Argonaute proteins are the core component of an RNA silencing complex. The human genome encodes four Argonaute paralogs –Ago1, Ago2, Ago3 and Ago4– proteins that are guided to target mRNAs by microRNAs. More than 500 miRNAs are conserved between mammals, and each microRNA can repress hundreds of genes, regulating almost every cellular process. We still do not fully understand the molecular mechanisms by which miRNAs regulate gene expression. Although we understand many aspects of microRNA biogenesis and formation of the RNA-induced silencing complex, much less is known about the subsequent steps leading to target mRNA regulation. Mammalian microRNAs rarely have complete complementarity to their target mRNAs so, instead of endonucleolytic cleavage by Ago2, microRNAs destabilize or repress translation of target mRNAs. Here I explored the functional limits of Argonaute proteins bound to their targets directly and indirectly through microRNAs in mammalian cells. I revealed the different abilities for Argonaute proteins bound at multiple sites in a target to generate cooperativity in silencing based on the extent of pairing between the microRNA and target mRNA. Further, I harnessed the endogenous microRNA silencing mechanism to repress an mRNA that is not a direct target of the microRNA by tethering the RNA-induced silencing complex to the 3´ UTR of an mRNA. This strategy allows tissue-specific gene silencing due to the limited endogenous expression profile of the recruited microRNA. Efforts made herein further our mechanistic knowledge of microRNA-induced gene silencing in mammalian cells and advance microRNA-based strategies toward treating human disease.

Fibroblast growth factor (Fgf) and Retinoic acid (RA) are known to be involved in patterning the posterior embryo. Work has shown that Fgf can convert anterior tissue into posterior fates and that embryos deficient in Fgf signaling lack posterior trunk and tail structures. Likewise, studies performed on RA have shown that overexpression of RA posteriorizes anterior tissue, while disrupting RA signaling yields a loss of posterior fates. While it appears these signals are necessary for posterior development, the role Fgf and RA play in development of the hindbrain is still enigmatic. A detailed study of the requirements for Fgf and RA in the early vertebrate hindbrain are lacking, namely due to a deficiency in gene markers for the presumptive hindbrain at early developmental stages. In this study, we make use of recently isolated genes, which are expressed in the presumptive hindbrain region at early developmental stages, to explore Fgf and RA regulation of the early vertebrate hindprain. We employed both overexpression and loss of function approaches to explore the role of Fgf in early vertebrate development with an emphasis on the presumptive hindbrain region in zebrafish embryos. By loss of function analysis, we show that Fgf regulates genes expressed exclusively in the hindbrain region (meis3 and hoxbla) as well as genes whose expression domains encompass both the hindbrain and more caudal regions (nlz and hoxb1b), thus demonstrating a requirement for Fgf signaling throughout the anteroposterior axis of the hindbrain (rostral to caudal hindbrain) by mid-gastrula stages. To further characterize early gene regulation by Fgf, we utilized an in vitro system and found that Fgf is sufficient to induce nlz directly and hoxb1b indirectly, while it does not induce meis3 or hoxb1a. Furthermore, in vivo work demonstrates that Fgf soaked beads can induce nlz and hoxb1b adjacent to the bead and meis3at a distance. Given the regulation of these genes in vitro and in vivo by Fgf and their position along the rostrocaudal axis of the embryo, our results suggest an early acting Fgf resides in the caudal end of the embryo and signals at a distance to the hindbrain. We detect a similar regulation of hindbrain genes by RA at gastrula stages as well, suggesting that both factors are essential for early hindbrain development. Interestingly however, we find that the relationship between Fgf and RA is dynamic throughout development. Both signals are required at gastrula stages as disruption of either pathway alone disrupts hindbrain gene expression, but a simultaneous disruption of both pathways at later stages is required to disrupt the hindbrain. We suggest that Fgf and RA are present in limiting concentrations at gastrula stages, such that both factors are required for gene expression or that one factor is necessary for activation of the other. Our results also reveal a changing and dynamic relationship between Fgf and RA in the regulation of the zebrafish hindbrain, suggesting that at segmentation stages, Fgf and RA may no longer be limiting or that they are no longer interdependent. As we have demonstrated that an early Fgf signal is required for gastrula stage hindbrain development, we next questioned which Fgf performed this function. We have demonstrated that the early Fgf signal required for hindbrain development is not Fgf3 or Fgf8, two Fgfs known to be involved in signaling centers at the mid-hindbrain boundary (MHB) and rhombomere (r) 4. We further show that two recently identified Fgfs, Fgf4 and Fgf24 are also insufficient alone or in combination with other known Fgfs to regulate hindbrain gene expression. However, as Fgfs may act combinatorially, we do not rule out the possibility of their involvement in early hindbrain gene regulation. However, as time passes and additional Fgfs are isolated and cloned, the elusive Fgf signal required for early hindbrain development will likely be identified. Taken together, we propose that an early acting Fgf residing in the caudal end of the embryo regulates hindbrain genes together with RA at gastrula stages. We suggest that both Fgf and RA are required for gene expression at gastrula stages, but this requirements changes over time as Fgf and RA become redundant. We also demonstrate that the Fgf required for gastrula stage hindbrain development has yet to be identified.

Heavy chain isotype switch recombination is preceded by the appearance of RNA initiating 5' of the specific switch region which will undergo recombination. In an effort to understand the potential function of germline transcripts in switch recombination and the degree to which the regulation of germline transcripts correlates with the regulation of switching, we studied this process in the murine B-lymphoma cell line I.29μ, which in the presence of bacterial lipopolysaccharide (LPS) switches primarily to IgA and less frequently to IgE. Levels of α-germline transcripts initiating upstream of α switch (Sα) sequences are elevated in clones of this line which switch well as compared to clones which switch less frequently. TGFβ1 has been shown to increase α-germline transcripts and switching to IgA expression in LPS-stimulated murine splenic B-cells. We now demonstrate in I.29μ cells that TGFβ also increases switching to IgA and increases the level of α-germline transcripts 5 to 9 fold. Nuclear run-on analysis shows that this increase is at the level of transcription. Thus, TGFβ appears to direct switching to IgA by inducing transcription from the unrearranged Sα- CαDNA segment. Germline α RNA is quite stable in I.29μ cells, having a half life of about 3 to 5 hours, and we find only slight stabilization in the presence of TGFβ. Levels of ε-germline transcripts are not increased by TGFβ . IL-4, which modestly increases switching to IgA in I.29μ cells, slightly increases trancription of α-germline RNA. However, we present evidence suggesting that endogenously produced IL-4 may also act at additional levels to increase switching to IgA. IFNγ, which reduces IgA expression in these cells, also reduces the level of α-germline transcripts. IFNγ also reduces the level of ε-germline transcripts induced by IL-4. Our results support the hypothesis that the regulation of transcription of particular switch sequences by cytokines in turn regulates the specificity of recombination. In studies aimed at identifying other signalling pathways that promote class switching, we discovered that inhibitors of the nuclear enzyme poly(ADP-ribose) polymerase (PARP) increase lipopolysaccharide (LPS)-induced switching to IgA in the B cell lymphoma I.29μ and to IgG1 in LPS + IL-4-treated splenic B cells. PARP, which binds to and is activated by DNA strand breaks, catalyzes the removal of ADP-ribose from NAD+ and poly(ADP-ribosylation) of chromatin-associated acceptor proteins. This enzyme is believed to function in cellular processes involving DNA strand breaks as well as in modulating chromatin structure. In I.29μ cells, PARP inhibitors increase IgA switching by day 2 and cause a 5-fold average increase in switching on day 3 as assayed by immunofluorescence microscopy. The PARP inhibitor, nicotinamide, also causes a reduced intensity of hybridization of Cμ and Cα specific probes to genomic DNA fragments containing the expressed VDJ-Cμ and the unrearranged Sα - Cα segments, respectively, indicating that PARP inhibition increases rearrangment of these fragments. Induction of switching by PARP inhibitors is not mimicked by treatment with cAMP analogs or reduced by inhibitors of protein kinase A (PKA). Induction of switching by PARP inhibitors does not appear to involve increased levels of transcription of the unrearranged Cα gene, although TGFβ is required for optimal induction by PARP inhibitors, consistent with a requirement for transcription of the unrearranged CH gene. PARP inhibitors do not overcome the requirement for endogenously produced IL-4.

Transcription by RNA polymerase II is a highly regulated process requiring a number of general and promoter specific transcription factors. Although many of the factors involved in the transcription reaction are known, exactly how they function to stimulate or repress transcription is not well understood. Central to understanding gene regulation is understanding the mechanism by which promoter specific transcription activators (activators) stimulate transcription. A group of factors called coactivators have been shown to be required for activator function in vitro. The best characterized coactivators to date are members of the TFIID complex. TFIID is a multisubunit complex composed of the TATA box binding protein (TBP) and 8-12 TBP associated factors (TAFIIs). Results from numerous in vitro experiments indicate that TAFIIs function by binding to activators and forming a bridge between the activator and the basal transcription machinery. In order to gain insight into the mechanism by which activators stimulate transcription, we chose to analyze the in vivo function of TAFIIs, their proposed targets. Results from the genetic disruption of a number of TAFIIs in the yeast Saccharomyces cerevisiae showed that most are encoded by essential genes. In order to study their function, temperature-sensitive and conditional alleles were constructed. Cells depleted of individual TAFIIs by either of these two methods displayed no defect in global transcription activation. Inactivation of yTAFII17, however, resulted in a promoter specific defect. In addition, inactivation of yTAFII145, yTAFII90, or TSM1, resulted in an inability of cells to progress through the cell-cycle. In an attempt to identify genes whose expression required yTAFII90, we performed subtractive hybridization on strains containing wild-type and temperature-sensitive alleles. Although this technique successfully identified genes differentially expressed in the two strains, it failed to identify genes whose expression required yTAFII90. These results indicate that TAFIIs are not the obligatory targets of activators, and that other factors must provide this role in vivo. Furthermore, that many of TAFIIs are required for cell-cycle progression.

Coordination of mitosis and cytokinesis is crucial to generate healthy daughter cells with equal amounts of genetic and cytoplasmic materials. In the fission yeast Schizosaccharomyces pombe, an evolutionarily conserved Cdc14-like phosphatase (Clp1) functions to couple mitosis and cytokinesis by antagonizing CDK activity. The activity of Clp1 is thought to be regulated in part by its subcellular localization. It is sequestered in the nucleolus and the spindle pole body (SPB) during interphase. Upon mitotic entry, it is released into the cytoplasm and localized to the kinetochores, the actomyosin ring, and the mitotic spindle to carry out distinct functions. It is not clear how Clp1 is released from the nucleolus, however, once released, a conserved signaling pathway termed Septation Initiation Network (SIN) functions to retain Clp1 in the cytoplasm until completion of cytokinesis. The SIN and Clp1 function together in a positive feedback loop to promote each other’s activity. That is, the SIN promotes cytoplasmic retention of Clp1, and cytoplasmic Clp1 antagonizes CDK activity and reverses CDK inhibition on the SIN pathway to promote its function and activity. However, at the start of this thesis, the mechanism by which the SIN regulated Clp1 was unknown. The SIN pathway is also required to promote constriction of the actomyosin ring, and the septum formation. However, its downstream targets were still uncharacterized. In two separate studies, we studied how Clp1 is released from the nucleolus at mitotic entry and how the SIN kinase Sid2 acts to retain Clp1 in the cytoplasm. We identified several Sid2 candidate substrates, and revealed other functions of the SIN pathway in coordinating mitotic events.

The Smad proteins transduce the signal of transforming growth factor-β (TGF-β) and related factors from the cell surface to the nucleus. Following C-terminal phosphorylation by a corresponding receptor kinase, the R-Smad proteins form heteromeric complexes with Smad4. These complexes translocate into the nucleus, bind specific transcriptional activators and DNA, ultimately modulating gene expression. Though studied through a variety of means, the stoichiometry of the R-Smad/Smad4 complex is unclear. We investigated the stoichiometry of the phosphorylation-induced R-Smad/Smad4 complex by using acidic amino acid substitutions to simulate phosphorylation. Size exclusion chromatography, analytical ultracentrifugation, and isothermal titration calorimetry analysis revealed that the R-Smad/Smad4 complex is a heterotrimer consisting of two R-Smad subunits and one Smad4 subunit. In addition, a specific mechanism for phosphorylation-induced R-Smad/Smad4 complex formation was studied. Although it had been previously established that part of the mechanism through which phosphorylation induces Smad oligomerization is through relieving MH1-domain mediated autoinhibition of the MH2 (oligomerization) domain, it is also evident that phosphorylation serves to energetically drive Smad complex formation. Through mutational and size exclusion chromatography analysis, we established that phosphorylation induces oligomerization of the Smads by creating an electrostatic interaction between the phosphorylated C-terminal tail of one R-Smad subunit in a Smad trimer with a basic surface on an adjacent R-Smad or Smad4 subunit. The basic surface is defined largely by the L3 loop, a region that had previously been implicated in R-Smad interaction with the receptor kinase. Furthermore, the Smad MH2 domain shares a similar protein fold with the phosphoserine and phosphothreonine-binding FHA domains from proteins like Rad53 and Chk2. Taken together, these results suggest that the Smad MH2 domain may be a distinct phospho serine-binding domain, which utilizes a common basic surface to bind the receptor kinase and other Smads, and takes advantage of phosphorylation-induced allosteric changes dissociate from the receptor kinase and oligomerize with other Smads. Finally, the structural basis for the preferential formation of the R-Smad/Smad4 heterotrimeric complex over the R-Smad homotrimeric complex was explored through X-ray crystallography and isothermal titration calorimetry. Crystal structures of the Smad2/Smad4 and Smad3/Smad4 complexes revealed that specific residue differences in Smad4 compared to R-Smads resulted in highly favorable electrostatic interactions that explain the preference for the interaction with Smad4.

The immune response to viral infection involves a complexity of both innate and adaptive pathways at the cellular and the molecular level. There are many approaches to begin to define the pathways at work to control viral pathogenesis. The approach favored in this thesis was to conduct a broad screen of the innate immune response at the gene expression level of infected cells. The innate immune response is critical to the control of viral infections. Type I interferons (IFN), IFNα and IFNβ, are antiviral proteins that are an integral part of the innate immune response. Furthermore, by virtue of their effects on maturation and activation of antigen-presenting cells, IFNs are a pivotal link between the innate and adaptive immune systems. Most cell types produce type-I IFN when exposed to viruses. However, viruses have evolved multiple strategies to suppress IFN production or signaling. It is imperative to understand the virus-host interaction at the molecular level in order to identify as yet unknown mechanisms of the host antiviral response; these additional pathways may be useful in counteracting the viral suppression of IFN. Type-I IFNs regulate expression of at least five hundred genes, suggesting a complex network of signaling pathways. Depending on the cell type different proteins regulate the induction of IFN or the expression of IFN-inducible genes. Identification of proteins that induce selected IFN-inducible genes may provide synergistic activity with or may have an advantage over type-I IFN for anti-viral therapy in the future. Many diseases are untreatable if identified late in their progression. In resource-limited countries, many diseases are diagnosed clinically, which can lead to incorrect or delayed diagnosis and treatment. The identification of biomarkers of disease has the potential to guide the correct therapy in a timely fashion. The objective of this thesis was to identify novel anti-viral therapies and disease biomarkers for dengue virus (DENV) infection. DENV is a mosquito-borne positive-sense single-stranded RNA virus, which causes an estimated 50 million infections annually. Most DENV infections result in a febrile illness called Dengue fever (DF). Less frequently, infections cause Dengue hemorrhagic fever (DHF), a potentially fatal vascular leakage syndrome associated with the production of pro-inflammatory cytokines. At present patients infected with DENV can only be treated by intravenous fluid support to prevent hypovolemia and hypotensive shock. This treatment is less effective in severe cases if the diagnosis is delayed. Identification of therapeutics with both antiviral and immune-modulatory activity may lower patient mortality and reduce the burden of DENV on society. DENV infection is cleared in most individuals after a short period of viremia {Libraty, 2002 #2225}. Based on in vitro and mouse models, type-I and type-II IFN signaling pathways are thought to be critical in the regulation of DENV infection. Higher serum levels of type I and type II IFNs during acute DENV infection in patients lend support to the above hypothesis {Kurane, 1993 #2152; Libraty, 2002 #2225}. To understand the DENV-human host cell interaction at the molecular level, we performed global gene expression analysis on DENV-infected primary human cells using Affymetrix GeneChips (HG-U133A). We studied dendritic cells (DC), monocytes, B cells and human umbilical vein endothelial cells (HUVECs), all of which are known to be permissive to DENV infection. We first identified genes commonly regulated in multiple cell types in response to DENV infection; we hypothesized that understanding this common gene expression profile would identify signaling pathways involved in regulation of viral spread, activation of immune cells or induction of inflammation. Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), one of the 23 common response genes, was identified as a key link between type I and type II interferon response genes. Pretreatment of cells with recombinant TRAIL (rTRAIL) inhibited DENV replication in monocytes, B cells, HUVECs and DCs. Using the DC infection model, we showed that this inhibition of viral replication was apoptosis-independent. Type-I IFN receptor (IFNR) blocking experiments showed that signaling through the type-I IFN receptor played an important role in the antiviral activity of exogenous rTRAIL. Furthermore, TRAIL also significantly reduced the expression of mRNA and protein of pro-inflammatory cytokines (TNFα, MIP-1β and IFNα) and chemokines (MCP-2, IP-10 and IL-6) in response to DENV infection. The data that TRAIL inhibits both viral replication and pro-inflammatory cytokine production suggest that TRAIL has therapeutic value in dengue. The endothelial cell is the site of pathology in DENV infection in vivo (vascular permeability and plasma leakage). To understand the direct effect of DENV infection on endothelial cells and its role in the induction of genes regulating vascular permeability, we compared gene expression in DENV-infected HUVECs to that of uninfected cells and cells infected with other RNA and DNA viruses, including flaviviruses (West Nile, yellow fever, and Japanese encephalitis viruses), bunyaviruses (Sin Nombre and Hantaan viruses), Epstein-Barr virus and vaccinia virus. Among the genes confirmed for their differential expression, ST2 (Interkeukin-1 receptor-like-1 protein-IL1RL1) and indoleamine 2,3-dioxygenase (IDO) were identified to be upregulated specifically in response to DENV infection. Higher serum soluble ST2 (sST2) levels were detected in DENV-infected patients than in patients with other febrile illnesses (OFI) at the end of the febrile stage and at defervescence (p=0.0088 and p=0.0004, respectively). In addition, patients with secondary DENV infections had higher serum sST2 levels compared with patients with primary DENV infections (p=0.047 at the last day of fever and p=0.030 at defervescence). Higher levels of IDO activity (pIn conclusion, global gene expression analysis identified novel proteins with promising characteristics for the treatment and/or diagnosis of DENV infection. Although further studies will be needed to validate the clinical utility of TRAIL, sST2, and IDO, these studies demonstrate the utility of this unbiased genomics approach to identify therapies to currently incurable diseases.

The study of DNA templated events is not complete without considering the chromatin environment. Histone modifications help to regulate gene expression, chromatin compaction and DNA replication. Because DNA damage repair must occur within the context of chromatin, many remodeling enzymes and histone modifications work in concert to enable access to the DNA and aid in restoration of chromatin after repair is complete. CK2 has recently been identified as a histone modifying enzyme. In this study we identify CK2 as a histone H3 tail kinase in vitro, identify the phospho-acceptor site in vitro, and characterize the modification in vivo in S. cerevisiae. We also characterize the DNA damage phenotype of a strain lacking a single catalytic subunit of CK2. We further characterize the CK2- dependent phosphorylation of serine 1 of histone H4 in vivo. We find that it is recruited directly to the site of a DSB and this recruitment requires the SIN3/RPD3 histone deacetylase complex. We also characterize the contribution of H4 serine 1 phosphorylation in chromatin compaction by using reconstituted nucleosomal arrays to study folding in the analytical ultracentrifuge.

Long-range chromosomal interactions have been discovered in a number of organisms, suggesting that gene regulation through direct physical association with regulatory elements and/or other genes is a common and conserved phenomenon. This thesis investigates the relationship between direct physical contact of genomic loci and how these interactions may play a role in gene regulation. Analysis of such levels of chromosomal organization has been made possible in part by the emergence of Chromosome Conformation Capture (3C). This technique makes use of formaldehyde crosslinking to trap interacting chromosomal fragments, which can be detected after a number of manipulations. By adapting the 3C protocol for use in intact Saccharomyces cerevisiaecells, one can determine the structure of a chromosome or chromosomal region without introducing artifacts due to the harsh isolation of nuclei. A number of 3C-based technologies, such as 4C (Circular 3C or 3C-on-Chip) and 5C (3C Carbon Copy) have added to the knowledge of physical association of genes with regulatory elements and/or other genes. Here, we present a new non-biased technology that allows for determination of chromosomal interactions between all fragments throughout a genome. We present two-dimensional heatmaps of chromosomal interactions for all 16 chromosomes in yeast. These techniques promise to shed light onto the biochemical process by which clustering of genes and elements can result in up- or down-gene expression, which is still poorly understood. To understand how chromosomal interactions play a role in gene regulation, we study clustering of heterochromatic loci. Clustering of heterochromatic loci in silenced nuclear compartments is a phenomenon that has been observed throughout evolution. These clusters are thought to represent nuclear sub-compartments that are enriched in silencing proteins, while the rest of the nucleus is depleted in such factors. Chromosome III in Saccharomyces cerevisiae contains four heterochromatic regions: the two telomeres and the silent mating type loci, HML and HMR, located on either end of the chromosome. Our work describes a long-range interaction between the heterochromatic regions on chromosome III. We analyze the mechanism that drive these interactions and reveal roles for silencing proteins and proper nucleosome assembly in mediating heterochromatic clustering. In addition we identify a novel step in heterochromatin formation that is not essential for gene silencing but is required for long-range interactions.

Chronic myeloid leukemia (CML) is a disease characterized by the expansion of granulocytic cells. The BCR-ABL tyrosine kinase inhibitor imatinib, the frontline treatment for Ph+ leukemias, can induce complete hematologic and cytogenetic response in most chronic phase CML patients. Despite the remarkable initial clinic effects, it is now recognized that imatinib will unlikely cure patients because a small cell population containing leukemic stem cells (LSCs) with self-renewal capacity is insensitive to tyrosine kinase inhibitors. In Chapter I, I briefly review the BCR-ABL kinase and its related signaling pathways. BCR-ABL kinase activates several signaling pathways including MAPK, STAT, and JNK/SAPK. BCR-ABL also mediates kinase-independent pathways through SRC family kinases. I will also discuss pathways involving β-catenin, hedgehog, FoxO and Alox5 are critical to the regulation of self-renewal and differentiation in LSC of CML. As detailed in Chapter II, I describe our work evaluating the effects of omacetaxine, a novel CML drug inducing cell apoptosis by inhibition of protein synthesis, on self-renewal and differentiation of LSCs and BCR-ABL-induced CML and acute lymphoblastic leukemia (B-ALL) in mice. We found that treatment with omacetaxine decreased the number of LSCs and prolonged the survival of mice with CML or B-ALL. In chapter III, I describe that Alox5 is an essential gene in the function of LSCs and CML development. We show evidence that Alox5 affects differentiation, cell division, and survival of long-term LSCs. Treatment of CML mice with a 5-LO inhibitor also impaired the function of LSCs similarly and prolonged survival. In chapter IV, I present evidence of our work showing a further dissection the Alox5 pathway by comparing the gene expression profiles of wild type and Alox5-/- LSCs. We show that Msr1 deletion causes acceleration of CML development. We also show that Msr1 affects CML development by regulating the PI3K-AKT pathway and β-catenin. Taken together, these results demonstrate that some pathways including Alox5 and Msr1 play an important role in regulating the self-renewal and differentiation of LSC. More efforts should be put into developing the novel strategies that may effectively target LSCs and thus cure CML.