Architectures and complex functions of tandem riboswitches

Abstract

Riboswitch architectures that involve the binding of a single ligand to a single RNA aptamer domain result in ordinary dose-response curves that require approximately a 100-fold change in ligand concentration to cover nearly the full dynamic range for gene regulation. However, by using multiple riboswitches or aptamer domains in tandem, these ligand-sensing structures can produce additional, complex gene control outcomes. In the current study, we have computationally searched for tandem riboswitch architectures in bacteria to provide a more complete understanding of the diverse biological and biochemical functions of gene control elements that are made exclusively of RNA. Numerous different arrangements of tandem homologous riboswitch architectures are exploited by bacteria to create more 'digital' gene control devices, which operate over a narrower ligand concentration range. Also, two heterologous riboswitch aptamers are sometimes employed to create two-input Boolean logic gates with various types of genetic outputs. These findings illustrate the sophisticated genetic decisions that can be made by using molecular sensors and switches based only on RNA.

Keywords: Aptamer; gene regulation; logic gate; noncoding RNA; transcription control; translation control.

Figure 1.

Previously known tandem architectures for riboswitches and their established functions. (A) Cooperative riboswitch aptamers carry highly similar aptamer domains that bind chemically identical ligands and associate with a single expression platform. Examples of this riboswitch architecture demonstrate cooperative ligand binding and a steeper dose-response curve [27]. (B) Pseudo-cooperative [27] and bi-mechanism [55] riboswitches involve the tandem arrangement of independently functioning riboswitches that respond to chemically identical ligands. For a bi-mechanism system, each riboswitch operates with a different regulatory mechanism (e.g. one transcriptional and one translational). (C) Dual riboswitch logic gates [25] involve the tandem arrangement of independently functioning riboswitches that respond to different target ligands, here depicted as X and Y. (D) Interactive aptamer logic gates [56] are formed by two adjacent aptamers that respond to different target ligands and associate with a single expression platform. Ligand binding by one aptamer affects the function of the adjacent aptamer. (E) Allosteric ribozyme logic gates [59] involve allosteric regulation by an aptamer for the function of a ribozyme that requires a second distinct compound for its activity.

Figure 2.

All theoretical two input Boolean logic gate systems and their relationship to riboswitch gene control systems. Depicted are the truth tables [28] for all possible gene regulation logic gate systems based on the presence (+) or absence (−) of ‘inputs’ A and B, where A represents the cognate ligand of the first RNA domain and B represents the cognate ligand for the second domain (e.g. ligands X and Y as presented in Figure. 1). There are four possible combinations of ‘inputs’ (presence or absence of A or B occupying the aptamer binding pocket) and 16 possible gene expression ‘outputs’, i.e. whether expression of the downstream gene is on (+) or off (−). Each output is named for its corresponding Boolean logic function [28] and is coloured to reflect its known or possible riboswitch manifestation. Note that ‘FALSE’ and ‘TRUE’ outputs (dark red) are not switches, and therefore have no utility for gene control. ‘A’, ‘B’, ‘NOT A’, and ‘NOT B’ outputs (light red) represent gene control outcomes that are identical to those achievable by single riboswitches with a single ligand input, and therefore tandem arrangements are unnecessary. Natural examples of five of the remaining ten genetically practical two-input Boolean logic gate functions are known to exist (blue). White boxes indicate that a tandem, two-input riboswitch system has not yet been observed in nature. Question marks indicate that it is not apparent how tandem riboswitches would function to create XOR and XNOR logic gates. Published examples are presented for the five known logic gates as follows: C-di-GMP-II/Group I Rz [59], ZTP/THF [89], TPP/HMP-PP [105], SAM-I/AdoCbl [25], and Guanine/PRPP [56]. Additional examples include (I) T box^Leu/ppGpp [56] and glycine/c-di-GMP (this study); (ii) ZTP/glycine (this study), Na⁺-I/c-di-AMP [118], and SAM-I/T box^Met (this study); (iii) T box^Met/SAM-I (this study), AdoCbl/C-di-GMP-I (this study), AdoCbl/C-AMP-GMP (this study), Adenine/T box^Tyr (this study), Glycine/T box^Ser (this study) and AdoCbl/T box^Val (this study); (iv) AdoCbl/SAM-I (this study) (Table 2).

Figure 3.

An unusual series of tandem riboswitches that sense AdoCbl, c-di-GMP, and c-AMP-GMP ligands. (A) Schematic representation of seven riboswitches occurring within a seven kb region of the genome of the bacterium D. acetoxidans. Asterisks denote the locations of putative intrinsic transcription terminators, indicating each aptamer functions as an independent riboswitch. Aptamers labelled 1, 2, and 3 were subjected to ligand binding assays as depicted in B. (B) Sequences and secondary structure models for aptamers evaluated by in-line probing assays. Circled nucleotides identify a position directly contacting the ligand that is known to be important for the selective binding of c-di-GMP (G nucleotide) or c-AMP-GMP (A nucleotide) [43]. (C) Binding curves resulting from in-line probing assays conducted on aptamers 1 through 3 (left to right) with c-di-GMP and c-AMP-GMP. Data points represent the average modulation at three locations, where the error bars represent the standard deviation at each concentration tested.