Elucidating Rules for Composing Synthetic Genetic Programs Using a Model-Guided Approach

Advances in the design of synthetic genetic programs have enabled the engineering of cells with new sense-and-respond capabilities. As increasingly sophisticated functions are being developed for basic science and medical applications, there remains a fundamental need for cells to carry out programs precisely and as intended. Achieving this level of control requires molecular components that are well-characterized, tunable, and can be predictably composed. To this end, we developed the Composable Mammalian Elements of Transcription (COMET)—a toolkit comprising zinc-finger transcription factors (ZF-TFs) and cognate promoters for regulating gene expression. We then created a mathematical framework that accurately explains COMET behavior, by accounting for cell heterogeneity using a statistical model along with gene regulation using a dynamical model. The computational approach enabled us to uncover design rules for how to reliably tune gene expression based on modular features of the TFs (zinc finger, mutant variant, and activation domain) and promoters (number, spacing, and arrangement of binding sites). Notably, the concise formalism differs from standard models of transcriptional regulation and uses fewer parameters. We found that COMET provides several advantageous properties, including (i) predictable dose response profiles, (ii) direct relationships between model parameters and physical features of TFs and promoters, and (iii) a unique mechanism for inhibitory TFs that combines competitive inhibition, steric occlusion, and effective modification of promoter properties. Lastly, we validated how applying the inhibitor mechanism ‘in reverse’ enables tight single-layer Boolean logic with multiple activating and inhibitory inputs. In summary, we developed a rigorous approach to define rules and parameters that govern COMET behavior, and discovered properties that are fundamentally distinct from those of standard TFs and promoters. We propose how principles learned from this investigation can be extended to quantitatively characterize other synthetic signaling and regulatory systems, towards the design of more complex genetic programs in mammalian cells.