Computational approaches for drug repurposing against cancer

Cancer remains a leading cause of death worldwide, particularly in its advanced and metastatic stages, where resistance to current treatments is common. In response to this challenge, drug repurposing offers a viable approach for cancer therapy, leveraging existing drugs approved for treating other conditions to reduce development risks, costs, and time. Notably, cationic amphiphilic drugs (CADs), commonly used for allergies and psychiatric disorders, have shown potential as anti-cancer agents. CADs target and destabilize lysosomal membranes in cancer cells, inhibiting enzymes such as acid sphingomyelinases (ASM) and other lysosomal lipases. This vulnerability leads to cell death, making them promising candidates for cancer treatment. In the context of lysosomal function, the role of ASM is particularly significant in Niemann-Pick disease, a genetic disorder characterized by sphingomyelin accumulation in the lysosomes of several tissues due to deficient ASM activity. Indeed, as for cancer, the growing number of variants of uncertain significance (VUS) identified in patient data highlights the need for an in-depth understanding of these variants and their roles in this disease.
Another example of successful repurposing is disulfiram (DSF), traditionally used for treating alcoholism. One of the metabolites of DSF, the diethyldithiocarbamate-copper complex (CuET), specifically accumulates in cancer cells and targets the NPL4 protein, a crucial component of the p97 segregase complex involved in protein degradation within the proteasome. The interaction between CuET and NPL4 disrupts the function of the protein function, leading to cancer cell death.
These cases emphasize the potential of repurposed drugs in developing new cancer treatments, providing a novel approach to combating this global health challenge.
This thesis represents the first step in investigating the molecular mechanisms underlying the tumor-suppressing action of CADs and DSF that our collaborators in the Unit of Cell Death and Metabolism (CDM) and in the Unit of Genome Integrity (GI) at the Danish Cancer Institute (DCI) have identified, using computational methods from the field of molecular modeling and simulations.
Five manuscripts are included. Manuscript I laid the groundwork for the future investigation of CADs and lysosomal membranes. We developed a Python-based pipeline to streamline the analysis of molecular dynamics (MD) simulations, which can handle biological membranes with diverse compositions and account for the presence of proteins. The analyses include different biophysical membrane properties that can be better rationalized using supporting plotting tools. Manuscript II presents the MAVISp framework, specifically designed for interpreting VUS, which holds particular relevance in cancer research. This extensive framework includes several modules, each capable of analyzing individual protein structures, their complexes, and ensembles of structures.
These modules are uniquely configured to predict the impact of specific mutations on protein function or structural stability. My contribution to this work included curating and validating the MAVISp database, focusing particularly on a case study on the NPL4 protein, and developing two Python-based pipelines for protein alignments and computation of EVE pathogenicity scores, used for assessing the potential disease-causing impact of human variants. In Manuscript III, we applied the MAVISp framework focusing on the ASM protein, one of the targets of the thesis. Over 400 ASM variants, identified from ClinVar, literature curation, or cancer samples, were studied, creating a comprehensive atlas of their structural effects on ASM. Additionally, the study offers a reassessment of several previously known variants. Manuscript IV focuses on MD simulations to characterize the interaction between the ASM protein and a lysosomal-like membrane. This study aimed to understand the impact of ebastine, chosen as representative of CADs, on both the lysosomal membrane and the ASM protein. Ultimately, Manuscript V presents a thorough comparison of various force-field parameters, including CHARMM36m, ff99SB-ILDN, and f99SB*-ILDN, evaluating their efficacy in MD simulations for depicting the structure and dynamics of yeast NPL₄. The study encompasses the modeling of zinc ions, a cofactor of NPL4, and cupric ions, which are hypothesized to play a role in the mechanism where copper might replace zinc in the Zinc Finger (ZF) domains of NPL₄.
Overall, these manuscripts collectively contribute to the field of drug repurposing by providing innovative computational tools to evaluate MD simulations of biological membrane and protein, and in-depth analyses of relevant protein structures, protein-drug interactions, and the structural effect of specific variants.