An adaptable in silico ensemble model of the arachidonic acid cascade

TL;DR

This paper introduces a flexible computer model of the arachidonic acid cascade that can adapt to different cell types and conditions, improving understanding of inflammatory processes.

Contribution

The first ensemble modeling approach applied to the arachidonic acid cascade, enabling adaptable and tuneable simulations.

Findings

The model accurately predicted eicosanoid production in different cell types and under various stimuli.

Quantitative agreement between predictions and experiments can be improved by expanding the model's network.

The model demonstrates the value of integrating in silico and in vitro methods for studying complex biological systems.

Abstract

Eicosanoids are a family of bioactive lipids, including derivatives of the ubiquitous fatty acid arachidonic acid (AA). The intimate involvement of eicosanoids in inflammation motivates the development of predictive in silico models for a systems-level exploration of disease mechanisms, drug development and replacement of animal models. Using an ensemble modelling strategy, we developed a computational model of the AA cascade. This approach allows the visualisation of plausible and thermodynamically feasible predictions, overcoming the limitations of fixed-parameter modelling. A quality scoring method was developed to quantify the accuracy of ensemble predictions relative to experimental data, measuring the overall uncertainty of the process. Monte Carlo ensemble modelling was used to quantify the prediction confidence levels. Model applicability was demonstrated using mass spectrometry mediator lipidomics to measure eicosanoids produced by HaCaT epidermal keratinocytes and 46BR.1N dermal fibroblasts, treated with stimuli (calcium ionophore A23187), (ultraviolet radiation, adenosine triphosphate) and a cyclooxygenase inhibitor (indomethacin). Experimentation and predictions were in good qualitative agreement, demonstrating the ability of the model to be adapted to cell types exhibiting differences in AA release and enzyme concentration profiles. The quantitative agreement between experimental and predicted outputs could be improved by expanding network topology to include additional reactions. Overall, our approach generated an adaptable, tuneable ensemble model of the AA cascade that can be tailored to represent different cell types and demonstrated that the integration of in silico and in vitro methods can facilitate a greater understanding of complex biological networks such as the AA cascade.

Ensemble modelling approaches, which account for the uncertainty surrounding model parameters, were applied to the arachidonic acid cascade for the first time. The adaptable, tuneable model was tailored to represent different cell types and stimuli.

Figures