Poly (β-amino Ester) Nanoparticles Modified with a Rabies Virus-derived peptide for the Delivery of ASCL1 Across a 3D In Vitro Model of the Blood Brain Barrier

Abstract

Gene editing has emerged as a therapeutic approach to manipulate the genome for killing cancer cells, protecting healthy tissues, and improving immune response to a tumor. The gene editing tool achaete-scute family bHLH transcription factor 1 CRISPR guide RNA (ASCL1-gRNA) is known to restore neuronal lineage potential, promote terminal differentiation, and attenuate tumorigenicity in glioblastoma tumors. Here, we fabricated a polymeric nonviral carrier to encapsulate ASCL1-gRNA by electrostatic interactions and deliver it into glioblastoma cells across a 3D in vitro model of the blood−brain barrier (BBB). To mimic rabies virus (RV) neurotropism, gene-loaded poly (β-amino ester) nanoparticles are surface functionalized with a peptide derivative of rabies virus glycoprotein (RVG29). The capability of the obtained NPs, hereinafter referred to as RV-like NPs, to travel across the BBB, internalize into glioblastoma cells and deliver ASCL1-gRNA are investigated in a 3D BBB in vitro model through flow cytometry and CLSM microscopy. The formation of nicotinic acetylcholine receptors in the 3D BBB in vitro model is confirmed by immunochemistry. These receptors are known to bind to RVG29. Unlike Lipofectamine that primarily internalizes and transfects endothelial cells, RV-like NPs are capable to travel across the BBB, preferentially internalize glioblastoma cells and deliver ASCL1-gRNA at an efficiency of 10 % causing non-cytotoxic effects.

Graphical Abstract

1. INTRODUCTION

Glioblastoma Multiforme (GBM) is the most common, aggressive, and lethal form of brain cancer. GBM is thought to form through communication between tumor-associated reactive astrocytes and glial cells. This connection promotes the aggressiveness, and progression, along with the survival of the tumor¹. Existing GBM therapies consist of surgical resection with succeeding localized radiation therapy and chemotherapy. All current treatments have limitations. For instance, surgical resection is invasive and normally only performed depending on the tumor shape, size, and location. Localized radiation is known to cause severe DNA damage resulting in cell apoptosis of surrounding healthy tissues². Chemotherapeutic drugs such as Temozolomide, the most utilized for GMB, also cause damage to the DNA. Unfortunately, these therapeutic treatments not only target tumor cells but also destroy healthy brain tissues^{2, 3}, neither of these treatments are specific to cell type⁴, and even after all treatments are applied, the survival rate is merely 15 months^{5, 6}.

The blood brain barrier (BBB) is the biggest biological challenge to overcome in the development of therapies for the treatment of GBM. This semi-permeable membrane is highly selective, limiting the blood borne particles allowed to cross⁷. The BBB not only averts small molecules from entering but contains transporters that evacuate foreign particles⁸. Molecules and nutrients can cross the BBB through carrier-mediated transport, active transport, passive diffusion, and endocytosis⁸. Chemical and physical approaches have been developed to cross the BBB such as vasoactive therapeutics^{9, 10}, osmotic pressure^11–13, and microbubbles¹¹. Vasoactive agents such as adenosine, histamine, and bradykinin, among others are utilized to open the BBB vasculature for a short period, which not only allows for drugs to be delivered into the brain but also toxins to enter^{14, 15}. Compounds such as mannitol open the tight junctions in the endothelial cell layer hyperosmotically allowing large molecules across the BBB by passive diffusion. Because of the nonselective gating, the uncontrolled inflow of molecules increases the amount of brain fluids causing aphasia, hemiparesis and neurological toxicity¹². Receptor mediated transcytosis pathways have been explored as a more selective pathway across the BBB. An example of such specific approaches can be found in the use of the rabies virus glycoprotein (RVG)-derived peptides^{16, 17} for mimicking rabies virus (RV) transient pathway across the BBB. RVG29 is a 29-residue peptide derived from RVG, which binds to the nicotinic acetylcholine receptor (nAchR). RVG29 is being studied as a targeting ligand to cross the BBB and deliver therapeutics to the central nervous system¹⁸. Using peptides as targeting ligand presents several advantages such as their relatively low molecular weight, easier obtention, relatively low cytotoxicity and immunogenicity, and degradability in vivo to naturally occurring compounds¹⁹.

Gene editing systems have grown as a new approach to treat diseases, including various types of cancer. Gene editing treatments consist in silencing, inserting, replacing, modifying or deleting parts of the genomic DNA of the tumorous cells²⁰. For instance, small interfering RNAs (siRNA) are used as treatment to silence specific gene(s)²¹, while meganucleases, zinc finger nucleases (ZFN)^{22, 23}, transcription activator-like effector nucleases (TALENs)²⁴, clustered regularly interspaced short palindromic repeats (CRISPR/Cas9)^{22, 24, 25} and piggyBac transposon/transposase²⁶ are technologies utilized to modify the genetic makeup by removing and inserting genes^{26, 27}. Gene editing tools have become important in biological engineering as tunable platforms for precise gene targeting. These therapeutic instruments have the potential to repair the mutations at the genetic level that cause diseases^{25, 28}. The achaete-scute family bHLH transcription factor 1 CRISPR guide RNA (ASCL1-gRNA) gene editing system has been investigated for GBM treatment. ASCL1 reprograms neuronal glioma cells blocking proliferation²⁹ through the removal from the cell cycle³⁰. ASCL1 is also thought to attenuate tumorigenicity in glioblastoma tumors³¹ as well as repair neuronal lineage potential, and promote terminal differentiation³².

Exogeneous plasmids, including the most common gene editing systems, are unable to spontaneously penetrate mammalian cells. Thus, viral vectors are the current delivery system in gene therapies. Viral vectors are effective as they rely on the virus natural ability to introduce genetic material into mammalian cells, however, they carry the possibility of immunogenic toxicity and insertional oncogenesis³³. Moreover, virus are limited to carry payloads with sizes between ~ 4.5 – 5 kb, making them unsuitable for most of the gene editing systems³⁴. Other avenues that have been used for the intracellular delivery of genes utilize physical or chemical manipulations. An example of physical manipulation is electroporation³⁵. This technique employs electric shock with high-voltage producing holes in the cell membrane allowing the introduction of genetic material³⁶. For chemical manipulation the most widely adopted technology is the Lipofectamine reagent, a cationic lipid-based system^{35, 37}, the benchmark for in vitro transfection³⁸. However, Lipofectamine is highly cytotoxic, limiting its application in vivo³⁹. Synthetic carriers such as are lipids, polymers, or inorganic nanoparticles have been proposed for gene delivery^{40, 41}. Although synthetic carriers have not been able to match the competence of a virus in terms of transfection efficiency, scientists can utilize materials chemistry and engineering to design them at low cost, with high reproducibility on a large scale, displaying low immunogenicity^{35, 40}, with add-on surface recognition functions to overcome biological barriers.

In this work, we have fabricated nanoparticles (NPs) composed of the block copolymer poly (ethylene glycol)–block–poly(1,4-butanediol)–diacrylate- β, –hydroxyamylamine–block–poly (ethylene glycol) (PEG-PDHA) for the stable encapsulation and intracellular delivery of the ASCL1-gRNA gene editing tool. Chemical crosslinking of PEG-PDHA polymer chains was used for nanoparticle stabilization in physiological conditions following previous investigations from our group on approaches to stabilize PEG-PDHA plasmid nanocarriers²⁷. For intercellular nuclear targeting, PEG-PDHA was conjugated with microtubule associated nuclear localization peptide (MTAS-NLS). PEG-PDHA NPs formed after complexation of the polymer with ASCL1-gRNA were surface engineered with RVG29 to mimic RV biorecognition pathway for crossing the BBB¹⁶. To assess their therapeutical potential RV-like NPs fabricated here were investigated using a 3D BBB in vitro model. We studied the ability of the NPs to cross the BBB in vitro model, penetrating human glioblastoma cells and selectively delivering the ASCL1-gRNA editing tool.

2. EXPERIMENTAL SECTION/METHODS

2.1. Materials.

1,4-butanediol diacrylate, 4-amino-1-butanol, Dimethyl Sulfoxide (DMSO), Triethylamine (TEA), Pyridine (Py), Sodium Chloride (NaCl), Poly(ethylenimine) (PEI), Sodium Hydroxide (NaOH), Hydrochloric acid (HCl), Triton, 4’,6-diamidino-2-phenylindole (DAPI), (1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), Hydroxy succinimide (NHS), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphynyltetrazolium bromide (MTT), Phosphate Buffer Saline (PBS), Fetal Bovine Serum (FBS), Ampicillin Sodium Salt and Opti MEM 1 reducing serum media were purchased from Sigma Aldrich. Amine-poly(ethylene glycol)-carboxymethyl (NH2-PEG -COOH) was purchased from Laysan Bio. Endothelial Cell Growth Basal Medium (EBM) and Endothelial cell Growth Medium SingleQuots kit was purchased from LONZA. Dulbecco’s Modified Eagle Medium (DMEM), Eagle’s Minimum Essential Medium (EMEM), human primary umbilical vein endothelial cell line (HUVEC), Uppsala 87 malignant glioma cell line (U87MG), and immortal mouse clone type I neuronal astrocyte cell line (C8-D1A) were obtained from ATCC. TrypLE Express was purchase from Gibco. Ultra-Pure Agarose, Lipofectamine 3000, LB broth (Miller), SYBR Safe DNA gel stain, and snakeskin dialysis tubing 10 kDa were received from Thermo Fisher Scientific. IgG (H+L) Cross-Adsorbed Goat anti-Rat, Alexa Fluor 647 secondary antibody, Triton X-100, Sytox Red and Calcein Green were purchased from Invitrogen. TWEEN-20 was purchased from Fisher Bioreagents. Albumin, Bovine (BSA) was obtained from VWR Life Science. Recombinant Anti-Nicotinic Acetylcholine Receptor alpha 1/CHRNA1antibody bought from abcam. Cy5.5 Mono NHS Ester was purchased from Cytiva. Human Achaete-scute homolog 1 (ASCL1) enzyme-linked immunosorbent assay (ELISA) kit was purchased from MyBioSource. Tris/Borate/EDTA (TBE) Buffer was obtained from Alfa Aesar. Plasmid Achaete-scute family bHLH transcription factor 1 guide ribonucleic acid (ASCL1-gRNA) containing a CMV-puro-t2A-mCherry expression cassette was purchased from Addgene as an E. coli agar stab. Bacterial amplification was performed using sterile LB broth containing ampicillin and refined utilizing an Endotoxin-free DNA purification kit from QIAGEN. Microtubule associated nuclear localization (MTAS-NLS) peptide was purchased from Eurogentec. 29-residue peptide derived from rabies virus glycoprotein (RVG29) was purchased from GenScript (RP20464). Rhodamine B (RhB) purchased from Chem-Impex Int’l Inc. Actin-stain 488 Phalloidin was obtained from Cytoskeleton Inc. Millicell ERS-2 volt-ohm meter and 3.0 μm pore size Millicell polyethylene terephthalate hanging cell culture insert were purchased from Millipore Sigma.

2.2. Synthesis of PEG-PDHA.

Michael addition polymerization technique was utilized for synthesizing poly (ethylene glycol)-block-poly[(1,4-butanediol)-diacrylate-β−5-hydroxyamylamine]-block-poly (ethylene glycol) (PEG-PDHA) as we reported previously^{27, 42}. Briefly, PDHA was synthesized by mixing 4-amino-1-butanol (7.48 mmol) with 1,4-butanediol diacrylate (8.23 mmol) and heating the mixture to 90 °C for 45.5 h under magnetic stirring and N₂ gas flow. Then, the product was resuspended in DMSO and sonicated with heat. The mixture was centrifuged, and the suspended portion was decanted and dried in a rotary evaporator. The purity of the PDHA was verified via proton nuclear magnetic resonance (¹H-NMR). PDHA was dissolved in DMSO and stored at −20 °C until further needed. NH2-PEG-COOH (0.05 mmol) was dissolved in DMSO with pyridine (6.18 mmol) and TEA (3.59 mmol) under magnetic stirring. PDHA was added dropwise to the solution every twelve minutes at a 1:2 PDHA: NH2-PEG-COOH molar ratio). The reaction was carried out for 24 h to produce COOH terminated PEG-PDHA. PEG-PDHA was purified through dialysis against Millipore water for two days. The purity of PEG-PDHA was assessed by ¹H-NMR. Only the polymer with purity above 80 % was utilized in this work. Purified PEG-PDHA was lyophilized and stored at −20 °C.

2.3. PEG-PDHA Functionalization.

MTAS-NLS was conjugated to the end-chain of PEG block from PEG-PDHA. The conjugation occurred through carbodiimide chemistry between carboxylic end-groups in the PEG block and the amine groups from residual lysine in MTAS-NLS. First, carboxylic groups were activated by suspending 0.006 mmol of PEG-PDHA in 3 mL of 10 mM 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) and 10 mM N-Hydroxysuccinimide (NHS). The suspension was sonicated for dispersion. Activation was carried out on an orbital shaker at 24 °C for 1 h. Then, the pH was adjusted to 8.6, MTAS-NLS (2 nmol) was added and the suspension was mixed in the orbital shaker at 4 °C for 2 days. The obtained suspension was purified by dialysis utilizing a 10 kDa membran. Finally, the PEG-PDHA conjugated MTAS-NLS polymer was lyophilized and purity was verified by ¹H-NMR.

2.4. Characterization of PEG-PDHA Nanoparticles.

Zeta potential and hydrodynamic diameter were measured by Dynamic Light Scattering (DLS) utilizing a NanoSizer (Malvern Nanosize, U.K.). DLS measurements were performed using Millipore water at 25 °C on and at a cell drive voltage of 30 V, utilizing the monomodal analysis model. Transmission electron microscopy (TEM) was performed utilizing a JOEL JEM-2010F high-resolution transmission electron microscope at 150 kV to further characterize the particles for size and morphology. Samples for TEM were stained with uranyl acetate prior imaging.

2.5. Plasmid Encapsulation and Crosslinking.

PEG-PDHA NPs were prepared by mixing bare PEG-PDHA with 10 % weight of MTAS-NLS conjugated PEG-PDHA in 0.15 M NaCl above the critical micelle concentration of PEG-PDHA (CMC_PEG-PDHA = 0.015 mg mL⁻¹)²⁷. ASCL1-gRNA plasmid was added carefully at a polymer/plasmid molar ratio of 0.36 for complexation. Encapsulation was carried out on an orbital shaker at 4 °C for 1 h. Then, 1 mL of a 10 mM EDC/NHS solution was added to the nanoformulation and the sample was mixed on the orbital shaker at 4 °C for 1 h. Afterward, the pH was adjusted to 8.6 and left overnight on orbital shaker for cross-linking. After, the sample was centrifuged at 18,000 rpm for 5 minutes at 4 °C. The supernatant was discarded, and PBS was added to resuspend the pellet. The nanoformulation was placed on an orbital shaker at 4 °C until needed. Plasmid encapsulation was confirmed by 1 % agarose gel electrophoresis using TBE as running buffer. Gel imaging was performed using a Lonza FlashGel Imaging unit and SYBR Safe for plasmid staining. UV-spectroscopy (Nanodrop One^C) was performed to quantify plasmid encapsulation.

2.6. Surface Modification of PEG-PDHA NPs with RVG29.

The crosslinked PEG-PDHA nanoformulation was modified via electrostatic interactions to embed RVG29 on the surface. RVG29 was added to the nanoformulation in 0.15 M filtered NaCl pH 4 at a molar ratio of 0.36 PEG-PDHA/RVG29 and sonicated until fully dispersed and placed on the orbital shaker at 4 °C for 3 days. Then, the nanoformulation was centrifuged at 18,000 rpm for 5 minutes at 4 °C. The supernatant was discarded, and the obtained RV-like PEG-PDHA NPs were suspended in PBS and placed on orbital shaker until used.

Quartz crystal microbalance with dissipation (QCM-D) monitoring was performed to confirm the surface chemistry protocol. QCM-D measurements were performed on a QSense Explorer microbalance (Biolin Scientific). A 14 mm diameter silicon dioxide coated quartz crystal with a fundamental frequency of 5 MHz was used as a sensor (QSX 303, Biolin Scientific). First, a baseline of 0.15 M NaCl at pH 7 was recorded. Then, a layer of PEI was assembled on top of the sensor by injecting a solution of 0.5 mg mL⁻¹ PEI in 0.15 M NaCl into the system to provide a positive surface for further assembly. A wash with 0.15 M NaCl at pH 7 was performed to remove excess of PEI. Afterward, 0.5 mg mL⁻¹ of PEG-PDHA in 0.15 M NaCl were absorbed. After 20 minutes incubation, excess of PEG-PDHA was removed through a wash with 0.15 M NaCl at pH 7. Then, PEG-PDHA was crosslinked with 10 mM EDC/NHS solution following the same protocol described above for colloidal NPs. Finally, 0.03 nmol of RVG29 were assembled onto the crosslinked PEG-PDHA surface. After 30 minutes incubation, the QCM-D sensor was washed with 0.15 M NaCl pH 7. The QCM-D frequency and dissipation were recorded for seven odd overtones (1st-13th). Changes in resonance frequency (ΔF) and dissipation (ΔD) were monitored during all the experiment. The relationship between ΔF and adsorbed mass were calculated using Sauerbrey Equation^{43, 44}:

$$m_{QCM-D}=-C\frac{\Delta f_i}{i}$$

with the mass sensitivity constant of the crystal C = −17.7 ng cm⁻²Hz⁻¹ and i the overtone number.

The normalized frequency shifts, $\Delta f= \frac{\Delta fi}{i}$ of the average frequency changes were employed to calculate the mass of polymer and RVG29 absorbed on the QCM-D quartz sensor. The ratio of dissipation and normalized frequency shifts, ΔD_i/(−ΔF_i/i), were smaller than 0.4 × 10⁻⁶ Hz^{−1 45}, fulfilling the conditions to use Sauerbrey equation⁴³.

Fluorescence correlation spectroscopy (FCS) was performed to verify that the RVG29 was attached to the surface of RV-like NPs. First, RVG29 was fluorescently labeled with Cy5.5 Mono NHS Ester via carbodiimide chemistry to produce RVG29-Cy5. FCS experiments were carried out utilizing the confocal laser scanning microscope (CLSM) LSM 880 from Carl Zeiss outfitted with the Confocor 3 FCS module. Briefly, the samples were excited with a helium-neon laser with the wavelength 633 nm, and the confocal volume was calibrated with the free dye ATTO-633 at a concentration of 40 nm. The RVG29-Cy5 as well as non-fluorescent PEG-PDHA NPs were dissolved in water containing 4 % SDS. First, the peptide was measured to determine the diffusion coefficient by using the fit-model 3D normal diffusion with the fit algorithms Simulated Annealing/Levenberg-Marquardt and considering one diffusing component. The estimated diffusion coefficient of the peptide was then used for the 2-component fitting of the peptide-nanoparticle complex. The formation of the complex was done by adding the peptide to 0.5 mg mL⁻¹ of PEG-PDHA NPs followed by an incubation at 37 °C for 1 h and with an agitation of 200 rpm. The measurement of the peptide-nanoparticle complex was done in situ by triplicate.

2.7. BBB In Vitro Model.

An in vitro triple co-culture model of the BBB was adopted from^{46, 47}, where glass cover slides on the bottom of the well of 24-well plate and Millicell hanging culture inserts were used. Independently, glioblastoma cells were cultured in EMEM culture media supplemented with 10 % FBS, astrocyte cells were cultured in DMEM culture media supplemented with 10 % FBS, and endothelial cells were cultured in EBM culture media supplemented with growth factors and 10 % FBS. All cell lines were maintained at 37 °C in 5 % CO₂. When cells reached ~ 80 % confluency, they were split utilizing TrypLE Express for cell releasing. Glass cover slides on the bottom of wells of the 24-well plate were seeded with 40K of glioblastoma cells. For the Millicell transwell, 60 K of astrocyte cells were seeded on the basolateral side of the membrane. After 3 h, the transwell was carefully flipped over and media was added to the well. After 2 days, 60 K of endothelial cells were seeded within the apical chamber of the transwell membrane. After 3 h of co-culturing, media were added to the apical chamber. After 5 days from astrocyte cell seeding, the transwell was hung on the wells of the 24-well plate containing glioblastoma cells in the bottom. The combined co-culture was maintained in both EBM and DMEM culture media supplemented with 10 % FBS at 37 °C in 5 % CO₂. Half the volume of the culture media was carefully replaced every other day until each cell compartment formed a monolayer.

2.8. Cellular Uptake.

For these studies, RV-like NPs were labeled with Rhodamine B isothiocyanate by covalently binding to the carboxylic end groups of PEG⁴⁸. The uptake studies were performed on the BBB 3D in vitro model. After cell confluency of 80 % and 24 h prior performing experiments, the media were changed. Prior to addition of NPs, transendothelial electrical resistance (TEER) was measured to assess the barrier function of the endothelial cells contained on the transmembrane surfaces⁴⁹. Labeled RV-like NPs were added carefully to the middle of the apical chamber of the transwell at plasmid concentrations of 5 μg μL⁻¹. Cells were fixed in the transwell and wells at different time intervals, 0, 4, 8 and 12 h with 4 % paraformaldehyde solution and permeabilized with 0.1 % Triton X-100 for 5 minutes. Actin-stain 488 Phalloidin and DAPI were used for F-actin and nucleus staining, respectively. Every membrane was carefully removed from support with delicate precision as to not disturb the cells for imaging. A Leica TCS SP8 Confocal Microscope was used to image cells co-cultured with labeled RV-like NPs. Flow cytometry (BD Accuri C6 Plus) was utilized to determine the uptake efficiency of cells co-cultured with labeled RV-like NPs. A fluorescence threshold was set utilizing intact glioblastoma, astrocytes, and endothelial cells. Measurement parameters were set the same for all samples and each run was set to count 10⁴ events. The total percentage of each cell layer containing the fluorescent labeled NPs was determined from the area corresponding to higher intensities than the threshold.

2.9. Transfection Efficiency.

The different cell lines were plated in the BBB 3D in vitro model as described above. Non fluorescent RV-like NPs carrying ASCL1-gRNA prepared as above, were carefully added in 20 μL of Opti-MEM to the wells at concentrations of 5 and 10 μg μL⁻¹ of plasmid ASCL1-gRNA. RV-like NPs were co-cultured in the BBB in vitro model for 2, 3 and 4 days. Lipofectamine 3000 was used as the positive control for transfection efficiency comparison against RV-like NPs. As ASCL1-gRNA plasmid construct contains fluorescent mCherry, flow cytometry (BD Accuri C6 Plus) was utilized to quantify transfection efficiency as means of mCherry protein cell expression. A fluorescence threshold was set utilizing intact glioblastoma, astrocytes, and endothelial cells. Measurement parameters were set the same for all samples and each run was set to count 10⁴ events. The total count of cells that expressed mCherry protein was determined from the area corresponding to higher intensities than the threshold. Confocal laser scanning microscopy (CLSM) was utilized to qualitatively characterize mCherry protein cell expression using a Leica TCS SP8 Confocal Microscope. ELISA was performed on the 3D BBB in vitro model to quantify intracellular ASCL1 protein concentrations. The cells were plated as stated previously. At 2 and 4 days after RV-like NPs addition, media was removed. Transwell membranes were moved to fresh cell-free wells. 200μL of TripLE was added to glioblastoma cells on bottom of the well. 200μL of TripLE was added to the fresh well containing astrocytes attached to underside of transmembrane along with 200μL of TripLE inside well with the endothelial cells. After detachment 500μL of PBS was used to wash and lift detached cells. Each cell layer was placed in a microcentrifuge tube and cell lyses was achieved by repeated freeze-thaw cycles. Following steps were performed according to the manufacturer’s protocol. Within 10 minutes of stop solution addition, optical density values at 450 nm were obtained using a plate reader.

2.10. Expression of nAchR in the BBB 3D In Vitro Model.

RV-like NPs specific recognition by the BBB in vitro model was evaluated by immunostaining of the anti-nicotinic acetylcholine receptors (nAchR), which bind to RVG29. After fixing and permeabilizing the cells from the 3D in vitro model, cells were incubated in a 1 % BSA solution for 30 minutes. Then, BSA solution was removed and recombinant anti-nicotinic acetylcholine receptor alpha 1/CHRNA1 antibody was added to the cells. The antibody was incubated for 1 h. Cells were washed three times with 0.05 % TWEEN-20 (washing solution) for 5 minutes each wash to assure free anti-body removal. A solution of the secondary antibody, IgG (H+L) Cross-Adsorbed Goat anti-Rat conjugated with Alexa Fluor 647, along with Actin-stain 488 Phalloidin containing the primary antibody (for cytoskeleton staining) were added. After 45 minutes of incubation, samples were washed three times with washing solution. DAPI solution was added for 5 minutes to all samples to stain the nuclei. Samples were washed again with washing solution and PBS was added. A Leica TCS SP8 Confocal Microscope was used for cell imaging.

2.11. Cell Viability.

Cell viability was evaluated by Live/dead assay using flow cytometry. Glioblastoma cells were seeded in a 96-well plate at a density of 15 K cells per well. Cells were incubated at 37 °C with 5 % CO₂ for 24 h to allow adhesion. The samples tested were as follows: naked ASCL1-gRNA, empty PEG-PDHA NPs, RV-like NPs formulation and Lipofectamine 3000 carrying ASCL1-gRNA. After incubation for 2, 3 and 4 days, TrypLE Express was utilized to detach cells from the 96-well plate surface. Cells were then suspended in FACS solution and Calcein (live stain) and Sytox (dead stain) were added to the samples right before measuring with flow cytometer.

2.12. Statistical Analysis.

Unless otherwise stated, all experiments were performed in triplicate with data being reported as mean ± standard error. To verify statistical differences, one-way analysis of variance (ANOVA) was performed along with Fisher’s least significant difference test for means comparison (p-value of 0.05, OriginPro 2016).

3. RESULTS AND DISCUSSION

3.1. Fabrication and Characterization of RV-like NPs.

Cross-linked PEG-PDHA NPs were fabricated using MTAS-NLS conjugated and unconjugated PEG-PDHA at a mass concentration of 10 % MTAS-NLS-conjugated polymer. ASCL1-gRNA gene editing tool was encapsulated within PEG-PDHA NPs via electrostatic interactions between the negatively charged plasmid and the amine groups of the PDHA block in the polymers that lead to the nanoparticle formation. After encapsulation, PEG-PDHA NPs were crosslinked with 10 mM EDC/NHS solution. Then, PEG-PDHA NPs were functionalized with RVG29 peptide through self-assembly on the NPs surface to form RV-like NPs (Figure 1).

Figure 1.

Schematic representation of the fabrication of RV-like nanoparticles encapsulating ASCL1-gRNA.

Transmission electron microscopy (TEM) and dynamic light scattering (DLS) were utilized to characterize the morphology and size of the NPs. TEM images confirmed that bare PEG-PDHA NPs (Figure S1) are spherical in shape with a uniform size distribution and a mean diameter of 8 ± 2 nm. DLS characterization showed that PEG-PDHA NPs (Figure 2b and Figure S1) exhibit a hydrodynamic diameter in the range of 90−250 nm with polydispersity index (PDI) between 0.25 and 0.55. RV-like NPs’ morphology and size were also characterized by TEM and DLS. TEM shows that the RV-like NPs also are spherical in shape with a uniform size distribution and the mean diameter size of 6 ± 2 nm (Figure 2a). DLS shows that the RV-like NPs had a hydrodynamic diameter in the range of 90−200 nm with PDI between 0.30 and 0.55 (Figure 2b and Figure S1). The NPs’ diameter measured by TEM is ~30× smaller than the hydrodynamic diameter measured by DLS. The size discrepancy between TEM and DLS is expected because the latter is an indirect measurement of the hydrodynamic diameter of PEG-PDHA micelles in a solution, which includes the hydration layer around the particle, whereas the former reflects the morphology size of the dehydrated/collapsed micelles^50–52. Moreover, the PDI values measured by DLS indicate that the hydrated samples are moderately polydisperse, possibly as a result of NP aggregation in PBS. The surface charge of PEG-PDHA NPs was determined by ζ-Potential measurements after each step in the functionalization (Figure 2c). The ζ-Potential of unmodified PEG-PDHA NPs is of −5.4 ± 0.1 mV and when NPs are prepared using 10 % of MTAS-NLS conjugated PEG-PDHA the ζ-Potential is −1.8 ± 0.6 mV. In both cases, PEG-PDHA NPs display an almost neutral surface charge as expected from the PEG blocks of the polymer that arrange around the core of the NPs where the nucleic acid is located. At physiological pH the PDHA block of the co-polymer is not charged, thus, to encapsulate negatively charged plasmid DNA via electrostatic interactions, we first protonated the PDHA block of the co-polymer at pH 4^{27, 53–55}. After allowing for complexation with ASCL1-gRNA, the complex was neutralized to pH 7 to form NPs entrapping ASCL1-gRNA. The ζ-Potential after encapsulation of ASCL1-gRNA at pH 7 changed towards positive values of +6.9 ± 0.7 mV. Encapsulation of the plasmid was also corroborated by gel electrophoresis (Figure 2d). We previously studied the stability of PEG-PDHA NPs as carriers for gene delivery, finding that PEG-PDHA NPs require crosslinking to avoid premature dissociation²⁷. Thus, after ASCL1-gRNA encapsulation PEG-PDHA NPs were crosslinked with EDC/NHS, which shifted the ζ-Potential towards a negative surface charge of −11.0 ± 0.6 mV. The shift to negative ζ-potential values is indicative of amines groups being crosslinked and come from free carboxylate end groups of the PEG block. It could also be that in the crosslinked NPs at pH 7, though the plasmid remains trapped within the crosslinked polymer, the interaction plasmid-polymer is weaker, thus the plasmid is more exposed to the surface, which would also explain the negative charge. Finally, RVG29 peptide was embedded into PEG-PDHA nanoformulation through electrostatic interactions to form RV-like NPs. The ζ-Potential of RV-like NPs was −25.3 ± 0.3 mV, which hinted the attachment of the negatively charged RVG29 on the surface of the NPs. NPs have an overall net negative charge before RVG29 functionalization. However, we can assume that some regions of the polymer with a positive charge are available for complexation with the negatively charged RVG29.

Figure 2.

RV-like PEG-PDHA NPs characterization. a) TEM images of RV-like NPs, scale bar = 10 nm. b) Hydrodynamic size distribution of PEG-PDHA NPs and RV-like NPs. c) PEG-PDHA NPs surface functionalization steps followed by ζ-Potential measurements at pH 7. d) Gel Electrophoresis of ASCL1-gRNA laden into PEG-PDHA NPs. Well 1: 1 kb Plus DNA ladder, well 2: Low DNA mass ladder, well 3: ASCL1-gRNA (8.3 kb), well 4: PEG-PDHA NPs and well 5: RV-like NPs.

To corroborate the surface chemistry followed to embed RVG29 onto PEG-PDHA NPs surface experiments with the QCM-D were performed (Figure S2). PEG-PDHA was electrostatically deposited on top of a PEI coated quartz crystal sensor. Then, PEG-PDHA was crosslinked and RVG29 was assembled on top of the crosslinked polymer. We observed that ~13 ng cm⁻² of RVG29 were embedded per each 157 ng cm⁻² of crosslinked PEG-PDHA. To further confirm the attachment of RVG29 to the surface of colloidal crosslinked PEG-PDHA NPs, we performed FCS measurements (Figure S3). RVG29 was chemically bonded to the fluorophore Cy5. FCS allows us to measure the change in the diffusion coefficient of the labelled peptide from a free molecule in bulk to a complex with the NPs. Free RVG-Cy5 had a diffusion coefficient of 73.9 ± 3.0 μm² s⁻¹. When RVG29-Cy5 was attached to PEG-PDHA NPs the diffusion coefficient decreased to 3.3 ± 1.0 μm² s⁻¹, indicating that RVG29-Cy5 is no longer freely diffusing but attached to the larger PEG-PDHA NPs.

3.2. Development of a 3D In Vitro Model of the BBB.

To investigate the potential of RV-like NPs as carriers for gene-editing tools across the BBB, we developed a 3D in vitro model of the BBB based on a triple co-culture adapted from McCarthy, D.J. et al.⁴⁷ and Stone, N.L. et al.⁴⁶. During the formation process of the 3D BBB in vitro model, it is critical to monitor the transendothelial electrical resistance (TEER) and to confirm the formation of tight junctions and barrier functions of epithelial cells on both sides of the porous transmembrane of the transwells⁴⁹. Measurements were carried out at each step in the formation of the 3D BBB model. The TEER (Ω) measurement from the cell-free transwell were subtracted when determining the TEER value of the samples. Moreover, TEER values are multiplied by the surface area of the transwell membrane to obtain the resistance flux (Ω cm²). A cell-free transwell sample with culture media was used as control. The cell-free transwell displayed a TEER of 18 Ω cm². The 3D BBB in vitro model was formed by seeding endothelial cells on the apical membrane side representing the endothelial cell layer, astrocytes cells on the basolateral membrane side, and in the bottom of the well glioblastoma cells (Figure 3a). Physiological TEER values are typically higher than 1500 Ω cm^{2 56}, however 3D in vitro BBB models have been reported to display TEER values between 10 and 260 Ω cm^{2 46, 47, 57, 58} depending on pore size and area of transwell membrane, as well as on the types of cells utilized. It has also been reported that immortalized cells tend to have TEER values lower than 150 Ω cm^{2 56}. The formation of the 3D BBB was confirmed by the increase of TEER. Prior to TEER measurements, the electrode was sterilized and equilibrated according to manufacture protocol. TEER measurements started at day 0 as shown in Figure 3b, which displays the TEER changes from intact 3D BBB in vitro model (control) to the exposure to RV-like NPs. The TEER values gradually dropped within one day after addition of RV-like NPs on day 2. Decreasing of TEER values after nanomaterials addition can be attributed to changes in the cells’ tight junctions due to NPs exposure⁵⁹, and to cell height increase over time⁴⁹, particularly of glioblastoma cells which tend to form organoid-like structures overtime. We further confirm the formation of the 3D BBB in vitro model by CLSM imaging of the three different cell layers composing the model (Figure 3c). Through CLSM imaging it is possible to confirm not only the presence of the three different cell cultures in each layer but the formation of cell monolayers in each compartment of the transwell (apical and basolateral sides of the transmembrane, and glass coverslip on the bottom of the well) within 6 days. The cell monolayers were very consistent and overlapping cells were observed in each transwell component of the 3D BBB in vitro model.

Figure 3.

3D BBB in vitro model. a) Schematic representation of the 3D in vitro model of the BBB. b) TEER monitoring of the BBB in vitro model. c) CLSM images of each cell layer forming the 3D BBB in vitro model. Nuclei are stained DAPI (cyan), and cytoskeletons were stained with actin-stain 488 phalloidin (green). Scale bar = 50 μm.

3.3. Cellular Uptake of RV-like NPs.

A necessary component to consider when engineering gene delivery nanocarriers is intercellular uptake. The nanocarrier must be designed to protect the payload from degradation when trafficking the biological system, while effectively penetrate and deliver the cargo intercellularly. RV-like NPs fluorescently labeled with Rhodamine B (RhB) were investigated for intercellular uptake across the 3D BBB in vitro model. RhB fluorescently labeled NPs were added to the apical chamber of the transwell membrane. All three cell lines were co-cultured with NPs for 4, 8, and 12 h, at which the well plates were removed, and cells were fixed and stained. CLSM imaging was performed to qualitatively assess NPs uptake in the 3D BBB in vitro model (Figure 4b–c and Figure S4). Intact cells from each compartment of the 3D BBB in vitro model were imaged as control (Figure 4a). PEG-PDHA NPs were tracked in the 3D BBB model after 8 h of co-culturing (Figure 4b). CLSM images show that PEG-PDHA the NPs are taken up mainly by endothelial cells. Few PEG-PDHA NPs can cross the endothelial cell layer and be uptaken by astrocytes. No PEG-PDHA NPs were able to travel across the 3D BBB model and internalize glioblastoma cells. When RV-like NPs were added to the 3D BBB in vitro model, most of the NPs were found internalized into the astrocytes and glioblastoma cells after 8 h of co-culturing (Figure 4c). This result infers that RV-like NPs are capable to travel through the endothelial and astrocyte cell layers from the transwell to reach glioblastoma cells located in the bottom of the well. Thus, we quantified RV-like NPs cell uptake using flow cytometry. Flow cytometry results verified that 99 ± 1 % of the endothelial cell layer, 89 ± 3 % of the astrocyte layer, and 84 ± 4 % of the glioblastoma cells layers from the 3D BBB in vitro model had taken up RV-like NPs within 8 h of co-culturing (Figure 4d). Cellular uptake was also tracked for PEG-PDHA NPs and RV-like NPs at co-culture times of 4 and 12 h, which showed similar trend as for 8 h (Figure S4).

Figure 4.

Cellular Uptake. CLSM images of each layer composing the 3D BBB in vitro model for: a) control (intact cells), b) co-cultured with PEG-PDHA NPs, and c) co-cultured with RV-like NPs. Images were taken after 8 h of co-culturing with the NPs. NPs are labeled with RhB (magenta), cell nuclei are stained by DAPI (cyan) and cytoskeleton stained with actin-stain 488 phalloidin (green). Scale bar = 50 μm. d) Flow cytometry of cells from each layer of the 3D BBB model exposed to RV-like NPs labelled with RhB for 8 h.

3.4. RV-like NPs Transfection Efficiency in Glioblastoma Cells.

Transfection efficiency of the gene editing tool ASCL1-gRNA was first investigated in glioblastoma cells using RV-like NPs. The ASCL1-gRNA plasmid construct utilized here contains fluorescent mCherry coupled within a self-cleaving T2A linker. Thus, transfection efficiency was measured by mCherry protein expression in cells transfected with ASCL1-gRNA through flow cytometer and CLSM. Intact glioblastoma cells were used in this experiment as the negative control to set the fluorescence threshold that will allow for distinguishing mCherry positive from mCherry negative cell populations (Figure 5a and 5d). Glioblastoma cells transfected with Lipofectamine 3000 reagent were used as positive control. Transfection with Lipofectamine 3000 resulted in 21.0 ± 1.5 % of the glioblastoma cell population expressing mCherry protein (Figure 5b and 5e). Transfection with RV-like NPs formulation resulted in 10 ± 1 % of glioblastoma cells expressing mCherry protein (Figure 5c and 5f). The transfection efficiency by RV-like NPs is half of what can be achieved when used commercially available Lipofectamine 3000 reagent. We previously reported the optimization of PEG-PDHA NPs for the encapsulation and intracellular delivery of PiggyBac transposon²⁷, which yielded to higher transfection efficiency in glioblastoma cells (~55 % and ~ 60 % with Lipofectamine 3000). We attribute the lower transfection efficiency of both, Lipofectamine 3000 and the RV-like nanoformulation to the cargo size. ASCL1-gRNA plasmid size is 8.3 kbp, while PiggyBac is only 6.3 kbp. Plasmid size has been identified as a critical factor in transfection efficiency. Previous studies have shown that regardless the transfection method, large plasmids are more difficult to be delivered into the cells^60–62. Plasmid size is particularly significant in gene editing therapies, where the therapeutic plasmids required for inserting, modifying, replacing, or deleting parts of the genomic DNA are generally large. Nevertheless, prior research studies have identified the minimum stable transfection efficiency for a successful gene editing therapy to be between 6 and 37 %^63–65.

Figure 5.

Transfection efficiency into glioblastoma cells. Flow cytometry of glioblastoma cells: a) intact, b) transfected with ASL1-gRNA using Lipofectamine reagent, and c) transfected with ASCL1-gRNA using RV-like NPs. CLSM images taken at day 3 post-transfection for: d) control (intact cells), e) transfected with ASL1-gRNA using Lipofectamine reagent, and f) transfected with ASCL1-gRNA using RV-like NPs. Nuclei are stained by Hoechst 33342 (cyan), cytoplasm is stained with Calcein (grey), and ASCL1-gRNA transfection is visualized by mCherry protein (yellow) expression. Scale bar = 50 μm. Enlarged images (2X) of the areas indicated by squares are included.

3.5. RV-like NPs Transfection Efficiency across the 3D BBB in vitro Model.

Following, we evaluated transfection efficiency in the 3D in vitro model of the BBB. RV-like NPs were investigated for intercellular delivery of ASCL1-gRNA across the three cell layers of the BBB in vitro model. The nanoformulation was added to the apical chamber of the transwell membranes. At the co-culturing times of 2, 3 and 4 days the 3D BBB model was dissembled, and cells were detached for flow cytometry measurements. As previously described, flow cytometry was utilized to quantify transfection efficiency. Intact endothelial, astrocytes and glioblastoma cells from the 3D BBB model were used as the negative control to set the fluorescence thresholds for cell transfection. For the positive control we utilized commercially available Lipofectamine 3000 reagent carrying ASCL1-gRNA. After 4 days of co-culturing, the highest transfection efficiency when using Lipofectamine as carrier of ASCL1-gRNA was of 15.7 ± 0.2 %, observed in the endothelial cell layer (Figure 6a). Around 8.9 ± 0.8 % transfected astrocyte cells were observed, but the least transfected cells were found to be in the glioblastoma cell compartment (5.8 ± 0.2 %). The observed trend indicates that Lipofectamine 3000 transfection efficiency decreases as the cell layers get farther from the administration site. Lipofectamine 3000 reagent was added to apical chamber of the transwell which contains endothelial cells. Endothelial cells and astrocytes (basolateral side of the transwell) restrict Lipofectamine traffic across the transmembrane and into glioblastoma cells located at the bottom of the well. When investigating RV-like NPs as carriers of ASCL1-gRNA an opposite trend than Lipofectamine 3000 was observed. At 4 days of co-culturing with RV-like nanoformulation, the highest transfection efficiency was observed in glioblastoma cells with 12.0 ± 2.3 % cells transfected (Figure 6b), indicating that RV-like NPs are capable to pass through the endothelial (2.4 ± 0.2 % transfected) and astrocytes (8.8 ± 0.8 % transfected) cell layers in the transwell membrane and internalize into glioblastoma cells. ASCL1-gRNA transfection efficiencies at day 2 and 3 using Lipofectamine and RV-like NPs are shown in (Figure S5 and Figure S6) and are alike the trends observed at day 4. CLSM imaging was performed to qualitatively assess the intracellular delivery of ASCL1-gRNA, which ultimately should result in the cell expression of fluorescent mCherry protein. At day 2, 3 and 4 post incubation with Lipofectamine 3000 or RV-like NPs, the 3D cell cultures were removed from the BBB model, and cells were stained with Hoechst 33342 for nuclei labelling and with Calcein green to label the cytoplasm of live cells (Figure 6c–e, Figure S5 and Figure S6). Intact cells were imaged as control (Figure 6c). CLSM imaging confirmed the transfection results measured by flow cytometry. When using Lipofectamine 3000, higher mCherry protein expression is observed in the endothelial and astrocyte cell layers than in glioblastoma cells. Contrary, RV-like NPs can reach the bottom of the 3D BBB in vitro model, internalize into glioblastoma cells, and deliver ASCL1-gRNA (Figure 6e) as indicated by mCherry protein expression in glioblastoma cells.

Figure 6.

Transfection efficiency across the 3D BBB in vitro model. Flow cytometry of cells from the BBB in vitro model transfected with ASL1-gRNA using a) Lipofectamine reagent, and b) RV-like NPs. CLSM images of each layer composing the 3D BBB in vitro model for: c) control (intact cells), d) transfected with ASL1-gRNA using Lipofectamine reagent, and e) transfected with ASCL1-gRNA using RV-like NPs. Flow cytometry and CLSM images were taken 4 days post-transfection. Nuclei are stained by Hoechst 33342 (cyan), cytoplasm is stained with Calcein (grey), and ASCL1-gRNA transfection is visualized by mCherry protein (yellow) expression. Scale bar = 50 μm. Enlarged (2X) images of the areas indicated by squares are included.

Additionally, to assess the quantity of transfection by RV-like NPs we measured the amount of ASCL1 protein produced by cells using an ELISA. At the co-culturing times of 2 and 4 days the 3D BBB in vitro model was disassembled, and cells were detached utilizing TrypLE. The cells were suspended in PBS and ASCL1 protein quantification was performed following the ELISA protocol from the manufacturer. Intact cells from the 3D BBB in vitro model were used as negative controls. ASCL1 concentration in the controls was zero as expected since the gene is absent in intact cells. Intracellular ASCL1 protein concentration in all cell lines exposed to RV-like NPs formulation increased over time, observing the higher concentrations in glioblastoma cells (Figure 7). On day 2 the ASCL1 protein concentration in glioblastoma cells was ~ 450 μg mL⁻¹ increasing to ~ 1310 μg mL⁻¹ on day 4 post-transfection with RV-like NPs. After performing statistical analysis using one-way ANOVA, we found that glioblastoma ASCL1 protein expression was significantly higher than in endothelial and astrocyte cells from the 3D BBB in vitro model.

Figure 7.

Intracellular concentration of ASCL1 protein mediated by RV-like NPs transfection. Concentration of ASCL1 protein expressed in the different cell layers of the 3D BBB in vitro model measured by ELISA assay 2- and 4-days post-transfection. Error bars show the means ± standard deviation (n=3). One-way ANOVA test was used to determine significant difference. * Means significant difference was found with respect to other cells from the corresponding day (p value <0.05).

3.6. Specific Recognition of RV-like NPs in the 3D BBB in vitro Model.

An important design component of RV-like NPs is the surface embedding of RVG29. This peptide is known to bind the nAchR and facilitate NPs penetration across the BBB through receptor-mediated transcytosis pathway⁶⁶. Verifying that the 3D BBB in vitro model developed here expresses nAchR receptors is of utmost importance to understand RV-like NPs cell translocation mechanism. We identified nAchR through immunochemistry using recombinant anti-nicotinic acetylcholine receptor alpha 1/CHRNA1 antibody. First, we studied nAchR in endothelial, astrocyte and glioblastoma cells cultured independently (Figure S7). Endothelial and glioblastoma cells showed no presence of nAchR while astrocytes expressed nAchR. This observation was expected since it is known that astrocytes express multiple subunits of nAchR^{67, 68}. Then, the endothelial, astrocyte and glioblastoma cell layers from the 3D BBB in vitro model were stained for recombinant anti-nicotinic acetylcholine receptor alpha 1/CHRNA1 antibody (Figure 8) reveling the presence of nAchR receptors in all cell lines across the BBB in vitro model. We hypothesized that surface functionalization with RVG29 will endow specific biorecognition functions to the NPs for passaging the BBB. Specific biorecognition should occur between RVG29 and nAchR. Here we showed that in our 3D BBB in vitro model, nAchR receptors are naturally formed when co-culturing the three cell lines. This has been previously shown and occurs to facilitate cell-to-cell communication within the co-cultures^{16, 18}.

Figure 8.

Expression of nAchR in the 3D BBB in vitro model. CLSM images of immunostaining for recombinant anti-nicotinic acetylcholine receptor alpha 1/CHRNA1 antibody in: a) endothelial, b) astrocytes, and c) glioblastoma from the 3D BBB in vitro model. Nuclei are stained by DAPI (cyan), cytoskeleton is stained with actin-satin 488 phalloidin (green), and nAchR antibody (magenta). Scale bar = 50 μm. Enlarged (2X) images of the areas indicated by squares are included.

3.7. Cytotoxicity of RV-like NPs.

Cell viability of glioblastoma cells when using RV-like NPs as delivery carrier of ASCL1-gRNA was investigated through Live/Dead assay. Calcein and Sytox Red were used to label live and dead cells, respectively, to further quantify the cell populations using flow cytometry (Figure 9). Intact glioblastoma cells were used as the control group. Around 90 ± 0.8 % of glioblastoma cells co-cultured with either naked ASCL1-gRNA, empty PEG-PDHA NPs or RV-like NPs were alive, while 10 ± 0.8 % of the population were dead cells during the four days of co-culturing. The same percentage of live and dead cell populations are observed in intact glioblastoma cell cultures (control group), indicating no cytotoxic effects from the plasmid nor from the NPs and rather a 10.0 ± 0.8 % of death in cells due to handling procedures such as detachment from substrate with enzymes. When glioblastoma cells are transfected using Lipofectamine 3000 only 63.0 ± 0.8 % of the cells are alive after two days of co-culturing, which is significantly lower than the control. After four days of exposure to Lipofectamine 3000 half of the cell population was dead. The detrimental effects of Lipofectamine on cells is expected because of the highly positive charge of the lipid-based carrier. The cytotoxicity of Lipofectamine 3000 has been previously reported utilizing various plasmids^{27, 41, 69, 70}.

Figure 9.

Cell Viability. Percentage of a) live and b) dead cells from populations of 10,000 glioblastoma cells exposed to different conditions. Error bars show the means ± standard error of mean (n=3). One-way ANOVA test was used to determine significant difference. * Means significant difference was found with respect to the control (p value <0.05).

4. CONCLUSION

In this work, we developed a non-viral platform for the intracellular delivery of gene editing tools across the BBB. A co-polymer from a poly (β amino ester), PEG-PDHA, was utilized to encapsulate ASCL1-gRNA. We previously demonstrated the potential of this polymer for gene therapy as it can be internalized by cells and translocate into their nucleus within a few hours of co-culturing. The surface of this nanoformulation was further engineered here to provide specific biorecognition functions for crossing the BBB and delivering the therapeutic cargo intracellularly. The surface engineering protocol followed mimicry the surface chemistry of RV to form RV-like NPs. A 3D in vitro model of the BBB was established to evaluate the nanoformulation. RV-like NPs are capable of transiently crossing the 3D BBB in vitro model and internalizing into glioblastoma cells within 8 hours of co-culturing. Once taken up by the cells, RV-like NPs dissociate due to polymer protonation at intracellular pH (< 6), releasing the genetic material. RV-like NPs are preferentially internalized by glioblastoma cells from the 3D BBB in vitro model, transfecting them at a rate of 12 % causing no cytotoxicity. On the contrary, Lipofectamine 3000 reagent is not efficient in crossing the 3D BBB in vitro model, transfecting primarily the contact endothelial cell layer of the model at a rate of 15 % and decreasing the viability of cells below 80 %.

Mimicking the surface chemistry of rabies virus into synthetic nanocarriers by incorporating RVG29 enables a non-invasive strategy to transiently overcome the BBB. The surface functionalization approach presented here opens the possibility for new biomimicry approaches to overcome other challenging biological barriers. The developed nanoformulation has the potential to encapsulate and deliver other gene editing tools regardless of their size, limiting factor in viral carriers. Further studies will explore in vivo gene delivery across the BBB for potential clinical translation. Understanding brain cellular pathways, and the ability to engineer non-invasive routes across the BBB are expected to accelerate the development of new therapies for brain cancers and other brain diseases.

ABBREVIATIONS

ASCL1-gRNA

achaete-scute family bHLH transcription factor 1 CRISPR guide RNA

BBB

blood brain barrier

RV

rabies virus

RVG

rabies virus glycoprotein

RVG29

29-residue peptide derived from RVG

RV-like NPs

nanoparticles with RVG29

CLSM

confocal laser scanning microscope

GBM

glioblastoma multiforme

nAchR

nicotinic acetylcholine receptor

siRNA

small interfering RNAs

ZFM

zinc finger nucleases

TALENS

transcription activator-like effector nucleases

CRISPR/Cas9

clustered regularly interspaced short palindromic repeats

NPs

nanoparticles

PEG-PDHA

poly (ethylene glycol)–block–poly(1,4-butanediol)–diacrylate- β, –hydroxyamylamine–block–poly (ethylene glycol)

MTAS-NLS

microtubule associated nuclear localization peptide

RhB

Rhodamine B

DMSO

Dimethyl Sulfoxide

TEA

Triethylamine

Py

Pyridine

NaCl

Sodium Chloride

PEI

Poly(ethylenimine)

NaOH

Sodium Hydroxide

HCl

Hydrochloric acid

DAPI

4’,6-diamidino-2-phenylindole

EDC

(1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride

NHS

Hydroxy succinimide

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphynyltetrazolium bromide

PBS

Phosphate Buffer Saline

FBS

Fetal Bovine Serum

EBM

Endothelial Cell Growth Basal Medium

DMEM

Dulbecco’s Modified Eagle Medium

EMEM

Eagle’s Minimum Essential Medium

QCM-D

quartz crystal microbalance with dissipation

HUVEC

human primary umbilical vein endothelial cell line

U87MG

Uppsala 87 malignant glioma cell line

C8-D1A

immortal mouse clone type I neuronal astrocyte cell line

TEER

transendothelial electrical resistance

ELISA

enzyme linked immunosorbent assay

TEM

transmission electron microscopy

DLS

dynamic light scattering

FCS

fluorescence correlation spectroscopy