Synthetic logic circuits encoded on toehold strand-displacement switchable CRISPR guide RNAs

Abstract

The field of biological computing offers the potential to construct devices using complex and dynamic regulation for applications ranging from theranostics to the production of high value chemicals in bioreactors. However, building these complex regulatory systems depends on the creation of effective logic gates, which form the basis of digital systems. The low metabolic load, small genetic footprint and the low latency of expression that can be achieved with a small RNA-based regulatory system in contrast to protein regulators, emphasises the potential within RNA regulation. From an engineering perspective, the great advantage of an RNA approach is the highly predictable nature of RNA folding and the availability of established in silico tools.

This thesis describes a novel de novo mechanism for NAND gate implementation in vivo using two RNAs, both of which must be expressed for the repression of an output gene. To construct this regulatory system a guide RNA of the CRISPR-Cas9 system is modified through the addition of a cis-repressing element, which complements part of the guide RNA and represses its activity. The activity of the cis-repressed guide RNA (crgRNA) can be rescued by the expression of an antisense RNA, which complements the cis-repressing element. This allows the guide RNA to return to an active conformation and repress the target promoter through CRISPRi (in strains expressing dCas9). This represents a NAND gate, as the output is repressed (OFF) only when both input RNAs are expressed. The design and optimisation of this system was performed using modelling of system energy states and dynamics and machine learning optimisation in a process which was automated into a single pipeline for future users. This system was characterised over a range of crgRNA and dCas9 expression levels and temperatures, and in different growth phases. Eight designs were tested and the optimal variant, for which output gene expression most closely approximated the OFF (repressed) and ON (un-repressed) states required for a logic gate, was chosen. The resulting NAND gate has a 10-fold repression of the output promoter when both RNAs were present; in contrast, only 1.2 fold repression was obtained when only the crgRNA was expressed.

Consequently, multiple versions of the optimal variant were synthesised, each with different sequences but the same design principles. These performed similarly when applied to the repression of different reporter genes. Finally, an in silico approach was used to maximise orthogonality of different versions of the optimal variant which was then demonstrated in vivo. This novel NAND gate design offers the ability to build large libraries of logic gates with small genetic footprints (304 bp) and the potential to be combined to produce complex regulatory networks.