The TOL network of Pseudomonas putida mt-2 processes multiple environmental inputs into a narrow response space Rafael Silva-Rocha and Víctor de Lorenzo* Systems Biology Program, Centro Nacional de Biotecnología CSIC, Cantoblanco-Madrid, 28049 Spain. Summary The TOL system encoded by plasmid pWW0 of Pseu- domonas putida mt-2 is able to sense a large number of both exogenous and endogenous signals as inputs for the genetic and metabolic circuit that determines the biodegradation of m-xylene. However, whether the enormous combinatorial space of inputs is translated into an equally variable response landscape or is pro- cessed into very few outcomes remains unclear. To address this question, we set out to define the number of states that can be obtained by a network of a given set of genes under the control of a specified regulatory circuit that is exposed to all possible combinations of inputs. To this end, the TOL network and its regula- tory wiring were formalized as a synchronous logic Boolean circuit that had 10 signals (i.e. pathway sub- strates, temperature, sugars, amino acids, metabolic regimes and global regulators) as possible inputs. The analysis of the attractors of the circuit using a satis- fiability (SAT) algorithm revealed that only eight transcriptional states are reached in response to the 1024 possible combinations of stimuli. The experi- mental validation resulted in a refinement of the model through the addition of a previously unknown interac- tion that controls the meta catabolic pathway. The full induction of the two xyl operons occurred with only 1.6% of the input combinations, which suggests that the architecture of the network allows the expression of the xyl genes only under a very narrow range of conditions. These data not only explain much of the unusual layout of the TOL circuit but also strengthen the view of the regulatory circuits of environmental bacteria as digital decision-making devices. Introduction The environmental bacteria that inhabit sites with unusual, e.g. xenobiotic, or recalcitrant carbon sources offer optimal experimental systems for the examination of signal inte- gration in gene regulatory networks (GRNs). The core biological function of GRNs is the sensing of a plethora of exogenous and endogenous inputs and the processing of these signals to form responses that promote the survival of the organism under the given conditions (Lee et al., 2002; Babu et al., 2004; Lozada-Chavez et al., 2006). The regulatory network encoded by the catabolic plasmid pWW0 of the soil bacterium Pseudomonas putida mt-2 (Silva-Rocha et al., 2011a) is one example case of envi- ronmental signal integration. The TOL system includes two metabolic operons that encode the enzymes that allow this organism to utilize m-xylene as the sole carbon source (Williams and Murray, 1974; Ramos et al., 1997). However, the expression of these operons is not only intertwined to an already complex genetic circuit for the activation of cognate promoters in response to pathway substrates (m-xylene and 3-methylbenzoate, Fig. 1) but is also wired to a multitude of environmental and physiological signals through an intricate GRN (Ramos et al., 1997; Silva-Rocha et al., 2011a). Although the molecular mechanisms and functions of some of the regulatory nodes of the TOL circuit are known in considerable detail, the global properties of the system are not yet understood (Koutinas et al., 2010; 2011; Silva-Rocha et al., 2011b). As more GRN architectures are disclosed in different organisms, the logic of the decision-making process that the cells undertake in response to changing environ- mental signals becomes more evident (Helikar et al., 2008; Domedel-Puig et al., 2011). A number of tools have been developed to analyse the complexity of this type of regu- latory systems and to expose the properties that are encrypted in the corresponding network architectures. One approach involves the formulation of Boolean models to represent the interactions that take place in any given GRN (Karlebach and Shamir, 2008). In a Boolean model, all of the interacting elements (nodes) are discretized to one of two opposite states: absent/inactive (or 0) and present/ active (or 1). In addition, a node is allowed to switch from one state to another according to a set of logic rules that Received 19 July, 2012; revised 23 September, 2012; accepted 27 September, 2012. *For correspondence. E-mail vdlorenzo@cnb.csic. es; Tel. (+34) 91 585 45 36; Fax (+34) 91 585 45 06. Environmental Microbiology (2013) 15(1), 271–286 doi:10.1111/1462-2920.12014 © 2012 Society for Applied Microbiology and Blackwell Publishing Ltd