Genetic flexibility of regulatory networks

Abstract

Gene regulatory networks are based on simple building blocks such as promoters, transcription factors (TFs) and their binding sites on DNA. But how diverse are the functions that can be obtained by different arrangements of promoters and TF binding sites? In this work we constructed synthetic regulatory regions using promoter elements and binding sites of two noninteracting TFs, each sensing a single environmental input signal. We show that simply by combining these three kinds of elements, we can obtain 11 of the 16 Boolean logic gates that integrate two environmental signals in vivo. Further, we demonstrate how combination of logic gates can result in new logic functions. Our results suggest that simple elements of transcription regulation form a highly flexible toolbox that can generate diverse functions under natural selection.

Fig. 1.

(A) Representation of a promoter as a logic gate. Small molecule signals (here cAMP and D-gal) interact with specific transcription factors (here GalR and CRP) that bind a specific sequence in the regulatory region and influence transcription of the reporter gene (here uidA). The logic of signal integration, i.e., the activity of the promoter at different combinations of signals (B), depends on the activity and on the combined action of regulatory proteins. In this case, (Lower Right) (D-gal) represents the basal promoter activity (none of the regulators bind to DNA), whereas (Lower Left) (no signals) represents promoter activity when GalR is bound. In the presence of cAMP (Upper Left) both GalR and cAMP-CRP can bind DNA, whereas in the presence of both signals (Upper Right) only cAMP-CRP can regulate the promoter. (C) Schematic drawing of the reporter construct used for testing signal integration at the engineered promoters. Red arrows represent ORFs. Regulatory sequences are inserted between the EcoRI and PstI sites.

Fig. 2.

All 16 possibilities for the Boolean-type integration of two input signals. For the 12 gates successfully created the schematic structures of regulatory regions are shown (Left) of the logic gates (boxes). The figure is not drawn to scale. Red boxes represent predicted GalR binding sites, yellow boxes represent predicted cAMP-CRP binding sites, black arrows represent promoters. Numbers in the boxes show the distance of the operator site (Center) from the transcription start site of the promoter. The activity of the reporter gene (uidA) was monitored in the four possible combinations of D-gal and cAMP. The presence of signals is indicated in (A), and the same arrangement was used in B–L. Blue color results from successful conversion of the chromogenic substrate (X-gluc) by the UidA protein, indicating that the reporter gene is expressed (ON) in a given combination of input signals. The lack of blue colors indicates that expression is OFF. Examples of reporter enzyme activities in logarithmic phase cells are provided in the Supporting Information. GenBank accession numbers are shown for constructs longer than 50 bp.

Fig. 3.

Strategies for combining logic gates. In the serial connection (A) the regulatory regions (gate 1 and gate 2) are placed sequentially upstream of the reporter gene (red arrows). In the parallel combination (B) the two regulatory regions are fused to the reporter gene separately.

Fig. 4.

Logic selector circuits. Schematic for showing how logic gates can be combined to get more complex behavior. Depending on the position of the switch, the output responds to the two inputs with different logic. In cells such schemes could be used to adapt the logic gates to environmental conditions. For example, scheme (A) could be implemented by a transcription factor that regulates the lower gate. A change in the environmental condition (e.g., presence of a small molecule) can activate or inactivate the transcription factor. In this case contribution of the lower gate to the output depends on the activity of the TF. In scheme (B) the output logic depends on which of the two gates is active.