Resistance is Futile: Physical Science, Systems Biology and Single-Cell Analysis to Understanding the Plastic and Heterogeneous Nature of Melanoma and Their Role in Non-Genetic Drug Resistance

Abstract

Melanoma is the most deadly form of skin cancer due to its great metastatic potential. Targeted therapy that inhibits the BRAF-V600E driver mutation has shown impressive initial responses in melanoma patients. However, drug resistance, as the universal phenomenon for any cancer therapy, always limits treatment efficacy and compromises outcomes. As the early-step of resistance development, non-genetic mechanisms enable cancer cells to transition into a drug-resistant state in as early as a few days after drug treatment without alteration of the genome. This early mechanism is, to a large extent, due to the heterogeneous and highly plastic nature of tumor cells. Therefore, it imperative to understand the plastic and heterogeneous nature of the melanoma cells in order to identify combination therapies that can overcome resistance. In this thesis, we investigate these two fundamental natures of non-genetic drug resistance using BRAF inhibition of BRAF-mutant melanomas as the model system. These melanoma cells undergo multi-step, reversible drug-induced cell-state transitions from the original sensitive phenotype to a drug-resistant one. We first conducted bulk analysis to characterize the detailed kinetics of the entire transition from drug-sensitive state towards drug-resistant state, revealing expression changes of thousands of genes and extensive chromatin remodeling. A 3-step computational biology approach greatly simplified the complexity and revealed that the whole cell-state transition was controlled by a gene module activated within just the first three days of drug treatment, with the RelA transcription factor driving chromatin remodeling to establish an epigenetic program encoding long-term phenotype changes towards resistance. From there, a detailed mechanism connecting tumor epigenetic plasticity with non-genetic drug resistance was resolved through in-depth molecular biology experiments. The mechanism was validated in clinical patient samples. We further investigated heterogeneity by moving from bulk cellular studies to single-cell analysis. The single-cell view further revealed that two driving forces from both cell-state interconversions and phenotype-specific drug selection control the cell-state transition dynamics. The single-cell studies also pinpointed the signaling network hub, RelA, as the driver molecule of the initiation of the adaptive transition. These two competing driving forces were further quantitatively modeled via a thermodynamic-inspired surprisal analysis and a modified Fokker-Planck-type kinetic model. Finally, using integrated single-cell proteomic and metabolic technology I developed to characterize the early-stage signaling and metabolic changes upon initial drug responses, we further identified two distinct paths connecting drug-sensitive and drug-tolerant states. Melanoma cells exclusively traverse one of the two paths depending on the level of MITF in the drug-naive cells. The two trajectories are associated with distinct signaling and metabolic susceptibilities and are independently druggable. In total, this thesis combines and synergizes various physical science and systems biology approaches together with several unique single-cell technologies and analysis to obtain a deep and comprehensive understanding of non-genetic drug resistance in cancer. The findings from this thesis provide several novel insights into the rational design of effective combination therapy for overcoming the development of resistance in response to cancer treatments.