ERC FUNDED PROJECTS

Homologous recombination (HR) is a conserved DNA double-strand breaks (DSB) repair pathway that uniquely uses an intact DNA molecule as a template. Genome-wide homology search is carried out by a nucleoprotein filament (NPF) assembled on the ssDNA flanking the DSB, and whose product is a “D-loop” joint molecule. Beyond accurate DSB repair, this capacity of HR to spatially associates homologous molecules is also harnessed for homolog pairing in meiosis. The goal of “3D-loop” is to tackle two long lasting conundrums: the fundamental homology search mechanism that achieves accurate and efficient identification of a single homologous donor in the vastness of the genome and nucleus, and how this mechanism is adapted for the purpose of homologs attachment in meiosis. I overcame the main hurdle to study these core steps of HR by developing a suite of proximity ligation-based methodologies and experimental systems to physically detect joint molecules in yeast cells. It revealed elaborate regulation controlling D-loop dynamics and a novel class of joint molecules. This proposal builds upon these methodologies and findings to first address basic properties of the homology sampling process by the NPF and the role of D-loop dynamics, with the long-term goal to establish a quantitative framework of homology search in mitotic cells (WP1). Second, the meiosis-specific regulation of homology search leading to homolog pairing likely integrates chromosomal-scale information. Genome re-synthesis and engineering approaches will be deployed to (i) achieve a quantitative and dynamic cartography of the cytological and molecular events of meiosis over a large chromosomal region, (ii) probe cis-acting regulations at the chromosomal scale, and (iii) revisit the molecular paradigm for crossover formation (WP2). We expect this project to shed light on the fundamental process of homology search and its involvement in the chromosome pairing phenomenon lying at the basis of sexual reproduction.

1 499 779 €

Start date: 2020-01-01, End date: 2024-12-31

Large protein complexes carry out some of the most complex functions in biology. Such structures are often assembled spontaneously from individual components through the process of self-assembly. If self-assembled protein complexes could be engineered from first principle it would enable a wide range of applications in biomedicine, nanotechnology and materials science. Recently, approaches to rationally design proteins to self-assembly into predefined structures have emerged. The highlight of this work is the design of protein cages that may be engineered into protein containers. However, current approaches for self-assembly design does not result in the assemblies with the required structural complexity to encode many of the sophisticated functions found in nature. To move forward, we have to learn how to engineer protein subunits with more than one designed interface that can assemble into tightly interacting complexes. In this proposal we propose a new protein design paradigm, shape directed protein design, in order to address shortcomings of the current methodology. The proposed method combines geometric shape matching and computational protein design. Using this approach we will de novo design assemblies with a wide variety of structural states, including protein complexes with cyclic and dihedral symmetry as well as icosahedral protein capsids built from novel protein building blocks. To enable these two design challenges we also develop a high-throughput assay to measure assembly stability in vivo that builds on a three-color fluorescent assay. This method will not only facilitate the screening of orders of magnitude more design constructs, but also enable the application of directed evolution to experimentally improve stable and assembly properties of designed containers as well as other designed assemblies.

2 325 292 €

Start date: 2018-06-01, End date: 2023-05-31

This project investigates the two-way relationship between spatio-temporal genome organization and coordinated gene regulation, through an approach at the interface between physics, computer science and biology. In the nucleus, preferred positions are observed from chromosomes to single genes, in relation to normal and pathological cellular states. Evidence indicates a complex spatio-temporal coupling between co-regulated genes: e.g. certain genes cluster spatially when responding to similar factors and transcriptional noise patterns suggest domain-wide mechanisms. Yet, no individual experiment allows probing transcriptional coordination in 4 dimensions (FISH, live locus tracking, Hi-C...). Interpreting such data also critically requires theory (stochastic processes, statistical physics…). A lack of appropriate experimental/analytical approaches is impairing our understanding of the 4D genome. Our proposal combines cutting-edge single-molecule imaging, signal-theory data analysis and physical modeling to study how genes coordinate in space and time in a single nucleus. Our objectives are to understand (a) competition/recycling of shared resources between genes within subnuclear compartments, (b) how enhancers communicate with genes domain-wide, and (c) the role of local conformational dynamics and supercoiling in gene co-regulation. Our organizing hypothesis is that, by acting on their microenvironment, genes shape their co-expression with other genes. Building upon my expertise, we will use dual-color MS2/PP7 RNA labeling to visualize for the first time transcription and motion of pairs of hormone-responsive genes in real time. With our innovative signal analysis tools, we will extract spatio-temporal signatures of underlying processes, which we will investigate with stochastic modeling and validate through experimental perturbations. We expect to uncover how the functional organization of the linear genome relates to its physical properties and dynamics in 4D.

1 499 750 €

Start date: 2018-04-01, End date: 2023-03-31

Serotonin (5-HT) is a central neuromodulator and a major target of therapeutic psychoactive drugs, but relatively little is known about how it modulates information processing in neural circuits. The theory of predictive coding postulates that the brain combines raw bottom-up sensory information with top-down information from internal models to make perceptual inferences about the world. We hypothesize, based on preliminary data and prior literature, that a role of 5-HT in this process is to report prediction errors and promote the suppression and weakening of erroneous internal models. We propose that it does this by inhibiting top-down relative to bottom-up cortical information flow. To test this hypothesis, we propose a set of experiments in mice performing olfactory perceptual tasks. Our specific aims are: (1) We will test whether 5-HT neurons encode sensory prediction errors. (2) We will test their causal role in using predictive cues to guide perceptual decisions. (3) We will characterize how 5-HT influences the encoding of sensory information by neuronal populations in the olfactory cortex and identify the underlying circuitry. (4) Finally, we will map the effects of 5-HT across the whole brain and use this information to target further causal manipulations to specific 5-HT projections. We accomplish these aims using state-of-the-art optogenetic, electrophysiological and imaging techniques (including 9.4T small-animal functional magnetic resonance imaging) as well as psychophysical tasks amenable to quantitative analysis and computational theory. Together, these experiments will tackle multiple facets of an important general computational question, bringing to bear an array of cutting-edge technologies to address with unprecedented mechanistic detail how 5-HT impacts neural coding and perceptual decision-making.

2 486 074 €

Start date: 2016-01-01, End date: 2020-12-31