Antibody screening using a human iPSC-based blood-brain barrier model identifies antibodies that accumulate in the CNS

Abstract

Drug delivery across the blood-brain barrier (BBB) remains a significant obstacle for the development of neurological disease therapies. The low penetration of blood-borne therapeutics into the brain can oftentimes be attributed to the restrictive nature of the brain microvascular endothelial cells (BMECs) that comprise the BBB. One strategy beginning to be successfully leveraged is the use of endogenous receptor-mediated transcytosis (RMT) systems as a means to shuttle a targeted therapeutic into the brain. Limitations of known RMT targets and their cognate targeting reagents include brain specificity, brain uptake levels, and off-target effects, driving the search for new and potentially improved brain targeting reagent-RMT pairs. To this end, we deployed human-induced pluripotent stem cell (iPSC)-derived BMEC-like cells as a model BBB substrate on which to mine for new RMT-targeting antibody pairs. A nonimmune, human single-chain variable fragment (scFv) phage display library was screened for binding, internalization, and transcytosis across iPSC-derived BMECs. Lead candidates exhibited binding and internalization into BMECs as well as binding to both human and mouse BBB in brain tissue sections. Antibodies targeted the murine BBB after intravenous administration with one particular clone, 46.1-scFv, exhibiting a 26-fold increase in brain accumulation (8.1 nM). Moreover, clone 46.1-scFv was found to associate with postvascular, parenchymal cells, indicating its successful receptor-mediated transport across the BBB. Such a new BBB targeting ligand could enhance the transport of therapeutic molecules into the brain.

Keywords: antibody‐mediated brain drug delivery; blood‐brain barrier; nicotinamide adenine dinucleotide.

Figure 1

Antibody library screening on in vitro BBB model. A) Scheme for the phage display screen. Step 1: pre-subtraction of phage scFv library on cultured human heart and lung endothelial cells in an effort to promote brain selectivity. Step 2: Supernatant from pre-subtraction step is next incubated with iPSC-derived BMECs to allow for binding and internalization of the antibody bearing phage. The antibody pools underwent three rounds of Step 2 internalization screening. Step 3: Internalizing phage were next dosed onto the blood side of BMECs in a Transwell format for 3 hours to allow for transcytosis. Recovered scFv-bearing phage particles were subjected to further analysis. B) Enrichment of BMEC-binding phage displayed scFvs as observed by FACS-analysis of phage antibody pools during screening Step 2. Shown are representative histograms of BMECs labeled with phage-displayed scFvs after respective screening rounds. The number of cells (counts: Y-axis) is given as function of the fluorescence intensity of phage antibody labeling of the cells (X-axis). In all experiments, BMECs were incubated with phage antibody pools, and cell-binding was detected by anti-M13 antibody. Geometric means are round R3 – 80.6, R2 – 27, R1 – 10.9, Non-binding phage – 3.16, respectively. C) Representative images from clonal phage immunocytochemistry with BMECs to determine clones displaying a BMEC binding phenotype. Scale bar, 50μm. For panels B) and C) Non-binding phage displays antibotulinum neurotoxin scFv.

Figure 2

Antibody binding properties. A) ScFv-Fc construct with scFv linked directly to rabbit IgG Fc region. B) Non-reducing, Coomassie blue-stained SDS-PAGE gel analysis of scFv-Fc antibodies following expression in HEK293F cells and protein A/G purification. Molecular weights are indicated. C) Binding and internalization of antibodies into BMECs. Live cells were incubated with antibodies (5 μg/ml) at 4°C and subsequently at 37°C for 30 min. The cell membrane was washed with cold buffer and the membrane bound fraction labeled with anti-rabbit Fc AlexaFluor555 antibody (red). After fixation and permeabilization the internalized antibodies were labeled with anti-rabbit Fc AlexaFluor488 antibody (green). Images were taken on an epifluorescent microscope. Scale bar, 20 μm. D) Temperature-dependent internalization of antibodies. Internalized antibody fluorescent signal values at 4°C are normalized to the total signal per cell at 37°C. Reported are means ±S.D., n=3, *p<0.05 determined by a two-tailed Student’s t-test assuming unequal variance. E) ScFv-Fcs binding to human and mouse brain microvessels. Cryosections of human and mouse brain were immunolabeled for CD31 (green) to visualize the blood vessels and incubated with 5 μg/ml scFv-Fcs (red) to identify scFv-Fc binding. Nuclei are visualized in blue. Images were taken on a confocal microscope. Scale bar, 20 μm.

Figure 3

Brain targeting of antibodies after intravenous administration in mice. A) Antibodies (5mg/kg) were injected intravenously in mice. One hour post-injection, mice were whole body perfused and brains collected. ScFv-Fcs (red) were labeled with fluorescent anti-rabbit Fc AlexaFluor555 antibody, blood vessels (green) were visualized with DyLight488 lectin that was present in the perfusion buffer. Four of five analyzed antibodies accumulate in brain vasculature as punctate structures in endothelial cells. Postvascular immunoreactivity was observed in the brain sections from mice injected with clone 46.1 (yellow arrows). Images were taken on an epifluorescence microscope. Scale bar, 20μm. B) Confocal images from a z-stack showing the localization of scFv-Fcs (red) with respect to collagen IV (blue). Blood vessels (green) as in A), nuclei (cyan). Clones 3, 26, and 46.1 colocalize with collagen IV (purple in merge and white arrows). Clone 17 shows no colocalization with collagen IV, but diffuse parenchymal staining was detected. C) Co-localization of scFv-Fcs (red) and GFAP+ astrocytes (blue) can be observed in merged confocal images (purple). The yellow arrows in panels B) and C) indicate accumulation of antibodies in postvascular, GFAP- brain cells. In all panels, the grayscale images are included to assist evaluation of the individual channels depicted in the merged images. Scale bar, 20 μm.

Figure 4

Organ biodistribution of antibodies. Antibodies (5 mg/kg) were injected intravenously in mice. One hour post-injection, mice were whole body perfused and organs collected. ScFv-Fcs were immunolabeled with fluorescent anti-rabbit Fc AlexaFluor555 antibody (red), blood vessels were visualized with the perfused DyLight488 lectin (green). White and yellow arrows point junctional localization of clone 46.1 in hepatocytes and renal epithelial cells, respectively. Images were taken on an epifluorescence microscope. Scale bar, 20 μm.

Figure 5

Quantification of scFv-Fc brain accumulation. Clones 17 and 46.1 were intravenously injected into mice (n=4). After one hour, the mice were whole body perfused, brains collected and antibodies extracted. Concentration of antibodies in brain extracts were determined with ELISA as described in Materials and Methods. Reported are means ±S.E.M., † p<0.005, * p<0.05 compared to Ctrl-Fc by two-tailed unpaired Student’s t-test.