RNA-based computation in live cells

Abstract

Man-made molecular 'computers' that operate inside live cells will enable unprecedented level of control over cellular physiology. A promising approach to building these computers uses RNA molecules and RNA-based regulation. RNA naturally lends itself to create 'digital' molecular networks that embody standardized (normal) forms of logic functions. The network's inputs, that may or may not be inverted by single-input NOT logic gates, feed into multi-input AND gates whose outputs are in turn integrated in a multi-input OR gate. Below I review recent steps that have been taken toward implementing these networks with allosteric riboswitches and ribozymes in bacteria and yeast, and RNAi in mammalian cells. I also propose how to co-opt recently discovered additional RNA regulation mechanisms into future construction efforts.

Figure 1. Basic definitions and schematics of logic networks

A, A given logic function as represented by input/output mapping (left panel, F=False, T=True) can be implemented by cascading universal gates such as NAND (middle panel) or by constructing a circuit in a normal form (right panel). B, Basic regulatory links that create complex molecular logic networks. Generic interactions of activation and inhibition are subdivided into classes based on the fine details of the interactions. These details, while not altering the high-level logic behavior of a network, can significantly influence its quantitative properties [17] and must be carefully considered in the design process. C, Designation of different ways in which concurrent regulation of a biological target by multiple effectors gives rise to different basic logic relations and their “analog” counterparts in an unsaturated regime.

Figure 2. Finite-depth layered logic circuits

A, Circuits that implement normal logic forms (Layouts 1-4). Colored circles represent external molecular inputs. They are linked by individual activating or inhibitory links to the immediate inputs of the molecular computations (gray color in top left diagram). Sub-networks that convert external inputs into the inputs for a computational core (collapsed into single activating or inhibitory links in the scheme) are called “sensors”. A purpose of sensors is to convert various external inputs into a uniform molecular format to promote scalability and modularity. The computational inputs are converted into the network’s output (red color) via a computational core implementing a particular mathematical relation, here a logic function. Layouts 1 and 2 use not-trivial converging regulatory links to implement disjunctive (DNF) and conjunctive (CNF) normal forms, respectively. Layouts 3 and 4 only use links that act independently of each other and hence are easier to design, but by virtue of their respective actions implement similar logic functions. B, Examples of layered circuits that, while being superficially similar to Layouts 1-4, do not implement normal forms but only one kind of logic operation. C, Collapsing sequential regulatory links for the purpose of circuit analysis. In a regulatory chain with no convergence two sequential links can be replaced by a single interaction according to the schematics in the figure.

Figure 3. Modes of RNA computation in bacteria based on riboswitches and sRNAs

A, A DNF-like computations (central panel) that involve multiple intrinsic terminators controlled by aptamers (left panel) or multiple sRNAs targeting the same transcript (right panel). Depicted diagrams only implement multi-input AND gates, but an OR operation could be included by adding a parallel mRNA expressing the same gene but targeted by a different set of inputs. B, A multi-input OR gate, a portion of a CNF circuit implemented by sRNAs that target a transcription silencing element. Extending this architecture to a full-fledged CNF requires that a number of cis silencing elements control the same mRNA. Such arrangement has not yet been reported but is conceivable.

Figure 4. RNA computations in eukaryotes

A, Yeast-based networks. Arrays of allosteric riboswitches in a gene’s 3’-UTR enables an AND clause of a DNF circuit, while combining two reporter constructs with an identical output can add the OR gate and extend the network to a full DNF. B, RNAi-based mammalian DNF and CNF circuits. A regulatory link between a small molecule metabolite B and an shRNA has been shown in a number of reports, while other classes of inputs (dotted lines) yet need to be demonstrated. In the DNF circuit the small RNAs target the output directly, while in the CNF layout they target a transcriptional repressor that controls the output via engineered promoter.