496. Elucidating Design Rules Governing Extracellular Vesicle-Mediated Therapeutic Protein Delivery

Extracellular vesicles (EVs) are secreted biological nanoparticles that have great potential as therapeutic delivery vehicles - they are well-tolerated in vivo and naturally capable of transferring RNA and proteins between cells. Our ability to engineer EVs as therapeutic delivery vehicles is limited by an incomplete understanding of how EVs load biomolecular cargo and deliver it to recipient cells. In particular, the biophysical rules governing mRNA and protein delivery by EVs have not been elucidated. Open questions include: Does size limit mRNA loading efficiency into EVs? To what extent is EV mRNA cargo translated in recipient cells? What factors impact the degree to which RNA and protein cargo are delivered to the cytoplasm of recipient cells?To quantitatively investigate the above questions, we leveraged our Targeted and Modular EV Loading (TAMEL) platform, which enables active loading of specific cargo RNA into EVs. TAMEL can enrich cargo mRNA loading into EVs up to 40-fold relative to passive loading. By directly comparing active loading efficiencies between mRNAs of different lengths, we characterized what type of RNAs can be loaded into EVs. While active loading of mRNA-length (> 1.5 kb) cargo molecules was significant, active loading was much more efficient for smaller (~0.5 kb) RNA molecules, providing the first direct evidence for the impact of cargo RNA size on loading into EVs. We next leveraged the TAMEL platform to elucidate the limiting steps in EV-mediated delivery of mRNA and protein to prostate cancer cells, as a therapeutically relevant model system. In this model system, we did not observe translation of EV-delivered mRNA in recipient cells, indicating this is a limiting step in functional delivery of EV cargo. In contrast, we observed robust EV-mediated delivery of dTomato reporter protein, and thus further explored EVs as therapeutic protein delivery vehicles.To probe the efficacy of EV-mediated therapeutic protein delivery, we investigated using EVs to deliver the prodrug converting enzyme cytosine deaminase fused to uracil phosphoribosyl transferase (CD-UPRT), which converts the prodrug 5-FC to the toxic 5-FU. Importantly, CD-UPRT can function without without requiring endosomal escape because both 5-FC and 5-FU are membrane permeable. We also explored strategies for EV-mediated delivery of Cas9 nuclease, which must overcome the EV loading barrier imposed by its NLS sequence as well as escape the endosome in recipient cells. To address the loading challenge, we investigated a strategy for conditional NLS reconstitution to allow enhanced loading of Cas9 into EVs. We then used Cas9 delivery to assess the degree to which EV-delivered proteins can escape the endosomal/lysosomal pathways and traffic to other subcellular locations. Altogether, our investigations elucidated key design rules and central limiting steps that may guide the further development and utilization of EVs as therapeutic biomolecule delivery vehicles.