Computational modeling of anthocyanin pathway evolution

Abstract

Alteration of metabolic pathways is a common mechanism underlying the evolution of new phenotypes. Flower color is a striking example of the importance of metabolic evolution in a complex phenotype, wherein shifts in the activity of the underlying pathway lead to a wide range of pigments. Although experimental work has identied common classes of mutations responsible for transitions among colors, we lack a unifying model that relates pathway function and activity to the evolution of distinct pigment phenotypes. One challenge in creating such a model is the branching structure of pigment pathways, which may lead to evolutionary trade-os due to competition for shared substrates. In order to predict the eects of shifts in enzyme function and activity on pigment production, we created a simple kinetic model of a major plant pigmentation pathway:the anthocyanin pathway. This model describes the production of the three classes of blue, purple and red anthocyanin pigments, and accordingly, includes multiple branches and substrate competition. We rst studied the general behavior of this model using a naïve set of parameters. We then stochastically evolved the pathway toward a dened optimum and analyzed the patterns of xed mutations. This approach allowed us to quantify the probability density of trajectories through pathway state space and identify the types and number of changes. Finally, we examined whether our simulated results qualitatively align with experimental observations, i.e., the predominance of mutations which change color by altering the function of branching genes in the pathway. These analyses provide a theoretical framework that can be used to predict the consequences of new mutations in terms of both pigment phenotypes and pleiotropic eects.We thank members of the Smith lab for useful conversations regarding implementation and interpretation of our computational model. We thank Jacob Stanley and Rutendo Sigauke in the Dowell group at CU Boulder for helpful conversations regarding development of the simulation framework and appropriate implementation of the subsequent analyses. We also thank Boswell Wing, Sebastian Kopf, and the other members of the CU Boulder Geobiology Super Group for their helpful feedback.