Gene Regulatory Networks
Gene regulatory networks (GRNs) are the on-off switches and rheostats of a cell operating at the gene level. They dynamically orchestrate the level of expression for each gene in the genome by controlling whether and how vigorously that gene will be transcribed into RNA. Each RNA transcript then functions as the template for synthesis of a specific protein by the process of translation. A simple GRN would consist of one or more input signaling pathways, regulatory proteins that integrate the input signals, several target genes (in bacteria a target operon), and the RNA and proteins produced from those target genes. In addition, such networks often include dynamic feedback loops that provide for further regulation of network architecture and output. As indicated in the schematic below, input signaling pathways transduce intracellular and/or extracellular signals to a group of regulatory proteins called transcription factors. Transcription factors activated by the signals then interact, either directly or indirectly, with DNA sequences belonging to the specific genes they regulate. The factors also interact with each other to form multiprotein complexes bound to the DNA.
Gross anatomy of a minimal gene regulatory
network (GRN)
embedded in a regulatory network.
A regulatory network can be viewed as a cellular input-output device. At minimum, a gene regulatory network typically contains the following components: (1) an input signal reception and transduction system that mediates intra and extracellular cues (left box; often, more than one signal impinges on a given target gene); (2) a "core component" complex composed of transacting regulatory proteins and cognate cis-acting DNA sequences (circle; functionally similar components may be associated with multiple target genes, resulting in similar gene-expression patterns); and (3) primary molecular outputs from target genes, which are RNA and protein (box to right of circle). The net effects are changes in cell phenotype and function (right box). Direct and indirect feedbacks typically are important. More realistic networks often feature multiple tiers of regulation, with first-tier gene products regulating expression of another group of genes, and so on. Beyond GRN boundaries are signaling responses and feedbacks, such as those that drive bacterial chemotaxis, which do not involve regulation of gene expression but instead act directly on proteins and protein machine assemblies (dashed arrows). Some regulatory networks have no embedded GRN component.
GRNs act as analog biochemical computers to specify the identity and level of expression of groups of target genes. Central to this computation are DNA recognition sequences with which transcription factors associate. Every gene has its own novel "cis-acting" sequence elements. They vary greatly in complexity from one gene to another and from generally simpler structures in bacteria to more complex structures in multicellular organisms. When active transcription factors associate with the cis-elements of their cognate target genes, they can function to specifically repress (down-regulate) or induce (up-regulate) synthesis of the corresponding RNA. The immediate molecular output of a gene regulatory network is the constellation of RNAs and proteins encoded by network target genes. The resulting cellular readouts are changes in the structure, metabolic capacity, or behavior of the cell mediated by new expression of up-regulated proteins and elimination of down-regulated proteins.
Click to enlarge.
GRNs are remarkably diverse in their structure, but several basic properties are illustrated in the figure to the left. In this example, two different signals impinge on a single target gene where the cis-regulatory elements provide for an integrated output in response to the two inputs. Signal molecule A triggers the conversion of inactive transcription factor A (green oval) into an active form that binds directly to the target gene's cis-regulatory sequence. The process for signal B is more complex. Signal B triggers the separation of inactive B (red oval) from an inhibitory factor (yellow rectangle). B is then free to form an active complex that binds to the active A transcription factor on the cis-regulatory sequence. The net output is expression of the target gene at a level determined by the action of factors A and B. In this way, cis-regulatory DNA sequences, together with the proteins that assemble on them, integrate information from multiple signaling inputs to produce an appropriately regulated readout. A more realistic network might contain multiple target genes regulated by signal A alone, others by signal B alone, and still others by the pair of A and B.
Co-regulated target genes often code for proteins that act together to build a specific cell structure or to effect a concerted change in cell function. For example, genes encoding components of the multiprotein proteasome machine (see The Machines of Life) are co-regulated at the RNA level. This was shown by microarray gene chip analyses in yeast cells, and each gene was found to possess a similar cis-regulatory DNA sequence that mediates binding of a particular transcription factor. Similarly, a bacterium may respond to a shortage of its preferred energy source by activating expression of genes whose protein products function in a biochemical pathway that allows it to use a different, more abundant source of energy.
Some genes are regulated by a single input mechanism, but, especially in higher organisms, a gene often responds to information from multiple signals via the activity of diverse transcription factors. For example, in human fibroblast cells responding to a "growth factor" impinging on cell surface signal receptors, a platoon of "immediate early" genes is up-regulated as the first step in the complex process of cell proliferation. Some of the same genes, though not all of them, can be activated in brain cells by the distinct stimulus of seizure. This network also illustrates that GRNs can be multitiered. The first signal (growth factor) initiates expression of "immediate early" target genes, which include transcription factors such as c-Fos and c-Jun. These transcription factors then cause a second group of target genes, called "early genes," to be expressed. Among these early genes are other transcription factors such as c-Myc. They regulate expression of yet another group of genes of a "delayed early" group. In this way a multitiered GRN cascade can be constructed, replete with feedbacks and crosstalk to other networks.
Click to enlarge.
A major gene regulatory network in the bacterium Caulobacter is now beginning to be mapped in a comprehensive manner based on genome-wide expression analyses coupled with genetic methods [M. T. Laub et al., Science 290, 214448 (2000)]. Caulobacter has about 3000 genes, of which almost 20% were found to be differentially expressed during the cell division cycle. Of the 553 responding genes, 38 are likely to be direct targets of a sequence-specific DNA-binding protein called CtrA and another 144 are indirectly regulated by CtrA. A firstpass connectivity map of the CtrA gene regulatory network derived from this study is summarized in this figure. Green indicates previously known relationships, while red indicates relationships that emerged from this global gene-expression study performed using microarray technology.
