Transcriptional kinetic synergy: a complex landscape revealed by integrating modelling and synthetic biology

Abstract

1AbstractEukaryotic genes are combinatorially regulated by a diversity of factors, including specific DNA-binding proteins called transcription factors (TFs). Physical interactions between regulatory factors have long been known to mediate synergistic behaviour, commonly defined as deviation from additivity when TFs or sites act in combination. Beyond binding-based interactions, the possibility of synergy emerging from functional interactions between TFs was theoretically proposed, but its governing principles have remained largely unexplored. Theoretically, the interplay between the binding of TFs and their effects over transcription has been challenging to integrate. Experimentally, probing kinetic synergy is easily confounded by physical interactions. Here we circumvent both of these limitations by focusing on a scenario where only one TF can be specifically bound at any given time, which we build using a synthetic biology approach in a mammalian cell line. We develop and analyze a mathematical model that explicitly incorporates the details of the binding of the TFs and their effects over transcription. The model reveals that synergy depends not only on the biochemical activities of the TFs, but also on their binding kinetics. We find experimental evidence for this result in a reporter-based system where fusions of mammalian TFs with engineered zinc fingers bind to a single, shared site. A complex synergy landscape emerges where TF activity, concentration and binding affinity shape the expression response. Our results highlight the relevance of an integrated understanding of TF function in eukaryotic transcriptional control.

Quantifying the regulatory role of individual transcription factors in Escherichia coli

“…Indeed, thermodynamic models have been employed to interrogate the regulatory function of eukaryotic promoters involved in key processes such as cellular differentiation, body patterning, and a host of other biological roles ( Segal et al, 2006 , 2008 ; Sayal et al, 2016 ; Chen et al, 2008 ; Ay and Arnosti, 2011 ; Fakhouri et al, 2010 ; Bashor et al, 2019 ). The ability to distinguish and quantify the different modes of regulation (stabilization and acceleration) and characterize TFs based on them is important for developing general theories of regulation that include multiple TFs that act on different kinetic steps of the transcription process ( Scholes et al, 2017 ; Martinez-Corral et al, 2020 ; Wong and Gunawardena, 2020 ); predictions for the combined regulatory effect of two stabilizing TFs should be different than predictions for a stabilizing TF acting together with an accelerating TF ( Scholes et al, 2017 ). With each characterized TF, we can develop an empirical baseline or null hypothesis for what a TF should do on a gene; departures from this expectation, because of emergence of complex regulatory phenomenon brought about by TF-TF interactions ( Weingarten-Gabbay and Segal, 2014 ), allosteric interactions ( Rosenblum et al, 2020 ; Kim et al, 2013 ), or other effects indicate surprises that warrant testing in these expanded models.…”

Quantifying the regulatory role of individual transcription factors inEscherichia coli

“…We find that the contributions of these two mechanisms do not correlate with position; in some locations we found sta- Here we focus entirely on the isolated regulatory role of each TF, however it is clear that one of the next steps is to probe how quantified TFs regulate together. the ability to distinguish and quantify the stabilization and acceleration mode of regulation and characterize TFs based on them is important for developing general theories of regulation that include TFs acting on different kinetic steps of the transcription process [77,78,79]; predictions for the combined regulatory effect of two stabilizing TFs should be different than predictions for a stabilizing TF acting together with an accelerating TF [77]. With each TF characterized, we can develop an empirical baseline or null hypothesis for what a TF should do on a gene: departures from this expectation, due to the emergence of complex regulatory phenomenon brought about by TF-TF interactions [63], allosteric interactions [80,81] or other effects indicate surprises that warrant testing in these expanded models.…”

The linear framework: using graph theory to reveal the algebra and thermodynamics of biomolecular systems

“…What happens to an output function on G ? It can be shown that, provided G itself is at thermodynamic equilibrium, any output function on G can be rewritten as a non-negative linear combination of steady-state probabilities of C ( G ),∑i∈νfalse(Cfalse(Gfalse)false)λiui∗false(Cfalse(Gfalse)false),1emλi≥0,where, crucially, the coefficients λ i do not depend on x , as long as G is at thermodynamic equilibrium [78]. Comparing with equation (5.1), we see that equation (5.4) defines an input–output function on C ( G ), which, as noted above, has the hypercube structure scriptCm.…”

“…Equilibrium parameter values are randomly chosen in some range, the corresponding false(p, sfalse) points calculated and the resulting point cloud is incrementally grown until a boundary is reached. The algorithms are elaborated in [14,15,78]. Part of the equilibrium position–steepness region for the regulated-recruitment gene-regulation model described above, based on the hypercube structure scriptC4+1, is shown in figure 3 c .…”