Design principles in biological networks: Balancing efficacy with robustness

Successful systems, biological or otherwise, must be robust to uncertainties and disturbances within their local environments as well as within their internal networks. Evolution has enabled biological systems to develop defenses against experienced or perceived uncertainties, but these regulatory defenses ultimately complicate the means by which a system may recover once it has gone awry. Here, three biological signaling networks are explored via robustness analysis to identify ideal candidate targets for network manipulation.;A model of ovarian steroidogenesis is developed to identify missing regulation elicited during endocrine stress. Combined with microarray data, the model predicts that additional regulation exists involving the inter-organ feedback of luteinizing hormone (LH); highlighting to importance of interpreting organ-specific data in a whole organism context. Intriguingly, steroid synthesis is not highly sensitive to LH regulation. It seems that the steroid genesis network must trade sensitivity for robustness to noise in the kinase cascades.;This trade-off ultimately defines ideal drug candidates as well. Robust performance (RP) analysis (in the form of the structured singular value) is applied to an existing model of Fas induced apoptosis. RP analysis of the apoptosis model is capable of identifying known combinatorial perturbations which strongly control apoptotic behavior. This result supports extending model development and RP analysis into drug target discovery. RP analysis is applied to an insulin signaling model to identify combinatorial targets which are robust to intracellular uncertainty.;Analysis of the insulin signaling network shows that drug targets must follow similar robustness criteria as regulatory targets in biological networks. Many targets exist which can quite strongly manipulate the GLUT4 output of the insulin signaling model. But the controllability of these targets is often compromised by noise occurring in the intracellular network. Both in network elucidation and drug target identification, the efficacy of the proposed regulation must be necessarily balanced with its controllability. Robust performance analysis can rapidly identify robust targets network manipulation, greatly assisting experimental design and hypothesis validation.