Recent advances in genomic technologies have enabled comprehensive profiling of these alterations and allow us not only to elucidate the molecular mechanism underlying carcinogenesis but also to identify useful diagnostic markers for translational medicine.

To identify aberrant methylation events in cancer, we used methylated DNA immunoprecipitation coupled with either genomic tiling array or high throughput short read sequencing. Aberrant hypermethylation was highly observed in polycomb-targeted genes in embryonic stem cells. Methylation markers specific for cancer were further validated by mass spectrometry analysis and used to classify colon cancer based on their epigenotypes.

We also used 28K illumina epigenotyping beadarray to analyze DNA methylation status in cancer and normal tissues. In liver cancer, high CpG regions were hypermethylated, while low CpG regions were hypomethylated. Finally, integration of methylome information with genetic alteration and chromosomal alteration will be discussed.
Hide

To test this, we developed a novel technology called Chromatin Interaction Analysis-Paired End diTag (ChIA-PET) for the global, high-throughput detection of long-range chromosome interactions mediated by any given transcription factor. In this approach, we combined ChIP, Chromosome Conformation Capture (3C) and PET together with massive parallel sequencing to identify chromosome interactions on a global-scale. In my talk, I will highlight how we have applied this technique to uncover hundreds of novel long-range chromosome interactions mediated by ERα in breast cancer cells. In addition, I will also present some of our recent work on the mechanism of long-range transcriptional regulation by ERα.
Hide

Comprehensive analysis of this genomic network through large-scale chromatin interaction mapping will provide important information about the spatial organization of chromosomal domains, the relationships between genomic elements, and will reveal the logic of long-range gene regulation. Recently we developed the Chromosome Conformation Capture - Carbon Copy (5C) technology that combines 3C with deep-sequencing for high-throughput mapping of interactions between genomic elements (Dostie et al. Genome Res. 2006). Using 5C we have generated comprehensive interaction maps of Mb-sized domains in the human genome. We will present results that indicate widespread long-range interactions among genes, enhancers and insulator/boundary elements, providing new insights into the nature and potential functional role of these associations. The unique power of 5C as compared to other high-throughput 3C based methodologies is that exceptionally dense interactions matrices can be generated for large chromosomal domains. The comprehensive nature of these interaction maps allows the determination of the overall spatial conformation of chromatin at unprecedented resolution. We will present a novel approach to convert 5C interaction maps into 3D models.

Supported by NIH (HG003143) and the Keck Foundation.
Hide

Indeed, each cell type is the beneficiary of more than thousands of committed modifications on the epigenome, which instruct critical nuclear functions. Although we are beginning to understand the flow of epigenetic information that govern gene activity patterns, we still do not know the regulatory machineries that segregate permissive from suppressive chromatin landscapes. Key experimental data in our laboratory provide evidence that hyperglycemic memory and cardiovascular injury are mediated by specific epigenetic events that not only remodel chromatin, but also, drive endothelial cell dysfunction in the aorta. The “thousand natural shocks” resemble the specific epigenetic marks that regulate gene expression and chromatin behaviour, whereas, the increased risk of atherosclerosis in diabetes is akin to “flesh is heir to” which bear resemblance to the physiology of hyperglycemic memory and the persistence of epigenetic marks. This talk will examine and discuss the importance of histone modifications mediated by methyl-writing and methyl-erasing enzymes in models of hyperglycemic variability.
Hide

One of the next challenging questions in biological research is to understand how this epigenetic information is then transduced by chromatin-associated proteins to downstream effects. The project aims at identifying the network of interaction between protein-motifs that are commonly found among chromatin-associated proteins and methyl-histone modifications (HMs).

The experimental strategy involved the cloning of a library of protein domains into a vector, which allows in vitro transcription and translation. The library includes the entire set of human protein domains potentially associated with modified histone tails: Chromo-, Tudor-, MBT-, Plant Agenet-, PWWP-, all belonging to the Tudor domain "Royal family", as well as BRK domains and selected PHD fingers. The second step of the project consisted in hybridizing the library onto an array of Histone H3 and H4 modified peptides. This has resulted in the identification of new histone binders, like the chromodomain of CDYL1 and CDYL2 that recognize the trimethylation of lysine 9 of histone H3 (H3K9me3), and the WD40 domain of WDR5 that binds to the symmetric dimethylation of histone H3 (H3R2me2s).

We have further validated the newly discovered interactions by alternative methods such as standard peptide immunoprecipitations from bacterial lysates, immunofluorescence, BIAcore and crystalization. The project, which is still ongoing, as allowed so far to increase the information available on how proteins "read" the histone epigenetic information and to further investigate the function of some of these interactions in downstream signaling events.
Hide

PRC2 involvement has been shown for colon cancer, but it is not known if these findings are applicable to other cancer types. We used a SEQUENOM’s EpiTYPER method to measure DNA methylation profiles in more than 400 cancer related genes across 59 cancer cell lines derived from 9 different types of cancer. We validated our cancer cell lines findings in a set of colon cancer and adjacent normal tissue samples. We confirmed > 85% of genes with altered methylation in the cell lines to be differentially methylated in the clinical specimen. The set of examined genes included 70 that are known to be targets for PRC2. We were able to identify PRC2 target methylation in multiple forms of cancer. In six out of seven tumor types we find the fraction of hypermethylated genes to be enriched for those with PRC2 binding sites in the promoter region. These results suggest that PRC2 target silencing is a common event in cancer. The only tumor type with no such enrichment is melanoma, a cancer type with large environmental component that retains a high degree of differentiation. Our methylation data provides an ideal starting point for genome wide methylation research to identify all differentially methylated genes. The results can be combined with existing large scale datasets to develop an approach that integrates epigenetic, transcriptional and mutational findings. This study demonstrates the effectiveness of candidate gene approaches in the discovery of novel epigenetic biomarkers. We will introduce the concept of the underlying EpiTYPER technology, and give you an outlook to SQNMs current development efforts.
Hide

Indeed dysregulated epigenetic modifications, especially in early neoplastic development, may be just as significant as genetic mutations in driving cancer development and growth. The reversal of aberrant epigenetic changes has therefore emerged as a potential strategy for the treatment of cancer. A number of compounds targeting enzymes that regulate histone acetylation, histone methylation and DNA methylation have been developed as epigenetic therapies, with some demonstrating efficacy in hematological malignancies and solid tumors. Histone deacetylase inhibitors (HDACi) are emerging as important and efficacious anti-cancer agents and although their mechanisms of action remain to be fully defined, these agents can clearly induce tumor cell death (apoptosis) and inhibit tumor cell proliferation. We have utilised genetically engineered mouse models of cancer to identify the genes and molecular pathways necessary for HDACi-mediated apoptosis and subsequent therapeutic efficacy. Using this information, we have rationally designed combination therapy regimens that demonstrate superior anti-tumor activity compared to single agents.
Hide

The methylation of cytosine residues in these regions is associated with alterations in chromatin structure including the binding of methylated DNA binding proteins and changes in the state of modification of histone residues in nucleosomes. These alterations serve to reinforce each other and may lead to the heritable silencing of genes which can have profound implications for human cancer development. Unlike mutational changes, epigenetic alterations are acquired in a gradual process which is associated with cellular division. Thus, these progressive alterations are potentially susceptible to interventions to reverse silencing. Epigenetic changes can be observed in premalignant tissues so that understanding what causes the alterations and development of potential strategies to reverse them could have an impact on carcinogenesis. The mechanisms underlying progressive methylation of CpG islands leading to altered chromatin configuration are now beginning to be understood. Drugs such as 5-aza-2’-deoxycytidine can reverse DNA methylation changes and reactivate gene expression by changing not only the methylation, but also the nucleosomal occupancy of the promoter. Further understanding of epigenetic changes in cancer and the mechanisms by which they are acquired may therefore help in the search for new therapeutics.
Hide

To test whether chromatin modification is also involved in osteoblast differentiation, we examined the levels of histone deacetylases (HDACs) and the degree of histone acetylation at the promoter regions of osteogenic genes. During osteogenesis, total HDAC enzymatic activity was decreased, concomitantly with decreased HDAC1 expression. Consistently, recruitment of HDAC1 to the promoters of osteoblast marker genes such as osterix and osteocalcin was downregulated. In contrast, the level of histone H3 and H4 aceylation was increased during osteoblast differentiation and hyperacetylated histone H3 and H4 were monitored at the promoter of osteogenic marker genes. Furthermore, osteogenesis-inducing molecules augmented the recruitment of Runx2, a key osteogenic transcription factor, onto the promoter of osteogenic genes. For instance, berberine, an isoquinoline chemical with osteognic activity, promoted the binding of Runx2 and p300 to the promoters of osteopontin and osteocalcin genes. Simultaneously, berberine repressed the association of HDAC1 at those genes, resulting in the stimulation of expression of osteogenic genes. Adiponectin also elevated the binding of Runx2 and p300 to the promoters of osteogenic genes. Taken together, these results suggest that the epigenetic regulation with osteogenic transcription factor and its coregulator upon signaling cues appears to be crucial to coordinate proper expression of osteogenic genes during osteogenesis.
Hide

Here we show that expression variability is an intrinsic property that persists at those different levels. Each promoter seems to have a unique capacity to respond to external signals that can be environmental, genetic or even stochastic. Our investigation into nucleosome organization of variably responding promoters revealed a commonly positioned nucleosome at a critical regulatory region where most transcription start sites and TATA elements are located, a deviation from typical nucleosome-free status. The nucleotide sequences in this region of variable promoters showed a high propensity for DNA bending and a periodic distribution of particular dinucleotides, encoding preferences for DNA–nucleosome interaction. Variable expression is likely to occur during removal of this nucleosome for gene activation. This is a unique example of how promoter sequences intrinsically encode regulatory flexibility, which is vital for biological processes such as adaptation, development and evolution.
Hide

Recently, the existence of tumor-propagating cells (TPCs) has been demonstrated in human neoplasia including brain tumor-propagating cells (BTPCs). Based on the TPC scenario during tumorigenesis, the heterogeneity of GBM may partially be explained by the existence of stem-cell like tumor cells. To understand epigenetic machinery regulating BTPC differentiation in GBM, we examined the global targets of polycomb-repressive complex-mediated histone H3 lysine (K) 27 tri-methylation (H3K27me3), H3-K4 methylation (H3K4me2) and DNA methylation, which are known to be transcriptional regulating machineries for tumor-associated genes. Continuous cultures of BTPCs in serum conditions resulted in profound morphological changes, adherent and spindle shape. Immunohistochemistry and RT-PCR analyses supported that the morphological changes were partially compatible with cell differentiation (differentiated tumor cells, DTCs). Analysis of H3K27me3 target genes with chromatin immunoprecipitation coupled with microarray (ChIP-chip) in BTPCs and DTC revealed that a number of genes gain H3K27me3 during differentiation. In addition, a subset of genes was commonly enriched with H3K27me3 in both BTPCs and DTC, but was not in neural stem cells, suggesting that these genes were required to perpetuate the malignant phenotype. By contrast, very few DNA methylation changes were observed during differentiation. Consistent with the H3K27me3 status, down-regulation of EZH2 histone methyltransferase inhibited the growth and morphological changes of BTPC. These results established the polycomb mediated H3K27me3 modification as a key epigenetic mechanism for TPC-based tumorigenesis and could be a valid therapeutic target in human neoplasia.
Hide

The technologies used were ChIP-seq, detailed expression array analysis, ChIP-chip of a custom ER binding site array, and comparative genomic analysis. Of the ~16,000 estrogen receptor α binding sites (ERABS) in non-amplified regions of the MCF7 cell line, ERα binding was found distributed up to 100 kb upstream and downstream of genes, and within intragenic regions with only 12% in gene desert regions. We analyzed the regions of open chromatin as defined by Formaldehyde-Assisted Isolation of Regulatory Elements and genome-wide occupancy of six histone methylation/acetylation marks (H3K4Me1, H3K4Me3, H3K9Ac, H3K14Ac, H3K9Me3 and H3K27Me3), and of RNA polII in the MCF-7 cells. By overlapping these marks with ERα binding site (BS) map, we aimed to define rules of ERABS usage and downstream transcriptional regulatory control of ERα. When positional bias relative to gene boundaries was removed, we found that ERα binding was characterized by chromatin configuration, and association with specific histone marks. These characteristic marks correlated well with ERα binding intensity. Using, the sequence tag counts as a measure of ERα occupancy, we calculated an extended position weighted matrix for ERα binding and developed an energetically based score for ER binding based on match to this optimized motif (COSMIC). We found that the great majority of the binding sites have a discernable motif, either a full or half site or a composite mix. Combinatorial analysis suggested that the presence of a good ERE was sufficient to drive ERα binding, but that in sites with less optimal ERE, histone modification had a greater role in determining binding intensity. Though no specific characteristic could be discerned between binding sites in the vicinity of either up or down regulated genes, we observed that regulated genes showed enrichment of RNA polII marks at the TSS (start) and TES (stop) of every gene. Using a sequencing strategy to identify all binding site interactions called Chromatin Interaction Analysis using Paired End diTag Sequencing (ChIA-PET), and discovered that most binding sites are involved in distance interactions.
Hide

A comprehensive genomewide catalog of epigenetic control elements and how these vary across cell states could offer critical insight into the relationships between genotype, phenotype and environment, and serve as a catalyst for future studies of the epigenetic mechanisms that regulate normal physiology and human disease. Our ability to characterize mammalian epigenomes has been markedly enhanced by technological developments in recent years. In particular, the introduction of ultra high-throughput sequencing has improved the precision, comprehensiveness and throughput of techniques for mapping chromatin and DNA methylation. I will focus on these new applications and their promise for high resolution interrogation of mammalian epigenomes.
Hide

Two striking examples are testicular cancer and autism. In both cases heritability is high but the explained genetic variance is low. This ‘missing heritability’ has also been found in many other genetic diseases. The usual explanation involves undetected genetic variants with weak and heterogeneous actions in affected individuals. Although this explanation is reasonable given the limited statistical power in most genetic studies, high heritability under these conditions is less readily explained.

We recently discovered three examples of transgenerational gene interactions (epistasis). In each case, we found pairs of interacting genes that act in different generations, with one of the variants acting in affected individuals and the other in the parents. Individually, these variants have modest phenotypic effects, but their interaction across generation leads to much more dramatic phenotypic outcomes. Remarkably in these cases, the genotype of the parents is a better predictor of phenotypic outcome than genotype of the affected individuals. The interacting genes encode RNAs and proteins that are involved cell signaling, stress response, RNA editing, and miRNA biology. Presumably epigenetic or cytoplasmic inheritance is accounts for these transgenerational effects.

We are conducting detailed genetic and phenotypic characterization of these three examples, we are testing hypotheses about the molecular mechanism for transgenerational interactions, and finally we are testing whether this novel mode of inheritance accounts for missing heritability for several traits and diseases in humans.
Hide

As a result, developing germ cells are thought to revert their epigenetic settings to enable totipotency after conception. Genomes are also subjected to changes in their chromatin states during early pre-implantation development. The potential of reprogramming in early embryos was illustrated by nuclear transfer experiments, where the epigenome of a fully differentiated somatic cell could be reprogrammed by the cytoplasm of an enucleated oocyte. The success rate of reproductive cloning is, however, low compared to natural reproduction. This suggests that resetting of epigenetic modifications on chromatin during germ cell development provides parental genomes the competence to effectively support changes in cell identity during early embryogenesis.

Recently we identified transgenerational transmission of methylated histones as a novel form of maternal contribution. Specifically, we showed that the transmission of H3K9me3, established in oocytes by the Suv39h HMTs, is required for maintaining the canonical constitutive heterochromatic state at major satellite repeats of the maternal genome in mouse early embryos. In the absence of this germ-line signal, as observed at the paternal genome, an alternative repressive chromatin state is formed by Polycomb group proteins at major satellites in early embryos. These data underscore the importance of transgenerational transmission of histone methylation via the maternal lineage. This study further revealed a mechanistic hierarchy between the H3K9 methylation and Polycomb repressive pathways.

In contrast to oogenesis, the majority of nucleosomes are replaced by protamines during spermatogenesis. It is currently unclear whether the remaining histones (and/or their possible modifications) could function in paternal transmission of epigenetic information. Ongoing work in the laboratory shows that modified histones are present in human and mouse spermatozoa. I will discuss our recent genome wide profiling data of various histone modifications in human spermatozoa in light of inheritance of epigenetic information across generations.

Hide

While each class of cis-regulatory elements may contribute to cell-type dependent gene expression, previous studies have mainly focused on the role of promoters as a driving force behind tissue-specific and differential expression, partly due to a scarcity in our knowledge of the long range regulatory elements. In order to better characterize the mechanisms of cell-type specific gene expression, we have performed experiments to localize the genomic binding sites of general transcription factors, active chromatin modifications, and insulator binding protein CTCF in the human genome in multiple cell types. We find that transcriptional promoters and enhancers are associated with distinct chromatin signatures. Further, we show that the chromatin signatures at promoters, as well as the localization pattern of the insulator-binding protein CTCF, remains largely invariant across cell types. By contrast, the majority of enhancers are marked by specific histone modifications in a cell-type dependent manner. We observe that cell-type specific gene expression correlates with not only changes in chromatin marks at promoters, but also changes at enhancers, and that the effects of multiple enhancers toward the adjacent genes tend to be synergistic. These results show at a large scale that enhancers play an important role in cell-type dependent gene expression, and highlight the necessity to identify these sequences for understanding mechanisms of cell-type specific gene expression.
Hide

Our aim is to understand the molecular mechanisms, of which little is understood, of how the totipotent blastomeres of the 8-cell embryo give rise to these three cell-types. By utilizing TaqMan real-time PCR combined with high-throughput microfluidic devices we have screened the expression dynamics of >800 genes (predominantly transcription factors) through preimplantation development at the whole embryo level. Subsequently, to best define molecular differences between the three cell-types, we have screened through 120 genes from single cells of the 32-celled blastocyst. Principal component analysis on the expression of these 120 genes from 80 single cells molecularly defines three distinct cell populations within the first two principal components. Next, we have selected the most informative 45 genes (& three housekeeping genes) in defining these populations at the 32-cell stage (including 29 transcription factors) for single cell analysis at earlier developmental time points. The analysis of the expression of this set of genes, in more than 450 individual cells, from the 1-cell through to the 64-cell stage of preimplantation development provides insight into the molecular network and hierarchical expression of genes regulating lineage fate, commitment, and segregation in the formation of the blastocyst.
Hide

Thus, there is a continuing and urgent need to identify the underlying cause of disease processes and to target these with novel medicines. That is easier said than done. With the array of technologies available and the state of understanding of biology as advanced as it now is, novel biological processes such as epigenetic regulation, are being discovered constantly; dysregulation of these processes can contribute to disease development and therefore hold considerable promise for therapeutic intervention. However, discovering and developing new medicines against these novel target classes or broad new processes that are 'unprecedented' from the perspective of drug development places additional hurdles for the drug hunter in a process that is very complex, even for those target classes (e.g., kinases, GPCRs, growth factors) for which many small molecule as well as biological drugs exist. In this presentation, typical approaches for the development of new drugs will be reviewed, and in parallel, the discussion will focus to some of the potential challenges, as well as possible solutions, that may be encountered during the development of new medicines in an unprecedented target class. These concepts will set the stage for the following speakers in this session, who will present work related to the preclinical and clinical development of compounds affecting three different types of epigenetic modifications – histone acetylation, histone methylation, and DNA methylation.
Hide

However, little is known regarding its function in early hematopoietic development. By inactivating the SWI/SNF complex in a hematopoietic system, here we found that the complex is an essential transcriptional regulator in early B cell development. B lymphoid specification and commitment was associated with upregulation of key subunits of the complex. Mice with either stable knockdown of BRG1 or SRG3, an essential ATPase or scaffold subunit, respectively, or constitutive expression of BAF57ΔN, a dominant-negative mutant, showed that the complex was required for early B lymphopoiesis, but largely dispensable for myelopoiesis. B cell development in these mice was arrested at the early pro-B cell stage by defects in B lineage-specific gene expression regardless of cell death. Our studies suggest that the SWI/SNF complex has a pivotal function in early B lymphopoiesis by epigenetically modulating gene regulatory networks.
Hide

To this end, Sangamo Biosciences and Sigma-Aldrich have recently partnered to commercialize a novel technology that enables high-frequency genome editing via the application of designed zinc finger nucleases (ZFNs). Within these chimeric proteins the DNA binding specificity of the zinc finger protein determines the site of nuclease action. Such engineered ZFNs are able to recognize and bind to a specified locus and evoke a double-strand break (DSB) in the targeted DNA with high efficiency and base-pair precision. The cell then employs the natural DNA repair processes of either homology-directed repair (HDR) or non-homologous end joining (NHEJ) to heal the targeted break. These two pathways provide the investigator with the ability to provoke three unique outcomes in genome editing – gene correction, gene deletion and targeted gene addition. Furthermore, the speed and efficiency of this process enables us to knockout multiple genes in the same cell. Drawing from our work with transformed cell lines, primary human cells, and multi-potent stem cells, we will present examples of single, double and triple gene knockout, as well as targeted gene insertion into native chromosomal loci.
Hide

However, the molecular mechanisms governing the stemness of tumor-initiating cells remain obscure, and present methods that select cancer stem cells are not efficient. Here we developed a model to selectively enrich breast tumor-initiating cells by consecutive passaging SKBR3 cells in imunocompromised NOD/SCID mice under the pressure of low-dose chemotherapy, and analyzed the expression profile as well as function of microRNAs (miRNAs) in breast cancer stem cells. Under weekly dosing of Epirubicin, the resulted cells from the 3rd generation of tumor xenografts (SK-3rd cells) displayed higher capacity of mamosphere formation in suspension cultures and phenotypes of breast cancer stem cells. MiRNA microarray, together with northern blot and quantitative PCR, revealed a marked reduction of let-7 miRNAs in mamosphere-derived SK-3rd cells as compared with the differentiated/adherent SK-3rd and the SKBR3 cells, which was parallel to elevation of ha-ras oncoprotein. Introduction of let-7 miRNA into the mamosphere-derived SK-3rd cells reduced the capacity of mamosphere formation and the potential of proliferation in vitro, as well as the potential of tumorigenesis and metastasis in vivo. Such effect can be partially achieved by RNAi-mediated silencing of ha-ras oncogene. This study demonstrates that let-7 miRNA loss is responsible for the stemness of breast tumor-initiating cells, and insinuates that supplementing the miRNA may serve as an effective therapy to eradicate cancer stem cells.
Hide

Nevertheless, very limited information is available on its inducers and target genes. We previously demonstrated that Helicobacter pylori (H. pylori) infection is associated with high methylation levels in human gastric mucosae, and that a high methylation level is associated with a risk of gastric cancer development [Maekita, Clin Cancer Res, 2006; Nakajima, CEBP, 2006]. Using Mongolian gerbils infected by H. pylori, followed by its eradication, we were able to demonstrate that inflammation induced by H. pylori infection, not H. pylori itself, was responsible for methylation induction [Niwa, in preparation]. Methylation analysis of 48 genes in gastric mucosae of human individuals with and without H. pylori infection revealed that there is clear target gene specificity in methylation induction [Nakajima, Int J Cancer, 2009]. To clarify the molecular mechanisms for the target gene specificity, genes susceptible and resistant to induction of DNA methylation were isolated by methylated DNA immunoprecipitation-microarray analysis of normal and cancer cells derived from the prostate and breasts. Expression microarray analysis showed that the susceptible genes consisted of a significantly higher fraction of genes with low transcription in normal cells. Chromatin immunoprecipitation and microarray analysis showed that, even among genes with low transcription, the presence of active histone marks, H3Ac and H3K4me3, in normal cells protects genes from becoming methylated in cancer cells, and that the presence of H3K27me3 promotes genes to become methylated [Takeshima, submitted]. This showed that gene usage and histone modifications mark genes that become methylated, and suggested that, since the gene usage and histone modifications are dependent upon inducers of aberrant DNA methylation, these inducers can induce methylation of specific target genes.
Hide

We used our DamID technology to construct a full-genome, high resolution map of the in vivo interactions of interphase chromosomes with NL components in human fibroblasts. This map shows that genome-NL interactions occur via more than 1,300 sharply defined large domains of 0.1-10Mb in size. Various FISH microscopy experiments confirm that these domains are indeed preferentially located at the nuclear periphery.

Lamina-associated domains (LADs) together harbor about 25% of all genes. Strikingly, the vast majority of these genes have very low gene expression levels, indicating that LADs represent a strongly repressive chromatin environment. New data will be presented that provide insights into the possible mechanism of gene repression in LADs. Among others, we found that most transcription factors are unable to bind to their binding motifs within LADs, even though these motifs are equally abundant inside and outside LADs.

We have also begun to study the dynamics of NL associations during differentiation. DamID maps obtained from different cell types show a dramatic restructuring of genome-NL associations, sometimes involving megabases of DNA.

Combining these and other data, we propose a model in which the NL plays an important role in the 'channeling' of signals to the genome, by rendering large gene sets refractory to binding and activation by transcription factors. Reshaping of LADs during differentiation may determine the response of the genome to various stimuli in a cell type – specific manner.
Hide

Indeed, changes in differential gene expression caused by mutations in transcription factors (TFs) or the cis-regulatory genomic DNA elements they bind to (TF binding sites, or TFBSs) can result in a variety of human diseases, including several congenital disorders and cancer. In order to fully understand both normal development and pathologies, and to design effective therapeutics it is critical to understand which TF regulates the expression of which gene, where and under which (developmental) conditions. In addition, it is essential to know the elements each TF binds to and where in the genome these TFBSs are located. This is a major challenge in genomic science as very little is known about the targets, binding sites, transcriptional activity and biological function for the majority of metazoan TFs. We use the nematode C. elegans (worm) as a model to address this challenge. The C. elegans genome contains ~20,000 protein-coding genes, 940 of which we predict encode regulatory TFs. We use gene-centered, rather than TFcentered (e.g. ChIP), protein-DNA interaction mapping methods to map transcription regulatory networks. We then integrate these networks with other networks and data types and analyze how regulatory networks and their properties relate to healthy as well as pathological system states. Progress on network mapping and characterization will be discussed.
Hide

However, little is known about the mechanisms. To understand the genome architectures delineating lineage specification, we determined the whole genome insulator regions bound by CTCF, enhancers by p300 and nuclear lamina chromatin anchors in embryonic and neuronal stem cell genomes. Through these integrated analyses, we found that CTCF, functioning as a barrier, demarcates discrete chromatin domains through chromatin looping that exhibit unique epigenetic signatures. The associations of p300 enhancers and lamin domains within the CTCF defined chromatin blocks are crucial for determining transcription activities. Furthermore, the highly organized spatial distributions of the CTCF-mediated domains suggest functional interplays between self-renewal and pluripotent natures of the genomes. The high resolution, genome-wide map of functional chromatin domains elucidated here and their dynamics in cell specificity should further our understanding on how genome organization orchestrates gene regulation and mammalian development.
Hide

In our previous studies, we found that methyltransferases Dnmt3a and Dnmt3b are associated with chromatin (1). To understand how the enzymes interact with chromatin, we have been trying to identify associated proteins which might mediate this interaction. ES cells that express the enzymes at high levels were chosen as a source. Affinity purification with specific monoclonal antibodies revealed the presence of core histones in the purified protein complex. Histone co-purification was not affected in Dnmt3L-knockout cells, suggesting that Dnmt3a and 3b interact with histones independent of the regulator Dnmt3L that is known to bind to histone H3 tails (2). We are now determining the characteristics of histones which attract Dnmt3s. Specific modification such as histone lysine methylation can serve as a chromatin signal for Dnmt3 recruitment.

We identified through yeast two-hybrid screening the polycomb protein Cbx4 (also known as mPc2) as a potential chromatin mediator for the DNA methyltransferases. The Cbx4 protein contains a chromodomain which recognizes histone H3 tail methylated at lysine 9. Biochemical studies also showed that Dnmt3a can be sumoylated by Cbx4 (3), which also possesses SUMO E3 ligase activity (4). To assess the significance of this modification in epigenetic regulation in development, we have disrupted the Cbx4 gene in the mouse. Our results indicate that deficiency in Cbx4 mainly affects thymus development. The role of Cbx4 in the control of DNA methylation and transcriptional regulation of target genes will be discussed.
Hide

The complex epigenetic process is achieved through multiple mechanisms and understanding this process is necessary to develop effective approach for cancer epigenetic treatment. We have developed a roust drug combination approach that target EZH2-mediated epigenetic gene silencing. With the development of high-resolution, genome-wide approaches to understand chromatin modifications, including DNA methylation and histone modifications, we are able to spur systematic efforts to characterize cancer epigenome. Through integrative analysis of gene expression, chromatin modifications, coupled with pharmacological interventions, we have dissected various epigenetic events in cancer, and. This effort leads to the identification of several putative tumor suppressors with distinct silencing mechanisms affecting multiple oncogenic signaling pathways, including Wnt/beta-catenin pathway in colon cancer. This study provides a framework for further understanding the cancer epigenome and potential development of more effective strategy for cancer epigenetic treatment.
Hide

DNA methylation is essential for mammalian development and health, and perturbation of its dynamics contributes to the development of disease, including cancer. The discovery and validation of the biological indicators (biomarkers) for human cancers are essential steps in the development of methods for accurate subtype classification and outcome prediction in clinic oncology. While genetics (SNP, LOH and mutation) and expression profiling (mRNA and protein) of biomarkers have been extensively assessed for cancer diagnosis and prognosis, the potential for using epigenetic fingerprints for early diagnosis and outcome prediction in clinic oncology propels the exploration of using DNA methylation as a biomarker for cancer prognosis. Our view on the biomarker field in cancer research will be discussed by describing our own research efforts: 1, the DNA methylation profiling of the urine sediments for sensitive/specific detection of bladder cancer; 2, the DNA methylation profiling for the staging and classification of liver cancer; 3, the MBD affinity chromatographic approach based approaches to the cancer methylome and 4, the deep-sequencing of the bisulphite converted DNA for the sequence resolution reference map of the human genome.
Hide