Scalable System-Wide Design of TF- and dCas9-Based Genetic Circuits

Engineered genetic circuits can reprogram organisms with complex decision-making capabilities that are only limited by the availability of genetic regulators, the designed architecture of their interactions, and the host's capacity. As more well-characterized regulators have become available, it remains a significant challenge to design larger genetic circuits, while accounting for the physical differences in the regulators' interactions, sequence-dependent changes in expression, and non-linear host-circuit coupling.  To accelerate the design of large genetic circuits with many regulators, we have developed and experimentally validated a system-wide mechanistic model of genetic circuits that predicts a circuit's input-output dynamical behavior in terms of the circuit's DNA sequence, biophysical measurements of the regulators, and the host's specifications. The model encompasses genetic circuits that use either traditional transcription factors as well as dCas9-based transcriptional regulation using guide RNAs. The model combines system-wide statistical thermodynamic models of transcription and translation that predict how changing transcription factor levels, Cas9 levels, guide RNA levels, guide RNA sequences, RBS sequences, and genome specifications alters the transcription of every promoter and the translation of every coding sequence in the circuit. The model then uses kinetics to determine circuit and host-circuit dynamics.

We experimentally validated our modeling approach by constructing and characterizing several genetic circuits.  First, we developed a new approach to measure the binding free energies of a transcription factor by designing and characterizing a series of genetic circuits incorporating auxiliary binding sites. Second, we tested the model's calculations by characterizing how different circuit characteristics (plasmid copy number, translation rates, and TF binding sites) and host growth conditions controlled the circuit's input-output relationship. Third, we characterized a series of signal amplification genetic circuits using different regulators and regulator expression levels to critically test the model's ability to account for changes in regulator binding free energies and binding occupancies. Fourth, we characterized a series of dCas9-based NOT gates to test the effects of changing Cas9 levels, guide RNA levels, and the guide RNA's binding sites on circuit function. Overall, we found that the model could accurately predict circuit behaviors, while accounting for several non-intuitive and non-linear behaviors, including the effects of altering TF/guide RNA levels and adding additional TF/guide RNA binding sites.  

Finally, based on these results, we propose a new dimensionless unit, the Ptashne number (Pt), that combines many circuit characteristics into a single number that accurately predicts the transcription rates of the circuit's promoters. Using the Pt number, the design space for an N-regulator circuit is transformed from a 5^N-dimensional space (5 variables per regulator) into an N-dimensional space (1 variable per regulator). We show how using the Pt number will accelerate the rational design of genetic circuits with many regulators.