Transcriptional regulation refers to the molecular systems that control the concentration of mRNA species within the cell. Variation in these controlling systems is not only responsible for many diseases, but also contributes to the vast phenotypic diversity in the biological world. There are powerful experimental approaches to probe these regulatory systems, and the focus of my doctoral research has been to develop and apply effective computational methods that exploit these rich data sets more completely. First, I present a method for mapping genetic regulators of gene expression (expression quantitative trait loci, or eQTL) using Random Forests. This approach allows for flexible modeling and feature selection, and results in eQTL that are more biologically supportable than those mapped with competing methods. Next, I present a method that finds interactions between genes that in turn regulate the expression of other genes. This is accomplished by finding recurring decision motifs in the forest structure that represent dependencies between genetic loci. Third, I present a method to use distributional differences in eQTL data to establish the regulatory roles of genes relative to other disease-associated genes. Using this method, we found that genes that are master regulators of other disease genes are more likely to be consistently associated with the disease in genetic association studies. Finally, I present a novel application of Random Forests to determine the mode of regulation of toxin-perturbed genes, using time-resolved gene expression. The results demonstrate a novel approach to supervised weighted clustering of gene expression data.

Recent statistics reveal breast cancer as the most common cancer among women and accounts for approximately 41,000 mortalities per year. In diagnosis, features such as stage, grade, lymph node metastasis are important prognostic indicators that help guide physicians and oncologist towards optimal patient care. Presence of established pathological markers such as ER, PR, and Her2/neu status would indicate ideal adjuvant therapy situation. Although treatment of these types of breast cancer is well established, cancer that lack all three receptors, “triple negatives” or “basal like” do not respond to adjuvant therapy and are considered more aggressive in that patients tend to recur early and experience visceral metastasis. Although scientists have uncovered numerous molecular biology mechanisms in search of an understanding in cancer, leading to development of fields such as apoptosis or growth pathways; cell cycle; angiogenesis; metastasis; and more recently cancer stem cells, much work remains as cancer is still not eradicated. MicroRNAs (miRNAs) are post transcriptional regulators of gene expression. Their discovery and functional understanding have only been uncovered in the past ten years. Long pri-miRNAs are transcribed from the genome and processed into premiRNAs by Dicer; and then into short single stranded mature miRNAs complexed with Argonaute proteins to inhibit protein translation. The first link of miRNAs to cancer was made only relatively recently, but the field has expanded exponentially since. TGF- β induced Epithelial to Mesenchymal Transition model in Normal Mouse Mammary Gland Epithelia Cells (NMuMG) is a commonly used model to dissect the molecular processes of breast cancer metastasis. Using miRNA microarray, we demonstrated miR-155 was upregulated along with alterations of other miRNAs. This observation was validated with Northern and qRT-PCR analysis. Promoter and ChIP analysis revealed TGF- β activated the Smad4 transcriptional complex to induce the expression of miR-155. The reduction of RhoA protein levels by ubiquitination has been described to be a critical step during EMT, and we showed miR-155 down regulates RhoA proteins without degrading its mRNA levels; therefore, preventing de novo synthesis of RhoA proteins in the course of EMT. The interaction between miR-155 and RhoA’s 3’UTR was confirmed by reporter assays. In summary, we reported the importance of miR-155 during TGF β induced EMT in NMuMG cells. FOXO3a is a well studied tumor suppressor transcriptional factor and resides in the nucleus to transcribe pro-apoptotic genes such as Bim, or p27 in the active state. During conditions when cells are signaled to grow and divide, it is phosphorylated by oncogenes such as AKT or IKK β, becomes inactivated and translocates into the cytoplasm. We have shown for the first time that FOXO3a activity is also regulated by miRNAs, specifically miR-155. Western and Northern analysis revealed a correlation between FOXO3a protein and mature miR-155 RNA levels in breast cancer cell lines along with breast tumor and normal tissues. Specifically, miR-155 expression is low in BT474 and high in HS578T, and inversely correlates with endogenous FOXO3a protein levels. Overexpression of miR-155 decreased endogenous FOXO3a protein and knockdown of miR-155 HS578T rescued its expression. Reporter assay experiments validated the interaction between miR-155 and FOXO3a 3’UTR. More importantly, overexpression of miR-155 in BT474 protected the cells from apoptosis induced by drugs while knockdown of miR-155 in HS578T initiated cell death even in the absence of drugs. In summary, we have shown the importance of miR-155 in chemosensitivity by targeting FOXO3a in breast cancer. MiR-155 has been previously shown up-regulated in multiple types of malignancies, including breast cancer. In addition, miR-155 expression was reported to correlate very strongly to survival in lung and pancreatic cancer. We validated by qRTPCR and Northern analysis that miR-155 expression is detected only in breast tumors and not normal breast tissue. In situ hybridization of breast cancer tissue microarrays revealed similar results. In light of previous studies that showed a correlation between miR-155 and survival in lung and pancreatic cancers, we performed an X-tile analysis to determine an optimal cut point for miR-155 level in our breast cancer sample population that would correlate to ten years overall survival. Verification using Kaplan-Meier validated a cut point at 90.14 to significantly correlate to overall survival (P=0.007). In addition, Chi-square analysis revealed miR-155 expression to correlate with high tumor stage, grade and lymph node metastasis. However, miR-155 expression did not correspond to ER, PR, or HER2/neu status, but this is hardly surprising since computational analysis does not predict miR-155 to target these genes. In summary, we have shown deviant expression of miR-155 in breast cancer. Due to its correlation with overall survival; higher grade and stage; lymph node metastasis, and triple negative subtype, miR-155 may prove to be a valuable prognostic marker and therapeutic target for breast cancer intervention.

Members of the forkhead class O (FOXO) transcription factors are tumor suppressors and key molecules that control aging and lifespan. The stability of mammalian FOXO proteins is controlled by proteasome-mediated degradation but general ubiquitin E3 ligases for FOXO factors remain to be defined. The current studies demonstrate that MDM2 bound to FOXO1 and FOXO3A and promoted their ubiquitination and subsequent degradation, a process apparently dependent on FOXO phopshorylation at PKB sites and on the E3 ligase activity of MDM2. The binding occurred between endogenous proteins and was involved the forkhead box of FOXO1 and the region of MDM2 that controls its cellular localization. MDM2 promoted the ubiquitination of FOXO1 in vitro in a cell free system. Knocking down MDM2 by siRNA caused the accumulation of endogenous FOXO3A protein, and enhanced the expression of FOXO target genes. In addition, MDM2 promoted the transcriptional activity of FOXO in a transient transfection system. In cells stably expressing a temperature sensitive mutant p53, activation of p53, by shifting to permissive temperatures led to MDM2 induction and the degradation of endogenous FOXO3A. These data suggested that MDM2 acts downstream of p53 as an E3 ubiquitin ligase to promote the degradation of mammalian FOXO factors.

Variable selection methods play important roles in modeling high-dimensional data and are key to data-driven scientific discoveries. In this thesis, we consider the problem of variable selection with interaction detection. Instead of building a predictive model of the response given combinations of predictors, we start by modeling the conditional distribution of predictors given partitions based on responses. We use this inverse modeling perspective as motivation to propose a stepwise procedure for effectively detecting interaction with few assumptions on parametric form. The proposed procedure is able to detect pairwise interactions among p predictors with a computational time of \(O(p)\) instead of \(O(p^2)\) under moderate conditions. We establish consistency of the proposed procedure in variable selection under a diverging number of predictors and sample size. We demonstrate its excellent empirical performance in comparison with some existing methods through simulation studies as well as real data examples. Next, we combine the forward and inverse modeling perspectives under the Bayesian framework to detect pleiotropic and epistatic effects in effects in expression quantitative loci (eQTLs) studies. We augment the Bayesian partition model proposed by Zhang et al. (2010) to capture complex dependence structure among gene expression and genetic markers. In particular, we propose a sequential partition prior to model the asymmetric roles played by the response and the predictors, and we develop an efficient dynamic programming algorithm for sampling latent individual partitions. The augmented partition model significantly improves the power in detecting eQTLs compared to previous methods in both simulations and real data examples pertaining to yeast. Finally, we study the application of Bayesian partition models in the unsupervised learning of transcription factor (TF) families based on protein binding microarray (PBM). The problem of TF subclass identification can be viewed as the clustering of TFs with variable selection on their binding DNA sequences. Our model provides simultaneous identification of TF families and their shared sequence preferences, as well as DNA sequences bound preferentially by individual members of TF families. Our analysis may aid in deciphering cis regulatory codes and determinants of protein-DNA binding specificity. / Statistics

Regulation of gene expression is essential for many biological processes. Binding of transcription factors to DNA is a key regulatory step in the control of gene expression. It is commonly observed that DNA sequences with high affinity for transcription factors occur more frequently in the genome than the instances of genes bound or regulated by these factors. However, the mechanism by which transcription factors selectively identify and regulate these genes was unclear. I utilized the transcriptional control of the phosphate-responsive signaling pathway (PHO) in Saccharomyces cerevisiae as a model system to address this problem.

Developing targeted genome regulation approaches holds much promise for accelerating trait discovery and development in agricultural biotechnology. Clustered Regularly Interspaced Palindromic Repeats (CRISPRs)/CRISPR associated (Cas) system provides bacteria and archaea with an adaptive molecular immunity mechanism against invading nucleic acids through phages and conjugative plasmids. The type II CRISPR/Cas system has been adapted for genome editing purposes across a variety of cell types and organisms. Recently, the catalytically inactive Cas9 (dCas9) protein combined with guide RNAs (gRNAs) were used as a DNA-targeting platform to modulate the expression patterns in bacterial, yeast and human cells. Here, we employed this DNA-targeting system for targeted transcriptional regulation in planta by developing chimeric dCas9-based activators and repressors. For example, we fused to the C-terminus of dCas9 with the activation domains of EDLL and TAL effectors, respectively, to generate transcriptional activators, and the SRDX repression domain to generate transcriptional repressor. Our data demonstrate that the dCas9:EDLL and dCas9:TAD activators, guided by gRNAs complementary to promoter elements, induce strong transcriptional activation on episomal targets in plant cells. Moreover, our data suggest that the dCas9:SRDX repressor and the dCas9:EDLL and dCas9:TAD activators are capable of markedly repressing or activating, respectively, the transcription of an endogenous genomic target. Our data indicate that the CRISPR/dCas9:TFs DNA targeting system can be used in plants as a functional genomic tool and for biotechnological applications.

Herein we describe the regulatory significance and molecular targeting of novel non-B-DNA secondary structures formed from the PDGFR-Beta core promoter nuclease hypersensitivity element.

Huntington’s disease (HD) is caused by expression of the huntingtin gene containing an expanded CAG repeat. N-terminal mutant huntingtin protein (N-mHtt) accumulates in the nucleus and impairs transcription of a subset of genes through incorporation into transcriptional complexes or sequestration of proteins away from the promoter. CREB-binding protein (CBP) is a transcriptional co-activator and acetyltransferase (AT) that binds to N-mHtt. We hypothesized that overexpressing CBP fragments that lack a promoter association domain would block N-mHtt-mediated transcriptional dysregulation and toxicity. We found that overexpressing full-length CBP or CBP fragments did not reverse transcriptional dysregulation, but did decrease toxicity in a cell model of HD. Overexpressing fragments of CBP containing the AT domain increased toxicity in wild-type cells, while overexpressing a fragment lacking this domain had no effect. We conclude that excess AT activity was detrimental in wild-type cells, while overexpressing CBP or CBP fragments was protective in HD cells.

Both microRNAs (miRNAs) and RNA-binding proteins (RBPs) regulate post- transcriptional events, but the post-transcriptional contribution to the global mammalian transcriptomes is still not well understood. In this study we study the synergistic interaction between microRNAs that inhibit gene production, and a special RBP, HuR, that positively regulates mRNA stability. We examined their relationship in terms of spatial, conservational and expressional perspective. We show comprehensive mapping of HuR binding sites by combination of its structural and sequential preferences; and cross-platform normalization method within a process of refining miRNA and HuR binding site mapping. Finally, we observed co-evolution of miRNA and HuR binding sites by looking at their proximity and conservation levels. Collectively, our data suggest that mammalian microRNAs and HuR, with seemingly opposing regulatory effects, cooperatively regulate their mutual targets.

No description available.