Combinatorial influence of environmental parameters on transcription factor activity

Abstract

Motivation: Cells receive a wide variety of environmental signals, which are often processed combinatorially to generate specific genetic responses. Changes in transcript levels, as observed across different environmental conditions, can, to a large extent, be attributed to changes in the activity of transcription factors (TFs). However, in unraveling these transcription regulation networks, the actual environmental signals are often not incorporated into the model, simply because they have not been measured. The unquantified heterogeneity of the environmental parameters across microarray experiments frustrates regulatory network inference.

Results: We propose an inference algorithm that models the influence of environmental parameters on gene expression. The approach is based on a yeast microarray compendium of chemostat steady-state experiments. Chemostat cultivation enables the accurate control and measurement of many of the key cultivation parameters, such as nutrient concentrations, growth rate and temperature. The observed transcript levels are explained by inferring the activity of TFs in response to combinations of cultivation parameters. The interplay between activated enhancers and repressors that bind a gene promoter determine the possible up- or downregulation of the gene. The model is translated into a linear integer optimization problem. The resulting regulatory network identifies the combinatorial effects of environmental parameters on TF activity and gene expression.

Availability: The Matlab code is available from the authors upon request.

Supplementary information: Supplementary data are available at Bioinformatics online.

Fig. 1.

Expression levels of a gene (COX5A) across the 55 cultivation conditions. The colored matrix is a schematic representation of the settings of the nine cultivation parameter types across the 55 conditions. The colored lanes indicate the cultivation parameter types that are employed to order the experiments, in this case, aeration type and limiting element. The regression model which models the gene expression as a function of the cultivation parameters, selected one single effect, i.e. aeration type, and one combinatorial effect, i.e. aeration type anaerobic together with limiting element carbon. The reconstructed expression pattern based on these two effects is indicated by the shaded area.

Fig. 2.

Schematic overview of the approach. The goal is to build , the optimal approximation of the discretized regression coefficients in R. (a) The coefficients in R are derived from a regression analysis, which assesses the influence of cultivation parameters on gene expression by employing these parameters as predictors in the regression model. The discretization procedure maps non-zero regression weights to 1 or −1, depending on their sign. (The schematic representation of R is given for five genes and three cultivation parameters.) (b) The elements of are determined by T and M. T is fixed and indicates binary TF binding potential to gene promoters. The elements of M are estimated and indicate the activity of TFs as enhancers or repressors under the different (combinatorial) cultivation parameters. A logic circuit derived from M is graphically depicted above the representation of M. (c) Visualization of the active TFs on the gene promoters of genes g1, g2 and g3 under cultivation parameter A. Enhancers are depicted as red boxes; repressors are depicted as green boxes. (TF γ can bind the promoter of g1, but is not active under A.) The height of a box indicates the enhancer or repressor strength. The strength of a particular enhancer or repressor is the same for all genes. A gene is upregulated when its activator strength, i.e. the sum of the heights of the red boxes, is larger than the repressor strength, which equals the sum of the heights of the green boxes. Downregulation is inferred in the opposite situation. See text for details.

Fig. 3.

Overview of the results obtained for the oxygen and carbon limitation data. (a) Inferred influence of cultivation parameters aerobic growth (Aer), anaerobic growth (Ana) and carbon limitation (Clim) on TF activity. Only the three dominating TFs are reported. (b) Representation of S, indicating the strength of the activated TFs under each of the four cultivation parameters. Enhancers are depicted in red; repressors are depicted in green. (c) Representation of T, indicating which gene promoters can be bound by the activated TFs. The enhancer or repressor strengths for the four cultivation parameters are visualized by the colored blocks inside the rectangle that represents a binding site. (d) Representation of , indicating the inferred regression coefficients. Upregulation is indicated by red; downregulation is indicated by green. Incorrectly inferred elements are marked with a gray cross. White boxes without a cross are the zero elements of R. These elements are not part of the optimization scheme.

Fig. 4.

CV errors for different values of λ.

Fig. 5.

Inferred TF activity derived from genes, which are involved in nitrogen metabolism. Preferred nitrogen sources are printed in bold; non-preferred nitrogen sources are printed in italic style. Abbreviations for the nitrogen and sulfur sources are explained in the text.

Fig. 6.

Representation of S for the regulatory program inferred using the compendium. Color coding is identical to Figure 3b.