PMC:4331679 / 1524-7149 JSONTXT

{"project":"2_test","denotations":[{"id":"25707537-20221256-14870708","span":{"begin":1730,"end":1731},"obj":"20221256"},{"id":"25707537-19426782-14870709","span":{"begin":2055,"end":2056},"obj":"19426782"},{"id":"25707537-22967854-14870710","span":{"begin":2211,"end":2212},"obj":"22967854"},{"id":"25707537-22967854-14870711","span":{"begin":2987,"end":2988},"obj":"22967854"},{"id":"25707537-22579283-14870712","span":{"begin":3147,"end":3148},"obj":"22579283"}],"text":"Introduction\nThe objective of anti-cancer therapeutics is to kill cancer cells with minimum damage to the healthy cells. To this end, a solid understanding of the cell fate decisions (e.g. apoptosis, proliferation) of different cells under various conditions would be required. It is well known that signalling pathways play crucial roles in the regulation of cancer cell fate [1]. However, it is challenging to understand the dynamics of signal transduction at systems level, due to non-linearity of the network dynamics, e.g. feedback and crosstalk. In cancer cells, this becomes even more complicated due to various types of alterations (e.g. DNA mutations, genome rearrangement, epigenetic changes, and pathway alterations). These alterations allow cancer cells to adapt to new conditions and evolve drug resistance. Therefore, to find effective anti-cancer therapies, cancer-specific alterations in the signalling pathways must be taken into account. Moreover, it is desirable to understand how cancer cells respond to different combinations of drugs, and how drug sensitivity can be enhanced. Genomic and proteomic data of cellular responses to drugs, synchronized with cell fate observations, would shed light on cancer drug effects at systems level. However, even if sufficient data are available, it is challenging to construct a model of cell signalling to explain the data and make accurate predictions. As more \"omics\" data about cancer are available recently, computational methods for modelling and discovery of cancer cell fate are becoming more important.\nCancer cell fate in response to drugs has been studied with both discrete and continuous models using knowledge driven approaches. For example, the study in [2] focused on discrete modelling of the apoptosis network, by constructing a model for cell fate with 25 key regulatory genes (e.g. Casp3, BCL2, XIAP, etc.). Cell survival pathways with cell death (necrosis and apoptosis) pathway were combined to model three cell fates, namely apoptosis, proliferation and survival. GINsim [3] software was used to perform simulation to assess the importance of Cytokines (e.g. TNF, FASL) in deciding the cell fate. In another study, Hong et al. [4] analysed the continuous model of apoptosis triggered by the drug Cisplatin. They initially constructed apoptosis pathways using existing literature. This model was configured to respond to the external stimulus, i.e. the drug Cisplatin. To analyse the functions of various signalling pathways and their crosstalk, Hong et al. integrated three apoptosis pathways, namely death receptors (e.g. FasL) induced pathway, mitochondrial pathway and ER stress pathway. Using their differential equation based model they found that the apoptosis caused by Cisplatin was dependent on doses and time. The level of apoptosis was almost stagnant at higher concentration of drug. They also observed that mitochondrial pathway has strongest effect on apoptosis (among the three pathways) [4].\nOn the other hand, data driven modelling of signalling pathways is a promising approach to uncover regulatory mechanism for cancer cell fate. The study in [5] investigated three lines of breast tumour, namely BT-20, MDA-MB-453 and MCF7, for their responses to various combinations of exposure to several genotoxic drugs and signalling inhibitors. The authors found that it was the pre-treatment, rather than co-treatment or post-treatment, of a subset of TNBCs with Epidermal Growth Factor Receptor (EGFR) inhibitors that can enhance the sensitivity of tumour cells in their apoptosis response to DNA-damaging chemotherapy. The study suggested that such a treatment may lead to the rewiring of oncogenic signalling pathways which has the potential to make cancer cells more susceptible to death. It was further reported that the inhibition of EGFR in a time-staggered way may be responsible for sensitising tumour cells to genotoxic drugs. However, simultaneous co-administration of inhibitors and genotoxic drug could not make the tumour cells less tumorigenic. To test these hypotheses, the study systematically investigated a series of drug combinations for their effects on breast cancer cells. The study suggests that rewiring inside the tumour cells is responsible for the increase of drug sensitivity, but it is not clear where and how the rewiring happens. The rewiring involves alterations of signalling pathways such as addition or deletion of edges in the network, change in reactions rates, and change in the concentration of molecules. Since it is more challenging to directly observe rewiring signalling pathways by wet lab experiments, computational methods for predicting rewiring from data would be useful for study of cancer drug effect and cell fate decisions.\nIn this paper, we propose a hybrid modelling approach that combines the advantages of knowledge-driven and data-driven approaches for modelling cancer signalling pathways and cell fate decisions. From a generic apoptotic network, we simulate the time series data of signalling proteins using ODEs based on a network candidate modified from the generic network by in silico rewiring; then the simulated data are compared with the real data. This in in silico rewiring, simulation and data fitting is repeated iteratively to improve the goodness of fit using Genetic Algorithm. Through such an optimisation we can detect the topology alterations of the network that allows close fitting of model to the real data. As such, network rewiring can be inferred from real data. Most of rewiring events predicted by our model are supported by existing literature."}