Gene regulatory logic of the interferon-β enhancer contains multiple selectively deployed modes of transcription factor synergy
Schiffman, A.; Cheng, F.; Ourthiague, D.; Hoffmann, A.
Show abstract
Type I interferon IFN{beta} is a key regulator of the immune response, and its dysregulated expression causes disease. The regulation of IFN{beta} promoter activity has been a touchpoint of mammalian gene control research since the discovery of functional synergy between two stimulus-responsive transcription factors (TFs) nuclear factor kappa B (NF{kappa}B) and interferon regulatory factors (IRF). However, subsequent gene knockout studies revealed that this synergy is condition-dependent such that either NF{kappa}B or IRF activation can be dispensable, leaving the precise regulatory logic of IFN{beta} transcription an open question. Here, we developed a series of quantitative enhancer states models of IFN{beta} expression control and evaluated them with stimulus-response data from TF knockouts. Our analysis confirmed that TF synergy is a hallmark of the regulatory logic but that it need not involve NF{kappa}B, as synergy between two adjacent IRF dimers is sufficient. We found that a sigmoidal binding curve at the distal site renders the dual IRF synergy mode ultrasensitive, allowing it only in conditions of high IRF activity upon viral infection. In contrast, the proximal site has high affinity and enables expression in response to bacterial exposure through synergy with NF{kappa}B. However, its accessibility is controlled by the competitive repressor p50:p50, which prevents basal IRF levels from synergizing with NF{kappa}B, such that NF{kappa}B-only stimuli do not activate IFN{beta} expression. The enhancer states model identifies multiple synergy modes that are accessed differentially in response to different immune threats, enabling a highly stimulus-specific but also versatile regulatory logic for stimulus-specific IFN{beta} expression. SIGNIFICANCEPrecise regulation of the immune cytokine IFN{beta} is essential for human health. Classic studies established that the transcription factors NF{kappa}B and IRF function synergistically in activating IFN{beta} expression. However, more recent data revealed that either factor may be dispensable in certain conditions, leaving the regulatory logic of IFN{beta} transcription an open question. In this study, we evaluated several quantitative models to determine what regulatory logic can account for the available data. We found that the enhancer exhibits multiple synergy modes with specific transcription capabilities that are deployed in a condition-dependent manner. The resulting IFN{beta} regulatory logic model reveals how stimulus-specificity is achieved, and may be tuned, thereby overcoming a bottleneck in predictive modeling the control of innate immune responses.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- A new pipeline SPICE identifies novel JUN-IKZF1 composite elements 94%
- Synthetic analysis of chromatin tracing and live-cell imaging indicates pervasive spatial coupling between genes 94%
- Quantification of circadian interactions and protein abundance defines a mechanism for operational stability of the circadian clock 94%
Similar papers in this journal
- A transcriptional cycling model recapitulates chromatin-dependent features of noisy inducible transcription 94%
- Transcription factor interactions explain the context-dependent activity of CRX binding sites 93%
- Discovering differential genome sequence activity with interpretable and efficient deep learning 93%
Similar papers in this journal
- Deciphering enhancer sequence using thermodynamics-based models and convolutional neural networks 95%
- Bimodal Specificity of TF-DNA Recognition in Embryonic Stem Cells 94%
- Predicting the DNA binding specificity of transcription factor mutants usingfamily-level biophysically interpretable machine learning 94%
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.