The overall goal of my research project is to identify key DNA/RNA cis-regulatory elements (CREs) in pluripotent cells and to define their interplay with trans-regulatory factors, in particular transcription factors and microRNAs.
The first step of this project will be the analysis of the huge amount of sequencing data produced by academic consortiums to locate the position of pluripotency CREs (Aim 1). We plan to verify the activity of these CREs using reporter assays and test their interactions with the genes of interest using chromosome conformation capture technology.
We will then perform a rigorous genetic analysis to study the dependency between these CREs and the ability of induced-pluripotent stem cells (iPSCs) to establish or maintain a pluripotent state (Aim 2). This will be done deleting CREs of interest in iPSCs by TALENs technology.
Finally, we will apply biochemical, proteomic and molecular biology tools to dissect the relationship between the sequence and regulatory activity of these CREs, examining whether CRE activation kinetics are primarily determined by cis- or trans-factors (Aim3). In particular we will dissect CRE activity at single nucleotide resolution using the pioneering massively parallel reporter assay (MPRA) under development in the host research group.
The generation of quantitative models of pluripotency CRE activation will allow us to better understand stem cell biology and will enable the development of new IPSCs derivation protocols and more accurate tools to shape cellular states.

It is becoming increasingly clear that the genomes are pervasively transcribed. Presently, the function(s) of these non-coding RNAs are unknown. We will pursue the hypothesis, based on our preliminary findings, that these transcripts contribute to genome stability. By a next generation sequencing approach we will analyses these transcripts and we will study them in relation to site-specific DNA damage and the activation of local DNA damage response activation.

Our ability to behave meaningfully in our environment depends critically on having a transient memory buffer in which to upload and hold on-line past experiences, goals or aspects about the current context, so that they can be integrated in choosing our course of action. This buffer is referred to as working memory and is critical to cognition because it allows behavior to be guided by internal representations acquired in the past, rather than by the immediate contingencies of our surroundings. Without it, our behavior would not amount to much more than a series of reflexes. Experiments with patients with brain lesions together with measurements of the activity of neurons from behaving animals show that the frontal part of the brain, in particular a brain area called the prefrontal cortex, is critically implicated in working memory.
In the laboratory, working memory is studied using simple tasks. For instance, a particular image displayed for 1 s might require pressing a particular key, but the subject is instructed to only perform the action after a ‘delay period’ of 5 s subsequent to the image presentation has elapsed. Animal experiments have shown that particular neurons in the prefrontal cortex are activated during the delay period while animals remember certain images but not others. These neurons might, thus, be part of a neuronal representation of the memory of the image.
Despite decades of working memory research, a number of fundamental questions remain unanswered: What properties of neurons and circuits allow their activity to persist in time without sustaining inputs from the senses? How robust are neuronal representations in working memory? This last point is crucial because the brain is constantly active, so the neurons participating in the memory representations are always receiving unrelated information which must be filtered out. Without stability against these perturbations, working memory function would not be possible. Because networks of neurons are very complex dynamical systems, answering these questions will ultimately require, not only new experimental approaches, but also a quantitative conceptual framework, in order to guide the design of experiments and to interpret the experimental data.
In this proposal, we will use a parallel experimental-theoretical approach to address these questions. Recently developed techniques for measuring and manipulating brain activity in mice performing simple working memory tasks will allow us, for the first time, to measure directly the stability of neural memory representations and to elucidate which properties of prefrontal neurons and circuits are responsible for generating them. At the same time, the data from our experiments will be used to assess the validity of current theoretical frameworks of working memory, to refine them, and to generate novel hypothesis regarding the circuit basis of working memory function in the prefrontal cortex.

Copper is a trace element that it is critical to all living organisms. Ironically, a number of severe neurodegenerative disorders (i.e., Alzheimer’s disease) are linked to the disruption of copper homeostasis. An attractive non-invasive approach to study copper distributions is to use of small-molecule fluorescent probes in imaging experiments. However, a vast majority of these tools cannot be delivered to a specific organelle in a highly controlled fashion. An alternative strategy, one that has recently attracted much attention, is to use genetically engineered protein-based sensors. It is well established that localization signals ‘tag’ a protein for transport to a specific local within the cell. These small amino acid sequences can be easily incorporated into protein sensors which will allow targeting to specific organelles. Thus, the goal of the proposed research project is to design a genetically encoded fluorescent copper(I) sensor capable of achieving high spatial resolution. Such probes hold great potential significance since the mechanisms of copper accumulation, trafficking, and function has been inadequately studied.

Resurrecting ancestral proteins can provide fascinating insight into the evolutionary history of molecules, a window on the past revealing how present day adaptive specializations may have occurred. Visual pigments form the first critical step in the phototransduction cascade within the photoreceptor cells of the eye. Some of these pigments, the rod opsins, are known to possess extraordinary adaptations for maximizing photosensitivity under low light conditions, including ultrafast femtosecond photoactivation speeds coupled with remarkable thermal stability. How did such extreme specialization evolve? Ancestral reconstruction approaches have been successful in resurrecting and functionally characterizing a 200 million year old ancestral archosaur rod visual pigment. However, due to the effort and expense associated with laboratory resurrection, this approach is necessarily limited to small numbers of inferred sequences, making the detailed description of important evolutionary transitions and the testing of different evolutionary hypotheses extremely difficult. In this proposal we argue that these difficulties can be overcome by exploiting recent advancements in computational photochemistry that, in principle, allow for the systematic reconstruction of ancient visual pigments in silico. Indeed, it has been recently reported that computer models of visual pigments based on ab initio multiconfigurational quantum chemistry, not only reproduce the changes in ?max values induced by different mutations, but also simulate the effect of the sequence variation on pigment photochemical and thermal activation. Such technology can thus be used to compute the optical and photochemical properties of inferred ancestral sequences, for comparison to modern visual pigments. Here, we propose a research project that integrates, for the first time, the diverse disciplines of ab initio quantum chemistry and ancestral protein resurrection to achieve an unprecedented understanding of the functional divergence of dim-light visual pigments. Accordingly, we will develop and evaluate a computational protocol for the systematic construction of computer models of ancestral pigments. With this technology at hand, large sets of pigments can be screened in silico for interesting functional shifts. These computational investigations will be combined with laboratory investigations of resurrected ancestral pigments focusing on a small number of key targets identified by the quantum chemistry modelling. Our highly innovative combined approach should lead to a better understanding of the evolution of rod visual pigments, and how they achieved their remarkable adaptations for dim light vision.

Using FDRs as a model, I will address the molecular basis of promiscuity and functional divergence among gene duplicates via an integrative approach: (1) Sequences of FDRs available from public databases will be analyzed to reconstruct the evolutionary history of FDRs. (2) Based on statistical inference, ancestral FDRs at critical evolutionary transitions will be resurrected. The structure, stability, and substrate range of ancestral and extant FDRs will be examined through experiments or computational modeling to identify the mechanistic basis of functional shifts. FDRs that show a broad substrate range will be compared to those with restricted catalytic capacity to probe the cause of promiscuity and the tradeoff of functional specialization. (3) Promiscuous FDRs, particularly mercuric reductase that exhibits cross-reactivity with several heavy metals and thus the potential for bioremediation, will be subject to directed evolution in order to enhance these secondary functions. These FDRs will go through several rounds of mutagenesis and high-throughput screening until the desired activity is reached. (4) To monitor how natural selection drives functional diversification of gene duplicates in real-time, I will force a promiscuous FDR to take over multiple catalytic reactions originally performed by separate FDR members in Escherichia coli. The promiscuous FDR will be introduced into a multi-FDR E. coli knockout through genetic manipulation. The strain relying on this promiscuous FDR for growth will be subject to long-term laboratory evolution. Observing how mutations cause shifts of protein functions in (3) and (4) will yield critical insights into molecular evolution.

The ability of animals and humans to learn new tasks is astonishing. How does the brain learn to choose actions that will maximize rewards in a new scenario? Much is known about how we learn to evaluate different stimuli in the environment in terms of how much future reward they predict – this process is thought to depend on learning in the striatum, and on prediction error signals conveyed to the striatum by the neuromodulator dopamine. However, in realistic learning situations the problem is more complex: although our environment consists of a myriad of stimuli, only few are relevant to any specific task. The striatum would be extremely inefficient if it were to learn about all the available stimuli. Indeed, the speed by which we learn new tasks suggests that there are neural processes that identify the task-relevant stimuli and direct learning towards these alone. Here we propose that this function is fulfilled by an understudied neuromodulatory system: the intrinsic striatal cholinergic system. This unique and mysterious system has been implicated in a variety of psychopathologies, including ADHD, schizophrenia and Parkinson’s disease, however, it has been the focus of relatively little scientific research, and as a result the precise function of striatal cholinergic neurons is far from clear. To remedy this, we propose an international multi-disciplinary collaboration that will use a variety of neuroscientific methods to test an overarching hypothesis: that striatal cholinergic neurons provide a “filter”, directing learning towards task-relevant stimuli and away from distractors. We will develop a task that is especially tailored to test this hypothesis, and furthermore, can be used with both humans and animals. This will allow us to use complementary research methods including single cell recordings, optogenetic manipulations, and brain imaging, to provide converging evidence regarding the role of the striatal cholinergic system in learning, and its interaction with dopaminergic processes. Tying all these together will be a computational model of how animals and humans are able to learn in our task, and how different attentional spotlights affect their performance. The proposed project will shed light on a functionally and clinically important neuromodulatory system, with far reaching implications in terms of understanding the neural basis of learning in real-world complex scenarios.

The cells found circulating in blood, including platelets, red cells, and the white blood cells of the immune system are produced continuously throughout our lives. Despite their diverse range of functions, all of these develop from a single progenitor cell type found in the bone marrow. How apparently similar progenitor cells give rise to, and regulate, the tremendous cellular diversity that is required for the blood system to function, is poorly understood. A better understanding of how these blood cells are generated is critical to the development of new therapies for replacement and restoration.
Schumacher’s laboratory has developed the molecular equivalent of ‘bar-coding’, a mechanism to label individual progenitor cells uniquely with a tag that is inherited by all their offspring. This technique enables a much more detailed study of the contribution of individual progenitors to the entire blood system than earlier methods. Complementing this high-level view, members of Schumacher’s team have been developing a methodology for time-lapse imaging of individual progenitors, capturing the patterns of development for family trees of cells and their progeny.
Traditional approaches to extracting family trees from movies have a human study images on a frame-by-frame basis; a costly and time-consuming process. Automated, computer-based extraction is challenging as cells move, change shape and pass under one another. Our approach is to develop a hybrid strategy. Cohen and his computer-engineering laboratory will develop algorithms based on machine learning that automate processing where possible and request human intervention in cases of uncertainty. These algorithms will be packaged into a user-friendly program distributed freely to benefit all scientists.
The data that results from these experimental methods is highly complex. The Hodgkin laboratory, which contains both experimental and theoretical scientists, has a long history of developing mathematical models of biological systems, providing concise, analytic descriptions and using statistical techniques to identify hidden order; their role in this project to interrogate both sources of data and abstract key features. These models will be fed back to Cohen to aid in automated data extraction and forward to Duffy for further investigation.
Duffy’s laboratory consists of experts in mathematical analysis of models. They will collaborate in the statistical analysis of data and perform a stochastic analysis of the best model of the blood system to enable extrapolation and prediction from the model. In collaboration with the Hodgkin laboratory, the most significant findings of this analysis will be fed back to Schumacher’s laboratory to inform further experiments that challenge and refine our understanding.
Ultimately, this grant’s aim is to utilize these new experimental technologies in concert with advanced computer science and mathematics to gain a deeper understanding of the blood system.

The collective behavior seen in assemblies of cells as well as groups of animals is the cornerstone of biological organization and distributed information processing. We propose to bring together two distinct approaches - dynamical self-organization and information-theoretic analyses - to probe how collective behavior emerges in fish groups.
We will base our research on new high-resolution recordings of fish group motion where hundreds of individuals can be precisely tracked for hours. This unprecedented amount of fish trajectory data collected by the Couzin lab, as well as plans to measure new behaviorally relevant quantities (the visual field of the fish, responses to various perturbations like laser pulses, trained individuals, robot-controlled models and virtual interacting fish), will enable us to apply new modeling approaches. Rather than assume an ad-hoc model and tune it to qualitatively reproduce group behaviors, we will use data-driven dimensionality reduction and probabilistic models from neuroscience (from Tkacik and Schneidman labs) to reconstruct the computations that the fish perform to determine their future movement, and to infer how information flows between the group members.
Specifically, we propose:
(i) Using “maximally informative dimensions”, we will ask what combination of the neighbors’ states is most informative about the individual’s future motion. This data-driven approach can infer how the fish integrate the information about neighbors, and how much of this information (in bits per unit time) they collect.
(ii) Using information theoretic measures we will quantify which kind of information (speed, direction, local order), how much of it, and how quickly, propagates beyond the range of the receptive field.
(iii) From these measured quantities we will build bottom-up models of collective behavior, and ask how much information minimally needs to be exchanged to support coherent motion; this can be compared with the measurements in (i,ii) and experimentally tested by decreasing the amount of available information by e.g. lowering the light level.
(iv) We will build top-down models for the whole group directly from data using maximum-entropy and generalized-linear-model approaches. We will look for the least complex interactions between individuals that explain the data and for signatures of hierarchical organization.
(v) We will ask if certain group configurations and geometries are better suited to support the rapid spread of information and generate quick collective responses. This can be tested using perturbation experiments and interpreted in an ecological context.
This research plan pursues the idea that organisms need to actively gather noisy information from their neighbors and the environment to function in a synergistic way across scales. Information theory can be used to quantify the information flows in large new data sets, and to reproduce them consistently in top-down and bottom-up models of collective dynamics.

How animals climb smooth surfaces using adhesive mechanisms is a fundamental biological question. Fibrillar and smooth adhesive structures are observed across a range of animals from insects to lizards, all of which rely upon controllable adhesion to climb, mate, or capture prey. The presence of adhesive hairs in many climbing and clinging organisms in Nature has led biologists to focus on the fibrillar design as the key principle for adhesion-based locomotion. Recent studies by the co-applicants and others suggest that sub-surface and internal structures, particularly with regards to compliance, represent critical ingredients for adhesion. These structures may be essential for the up-scaling of adhesive forces in large animals (e.g., geckos), allowing them to achieve adhesive capacities similar to animals much smaller in mass (e.g., beetles). We hypothesize that the ability for an organism to use adhesion-based climbing depends upon the ratio of an organism’s attachment area to the compliance of the whole system. Specifically, we predict that larger organisms should evolve increasingly stiff sub-surface structures to allow them to use adhesion to climb. We argue that evolutionary changes in the compliance of subsurface structures represent one of the reasons for why animals of different sizes can adhere to surfaces. We test this hypothesis (Fig. 1) through laboratory and field experiments with two groups that have evolved adhesive pads (insects and geckos). We will perform (A) Laboratory studies of subsurface morphology and material properties, (B) Laboratory studies examining effects of compliance on locomotion and adhesion, (C) Field studies examining compliance of surfaces in nature, and (D) Modeling of Subsurface Mechanics and Adhesion Capacity. To accomplish these experiments, our proposal unites three researchers (Irschick, Federle, and Crosby) who bring different expertise to the broader study of adhesion. Irschick is a functional morphologist who has studied how animals perform (jump, run, etc.) in the laboratory and in nature. Federle’s research combines an ecological/evolutionary perspective with a mechanistic and biophysical approach, and Crosby is an expert in materials design and fabrication, especially in the context of soft polymer adhesion and mechanics. Each will conduct different experiments, and will also work together to achieve these different goals. We aim to uncover fundamental scaling laws from insects to lizards by integrating field and laboratory work, and thus our project will provide a critical step in establishing how nature has accomplished adhesion-based locomotion for organisms that range in mass over several orders of magnitude.

The goal of organismal neuroscience is to understand how the circuits of neurons that make up the central nervous system interpret sensory information about the outside world and use that to generate behavior. Implicit in the execution of any behavior is a choice made by the organism, with the nervous system evaluating the costs of that behavior (e.g. effort of movement towards a goal) versus the perceived pay off of the behavior (e.g. reaching a satisfying meal). Behavior is shaped by neural circuits, which are in turn shaped by genetics and by life stage (i.e. different individuals have different genes and experiences, and therefore different neural circuits which consequently will make different choices, but so too will the same animal when it is young vs. old, virgin vs. pregnant, etc.). Progress in neuroscience is hindered by the difficulty of quantitatively measuring the different values that go into choices made by animals in complex, natural settings. Human observers often miss subtle behavioral variance, food or mate choices may involve too many or unknown parameters. Issues of scale arise from time a skilled observed must put into analysis, and the complexity of nervous systems (e.g. 200 billion neurons in a human brain). Open questions include what neural circuits are involved in valuation, and what and how genes shape these circuits. The fruit fly Drosophila melanogaster is a model organism that presents a balance between neurological and behavioral complexity, and approachable simplicity. Drosophila have ~100,000 neurons. They navigate the world using visual, tactile, and chemical cues and make cost-benefit analysis involving the use of limited energy stores to find the best feeding sites, where they eat, mate, and lay eggs. While this provides a rich animal-environment interactions, the flies’ neural circuits and genetics are relatively simple compared to mammals. Recent advances in genetics allow targeting specific neuronal populations in the fly. Advances in machine vision and computation allow high throughput analysis of behaving Drosophila, and advances in mathematics allow sophisticated models to be extracted from large and complex behavioral data sets. We have designed behavioral arenas to enable automated tracking and analysis of fruit flies making choices between food sources of different qualities while avoiding naturalistic obstacles. We aim to combine these experiments with genetic manipulations of specific neuronal circuits to assess the role these circuits play in behavioral decision making. This set up will allow an unprecedented rate of discovery in the neurogenetic underpinnings of behavioral choice in awake, behaving animals presented with quantifiable naturalistic cost and benefit considerations. Finally, we will use this vast cache of behavioral data to create models of neural-circuit driven decision making in Drosophila melanogaster.

To understand and interfere with the stages of the virus life cycle, knowledge of the structural properties of viruses and their assembly intermediates is required. A case in point is the human immunodeficiency virus type 1 (HIV-1), which, despite intense study, still presents challenges coming from a limited knowledge of its architecture and the transformations associated with its passage from the noninfectious immature state to the infectious mature state during the viral life cycle. An intriguing feature of the immature HIV-1 is the sizable gaps in its protein lattice observed by electron microscopy. The origin of the gaps is not understood but they may be important for the timing of release of
the viral particle from the cellular membrane before assembly is completed. We hypothesize that the gaps are unavoidable consequences of the high aspect ratio of the protein subunits constrained to assemble on a spherical shell. The hypothesis represents a new paradigm in virus assembly and will be tested through a multi-pronged interdisciplinary approach including: a) mutagenesis, b) complementary structural studies of immature particles by
cryo-TEM and SAXS, c) theoretical modeling of rod-like subunits spherical assembly, d) directed assembly on templates of various radii.

The proliferation of all organisms depends on the precise coordination of a large number of enzymatic events within multiprotein replisomes to ensure that chromosome duplication occurs in a timely and faithful manner. While the core architectural characteristics of replisomes are known, the inner workings of these sophisticated molecular machines remain poorly understood. In particular, it remains unclear how discontinuous synthesis on the lagging strand, which requires several slow enzymatic steps, can occur without causing loss of coordination with continuous synthesis of the leading strand. Even less is known about the fate of this coordination during encounters with obstacles, such as DNA lesions, which is a common event during chromosome duplication. To understand the coordination of enzymatic events during replication, I will directly observe the structural and functional properties of single E. coli replisomes while in operation. I will simultaneously monitor synthesis occurring on each daughter strand using labels attached to the ends of DNA substrates while they are being replicated. By introducing DNA lesions into these substrates and working with fluorescently labeled replisome components, I will observe the ultimate fate of replisome subsystems during encounters with DNA lesions. These studies will provide detailed information about the multitude of short-lived states and dynamic events that occur at the replication fork and reveal how encounters with DNA lesions alter replisome coordination and composition at the single-complex level.

Computational protein design is shedding light on molecular recognition and holds the promise of generating molecular functions not seen in nature. During my HFSP-supported postdoctoral training I developed the first general method for designing protein interactions and produced proteins that neutralized influenza by inhibiting its hemagglutinin surface protein. The overall goal of the proposed research is de novo design of unstructured protein segments (loops) for function. Loops are the principal mediators of molecular recognition in immune-system antibodies and many enzymes, and are particularly important for binding specificity, but due to their flexibility have been recalcitrant to design and engineering. The proposed research aims to (a) characterize the sequence and structural features by which loop conformations in natural protein complexes are rigidified, setting the stage for incorporating these principles in computational design methodology; (b) develop a novel method for design of rigid loops by introducing substitutions that energetically destabilize conformations other than the target one; and (c) produce exquisite specificities by de novo design of loops in a natural interacting pair that destabilize binding to undesired targets and favor association with new targets. This interdisciplinary research combines high-performance computing and high-throughput assays of protein function to clarify a poorly understood aspect of molecular recognition: conformational specificity. This understanding may open the way to design of exquisite substrate specificities in enzymes as well as to antibody design, leading to novel therapeutics and diagnostics.

The functional identification of a sensory system requires characterizing the relationship between a set of stimuli and the related evoked responses, so as to identify the sensory features for which this system is selective and the computations that it performs. When applied to mammalian visual cortex, this approach has revealed that the estimated input/output relationship of cortical neurons depends on the statistics of the stimulus that is presented, reflecting the extraordinary complexity of the nonlinear computations performed by this cortical network. Detailed analysis of these contextual nonlinearities is critically needed to claim that we understand cortical vision in natural viewing conditions. In this project, my approach will be to investigate the nonlinear dynamics of sensory cortical networks in a structurally simpler and evolutionary ancient system, the three-layered visual cortex of a reptile, the turtle. I will assess the functional properties of this visual cortex in the anaesthetized turtle with a set of visual protocols designed to dissect out the functional operations involved in the processing of natural stimuli. Using a combination of electrophysiological and imaging techniques, I will probe in vivo the encoding of the stimulus at different levels of integration (from the subcellular to the population level) and explore in vitro the connectivity schemes underlying the coding properties of functionally identified cortical neurons. This approach should provide useful data to help understand the functional logic and basic mechanisms underlying the emergent nonlinearities observed in vertebrate sensory cortices.

How environment modifies the genetic basis of insecticide resistance is a fundamental question. In changing environments, studying such a mechanism helps predict how insects develop resistance given the environment they evolved in. In Drosophila melanogaster (Dm), resistance to DDT is widely observed and often mediated by a major gene, Cyp6g1. However, quantitative resistance is also observed but has received little attention. Climate adaptation was reported for Dm populations but the implication of adaptation on functional traits is poorly investigated. We aim to identify the genetic factors involved in quantitative resistance to DDT in Dm and how these are expressed in populations adapted to particular temperature and precipitation regime. We will first identify the molecular variation associated with resistance for various level of DDT exposure in different combinations of high and low temperature and humidity using a fully sequenced diversity reference panel of Dm. This will draw a list of gene polymorphisms involved in DDT resistance for each set of in-vitro conditions. We will then sample natural Dm populations that experienced in-natura conditions similar to the one imposed in-vitro and we will study the pattern of functional polymorphism involved in DDT resistance in the wild. This will help assess the potential of the natural population to develop quantitative resistance to DDT. We will then expose naturally adapted populations of Dm to DDT both in their native conditions and in exotic conditions. This will test whether populations evolved in a particular environment are more likely to develop quantitative resistance in their native conditions than in new ones.

The vertebrate embryo body is segmented and elongated along the antero-posterior axis. Segmentation of the vertebrate body into vertebrae precursors takes place simultaneously with the extension of the body axis. For the synergistic progression of these two processes, cell behavior and tissue patterning needs to be precisely coordinated. To reach a comprehensive description of the complexity of this system we propose a multidisciplinary project involving biologists, image processing experts and modelers. High-resolution time-lapse imaging of quail transgenic embryos will be used to collect data of the molecular, cellular and tissue dynamics underlying axis formation and patterning with a previously unmatched precision. Analyzed data will be used to build, for the first time, a 4D model integrating morphogenetic and patterning processes. This model will be used to challenge existing and formulate new hypotheses in the field of axis elongation and segmentation. Our collaboration significantly profits in the way that novel in silico predicted properties will be reiteratively challenged experimentally in vivo to feed back and improve our model.

The brain creates dynamic representations of the sensory environment by extracting stimulus features at early processing stages and synthesizing more abstract object representations in higher brain areas. However, the nature of the underlying neuronal circuits and their strategies to perform this task are still poorly understood. This research project aims at identifying neuronal computations representative for basic cortical circuits and uncovering the underlying cellular mechanisms. To this end, I will visualize and manipulate different types of interneurons (INs) in the posterior zone of the dorsal telencephalon (Dp) of zebrafish, which is homologous to olfactory cortex in mammals and assumed to be involved in olfactory object representations and odor memory. Using transgenic marker lines, I will measure odor responses of different IN types by 2-photon Ca2+ imaging and targeted whole-cell recordings. I will analyze the synaptic architecture of Dp by electrophysiology and optogenetics and derive hypotheses for the function of different IN types in the context of the circuit. These hypotheses will then be tested by reversibly silencing INs in transgenic lines, and the resulting cell-type specific effects will be examined at the level of individual neurons and large-scale activity patterns. Behavioral experiments will finally address the roles of the different IN circuit elements in odor discrimination learning. In parallel, I will examine the modulatory role of acetylcholine on circuit function and plasticity in Dp and on odor discrimination learning. The results will provide insights into canonical computations and control of plasticity in basic cortical circuits.

When an organ responds to a stimulus, the response requires the coordination between multiple cell types. How do different cell types respond to a stimulus? Do all cell types within an organ respond? Do they respond identically? In plants, cell-type specific transcriptome profiling in root has ruled out the possibility that the stimulus triggers the same response in all cell types. Moreover, targeted misexpression of specific components of individual hormone signalling pathways has revealed that only certain cell types respond to individual hormones. However, how different cell types play an integrated role in the growth response initiated by changes in light or temperature is not understood.
The aim of this project is to unravel the spatial and temporal dynamics of how individual cell types within shoot organs respond to variation in ambient light or temperature.
We will generate different shoot cell-type tagged Arabidopsis transgenic lines and perform RNAseq analysis of the different cell types during the response to changes in two environmental stimuli that induce shoot growth: a change in light quality and an increase in temperature. We will look for cell-type specific transcriptional networks and GO enrichment during the responses to find the contribution of each cell type to the overall process. Also, by ChIPseq, we will study three histone modifications known to respond to changes in light and temperature in two selected cell types that respond differently to the stimuli. Then we will integrate the transcriptomic and epigenetic data to try to answer whether changes in epigenetic marks between cell types explain their different ability to respond to a stimulus.

Accurate segregation of replicated chromosomes during mitosis is critical for the maintenance of a stable genome. Chromosome segregation depends on the enzyme separase, which cleaves the cohesin proteins that hold sister chromosomes together. In yeast, separase activity has been found to be dynamically controlled by phosphorylation. However, very little is known about the regulation of separase activity in higher eukaryotes, largely due to the complexity of its regulation and the lack of good tools to study its activity in space and time. In this project, an innovative experimental setup will be established to allow the in-depth molecular characterization of separase activation in the multicellular organism C. elegans. An in vivo biosensor will be developed to accurately measure the rate of cohesin cleavage in each dividing cell of the developing organism. This unique tool will allow us to screen for kinases and phosphatases that control separase activity in multicellular organisms. The identified kinases and phosphatases will be purified and their enzymatic activity will be tested towards C. elegans separase SEP-1 and cohesin subunit SCC-1. To analyze phosphoregulation of SEP-1 or SCC-1, phosphomutants will be developed and functionally characterized in vivo using the separase biosensor. In addition, we will perform biochemical reconstitution of cohesin cleavage in vitro, which will allow us to further test our in vivo findings and analyze the regulation of separase activity at the molecular level. Together, these approaches will provide a powerful system to gain novel insights into sister-chromatid separation in mitosis.