Pretargeted delivery of PEG-coated drug carriers to breast tumors using multivalent, bispecific antibody against polyethylene glycol and HER2

Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

nanomedjournal.com

Pretargeted delivery of PEG-coated drug carriers to breast tumors using multivalent, bispecific antibody against polyethylene glycol and HER2 Christina L. Parker, PhD a , Morgan D. McSweeney, BS a , Andrew T. Lucas, PharmD b, c, d , Timothy M. Jacobs, PhD a , Daniel Wadsworth a , William C. Zamboni, PharmD, PhD b, c, d , Samuel K. Lai, PhD a, e, f,⁎ a

Division of Pharmacoengineering and Molecular Pharmaceutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, United States Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, United States c UNC Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, United States d Carolina Center for Nanotechnology Excellence, University of North Carolina at Chapel Hill, United States e Department of Biomedical Engineering, University of North Carolina at Chapel Hill, United States f Department of Microbiology and Immunology, University of North Carolina at Chapel Hill, United States Revised 18 July 2019

b

Pretargeting is an increasingly explored strategy to improve nanoparticle targeting, in which pretargeting molecules that bind both selected epitopes on target cells and nanocarriers are first administered, followed by the drug-loaded nanocarriers. Bispecific antibodies (bsAb) represent a promising class of pretargeting molecules, but how different bsAb formats may impact the efficiency of pretargeting remains poorly understood, in particular Fab valency and Fc receptor (FcR)-binding of bsAb. We found the tetravalent bsAb markedly enhanced PEGylated nanoparticle binding to target HER2 + cells relative to the bivalent bsAb in vitro. Pretargeting with tetravalent bsAb with abrogated FcR binding increased tumor accumulation of PEGylated liposomal doxorubicin (PLD) 3-fold compared to passively targeted PLD alone, and 5-fold vs pretargeting with tetravalent bsAb with normal FcR binding in vivo. Our work demonstrates that multivalency and elimination of FcRn recycling are both important features of pretargeting molecules, and further supports pretargeting as a promising nanoparticle delivery strategy. Published by Elsevier Inc. Key words: Pretargeting; Bispecific; Drug delivery; Polyethylene glycol

Conflicts of Interest: The authors declare no potential conflicts of interest. Funding: This work was supported by the National Science Foundation Graduate Research Fellowship Program (DGE-1144081, C. L.P; DGE-1650116, M.D.M.), GlaxoSmithKline Fellowship from UNC Eshelman School of Pharmacy Foundation (C.L.P.), David and Lucile Packard Foundation (2013-39247, S.K.L.), UNC Research Opportunities Initiative Grant in Pharmacoengineering (S.K.L.), Eshelman Institute for Innovation (S.K.L), and startup funds from the Eshelman School of Pharmacy and Lineberger Comprehensive Cancer Center (S.K.L.). Flow cytometry was performed at the Flow Cytometry Core Facility, which is supported in part by P30 CA016086 Cancer Center Core Support Grant to the UNC Lineberger Comprehensive Cancer Center. Research reported in this publication was supported by the North Carolina Biotech Center Institutional Support Grant 2015-IDG-1001. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies. Disclosures ⁎Corresponding author at: Division of Pharmacoengineering and Molecular Pharmaceutics, University of North Carolina at Chapel Hill. E-mail addresses: [email protected]. URL: https://pharmacy.unc.edu/lai-research-group/ (S.K. Lai). https://doi.org/10.1016/j.nano.2019.102076 1549-9634/Published by Elsevier Inc.

In cancer therapy, targeted drug delivery aims to maximize the dose of anti-cancer or imaging agents delivered to cancer cells/tissues while minimizing exposure and toxicity to healthy, non-targeted tissues. One broadly studied approach, termed passive targeting, is to encapsulate anti-cancer agents into nanocarriers coated with “stealth” polymers, such as liposomes, micelles, and polymeric nanoparticles, that can in turn accumulate in tumors due to tumors' inherent leaky vasculature (i.e. the Enhanced Permeability and Retention (EPR) effect). 1–3 Surface modification of nanocarriers with polymer polyethylene glycol (PEG) is a commonly used strategy for formulating longcirculating nanocarriers by effectively reducing nonspecific protein binding and cell clearance, and enhancing nanoparticle uptake via EPR. 4–7 Unfortunately, coating polymers that minimize opsonin absorption also limit binding and internalization into target cells. To increase particle uptake, antibodies and other moieties that exploit the differential expression of select surface receptors on cancer cells compared to healthy cells are often attached to the nanocarriers' surface (i.e. active targeting) with the expectation that the actively targeted carriers would

2

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

more effectively deliver cargo therapeutics to target cells. 8–11 Interestingly, numerous studies have found that increasing antibody grafting beyond certain thresholds actually reduced overall targeting efficiency in vivo 12–17, presumably because a high density of conjugated ligands compromised the anti-fouling nature of “stealth” PEG coatings. This leads to premature elimination of the carriers from circulation by the mononuclear phagocyte system (MPS), 8,12,18,19 and results in a smaller fraction of the administered nanocarriers that could extravasate and accumulate at target site. This led a number of studies to conclude that actively targeted nanoparticles did not improve particle delivery to tumors compared to passively targeted nanoparticles. 20–22 To overcome the aforementioned limitations with both activeand passive targeting, some researchers are exploring “pretargeting” as a strategy to preserve prolonged circulation of coated nanoparticles while simultaneously enabling active targeting to specific cells. 23 Pretargeting is a two-step strategy that relies on the use of bispecific pretargeting molecules that can bind both cellular epitopes and subsequently administered effector molecules. Ideal pretargeting molecules would extravasate from systemic circulation and accumulate on the surface of target cells, or be quickly eliminated from the circulation. Drug-loaded carriers are then administered, and a fraction of the extravasated carriers would be captured by cell-bound pretargeting molecules, followed by endocytosis into target cells. This approach was initially tested for the treatment of hematological malignancies in the form of pretargeted radioimmunotherapy, which improved imaging contrast and tumor suppression as well as reduced radioactivity in healthy organs. 24 Later studies have extended the use of bispecific proteins to pretarget nanocarriers to specific cell populations. 25–28 While bsAb are commonly designed in Ig-like formats with a Fc domain, 29,30 other pretargeting molecules are designed as bsAb fragments lacking a Fc domain. 25,27,31 To date, no studies have compared how the design of the bsAb format may impact the pretargeting efficiency of nanocarriers. A longstanding challenge in bsAb engineering has been the proper pairing of heavy and light chains leading to high purity and yield of the final product. Here, we used a recently developed bsAb platform called OrthoMab to investigate the optimal bsAb design for pretargeting molecules. By introducing orthogonal mutation pairs into heavy and light chains, the OrthoMab platform yields high fidelity pairing of the correct heavy and light chains for functional bsAb. 32 We designed two bispecific pretargeting molecules (tandem Fab and Fab-IgG1) that recognize both HER2 receptors overexpressed on breast cancer cells, and PEG present on PEGylated liposomal doxorubicin (PLD) and PEGylated polystyrene beads. By utilizing Fab that specifically recognize and bind PEG, we need not modify the PEGylated therapeutic or carrier, which we expect will maximize the fraction of PEGylated carriers that can circulate and extravasate at target tissues. While the bivalent tandem Fab (two Fab domains connected by a flexible linker) has one binding domain per antigen, the tetravalent Fab-IgG1 (additional Fabs are covalently linked to the Fab domains of a traditional IgG molecule) has two binding domains per antigen (Figure 1, A). This allows us to evaluate the effect of Fab valency (number of binding domains per antigen) and impact of FcR-

binding on targeting and distribution of pretargeted PEG nanocarriers to HER2 + breast cancer cells both in vitro and in an orthotopic xenograft HER2 + breast tumor model in mice.

Methods Cell lines and animals Human SKBR3 (HER2 +), A2780 (HER2 −), and BT474 (HER2 +) were purchased from the UNC-CH Tissue Culture Facility. SKBR3 cells were cultured in McCoy's medium containing 15% fetal bovine serum, and A2780 cells were cultured in RPMI 1640 containing 10% fetal bovine serum and 1% L-glutamine. BT474 cells were maintained in RPMI 1640 media with 2 g/L sodium bicarbonate and 2 mM L-glutamine, and supplemented with 10% fetal bovine serum, 4.5 g/L glucose, 10 mM HEPES, 1.0 mM sodium pyruvate, and 0.01 mg/ml human insulin. All cells were maintained at 37 °C and 5% CO2. Female athymic nude (6–8 week old) mice were obtained from Charles River Laboratories (Wilmington, MA, USA) or bred in-house by UNC Animal Services Core (Chapel Hill, NC), and maintained in a sterile housing suite. All animal experiments were carried out in accordance with an animal use protocol (16–333.0) approved by the University of North Carolina Animal Care and Use Committee. Mice were randomly assigned to treatment groups and investigators were blinded to the treatments. Cell uptake assay Cells were seeded at 5 × 10 4 cells/well into 96-well plates. Next day, the cells were incubated with 10 nM mAb controls or bsAb for 4 h at 37 °C. After washing to remove unbound Ab, the cells were then incubated with fluorescent, PEGylated polystyrene beads (1:10 4 cell:bead ratio) for 12 h at 37 °C. Cells were washed to remove unbound beads and flow cytometry was performed using iQue Screener PLUS (Intellicyt, Albuquerque, NM). Data were analyzed using ForeCyt and BD FACSDiva software. To determine if bsAb remained on the surface of cells 24 h after bsAb incubation, cells were treated using extended uptake conditions. After cells were treated with 10 nM mAb or bsAb for 4 h at 37 °C, cells were washed and incubated in fresh media for 24 h at 37 °C. Then, cells were washed and incubated with fluorescent, PEGylated polystyrene beads (1:10 4 cell:bead ratio) for 4 h at 37 °C. Unbound beads were removed through washing, and cell-associated fluorescence was analyzed via flow cytometry. Pharmacokinetics of bispecific antibody in the presence and absence of high dose IVIg Female athymic nude mice (6–8 weeks old) either received a single intravenous injection of 30 μg bispecific Fab-IgG1 or two intravenous injections separated by 15 minutes of 30 μg bispecific Fab-IgG1 and 30 mg human intravenous immune globulin (IVIg, Provigen) via tail vein. Blood was collected from mice at different time points (5 min, 1 h, 3 h, 5 h, 7 h, 24 h, 48 h, 72 h; n = 8 mice per treatment group, n = 4 mice per time point).

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

3

Figure 1.. Characterization of monospecific and bispecific antibodies (Ab). A) Schematic illustrating differences in size and number of antigen-binding domains for each Ab. The theoretical molecular weight of bispecific Fab-IgG1 and tandem Fab are ~250 kDa and ~100 kDa, respectively. B) Nonreducing (left) and reducing (right) protein gel showing Coomassie blue staining of IgG1PEG, IgG1HER2, tandem Fab HER2xPEG, Fab-IgG1HER2xPEG, and non-specific IVIg. C) Binding affinity of IgG1HER2, tandem Fab HER2xPEG, and Fab-IgG1HER2xPEG to HER2-Fc chimera analyzed by ELISA (n = 2). D) Binding affinity of IgG1PEG (left), tandem Fab HER2xPEG (middle), and Fab-IgG1HER2xPEG (right) to DSPE-PEG5k in the presence and absence of free PEG8K competition analyzed by ELISA (n = 2).

Whole blood was stored undisturbed at room temperature for 20 min to allow clotting. Samples were then centrifuged at 2000×g in a refrigerated centrifuge for 15 min to isolate serum. PEGspecific ELISAs 33 were used to quantify the serum concentration of bispecific Fab-IgG1HER2xPEG at various time points by detecting antibody with goat anti-human IgG F(ab)’2 (Rockland Immunochemicals, cat no. 209–1304, 1:10,000 dilution). PK analysis of the blood concentration of bsAb was conducted using PKSolver with a two-compartment model. 34 Biodistribution of pretargeted PLD in tumor-bearing mice Female athymic nude (nu/nu) mice received a subcutaneous implantation of a single 60-day, 0.36 mg 17β-estradiol pellet six days prior to left mammary pad injection of BT474 cells; estradiol supplement was used to support growth of HER2 +

BT474 cells. BT474 cells were resuspended in a 1:1 Matrigel/ PBS solution to a final concentration of 4 × 10 7 cells/ml. Each mouse received a 100 μl injection of BT474 (4 × 10 6) cells into the left mammary pad. Once tumors were ≥ 100 mm 3, mice were randomized into antibody treatment groups (n = 4 mice per group). PBS, bsAb (30 μg Fab-IgG1), and bsAb + IVIg (30 μg Fab-IgG1 with 30 mg IVIg) were administered i.v., followed by generic PEGylated liposomal doxorubicin (PLD, 3 mg/kg i.v.) 24 h after Ab dose. Due to limitations in concentrating bsAb, the bsAb + IVIg dose was separated into two injections separated by 4 h (1st injection: 30 μg Fab-IgG1 + 10 mg IVIg; 2nd injection: 20 mg IVIg). Forty-eight hours after PLD dose, mice were sacrificed, and tissues (heart, liver, kidneys, spleen, lungs, tumor) and blood via cardiac puncture were collected. Total doxorubicin concentration in serum and tissue homogenate was quantified using a validated

4

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

LC–MS/MS assay (LLOQ = 5 ng/mL; LLOD = 1 ng/mL). Generic PLD (Sun Pharmaceutical Industries Ltd.), comparable to Doxil® liposome, was gifted by the Zamboni lab at UNC-CH and purchased through UNC Shared Services Center Pharmacy. Statistical analysis All data are presented as mean ± SD unless specified as mean ± SEM. Graphs and statistical tests were performed using GraphPad Prism 8 software. Group comparisons were analyzed using two-way ANOVA and post hoc multiple comparisons Tukey's test unless specified as one-way ANOVA with post-hoc Tukey's. A P-value b0.05 was considered to indicate statistical significance.

Results OrthoMab platform preserves antigen binding affinity We engineered monoclonal IgG1PEG, IgG1HER2, as well as tetravalent bsAb “Fab-IgG1” and bivalent “tandem Fab” formats by merging human IgG1 backbones with HER2- and PEGbinding VH and VL domains previously isolated from mouse IgG (Figure 1, A). 35,36 Purified Ab from the culture supernatant displayed the expected molecular weights as visualized on nonreduced and reduced 4–12% bis-tris protein gels (Thermo Fisher Scientific) (Figure 1, A and B). We next evaluated specific binding to both HER2 and PEG using antigen-specific ELISA assays. Both bsAb possessed similar HER2-binding affinities to the monoclonal anti-HER2 IgG1: the KD for IgG1HER2, tandem Fab HER2xPEG, and Fab-IgG1HER2xPEG against HER2 proteins were 0.76 ± 0.11 nM, 2.76 ± 0.18 nM, and 1.20 ± 0.13 nM, respectively (Figure 1, C). In contrast, Ab with bivalent Fabs possessed comparable binding affinity for PEG (KD for IgG1PEG and Fab-IgG1 against PEG were 4.24 ± 0.48 nM and 4.16 ± 0.60 nM respectively), while tandem Fab HER2xPEG, with one PEG-binding Fab, had markedly weaker affinity with KD ~ 1166 ± 182.4 nM. All three mAb constructs bound PEG specifically, as incubation with excess free PEG8K completely eliminated their binding signal (Figure 1, D). Altogether, these results confirmed that we were able to produce functional bsAb, and that the orthogonal mutations introduced at the heavy and light chain interface did not impair the binding to either HER2 or PEG compared to their respective parent monospecific IgGs. Pretargeted delivery of PEGylated nanocarriers in vitro We next assessed the pretargeting efficiencies of the different bsAb by measuring the cellular association of fluorescent PEG beads in both SKBR3 (HER2 +) and A2780 (HER2 −) cells pretargeted with tetravalent Fab-IgG1, bivalent tandem Fab, or combination of the parent mAbs (IgG1HER2 and IgG1PEG). We observed minimal fluorescence in A2780 cells across all conditions, irrespective of the specific bsAb or mAb used (Figure 2). The mean fluorescence intensity (MFI) of PEG beads pretargeted with Fab-IgG1 was 25-fold higher in HER2 + SKBR3 cells compared to both monoclonal IgG controls and tandem Fab (P b 0.0001), and also to A2780 cells pretargeted with the FabIgG1 (Figure 2, C). Similarly, the percentage of GFP positive

cells for PEG beads pretargeted with Fab-IgG1 was 20-fold higher in HER2 + SKBR3 cells compared to all other conditions (P b 0.0001; Figure 2, B). These results not only validated the specificity of Fab-IgG1 to both HER2 and PEG, but also underscored that pretargeting effectiveness can be influenced by the Fab valency on the bsAb. The results also confirmed that pretargeting molecules must be bispecific, as defined by a covalent linkage between anti-HER2 and anti-PEG binding domains, in order to enhance nanoparticle delivery to target cells. To minimize the fraction of free pretargeting molecules in the circulation at the time of nanoparticle dosing, we must afford sufficient time for them to be eliminated by natural renal clearance. We thus tested next the pretargeting effectiveness following an extended lag time (ie. 24 h) between Ab and PEG bead incubations compared to the 4 h duration in the above study to determine if Fab-IgG1 could remain on target cell surface over the entire duration. We observed a ~2-fold reduction in MFI of cells pretargeted with Fab-IgG1 under the extended uptake (4 h incubation, wash, wait) condition compared to the standard uptake condition (4 h incubation only; P b 0.0001; Supplemental Figure 1, B). However, the fraction of cells interacting with nanoparticles remained largely unchanged between the two time points (N80%), and was markedly higher than all other tested conditions (P b 0.0001; Supplemental Figure 1, A). These results suggest that Fab-IgG1 can facilitate effective pretargeting even when introduced up to 24 h in advance of nanoparticle dosing. Due to its superior pretargeting efficiency in vitro, we advanced the Fab-IgG1 for mouse studies. Multivalent Fab-IgG1 circulation kinetics was reduced in the presence of high dose IVIg We evaluated the pharmacokinetics (PK) of Fab-IgG1 following intravenous administration, and found that the terminal half-life of Fab-IgG1 was ~22 hours (Figure 3, A), most likely due to the Fc domain of the Fab-IgG1 engaging neonatal Fc receptors (FcRn) resulting in their recycling and efficient retention in the circulation. To validate the role of FcRn recycling on Fab-IgG1 PK, we administered a high dose of human intravenous immune globulin (i.e. 30 mg IVIg) prior to dosing with Fab-IgG1, given the effectiveness of IVIg replacement therapy in accelerating catabolism of autoantibodies in patients with autoimmune diseases. 37,38 We observed a 3-fold reduction in serum half-life of Fab-IgG1 when administered with high dose IVIg (t1/2 ~7.2 h, Figure 3, B), implicating the role of FcRn in the prolonged circulation of Fab-IgG1. Biodistribution and tumor accumulation of pretargeted PLD in tumor-bearing mouse model Next, we evaluated the pretargeting effectiveness of Fab-IgG1 in an orthotopic breast cancer mouse model with estradiol supplementation to better represent the tumor physiology and stromal microenvironment of human breast cancer. Compared to mice receiving PEGylated liposomal doxorubicin (PLD) (i.e. passive targeting only), pretargeting with Fab-IgG1 lacking FcRn recycling increased the concentration of doxorubicin in tumors by 3-fold (P = 0.0124, Figure 4, B). Pretargeting efficiency appears to be critically dependent on minimizing the serum

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

5

Figure 2.. Pretargeted delivery of PEGylated nanoparticles to HER2 + vs HER2 − cells. A) Distribution of fluorescence in HER2 − and HER2 + cells upon incubation with monospecific and bispecific Ab followed by 100 nm fluorescent PEGylated polystyrene beads measured by flow cytometry. B) Percentage of GFP positive cells and (C) mean fluorescence intensity (MFI) of cell-associated fluorescent PEG beads. The data represents n ≥ 2 independent experiments performed with ten replicates. **** indicates P b 0.0001 vs bispecific Fab-IgG1 + PEG beads incubated on SKBR3 cells.

levels of Fab-IgG1 at the time of PLD dosing: there was ~5-fold greater doxorubicin in tumors of mice where Fc-FcRn binding was blocked with excess IVIg compared to tumors with normal Fc-FcRn binding (P = 0.0035, Figure 4, B and D). At the time of PLD dose, we detected a higher concentration of Fab-IgG1 molecules in the serum of mice treated with Fab-IgG with normal Fc-FcRn binding (without IVIg) than mice treated with PLD alone or Fab-IgG1 with saturated Fc-FcRn binding (with IVIg) (Supplemental Figure 2). Thus, having residual pretargeting molecules present in the circulation at the time of nanoparticle dosing actually led to less effective nanoparticle delivery to tumors compared to passive targeting alone. Across treatment groups, there were no statistical differences in the serum doxorubicin concentration 48 h post-PLD dose (Figure 4, A, Table 1). There were also no statistical differences in the concentration of doxorubicin in non-targeted organs (liver, spleen, heart, and lungs) across treatment groups (Figure 4, C and E). This is particularly important for the accumulation of

doxorubicin in the heart because it is a major site of doxorubicin toxicity. 39 The lack of increased liver accumulation means that the ratio of doxorubicin concentration in the tumor to liver was ~3-fold higher in mice treated with a pretargeting dose of bsAb + IVIg compared to mice treated with either PLD alone or pretargeting dose of bsAb (Figure 4, F). This effect was also observed in the ratio of doxorubicin concentration in the tumor to spleen, tumor to heart, and tumor to lungs (Supplemental Figure 3, A–D). Similarly, we observed ~3-fold increase in the ratio of doxorubicin concentration in the tumor to serum for mice treated with a pretargeting dose of bsAb + IVIg versus mice treated with either PLD alone or pretargeting dose of bsAb (Supplemental Figure 3, E). However, we did not observe any appreciable differences in the ratio of doxorubicin concentration in liver to serum or spleen to serum across treatment groups (Supplemental Figure 3, F–G), suggesting that the effects of the pretargeting dose of Fab-IgG1 + IVIg are primarily in the tumor and do not inhibit MPS clearance of PLD. Altogether, our data suggest that

6

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

Figure 3.. High dose IVIg reduces the circulation kinetics of bispecific Ab by 3-fold. A) The serum circulation profile of bispecific Fab-IgG1 (30 μg) in athymic nude mice (n = 8 total mice, 4 mice per time point). B) The serum circulation profile of bispecific Fab-IgG1 (30 μg) in the presence of high dose IVIg (30 mg) in athymic nude mice (n = 8 total mice, 4 mice per time point). The solid line for both figures represents the predicted fit for a two-compartment model used to calculate the elimination half-life (t1/2), volume of distribution (VD), and clearance (CL).

coupling a tetravalent bsAb with reduced FcRn recycling can effectively enhance the delivery of otherwise passively targeted nanocarriers to tumors.

Discussion Pretargeting is a promising strategy that can combine the benefits of both passive targeting (longer particle circulation increases the fraction of particles that can extravasate at target tissue) and active targeting (cell-specific binding and uptake of particles to target cells/tissues). By eliminating the need to conjugate ligands onto the particles, particles can undergo longer circulation afforded by “stealth” polymer coating. Extravasated particles can then be internalized and retained at target sites by cell-bound pretargeting molecules. Here, we systemically evaluated how the format of bsAb-based pretargeting molecule may influence the pretargeting efficiency of PEGylated nanoparticles to tumors by first assessing the impact of multivalency on pretargeting efficiency in vitro comparing the uptake of FabIgG1 (4 Fabs, 2 each against PEG and HER2; 1 Fc domain) versus tandem Fab (2 Fabs, 1 each against PEG and HER; no Fc domain), followed by assessing the role of Fc domain in vivo. We found that increasing Fab valency and eliminating FcRn-

mediated prolonged circulation are both important to maximize pretargeting efficiency. Our findings lay down the blueprint for how to engineer bsAb-based pretargeting molecules that maximize pretargeting efficiencies. The two pretargeting molecules tested (tandem Fab vs. FabIgG1) differ in both their valency and the presence of Fc domain. The ideal study would have been to synthesize a quad Fab (4 Fabs, 2 each against PEG and HER2; no Fc domain), and compare the pretargeting potencies against the tandem Fab and Fab-IgG1. Unfortunately, the cloning of highly repetitive sequences and production of sufficient quantities of quad Fab were both exceptionally challenging. It is well established that the presence of Fc domain can enhance the production and secretion of proteins, as evidenced by the frequent inclusion of Fc in protein conjugates. 40–42 Our study shows that the Fc domain of the pretargeting molecules does not directly facilitate uptake in HER2 + cells, as evidenced by the lack of pretargeted delivery with control IgG1PEG versus no treatment control (Figure 2). This confirms the Fc domain is simply playing a bystander role, thus supporting our comparison of pretargeted delivery with Fab-IgG1 versus tandem Fab. The valency of pretargeting molecules plays an important role in pretargeting efficiency. Specifically, multivalency - increasing the number of antigen-binding domains per pretargeting

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

7

Figure 4.. Biodistribution of passively targeted and pretargeted PEGylated liposomal doxorubicin (PLD) at 48 h post-PLD dose. Tumor-bearing mice received PBS or pretargeting antibody treatment 24 h prior to PLD (n = 6–8 mice per treatment group). A) Concentration of doxorubicin in mouse serum. B) Concentration of doxorubicin in homogenized tumors (one-way ANOVA and post-hoc Tukey test; * P = 0.0124, ** P = 0.0035). C) Concentration of doxorubicin in homogenized non-targeted tissues: liver, spleen, heart, and lungs (mean ± SEM). D) Percent injected dose per gram of tumor (one-way ANOVA and post-hoc Tukey's test; * P = 0.0125, ** P = 0.0045). E) Percent injected dose per gram of tissue (mean ± SEM). F) Ratio of doxorubicin concentration in tumor to liver (one-way ANOVA and post-hoc Tukey's test; ** P = 0.0083, ** P = 0.0046). All biodistribution data is representative of n = 2 independent experiments. Means ± SD are shown (A, B, D), and means ± SEM are shown (C, E, F). Doxorubicin concentration in serum and homogenized tissues was quantified by HPLC, and was used to quantify tissue injected dose/g (%ID/g).

molecule - enhances the affinity to the particle, decreases dissociation rates when bound to cell-surface antigens, and maximizes tumor uptake and retention. 43 Our result is consistent with other findings that tetravalent bsAb are more potent at enabling pretargeted delivery of nanoparticles to tumors than bivalent bsAb. For instance, Harwood et al. engineered a tetravalent T-cell recruiting bsAb composed of three EGFRbinding domains and a single CD3-binding domain, and a tandem bispecific with one EGFR-binding domain and a single CD3-binding domain. 44 The multivalent bsAb was 15- to 20fold more potent at redirecting human T cells to lyse EGFRexpressing cells in vitro compared to bivalent bsAb. Our antigen-specific ELISAs showed both Fab-IgG1 and tandem Fab possess comparable affinity to HER2, whereas tandem Fab had nearly a 1000-fold lower affinity to PEG compared to Fab-IgG1. We believe the difference likely reflects in part to differences in the density of available HER2 versus PEG in our ELISA assays. To generate a high density of PEG chains on the ELISA plates, we coated plates with 5 kDa DSPEPEG, which presumably presents the hydrophilic PEG extending out from the surface of the ELISA plate. Because the anti-PEG Fab domains target the PEG backbone 45 (i.e. can bind anywhere along the PEG chain and multiple Fab can bind the same PEG chain), our PEG-ELISA setup most likely affords bivalent

binding of anti-PEG Fabs with Fab-IgG1HER2xPEG and IgG1PEG, as confirmed by the difference in binding observed with tandem Fab vs. Fab-IgG1. In contrast, our HER2-ELISA is generated by coating much larger HER2-Fc chimeric proteins (220–250 kDa) onto polystyrene plates. The much greater MW of HER2-Fc directly translates to a lower antigen density on the ELISA plate. Moreover, it is likely that only a portion of the HER2 protein is presented in a way that allows for binding by the anti-HER2 Fab/ IgG1/Fab-IgG1; it is frequently assumed that only ~5% of the coated antigen on ELISA is actually available for binding. Thus, only one of the two anti-HER2 Fabs on the Fab-IgG1 and IgG1HER2 is likely engaging HER2 coated on the ELISA plate, which would be reflected by comparable binding affinity between Fab-IgG1 and tandem Fab. We anticipate that Fab-IgG1 would associate much more tightly with HER2 targets on cancer cells than the tandem Fab in vivo. Pretargeting molecules present in the circulation can accumulate on injected nanoparticles before the particles have the opportunity to extravasate at target tissues; this would effectively be no different that injecting actively-targeted nanoparticles. Thus, we postulated that the ideal pretargeting molecules should have a relatively short serum half-life such that any unbound pretargeting molecules are quickly cleared from systemic circulation prior to nanocarrier dosing, leaving only

8

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

Table 1. Doxorubicin concentration in organs across treatment groups. Doxorubicin (Dox) concentration was determined by HPLC, and data are shown as mean ± SD. Data is representative of n = 2 independent experiments with n = 6–8 mice per treatment group. Treatment group PBS + PLD Fab-IgG1 + PLD Fab-IgG1 + IVIg + PLD

Serum [Dox] (ng/ml)

Tumor [Dox] (ng/ml)

Liver [Dox] (ng/ml)

Spleen [Dox] (ng/ml)

Heart [Dox] (ng/ml)

Lungs [Dox] (ng/ml)

1035.4 ± 506.6 1468.7 ± 817.9 1328.2 ± 837.3

221.2 ± 122.3 142.0 ± 130.4 649.7 ± 349.1

724.0 ± 470.3 1126.5 ± 1382.4 729.3 ± 292.3

284.8 ± 289.3 255.1 ± 191.5 340.4 ± 221.5

137.0 ± 24.4 254.3 ± 196.7 319.0 ± 229.7

415.4 ± 151.0 428.0 ± 187.1 527.8 ± 315.1

those pretargeting molecules that are already bound to the target cells. BsAb molecules possessing Fc domain engage with neonatal Fc receptors (FcRn) for extended serum half-life. To evaluate the impact of prolonged circulation due to FcRnmediated recycling, we administered a high dose IVIg together with the Fab-IgG1 to reduce FcRn-mediated recycling of the FabIgG1. In good agreement with our expectation, the consequent shorter serum half-life of Fab-IgG1 led to a 5-fold greater accumulation of PLD in the tumor compared to pretargeting with Fab-IgG1 alone. These findings provide a blueprint that guides the development of the next generation pretargeting molecules, specifically emphasizing the need for bivalent binding to PEG (and possibly HER2) as well as elimination of FcR binding. Biodistribution studies comparing drug/particle accumulation in tumors of mice treated with nontargeted particles (ie. passive targeting) versus targeted particles (ie. active targeting) produced variable results. Several studies reported active targeting improved tumor drug accumulation 3–4-fold compared to passive targeting. 14,46–48 However, other studies reported that actively targeted nanoparticles did not improve particle delivery to tumors compared to passively targeted nanoparticles. 20–22,49 Variations in therapeutic efficiency and drug accumulation in tumor for actively targeted versus nontargeted nanoparticles may be attributed in part to differences in the functionalization of particles. For some ligands, a narrow window of targeting ligand density exists that maximizes both tumor targeting and stealth properties. 14 Further, the local biodistribution of actively targeted nanoparticles may differ from passively targeted particles, leading to differences in therapeutic benefit. 20 For instance, both Kirpotin et al. 20 and Zahmatkeshan et al. 50 found that nontargeted drug-loaded nanoparticles predominantly accumulated in tumor extracellular space while targeted nanoparticles were efficiently internalized by tumor cells, which correlated with superior antitumor activity. These results underscore the importance of cell-specific delivery, a feature preserved with pretargeted nanoparticle delivery. Across tumor models (ie. target cell receptor, tumor type) and particle formulation (ie. chemotherapeutic drug-loaded nanoparticles, radiolabeled effector molecules and particles) pretargeting demonstrated therapeutic benefit to tumor-bearing mice as indicated by suppressed tumor growth and extended survival.25,29,51–53 Consistent with this body of literature, we showed that bsAb-based pretargeting markedly improved tumor accumulation and tumor-to-non-target organ ratio. Our results are similar to the work by Rauscher et al., which reported a 2-fold increase in specific tumor uptake of radiolabeled PEGylated liposomes with bsAb pre-injection versus without bsAb pre-injection in two separate studies. 51,52 Both studies displayed ~8% ID/g in tumor

for pretargeting formulations compared to nontargeted formulations (~4% ID/g). Other studies report much greater tumor uptake with pretargeting compared to passively targeted effector molecules, in large part due to very low baseline tumor uptake for passively targeted effector molecules: 30-fold increase (0.8 ± 0.02% ID/g vs 0.03 ± 0.01% ID/g) 53 and 100-fold improvement (7.58 ± 0.78% ID/g vs 0.07 ± 0.01% ID/g). 29 Although we saw less total accumulation and lower improvement than some of the studies, it is important to note that nearly all prior studies utilized subcutaneous xenograft models. Several other factors may contribute to differences in fold change of effector accumulation in tumors across studies, including choice of radiolabel and effector formulation, antigen density on target cells, dosing concentration of pretargeting and effector molecules, anatomical location of tumor xenograft, blood vessel density, and stromal content. The anti-PEG Fab (clone 6.3) 35 used in our study binds PEG of various sizes (spanning the typical MW of PEG used in PEGylated therapeutics), and importantly binds PEG irrespective of the terminal group, including methoxy- and amino-terminated PEG chains. 45 By engineering bispecific pretargeting molecules that can directly bind PEG on nanocarriers, we aim to improve the treatment of diseases using PEGylated therapeutics that are already FDA-approved without re-engineering or re-developing the therapeutic molecules. Thus, we only need to focus on the development of pretargeting molecules rather than both the pretargeting molecule and therapeutic moiety. With at least 15 PEGylated therapeutics currently in clinical trials, we anticipate that more patients will be prescribed PEGylated therapeutics, affording more opportunities for treatment using our pretargeting strategy. The modular nature of our bsAb-based pretargeting enables facile targeting of the same nanocarrier to diverse tissues/ cells simply by modifying the cell-binding Fab. Combining multiple pretargeting molecules as a cocktail may enhance delivery of PLD to diverse cell types within the tumor for greater drug exposure and better distribution throughout tumor. By maintaining the anti-PEG Fab binding domain on the pretargeting molecule, we can instantly enable cell-specific delivery of FDA approved PEGylated therapeutics, providing an opportunity to further improve the efficacies of these therapeutics.

Acknowledgements We would like to thank Charlene Williams and the UNC-CH Animal Services Core for their help in design and execution of animal studies. Animal Studies were performed within the UNC Lineberger Animal Studies Core Facility at the University of

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

North Carolina at Chapel Hill. The UNC Lineberger Animal Studies Core is supported in part by an NCI Center Core Support Grant (CA16086) to the UNC Lineberger Comprehensive Cancer Center.

Appendix A. Supplementary methods and data Supplementary data to this article can be found online at https://doi.org/10.1016/j.nano.2019.102076.

References 1. Fang J, Nakamura H, Maeda H. The EPR effect: unique features of tumor blood vessels for drug delivery, factors involved, and limitations and augmentation of the effect. Adv Drug Deliv Rev 2011;63:136-51. 2. Kobayashi H, Watanabe R, Choyke PL. Improving conventional enhanced permeability and retention (EPR) effects; what is the appropriate target? Theranostics 2014;4:81-9. 3. Torchilin V. Tumor delivery of macromolecular drugs based on the EPR effect. Adv Drug Deliv Rev 2011;63:131-5. 4. Fang J, Nakamura H, Maeda H. The EPR effect: unique features of tumor blood vessels for drug delivery, factors involved, and limitations and augmentation of the effect. Adv Drug Deliv Rev 2011;63:136-51. 5. Kobayashi H, Watanabe R, Choyke PL. Improving conventional enhanced permeability and retention (EPR) effects; what is the appropriate target? Theranostics 2013;4:81-9. 6. Torchilin V. Tumor delivery of macromolecular drugs based on the EPR effect. Adv Drug Deliv Rev 2011;63:131-5. 7. Torchilin VP. Passive and active drug targeting: drug delivery to tumors as an example. Handb Exp Pharmacol 2010;197:3-53. 8. Bertrand N, Wu J, Xu X, et al. Cancer nanotechnology: the impact of passive and active targeting in the era of modern cancer biology. Adv Drug Deliv Rev 2014;66:2-25. 9. Torchilin VP. Passive and Active Drug Targeting: Drug Delivery to Tumors as an Example. In: Handbook of Experimental Pharmacology. Springer, Berlin, Heidelberg, pp. 3–53. 10. Bazak R, Houri M, El Achy S, et al. Cancer active targeting by nanoparticles: a comprehensive review of literature. J Cancer Res Clin Oncol 2015;141:769-84. 11. Allen TM. Ligand-targeted therapeutics in anticancer therapy. Nat Rev Cancer 2002;2:750-63. 12. Reuter KG, Perry JL, Kim D, et al. Targeted PRINT hydrogels: the role of nanoparticle size and ligand density on cell association, biodistribution, and tumor accumulation. Nano Lett 2015;15:6371-8. 13. Elias DR, Poloukhtine A, Popik V, et al. Effect of ligand density, receptor density, and nanoparticle size on cell targeting. Nanomedicine Nanotechnology, Biol Med 2013;9:194-201. 14. Gu F, Zhang L, Teply BA, et al. Precise engineering of targeted nanoparticles by using self-assembled biointegrated block copolymers. PNAS 2008;105:2586-91. 15. Beech JR, Shin SJ, Smith JA, et al. Mechanisms for targeted delivery of nanoparticles in cancer. Curr Pharm Des 2013;19:6560-74. 16. Fakhari A, Baoum A, Siahaan TJ, et al. Controlling ligand surface density optimizes nanoparticle binding to ICAM-1. J Pharm Sci 2011;100:1045-56. 17. Moradi E, Vllasaliu D, Garnett M, et al. Ligand density and clustering effects on endocytosis of folate modified nanoparticles. RSC Adv 2012;2:3025-33. 18. Cheng Z, Al Zaki A, Hui JZ, et al. Multifunctional nanoparticles: cost versus benefit of adding targeting and imaging capabilities. Science 2012;338:903-10.

9

19. Yang Q, Parker CL, McCallen JD, et al. Addressing challenges of heterogeneous tumor treatment through bispecific protein-mediated pretargeted drug delivery. J Control Release 2015;220:715-26. 20. Kirpotin DB, Drummond DC, Shao Y, et al. Antibody targeting of longcirculating lipidic nanoparticles does not increase tumor localization but does increase internalization in animal models. Cancer Res 2006;66:6732-40. 21. Bae YH, Park K. Targeted drug delivery to tumors: myths, reality and possibility. J Control Release 2011;153:198-205. 22. Leamon CP, Cooper SR, Hardee GE. Folate-liposome-mediated antisense oligodeoxynucleotide targeting to Cancer cells: evaluation in vitro and in vivo. Bioconjug Chem 2003;14:738-47. 23. Patra M, Zarschler K, Pietzsch H-J, et al. New insights into the pretargeting approach to image and treat tumours. Chem Soc Rev 2016;45:6415-31. 24. Larson SM, Carrasquillo JA, Cheung N-KV, et al. Radioimmunotherapy of human tumours. Nat Rev Cancer 2015;15:347-60. 25. Su Y-C. Burnouf P-A, Chuang K-H, et al. Conditional internalization of PEGylated nanomedicines by PEG engagers for triple negative breast cancer therapy Nat Commun 2017;8:15507. 26. Tung H-Y, Su Y-C, Chen B-M, et al. Selective Delivery of PEGylated Compounds to Tumor Cells by Anti-PEG Hybrid Antibodies. Mol Cancer Ther; 14: 1317–26. 27. Khaw B-A, Gada KS, Patil V, et al. Bispecific antibody complex pretargeting and targeted delivery of polymer drug conjugates for imaging and therapy in dual human mammary cancer xenografts. Eur J Nucl Med Mol Imaging 2014;41:1603-16. 28. Goldenberg DM, Chatal J-F, Barbet J, et al. Cancer Imaging and Therapy with Bispecific Antibody Pretargeting 29. Cheal SM, Xu H, Guo H, et al. Theranostic pretargeted radioimmunotherapy of internalizing solid tumor antigens in human tumor xenografts in mice: curative treatment of HER2-positive breast carcinoma. Theranostics 2018;8:5106-25. 30. Cheal SM, Fung EK, Patel M, et al. Curative multicycle radioimmunotherapy monitored by quantitative SPECT/CT-based theranostics, using bispecific antibody pretargeting strategy in colorectal cancer. J Nucl Med 2017;58:1735-42. 31. Yang Q, Parker CL, Lin Y, et al. Pretargeting with bispecific fusion proteins facilitates delivery of nanoparticles to tumor cells with distinct surface antigens. J Control Release 2017;255:73-80. 32. Lewis SM, Wu X, Pustilnik A, et al. Generation of bispecific IgG antibodies by structure-based design of an orthogonal Fab interface. Nat Biotechnol 2014;32:191-8. 33. Yang Q, Jacobs TM, McCallen JD, et al. Analysis of pre-existing IgG and IgM antibodies against polyethylene glycol (PEG) in the general population. Anal Chem 2016;88:11804-12. 34. Zhang Y, Huo M, Zhou J, et al. PKSolver: an add-in program for pharmacokinetic and pharmacodynamic data analysis in Microsoft excel. Comput Methods Programs Biomed 2010;99:306-14. 35. Cheng TL, Roffler SR, Chuang KH, et al. Anti-polyethylene glycol antibody expressing cell quantify any free polyethylene glycol and polyethylene glycol-derivatized molecules. US Appl 20120015380A1 , https://www.google.com/patents/US20120015380 (2012, accessed 18 July 2018). 36. Breece TN, Fahrner RL, Gorrell JR, et al. Protein Purification. Pat No US 6,870,034 B2, , https://patents.google.com/patent/US6870034B2/en (2005). 37. Bleeker WK, Teeling JL, Erik Hack C. Accelerated autoantibody clearance by intravenous immunoglobulin therapy: Studies in experimental models to determine the magnitude and time course of the effect Blood Epub ahead of print 2001, https://doi.org/10.1182/blood. V98.10.3136. 38. Jacob S, Rajabally YA. Current proposed mechanisms of action of intravenous immunoglobulins in inflammatory neuropathies. Curr Neuropharmacol 2009;7:337-42. 39. Chatterjee K, Zhang J, Honbo N, et al. Doxorubicin cardiomyopathy. Cardiology 2010;115:155-62.

10

C.L. Parker et al / Nanomedicine: Nanotechnology, Biology, and Medicine 21 (2019) 102076

40. Czajkowsky DM, Hu J, Shao Z, et al. Fc-fusion proteins: new developments and future perspectives. EMBO Mol Med 2012;4:1015-28. 41. Uhlén M, Forsberg G, Moks T, et al. Fusion proteins in biotechnology. Curr Opin Biotechnol 1992;3:363-9. 42. Soleimanpour S, Hassannia T, Motiee M, et al. Fcγ1 fragment of IgG1 as a powerful affinity tag in recombinant fc-fusion proteins: immunological, biochemical and therapeutic properties. Crit Rev Biotechnol 2017;37:371-92. 43. Cuesta ÁM, Sainz-Pastor N, Bonet J, et al. Multivalent antibodies: When design surpasses evolution. Trends Biotechnol 2010;28:355-62. 44. Harwood SL, Alvarez-Cienfuegos A, Nunez-Prado N, et al. ATTACK, a novel bispecific T cell-recruiting antibody with trivalent EGFR binding and monovalent CD3 binding for cancer immunotherapy. Oncoimmunology 2017;7e1377874. 45. Roffler SR. 6–3 high-affinity anti-PEG IgG monoclonal antibody, , http://www.ibms.sinica.edu.tw/~sroff/anti-PEG/6-3.html (2016). 46. Kao C-H, Wang J-Y, Chuang K-H, et al. One-step mixing with humanized anti-mPEG bispecific antibody enhances tumor accumulation and therapeutic efficacy of mPEGylated nanoparticles. Biomaterials 2014;35:9930-40. 47. Su C-Y, Chen M, Chen L-C, et al. Bispecific antibodies (anti-mPEG/ anti-HER2) for active tumor targeting of docetaxel (DTX)-loaded

48.

49.

50.

51.

52.

53.

mPEGylated nanocarriers to enhance the chemotherapeutic efficacy of HER2-overexpressing tumors. Drug Deliv 2018;25:1066-79. Zhao Y, Fletcher NL, Liu T, et al. In vivo therapeutic evaluation of polymeric nanomedicines: effect of different targeting peptides on therapeutic efficacy against breast cancer. Nanotheranostics 2018;2:360-70. Cui J, Ju Y, Houston ZH, et al. Modulating targeting of poly(ethylene glycol) particles to tumor cells using bispecific antibodies. Adv Healthc Mater 2019;1801607:1801607. Zahmatkeshan M, Gheybi F, Rezayat SM, et al. Improved drug delivery and therapeutic efficacy of PEgylated liposomal doxorubicin by targeting anti-HER2 peptide in murine breast tumor model. Eur J Pharm Sci 2016;86:125-35. Rauscher A, Frindel M, Maurel C, et al. Influence of pegylation and hapten location at the surface of radiolabelled liposomes on tumour immunotargeting using bispecific antibody. Nucl Med Biol 2014;41:e66-74. Rauscher A, Frindel M, Rajerison H. Improvement of the targeting of radiolabeled and functionalized liposomes with a two-step system using a bispecific monoclonal antibody (Anti-CEA × Anti-DTPA–In). Front Med 2015;2:83. Westerlund K, Vorobyeva A, Mitran B, et al. Site-specific conjugation of recognition tags to trastuzumab for peptide nucleic acid-mediated radionuclide HER2 pretargeting. Biomaterials 2019;203:73-85.