A Study Of Small Gene Regulatory Networks: Robustness, Dynamics, And Behavior

Abstract

Gene regulatory networks, like any evolving biological system, are subject to potentially damaging mutations. Much work has been done to study what types of networks are more robust to node deletions (knockouts of entire genes). Less well understood, however, is the question of which networks best maintain consistent behavior in the face of smaller mutations that affect binding affinity, protein half-life, and other regulatory parameters. Such mutations have subtler effects than whole-gene knockouts do, but because they are far more common than knockout mutations, their impact on network evolution may be substantial. The first chapter investigates the expression patterns of simulated gene regulatory networks as these types of parameters are varied, and explores which topologies allow the networks to "ignore" parameterchanging mutations and maintain their expression patterns relatively unchanged. In the simulations, networks containing mutual repression feedback consistently displayed a more robust response to simulated mutation. The presence of this variety of feedback in well-studied developmental regulatory networks suggests that it may be a widespread mechanism for reducing the phenotypic consequences of both noise and mutational perturbations. The second chapter also uses feedback loop module networks as a means to investigate and compare modeling approaches. It describes an algorithm to infer the best Boolean representation of the differential equation network models, as well as metrics for measuring how closely the Boolean model approached the dynamics of the continuous one. Using these tools allowed testing of the "Booleanizability" of networks containing mutual-activator and mutual-repressor feedback loops. The investigation revealed that Boolean models are better approximations of networks with repressor loops than of those without them, and this is explained in terms of the characteristics explored in the investigation of network robustness. Chapter 3 contains a model of the gap genes in Drosophila melanogaster, based on published experimental findings on the interactions among these genes and their products. The mechanistic mathematical model of gap gene expression was fitted to experimental data and placed in the context of other gap gene models. The chapter discusses the advantages and limitations of the various modeling techniques that have been employed for this system.