Gene-Specific Linear Trends Constrain Transcriptional Variability of the Toll-like Receptor Signaling

Abstract

Summary Single-cell gene expression is inherently variable, but how this variability is controlled in response to stimulation remains unclear. Here, we use single-cell RNA-seq and single-molecule mRNA counting (smFISH) to study inducible gene expression in the immune toll-like receptor system. We show that mRNA counts of tumor necrosis factor α conform to a standard stochastic switch model, while transcription of interleukin-1β involves an additional regulatory step resulting in increased heterogeneity. Despite different modes of regulation, systematic analysis of single-cell data for a range of genes demonstrates that the variability in transcript count is linearly constrained by the mean response over a range of conditions. Mathematical modeling of smFISH counts and experimental perturbation of chromatin state demonstrates that linear constraints emerge through modulation of transcriptional bursting along with gene-specific relationships. Overall, our analyses demonstrate that the variability of the inducible single-cell mRNA response is constrained by transcriptional bursting.

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

“…In this model, the transcriptional process is described by two main features: burst size, defined as the average number of mRNA produced per burst (i.e., gene activation event), and burst frequency, defined as the frequency with which the bursts occur (Nicolas et al, 2018). The observed mean-variance and mean-noise trends of the basal and TNF-stimulated transcript distributions of the target genes show deviation from Poisson behavior that is consistent with transcriptional bursting (Fig EV2A) (Singh et al, 2010;Skupsky et al, 2010;Wong et al, 2018;Bagnall et al, 2020). Thus, we expected the transcriptional bursting model would provide insight into the observed differences in transcriptional noise across NF-κB target genes.…”

“…This simple promoter model sufficiently captured differences in transcriptional bursting "modes" following TNF treatment in Jurkat T cells that was a main focus of our study. However, two recent studies analyzing transcriptional bursting in response to stimulation of NF-κB by TNF or LPS reported somewhat different results (Bagnall et al, 2020;Zambrano et al, 2020). Bagnall et al studied activation of Tnf and Il1b following LPS stimulation in macrophages and found gene-specific mean-noise trends for Tnf vs Il1b; while a two-state promoter model was sufficient to reproduce Tnf distributions, a three-state model with an additional unproductive (or "refractory") state was required to fit Il1b distributions.…”

“…Thus, we expected the transcriptional bursting model would provide insight into the observed differences in transcriptional noise across NF-κB target genes. We first estimated burst size and burst frequency using the sample variance σ 2 and the mean μ of the mRNA distributions (Raj et al, 2006;Suter et al, 2011;Nicolas et al, 2018;Bagnall et al, 2020). These estimates of burst size b m = σ 2 /μ (i.e., the Fano factor) and burst frequency f m = μ/(b m − 1) based on the sample moments are often used to describe deviation from Poisson distributions, for which b m = 1 and f m = ∞ (So et al, 2011;Nicolas et al, 2018;Bagnall et al, 2020).…”

“…We first estimated burst size and burst frequency using the sample variance σ 2 and the mean μ of the mRNA distributions (Raj et al, 2006;Suter et al, 2011;Nicolas et al, 2018;Bagnall et al, 2020). These estimates of burst size b m = σ 2 /μ (i.e., the Fano factor) and burst frequency f m = μ/(b m − 1) based on the sample moments are often used to describe deviation from Poisson distributions, for which b m = 1 and f m = ∞ (So et al, 2011;Nicolas et al, 2018;Bagnall et al, 2020). Analysis of burst size and burst frequency based on the moments of the smFISH count distributions revealed differences in how TNF affected Nfkbia, Tnfaip3, Tnf and Il8 vs Il6 and Csf2, with the former group exhibiting large increases in burst size (Fig EV2B ) and the latter group exhibiting much smaller increases in burst size but with increasing values of burst frequency (Fig EV2C).…”

Decoding the dynamics of multilayered stochastic antiviral IFN-I responses

“…IFN-I signaling transduction, with the phosphorylation of JAK1, TYK2, STAT1, and STAT2 identified as key events that drive IFN-I induction (reviewed in [52]). Finally, recent data interpretations (e.g., on scRNA-seq, single-molecule mRNA counting) and mathematical modeling suggest that seemingly noisy cellular decisions are defined by rather simple fundamental functional constraints, with variations in genome architecture being a major source of heterogeneity [38,[53][54][55].…”