The set of regulatory interactions between genes, mediated by transcription factors, forms a species' transcriptional regulatory network (TRN). By comparing this network with measured gene expression data one can identify functional properties of the TRN and gain general insight into transcriptional control. We define the subnet of a node as the subgraph consisting of all nodes topologically downstream of the node, including itself. Using a large set of microarray expression data of the bacterium Escherichia coli, we find that the gene expression in different subnets exhibits a structured pattern in response to environmental changes and genotypic mutation. Subnets with less changes in their expression pattern have a higher fraction of feed-forward loop motifs and a lower fraction of small RNA targets within them. Our study implies that the TRN consists of several scales of regulatory organization: 1) subnets with more varying gene expression controlled by both transcription factors and post-transcriptional RNA regulation, and 2) subnets with less varying gene expression having more feed-forward loops and less post-transcriptional RNA regulation.
Deep Dive into Patterns of subnet usage reveal distinct scales of regulation in the transcriptional regulatory network of Escherichia coli.
The set of regulatory interactions between genes, mediated by transcription factors, forms a species’ transcriptional regulatory network (TRN). By comparing this network with measured gene expression data one can identify functional properties of the TRN and gain general insight into transcriptional control. We define the subnet of a node as the subgraph consisting of all nodes topologically downstream of the node, including itself. Using a large set of microarray expression data of the bacterium Escherichia coli, we find that the gene expression in different subnets exhibits a structured pattern in response to environmental changes and genotypic mutation. Subnets with less changes in their expression pattern have a higher fraction of feed-forward loop motifs and a lower fraction of small RNA targets within them. Our study implies that the TRN consists of several scales of regulatory organization: 1) subnets with more varying gene expression controlled by both transcription factors and
An interesting topological feature of the transcriptional regulatory network (TRN) of the bacterium Escherichia coli is its almost tree-like structure with only few loops (see [1] for a detailed discussion and comparison with the TRN of the yeast Saccharomyces cerevisiae). This observation has several consequences. First, hierarchical levels in the network can be meaningfully defined and analyzed. Second, it leads to the question, on which level of organization information processing takes place in the TRN given a dominant directed flow dictated by the network's architecture. On a local scale, substructures in the TRN that appear significantly more often than in corresponding randomized networksso-called network motifs [2,3]-have been found to match specific information processing steps. Particularly feed-forward loops have been theoretically proposed [4] and experimentally supported [5,6] to function as noise-suppression units and delay devices.
Here we dissect the TRN into topological modules. We define the subnet of a node (root) as the subgraph consisting of all nodes topologically downstream of the root, including the root node itself (see Figure 1 for an illustration of the concept). Subnets can extend over multiple hierarchical layers if they contain a hierarchy of transcriptions factors (TFs). Moreover, they can overlap if genes are regulated by TFs from different subnets. Some network motifs such as the feed-forward loop or the single input motif are subnets themselves and therefore fully contained in at least one subnet. This approach is possible due to the topological properties of the E. coli TRN: apart from the few small cycles in the network (see Results), most subnets are directed acyclic graphs.
The search for the imprint of the transcriptional regulatory network in gene expression profiles is a search for very weak signals, often masked by the broad range of additional biological processes (beyond the regulation via transcription factors) shaping the expression of a gene. In two previous studies [7,8], the consistency between expression profiles and pairwise interactions in the TRN has been shown to be surprisingly low. The consistency on a larger scale has been studied for a specific type of subnets, named ‘origons’ [9]. There, the authors find that genes in some origons are selectively af- fected by specific environmental signals. In this contribution, we study patterns of subnet usage for two markedly different genome-wide gene expression data sets. As is [9], we use microarray expression profiles from the ASAP database, where wildtype expression under standard growth conditions is compared to a variety of profiles with external stimuli and genetic alterations. As a second data set, we use the time-course data of [10]. Here, E. coli strains are exposed to different media and stresses, and profiled at up to 16 time points. We analyze subnets with respect to their responsiveness to altered conditions in both data sets and classify them according to the observed subnet usage patterns. E. coli employs different scales of regulatory control to establish homeostasis (see, e.g., [11]) or to adapt to external stimuli. Recently, we introduced the concept of digital and analog control to differentiate between the regulatory response coordinated by dedicated TFs and DNA architectural proteins, respectively [12]. We found that as soon as one form is limited (by TF mutations or changes in the DNA superhelicity), the other form of control compensates, exhibiting a balance of regulatory control. An analysis employing methods from point process statistics has been able to further support the interplay of digital and analog control by analyzing gene distributions [13]. In the following, we want to delineate the interplay between the subnet usage as a TF mediated, topologically based form of control, and two other scales of regulatory control: translational inhibition and mRNA degradation induced by small non-coding RNAs (sRNAs) and the dynamic coordination of nodes connected in a feedforward loop.
We consider the most complete prokaryotic TRN available, the TRN of the bacterium E. coli. Nodes in our network correspond to genes (and the respective TF) while a directed edge represents a regulatory interaction mediated by a TF. Based on the version 6.3 of the Regulon database [14], the TRN comprises 1515 nodes and 3171 links, with 162 regulators (i.e. nodes which regulate at least one other gene) and 1432 target nodes (i.e. nodes which are regulated by at least one other gene).
We dissect the TRN into subnets, defined as subgraphs consisting of a root node with at least one regulatory interaction, and all downstream nodes (see Figure 1A for an illustrative example network consisting of three subnets). The 162 subnets of the TRN are overlapping and of very different sizes and hierarchical complexities (see the frequency distribution of subnet sizes in Figure 2A and the histogram of relative subnet overla
…(Full text truncated)…
This content is AI-processed based on ArXiv data.
