Systematic drug perturbations on cancer cells reveal diverse exit paths from proliferative state

During a cell state transition, cells travel along trajectories in a gene expression state space. This dynamical systems framework complements the traditional concept of molecular pathways that drive cell phenotype switching. To expose the structure that hinders cancer cells from exiting robust proliferative state, we assessed the perturbation capacity of a drug library and identified 16 non-cytotoxic compounds that stimulate MCF7 breast cancer cells to exit from proliferative state to differentiated state. The transcriptome trajectories triggered by these drugs diverged, then converged. Chemical structures and drug targets of these compounds overlapped minimally. However, a network analysis of targeted pathways identified a core signaling pathway - indicating common stress-response and down-regulation of STAT1 before differentiation. This multi-trajectory analysis explores the cells' state transition with a multitude of perturbations in combination with traditional pathway analysis, leading to an encompassing picture of the dynamics of a therapeutically desired cell-state switching.

The GEDI heatmap project a high-dimensional transcriptome into a 2D self-organizing map to visualize the overall gene expression. Each pixel in the GEDI map (grid element) represents a mini-cluster of highly similarly behaving genes. The pixel at the same position in each map represents the same genes. The color of each pixel represents the gene expression level. (Both Befemelane and Trimebutine have one transcriptome missed due to the failure to pass the quality control) Figure S5. Heatmap of the differentially expressed genes identified using SAM.   Day 1 CAP-Net shows the downstream commonly affected pathways after 1 day drug treatments. Diamonds and circles respectively represent targets and differentially expressed genes. The color of a node shows the average expression level of the corresponding gene. If a gene has a stemness function, it is marked by bold black line.

IFI27
A protein that promotes cell death and mediates IFN-induced apoptosis, characterized by a rapid and robust release of cytochrome C from the mitochondria and activation of BAX and caspases 2, 3, 6, 8 and 9.

Inhibition [2,3]
VEGFA A growth factor that regulates angiogenesis. The CAP-Net shows a low expression of VEGFA, which indicates a reduction in the angiogenesis and tumor invasion.

CSF1
A cytokine that enhances the tumor growth via tumor-associated macrophages. It is highly expressed in several subtypes of breast cancer and causes the macrophage differentiation by stimulating VEGFA. The down-regulation of CSF1 in the CAP-net supports previous findings as a positive effects on the breast cancer treatment.

AKT1
A Serine-Threonine protein kinase that regulates metabolism, proliferation, cell survival, growth and angiogenesis.
- [10] IL8 A chemotactic factor that attracts neutrophils, basophils and T-cells. It activates neutrophils. It is released from several cell types in response to an inflammatory stimulus.

EDN1
A protein that is used to produce vasoconstrictive peptides activation [15] HMOX1 An enzyme that cleaves the heme ring at the alpha methene bridge to form biliverdin, and exhibits cytoprotective effect since excess of heme sensitizes cells to apoptosis activation [16] MAOA A mitochondrial enzyme that degrades monoamines neuron transmitters and dietary amines, induces EMT through activation of VEGF Inhibition [17,18]

MUC1
A membrane protein that activated T-cells, influences the Ras/MAPK pathway, promotes tumor progression, regulates TP53-mediated transcription, determines cell fate in the genotoxic stress response and represses TP53 activity.