Abstract
Polymer-based nanoparticles (PNP) are receiving increasing attention as potential cancer therapeutics to replace conventional cancer treatments. What differentiates PNP from current cancer therapeutics is their specificity towards cancer cells,which, among many advantages, increases the efficacy of the drug, lowers systemic toxicity, and allows for higher tolerable doses. Despite these advantageous characteristics, PNP still have a long way to go before becoming standardized cancer therapeutic delivery vehicles, as their mechanisms of action remain unknown. This thesis employs a multi-scale simulation approach to elucidate the mechanisms of action of PNP formed by amphiphilic block co-polymers. In particular, all-atom (AA) and coarse-grain (CG) simulations are used to understand key processes of PNP that will contribute to their application as cancer therapeutics, such as their self-assembly process, drug encapsulation, and selectivity towards cancer cells. These processes are extremely dynamic, making them either impossible or very hard to study using experimental techniques. The motivation behind this thesis is to contribute to the knowledge of the rational design of PNP and to create platforms to standardize and automate the analysis of PNP simulations, enabling scientists to apply these methods to any other PNP of interest.
A fundamental problem in PNP rational design is to deduce the overall micelle characteristics from the individual polymers, which are normally not the same. To address this, in this thesis AA MD simulations of the same PEO-PMA polymers (same monomer and polymer numbers) but arranged in a different topology were performed. Here, it was demonstrated that polymer topology plays a key role in the overall micelle physical characteristics, drawing a link between topologies
and specific micelle characteristics such as size or hydration. It was also shown that polymers that form a micelle with a clear hydrophobic core and hydrophilic corona, adopt location-specific polymer conformations. Furthermore, in this thesis, CG simulations were conducted on an experimentally validated PEG-PLGA NP loaded with anti-cancer peptides, aiming to understand its cargo encapsulation and selectivity toward cancer cells. Regarding drug storage, it was shown that the polymers in the PEG-PLGA NP also take location specific conformation, and that these conformations form local microenvironments within the NP that lead to the solubilization of the peptide in several storage locations. Additionally, to study the experimentally validated selectivity of this NP towards cancer cells, the NP was simulated with a model cancer and healthy membrane. From these simulations, the changes induced by the NP-membrane interactions on both, the NP physicochemical characteristics and membrane disruption, were quantified and compared. It was found that changes in both -NP and membrane- were more substantial in the cancer simulation, meaning that the selectivity of this NP could also be observed in silico. The key force driving the NP-membrane interactions was the interactions between the PEO polymers and a specific lipid species, that is present in higher percentages in cancer membranes. This suggests that preferential polymer-lipids interactions may play a vital role in the PNP selectivity towards cancer cells. Moreover, a set of parameters to assess the selectivity of PNP towards cancer cells in silico are proposed, which can be applied to other NP-membrane systems.
To conduct such a detailed analysis of polymer systems, it was necessary to develop novel analysis tools. To this end, a graph theoretical cluster algorithm was developed to track changes in polymer aggregation throughout a simulation. Also,
to study the specific polymer conformations within the micelle, dimensionality reduction unsupervised machine learning and clustering techniques were applied. Finally, to be able to analyse simulation properly, an algorithm to make molecular structures whole across the periodic boundary when its size is greater than half the box size was created. These analysis tools, along with others developed in this thesis, have been incorporated into the publicly available PySoftK software package. This way, PySoftK aims to provide an automated computational analysis workflow to study complex properties of soft-matter.
| Date of Award | 1 Apr 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Chris Lorenz (Supervisor) & Martin Ulmschneider (Supervisor) |