Tr a n s l a t i o n a l O n c o l o g y
`www.transonc.com
`
`Volume 6 Number 5
`
`October 2013
`
`pp. 562–572 562
`
`The Effect of Molecular Weight,
`PK, and Valency on Tumor
`Biodistribution and Efficacy of
`Antibody-Based Drugs1,2
`
`Ruth Muchekehu*, Dingguo Liu*, Mark Horn*,
`Lioudmila Campbell*, Joselyn Del Rosario*,
`Michael Bacica*, Haim Moskowitz*,
`Trina Osothprarop*, Anouk Dirksen*,
`Venkata Doppalapudi*, Allan Kaspar*,
`Steven R. Pirie-Shepherd† and Julia Coronella*
`
`*CoxV Pfizer Worldwide Research and Development,
`San Diego, CA; †Pfizer Worldwide Research and
`Development, Oncology Research Unit, San Diego, CA
`
`Abstract
`Poor drug delivery and penetration of antibody-mediated therapies pose significant obstacles to effective treatment
`of solid tumors. This study explored the role of pharmacokinetics, valency, and molecular weight in maximizing drug
`delivery. Biodistribution of a fibroblast growth factor receptor 4 (FGFR4) targeting CovX-body (an FGFR4-binding
`peptide covalently linked to a nontargeting IgG scaffold; 150 kDa) and enzymatically generated FGFR4 targeting
`F(ab)2 (100 kDa) and Fab (50 kDa) fragments was measured. Peak tumor levels were achieved in 1 to 2 hours for
`Fab and F(ab)2 versus 8 hours for IgG, and the percentage injected dose in tumors was 0.45%, 0.5%, and 2.5%,
`respectively, compared to 0.3%, 2%, and 6% of their nontargeting controls. To explore the contribution of multi-
`valent binding, homodimeric peptides were conjugated to the different sized scaffolds, creating FGFR4 targeting
`IgG and F(ab)2 with four peptides and Fab with two peptides. Increased valency resulted in an increase in cell surface
`binding of the bivalent constructs. There was an inverse relationship between valency and intratumoral drug con-
`centration, consistent with targeted consumption. Immunohistochemical analysis demonstrated increased size
`and increased cell binding decreased tumor penetration. The binding site barrier hypothesis suggests that limited
`tumor penetration, as a result of high-affinity binding, could result in decreased efficacy. In our studies, increased
`target binding translated into superior efficacy of the IgG instead, because of superior inhibition of FGFR4 prolifera-
`tion pathways and dosing through the binding site barrier. Increasing valency is therefore an effective way to
`increase the efficacy of antibody-based drugs.
`
`Translational Oncology (2013) 6, 562–572
`
`Introduction
`Effective antibody therapies for targeting solid tumors are limited by
`poor penetration [1] and very low percent of injected dose (ID)
`reaching tumor [2]. Limited tumor penetration, caused by hetero-
`geneous antigen expression [3] and blood supply [4], increased inter-
`stitial fluid pressure [5,6], as well as a so-called “binding site barrier”
`caused by high-affinity binding [7,8] are thought to contribute to less
`effective therapy by leaving viable cells untargeted [6]. As a conse-
`quence, alternatives to full-length IgG drugs have been widely inves-
`tigated as a means of improving penetration [9,10].
`Using a fibroblast growth factor receptor 4 (FGFR4) targeting
`CovX-body (a nontargeting IgG, covalently linked to an FGFR4-
`targeting peptide; 150 kDa) and enzymatically generated FGFR4
`targeting F(ab)2 (100 kDa) and Fab (50 kDa), we addressed the role
`
`of size, pharmacokinetics (PK), and avidity in tumor uptake, pene-
`tration, and ultimately efficacy.
`Net drug levels in the tumor are driven by the PK properties
`(influenced by the dose and rate of plasma clearance), diffusion rate
`
`Address all correspondence to: Steven R. Pirie-Shepherd, PhD, Pfizer Worldwide
`Research and Development, Oncology Research Unit, 10777 Science Center Drive,
`San Diego, CA 92121. E-mail: steven.pirie-shepherd@pfizer.com
`1The authors disclose no potential conflicts of interest.
`2This article refers to supplementary materials, which are designated by Figures W1 to
`W4 and are available online at www.transonc.com.
`Received 17 May 2013; Revised 8 July 2013; Accepted 9 July 2013
`
`Copyright © 2013 Neoplasia Press, Inc.
`1944-7124/13 DOI 10.1593/tlo.13409
`
`Open access under CC BY-NC-ND license.
`
`1 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`Translational Oncology Vol. 6, No. 5, 2013
`
`Biodistribution of Antibody Scaffolds Muchekehu et al.
`
`563
`
`(determined by the size and properties of the biotherapeutic), bind-
`ing affinity, and rate of consumption of the drug [3,11,12].
`IgG drug scaffolds inherently have excellent PK properties compared
`to other protein therapeutics because of both their molecular weight
`and ability to bind to the neonatal FcRn receptor, which recycles
`molecules that bind to them back to the serum maintaining elevated
`levels. The limited and heterogeneous tumor penetration of IgGs,
`however, has led to the use of smaller IgG fragments such as Fabs,
`scFv′s, and diabodies [13–15], which can, in theory, diffuse more
`efficiently through tumors, translating into more favorable ID ratios
`at earlier time points [16]. The use of antibody fragments though must
`be balanced by the shorter serum half-lives of non–Fc-containing con-
`structs and the potential for more rapid distribution to normal tissues.
`As well as PK, increased valency may also drive tumor biodistribu-
`tion and efficacy, although the role of valency in tumor retention has
`yielded sometimes conflicting data. Increasing the valency increased the
`tumor uptake of human epidermal growth factor receptor 2 (HER2)
`binding diabodies [13,17], while increasing the valency of HER2 bind-
`ing DARPins, decreased tumor uptake [18]. In those studies, increased
`valency was achieved by doubling the molecular weight, and therefore,
`the role of increasing the size (and potentially decreasing clearance time)
`in tumor uptake and retention could not be distinguished from the
`role of increased valency. However, in other studies using divalent
`(scFv′)2 molecules with zero, one, and two binding sites (same molec-
`ular weight), three-fold greater tumor retention was achieved with the
`construct with two binding sites [19].
`A CovX-body is a peptide antibody fusion generated by conjugat-
`ing a peptide on an azetidinone linker to a nonbinding humanized
`IgG1 monoclonal aldolase antibody [20]. The CovX-body technol-
`ogy allows the increase in the number of targeting peptides on our
`scaffolds from two to four on the bivalent IgG and F(ab)2 and from
`one to two on the Fab using homodimeric FGFR4-targeting pep-
`tides. Increasing the valency of the constructs allows for the mea-
`surement of the role of increased valency on tumor uptake and
`penetration without significantly altering the molecular weight of
`the targeting scaffolds. Increasing the valency of our constructs in-
`creased cell binding of the bivalent constructs. It did not significantly
`increase tumor levels and decreased the penetration of the scaffolds
`into the tumor after a single dose, presenting a so-called “binding site
`barrier.” The binding site barrier is the phenomenon whereby high-
`affinity antibodies accumulate around the vasculature and fail to
`distribute evenly throughout the tumor [8]. This dynamic barrier
`can be overcome by increasing the dose of the antibody [7,21]. In a
`multi-dose efficacy study comparing the tumor growth inhibition
`(TGI) of the IgG homodimer peptide construct versus the IgG mono-
`mer peptide, superior efficacy is observed with the homodimer IgG.
`This current study demonstrates that in a single dose study, PK is
`the most important driver of maximal tumor levels. While higher levels
`of Fab were seen in the tumor after an hour than the IgG and F(ab)2,
`superior maximal tumor concentrations are achieved with the IgG con-
`structs. Increasing the avidity of an IgG is an effective way to maximize
`the efficacy of our targeting scaffolds.
`
`Materials and Methods
`
`Generating F(ab)2 and Fab Scaffolds
`CVX-2000, a humanized IgG1κ antibody [22], was digested over-
`night at 37°C with immobilized papain (Thermo Scientific, Waltham,
`
`MA) to produce Fab or immobilized pepsin (Thermo Scientific) to pro-
`duce F(ab)2. Fab fragments were purified by size exclusion followed by
`binding and elution to Protein L to separate Fab fragments from Fc
`fragments. F(ab)2 fragments were purified by size exclusion, followed
`by cation exchange. Final fractions were analyzed on a Bioanalyzer
`(2100) protein electrophoresis chip (Agilent Technologies, Santa Clara,
`CA; Figure W1).
`CVX-2000, F(ab)2, and Fab constructs were conjugated with a
`monomeric or homodimeric FGFR4-targeting peptide through an
`azetidinone linker on the peptide, which reacts specifically with a lysine
`in the Fab arm [23]. The original FGFR4-targeting peptide was dis-
`covered by phage display and synthesized as described previously
`[20]. Nontargeted controls of the scaffolds were conjugated with a
`nonbinding peptide.
`
`Direct binding ELISA. High-binding half-well 96-well plates
`were coated overnight at 4°C with recombinant human FGFR4-Fc
`(R&D Systems, Minneapolis, MN) or CVX-2000 anti-idiotype anti-
`body for total scaffold measurements. After washing and blocking,
`titrated compounds were then added to the plates and incubated
`at room temperature for 1 hour, followed by incubation with goat
`anti-human IgG-HRP (Jackson ImmunoResearch Laboratories, West
`Grove, PA) at room temperature for 1 hour. Tetramethylbenzidine
`substrate solution (KPL, Gaithersburg, MD) was added, and OD450
`was measured. Half maximal effective concentration (EC50) value was
`obtained from the dose-response curve from the experiment.
`
`Competitive ELISA. High-binding half-well 96-well plates were
`coated with goat anti-human IgG-Fc (Bethyl Laboratories, Montgomery,
`TX) at 4°C overnight. After washing and blocking, plates were incu-
`bated with recombinant human FGFR4-Fc for 1 hour at room tem-
`perature. Plates were washed, and titrated compounds were added in
`the presence of 50 ng/ml recombinant human FGF19 (R&D Systems)
`and 1 μg/ml heparan sulfate (Seikagaku/Amsbio, Lake Forest, CA) and
`incubated for 2 hours at room temperature. The bound compounds
`were detected by biotinylated anti-FGF19 antibody (R&D Systems),
`followed by incubation with streptavidin-HRP (Fitzgerald Industries,
`Acton, MA). OD450 was measured.
`
`Surface plasmon resonance binding analysis.
`Surface plasmon
`resonance (SPR) binding analyses of anti-FGFR4 compounds were
`performed on ProteOn XPR36 instrument (BioRad, Hercules, CA)
`at 25°C. For kinetic analysis, recombinant human FGFR4-Fc protein
`(R&D Systems) was immobilized on parallel surfaces of a GLM chip
`(BioRad) by amine coupling according to the manufacturer’s protocol.
`Running buffer was phosphate-buffered saline (PBS) with 300 mM
`sodium chloride and 0.05% (vol/vol) Tween 20. FGFR4 immobili-
`zation level was 990 RU for Fab binding and 640 RU for F(ab)2 and
`IgG binding. Compounds were tested for binding to FGFR4 starting
`at 200 nM at a flow rate of 50 μl/min. Association was monitored for
`180 seconds, and dissociation was monitored for 600 seconds. Chip
`was regenerated with 0.85% (vol/vol) phosphoric acid in water. Data
`were double-referenced to blank chip surface and buffer injection and
`fitted to 1:1 binding model with local Rmax using ProteOn Manager
`software (BioRad) to determine kinetic rate constants and K D. Kinetic
`constants are averaged from three independent experiments.
`For evaluation of monomer and homodimer peptide–conjugated
`compounds by capture of the compounds, anti-idiotype monoclonal
`antibody for CVX-2000 was immobilized across all surfaces of a
`
`2 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`564
`
`Biodistribution of Antibody Scaffolds Muchekehu et al.
`
`Translational Oncology Vol. 6, No. 5, 2013
`
`GLM chip (BioRad) to 10,000 RU. Running buffer was PBS with
`0.01% (vol/vol) Tween 20; 50 nM Fab, 10 nM F(ab)2, 10 nM IgG
`monomer and 80 nM Fab, 20 nM F(ab)2, 20 nM IgG homodimer-
`conjugated constructs were captured with anti-idiotype CVX-2000
`antibody. FGFR4-Fc (10 nM; R&D Systems) protein was tested
`for binding to anti-FGFR4 compounds captured on the chip. Data
`were double-referenced to chip surface and buffer blank.
`
`Huh-7 cell line. Huh-7 cells were obtained from Japan Health
`Science Research Resources Bank (Osaka, Japan; Cat. JCRB0403) and
`were in Dulbecco’s modified Eagle’s medium (Invitrogen, Carlsbad,
`CA) containing 10% FBS and maintained at 37°C and 5% CO2.
`
`Flow cytometry (FACS). Huh-7 cells were harvested with cell
`stripper and resuspended in FACS buffer (PBS + 10% FBS + 1%
`sodium azide). Cells were incubated with FGFR4-targeting scaffolds
`for 1 hour on ice. Cells were then washed and incubated with phy-
`coerythrin (PE)-labeled Goat Anti-Human IgG, F(ab′)2 Fragment
`Specific secondary antibody (Jackson ImmunoResearch Laboratories)
`for 30 minutes. Cell binding was measured using a flow cytometer
`(BD Biosciences, San Diego, CA). Data were analyzed using FloJo
`software (TreeStar Inc, Ashland, OR).
`
`PK in rodents.
`PK properties of constructs were assessed in 5-week-
`old male Swiss Webster mice weighing 18 to 20 g (Charles River
`Laboratories, Wilmington, MA). Compounds were administered
`intravenously (i.v.) at 10 mg/kg (n = 3), and blood samples were
`collected over a period of 5 days. Serum samples were prepared and
`analyzed for FGFR4 binding activity in binding ELISAs as described
`previously. Total scaffold levels were measured by binding to a CVX-
`2000 anti-idiotype antibody capture ELISA. Data were analyzed using
`WinNonlin software (Pharsight, St Louis, MO) to generate PK
`parameter estimates.
`
`Preparation of near-infrared–conjugated constructs.
`FGFR4
`targeting CovX-bodies (IgG) and F(ab)2 and Fab constructs were
`labeled with IRDye 800CW (LI-COR Biosciences, Lincoln, NE).
`In brief, constructs were buffer exchanged into 50 mM sodium phos-
`phate buffer (pH 7) and incubated with two equivalents of the dye to
`the antibody solution overnight at room temperature in the dark.
`Constructs were buffer exchanged several times in Amicon spin filters
`[50 kDa molecular weight cut off (MWCO)] to remove free dye.
`Dye-to-protein ratios were calculated according to the manufacturer’s
`instructions using the A780 and A280 measurements.
`
`In vivo xenograft studies. Xenografts were induced by subcuta-
`neous implantation of Huh-7 tumor cells into 5- to 7-week-old female
`nu/nu mice (18-20 g at start of experiment) and allowed to grow to a
`volume of 200 to 400 mm3 before initiation of treatment. Once tumors
`were established, mice were randomized to treatment groups on the
`basis of their tumor volumes for all in vivo studies described below.
`
`In vivo animal imaging. Near-infrared–conjugated compounds
`were administered at 3 and 10 mg/kg by intraperitoneal (i.p.) injection
`(no significant difference in tumor uptake was observed in a pilot study
`comparing i.p. vs i.v. injection). Mice were anesthetized with 5% iso-
`flurane for induction and maintained at 2% during image capture.
`The images were acquired at the indicated time points with an IVIS
`Lumina II Imaging System (PerkinElmer, Waltham, MA). A charge-
`
`coupled device (CCD) camera was used to collect the images. The
`images were analyzed using Living Image Software 4.0 (PerkinElmer).
`Regions of interest were quantified for mean pixel values.
`
`Biodistribution studies. Mice were dosed at 30 mg/kg i.p. injection,
`and tumors were harvested at maximal accumulation time points
`derived from imaging study (1 hour for Fab, 2 hours for F(ab)2, and
`8 hours for IgG). Tumors and normal tissue were harvested for
`biodistribution and histologic evaluation. For total scaffold accumula-
`tion, tissues were homogenized using FastPrep Lysing Matrix D Tubes
`(MP Biomedicals, Santa Ana, CA). Tissues were placed in tubes in a
`cell lysis buffer (Cell Signaling Technology, Danvers, MA) containing
`HALT protease and phosphatase inhibitor cocktail (Thermo Scientific).
`Tubes were then pulse homogenized using a FastPrep-24 instrument
`(MP Biomedicals) followed by incubation on a shaker at 4°C for an
`hour. Samples were then spun at 14,000 rpm for 10 minutes, and the
`supernatant was removed to a fresh tube. Samples were then applied
`directly to CVX-2000 anti-idiotype capture binding ELISA plates for
`total scaffold measurement.
`
`In vivo efficacy study. Mice were randomized into groups of
`10 mice per group. All compounds were administered once weekly at
`30 mg/kg by i.p. injection. Tumor volumes were measured once or
`twice weekly using calipers. Once the mean tumor volume of each
`treatment group exceeded 2000 mm3, mice were killed by CO2
`asphyxiation followed by cervical dislocation. Tumors and normal
`tissue were harvested for histologic evaluation.
`
`Immunohistochemistry
`Tumors from the biodistribution study described above were fixed
`in formalin for 24 hours. Tumors were embedded in paraffin blocks,
`sectioned, and mounted for immunohistochemistry. After deparaffi-
`nization and rehydration, heat-mediated antigen retrieval was per-
`formed using antigen retrieval buffer (Abcam, Cambridge, United
`Kingdom) for 30 minutes. Slides were incubated in 1% H2O2 for
`10 minutes followed by a blocking step for 30 minutes and primary
`antibody incubation. Blood vessels were detected using rabbit anti-
`CD31 antibody (Abcam; ab28364) overnight at 4°C. Sections were
`washed with PBS containing 0.01% Tween-20 and incubated with
`biotin anti-rabbit antibody (Jackson ImmunoResearch Laboratories)
`for 1 hour at room temperature followed by alkaline phosphatase–
`conjugated streptavidin ( Jackson ImmunoResearch Laboratories)
`for 1 hour. Sections were washed and incubated with Vector Red
`Alkaline Phosphatase Substrate Kit (Vector Labs, Burlingame, CA)
`for 20 minutes. For dual staining, sections were washed and incu-
`bated in HRP-conjugated donkey anti-human IgG secondary antibody
`(Jackson ImmunoResearch Laboratories) overnight at 4°C. Sections
`were washed and incubated in DAB peroxidase substrate kit (Vector
`Labs), dehydrated, and mounted. Images were captured using a Leica
`SCN400 slide scanner (Leica Biosystems, Wetzlar, Germany) at ×40,
`and images were analyzed on Leica SCN400 Image Viewer software.
`Tumor penetration was quantified using Image-Pro plus software
`(Media Cybernetics Inc, Rockville, MD). Five isolated vessels were
`randomly selected per slide with the distance (in μm) the scaffolds
`penetrated measured twice with each measurement taken on opposite
`sides. Statistical significance was determined using Prism (GraphPad
`Software, San Diego, CA). FGFR4 staining was measured using a rat
`anti–hFGF-R4 primary antibody (R&D Systems), followed by an
`HRP-goat anti-rat IgG secondary antibody (Jackson ImmunoResearch
`
`3 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`Translational Oncology Vol. 6, No. 5, 2013
`
`Biodistribution of Antibody Scaffolds Muchekehu et al.
`
`565
`
`Figure 1. Characterization of FGFR4 binding scaffolds. (A) IgG, F(ab)2, and Fab constructs bind specifically to FGFR4. (B) In an FGF19
`competition ELISA, all constructs compete with FGF19 to bind to FGFR4. (C) All constructs bind to Huh-7 cells. (D) PK curves of IgG, F(ab)2,
`and Fab following a single i.v. dose of 10 mg/kg in Swiss Webster mice. Both total and FGFR4 binding were measured as described in the
`Materials and Methods section.
`
`Laboratories) for 1 hour at room temperature. Sections were washed
`and incubated using a DAB peroxidase substrate kit (Vector Labs).
`
`Phospho-p44 Mitogen-Activated Protein Kinase (Extracellular
`Signal-Regulated Protein Kinase) Assay
`Phospho-extracellular signal-regulated protein kinase (Erk1/2) was
`measured using a PathScan Phospho-p44 mitogen-activated protein
`kinase (MAPK) Sandwich ELISA Kit (Cell Signaling Technology;
`Cat. No. 7315). Homogenized tumor lysates from our biodistribu-
`tion study were used according to the manufacturer’s instructions. In
`brief, 100 μl was placed in microwells and incubated for 2 hours at
`37°C. Wells were then washed four times with provided wash buffer
`and incubated with detection antibody for 1 hour at 37°C. Wells were
`washed and incubated in HRP-conjugated secondary antibody for
`30 minutes at 37°C. Wells were washed and incubated with tetra-
`methylbenzidine substrate for 10 minutes at 37°C, STOP solution
`was added, and OD450 was measured.
`
`Internalization Studies
`All scaffolds were biotinylated using EZ-Link Sulfo-NHS-LC-
`Biotin (Thermo Scientific) according to the manufacturer’s instruc-
`tions. Huh-7 cells were harvested using Cellstripper (Cellgro, Manassas,
`VA) and seeded at 100,000 cells per well in PBS4 (1 mM MgCl, 1 mM
`CaCl2, 0.2% BSA, 5 mM glucose, and 10% FBS). Compounds
`were added and cell incubated at 37°C for 0, 15, 30, 60, and 120 min-
`utes. Plates were placed on ice for 5 minutes to stop internalization
`and washed three times in PBS4. Cells were then incubated in Avidin
`(Sigma, St Louis, MO; at 100 μg/ml) at 4°C for 1 hour. Biocytin
`(Sigma; 1 mg/ml) was then added for 10 minutes. Cells were cen-
`
`trifuged, the supernatant was discarded, and cells were solubilized for
`30 minutes at 4°C. Total amount of internalized scaffold was measured
`in supernatant by CVX-2000 anti-idiotype capture binding ELISA
`followed by incubation with streptavidin-HRP (Fitzgerald Industries).
`
`Statistical Analyses
`Data were analyzed either by two-way analysis of variance (ANOVA)
`with a Bonferroni post-test or by a two-tailed t test using Prism (Graph-
`Pad Software).
`
`Results
`
`In Vitro Characterization of Scaffolds
`CVX-2000, F(ab)2, and Fab constructs were conjugated with
`monomer FGFR4-targeting peptide as described in the Materials
`and Methods section.
`Binding to FGFR4 was measured by ELISA and SPR. The IgG
`and F(ab)2 constructs bind with an apparent binding affinity of
`
`Table 1. Binding Affinity of IgG, F(ab)2, and Fab Determined by SPR.
`
`Fab monomer
`Fab homodimer
`F(ab)2 monomer
`F(ab)2 homodimer
`IgG monomer
`IgG homodimer
`
`kon (M−1 s−1)
`
`1.1 × 106
`9.4 × 105
`1.2 × 106
`1.3 × 106
`1.4 × 106
`3.6 × 106
`
`koff (s−1)
`9.7 × 10−3
`7.1 × 10−3
`2.9 × 10−3
`2.9 × 10−3
`2.4 × 10−3
`1.9 × 10−3
`
`K D (nM)
`
`8.5
`7.6
`2.5
`2.3
`1.8
`0.53
`
`FGFR4 capture to measure monovalent interactions.
`
`4 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`566
`
`Biodistribution of Antibody Scaffolds Muchekehu et al.
`
`Translational Oncology Vol. 6, No. 5, 2013
`
`Figure 2. Biodistribution studies. (A) Time-dependent tumor uptake of IgG, F(ab)2, and Fab. In vivo optical imaging of near infra-red (NIR)-
`conjugated constructs. Average signal intensities were quantified using regions of interest (ROIs) from the tumor sites. Data are presented
`as mean fold increase from initial image capture at 30 minutes ± SEM of eight mice (***P < .001, *P < .05; IgG vs both F(ab)2 and Fab
`accumulation, *P < .05 and **P < .01; F(ab)2 vs Fab accumulation by two-way ANOVA with Bonferroni post-test). Tumor and normal tissue
`uptake of (B) IgG 8 hours post dose (*P < .05), (C) F(ab)2 2 hours post dose, and (D) Fab 1 hour post dose (***P < .001, **P < .01 by two-
`way ANOVA with Bonferroni post-test). (E) IgG, F(ab)2, and Fab tumor uptakes and serum levels compared at the early time points of 1 hour,
`2 hours, and 1 hour, respectively. At this early time point with equivalent serum levels, the targeted Fab shows maximal accumulation
`levels compared to the F(ab)2 and IgG (***P < .001, **P < .01 by one-way ANOVA with Bonferroni post-test). (F) Tumor to serum levels
`further demonstrate that the Fab construct accumulation is significantly higher than the IgG accumulation (*P < .05 by two-way ANOVA
`with Bonferroni post-test).
`
`0.7 and 0.8 nM, respectively, whereas the Fab binds with a binding
`affinity of 11 nM (Figure 1A and Table 1). In a competition ELISA,
`the IgG and F(ab)2 constructs compete for binding of FGF19 to
`FGFR4 with an half-maximal inhibitory concentration (IC50) of
`9 and 18 nM, respectively, whereas the Fab competes with an IC50
`of 500 nM (Figure 1B).
`Binding to Huh-7 cells, a hepatocellular carcinoma cell line that
`expresses high levels of FGFR4 [24], was measured by FACS. The IgG
`
`and F(ab)2 constructs have a cell binding affinity of 9 and 18 nM, respec-
`tively, whereas the Fab has a binding affinity of 198 nM (Figure 1C ).
`
`Removal of the Fc and Reduction in Size Significantly Impact
`the PK Properties
`PK studies were conducted by administering a single i.v. injection of
`10 mg/kg of all compounds. The serum concentrations over time were
`measured using an FGFR4 and CVX-2000 anti-idiotype capture binding
`
`5 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`Translational Oncology Vol. 6, No. 5, 2013
`
`Biodistribution of Antibody Scaffolds Muchekehu et al.
`
`567
`
`ELISA for total scaffold measurements (Figure 1D). The intact FGFR4
`targeting CovX-body has a β half-life in mice of 60 hours, compared to
`β half-lives of 4 to 6 and 1 to 3 hours for F(ab)2 and Fab, respectively.
`
`Time-Dependent Tumor Uptake
`The constructs were labeled with a near-infrared dye IRDye
`800CW to allow in vivo imaging of tumor penetration and retention
`in the Huh-7 hepatocellular carcinoma xenograft model. Compounds
`were administered at 3 mg/kg with a single i.p. injection, and tumor
`uptake was measured by image capture at the time points indicated
`(Figure 2A). Maximal accumulation of the IgG was seen after 8 hours
`(2.51 ± 0.47 fold increase in fluorescence relative to initial image
`captured at 30 minutes, n = 8; a representative image of IgG tumor
`uptake is shown in Figure W2), where accumulation was significantly
`higher than the Fab and F(ab)2 constructs. The maximal accumula-
`tion for the F(ab)2 construct was from 2 to 4 hours (1.63 ± 0.13 fold
`increase in fluorescence, n = 8), and at these time points, the F(ab)2
`accumulation was significantly higher than the Fab. Minimal accu-
`mulation of the Fab was observed after 1 hour in this imaging study
`(0.99 ± 0.08 fold increase in fluorescence, n = 8). In the more quan-
`titative biodistribution study, where the tumors were homogenized
`and accumulation was measured by ELISA, accumulation of the
`
`Fab was observed after 1 hour (Figure 2D). This accumulation was
`also confirmed by immunohistochemistry (Figure 4F ).
`
`Tumor and Normal Tissue Uptake of Scaffolds
`A biodistribution study was performed to determine the tumor
`and normal tissue uptake of the FGFR4-targeted constructs and their
`nontargeted controls. Animals were dosed at 30 mg/kg to quantify
`the tumor and normal tissue uptake. Tumor and normal tissue were
`harvested at the maximum accumulation time point derived from
`the previous imaging study: 8 hours post dose for the IgG (Figure 3B),
`2 hours for the F(ab)2 constructs (Figure 3C ), and 1 hour for the Fab
`constructs (Figure 3D). Total scaffold accumulation was quantified
`by a CVX-2000 anti-idiotype capture binding ELISA. After 8 hours,
`accumulation of the nontargeted IgG is significantly higher than the
`targeted IgG (5.80 ± 0.92% vs 2.3 ± 0.99% ID, respectively; P < .05,
`n = 5). After 2 hours, there is a similar trend for the nontargeted
`F(ab)2 vs targeted F(ab)2 accumulation; however, this is not statis-
`tically significant (1.62 ± 0.58% vs 0.48 ± 0.24% ID, respectively;
`P > .05, n = 5). There was no significant difference between the
`targeted and nontargeted Fab after an hour (0.46 ± 0.07% vs 0.26 ±
`0.06% ID, respectively; P > .05, n = 5). Both targeted and nontargeted
`constructs accumulate in tumors because of the enhanced permeability
`
`Figure 3. Increasing valency increases cell binding of bivalent constructs. (A–C) Monomer and homodimer peptide–conjugated con-
`structs bind recombinant FGFR4 in a similar manner on an FGFR4 binding ELISA. Huh-7 cell binding of IgG (D) and F(ab)2 (E) homodimer-
`conjugated constructs demonstrates a boost in cell binding affinity, whereas the Fab monomer and homodimer (F) have similar cell binding
`affinities. (G) Anti-idiotype capture levels of anti-FGFR4 monomer and homodimer compounds to measure multivalent interactions. Increased
`levels of FGFR4 binding is seen with the IgG homodimer. (H) Tumor and normal tissue uptake of homodimer peptide–conjugated IgG, F(ab)2,
`and Fab (*P < .05 vs monomer IgG).
`
`6 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`568
`
`Biodistribution of Antibody Scaffolds Muchekehu et al.
`
`Translational Oncology Vol. 6, No. 5, 2013
`
`Figure 4. Increased avidity decreases penetration of scaffolds into the tumor. (A) FGFR4 staining in an adjacent section is shown. Dual
`staining for blood vessels (brick red) and human IgG (brown) is shown in (B) PBS-dosed animals, (C) nontargeted IgG (8 hours), monomer
`peptide–conjugated (D) IgG (8 hours), (E) F(ab)2 (2 hours), (F) Fab (1 hour) and homodimer peptide–conjugated (G) IgG (8 hours), (H) F(ab)2
`(2 hours) and (I) Fab (1 hour). Arrows indicate blood vessels (red). Perivascular or diffuse construct staining from those points can be
`seen. (J) Plot of average distance from randomly selected blood vessels (mean ± SEM; *P < .05, ***P < .001 by one-way ANOVA with
`Bonferroni post-test).
`
`7 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`Translational Oncology Vol. 6, No. 5, 2013
`
`Biodistribution of Antibody Scaffolds Muchekehu et al.
`
`569
`
`and retention effect; however, the differences in levels of the non-
`targeted versus targeted IgG and F(ab)2 constructs may be due to inter-
`nalization and “consumption” of the targeted constructs as shown in
`previous studies [25].
`High levels of the targeted Fab accumulated in both the lung and
`spleen (4.42 ± 2.03% and 6.20 ± 1.00% ID, respectively; Figure 2D).
`Both the lung and spleen normally express FGFR4 [26,27], and the
`accumulation of the targeted Fab suggests that the FGFR4 binding
`peptide binds both human and mouse FGFR4. It also demonstrates
`the ability of the smaller Fab to rapidly accumulate in normal tissue.
`Tumors were also harvested after 1 hour post dose with the IgG
`for an early time point comparison with the Fab and F(ab)2 con-
`structs (Figure W3). At this early time point, serum levels of the
`three constructs were comparable, allowing a comparison of tumor
`levels (Figure 2E ). Tumor levels as a percent of serum show that
`the targeted Fab and nontargeted Fab constructs (5.81 ± 1.51%
`and 3.30 ± 1.29% serum, respectively) accumulate at a higher level
`than the targeted and nontargeted IgG (0.86 ± 0.27% and 0.28 ±
`0.08% serum, respectively) and targeted F(ab)2 constructs (1.43 ±
`0.42% serum), whereas the nontargeted F(ab)2 accumulation was
`not significantly different at this early time point (3.95 ± 1.29%
`serum; P > .05, n = 5; Figure 2F).
`
`The Role of Increased Valency in Tumor Targeting
`Increasing the number of targeting peptides from two to four on
`the IgG and F(ab)2 constructs and from one to two on the Fab did
`not increase the apparent binding affinity to recombinant FGFR4 in
`an FGFR4 capture binding ELISA (Figure 3, A–C) or SPR (K D =
`
`0.5, 2.3, and 7.6 nM, respectively; Table 1). It did however increase
`the binding of the bivalent constructs to Huh-7 cells by 2 logs (Fig-
`ure 3, D and E) but not of the Fab (Figure 3F).
`In an SPR assay where the compounds (10 nM) were captured on
`the binding surface by the anti-idiotype CVX-2000 antibody, in-
`creased levels of FGFR4 binding could be measured on the tetra-
`valent constructs, IgG (Figure 3G ) and F(ab)2 (data not shown)
`homodimer-conjugated constructs. These bind twice the amount
`of FGFR4 as the rest of the compounds, as would be expected as they
`each display four peptides. Comparable levels of monomer and
`homodimer constructs were captured with anti-idiotype CVX-2000
`(Figure W4). This correlates with the increased cell binding we see
`with these constructs in the FACS assay (Figure 3, D and E ).
`Tumor and normal tissue accumulation of the homodimer-
`conjugated peptides showed that increasing the valency of the IgG,
`F(ab)2, and Fab did not increase the tumor levels versus their monomer
`peptide–conjugated constructs (homodimer tumor accumulation of
`0.82 ± 0.08%, 0.55 ± 0.12%, and 0.32 ± 0.06% ID, respectively,
`n = 5; Figure 3G).
`
`The Role of Size and Valency in Tumor Penetration
`High and evenly distributed levels of FGFR4 are seen in Huh-7
`tumors (Figure 4A). Adjacent FGFR4 staining of all harvested tumors
`shown in Figure 4 all showed a similar high even distribution of
`FGFR4. A representative section in penetration of the nontargeted
`IgG construct in the tumor appears to be nonrestricted, and homog-
`enous antibody staining is seen throughout the tumor (99.4 ± 10.4 μm
`penetration; Figure 4C).
`
`Figure 5. Increased avidity leads to superior efficacy. (A) In vivo xenograft study (i.p. dosing, 30 mg/kg once weekly) and (B) in vitro cell
`proliferation assay (12 nM compounds present in media for whole experiment). *P < .05, ***P < .001 by two-way ANOVA with Bonferroni
`post-test. Arrows indicate dosing. (C) Phospho-Erk levels measured in tumors harvested at the end of the efficacy study; *P < .05 versus
`PBS-dosed tumors by two-tailed Student’s t test. Dual staining of blood vessels and human IgG of the tumors at the end of the efficacy
`study; (D) monomer peptide IgG dosed and (E) homodimer peptide IgG dosed.
`
`8 of 12
`
`OnCusp
`Ex. 1033
`
`

`

`570
`
`Biodistrib