Humanized Biomimetic Nanovesicles for Neuron Targeting

  If we ever could figure out the signal that causes neuroplasticity this could be a good delivery vehicle,even better if we magnetized them.

Humanized Biomimetic Nanovesicles for Neuron Targeting

Assaf Zinger, Caroline Cvetkovic, Manuela Sushnitha, Tomoyuki Naoi, Gherardo Baudo, Morgan Anderson, Arya Shetty, Nupur Basu, Jennifer Covello … See all authors

First published: 11 August 2021

https://doi.org/10.1002/advs.202101437

Zinger and Cvetkovic contributed equally to this work.

Sections

PDF

Tools

Share

Abstract

Nanovesicles (NVs) are emerging as innovative, theranostic tools for cargo delivery. Recently, surface engineering of NVs with membrane proteins from specific cell types has been shown to improve the biocompatibility of NVs and enable the integration of functional attributes. However, this type of biomimetic approach has not yet been explored using human neural cells for applications within the nervous system. Here, this paper optimizes and validates the scalable and reproducible production of two types of neuron-targeting NVs, each with a distinct lipid formulation backbone suited to potential therapeutic cargo, by integrating membrane proteins that are unbiasedly sourced from human pluripotent stem-cell-derived neurons. The results establish that both endogenous and genetically engineered cell-derived proteins effectively transfer to NVs without disruption of their physicochemical properties. NVs with neuron-derived membrane proteins exhibit enhanced neuronal association and uptake compared to bare NVs. Viability of 3D neural sphere cultures is not disrupted by treatment, which verifies the utility of organoid-based approaches as NV testing platforms. Finally, these results confirm cellular association and uptake of the biomimetic humanized NVs to neurons within rodent cranial nerves. In summary, the customizable NVs reported here enable next-generation functionalized theranostics aimed to promote neuroregeneration.

1 Introduction

Restoration of neural function after traumatic injury, neurodegeneration, or neuroinflammation is currently hindered by a lack of effective and clinically practicable biotechnologies for precise, cell-targeted therapies or diagnostics. As such, there remains a need for biotechnological breakthroughs that can enhance and sustain the delivery of therapeutic cargos (e.g., genetic material and chemical compounds using nanomaterials),^[1] while also mimicking the microenvironment of the brain to avoid foreign body response.^[2] One promising pathway is the utilization of nanotechnologies inspired by nature, more commonly referred to as bio-inspired or biomimetic tools. By mimicking the composition and biological functions of the cells in our body, biomimetic tools avoid potential side effects that occur from systemic administration of potential therapeutics or imaging tools, such as the inflammation that can occur when using viral-based delivery approaches.^{[3, 4]} Thus, they offer the opportunity to gain insight into potentially safer and more tractable methodologies.^[5] For example, cell-derived exosomes are promising drug delivery systems as they are one mechanism for natural extracellular communication.^[6-10] Still, new approaches are needed, given that the complexity and variability of biomimetic tools from cellular sources appropriate for the nervous system reduces their potential for scalable precision medicine.^[11]

Functionalized nanoparticles (NPs) have high potential as well-defined carriers for the selective and targeted delivery of therapeutic cargo to neural cells due to their size scale.^{[12, 13]} For example, NPs have been used for the functional delivery of drugs to the rodent brain in multiple pathologies.^[14-16] Various surface modifications, such as coupling targeting peptides^[17] or antibodies^[18] to NPs or modifying surface charge for selective neuron-specific targeting,^[19] have been employed to increase targeting efficacy. Alternatively to NPs, exosome-like lipid nanovesicles (NVs) have been used as both contrast agents and drug delivery vehicles to the brain while mimicking neural cellular communication.^{[20, 21]} However, standardized protocols for the storage and characterization of NVs have yet to be fully established,^{[22, 23]} while the low yield from biological sample sources^[24] reduces the potential of translating NVs to clinical applications. Here, we sought to devise and optimize enhancements to existing lipid NVs using a well-defined and scalable cell source.

Previously, we endeavored to achieve enhanced bioactivity targeted to specific cell types by developing innovative hybrid biomimetic NVs that took advantage of specific cell types (e.g., native cellular targeting moieties) and synthetic NPs (e.g., ease of fabrication, scalability, and reproducibility) while bridging the gaps in therapeutic translation.^{[5, 23]} In particular, we demonstrated that the incorporation of leukocyte-derived plasma membrane proteins into the phospholipid bilayer of NVs enables immune system avoidance and association with inflamed endothelial cells while delivering a therapeutic payload.^[25] We also determined the integration location and orientation of these membrane proteins on the NV lipid membranes and revealed an equal distribution of the cytoplasmic and exoplasmic domains on one representative leukocyte-derived membrane protein, CD11b.^[26] Moreover, the biomimetic properties of these NVs resulted from the transfer of cellular adhesion proteins to the surface of NVs, which then mediated protein–protein interactions with target cells.^[27] Given that cellular interactions of neurons are in part attributed to cell–cell binding of adhesion proteins at the cell membrane surface, a similar approach for targeting neural cell types holds promising potential.^{[28, 29]} Nonetheless, testing of this approach with human neural cells (in order to aid in potential clinical translation) has been hampered by the lack of pure cell sources for reproducible and scalable production. Recent advances in the differentiation of human pluripotent stem cells (hPSC) into specific neural cell types^{[30, 31]} may enable the generation of biomimetic NVs and experimental testing platforms which, compared to platforms using rodent-derived neural cells, would be less likely to induce an immunogenic reaction in humans and thus more appropriate for clinical translation.

Based on this premise, we have developed and defined a new class of biomimetic human neural NVs (a.k.a. neurosomes) using a reproducible and scalable protein source from a pure population of rapidly derived hPSC-derived excitatory cortical neurons (iNeurons). Specifically, we used a bottom-up microfluidic-based synthesis method to bioengineer our novel NVs by combining phospholipids with the membrane proteins from iNeurons. We found that incorporation of neuron-derived membrane proteins does not affect the physicochemical properties of NVs and, in fact, enhances their uptake into cultured neurons. We further confirmed proof-of-principle cellular targeting efficacy both in vitro and in vivo using sphere (a.k.a. organoid) cultures and direct administration to the rodent trigeminal ganglion, respectively.^[32] These studies advance the current paradigm of NV bioengineering for improved cellular targeting within the nervous system.

2 Results

2.1 Preparation of Cell Source for Membrane Proteins and Workflow Scheme

First, we generated a pure population of neurons by directly inducing a genetically engineered hPSC line containing a doxycycline (dox)-inducible neurogenin 2 (ngn2) transgene (Figure 1A), as previously described in our established protocols,^{[30, 31]} to serve as a membrane protein source for functionalization of NVs. These hPSCs were directly induced into a uniform population of cortical glutamatergic excitatory neurons (iNeurons) with distinctive neuronal morphology by seven days of induction in a neural-supportive medium. To determine whether bioengineered proteins can be produced in cells and transferred to NVs, a stable membrane-bound green fluorescent protein (memGFP) transgene was incorporated via lentiviral delivery into hPSCs (Figure S1A, Supporting Information), which exhibited sustained memGFP expression during the differentiation process (Figure 1B). Proteins from the membranes of differentiated iNeurons and the parental hPSCs were extracted (Figure S1B, Supporting Information) for integration into NV lipid bilayers in order to generate two groups of biomimetic NVs: “neurosomes” (neuro-, N) and “plurisomes” (pluri-, P). As a control group, we prepared “liposomes” (lipo-, L) (i.e., NVs without incorporated membrane proteins) for comparison.

Figure 1

Open in figure viewerPowerPoint

Physiochemical and biomimetic characterization of neural biomimetic NVs. A) iNeurons were directly generated from a genetically engineered human pluripotent stem cell (hPSC) line containing a doxycycline (dox)-inducible neurogenin 2 (ngn2) transgene. A pure population was obtained within 7 days of differentiation. B) A stable membrane-bound green fluorescent protein (memGFP) transgene was incorporated into the hPSC line to track protein carry-over. Scale: 100 µm. C) A microfluidic approach was utilized for the synthesis of neural biomimetic NVs with cell-specific membrane proteins and two different lipid formulations (i.e., A and B). Three NV groups were fabricated using each lipid formulation: “liposomes” (lipo-, L) containing no protein, “plurisomes” (pluri-, P) containing hPSC-derived proteins, and “neurosomes” (neuro-, N) containing iNeuron-derived proteins. D) Immunoblotting revealed the transfer of mem-GFP in plurisomes and neurosomes (NVs originating from hPSCs and iNeurons, respectively) as well as the transfer of neuronal membrane protein (MP) marker NCAM1 in neurosomes of both formulations. (Bands are replicated from Figure S1E (Supporting Information), with dividing lines indicating splicing from original image.) E) Cryo-TEM images illustrated that all NV formulations had similar lipid bilayer morphologies containing a spherical bilayer structure. Scale: 50 nm. F) Physiochemical properties including NV size, PDI, and zeta potential were assessed. Though neither NV size nor PDI were significantly altered between NVs of different lipid formulations, NVs from lipid formulation B displayed a less negative zeta potential (n = 3–7 independent NV batches per group; see Figure S1F in the Supporting Information). For Figure 1F, results are shown as mean ± SEM. One-way ANOVA followed by Tukey's multiple comparison test was used to determine statistical probabilities between NVs of different protein sources within the same formulation (A or B), with *p < 0.05.