Browsing by Author "Keene, Jack D"

Post-transcriptional events are crucial determinants of gene expression, and aberrant expression patterns resulting from misregulation are evident in many pathological states. Cancer has traditionally been viewed as being driven by aberrant transcriptional regulation and signaling events, though, over the past several years, many RNA binding proteins and non-coding RNAs have emerged as critical players in tumor development. It is now recognized that regulation of post-transcriptional processes, such as mRNA stability and translation, robustly influence cancer-related gene expression patterns of proto-oncogenes, growth factors, cytokines, and cell cycle regulators. Despite its recognized importance, mechanisms of post-transcriptional regulation that influence molecular pathways at the mRNA level are understudied in the context of tumorigenesis. Additionally, cancer cells are derived from normal cells that often evolve step-wise and progressively to a neoplastic state, and the involvement of post-transcriptional regulation has not been looked at in the context of tumor initiation and step-wise progression. Thus, more studies are needed in order to fully understand the post-transcriptional mechanisms activated by cancer driver mutations that coordinate tumor initiation and progression.

In this dissertation, we aimed to elucidate mechanisms of post-transcriptional regulation coordinating tumorigenesis. We first established a genetically defined, primary cell-derived model of breast cancer initiation and progression. In this model, normal human mammary epithelial cells were immortalized through the expression of hTERT, p53DD, cyclin D1, CDK4R24C and c-MYCT58A, and subsequently converted to a tumorigenic state through expression of oncogenic H-RASG12V. Using RNA-sequencing and real-time PCR arrays, we comprehensively quantified changes in mRNA abundance, miRNA expression and alternative splicing in this system, and revealed thousands of changes during immortalization and relatively few changes during RAS transformation. Moreover, pre-malignant, immortalized cells had expression signatures consistent with an epithelial-to-mesenchymal transition (EMT), but they expressed low levels of mesenchymal protein markers and were non-invasive. Activation of RAS in these pre-malignant cells induced an invasive phenotype without major changes in global mRNA expression. Consistent with post-transcriptional mechanisms, RAS increased protein levels of Vimentin and N-cadherin without changing mRNA levels.

We then sought to investigate a mechanism of this RAS-induced post-transcriptional regulation. We used a method developed in our lab called Digestion-Optimized Ribonucleoprotein Immunoprecipitation coupled with RNA-sequencing (DO-RIP-seq) to identify and quantify transcriptome-wide binding sites for the RNA binding protein HuR. Our study is the first to identify and quantify transcriptome wide binding sites for any RBP during tumorigenesis, and we report that HuR quantitatively, but not qualitatively, changed association at individual mRNA binding sites during RAS transformation. We identified a GU-rich secondary motif associated with a decrease in HuR binding during transformation. Furthermore, our data suggest that HuR may cooperate with the CELF1 protein to positively regulate the translation of a subset of mRNAs and promote the EMT phenotype. We generated HuR CRISPR knockout cell lines and demonstrated that HuR expression was necessary for the maintenance of cancer traits, including proliferation, anchorage independent growth, migration and invasion, but it does not regulate mRNA stability in this context. Lastly, we identified a binding site position dependent mechanism by which HuR regulates alternative polyadenylation of mRNAs encoding proteins involved in cancer-related processes.

In conclusion, our findings indicate that EMT-associated invasion can be initiated through two sequential stages: transcriptional priming followed by oncogenic RAS-triggered post-transcriptional regulation. The HuR RNA binding protein is important for maintaining the cancer phenotypes induced by oncogenic RAS, and regulation by HuR may be, at least in part, determined by a GU-rich secondary motif as well as cooperation with the CELF1 RNA binding protein.

Precise regulation of the complex process of gene expression is essential for all aspects of life, and a large degree of this precision is mediated at the posttranscriptional level. The global and individual mechanisms by which posttranscriptional control is coordinated to maintain or alter levels of gene expression as necessary are not fully understood. Identification of the mRNA target sets of individual RNA binding proteins (RBPs) and characterization of the mechanisms by which RBPs regulate the expression of individual mRNAs provide some insight into the global structure of the posttranscriptional environment. However, few studies have integrated these findings into a global model of posttranscriptional control. We have explored the structure and function of the posttranscriptional regulatory system through a combination of global modeling approaches, global studies of mRNA translation and decay, and mechanistic studies of the function of individual RBPs, specifically HuR and Pum1.

By combining RBP-mRNA association data and transcription factor (TF) target data from separate global studies in yeast, we developed an integrated model of gene expression regulation. Evaluation of this model indicates that posttranscriptional regulation may be responsible for substantially greater contributions to the overall gene expression program than transcriptional regulation. Further, we identified a self-regulatory feature of the posttranscriptional network that suggests a `regulators of regulators' structure may be a defining feature of the posttranscriptional control of gene expression.

Additionally, we explored the mechanisms and functional consequences of the dynamic association between an RBP, HuR, and its target RNAs through a combination of modeling and experimental approaches, including polysome profile analysis and global measurement of RNA stability. Our model indicates that changes in total mRNA abundance are insufficient to fully explain the dynamics of association between HuR and its targets, suggesting a role for competition and cooperation with other RBPs. Our findings also indicate that HuR may play a role in inhibition of translation in a dynamic immunological system (T cell activation).

Finally, we performed a mechanistic analysis of the function of the Pum1 RBP and characterized the role of this protein in the translational regulation of several important target mRNAs through the use of luciferase reporter assays. We also provided the first in vivo evidence of a role for specific regions of the Pum1 protein in the mediation of gene expression. However, we were unable to verify previous in vitro reports of a role for Pum1 in the control of translation elongation of verified in vivo mRNA targets, suggesting that Pum1's regulatory function may be context dependent.

Ultimately, the approaches and findings in this study will provide a framework for the development of a global integrated model of posttranscriptional control. Through the iterative development of models and experimentation, hypotheses can be generated and, tested in the laboratory, and the results of these experiments will then further improve the development of the models. An integrated approach of this type will be necessary to fully understand the highly complex and interconnected nature of the gene expression regulatory system.

Orthopoxviruses encode many genes that are not essential for viral replication, which often account for differences in pathogenesis among otherwise closely related orthopoxviruses. Although dendritic cells (DCs) are essential to the generation of an effective anti-viral immune response, the effects of different orthopoxviruses on DC function is poorly understood. The objective of these studies was to determine the effect of different orthopoxviruses on DCs. Cowpox virus (CPXV) is ideally suited to this purpose because it encodes the largest and most representative set of accessory genes among orthopoxviruses, it is endemic in mouse populations, and can infect humans.

We hypothesized that CPXV would have novel mechanisms of evading the immune response that other orthopoxviruses lack, which may exert maximal effect in the context of antigen presenting cells such as DCs, allowing for discovery of novel viral strategies of immune evasion. To test this, CPXV was used to infected mouse bone marrow-derived DCs (BMDCs), and the effect of the virus on DC survival, expression of T-cell costimulatory molecules and cytokine production was determined. The effects of vaccinia virus strain Western Reserve (VV), the prototype of the species, and modified vaccinia virus strain Ankara (MVA), a promising vaccine vector, on mouse BMDCs were also determined. Confirming the hypothesis that CPXV would have different effects on mouse BMDCs from other orthopoxviruses, BMDCs infected with CPXV survived longer in culture than those infected with MVA or VV. In addition, CPXV specifically downregulated MHC I, MHC II, CD40, and CD86, and induced production of significant levels of IL-6 and IL-10.

Because IL-10 has many suppressive effects on the immune system, inducing IL-10 may provide a selective advantage to CPXV in vivo. To examine the role of IL-10 in a CPXV infection, wild type and IL-10 deficient mice were infected intranasally with CPXV. The effect of CPXV infection on disease morbidity, viral loads, inflammation and the protective immune response was determined. As expected, IL-10 was important in controlling inflammation during CPXV infection, but there was no effect on viral replication or clearance. Surprisingly, IL-10 was important in generation of a protective memory response to CPXV, which may reflect a novel role for IL-10 in the immune response.

In recent years, the pervasiveness and importance of post-transcriptional regulation has reshaped the underlying principles of the organizational logic of gene expression. RNA-binding proteins (RBPs) and non-coding RNAs are the regulatory molecules primary responsible for interaction with target mRNAs and thereby regulating post-transcriptional processes eventually influencing characteristics of the encoded protein. Many of the mRNA targets of RBPs encode functionally related proteins, which for post-trascriptional operons, resulting in coordination of macromolecular complexes or specific cellular processes. Thus, identifying RNA targets, precise binding sites, and the dynamics of these interactions will reveal how these important regulatory factors contribute to gene regulatory networks.

ELAV family of human RBPs consist of 4 members, which all have 3 RRM (RNA-recognition motif) domains the last separated by a hinge region. It predominant role is to positively regulate the stability and translation of target mRNAs through binding to ARE (AU-rich elements) in the 3' UTR (untranslated region) of protein coding transcripts. In response to certain stimuli, HuR is subject to post-translational modifications and changes subcellular localization, which impacts its regulatory capacity. In this study on a transcriptome-wide level, we interrogate the RNA targets, precise binding sites, as well as the remodeling of these interactions in response to stimuli.

We utilized two complementary methods, RIP-chip and PAR-CLIP, to identify targets of HuR and high-resolution binding sites on a transcriptome-wide scale. We discovered that HuR-mRNA interactions are not restricted to the 3' UTR and there are thousands of intronic binding sites. A significant proportion of intronic binding sites are contained in the poly-pyrimidine tract near 3' splice sites. Binding sites in the 3' UTR and intron are often approximately 30 nucleotides apart. HuR can bind to both AU-rich and U-rich sequences, the former more prevalent in 3' UTRs and the latter more prevalent at the 3' splice site.

Next we integrated the binding data with transcriptomics of HuR siRNA mediated knockdown. We found that the degree of binding is proportional to the degree of HuR-dependent stabilization. Moreover the ability to stabilize mRNA is not restricted to 3' UTR binding sites, as intronic binding sites also exhibited the binding degree correlated stabilization. We observed that the spatial pattern of HuR binding sites relative to exons influences exon usage decisions. Specifically, binding sites upstream of the exon promote exclusion, while binding sites downstream of the exon promote inclusion.

Much of the regulation of gene expression occurs at the posttranscriptional level, and much of this regulation is controlled and coordinated by RNA binding proteins (RBPs). Many RBPs have multiple mRNA targets, and the proteins encoded by these targets often share functional relationships, forming posttranscriptional RNA operons. These operons often reflect the function of the RBP, thus determination of the genome-wide targets of RBPs allows insight into their functions.

The PUF family of RBPs is characterized by the presence of an extremely well conserved RNA binding domain, typically consisting of 8 repeats of an RNA binding motif, with each repeat binding to one RNA base. PUF proteins are proposed to have an ancestral role in self-renewal of stem cells and have been shown to affect a number of developmental processes. Human and other vertebrate genomes contain two canonical PUF genes, Pum1 and Pum2, and at the outset of this study there was very little known about functions or targets of either protein, especially Pum1.

In order to identify the genome-wide targets of human Pum1 we used RNA immunoprecipitation followed by microarray, or RIP-Chip, analysis. RIP-Chip allowed us to identify Pum1 target mRNAs in human HeLa cells. We found that there were numerous functional relationships among the proteins encoded by these mRNAs, forming putative RNA operons. Some of these potential operons are progression of cell cycle, cell differentiation and proliferation, and regulation of transcription. We were also able to find a consensus Pum1 binding motif, UGUAHAUA, in the 3' UTRs of Pum1 target mRNAs.

The genome-wide targets of PUF proteins from other species have been previously identified, and by comparing the targets of human Pum1 to targets of Drosophila Pumilio and yeast Puf3, both of which bind to the same RNA sequence as Pum1, we determined that there has been evolutionary rewiring of regulation by Puf proteins. While the PUF RNA binding domain and consensus binding sequence have remained almost identical through evolution, the surrounding protein sequence and the mRNAs bound have changed dramatically, indicating that evolutionary rewiring is occurring in a modular fashion.

After identifying Pum1 associated mRNAs, we went on the study the function of Pum1. Through Pum1 knockdown assays we found that Pum1 enhances decay of target mRNAs, and that this effect is likely due to Pum1 enhancing deadenylation of these mRNAs. We also showed by immunofluorescence that Pum1 protein has a cytoplasmic granular subcellular localization and upon oxidative stress relocates to stress granules but not processing bodies. We were, however, unable to detect any difference in Pum1 mRNA targeting after stress. We were also unable to detect any changes in progression through cell cycle after Pum1 knockdown.

In this study we identified the genome-wide mRNAs associated with Pum1, determined functional relationships among these targets related to the proposed ancestral role of PUF proteins in self-renewal of stem cells, and identified a sequence motif to which Pum1 binds in these mRNAs. We also demonstrated that Pum1 enhances decay of associated mRNAs, and that this effect is likely due to Pum1 enhancing deadenylation of associated mRNAs. These results provide a description of mRNA targets and mechanisms of action of Pum1 proteins, which will provide a strong foundation for future experiments to further explore the functions of the Pum1, especially as they relate to human stem cells.

Posttranscriptional gene expression is controlled and coordinated by RNA binding proteins (RBPs), many of which recognize specific RNAs through cis-regulatory RNA elements. One of the most highly studied classes of cis-regulatory RNA elements is the AU-rich elements (AREs). AREs are bound by a class of RBPs called ARE binding proteins (ARE-BPs), of which there are over a dozen in humans including HuR, tristetraprolin (TTP) and butyrate response factors 1 and 2 (BRF1 and BRF2). TTP, BRF1 and BRF2 belong to a family of tandem C3H zinc finger proteins that destabilize ARE-containing mRNAs. HuR acts to enhance the stability and translation of ARE-containing mRNAs, a function that is rare among ARE-BPs. While each of these ARE-BPs regulates the expression of ARE-containing mRNAs, some ARE-BPs themselves are also encoded by ARE-containing mRNAs, raising the possibility that each of these ARE-BPs may regulate one another's expression. In order to determine how these ARE-BPs influence each others expression and how this affects the regulation of global gene expression programs we have focused on three different aspects of these ARE-BP networks: control, response to stimuli, and global effects.

To address of network control of ARE-BPs we have focused on how HuR regulates a network of mRNAs including TTP, BRF1 and HuR's own mRNA. We demonstrate that HuR can bind to TTP's, BRF1's and its own mRNA. Furthermore, by employing overexpression and siRNA knockdown approaches we demonstrate that these mRNAs and their corresponding 3'UTR luciferase reporters are resilient to fluctuations in HuR levels and that the degree of this resiliency is cell type and condition specific.

To address the temporal responses within an ARE-BP network we focused on how each of the members of the TTP family of ARE-BPs reacts following the induction of the other family members by using epidermal growth factor (EGF) stimulation. Here we show that induction of TTP family member mRNAs during EGF stimulation is partially attributable to changes in mRNA stability. Furthermore, we also show that TTP and BRF1 are able to bind each of the TTP family member mRNAs and subsequently affect their expression by altering their mRNA degradation rates. In addition, we demonstrate that the unique temporal induction patterns of the TTP family member RBPs is correlated with the EGF stimulated induction of TTP-bound mRNAs, suggesting that a network comprised of TTP family members is able to influence the timing of complex gene expression patterns.

Finally, to address the influence of these networks on regulation of global gene expression programs we have focused on how HuR recognizes AREs and whether it can globally recognize multiple classes of ARE-containing mRNAs, including the canonical class of AREs recognized by the TTP family members. To investigate how the three RNA recognition motifs (RRMs) of HuR contribute to ARE recognition we generated a series of RRM point mutants and test their ability to disrupt RNA recognition of each of the RRMs. To identify different classes of ARE-containing mRNAs we examined these mutants with a global RNA binding site detection method called photoactivatable ribonucleoside crosslinking immunoprecipitation (PAR-CLIP). Together these techniques suggest that the RRMs of HuR cooperate to recognize mRNA targets and that HuR's ability to bind RNA is coupled to the cellular distribution of HuR, and thus, are important in its role for regulating expression of bound mRNAs.

Together these studies indicate that ARE-BP posttranscriptional networks are highly interconnected and display complex regulatory interactions depending on cell type and stimuli. Furthermore, these networks can create complex behaviors such as timing of expression events or resiliency to fluctuations in protein levels. Finally, the components of these ARE-BP networks target partially overlapping sets of mRNAs to impact global gene expression patterns that ultimately coordinate the cellular responses to external stimuli.

RNA binding proteins (RBPs) and non-coding RNAs orchestrate gene expression in part through the recognition specific sites in mRNA. Thus understanding the connection between binding to specific sites and regulation of the whole transcript is essential. Current methods to do this can either identify the binding sites or quantitate binding to whole transcripts, but not both. Furthermore reliance of binding site detection on ultraviolet crosslinking results in inefficient identification of binding sites, and insufficient data to assess binding strength at sites. I have overcome these limitations by combining aspects of current methods to develop a new method called DO-RIP-seq (digestion optimization RNA immunoprecipitations with deep sequencing) that can quantitate the binding strength of RBPs at sites in mRNA, and also relate binding sites to binding of the whole mRNA. DO-RIP-seq was developed using the well-studied RBP ELAVL1/HuR as a test case, and applied to the less well-studied RBP known as RBM38/RNPC1. The quantitative data from DO-RIP-seq out-performed current binding site methods at predicting other features of the binding sites of HuR and RBM38, for example the lack of RNA secondary structure, and preferences in binding to particular sub-motifs. My studies indicate that DO-RIP-seq will be useful in uncovering the determinants of RNA-protein interactions, and studying dynamic biological processes that could modulate these interactions.