Promoter Decoding of Transcription Factor Translocation Dynamics

Abstract

Many cellular signaling pathways exhibit a bowtie topology: multiple distinct signal inputs converge on a single master transcription factor, which controls the expression of downstream genes. Recent evidence suggests that information about signal inputs can be encoded by regulating the activation dynamics of the master transcription factor. However, it was unclear whether this is sufficient to obtain specificity in gene expression, such that each input induces a specific set of output genes. Using the budding yeast transcription factor Msn2 as a model system, we address this question.

We systematically dissect how different promoters decode transcription factor translocation dynamics in single cells (Chapter 2). We find that promoters fall into four main classes depending on the threshold of signal required for activation and on the timescale of activation. Furthermore, we provide insight into the mechanistic basis for promoter class. We show that it is possible to differentially control expression of the different promoter classes by control of Msn2 dynamics. We find that slow promoters exhibit dramatically higher noise in gene expression, but are able to filter out Msn2 oscillations. This highlights a general trade-off: for promoters, implementing a high-pass temporal filter comes at the cost of much higher noise in gene expression.

Applying tools from information theory and focusing on gene expression, we rigorously quantify the limits on information transduction through regulation of Msn2 dynamics (Chapter 3). Although we find that the amount of information transduced by Msn2 to target genes is only sufficient for reliable binary decisions, information transduction can be improved by modulating promoter cis-elements or by integrating information from multiple genes. We find amplitude-encoding to be more reliable than frequency-encoding. Taken together, our results suggest that information transduction through regulation of Msn2 dynamics is limited to reliable transduction of signal identity, but not signal intensity.

The work we describe in this dissertation would not have been possible without the development of high-throughput microfluidic technologies (Chapter 4). We describe our development of a multiplexed microfluidic device, which we combine with four-color quantitative time-lapse microscopy to control nucleocytoplasmic shuttling of Msn2 and measure gene expression dynamics in single cells. We provide a detailed protocol for future studies.

Our work demonstrates that it is possible to encode multiple distinct gene expression programs in the dynamics of a single transcription factor. Nonetheless, at the level of individual genes, noise in the decoding step places an upper limit on information transduction.