Exploring the Regulation of Transcriptional Noise in Nf-κB-Inducible Genes Across Chromatin Environments

Date of Award

Spring 2022

Document Type


Degree Name

Doctor of Philosophy (PhD)


Biomedical Engineering (ENAS)

First Advisor

Miller-Jensen, Kathryn


There is a significant amount of cell-to-cell heterogeneity inherent in gene expression. One source of this heterogeneity in eukaryotes is transcriptional bursting, in which a gene promoter infrequently transitions from an inactive state to an active state, producing bursts of transcripts. Cell to cell variations lead to phenotypic differences downstream, which are implicated in immunological function. One model often used to study transcriptional bursting and its phenotypic consequences is human immunodeficiency virus (HIV). The HIV provirus integrates into the host genome, and in a subset of genomic locations, remains silent, or latent, until a perturbation stochastically reactivates viral transcription. Work to remove these latent reservoirs is hampered by the heterogeneity of reactivation which is attributed to varied regulation of transcriptional bursting. Several factors have been implicated to regulate the bursting dynamics such as transcription factor activity, chromatin remodeling, polymerase binding, and positive feedback. However, how exactly HIV, and other endogenous genes, modify their transcriptional profiles is not conclusive. Here, we use computational modeling techniques, built upon molecular experiments, to explore how the role of NF-κB transcription and chromatin environments effect transcriptional bursting, and how positive feedback further amplifies phenotypic differences. We first utilized a widely used two-state promoter model to simulate latent HIV gene expression. Using experimentally derived transcriptional rates from latent-but-inducible HIV clones, we showed that a two-state promoter model can reproduce experimentally observed noise in HIV transcription. We then coupled this two-state promoter model to a positive feedback loop reflecting HIV’s own positive transcriptional transactivator. When this noise is amplified through viral positive feedback, we also qualitatively reproduced experimentally observed viral phenotypes and how they vary across different viral integration positions. We then explored a similar methodology for another NF-κB-regulated gene, tumor necrosis factor (TNF). Here, we also were able to replicate bursting dynamics of activation, and further amplified diversity with positive feedback. This work highlights the importance of NF-κB signaling dynamics and amplification of positive feedback. Finally, building upon discrepancies in experimental chromatin configuration measurements in the two-state model, we switched to a three-state promoter model which could both replicate chromatin remodeling as well as imitate polymerase pausing before transcription. This new model allowed for varied fractional promoter probability states and could be activated by increasing the paused polymerase release rate. We coupled this model to a positive feedback loop and could reproduce the experimentally observed phenotypes. Most importantly, however, the three-state model reproduced experimentally observed bimodal patterns of HIV expression, which the two-state model failed to do in our parameter space. This discovery illustrates the superiority of the three-state model to better represent biological chromatin mechanisms and capture viral expression and noise within a single activation profile. Altogether, this work highlights the importance of chromatin environment and NF-κB dynamics in the regulation of transcriptional noise, and the role of positive feedback to amplify this noise. Utilizing two- and three-state modeling, we can replicate activation of HIV and TNF, and postulate possible areas of hypothesized activation profiles. More broadly, our work highlights important areas of noise modulation that can change phenotypic outcomes to influence clinical applications, as in the case of latent HIV.

This document is currently not available here.