High Throughput and Cell-Free Protein Modification of Extracellular Vesicles for Improved Therapeutic Applications

Introduction: : Native extracellular vesicles (EVs) of mammalian cell origin are being tested for broad biomedical therapies; yet, each vesicle exposes fewer than twenty types of surface proteins. Geometric packing allows more than 1000 surface proteins on a 100nm EV; enriching just the first 500 slots will raise surface protein density by an order of magnitude and redirect vesicles toward diseased tissues. While several methods of EV surface modification are available, such as click chemistry, lipid insertion, or host cell engineering, these methods are tedious and challenging to scale-out to more than a few protein candidates. We circumvented this problem by utilizing cell-free protein synthesis. It enables users to provide DNA templates to the cell-free systems to add any desired proteins to EVs. In past work, our lab demonstrated the efficient display of ~39 EV proteins on liposome surfaces using cell-free protein synthesis. These domains are anchored to Aquaporin-Z (AQP), which facilitates co-translational insertion and anchoring to the membrane. Here, two donor lines of Placenta Mesenchymal Stem Cell (PMSC) EVs are modified using cell-free protein synthesis of therapeutically relevant proteins. These EVs are validated for protein display using single EV flow cytometry and super-resolution imaging. They are then studied using neuroprotection assays to determine the therapeutic effect and targeting of the modified EVs. The modified EV variants are also examined using mouse models. This work demonstrates the superiority of cell-free protein synthesis to add surface proteins to natural EVs, opening doors to rapid engineering and prototyping of modified EVs for therapeutic treatment.

Materials and

Methods: : Extracellular vesicles were generated from two PMSC lines using ultracentrifugation. Nanoparticle Tracking Analysis (NTA) was used to determine the size, standard deviation, and concentration of EVs before and after cell-free modification. Inserted proteins were generated by first developing a plasmid through golden gate synthesis, which incorporated a promoter, protein of interest, and terminator. Next, soluble/extracellular domains of various membrane proteins were synthesized using cell-free protein synthesis. When the EVs, cell-free systems, and plasmids were incubated overnight, the proteins were synthesized and inserted into the EV membrane. Protein display was validated using flow cytometry, ELISA, and super-resolution microscopy. We opted to use a universal HIS tag detection method. Single EV flow cytometry was done using Memglow 700-labeled EVs and a FITC HIS antibody for protein labeling. ELISAs were performed by using a HIS capture antibody and capturing whole extracellular vesicles that displayed the HIS protein, in which a primary and secondary antibody were added for detection. Success was quantified using absorbance measurements. Finally, the ONI Super-resolution microscope was used in conjunction with fluorescent extracellular vesicles and anti-HIS antibodies. EV function was characterized by a neuroprotection assay. Neuroprotection used SH-SY5Y cells damaged by staurosporine, with EVs being added and dendrite growth being quantified using Calcein-AM.

Results, Conclusions, and Discussions:: PMSC EVs were modified through cell-free protein synthesis (Figure 1A). Our cell-free system consisted of an E. Coli lysate, a supplemental salt buffer, and a plasmid coding for a protein of interest. When an EV membrane was present, co-translational insertion occurs where the protein of interest was generated and inserted into the membrane. NTA analysis showed no changes in the average size of the extracellular vesicles after cell-free protein synthesis (Figure 1B). Next, we determined if our cell-free system provided robust display of AQP-protein variants using single EV flow cytometry. Our system utilized extracellular membrane protein domains tagged to Aquaporin-Z to better facilitate co-translational insertion and anchoring. Our results showed that 10 different proteins can successfully be displayed on both sources of PMSC EVs (Figure 1C). These 10 proteins were selected from PMSC EV proteomics data and have been shown to interact directly with cells of the central nervous system. The efficiency of protein display ranged from 10 to 60% of the total EV population. To examine the effect the protein insertion had on the native EV membrane, we utilized super-resolution microscopy and stained our PMSC EVs pre- and post-protein modification for three tetraspanins. These EVs were imaged, and the number of fluorescent localizations was quantified (Figure 1D). There was no significant difference between native PMSC EV and cell-free modified EV regarding tetraspanin display. With robust display confirmed, we next showed that our cell-free modified EV can enhance or confer neuroprotection effects. We tested the original and modified EVs using a neuroprotection model, consisting of SH-SY5Y cells that were “damaged” by staurosporine. EVs were supplemented post-damage to examine if they can protect SH-SY5Y from the damage. We found that most of our modified EVs from both PMSC EV lines (>6 variants) resulted in a significant increase in neurite outgrowth, indicating neuroprotection. The top modified EVs with the highest effectiveness in vitro were applied to a mouse model to examine therapeutic efficacy and biodistribution. Our work demonstrates that cell-free protein synthesis systems enable direct anchoring of therapeutically relevant proteins to the surface of EVs, enabling the augmentation of EV functions.

Acknowledgements and/or References (Optional)::