Epigenomic, Nuclear Architectural, and ncRNA Integration of Regulated Gene Expression Programs in Homeostasis and Disease
Summary: Michael Rosenfeld investigates the molecular strategies responsible for integrating programs of genome-wide transcriptional responses to diverse signaling systems critical for physiological and pathological regulation. These strategies are modulated by the endocrine system, which is critical for physiological and behavioral processes in all vertebrates. His work has uncovered unexpected global roles of noncoding RNAs and regulated dynamic alterations in nuclear architecture, both in development and in homeostasis of specific epigenomic strategies. These strategies link regulated gene responses to other cellular response programs, including DNA damage/repair, proliferation, and inflammation. Discovery of these strategies has suggested new approaches to several diseases, including prevalent forms of cancer, neurodegeneration, growth defects, and diabetes.
Molecular Mechanisms of Tumor TranslocationsBy establishing a cellular model that mimics authentic translocation events without proliferative selection, we have uncovered a mechanism underlying nuclear receptor–dependent tumor translocations. The binding of liganded androgen receptor (AR) at introns first juxtaposes translocation loci and alters local chromatin architecture, triggering intra- and interchromosomal interactions. These long-distance chromosomal interactions are apparently required for translocation events. AR then promotes site-specific DNA double-stranded breaks (DSBs) at translocation loci, identified by applying a DSB/chromatin immunoprecipitation (ChIP) methodology that we developed at a genome-wide level, demonstrating the recruitment of two types of enzymatic activities induced by genotoxic stress and liganded AR. These activities include activation-induced cytidine deaminase (AID) and the LINE-1 repeat-encoded ORF2 endonuclease. Together, they generate site-selective DSBs at juxtaposed translocation loci that are ligated by nonhomologous end-joining and additional pathways, causing specific translocations. Specific repressive strategies, including corepressors and ORF2, protect against translocations in normal cells. These findings permit new screening approaches to identify new translocations in breast and prostate cancers and to screen for chemicals that prevent induced tumor translocations.
Because the potential connection between cancer and inflammation has been insufficiently explored, we investigated specific prostate cancer cell–macrophage interactions. We found that cytokine-dependent interactions impose an innate immunity-type response, mediating a switch from repression to activation of selective androgen receptor antagonists/modulators (SARMs) in vivo. This process is based on an evolutionarily conserved receptor N-terminal helical motif, which is present in sex steroid receptors. This motif recruits a sensor, TAB2, with inflammation-dependent recruitment of TRAF6 and MEKK1 to dismiss the nuclear receptor corepressor (NCoR) complex. This sensor strategy appears to subserve essential reproductive functions, mediating reversal of sex steroid-dependent repression of a "negatively regulated" gene cohort by E2.
In parallel studies of response to genotoxic stress, we discovered that a protein, tyrosine phosphatase, Eya, is involved in promoting efficient DNA repair rather than apoptosis during mammalian organogenesis. Eya executes a damage signal-dependent dephosphorylation of a γ-H2AX carboxyl-terminal tyrosine phosphate (Y142) by adjudicating the relative recruitment of either DNA repair versus pro-apoptotic factors.
Noncoding RNAs, Subnuclear Architecture, and DNA Repeat Transcription as Integrators of Regulated Gene Response Programs
Because the biological functions of the massive numbers of long noncoding RNAs (ncRNAs) remain enigmatic, we explore their potential roles as sensors and integrators of transcriptional response. We have shown that an RNA-binding protein, TLS (translocated in liposarcoma), is a key transcriptional regulatory sensor of DNA damage signals. TLS is allosterically modulated by RNA to permit binding to and inhibition of CBP/p300 histone acetyltransferase activities on repressed gene targets. Thus, recruitment of TLS by low-copy-number ncRNA transcripts induced by genotoxic stress and tethered to the 5' regulatory regions of growth-control genes causes gene-specific repression.
We extended these observations of allosteric effects of ncRNAs to the question of nuclear architecture, uncovering a highly instructive ncRNA/methylated cofactor interaction underlying activation of the serum-dependent growth-control transcriptional program. The potential interacting roles of distinct subnuclear architectural features and large numbers of ncRNAs in gene regulation remain elusive. We have found that Polycomb body (PcG)- and interchromatin granule–specific TUG1 and NEAT2 ncRNAs, respectively, license relocalization of growth-control genes between these two compartments, based on recognizing the methylation status of Polycomb 2 protein (Pc2) in response to growth signals, based on altering activities of a methyltransferase (Suv39h1) and a demethylase (KDM4C). These ncRNAs direct the assembly of transcriptional corepressor/coactivator complexes, thereby specifying the expression of growth-control genes in a spatially organized manner.
Unexpectedly, demethylated Pc2 is required for SUMOylation of E2F1, which causes recruitment of a ubiquitin E3 ligase mediating H2B monoubiquitination, culminating in activation of the growth-control gene program. We hypothesize that a general ncRNA-dependent sensor strategy relocates large subsets of regulated transcription unit cohorts. Furthermore, our data suggest that ncRNAs act as allosteric modifiers of readers and writers of the histone code, implying a critical role in regulation of gene expression programs.
This research also led to the discovery that a subset of human Alu DNA repeats containing canonical DR2 retinoic acid receptor (RAR)-binding sites (DR2 Alu repeats) are bound and regulated by RAR in human stem cells. This causes polymerase III (Pol III)-dependent transcription of ~4,000–5,000 DR2 Alu repeats, generally in proximity to RA-activated Pol II transcription units. The regions are rapidly translocated to the nuclear pores, and DR2 Alu transcripts exhibit apparent Ago-dependent cytoplasmic processing to generate novel small RNAs, referred to as repeat induced RNAS (riRNAs). These riRNAs appear to mediate degradation of a specific subset of "stem cell" mRNAs , including Nanog, that harbor complementary sequences in their 3'-untranslated regions (3'UTR) participating in the program that causes the pluripotent cells to leave the stem cell state.
Enhancers, ncRNAs, and Mechanisms of "Epigenomic" Regulation
We have continued to explore key regulatory strategies in higher eukaryotes dedicated to imposing robust control on the dismissal of corepressors, preventing illegitimate signal-independent gene activation. Our laboratory, which first identified NCoR, established the role of two highly related transducin β-like proteins, TBL1 and TBLR1, components of the NCoR/SMRT corepressor complex. TBL1 and TBLR1 are key in permitting regulated gene activation events by many classes of transcription factors, requiring the phosphorylation of five specific Ser/Thr phosphorylation sites, allowing interaction with the specific machinery required for corepressor/coactivator exchange. Small interfering RNA screens revealed the coordinated actions of TBL1 and TBLR1 for ubiquitylation and dismissal of CtBP1/2 and NCoR/SMRT, respectively. These studies uncovered mechanisms that underlie pathogen-specific responses and disease-specific programs of inflammation. The delineation of these mechanisms, in collaboration with Christopher Glass (University of California, San Diego), has proven that they underlie the biologically critical transrepression of distinct subsets of proinflammatory genes by nuclear receptors (PPARγ, LXRs, Nurr1, and RORα), employing analogous SUMOylation-dependent mechanisms. Although highly homologous, the corepressors NCoR and SMRT serve distinct functional programs. We found, for example, specific functional requirements for SMRT in the actions of both retinoic acid–dependent and Notch-dependent forebrain development. Our data revealed that SMRT represses expression of the Jumonji-domain containing gene, JMJD3, a direct retinoic acid–receptor target that others and we discovered. JMJD3 functions as a histone H3 trimethyl K27 demethylase, activating specific components of the neurogenic program.
We have also investigated specific roles for subsets of gene enhancers that dictate cell-type identity. Each enhancer "class" regulates distinct biological programs during in vivo organogenesis. Our initial data suggest that ncRNAs expressed at enhancers and induced by ligands/signals may participate in exchange of corepressor/coactivator complexes in the developmental stage of gene activation. We established that β-catenin acts as a binary switch to simultaneously activate expression of the critical lineage-determining transcription factor, and the initial documentation that homeodomain proteins can serve as a direct target for -catenin recruitment. Sustained Notch signaling in progenitor cells was required to prevent conversion of the late-arising cell lineages to the firstborn cell phenotype, a strategy commonly used in mammalian organogenesis.
Our investigation of specific enzymatic machinery contributed several insights into gene regulatory programs. We found that the histone lysine demethylase, LSD1, is required for late cell-lineage determination and differentiation in vivo during pituitary organogenesis, acting primarily on target gene activation programs. Postpartum estrogen-dependent, induced expression of a Krüppel-like repressor (ZEB1) and Pc2 each act as a molecular beacon for recruitment of the LSD1-containing CoREST/CtBP corepressor complex, repressing an additional cohort of genes that previously required LSD1 for activation. We identified and characterized a novel histone H2A K119 ubiquitin ligase (2A-HUB) and a histone H2A deubiquitinase (2A-DUB) that participate in androgen receptor–dependent gene activation.
We have uncovered a strategy based on the requirement for specific cohorts of inhibitory histone methyltransferases (HMTs) to impose gene-specific gatekeeper functions that prevent unliganded nuclear receptors from modulating gene activation in the absence of stimulating signals. We found that a PHD and Jumonji C (JmjC) domain-containing protein, PHF8, functions primarily as an H4K20me1 demethylase, the identity of which had previously been unknown. PHF8 is recruited to promoters by its PHD domain based on interaction with H3K4me2/3 and controls G1–S transition in conjunction with E2F1 mediating recruitment of HCF-1/SET1A. Phosphorylation-dependent PHF8 dismissal from chromatin in prophase is apparently required for the accumulation of H4K20me1 during early mitosis as a putative component of the condensin II–loading process. ChIP-Seq analysis revealed a significant overlap of condensin II and H4K20me1 sites in mitotic cells. Additionally, HEAT repeat clusters in two condensin II subunits (N-CAPD3 and N-CAPG2) were able to recognize H4K20me1, identifying a new class of modified histone "readers" and functionally linking regulators of chromatin architecture to specific epigenomic machinery/histone marks. In parallel, we have identified PHF2 as a histone H4K20 di- and tri-methyl-demethylase.
In conclusion, finding the interrelationship between ncRNAs, subnuclear architectural structures, and their ability to serve as allosteric modifiers of the "epigenomic code" should lead to insights into the integration of regulated transcriptional response programs in development, homeostasis, and disease.
This work was supported in part by grants from the National Institutes of Health, the National Cancer Institute, and the Department of Defense.
As of May 30, 2012