The NF-κB family of transcription factors (RelA [RELA], c-Rel [REL], RelB [RELB], p50 [NFKB1], and p52 [NFKB2]) regulates many critical biological processes including development, immunity, and inflammatory responses. It has a dominant role in regulating gene transcription in almost all mammalian cell types in response to a wide range of external, internal and environmental cues. Understandably, its misregulation results in the development of severe diseases, such as chronic inflammatory disorders, autoimmune diseases, and cancer. While the NF-κB signaling cascade has been extensively studied and each individual NF-κB knockout strain (Rela-/- for RELA protein, Rel-/- for CREL protein, Relb-/- for RELB protein, Nfkb2-/- for NFKB2 protein, and Nfkb1 -/- for NFKB1 protein) exhibited different phenotypes in mouse, it is still poorly understood whether there exists a distinctive subset of target genes for each NF-κB family member. Our lab previously reported that Il12b gene transcription is strongly dependent on a unique 46-residue segment of c-Rel which promotes strong binding of the protein to two non-consensus κB sequences in the Il12b promote. Here we further investigate NF-κB selectivity at a genomic scale using chromatin RNA-seq and ChIP-seq technologies. Strikingly, c-Rel appears to be exclusively dedicated to controlling the expression of Il12b during lipid A-stimulated mouse bone marrow-derived macrophages (BMDM), as no other genes demonstrated the same dependence and preference for c-Rel binding.

To carefully modulate NF-κB signaling, numerous regulatory layers are used by the cell, one of which is mediated by the nuclear IκB proteins (BCL3 or Bcl-3, NFKBID or IκBNS, and NFKBIZ or IκBz). Unlike prototypic IκBs (IκBa, IκBb, and IκBe), which sequester NF-κB in the cytoplasm during homeostasis and release it upon stimulation, nuclear IκBs are located in the cell nucleus and are thought to serve as co-activators or co-repressors. Furthermore, nuclear IκB selectively interacts with p50 (NFKB1) and p52 (NFKB2), two of the NF-κB proteins that do not contain transactivation domains (TADs). Both Bcl-3 and IκBz, but not IκBNS, contain TADs and thus are able to alter the repressive nature of p50 or p52 via protein-protein interaction. To study the regulatory effects of nuclear IκBs in transcription, we performed chromatin RNA-seq and ChIP-seq on lipid A-stimulated BMDM. Lipid A induces the expression of both primary and secondary response genes in BMDMs, including Nfkb1, Nfkb2, and all three nuclear IκBs. RNA-seq was performed with Nfkb1-/-, Nfkb2-/-, Bcl3-/-, Nfkbid-/-, and Nfkbiz-/- cells. Our results indicate that the absence of p52, Bcl-3, or IκBNS has only minor effects on Lipid A induced TLR4-mediated gene transcription. However, a subset of inducible genes depends strongly on the expression of p50 and IκBz. Taking into account the corroborative data from RNA- and ChIP-seq experiments, we delineate and classify genes that are either directly or indirectly regulated by p50 and IκBz.

The activation of tissue specific genes relies upon the precise orchestration of a number of events that result in the initiation of transcription upon lineage specification. This process is heavily dictated by the chromatin environment both at the promoter and distal sequences, as well as by the availability of transcription factors necessary for activation. A critical role is played at distal regulatory sequences, which often are the first sites to be engaged by key regulatory proteins. This interaction often promotes a chromatin environment that is necessary for the activation of the gene and results in the recruitment of additional sequence specific factors and a direct interaction with the promoter to initiate transcription. Understanding the properties of enhancer elements for tissue specific genes is important for a clear understanding of the mechanisms of activation. A number of studies have shown that enhancers are marked long before the activation of the gene takes place, in some cases as early as the embryonic stem cell stage. A detailed study described an unmethylated window within the enhancer of the Ptcra locus. Further analysis showed the enhancer mark to be regulated by sequence specific binding factors. These studies lacked the appropriate chromatin environment, which we know to be important. Here we use bacterial artificial chromosome containing the Ptcra locus to demonstrate that the enhancer mark persists in a chromatin context but is not regulated in the manner described in a non-native chromatin context. We then expand our studies to global tissue specific gene expression in order to understand more broadly the regulatory properties the define tissue specific genes. Parsing the mechanisms that drive tissue specific gene expression is critical for an understanding of pluripotency and tissue specificity. Here we use deep chromatin RNA-sequencing to accurately quantify the transcriptome of pluripotent stem cells and four primary differentiated cell types - E14.5 cortical neurons, CD4+ CD8+ thymocytes, bone marrow-derived macrophages and hepatocytes, in mouse. We define tissue specific genes with the broadest dynamic range in expression and define the chromatin properties at their promoters. Separating tissue specific genes with the largest dynamic range in expression allowed us to uncover cell type specific differences in the fundamental promoter properties.

Through billions of years of evolution, immune cells have developed diverse mechanisms to resolve pathogenic infection and heal injury. Proper immune cell activation includes receptor- ligand engagement, intracellular signaling cascades, and transcriptome-wide changes. Dysregulation of proinflammatory cytokines during infection can lead to severe complications. The NF-kB family regulates many pro-inflammatory cytokines associated with inflammation and immune system imbalance, such as Tnfa, Il1b, and Il6. Although NF-kB is one of the best-studied transcription factors, the molecular mechanisms underlying dimer specific function remain largely unknown. Our research explores the role of dimers that are not essential for cellular or organismal survival but have distinct roles in immune cell activation. This research employs a macrophage model system to define dimer-specific roles of NF-kB and the molecular mechanisms underlying their regulation. Using a reductionist system enables us to explore dimer-specific functions with unprecedented depth in the cell context. We focused on c-Rel and p50 due to their potent activation downstream of viral and bacterial PRRs and their highly specific roles in immune cellactivation. We employ both genomic and biochemical approaches such as RNA-seq, ChIP-seq, Sequential ChIP-seq, EMSAs, and co-immunoprecipitations to uncover the mechanistic role of c- Rel and p50. This research provides novel insights into the regulatory logic employed by NF-kB by revealing the underlying mechanisms that make dimer-specific functions possible.

Macrophages play a critical role in bridging the innate and adaptive immune systems, which requires timely and precise transcription control when responding to danger signals. Gene expression networks in macrophages have been profiled extensively in recent decades with the advancement of high-throughput sequencing technologies. However, unveiling critical mechanistic insights into control of diverse transcriptional networks has proven to be challenging. Recent evidence has suggested that the accessibility of chromatin structure can provide an integral means of transcription regulation. We have used the relatively recent Assay for Transposase-Accessible Chromatin using Sequencing (ATAC-seq) method to assess chromatin accessibility with optimal resolution in mouse bone marrow derived macrophages during the course of lipid-A stimulation. This advanced method enabled us to characterize the chromatin state of genome-wide regulatory regions. By applying a quantitative systematic approach, we obtained mechanistic insights into the highly selective role of the major transcription factors, NF-κB and IRF3, in nucleosome remodeling and transcription regulation.

Ikaros (Ikzf1) is a transcription factor necessary for the development of lymphoid and other hematopoietic lineages. Several studies have identified Ikaros as an important lymphoid tumor suppressor both in murine models and human cases of acute lymphocytic leukemia (ALL). Few gene targets of Ikaros regulation have been identified, and the mechanistic link to tumor suppression is unknown. Ikaros binds DNA through a region of four tandem C2H2 DNA-binding zinc fingers, and early characterization suggested that the first and fourth fingers have modulatory binding that allows for targeting of different consensus sequences. However, the binding sequence specificity of each of these fingers remained ambiguous. The Smale lab has developed a mutant mouse strain with a deletion of the exon encoding the fourth zinc finger, Ikzf1^δF4/>δF4, which among other phenotypes, develops spontaneous aggressive T cell lymphomas with 100% penetrance. These results suggest that the binding specificity of zinc finger 4 is necessary for regulating target genes that maintain Ikaros tumor suppressor function. This thesis examines the DNA binding specificity of Ikaros zinc fingers 1 and 4, and identifies a putative consensus sequence for each. The kinetics and stability of binding for purified proteins without zinc fingers 1 or 4 is also characterized using surface plasmon resonance. Mice with a heterozygous deletion of the Ikaros DNA binding domain, Ikzf1^DN/>+, also develop thymic lymphomas and have been observed to show a decreased threshold of stimulation for T cell activation. This phenotype is thought to have a connection with the malignant transformation of these cells. Ikzf1^δF4/>δF4 T cells are tested by in vitro assays and found to also have a significantly decreased threshold for activation. These assays are used to further investigate gene deregulation in activated Ikzf1^δF4/>δF4 T cells using RNA high throughput sequencing. Many deregulated genes with potential contribution to decreased activation threshold and loss of tumor suppressor function were identified by RNA-seq analysis. In combination with RNA-seq, Ikaros ChIP-seq data could be used to validate functional targets and associate loss of Ikaros occupancy with deregulation of gene expression. Ikaros target genes have remained largely elusive in part due to difficulty with traditional chromatin immunoprecipitation (ChIP) methods. A new technique is tested and optimized for future application to Ikaros ChIP-seq.

In an attempt to understand DNA methylation in various contexts, we have examined chromatin modification at enhancers and CpG islands. At both DNA features, we find the binding of transcription factors is the major determinant of methylation status.

At the enhancer for the tissue-specific inflammatory gene Il12b, we attempted to isolate the DNA sequence necessary for the establishment of a low methylation window usually present in most cell types. We cloned Il12b enhancer deletions into a bacterial artificial chromosome that could recapitulate native chromatin when stably transfected into murine ES cells, but were unable to remove the low methylation window without deleting the full enhancer sequence. The Il12b enhancer is uniquely methylated in embryonic stem cells compared to all other cell types; it has higher methylation than usual and responds to certain changes in growth conditions. DNA methylation increases globally during stem cell differentiation, but DNA methylation at the Il12b enhancer remains constant in successfully differentating cells. Finally, we take advantage of variable Il12b enhancer methylation in embryonic stem cells to demonstrate that, following differentiation to a macrophage fate, moderate enhancer methylation does not prevent Il12b expression.

To understand what factors influence the well studied low DNA methylation, histone 3 lysine 4 trimethylation (H3K4me3), and low nucleosome occupancy at CpG islands, we cloned CpG rich DNA into bacterial artificial chromosomes which were stably transfected into ES cells. Analysis of the integrated BACs revealed that CpG island features are each controlled through separate mechanisms. We determined several properties of CpG island features based on experimental deletions and fusions of a small CpG island and the 601 positioning sequence. Protection from DNA methylation at CpG islands can occur either by binding of a specific transcription factor, or by a size threshold mechanism in murine ES cells. H3K4me3 marking requires low DNA methylation, but unmethylated CpGs are not sufficient to recruit high levels. Nucleosome density is influenced by transcription factor binding and sequence positioning determinants, but is unaffected by low DNA methylation and moderate H3K4me3 levels.

We expanded our analysis of CpG islands to include all CpG rich regions in the human genome, which were computationally determined based on our own criteria. Using available chromatin datasets, we assayed the effect of nucleotide content on CpG island features. We found that CpG density and island size correlated with high levels of CpG island features. However, by far the strongest determinant of CpG island features was association with a promoter. Promoter CpG rich regions were strongly biased to accumulate high levels of all CpG island features, which could not be explained by nucleotide content. Instead, we showed that promoter CpG islands have much higher transcription factor binding then other CpG islands in the genome, and high binding is correlated with lower DNA methylation and higher H3K4me3. Finally, we observed a difference in the DNA methylation at human and mouse CpG islands. High CpG density mouse CpG islands are much more susceptible to demethylation.

This multifaceted study elucidates many previously undefined relationships between transcription factors and chromatin properties. These findings will be beneficial to describing the complex mechanisms that drive regulation of cell fate and gene expression.

Pluripotency factors Oct4, Sox2, and Nanog orchestrate an elaborate hierarchy of gene regulation governing embryonic stem cell (ESC) identity. The capability of differentiating into cell from any of the three germ layers offers a potential solution and path toward the discovery of novel therapies for devastating diseases. The differentiation of ESCs requires three fundamentally distinct transitions in the transcriptional state: 1) the activation of silent genes in ESCs that are transcribed in differentiated states, 2) the silencing of ESC-transcribed genes that are entirely inactive in the differentiated cells, 3) the modulation of transcription for genes that are expressed in both populations. Deciphering the mechanisms regulating the transitions of these transcriptional states will illuminate how the pluripotent state is established and maintained by the regulation of pluripotency factors. Because Oct4 and Sox2 are the core regulators of the transcriptional regulatory network for pluripotency, we focused our studies on mechanisms regulated by Oct4 and Sox2 in ESCs.

Oct4 and Sox2 bind to thousands of enhancer composite sites at pluripotency genes and differentiation-promoting genes. These Oct4/Sox2 target genes exhibit distinct transitions of transcriptional states for the establishment of pluripotency. As distinct mechanisms are likely to regulate different transcriptional states, a critical step toward a mechanistic understanding of pluripotency is to delineate the genes with well-defined transcription characteristic. Because most studies have relied on low stringency criteria to define differential expression to infer regulatory mechanisms, they did not rigorously evaluate the selective functions of Oct4/Sox2 composite binding critical for establishing pluripotency. To scrutinize how Oct4 and Sox2 account for possible differences in the regulatory mechanisms necessary for distinct transcriptional states, we refined the gene classification and interrogated the role of Oct4/Sox2 binding at enhancer composite motifs at distinct gene classes. By combining RNA-seq, ChIP-seq, and ATAC-seq with functional validation employed by CRISPR/Cas9 mutagenesis, we discovered that Oct4/Sox2 function differently across gene groups with various transcriptional states. In addition to a role in transcriptional activation at ESC-specific and Dynamic genes, my data suggest that Oct4/Sox2 motifs at silent genes may mediate the transcriptional repression in ESCs. In Chapter 3, we also employed a gene-centric approach to quantitatively compare transcription factor co-binding, histone modifications, chromatin accessibility, enhancer properties, and the genomic context of transcriptional regulation by Oct4 and Sox2 in ESCs. Together, we extended our understanding of the critical role played by Oct4 and Sox2 in the establishment of pluripotency.

The immune system is essential for host defense to pathogen infection, tissue repair, stress response, and other physiological functions. An unbalanced immune system is detrimental to homeostasis and mammalian survival. Immunodeficiency can cause susceptibility to pathogen infection. A hyperactive immune system can cause autoimmunity and chronic inflammatory diseases including rheumatoid arthritis, psoriasis, and atherosclerosis. Deciphering the mechanisms regulating the immune response will illuminate on the pathogenesis of inflammatory diseases and promote the development of new therapies to treat these diseases. Because innate immunity is the first line of host defense and transcription is a critical contributor to the innate immune response, we focused our studies on mechanisms regulating transcriptional activation of the innate immune response.

Toll-like receptor (TLR) signaling is a classical model to study pathogen recognition and the activation of innate immunity. TLR transcriptional cascades have been extensively studied using conventional systems approaches. Because conventional systems approaches rely on statistics and large sample sizes to uncover the common regulatory mechanisms, they often miss gene-specific regulatory mechanisms. To reveal gene-specific regulatory mechanisms, we used a stringent systems approach to dissect the TLR transcriptional cascades in chapter 2. To prevent biases towards the majority of weakly induced genes, we separated strongly induced genes from weakly induced genes. Combining high-resolution transcriptional profiles and perturbation studies, we classified TLR4-activated genes by their activation mechanisms. By integrating RNA-seq, ChIP-seq, ATAC-seq, and motif data, we found that several key inflammatory genes are regulated by highly selective mechanisms. TLR4-activated nuclear factor kappa B (NFκB) and interferon regulatory factor 3 (IRF3) selectively regulate a small subset of pro-inflammatory genes by chromatin and combinatorial regulation. We also found serum response factor (SRF) selectively regulates a few early transient primary response genes. In chapter 3, we used a gene-centric method and identified the motif rules governing the combinatorial regulation of serum response factor (SRF) and ternary complex factor (TCF). Ternary complex formation can enhance SRF binding and promote histone mark deposition. The active ternary complex can regulate transcription through either promoter or enhancer. Taken together, we demonstrated a gene-centric, stringent systems approach that can complement the conventional systems approach to unveil gene-specific regulatory mechanisms in ligand-induced transcriptional cascades.