Computational and chemical approaches to drug repurposing

Abstract

Drug repurposing, which entails discovering novel therapeutic applications for already existing drugs, provides numerous benefits compared to conventional drug discovery methods. This strategy can be pursued through two primary approaches: computational and chemical. Computational methods involve the utilization of data mining and bioinformatics techniques to identify potential drug candidates, while chemical approaches involve experimental screens oriented to finding new potential treatments based on existing drugs. Both computational and chemical methods have proven successful in uncovering novel therapeutic uses for established drugs. During my PhD, I participated in several experimental drug repurposing screens based on high-throughput phenotypic approaches. Finally, attracted by the potential of computational drug repurposing pipelines, I decided to contribute and generate a web platform focused on the use of transcriptional signatures to identify potential new treatments for human disease. A summary of these studies follows:

In Study I, we utilized the tetracycline repressor (tetR)-regulated mechanism to create a human osteosarcoma cell line (U2OS) with the ability to express TAR DNA-binding protein 43 (TDP-43) upon induction. TDP-43 is a protein known for its association with several neurodegenerative diseases. We implemented a chemical screening with this system as part of our efforts to repurpose approved drugs. While the screening was unsuccessful to identify modulators of TDP-43 toxicity, it revealed compounds capable of inhibiting the doxycyclinedependent TDP-43 expression. Furthermore, a complementary CRISPR/Cas9 screening using the same cell system identified additional regulators of doxycycline-dependent TDP43 expression. This investigation identifies new chemical and genetic modulators of the tetR system and highlights potential limitations of using this system for chemical or genetic screenings in mammalian cells. In Study II, our objective was to reposition compounds that could potentially reduce the toxic effects of a fragment of the Huntingtin (HTT) protein containing a 94 amino acid long glutamine stretch (Htt-Q94), a feature of Huntington's disease (HD). To achieve this, we carried out a high-throughput chemical screening using a varied collection of 1,214 drugs, largely sourced from a drug repurposing library. Through our screening process, we singled out clofazimine, an FDA-approved anti-leprosy drug, as a potential therapeutic candidate. Its effectiveness was validated across several in vitro models as well as a zebrafish model of polyglutamine (polyQ) toxicity. Employing a combination of computational analysis of transcriptional signatures, molecular modeling, and biochemical assays, we deduced that clofazimine is an agonist for the peroxisome proliferator-activated receptor gamma (PPARγ), a receptor previously suggested to be a viable therapeutic target for HD due to its role in promoting mitochondrial biogenesis. Notably, clofazimine was successful in alleviating the mitochondrial dysfunction triggered by the expression of Htt-Q94. These findings lend substantial support to the potential of clofazimine as a viable candidate for drug repurposing in the treatment of polyQ diseases.

In Study III, we explored the molecular mechanism of a previously identified repurposing example, the use of diethyldithiocarbamate-copper complex (CuET), a disulfiram metabolite, for cancer treatment. We found CuET effectively inhibits cancer cell growth by targeting the NPL4 adapter of the p97VCP segregase, leading to translational arrest and stress in tumor cells. CuET also activates ribosomal biogenesis and autophagy in cancer cells, and its cytotoxicity can be enhanced by inhibiting these pathways. Thus, CuET shows promise as a cancer treatment, especially in combination therapies.

In Study IV, we capitalized on the Molecular Signatures Database (MSigDB), one of the largest signature repositories, and drug transcriptomic profiles from the Connectivity Map (CMap) to construct a comprehensive and interactive drug-repurposing database called the Drug Repurposing Encyclopedia (DRE). Housing over 39.7 million pre-computed drugsignature associations across 20 species, the DRE allows users to conduct real-time drugrepurposing analysis. This can involve comparing user-supplied gene signatures with existing ones in the DRE, carrying out drug-gene set enrichment analyses (drug-GSEA) using submitted drug transcriptomic profiles, or conducting similarity analyses across all database signatures using user-provided gene sets. Overall, the DRE is an exhaustive database aimed at promoting drug repurposing based on transcriptional signatures, offering deep-dive comparisons across molecular signatures and species.

Drug repurposing presents a valuable strategy for discovering fresh therapeutic applications for existing drugs, offering numerous benefits compared to conventional drug discovery methods. The studies conducted in this thesis underscore the potential of drug repurposing and highlight the complementary roles of computational and chemical approaches. These studies enhance our understanding of the mechanistic properties of repurposed drugs, such as clofazimine and disulfiram, and reveal novel mechanisms for targeting specific disease pathways. Additionally, the development of the DRE platform provides a comprehensive tool to support researchers in conducting drug-repositioning analyses, further facilitating the advancement of drug repurposing studies.