Item

Holistic complex systems modelling approaches for cell signalling networks – Mammalian cell cycle control system : A thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy at Lincoln University

Abroudi, Ali
Date
2019
Type
Thesis
Fields of Research
ANZSRC::080108 Neural, Evolutionary and Fuzzy Computation , ANZSRC::080110 Simulation and Modelling , ANZSRC::060111 Signal Transduction , ANZSRC::060102 Bioinformatics , ANZSRC::0802 Computation Theory and Mathematics
Abstract
Cell cycle is a precisely regulated process in which cells of living organisms go through a sophisticated growth and division cycle that eventually leads to production of two daughter cells (Morgan, 2007; Satyanarayana & Kaldis, 2009). The progression of cell cycle is monitored through different checkpoints to guarantee the genome integrity (Kastan & Bartek, 2004; Novák et al., 2001; Nyberg et al., 2002; Walworth, 2000). Understanding the underlying mechanism and behaviour of these checkpoints is one of the most important topics in biology. In this study, Ordinary Differential Equation (ODE) and Petri Net (PN) modelling techniques and numerical analyses have been used to gain deeper insights into (i) the efficiency of checkpoints in detecting damaged/healthy cells, (ii) identifying the most significant parts of the system on G1-S and G2-M transitions, (iii) and the behaviour of the whole mammalian cell cycle control system (not just a part of it) from a systems perspective where all the crucial cell cycle sub-systems interact with each other to control cell cycle progression. This research begins with identifying the current gaps in the field of mammalian cell cycle modelling. The findings led us to developing a comprehensive mathematical model with all the missing essential components to integrate the important cell cycle sub-systems (Growth Factor, G1-S checkpoint, G2-M checkpoint and DNA damage signalling sub-systems) and their constituent modules (i.e., modules of G2-M sub-system are: Cdk1_Related, Tyrosine Phosphatase, Tyrosine Kinase, APC-Related and Plk1-Related modules). This arrangement maintains the functionality of the cell cycle control system while giving a better understanding of the underlying interactions by lessening the complexity of the system as a whole. The proposed comprehensive mathematical model was evaluated and further analysed in order to investigate the behaviour of mammalian cell cycle system and DNA damage response as well as to verify the role of the newly added components in cell cycle. The next part of this research is devoted to system behaviour analysis using parameters of the comprehensive mathematical model presented in the first section of the thesis. Therefore, we developed an analytical model (named SOMCCA) based on Self-Organizing Map (SOM) and Correlation Coefficient Analysis (CCA) to perform Global Sensitivity Analysis (GSA) on model parameters. Through this analysis, the most significant parameters, modules and sub-systems on G1-S and G2-M transitions were identified. The results of the SOMCCA model revealed that two sub-systems (the Growth Factor signalling and G1-S checkpoint) and seven parameters in the modules within them are significant on G1-S and G2-M transitions. The third part of this research investigates the efficacy of cell cycle checkpoints in correctly detecting damaged and healthy cells under DNA damage condition as there is biological evidence that G2-M checkpoint is relatively more efficient in detecting damaged/healthy cells in comparison to G1-S checkpoint. To do this analysis, we developed a model called Checkpoint Efficiency Evaluation (CEE) model based on statistical Type II error and Probability Density Function (PDF) of Peak Times (PTs) of G1-S and G2-M indicators (CycE_Cdk2 and CycB_Cdk1, respectively). The CEE model was applied to the comprehensive mathematical model presented in the first part of the thesis and the results were in good agreement with biological findings. Having developed a comprehensive mathematical model in the first part of this thesis, the main focus of the last part of the thesis is to perform a model abstraction in order to develop an abstract minimised model where the main characteristics of the model (such as dynamics of cell cycle core components, G1-S and G2-M transitions and response to DNA damage) remain qualitatively the same. First step towards model abstraction was to determine the key components that should be present in the abstract model. This was done based on the most significant parameters of the comprehensive model identified in the second part of the thesis. The next step was to determine the most suitable modelling approach. The presence of different time scales (from quick activations to slow synthesis processes) as well as presentation at different levels (from sub-systems down to modules and proteins) led us to choose Petri Net (PN) modelling for model abstraction because PN can be developed as a hybrid model and it is also an intuitive graphical approach that has the ability to present complex systems at different levels of abstraction. Therefore, we developed a hybrid PN-based model called Multi-Level Hybrid Petri Net (MLHPN). The MLHPN model has just four equations while the comprehensive model has 61 equations. In order to scrutinize the efficiency of G1-S and G2-M checkpoints in correctly detecting damaged and healthy cells, the CEE model was applied to the MLHPN model, and similar to the analysis results for the comprehensive model, the results for the MLHPN model showed that G2-M is more efficient than G1-S. In conclusion, this study showed the value of computational modelling (Ordinary Differential Equations (ODE) and Petri Net (PN)-based models) and Artificial Neural Networks/statistical methods (Self Organising Maps and Correlation Coefficient Analysis (SOMCCA) and Checkpoint Efficiency Evaluation (CEE) models) in comprehending complex signalling networks and obtaining deeper insights into the underlying mechanisms of the mammalian cell cycle control system from a systems point of view integrating all the essential cell cycle sub-systems including G1-S and G2-M checkpoints, the Growth Factor signalling and the DNA damage signalling pathway. Furthermore, the thesis showed the importance of newly added components and modules along with the Artificial Neural Networks and statistics based numerical investigation to identify the most significant parts of the system on system response. The demonstration of the value of the comprehensive mathematical model and the MLHPN model in investigating the efficiency of G1-S and G2-M checkpoints and verifying the greater efficiency of G2-M checkpoint in correctly detecting damaged/healthy cells is an important contribution of this research. Finally, this research demonstrated the value of model abstraction from a comprehensive model with 61 equations down to a hybrid PN-based model with just four equations while the system characteristics remained the same.
Source DOI
Rights
https://researcharchive.lincoln.ac.nz/pages/rights
Creative Commons Rights
Access Rights