From Systems Biology to Systems Medicine
Centre for Process Systems Engineering, Department of Chemical Engineering, Imperial College London, London SW7 2AZ, United Kingdom
Tel.: +44 (0)20 7594 6655
Systems biology is an established research field that seeks to integrate experimentation with computational approaches to improve our understanding of biological systems. In recent years it has been argued that the principles of this holistic approach are equally applicable to medical problems, where the biological system is perturbed from its native state. Systems medicine can help integrate disparate datasets, interpret the effect of mutations, and therefore identify opportunities for tailored approaches to treatment. In this work, we have focused on the unfolded protein response, a signalling cascade originating in the endoplasmic reticulum, which has been studied experimentally and computationally in the context of Bioprocessing. However, it is equally important in neurodegenerative diseases as its markers have been found to be upregulated in tissue samples from patients suffering from Alzheimer’s, Parkinson’s and Huntington’s disease. We have used an existing model for protein folding in the endoplasmic reticulum to examine how the accumulation of unfolded proteins can be relieved and have further conducted a sensitivity analysis to identify the rate-limiting steps in protein folding. Our results highlight the crucial role of protein disulfide-isomerase in protein processing in the organelle.
Systems biology involves the analysis of complex networks through the use of both biological experimentation and mathematical modelling, statistical or mechanistic, with a focus on how component interactions control system behaviour. There are several advantages to this holistic approach. Firstly, it facilitates the integration of data from a variety of experimental methods, e.g. transcriptomic data collected by quantitative PCR or DNA microarray technology, which provide relative gene expression levels, can be integrated with protein concentrations obtained with proteomic methods (e.g. mass spectrometry) to reveal regulation effects. Secondly, it allows the perturbation and dynamic analysis of the system, which is useful for understanding native and non-native behaviour arising, for example, from mutations (1). Thirdly, it enables us to question hypotheses. Although a mathematical model of a biological network, which tends to be significantly less complicated than the actual system it describes, can never be fully validated, it can be a useful means for testing various hypotheses by examining the fit of different model structures with experimental data (2). Certain hypotheses can be disproven, while others put forward for further scrutiny. Finally, a high-fidelity model can be used to guide experimentation through, for example, the design of experiments to distinguish between biological hypotheses, or the application of model simulation and analysis techniques to examine the effect of gene up/down-regulation or gene knockout/silencing (3).
The task of model validation or, more appropriately in this case, model confirmation, is not straight-forward. The availability of a high volume of experimental data does not guarantee successful model parameterisation. For this, the information-content of the data needs to be sufficient. In practice, this means that it is important to carefully determine which system perturbations will minimise the variance and maximise the information content of the data before a validation experiment is conducted. When the underlying biological system is well understood a mathematical model can be developed and used for optimal experimental design (4). Alternatively, statistical design of experiments can be used to generate informative data that can populate a statistical model or help establish a mechanistic model (5).
Recently, there have been several initiatives towards the application of systems biology principles to medicine. Ashall et al. (6) followed a combined experimental and computational approach to demonstrate that oscillations are an important characteristic of the response of nuclear factor kB (NF-kB) transcription factor, which regulates cellular stress responses and the immune response to infection, to, the addition of tumor necrosis factor-α to the cell culture environment, which mimics pulsatile inflammatory signals. Similarly, Toettcher et al. (7) mathematically described three molecular mechanisms known to control mammalian cell cycle arrest and used quantitative experimental measurements to show that the seemingly redundant mechanisms have, in fact, complimentary functions that together achieve a specific cellular outcome. Weston and Hood (8) predict that such integrative approaches will help advance predictive and personalised medicine. They argue that analysis of an individual’s genome and proteome will guide the introduction of pathogenic perturbations of common protein and gene regulatory networks, which will allow personalised predictions even before the onset of a disease.