About Us - Research

Summary

Cyclic regulatory programs are fundamental building blocks of living systems. In this project we aim to study three of these cycles, the circadian, cell-division, and nutrient-response cycles, and the interrelationships among them. The circadian cycle involves orchestrated genome regulation depending on a self-sustaining clock that adjusts to day and night conditions. The cell-division cycle corresponds to a series of molecular events that leads to genome duplication and segregation during cell proliferation, and is characterized by checkpoints that allow cycle progression only if certain conditions have been met. The nutrient-response cycle is initiated by exposure to nutrients, which leads a "fasting" cell to undergo a program of gene expression that returns to the fasting state once the nutrients have been exhausted. The phases of each of these cycles are characterized by genomic states that reflect a response to phase-specific signals. These genomic states set up transcriptional programs that then motor the cycle forward to the next phase. Although each individual cycle has been studied extensively, we still know little about the global genomic responses to the cycles and their associated transcriptional regulatory programs. Moreover, the current knowledge is largely limited to RNA polymerase (pol) II mRNA-encoding genes, with few or no studies examining pol III transcription units or pol II non-coding RNA (ncRNA) genes.

We know even less about how the three cycle transcription programs interconnect and influence each other. The "CycliX: transcription regulatory networks of three interconnecting cycles" project aims at a quantitative and comprehensive understanding of the global genomic responses and pol II and III transcriptional programs characterizing each cycle. Importantly CycliX will focus on how and when these transcription programs intersect at shared "core" regulatory networks to assure proper integration and coordination among the three cycling systems.

A special aspect of this project is the analysis of pol III transcription alongside that of pol II. Pol III synthesizes ncRNAs (e.g., tRNAs, 5S RNA, snRNAs, some microRNAs) that are highly regulated during nutrient responses and in proliferating versus non-proliferating cells, as they must double in mass upon each cell division but owing to their great stability need little synthesis in resting cells. Nevertheless, pol III transcription analyses on a genomic scale have been lacking because most of these transcription units are repeated in the genome, making hybridization approaches extremely difficult. However, we have found that by chromatin immunoprecipitation followed by deep sequencing (ChIP-Seq) analysis, we can determine the transcription state of individual repeated pol III transcribed genes. Thus, except for the ribosomal RNA genes transcribed by pol I and the few genes transcribed by the recently discovered spRNAP-IV, we plan to interrogate all transcription activity in a cell.

The project is divided into three subprojects, which combine experimental and computational approaches. In Subproject 1, titled "Identification of cyclic nodes", we will analyze the genomic states at different phases of the three cycles. We will quantify transcriptional activity by measuring levels of pol II and III on the entire genome and by mapping certain histone marks, specifically histone H3 trimethylated on lysine 4 (H3K4me3), H3K36me3, and H3K4me1, using ChIP-Seq. We will thus identify exhaustive sets of pol II and III genes whose transcription activity varies during each cycle as well as sets of genes whose activity varies during at least two of the three cycles, referred to as "cyclic-nodes". These efforts will be integrated with extensive mining and bioinformatics analyses of existing functional datasets, e.g. mRNA accumulation, during these cycles.

To elucidate shared core regulatory network(s) that interconnect the three cycling transcription systems, the regulatory factors controlling the expression of cyclic-nodes, referred to a "cyclic-node regulators", need to be identified. Therefore, in Subproject 2, titled "Interaction network at cyclic-nodes", we will first define functional regulatory regions of cyclic-nodes by identifying active transcriptional enhancer and silencer elements. We will use a novel method for unbiased, genome-wide mapping of DNase I hyper-sensitive sites combined with genome-wide localization of (i) the pol II co-activators p300/CBP and histone H3 that is monomethylated at lysine 4 to identify active enhancers, and (ii) the co-repressors NCOR1 and NCOR2 (SMRT) to identify active silencers. These co-activators and co-repressors mark the large majority of enhancer and silencer elements. From these data, we will be able to identify cis-regulatory elements with high spatial resolution.

Functional links between cyclic-node regulators and corresponding cis-elements will be inferred in two ways. First, we will predict computationally the transcription regulators acting at the mapped cis-regulatory sites using regression models. Second, in parallel, we will apply high-throughput yeast one-hybrid experiments to identify the transcription factors capable of binding to the experimentally identified cis-regulatory sites. We anticipate that some of these cyclic-node regulators will themselves be the products of cyclic-nodes as it is well known that the cell-division and circadian cycles are based on interlocked feedback loops. Once candidate transcription factors involved in cycle-cycle connections are identified, we will directly test whether these factors (or a subgroup of them) indeed bind to the expected regulatory elements in mouse cultured cells and mouse liver with ChIP-Seq and ChIP followed by real-time PCR approaches. By identifying cyclic-node regulators, we thus anticipate that this subproject will reveal a complex transcriptional regulatory network interconnecting the three cycles.

In Subproject 3, titled "Functional consequences of cyclic-node regulator perturbations", we will test the function of candidate regulators directly. We will both overexpress these genes and downregulate them, either by siRNA silencing or, when available, by using knock-outs or other mutations of the relevant genes, and test the effects on cycle-cycle connections. In parallel to the experimental validations, we will develop kinetic models to shed light on the dynamical consequences of interconnected molecular cycles.

Together, these interdisciplinary efforts will lead to a comprehensive understanding of the genome-wide pol II and III transcription programs characterizing the phases of the circadian, cell-division, and nutrient-response cycles. The complex networks underlying these fundamental processes will hence be probed in terms of transcription activity of the genes themselves, the corresponding regulatory elements, and transcription factors mediating the transcriptional programs of cyclic-nodes. Thus, this project will provide unique insight on the design principles of an interconnecting shared core regulatory network(s) that mediates the integration and coordination between three distinct biological systems with cycling capacity.

In many ways, the approach taken here - to examine the (near) totality of cyclic transcription responses and then to work in "reverse" to identify the (near) totality of regulators - follows the systems biology revolution made in large part possible by the sequencing of complete genomes. We do not identify a few key regulators and then try to define their regulatory targets but rather we use our knowledge of the mouse genome sequence combined with revolutionary technologies to take the global approach - an approach that allows us to define in their (near) entirety the transcriptional regulatory networks of three interconnecting cycles: circadian, cell division, and nutrient response.

Specific aims and general organization of the project

Cyclic regulatory systems are ubiquitous in cells and tissues, providing dynamical modalities that are fundamental for life. Circadian rhythms (24 hour cycles), cell division, and nutrition response (fasting/feeding) are characterized by physiological and molecular processes that all cycle through sequential states and ultimately revert to the starting point. Such periodic behavior is in stark contrast to irreversible molecular programs such as cell differentiation or developmental programs, and reflects the difference between oscillatory and switch-like molecular circuits. We expect that the design principles of the molecular networks that perform cyclic responses will be fundamentally distinct from those that govern irreversible processes like those involved in cell differentiation. Importantly, the three aforementioned cycles are known to interconnect. Since interacting cycles are noteworthy for rich dynamical behavior, e.g. the phenomenon of mode locking whereby the phase of one cycle is tied to that of another cycle, we anticipate that crosstalk between cycles should have implications for the dynamical and systems properties of these cycles in cells and tissues.

Although the stages of each cycle are characterized by post-transcriptional events such as protein modifications and degradations, when considering the cycle "engine", the purpose of these events is to set up, at each stage, transcription programs that will motor the cycle forward, programs that result from successive epigenetic changes that are sufficiently malleable that a return to the original state is possible. Thus, the project is based on the premise that the phases of each cycle are characterized by genomic states established in response to signals, that give rise to differential gene transcription programs: these programs then move the cycles to the next stage. Indeed, during the cell-division cycle for example, we know that post-transcriptional processes such as protein phosphorylation, degradation etc. ultimately result in ordered waves of gene expression that drive the cycle forward. Since gene regulatory networks control gene transcription, we need to define the gene transcription networks underlying each process and investigate when and how they interact.

As shown in Figure 1 below, the project is organized into three subprojects of increasing complexity, each of which involves experimental and theoretical components. Overview of the three subprojects and their interconnectivity (hs; hypersensitive. Y1H; yeast one-hybrid. ChIP, chromatin immunoprecipitation. rtPCR; real-time PCR. TF; transcription factor).

Identification of the cyclic-nodes, i.e. the parts, (Subproject 1) will be followed by the mapping of the interaction networks amongst the parts (Subproject 2). Finally, the functional consequences of the interactions between cycles for these cyclic systems will be further investigated by perturbation analysis of key cyclic-node regulators in the network (Subproject 3).

Our specific aims are:

Subproject 1: Identification of cyclic-nodes. We will perform ChIP-Seq to localize pol II and III and certain histone marks (e.g. H3K4me3), H3K36me3) associated with active transcription. Quantification of these datasets will directly reveal transcription activities in "real time" at different phases of the three cycles. Analysis of these datasets will allow us to identify, first, genes whose transcription activity varies during a cycle and, second, genes whose activity varies during at least two of the three cycles (cyclic-nodes). We anticipate that some of the trans-acting factors regulating these cyclic-nodes (cyclic-node regulators, which may often be themselves encoded by cyclic-nodes) will serve as the gears or cog wheels mediating interactions between the cycles. These efforts will be paralleled and integrated with extensive mining of existing datasets on mRNA accumulation during these cycles. A first analysis will be performed with existing microarray datasets to identify genes whose expression varies during any or several of the three cycles studied. Using these methods we expect to gather a comprehensive set of genes encompassing key mediators of interactions between the different cycles.

Subproject 2: Interaction network at cyclic-nodes (core regulatory network). Once we have identified cyclic-nodes of the transcriptional networks involved in the three cycles, we will combine bioinformatics and experiments to identify key regulatory interactions amongst the oscillatory nodes. Specifically, we will identify active enhancer and silencer elements at the different phases of the cycles by using a novel method for an unbiased mapping of DNase I hyper-sensitive sites and by genome-wide ChIP-Seq analyses to localize the co-activators p300/CBP, histone H3 monomethylated at lysine 4 (H3K4me1), and the co-repressors NCOR1 and NCOR2 (SMRT). These co-activators and co-repressors are recruited by other factors to the large majority of enhancers and silencer elements. Benefiting from the high spatial resolution of ChIP-Seq, we will be able to identify precisely the most important cis-regulatory modules, and with this information to predict, using cis-regulatory bioinformatics, the most important transcription regulators acting at the mapped cis-regulatory sites. In parallel we will apply high-throughput yeast-one hybrid approaches to identify the transcription factors capable of binding to the experimentally identified elements and we will confirm binding in mouse cultured cells and liver. These analyses will identify cyclic-node regulators, thus enabling the generation of interconnecting core regulatory networks.

Subproject 3: Functional consequences of cyclic-node regulator perturbations. Once we have identified candidate cyclic-node regulators, we will perturb them in fibroblasts with both overexpression and siRNA silencing as well as, when available, knock-outs or other mutations of the relevant genes. As time allows, we will also study the perturbation of these regulators in the mouse. We will then investigate the mutual desynchronization of the respective cycling systems. In parallel, theoretical studies will investigate the consequences of crosstalk between the cycles in silico.