J. theor. Biol. (2000) 204, 311}326 doi:10.1006/jtbi.2000.2014, available online at http://www.idealibrary.com on Genetic Regulation of Root Hair Development in Arabidopsis Thaliana : A Network Model LUIS MENDOZA AND ELENA R. ALVAREZ-BUYLLA* ¸aboratorio de Gene H tica Molecular y Evolucio H n, Instituto de Ecologı H a ; NAM, Ap. Postal 70-275, Coyoaca H n, D.F. CP04510, Mexico (Received on 24 June 1999, Accepted in revised form on 31 January 2000) The root epidermis of Arabidopsis thaliana is formed by alternate "les of hair and non-hair cells. Epidermal cells overlying two cortex cells eventually develop a hair, while those overlying only one cortex cell do not. Here we propose a network model that integrates most of the available genetic and molecular data on the regulatory and signaling pathways underlying root epidermal di!erentiation. The network architecture includes two pathways; one formed by the genes ¹¹G, R homolog, G¸2 and CPC, and the other one by the signal transduction proteins ETR1 and CTR1. Both parallel pathways regulate the activity of AXR2 and RHD6, which in turn control the development of root hairs. The regulatory network was simulated as a dynamical system of eight discrete state variables. The distinction between epidermal cells contacting one or two cortical cells was accounted for by "xing the initial states of CPC and ETR1 proteins. The model allows for predictions of mutants and pharmacological e!ects because it includes the ethylene receptor. The dynamical system reaches one of the six stable states depending upon the initial state of the CPC variable and the ethylene receptor. Two of the stable states describe the activation patterns observed in mature trichoblasts (hair cells) and atrichoblasts (non-hair cells) in the wild-type phenotype and under normal ethylene availability. The other four states correspond to changes in the number of hair cells due to experimentally induced changes in ethylene availability. This model provides a hypothesis on the interactions among genes that encode transcription factors that regulate root hair develop- ment and the proteins involved in the ethylene transduction pathway. This is the "rst e!ort to use a dynamical system to understand the complex genetic regulatory interactions that rule Arabidopsis primary root development. The advantages of this type of models over static schematic representations are discussed.  2000 Academic Press Introduction The increasing amount of data on the molecular mechanisms underlying cellular di!erentiation and morphogenesis are putting forward complex networks of genetic interactions. These are dy- namical systems that operate in speci"c cellular * Author to whom correspondence should be addressed. E-mail: abuylla@servidor.unam.mx and spatial scales. Network models of genetic regulatory interactions are dynamical repres- entations of molecular interactions, and can be used to predict the stable activation states attained by a given genetic architecture within mature cells (Kau!man, 1993). These models provide a formal dynamical framework to analyse cellular di!erentiation, in contrast to schematic representations of gene and protein 0022}5193/00/110311#16 $35.00/0  2000 Academic Press