression of FcRn located in brain endothelial cells analyzed for FcRn binding variants. The PK profiles of each fl (Schlachetzki et al., 2002) may be involved in the ef ux of IgGs are shown in Fig. 1C and D. In both epithelia, the N434A from the brain. variant was cleared at a much faster rate than the H435A, and fi The aim of the current study was to de ne the role of FcRn the AUC values for each were significantly different (left AUC in IgG efflux from the rat brain using two variants of a H435A 2.2 107 ng min/g vs. left AUC N434A 1.4 107 ng min/g, recombinant human IgG1 mAb that either had increased P¼0.01; right AUC H435A 2.6 107 ng min/g vs. right AUC FcRn binding (IgG1 N434A) or decreased FcRn binding (IgG1 N434A 1.6 107 ng min/g, Po0.01). These data are indicative of H435A) compared to wild-type by incorporating mutations at an active FcRn mediated efflux mechanism removing the FcRn- the 434 and 435 amino acid positions (EU numbering) (Firan binding N434A variant from the olfactory epithelium. brain research 1534 (2013) 13–21 15

300 60 N434A High FcRn Binder H435A Low FcRn Binder 250 H435A Low FcRn Binder 50 N434A High FcRn Binder 200 40

150 30 100 20 Plasma Levels Plasma 50

Hemisphere Levels Hemisphere 10 B21M Variants (ng/ml) B21M Variants 0 Variants B21M (ng/g) 0 10 20 30 40 50 60 70 80 90 100 0 Time (min) 20 40 90 Time (min)

450,000 450,000

400,000 P<0.05 400,000 P<0.05

350,000 350,000

300,000 300,000

250,000 250,000

200,000 200,000

150,000 150,000 IgG conc (ng/g)IgG IgG conc (ng/g)IgG

100,000 100,000 Left Olfactory Epithelium 50,000 Right Olfactory Epithelium 50,000 20 40 90 20 40 90 Time (min) Time (min)

Fig. 1 – PK proﬁles of variants after intranasal-to-CNS administration: H435A or N434A were administered intranasally to rats. A total of 50 ll, of a 60 mg/ml solution of either variant, was applied to each nostril sequentially over 12 min (4 25 ll, right and left, 4 min interval). (A) Serum was collected at 20, 40, and 90 min post-dose and full-length IgG quantiﬁed. Animals were euthanized at 20, 40, and 90 min post-dose and (B) hemispheres (n¼12) and (C) left and (D) right olfactory epithelia (n¼6) analyzed for N434A (■) or H435A (●). Values are mean 7 SEM.

3 2.3. Pharmacokinetics of FcRn binding variants after intra-cranial administration 2 2.3.1. Serum levels of the FcRn binding IgG variant, N435A, increase with time ThecontributionofFcRninIgGbrainefﬂux was suggestive of 1 FcRn-mediated efﬂux but not conclusive after intranasal Serum levels of levels Serum

administration due to the relatively low brain levels and (ng/mL) N434A IgG1 differences in serum levels of the variants. Therefore, we LLOQ: 0.5 ng/mL complimented these studies by direct intracranial stereotaxic 0 0 1 2 3 4 administration. Preliminary experiments were performed to Time (hr) determine a dose that, when administered into the brain via stereotaxic coordinates to the parietal cortex, would result in Fig. 2 – Time-course of serum levels of IgG1 N434A: serum detectable serum levels. To do this, rats were maintained under levels (ng/mL) were measured from rats (n¼15) at 5 and anesthesia for 4 h after unilateral administration of the FcRn 3 min, and 1, 2, and 4 h following intra-cranial m m binding variant (N434A; 2.0 g/mL; 1.2 L) into the right anterior administration of IgG1 N434A (2.4 lg; 2 mg/ml). Intact IgG1 SiFl region of the somatosensory cortex. Serum levels of intact was quantified using the MSD assay. IgG were measured at 5, 30, 60, 120, 180, and 240 min. Following intra-cranial administration of the antibody, low but detectable levels of full-length IgG in serum were detected by 30 min. 2.3.2. Unilateral intra-cranial dosing of IgG variants Serum levels continued to increase up to the termination of the Having established that intact IgG serum levels following experiment at 4 h. The rate of efflux was fairly stable from 0 to intra-cranial administration increased over time, but had not 180min with an average efflux rate of 0.4 ng/mL/h. The rate reached maximal levels after 4 h, serum levels of FcRn binding increased to 0.9 ng/mL/h between 180 and 240 min, with serum variants (N434A, with the FcRn low binding control IgG, H435A) levels of 2.170.5 ng/mL at the final time point (Fig. 2). were measured up to 24 h. A 2.4 mgdose(2.0mg/mL) of either 16 brain research 1534 (2013) 13–21

2.3.3. Bilateral intra-cranial dosing of IgG variants 40 A high concentration of mAb administered unilaterally may B21M IgG1 N434A saturate the FcRn receptor in the vicinity of the injection, thus 30 B21M IgG1 H435A limiting the contribution of receptor-mediated efflux. This could explain the lack of a clear difference between the two variants 20 following unilateral administration; therefore, the variants were dosed bilaterally. The total brain dose of 2.4 mgwasthesameas 10 in the previous study, but a 1.2 mg(0.6mL) dose of either N434A or H435A was administered bilaterally. Serum levels were IgG1 variants (ng/mL) Serum levels of B21M of levels Serum 0 measured at 5 min, 4, and 24 h and rats were allowed to recover 0 4 8 12 16 20 24 from anesthesia immediately after surgery. After 24 h, N434A Time (hr) serum levels reached 25.975.2 ng/mL which was significantly greater (Po0.001) than the serum levels of the low FcRn binding variant, H435A, at 6.670.6 ng/mL. The rate of efflux of N434A 160 was calculated to be three times greater than H435A (Fig. 4A). 140 B21M IgG1 N434A In a separate group of animals, brain hemispheres were 120 B21M IgG1 H435A excised 5 min after either N434A or H435A antibody admin- 100 istration. Levels of N434A detected at 5 min were 2.270.5 mg/g 80 of tissue (average mass of the hemispheres was 1.0 g) which 60 was the expected range after administration. After 24 h, levels of N434A in the hemispheres decreased by 48%. Levels (ng / g tissue) / (ng 40 of the non-FcRn binding mAb, H435A, did not change over the

Percentage change Percentage 20

of B21M IgG1 variants B21M IgG1 of same time period (Fig. 4B). 0 5 mins 4 24 5 mins 4 24 In order to visualize the presence of human IgG immuno- Time (hr) histochemically in the rat brain, three animals were dosed either N434A or H435A bilaterally into the SiFl region of the Fig. 3 – Unilateral intra-cranial dosing: anesthetized Sprague cortex similar to the previous experiment. After 24 h rats were Dawley rats (n¼6/group) were intra-cranially (anterior right perfused and brains prepared for immunohistochemistry hemisphere) administered (2.4 lg; 2 mg/ml) of either IgG1 using an anti-human Fc rabbit polyclonal in thin (5 mm) brain N434A or IgG1 H435A. Intact IgG1 was quantified using the sections. Brains were processed for routine histology (hema- MSD assay. (A) Serum levels (ng/mL) were measured at toxylin and eosin and Luxol fast blue), as well as human IgG 5 min, and 4 and 24 h following intra-cranial administration and anti-rabbit IgG negative controls. At 24 h after dosing both of IgG1 N434A (■) or IgG1 H435A (●). (B) After the final blood N434A and H435A mAb variants were visualized in brain tissue draw, brains were removed, homogenized and intact IgG (Fig. 5). There was a range of staining intensity visualized in quantified IgG1 N434A (solid) or IgG1 H435A (stripes). sections at an equivalent coronal plane in the region of the somatosensory cortex in pyramidal and supporting cells that was localized to the cortex or spread into the corpus callosum region. The staining intensity of N434A (Fig. 5A) in the cortex N434A or H435A was administered into the right anterior SiFl was consistently less than brains injected with H435A (Fig. 5B). region of the cortex of anesthetized rats. The animals in this There was no specific staining detected in PBS microinjected study were anesthetized until after the 4 h blood draw then brains (Fig. 5C) or mAb injected brains incubated in serum allowed to recover. Consistent with the preliminary study, (Fig. 5D). Semi-quantitative immunostaining scores supported levels of full-length IgG in the serum at 5 min were below the the impression that less of the FcRn binding variant was LLOQ for all rats dosed, thus confirming that no surgical visualized in brain after 24 h (HA¼7.071.4 vs. NA¼4.471.7 damage was performed that would lead to systemic contam- of a total of 12.0 maximum staining; P¼0.3 by unpaired t-test), ination. Levels of N434A and H435A were similar 4 h after which is consistent with greater efflux of the FcRn binding administration (4.471.9 and 3.471.9 ng/mL, respectively), but mAb variant. after 24 h there tended to be higher levels of the N434A FcRn- binding variant (20.675.8 and 11.973.1 ng/mL, respectively) which did not attain a level of statistical significance (Fig. 3A). 3. Discussion In brain tissue at the earliest time point of 5 min, levels of N434A (FcRn binding variant) were 17167354 ng/g of tissue This study used two in vivo drug delivery models in rats, and similar to that expected based on dose administered intranasal-to-CNS and direct intra-cranial dosing, to determine (average mass of a hemisphere was 1.0 g). Levels decreased the contribution of FcRn to IgG receptor-dependent reverse- by approximately 40% after 24 h whereas levels of the non- transcytosis (efflux) from the brain parenchyma to the circula- binding variant H435A in the brain hemispheres were tion. Following administration into the brain by these two unchanged over time up to 24 h (Fig. 3B). Levels in the methods, greater amounts of the FcRn binding variant N434A cerebellum, brainstem, and lymph nodes were low and no were detected in the circulation compared to the low-binding difference was detected between the variants (data not FcRn variant control, H435A. Following intra-cranial adminis- shown). tration, levels of H435A in the brain hemispheres did not brain research 1534 (2013) 13–21 17

Thorne, 2012). Functional studies using both small and large 40 P < 0.001 molecules have shown delivery to the hemispheres via this B21M IgG1 N434A mechanism despite this potential hindrance. The main driving 30 B21M IgG1 H435A force for uptake is the concentration gradient of the test article (Barakat et al., 2006; Evseev et al., 2010; Furrer et al., 2009; 20 Gorbatov et al., 2010; Hoekman and Ho, 2011; Romanova et al., 2010; Sipos et al., 2010). However, there are two main factors 10 that do impinge on the efﬁciency of CNS uptake of a molecule using this technique: those related to formulation, and those 0 IgG1 variants (ng/mL) Serum levels of B21M of levels Serum 0 4 8 12 16 20 24 related to physicochemical characteristics. The formulation Time (hr) considerations include the type, pH, and tonicity of the buffer, and/or the presence of excipients representing transport enha- ncers, stabilizers, and muco-adhesives (Pujara et al., 1995; 160 Washington et al., 2000). The important physicochemical 140 characteristics include molecular size, hydrophobicity/hydro- 120 philicity, surface charge, and mucus compatibility (Vyas et al., 100 2006). The test articles used in these studies are similar in 80 P < 0.01 terms of their physicochemical characteristics. Both H435A 60 and N434A have pI values of 7.2, differ by just one amino acid, 40 and have virtually identical secondary and tertiary structures (ng / g tissue) / (ng as measured by circular dichroism. Functionally, they only 20 Percentage change Percentage ﬁ

of B21M IgG1 variants differ in their af nity for the FcRn receptor, have no known 0 5 min 24 hr 5 min 24 hr targets in the brain and therefore are uniquely suited to use B21M IgG1 B21M IgG1 address the role of FcRn in IgG efflux from the brain. N434A H435A Following intra-nasal administration, the data suggested that FcRn contributed to reverse-transcytosis of IgGs from the – Fig. 4 Bilateral intra-cranial dosing: anesthetized Sprague rat brain, but were inconclusive, in part because of different ¼ Dawley rats (n 10/group) were intra-cranially (anterior levels of mAb remaining in the olfactory epithelium. With lower l hemisphere: right and left) administered (1.2 g each site; than expected brain levels after dosing intranasally at 40 nmol/ 2 mg/ml) either IgG1 N434A or IgG1 H435A. Intact IgG1 was rat, and unexpected 100-fold higher serum levels (compared to fi quanti ed using the MSD assay. (A) Serum levels (ng/mL) intra-cranial administration), it is unclear whether variants were measured at 5 min, and 4 and 24 h following intra- entered the brain hemispheres prior to their entering the ■ ● cranial administration of IgG1 N434A ( ) or IgG1 H435A ( ). vascular compartment and confounded the interpretation of fi (B) After the nal blood draw, hemispheres were removed, brainFcRncontributiontolevelsintheserum.However,AUC fi homogenized, and intact IgG quanti ed (ng/g brain tissue) in calculated data of the variant levels that did enter the hemi- total hemispheres. Key: 5 min (solid), 24 h (squares). spheres did provide a statistical significant difference in the rate of efflux of the two variants that did enter the brain. These data suggest a role of FcRn in the reverse-transcytosis of IgGs from change over a 24 h period while the levels of N434A significantly the rat brain. decreased over time. Following unilateral stereotaxic administration, there was Intranasal-to-CNS delivery was used initially because it is a a tendency for greater amounts of N434A in the serum after non-invasive technique. For maximal delivery to the brain 24 h compared to H435A. Importantly, potential damage to hemispheres, the test article had to be applied directly to the the tissue vasculature during the surgery could have resulted olfactory epithelium of the nose (Thorne et al., 2004). Test in systemic contamination; however, the serum levels at article then moves in a paracellular fashion, driven by diffu- 5 min after administration were below the lower level of sion, past the olfactory epithelium, and into the nasal lamina quantification in all animals suggesting the absence of such propria beneath (Dhuria et al., 2010). This space is contiguous damage. Sample size (n¼6) may have been the reason for the with channels through the cribriform plate, which contain the lack of statistical significance, as well as potential saturation axons of the olfactory neurons. Earlier studies have shown of FcRn due to high local concentrations of the compound that a certain percentage of cerebrospinal fluid (CSF) exits the over the duration of the study (r24 h). To avoid this problem, brain through the cribriform plate and enters nasal lymphatics the study was repeated with the mAbs dosed bilaterally at the which drain into the cervical lymph nodes (Bradbury et al., same co-ordinates for both right and left brain hemispheres. 1981; Bradbury and Westrop, 1983; Cserr et al., 1992). This The total brain dose was identical to the unilateral dosing, process could represent a hindrance to molecular flow into the but the local concentration at either brain hemisphere was brain. However, subsequent ultra-structural studies have halved. A significant increase in serum levels at 24 h and a demonstrated the existence of neuronal channels that tra- decrease in total brain hemisphere levels were noted for the verse the subarachnoid space (Field et al., 2003; Li et al., 2005) higher affinity FcRn binding variant compared to the low thus preventing direct interaction with CSF. These channels FcRn binding variant. Together these data confirm the trends are believed to provide direct access into the parenchyma of shown in the intranasal and unilateral intra-cranial admin- the brain hemispheres (Dhuria et al., 2010; Lochhead and istration studies; and are in agreement with the previous 18 brain research 1534 (2013) 13–21

N434A; anti-human IgG H435A; anti-human IgG

PBS; anti-human IgG H435A; rabbit serum

Fig. 5 – Greater levels of IgG1 H435A than IgG1 N434A in brain: representative examples of IgG-immunoreactivity in rat brains 24 h after intra-cranial administration of (A) IgG1 N434A; anti-hIgG1; (B) IgG1 H435A; anti-hIgG1; (C) PBS, anti-hIgG1, and (D) IgG1 H435A, rabbit serum.

studies that propose FcRn mediated efflux (Deane et al., 2005; evidence that the IgG crossed the BBB into the brain parench- Deane et al., 2009; Zhang and Pardridge, 2001). yma as the group did not measure IgG levels in the brain The opposite conclusion with regard to the role of FcRn in directly, but instead measured levels in residual blood. It is IgG brain efflux was reached by two studies using two also unclear what the affinity of their IgG antibody was to different experimental methods. β-2-microglobulin knock- murine FcRn. Therefore, there is limited evidence that FcRn out mice, which lack functional FcRn, and wild-type controls had the ability to play a role in the efflux within that showed no difference in brain-to-plasma AUC ratios after previously published study. Another disadvantage of this intravenous administration of an I125 labeled mAb, leading protocol was their use of FcRn knock-out mice. With no FcRn, the investigators to conclude that FcRn does not significantly the recycling and salvation of IgGs would not be present in contribute the low exposure of mAb in brain (Abuqayyas and these mice so IgG half-life would be substantially decreased. Balthasar, 2013; Garg and Balthasar, 2009). In contrast another Although the study involving these mice was shortened to 4 d study reported that brain clearance after systemically admi- to compensate for this, there would be significantly less IgG in nistered mAb was reduced in FcRn/mice (Deane et al., the circulation after 4 d (95% less). This adds differences in 2005; Deane et al., 2009). AUC of mAb in WT and knock-out mice confounding brain The most significant difference between the protocol used exposure. Indeed, clearance was eight-fold faster in FcRn in the studies with opposing conclusions (Abuqayyas and knock-out mice compared to the other strains, as would be Balthasar, 2013; Garg and Balthasar, 2009), and the protocol expected (Abuqayyas and Balthasar, 2013). In addition, the used in the current study, and others (Zhang and Pardridge, observed brain to plasma AUC ratio was greater in mice in the 2001), was that IgG was administered directly into the brains of second study and the data was adjusted for differences in rats and serum and brain levels were measured. The other hematocrit (Abuqayyas and Balthasar, 2013; Garg and studies (Abuqayyas and Balthasar, 2013; Garg and Balthasar, Balthasar, 2009). The emphasis on mathematical modeling 2009) involved administration of IgG into the circulation of may account for the differences in their conclusions compared wild-type and FcRn knock-out mice and relied on the ability of to the observations in FcRn knock-out mice where brain IgG to cross into the brain to measure AUC differences clearance of a systemically administered mAb was lower than between brain content and serum levels. There was no direct wild type controls (Deane et al., 2005; Deane et al., 2009). brain research 1534 (2013) 13–21 19

In summary, this study demonstrates that FcRn plays an position on an acrylic support with their heads positioned at a important role in the efflux of IgGs. These results need to be 451 angle to the horizon. A microcannula (BioTime, Berkeley, taken into account in future studies evaluating therapeutic CA, USA) was inserted to a depth of 1.5 cm into the right nostril IgGs containing an Fc portion when targets in the brain are and either H435A or N434A (1.5 mg in 25 mLatarateof50mL/ investigated. As the variants in the present study did not min) was infused by syringe pump (Harvard Apparatus, Cam- have a neuronal target, future studies should consider the bridge, MA, USA). After 4 min, the same variant was applied impact of target receptor occupancy for the therapeutic target into the left nostril and the alternating procedure repeated for a to determine the maintenance of IgG brain levels or when total application of 50 mL/nostril, therefore 400 mmol/L. After the investigating the relevance of FcRn-dependent efflux. final dose, the microcannula was removed and inhalant anes- thetic was continued for 8 min with the animal supine and the head angle maintained at a 451 angle. At 20 min after the start 4. Experimental procedure of the first dose, animals were euthanized and tissues collected. For longer time points, anesthesia was maintained for 20 min 4.1. Animals andthenremovedandanimalswereallowedtoawaken.

– – Male Sprague Dawley rats, 7 10 weeks old (200 300 g) (Charles 4.5. Intra-cranial administration River, Wilmington, MA, USA) were kept in plastic filter-topped cages and allowed free access to food and water. All animal The animals were placed in a chamber for induction of studies were performed in accordance with the Federal Animal anesthesia with isoflurane (initially 2–4%) and then removed Welfare Act and methods approved by the Institutional Animal andplacedinastereotaxicdevice(Knopf)onasurgicalpad Care and Use Committee at Janssen R&D. maintained at 37 1C with a nose cone for maintenance of anesthesia (2% isoflurane) (Cetin et al., 2006). Buprenorphine 4.2. Monoclonal antibody preparation (0.01–0.05 mg/kg) analgesia was administered sub-cutaneous. A midline incision was made to expose the bregma and was The mAbs used in this study were variants of a recombinant used to locate the ipsilateral primary somatosensory forelimb chimeric human IgG1 monoclonal antibody specificforhuman (SiFl) (þ0.2 mm anterior and 4.0 mm lateral3.0 mm deep). For respiratory syncytial virus (RSV). For intranasal dosing the FcRn unilateral administration, only the right hemisphere was dosed. binding mutants, IgG1 H435A and IgG1 N434A, underwent A drill with a small burr attachment was used to introduce a buffer exchange from phosphate-buffered saline (PBS). The 1 mm hole into the skull to allow entry for a Hamilton syringe buffer used for exchange was 10 mM histidine/5.5% sucrose (5 mL) to inject the test article, either IgG1 N434A or IgG1 H435A (pH 5.3), 150 mM NaCl. After three rounds of exchange, into the brain (antibodies were dosed in pH-neutral buffers (pH the mutants were concentrated to 66.67 mg/mL and pro- 7.4), unless otherwise stated). For unilateral administration a pylene glycol was added to a final concentration of 10%, making 1.2 mL bolus of test article (2.0 mg/mL) was administered at a rate the final concentration of the mutants 60 mg/mL. These pre- of 0.4 mL/min (total dose¼2.4 mg, or 14 mmol/L). For bilateral parations were used for intranasal dosing. administration the volume administered to each side was halved (0.6 mL); therefore, the total dose administered was equal 4.3. Physicochemical characteristics of the FcRn binding to the unilateral administration. Care was taken to perform the variants surgical procedure following aseptic techniques and a surgical plane of anesthesia was maintained throughout the entire The physiochemical characteristics of the two variants were procedure. The incisions of animals were closed and the assessed and compared as this is a factor which can con- animals were allowed to recover from anesthesia following tribute to a difference in intra-nasal uptake. The predicted the 5 min baseline blood draw. isoelectric point (pI) values of the variants were derived using Vector NTI sequence analysis software (Invitrogen). Circular 4.6. Blood and brain tissue collection and homogenization dichroism (CD) spectroscopy to analyze structure was performed on the variants (0.25 mg/mL in PBS) and compared to Animals were given isoflurane to facilitate retro-orbital blood PBS alone. Spectral acquisition was measured at 6 spectra draws of approx 100 mL. All blood samples were allowed to from 190–260 nm at 1 nm path length and 1 nm intervals with stand for at least 30 min, but no longer than 1 h, centrifuged at a 2 s signal at 20 1C (Circular Dichroism Spectropolarimeter, 3500 rpm for 15 min and the serum separated and stored at Model 400, Aviv). The CD spectra were averaged and the net 80 1C. Following the final blood collection a full body perfu- spectrum of the variants obtained by subtracting the average sion was performed to remove residual blood from the brain PBS scores. Spectra were fitted for species content using an via cardiac puncture using 60 mL (15 mL/min, syringe pump) MWR of 106 g/mol (150 kDa). of a PBS solution containing protease inhibitors (1 tablet/ 10 mL, Complete Mini, Roche, Indianapolis, IN, USA) and 4.4. Intranasal-to-CNS administration 5 mM EDTA. Following perfusion, the anterior cervical lymph nodes were removed and placed in FastPrep tubes. The head Pre-dose plasma samples were collected from the animals via was then decapitated and the brain removed and halved into tail vein a day prior to dosing. On the day of dosing rats were two hemispheres (excluding the olfactory bulbs), brainstem anesthetized with sevoflurane (5.0–6.0% sevoflurane, 3.0 L/min (hypothalamus to level of cistern magna of hindbrain), and

O2; Abbott Labs., Princeton, NJ, USA) while placed in a supine cerebellum. The olfactory epitheliums were collected following 20 brain research 1534 (2013) 13–21

intranasal-to-CNS administration only. Brain tissues were were then added together to obtain a sample total score placed into pre-weighed lysis Matrix D tubes (MP Biomedical, (maximum¼12). Cat# 6913-500) or IKA tubes (IKA Cat# 3703100) then re- 1 weighed and quickly frozen on dry ice and stored at 80 C 4.9. Statistical analyses until homogenization. Brain hemispheres were homogenized in 1:10 homogeniza- 4.9.1. Intra-cranial administration tion buffer (assume 1 g of tissue is 1 mL volume) (TBS 25 mM Data were plotted as mean 7 standard error of the mean Tris [pH 7.8], 150 mM NaCl, 1% Triton X-100, 5 mM EDTA (SEM). Statistical tests used were either a one- or two-way and cOmplete, EDTA-free Protease Inhibitor Cocktail Tablets ANOVA with post-test analysis performed using Bonferroni's [1 tablet/10 mL]). The brain hemispheres were thawed and multiple comparison test. P-values less than 0.05 were con- homogenized in buffer solution using an IKA instrument (Ultra sidered significant. Turrax, IKA Cat# 3645001) (setting# 6, 20 s). The left and right olfactory epithelia were diluted in the same homogenization 4.9.2. Intranasal-to-CNS buffer, and homogenized using the Bio101 FastPrep instru- A fixed effects model, including group, time, and group by time ment (6.5 m/s; 40 s). Full length IgG in the brain and tissue interaction as fixed effects, was fittotheFcRnvariantconcen- samples was quantified within 24 h of homogenization. tration data (Wolfsegger and Jaki, 2005; Wolfsegger, 2007). For each FcRn variant group, with application of the trapezoidal rule, 4.7. Full-length human IgG detection area under the curve (AUC) measurements: AUC (0-Last) (AUC of mean concentration from 0 min to 90 min) were calculated based Full-length IgG antibodies were quantified using the Meso on the mean concentrations across three time points and under Scale Discovery (MSD) electrochemiluminescent assay. The the assumption of zero concentration at time 0. Group mean mAbs used in this study were a recombinant chimeric human concentrations by time were estimated and 95% confidence IgG1 monoclonal antibody specific for human respiratory intervals were constructed. FcRn variant effects were examined syncytial virus (RSV). An anti-idiotypic antibody specific for using group AUCs as linear functions of the mean concentrations the anti-RSV IgGs was labeled with sulfo-NHS-LC-biotin and at three times. The Delta method was applied for the inference used as a capture antibody. A pan anti-human IgG1/IgG2 of between-group AUC differences. All the 95% confidence antibody labeled with MSD Sulfo-TAG NHS Ester was used as intervals were two-sided t-type intervals and all P-values the detection antibody. The electrochemiluminescent labels were from two-sided t-tests. For all tests, P-values less than added emit light when electrochemically stimulated. Electro- 0.05 were considered significant (Wolfsegger and Jaki, 2005; des bound to the sample-sandwich initiate the detection Wolfsegger, 2007). process. A standard curve for the anti-RSV IgGs in neat rat serum was observed over the range of 50,000–50 pg/mL. The lower limit of quantification (LLOQ) was 0.5 ng/mL using 25 mL Acknowledgments of neat rat serum. We are grateful to Steve Jarantow, Deidra Bethea and Bethany Swencki-Underwood for their assistance in the physiochem- 4.8. Immunohistochemistry ical characterizing of the antibodies, and Bernie Scallon for helpful discussion. At 24 h after intracranial dosing (3 dosed N434A, 3 dosed H435A, and one control dosed PBS) and whole body blood perfusion references (10% neutrally buffered formalin, NBF), the brain hemispheres were removed and a 1 mm section proximal to the site of dosing was placed between two biopsy sponges and fixed in Abuqayyas, L., Balthasar, J.P., 2013. Investigation of the role of 10% NBF. The rats used in this study were separate from the FcγR and FcRn in mAb distribution to the brain. Mol. Pharm. fl previous ef ux study. Following dehydration, the tissues were 10, 1505–1513. exposed to the primary antibody, human IgG1 (1 mg/mL; 1 h; Barakat, N.S., Omar, S.A., Ahmed, A.A., 2006. Carbamazepine Epitomics), then stained and counter-stained and dehydrated uptake into rat brain following intra-olfactory transport. J. further. The above procedures were completely automated Pharm. Pharmacol. 58, 63–72. using the TechMate 500 (BioTek Solutions). Positive staining Bergman, I., Burckart, G.J., Pohl, C.R., Venkataramanan, R., Barmada, M.A., Griffin, J.A., Cheung, N.K., 1998. was indicated by the presence of a brown chromogen (DAB- Pharmacokinetics of IgG and IgM anti-ganglioside antibodies HRP) reaction product. Representative images were obtained in rats and monkeys after intrathecal administration. J. fi with an Olympus Micro re digital camera (M/N S97809) Pharmacol. Exp. Ther. 284, 111–115. attached to an Olympus BX60 microscope. Samples were Bickel, U., Kang, Y.S., Yoshikawa, T., Pardridge, W.M., 1994. In vivo individually scored on a semi-quantitative basis for human demonstration of subcellular localization of anti-transferrin IgG immunostaining. Each sample received an intensity human receptor monoclonal antibody-colloidal gold conjugate in IgG score per region of staining from each hemisphere. These brain capillary endothelium. J. Histochem. Cytochem. 42, 1493–1497. regions were Pyramidal Cells, Other Neural & Support Cells, and Boado, R.J., Zhang, Y., Zhang, Y., Pardridge, W.M., 2007. Corpus Callosum Area. Intensity scores were graded on a 5 Humanization of anti-human insulin receptor antibody for þ þ þ þ point scale; 0, 1 ,2 ,3 ,or4 staining intensity (where 0¼no drug targeting across the human blood–brain barrier. þ staining and 4 ¼completely saturated staining). The scores Biotechnol. Bioeng. 96, 381–391. brain research 1534 (2013) 13–21 21

Bradbury, M.W., Cserr, H.F., Westrop, R.J., 1981. Drainage of Pardridge, W.M., Buciak, J.L., Friden, P.M., 1991. Selective cerebral interstitial fluid into deep cervical lymph of the transport of an anti-transferrin receptor antibody through the rabbit. Am. J. Physiol. 240, F329–F336. blood–brain barrier in vivo. J. Pharmacol. Exp. Ther. 259, 66–70. Bradbury, M.W., Westrop, R.J., 1983. Factors influencing exit of Pardridge, W.M., Kang, Y.S., Buciak, J.L., Yang, J., 1995. Human substances from cerebrospinal fluid into deep cervical lymph insulin receptor monoclonal antibody undergoes high affinity of the rabbit. J. Physiol. 339, 519–534. binding to human brain capillaries in vitro and rapid Cetin, A., Komai, S., Eliava, M., Seeburg, P.H., Osten, P., 2006. transcytosis through the blood–brain barrier in vivo in the Stereotaxic gene delivery in the rodent brain. Nat. Protoc. 1, primate. Pharm. Res. 12, 807–816. 3166–3173. Poduslo, J.F., Curran, G.L., Berg, C.T., 1994. Macromolecular Cserr, H.F., Harling-Berg, C.J., Knopf, P.M., 1992. Drainage of brain permeability across the blood–nerve and blood–brain barriers. extracellular fluid into blood and deep cervical lymph and its Proc. Natl. Acad. Sci. USA 91, 5705–5709. immunological significance. Brain Pathol. 2, 269–276. Pujara, C., Shao, Z., Duncan, M., Mitra, A., 1995. Effects of Deane, R., Sagare, A., Hamm, K., Parisi, M., LaRue, B., Guo, H., formulation variables on nasal epithelial cell integrity: Wu, Z., Holtzman, D.M., Zlokovic, B.V., 2005. IgG-assisted age- biochemical evaluations. Int. J. Pharm. 114, 197–203. dependent clearance of Alzheimer's amyloid beta peptide by Romanova, G.A., Shakova, F.M., Gorbatov, V.Y., Kvashennikova, Y.N., the blood–brain barrier neonatal Fc receptor. J. Neurosci. 25, Davydova, T.V., 2010. Effect of antibodies to glutamate on 11495–11503. retention of conditioned passive avoidance response in rats with Deane, R., Bell, R.D., Sagare, A., Zlokovic, B.V., 2009. Clearance of ischemic injury of the prefrontal cortex. Bull. Exp. Biol. Med. 149, amyloid-beta peptide across the blood–brain barrier: 289–292. implication for therapies in Alzheimer's disease. CNS Neurol. Roopenian, D.C., Akilesh, S., 2007. FcRn: the neonatal Fc receptor Disord. Drug Targets 8, 16–30. comes of age. Nat. Rev. Immunol. 7, 715–725. Dhuria, S.V., Hanson, L.R., Frey 2nd, W.H., 2010. Intranasal Schlachetzki, F., Zhu, C., Pardridge, W.M., 2002. Expression of the delivery to the central nervous system: mechanisms and neonatal Fc receptor (FcRn) at the blood–brain barrier. J. experimental considerations. J. Pharm. Sci. 99, 1654–1673. Neurochem. 81, 203–206. Evseev, V.A., Vetrile, L.A., Zaharova, I.A., 2010. Effect of intranasal Sipos, E., Kurunczi, A., Feher, A., Penke, Z., Fulop, L., Kasza, A., administration of anti-glutamate antibodies after stress Horvath, J., Horvat, S., Veszelka, S., Balogh, G., Kurti, L., Eros, I., exposure on the stress response. Bull. Exp. Biol. Med. 149, Szabo-Revesz, P., Parducz, A., Penke, B., Deli, M.A., 2010. 551–553. Intranasal delivery of human beta-amyloid peptide in rats: Field, P., Li, Y., Raisman, G., 2003. Ensheathment of the olfactory effective brain targeting. Cell. Mol. Neurobiol. 30, 405–413. nerves in the adult rat. J. Neurocytol. 32, 317–324. Thorne, R.G., Pronk, G.J., Padmanabhan, V., Frey 2nd, W.H., 2004. Firan, M., Bawdon, R., Radu, C., Ober, R.J., Eaken, D., Antohe, F., Delivery of insulin-like growth factor-I to the rat brain and Ghetie, V., Ward, E.S., 2001. The MHC class I-related receptor, spinal cord along olfactory and trigeminal pathways following FcRn, plays an essential role in the maternofetal transfer of intranasal administration. Neuroscience 127, 481–496. gamma-globulin in humans. Int. Immunol. 13, 993–1002. Vyas, T.K., Tiwari, S.B., Amiji, M.M., 2006. Formulation and Furrer, E., Hulmann, V., Urech, D.M., 2009. Intranasal delivery of physiological factors influencing CNS delivery upon intranasal ESBA105, a TNF-alpha-inhibitory scFv antibody fragment to administration. Crit. Rev. Ther. Drug Carrier Syst. 23, 319–347. the brain. J. Neuroimmunol. 215, 65–72. Washington, N., Steele, R.J., Jackson, S.J., Bush, D., Mason, J., Gill, D.A., Garg, A., Balthasar, J.P., 2009. Investigation of the influence of Pitt, K., Rawlins, D.A., 2000. Determination of baseline human FcRn on the distribution of IgG to the brain. AAPS J. 11, nasal pH and the effect of intranasally administered buffers. Int. J. 553–557. Pharm. 198, 139–146. Gorbatov, V.Y., Trekova, N.A., Fomina, V.G., Davydova, T.V., 2010. Wolfsegger, M.J., Jaki, T., 2005. Estimation of AUC from 0 to Antiamnestic effects of antibodies to glutamate in infinity in serial sacrifice designs. J. Pharmacokinet. experimental Alzheimer's disease. Bull. Exp. Biol. Med. 150, Pharmacodyn. 32, 757–766. 23–25. Wolfsegger, M.J., 2007. Establishing bioequivalence in serial Hoekman, J.D., Ho, R.J., 2011. Enhanced analgesic responses after sacrifice designs. J. Pharmacokinet. Pharmacodyn. 34, 103–113. preferential delivery of morphine and fentanyl to the olfactory Yeung, Y.A., Leabman, M.K., Marvin, J.S., Qiu, J., Adams, C.W., epithelium in rats. Anesth. Analg. 113, 641–651. Lien, S., Starovasnik, M.A., Lowman, H.B., 2009. Engineering Kliwinski, C., Cooper, P., Perkinson, R., Mabus, J., Tam, S., human IgG1 affinity to human neonatal Fc receptor: impact of Wilkinson, T., Giles-Komar, J., Scallon, B., Powers, G., Hornby, P., affinity improvement on pharmacokinetics in primates. J. 2013. The contribution of FcRn binding to intestinal uptake of Immunol. 182, 7663–7671. IgG in suckling rat pups and human FcRn-transgenic mice. Am. Yu, Y.J., Zhang, Y., Kenrick, M., Hoyte, K., Luk, W., Lu, Y., Atwal, J., J. Physiol. Gastrointest. Liver Physiol. 304, G262–G270. Elliott, J.M., Prabhu, S., Watts, R.J., Dennis, M.S., 2011. Boosting Li, Y., Field, P.M., Raisman, G., 2005. Olfactory ensheathing cells brain uptake of a therapeutic antibody by reducing its affinity and olfactory nerve fibroblasts maintain continuous open for a transcytosis target. Sci. Transl. Med. 3 (84ra44). channels for regrowth of olfactory nerve fibres. Glia 52, Zhang, Y., Pardridge, W.M., 2001. Mediated efflux of IgG molecules 245–251. from brain to blood across the blood–brain barrier. J. Lochhead, J.J., Thorne, R.G., 2012. Intranasal delivery of biologics Neuroimmunol. 114, 168–172. to the central nervous system. Adv. Drug Delivery Rev. 64, 614–628.