Benjamin Martin, Maike M. K. Hansen, Dong Woo Hwang, Sonali Chaturvedi, Robert A. Coleman, Weihan Li, Matt Thomson, Leor S. Weinberger, Ravi V. Desai, Sheng Ding, Chen Yu, Xinyue Chen, and Robert H. Singer
INTRODUCTION: Fluctuations have long been known to dynamically shape microstate distributions in physical systems. Throughout engineering, “dithering” approaches that modulate fluctuations are used to enhance inefficient processes and, in chemistry, thermal fluctuations are amplified (e.g., by Bunsen burners) to accelerate reactions. In biology, a long-standing question is whether stochastic expression fluctuations originating from episodic transcription “bursts” play any physiologic role. RATIONALE: Stochastic fluctuations (noise), measured by the coefficient of variation, scale inversely with mean expression level. For example, transcriptional activators that increase the mean lead to decreased noise, whereas stressors that decrease the mean increase noise. However, this 1/mean “Poisson” scaling of transcriptional noise can be broken by certain processes (e.g., feedback) and, curiously, by small molecules such as pyrimidine nucleobases. We set out to determine the mechanism of action of nucleobases that amplify transcriptional noise independently of mean and explored their potential functional role. Specifically, we tested whether a noise-amplifying pyrimidine nucleotide and its naturally occurring base analogs decouple noise from the mean by disruption of a putative cellular noise control mechanism (i.e., a noise thermostat). RESULTS: We found that DNA surveillance and repair machinery decouple transcriptional noise from mean expression levels, homeostatically changing noise independently of mean, and this potentiates cell fate transitions in stem cells. Specifically, during removal of modified nucleotide substrates (e.g., idoxuri-dine) and naturally occurring nucleotide analogs [e.g., 5-hydroxymethylcytosine (hmC) and 5-hydroxymethyluridine (hmU)], transcriptional noise is amplified globally across the transcriptome. The amplified transcriptional noise is intrinsic (i.e., not cell extrinsic), independent of changes in the mean (i.e., occurs with minimal change in mean), and distinct from a stress response. Forward genetic screening identified AP endonuclease 1 (Apex1), a member of the base excision repair (BER) DNA surveillance pathway, as the essential mediator of homeostatic noise amplification, and up-regulation of BER enzymes upstream of Apex1 (e.g., glycosylases) also amplified noise. Single-molecule and live-cell imaging showed that this homeostatic noise amplification originated from shorter-duration, but higher-intensity, transcriptional bursts. Mechanistically, Apex1 amplified noise by altering DNA topology, i.e., by increasing negative DNA supercoiling, which impedes transcription but upon repair accelerates transcription, thereby homeostatically maintaining mean expression levels. We call this mechanism “discordant transcription through repair (“DiThR,” pronounced “dither”). Computational modeling predicted that DiThR could increase responsiveness to fate-determining stimuli and, indeed, experimental activation of DiThR potentiated both differentiation of embryonic stem cells into neural ectodermal cells and reprogramming of differentiated fibroblasts into induced pluripotent stem cells. CONCLUSION: Our data reveal that a DNA surveillance pathway uses the biomechanical link between supercoiling and transcription to homeostatically amplify transcriptional fluctuations. The resulting increase in expression excursions, or outliers, increases cellular responsiveness to diverse fate specification signals. Thus, DNA-processing activities that interrupt transcription could function in fate determination and may explain why naturally occurring base modifications, such as the oxidized nucleobase hmU, are enriched in embryonic stem cell DNA. The existence of a DiThR pathway that orthogonally regulates transcriptional fluctuations suggests that cells evolved mechanisms to exploit noise for the functional regulation of fate transitions and highlights the potential to harness these endogenous pathways for cellular reprogramming.