TNF stimulation primarily modulates transcriptional burst size of NF-κB-regulated genes

Abstract

Cell-to-cell heterogeneity is a characteristic feature of the tumor necrosis factor (TNF)-stimulated inflammatory response mediated by the transcription factor NF-κB, motivating an exploration of the underlying sources of this noise. Here we combined single-transcript measurements with computational models to study transcriptional noise at six NF-κB-regulated inflammatory genes. In the basal state, NF-κB-target genes displayed an inverse correlation between mean and noise. TNF stimulation increased transcription while maintaining noise, except for the most repressed genes. By fitting transcript distributions to a two-state model of promoter activity, we found that TNF primarily stimulated transcription by increasing burst size while maintaining burst frequency. Burst size increases were associated with enrichment of initiated-but-paused RNA polymerase II at the promoter, and blocking the release of paused RNAPII with a small molecule inhibitor decreased TNF-stimulated burst size. Finally, we used a mathematical model to show that TNF positive feedback further amplified gene expression noise resulting from burst-size mediated transcription, leading to diverse TNF functional outputs. Our results reveal potential sources of noise underlying intercellular heterogeneity in the TNF-mediated inflammatory response.

Variability of the innate immune response is globally constrained by transcriptional bursting

“…However, we found that many genes with non-zero intercept fits were associated with substantial basal expression in untreated cells, which was also observed previously for the more quantitative smFISH data (Bagnall et al, 2020). Basal expression of the related gene targets has been shown to exhibit different bursting kinetics and mode of regulation from the inducible expression (Bass et al, 2021), which in part may explain the fitted non-zero intercepts for a subset of genes. For α 0 = 0, linear constraints essentially imply that the burst size must be constant (and equal to the slope of the mean-variance line), while the frequency undergoes modulation with the population mean changes in response to stimulation.…”

“…While intuitively zero intercept is expected (i.e., no expression in untreated conditions), these in part reflect the empirical nature of these trends and the limited sample size, especially for those genes where α 0 is small (in relation to variance), for example, Cxcl10 (Figure 2C). However, many genes, including Pnf1 and Eif6 exhibit substantial basal expression in untreated cells (Bass et al, 2021), resulting in either elevated or reduced variability (in relation to true zero) being captured via nonzero intercept in the regression fit (Bagnall et al, 2020).…”

“…Recent advances have demonstrated key insights into regulation of transcriptional bursting. In general, the bursting kinetics are genespecific and subject to regulatory control via cellular signalling events (Suter et al, 2011;Larson et al, 2013;Megaridis et al, 2018;Wong et al, 2018;Bass et al, 2021) as well as genome architecture and promoter sequences (Dar et al, 2012;Dey et al, 2015;Zoller et al, 2015;Hagai et al, 2018;Ochiai et al, 2020;Einarsson et al, 2022). For example, core promoters control burst sizes, while enhancer elements modulate burst frequency to define cell-type specific (Larsson et al, 2019) or circadian gene expression outputs (Nicolas et al, 2018).…”

“…These analyses suggest that complex transcriptional regulation at a single cell level may be globally characterised by mean-variance relationships of genespecific mRNA outputs, providing new ways to characterise response variability. While quantitative smFISH provides important insights, this approach is often limited by the number of genes, which can be investigated (Raj et al, 2008;Zenklusen et al, 2008;Larson et al, 2013;Lee et al, 2014;Gomez-Schiavon et al, 2017;Bagnall et al, 2018;Bagnall et al, 2020;Bass et al, 2021), therefore further analyses of global gene expression patterns (Larsson et al, 2019;Ochiai et al, 2020) are required to fully understand the underlying regulatory constraints.…”

Enhanced transcriptional heterogeneity mediated by NF-κB super-enhancers

“…Furthermore, stimulation with other anti-IgM concentrations demonstrated a high bimodality of NFKBIA expression across doses, whereas expression of CD83 demonstrated a long-tailed distribution (Supplementary Fig. 5d), which may be explained by its larger transcriptional burst size 39 .…”