Poster S14 - Metabolic reprogramming drives melanoma differentiation: Genome-scale modeling perspective

Introduction: Melanoma exhibits significant heterogeneity in terms of its molecular and clinical characteristics. This heterogeneity poses challenges in understanding the underlying mechanisms driving melanoma progression and in developing effective therapeutic strategies. Previously, Tsoi et. al. (2018, Cancer Cell) identified a melanoma differentiation trajectory, revealing distinct molecular subtypes that correspond to different stages of melanocyte differentiation. These subtypes were shown to exhibit differential drug sensitivities, highlighting the potential for subtype-specific therapeutic approaches. Specifically, melanocyte differentiation is shown to be correlated to sensitivity to drugs that induce a type of cell death known as ferroptosis. Cellular metabolism plays a crucial role in cancer development and progression, with cancer cells often exhibiting metabolic reprogramming to support their growth, survival, and response to drug treatments. Genome-scale metabolic models are powerful tools for investigating these metabolic changes of cell behavior. This study aims to understand the metabolic basis of melanoma differentiation from a systems biology perspective. By integrating data from melanoma cell lines representing different stages of differentiation into a human metabolic model, we seek to identify metabolic reactions and pathways that exhibit consistent changes along the differentiation trajectory.

Materials and

Methods: Gene expression profiles of 16 melanoma cell lines, selected based on their position along a differentiation trajectory defined by Tsoi et al., were obtained from the DepMap Portal and integrated into the Human1 genome-scale metabolic model using ftINIT algorithm, individually. The 16 context-specific models predicted were then examined to identify metabolic differences related to pathways that were either present or absent in specific cases, highlighting changes that might arise with differentiation. Parsimonious flux balance analysis was then employed to simulate the growth of the cell lines and predict their maximum biomass production rates. To explore the metabolic flux space under suboptimal growth conditions and identify potential metabolic adaptations, the upper bound for biomass production was set to half of its maximum value for each cell line. Riemannian Hamiltonian Monte Carlo sampling was performed to sample the constrained flux space and obtain flux distributions for each reaction. Wasserstein distance matrices, which quantify the dissimilarity between probability distributions, were calculated to measure the differences between the flux distributions of individual reactions across the cell lines. Finally, a linear regression model was fitted to these distance matrices to identify reactions whose flux distributions exhibited consistent changes along the differentiation trajectory, potentially implicating them as drivers of the differentiation process.

Results, Conclusions, and Discussions: The integration of omics data and genome-scale metabolic modeling (Figure 1) revealed distinct metabolic profiles across the melanoma differentiation trajectory. Analysis of the context-specific metabolic models for each differentiation state uncovered differences in network topologies, suggesting that melanoma cells undergo metabolic reprogramming as they progress along the differentiation path. Notably, reactions involved in lipid metabolism, redox balance, and energy production exhibited flux distributions that consistently changed along the differentiation trajectory. Furthermore, we identified metabolic reactions that are essential for the growth and survival of melanoma cells at different differentiation states. These findings suggest that the identified metabolic processes may play a crucial role in driving melanoma differentiation and sensitivity to therapeutic approaches. Taken together, these results demonstrate the utility of integrative genome-scale metabolic modeling to identify metabolic drivers of melanoma differentiation. In conclusion, by combining gene expression data and data-driven computational analyses, we relate cell state and metabolic flux profile. This approach provides novel insights into the metabolic drivers of melanoma differentiation and paves the way for the development of subtype-specific therapeutic strategies targeting the metabolic vulnerabilities of different melanoma subtypes. Additionally, our approach can be applied to other cancers and cell types.

Acknowledgements (Optional):