EXPLORING ADVANCED PHARMA TECHNOLOGIES: FROM GENE DELIVERY TO IMPLANTABLE SYSTEMS AND ARTIFICIAL INTELLIGENCE-BASED TOOLS

In this PhD thesis, μMESH was investigated as a platform for the co-delivery of multiple drugs, ranging from small drugs to nanomedicines. The taxanes paclitaxel (PTXL) and docetaxel (DTXL) were loaded into the μMESH as free drugs and in nanomedicine formulations. In in vitro studies, the free forms of the drugs showed a more sustained release profile over 150 days and demonstrated a higher efficacy in halting the proliferation of the brain tumor in orthotopic models of GBM. Additionally, lipid nanoparticles (LNP) encapsulating plasmid DNA targeting GBM chemoresistance to TMZ were assessed for their spatial distribution, stability and release from the water-soluble layer of the μMESH. In particular, the plasmids were specifically designed to target the human MGMT and PARP1 promoters in two GBM cell lines, U87-MG and U118. In parallel with the applied research projects, a novel protocol based on the PULCON method was developed and validated to precisely quantify, within a routine Nuclear Magnetic Resonance analysis, the concentration of polymers and therapeutic agents in various delivery systems. The PULCON method, alongside gravimetric and potentiometric analysis, and electron microscopy and In vivo imaging, was used to assess the biodegradation kinetics of the μMESH, comparing the rate with platforms defined by other patterns. The architecture of the 20×20 µm μMESH enables closed adhesion to the cell surface, acting as a “fishing net”, and ensures a prolonged permanence of the device ̶ over 2 months ̶ making it ideal for the sustained drug release. Upon validating the loading of LNP in an implantable device, their transport along the axonal projections of neurons was also investigated, as an alternative strategy for direct brain delivery, bypassing the hematoencephalic barrier. Axonal transport, commonly used to move nutrients and cellular vehicles, is also exploited by viral vectors to infect the central nervous system’s cells. Herein, LNP carrying a lipid-dye and a fluorescent RNA molecule were monitored for their trafficking in primary cortical neurons cultured into compartmentalized microfluidic chips. LNP were efficiently taken up by the axonal termini and retrograde-transported to the soma of the neurons, resulting in the gradual accumulation of both the lipid-dye and the fluorescent cargo. Interestingly, the lipid-dye appeared more visually diffused than the signal of the RNA, as it was diluted between the various cellular membranes. Additionally, as an emerging and promising tool, this thesis also focused on the possible role of Artificial Intelligence (AI) in modern drug delivery and its potential impact on the daily life work of researchers at the bench. A comprehensive database of different formulations, lipidic and polymeric nanoparticles, was generated through the use of a microfluidic device. A predicting model was optimized through algorithm training, by providing the formulation parameters (flow rate, flow rate ratio and reagent concentrations) as input and the properties of the final particles (hydrodynamic diameter, polydispersity and encapsulation efficiency) as output. The ability of the model to provide optimal features for unseen formulations with selected attributes (e.g. diameter of 100 nm) was compared to two well-known generative AI platforms, such as ChatGPT and Google’s Bard. Surprisingly, the developed model and Bard succeeded with negligible errors, while ChatGPT was completely off-target. Generating a database consisting of already prepared formulations and using AI-based models can positively impact improving and speeding up the development of customized formulations.