Dmitry Rinberg, Ph.D. - US grants

? DESCRIPTION (provided by applicant): Two of the most fundamental questions of sensory neuroscience are: 1) how is stimulus information represented by the activity of neurons at different levels of information processing? And 2) what features of this activity are read by the higher brain areas to guide behavior? The first question has been the subject of a large body of work across different sensory modalities. To answer the second question, one needs to establish a causal link between neuronal activity and behavior. In many systems, fine spatiotemporal patterns of activity underlie the neural representation of information. In these systems, deciphering the salient neural code will require manipulating sets of neurons with fine spatiotemporal resolution while monitoring behavior. Ultimately, we would like to be able to manipulate arbitrary subsets of the neurons comprising neural circuits, and to do so with precise spatiotemporal control. Further, we would like to do so during natural behavior, to provide the ability to discern the salient spatiotemporal patterns that are behaviorally relevant from the broader cacophony of activity. Olfaction is emerging as an ideal system for investigating spatiotemporal coding. Recent studies have revealed that fine temporal scales are essential to olfactory information processing, both in terms of representation and the behavioral readout. Moreover, joint spatiotemporal activity perturbations affect this readout. Here, we propose to develop and apply within the context of early olfaction an advanced holographic optogenetic technology, which will enable flexible manipulation of the spatiotemporal firing patterns of dozens of neurons within a network, down to single-neuron resolution and several milliseconds of temporal resolution. The essential code underlying the evolutionarily crucial ability to detect and to discriminate odor stimuli is carried at a very fine spatiotemporal resolution and our goal with this new approach is to dissect which features of this code are behaviorally accessible. By combining state-of-the-art tools for patterned cellular-resolution optogenetic stimulation with olfactory- guided behavioral paradigms in head-fixed mice, the proposed research is expected to contribute a powerful new approach for dissecting which features of neural codes are behaviorally accessible.

DESCRIPTION (provided by applicant): Animals acquire information about the environment through their sensory systems. The first sensory stages play a critical role by formatting the representation of this information in an efficient manner that facilitates easy and flexible extraction by downstream areas. Some of the proposed computations in sensory areas of the brain include gain control, decorrelation, orthogonalization, redundancy reduction, and temporal coding. A diverse population of inhibitory interneurons in local brain areas is believed to play a crucial role in these computations. The mammalian olfactory system has multiple features, which make this system uniquely suited to test these ideas: the relative structural simplicity compared to other sensory systems, high relevance of olfaction to animal behavior, availability of modern genetic tools, and the combination of both sensory and state-dependent activity in a single network. Recent efforts of Koulakov and Rinberg groups helped to overcome some obstacles, which slowed down the progress in the field, such as experimental difficulties in controlling olfactory stimuli and insufficient connection between theory and experiment. Capitalizing on the advantages of the system and utilizing recently developed theoretical framework, new optogenetic methods for stimulus control, and multineuron recordings, we propose a collaborative project to test the basic principles of sensory processing in this system. We will investigate an emergence and a role of the temporal code in the representation of odorants in the OB, by studying the role of the interaction of the principal neurons of the olfactory bulb, mitral and tufted cells (MTCs), with the most abundant class of inhibitory interneurons, granule cells (GCs). We will test our recently proposed theory for the MTC-GC network called the Sparse Incomplete Representations (SIR) model. According to the SIR model, GCs form representations of odorants, while MTCs transmit to higher processing centers the errors of these representations. The SIR model predicts that, due to the balance between excitatory odorant-driven inputs from glomeruli and the inhibition from GCs, olfactory code carried by the MTCs becomes combinatorially and temporarily sparse. In the combinatorial form of sparseness, a small fraction of MTCs that receive excitatory inputs yield noticeable responses to odorants that persist through the sniff cycle. In the temporal sparseness, MTCs respond to odorants during a small temporal window within the sniff cycle. Complex temporal patterns of responses will be predicted and tested within this project. Both temporal and combinatorial sparseness are predicted to be dependent on the inputs from cortex, giving the olfactory code an enormous potential flexibility. To test the detailed predictions of the SIR model we propose the following specific aims (SAs): SA1: To investigate diversity of sparse temporal representations of odorants by the MTCs of the OB. Here we will investigate the temporal patterns of responses predicted by the SIR model and compare them with experimental data. We will also study the plasticity of the temporal code carried by the MTCs. SA 2: To investigate the temporal discreteness of sparse representations of odorants by the MTCs of the OB. Here we will study the synchronization of the transient events in MTC firing with ?-frequency oscillations and study the implications for the temporal coding in the OB. SA3: To study the context and state dependence of the sparse representations in the OB. In this Aim we will investigate how the olfactory code carried by the MTCs can be modified dynamically to better fit a particular task. Intellectual merit: Our project will help elucidate the general principles of information processing. We will demonstrate how sensory representations can be dynamically tuned to reflect particular tasks faced by the organism. We will show how networks can adapt to better detect novel features in the environment and disregard familiar ones. We will show how spatial information can be converted into temporal representations by the use of inhibitory interneurons. Broader impacts: The award will provide a unique cross-disciplinary environment for training of young neuroscientists. We expect that two postdoctoral fellows, specializing in theoretical and in experimental approaches, will receive training through this award. Undergraduate students at NYU and CSHL will be involved in the project as summer interns or as senior thesis writers. We will strive to involve minority graduate and undergraduate students, who will be recruited through dedicated programs at each institution. Health implications for the broader society: About 1-2% of people in North America experience a smell disorder. Since our sense of smell helps us enjoy life, may serve as a warning system alerting us of danger, such as spoiled food, fire, or a gas leak, and may be a sign of other health problems, any loss in the sense of smell can negatively affect our quality of life.

DESCRIPTION (provided by applicant): One of the basic questions of sensory neuroscience is how stimulus features are represented by changes in neuronal activity and how these changes affect perception. In olfaction, two critical features of the stimulus, odorant identity an concentration, can be independently perceived by animals. How the two are encoded independently is a long-standing unresolved question in the field. In mammals, odorants enter the nasal cavity with each respiration (sniff) and are transduced by a large repertoire of odorant receptors (ORs), each with a different response specificity and sensitivity. In the mouse, sensory neurons expressing each OR project to one or a few glomeruli in the olfactory bulb, from which each mitral/tufted cell (MTC) receives its main excitatory input. MTCs are the sole output neurons of the olfactory bulb and therefore play a central role in olfactory feature coding. The goal of the present proposal is to determine how MTCs encode odorant identity and concentration in awake animals. One general view in the field is that odorant identity can be expressed as a vector of OR (glomerular) activation across all inputs. It has been suggested that this odorant identity vector is transformed into a pattern of temporal delays (latencies) of activation across MTCs connected to different glomeruli1, 2. We propose that this can happen if odorant concentration in the nose increases gradually after inhalation onset with each sniff, and ORs with the highest affinities for a given odorant are activated earlier in the sniff cycle. If MTs inherit this temporal structuring of receptor activation, higher olfactory circuits could read thes latencies. One feature of this model is that the sequence of MTC activation provides a stable representation of odorant identity, independent of other factors such as sniff duration or concentration: changes in these factors would temporally scale the whole pattern of latencies without changing the sequence of activation (at least at the beginning of the inhalation). Consequently, a temporal shift of all MTC latencies would provide a representation for odor concentration independent of identity. Thus, both identity and concentration can be represented independently in time. Moreover, previous studies do not address whether the brain can or does read these latencies in behaving animals. The proposed research is methodologically innovative, in our opinion, because it achieves unprecedented control over the key components of olfaction: sniff monitoring, stimulus timing, individual receptor-type control, receptor-specifi M/T cell recording, awake/behaving animals. This research is conceptually innovative, in our opinion, because we 1) integrate pre-sensor physics with receptor-ligand interactions into a model which accounts for how odorant identity and concentration can be independently represented in the latencies of MTC responses, and 2) have developed an approach to test this model both in terms of encoding and readability in the same preparation, thus enabling particularly strong inferences regarding the representation and use of olfactory information in the nervous system.

Precise spike timing plays an important role in sensory encoding. Spikes may convey information about the time of events in the external world, as when an animal needs to escape quickly from a predator. Spike timing may carry information about other physical characteristics of an external world; for example, interaural time differences are informative about a location of a sound source. And in general, sensory information can be carried by temporal codes, which means that information is represented in spike trains at time scales that are faster than meaningful fluctuations of the external stimuli. However, these conclusions are based mostly on observed correlations between sensory stimuli and spikes. The question of how the timing of sensory signals relates to behavior remains elusive, mostly because appropriately precise tools have not been available to manipulate the timing of sensory representations. The current project will capitalize on a novel technique of two-photon holographic stimulation, which enables the perturbation of activity in many individual neurons on time scales of ~10 ms, which is relevant for sensory representation and behavioral readout. Application of this technique in three different sensory systems ? olfaction, vision, and audition ? in combination with novel statistical and theoretical approaches, will address fundamental questions about the role of the temporal structure of neural codes in behavior. Sensory information may be represented by multiple temporal features of the neural code, such as multi- neuronal synchrony, spike timing relative to other spikes, or to global neural dynamics, such as oscillations. However, the presence of the information alone is not enough to prove that these features can be used to evoke behavior. Aim 1 will examine which temporal features of sensory code can be used by animals to guide behavior by measuring which classes of temporal information animals are capable of discriminating. Aim 2 will provide complementary information by measuring the behavioral consequences of perturbing specific temporal aspects of neuronal responses to natural stimuli. Although the use of timing may differ between brain areas, the proposed comparative approach will reveal general principles of sensory coding and establish how neural networks adjust for specific computational demands of sensory information processing.

Project Summary To survive, organisms must both accurately represent stimuli in the outside world, and use that representation to generate beneficial behavioral actions. Historically, these two processes ? the mapping from stimuli to neural responses, and the mapping from neural activity to behavior ? have largely been treated separately. Of the two, the former has received the most attention. Often referred to as the ?neural coding problem,? its goal is to determine which features of neural activity carry information about external stimuli. This approach has led to many empirical and theoretical proposals about the spatial and temporal features of neural population activity, or ?neural codes,? that represent sensory information. However, there is still no consensus about the neural code for most sensory stimuli in most areas of the nervous system. The lack of consensus arises in part because, while it is established that certain features of neural population responses carry information about specific stimuli, it is unclear whether the brain uses (?reads?) the information in these features to form sensory perceptions. We have developed a theoretical framework, based on the intersection of coding and readout, to approach this problem. Experimentally informing this framework requires manipulating patterns of neuronal activity based on, and at the same spatiotemporal scale as, their natural firing patterns during sensory perception. This work must be done in behaving animals because it is essential to know which neural codes guide behavioral decisions. In the first phase of this project (funded by the BRAIN Initiative), we developed the technology necessary for realizing this goal. In the present proposal, we will extend our patterned neuronal stimulation technology and apply it to answer long-standing questions about neural coding and readout in the visual, olfactory, and auditory systems. We will pioneer the capacity to determine which neurons within a network are encoding behaviorally relevant information, and also to determine the extent to which temporal patterns of those neurons? activity are being used to guide behavior. Finally, we will study these neural coding principles across changes in behavioral state and during learning to determine how internal context and past experience shape coding and readout. The contributions of the proposed work will be three-fold. First, we will provide the neuroscience community with the tools needed to test theories of how neural populations encode and decode information throughout the brain. Second, we will reveal fundamental principles of spatiotemporal neural coding and readout in the visual, olfactory, and auditory systems of behaving animals. And third, our unifying theoretical framework for cracking neural codes will allow the broader neuroscience community to resolve ongoing debates regarding neural coding that have been previously stalemated by considering only half of the coding/readout problem.

Project Summary (Administrative Core) The Administrative Core will serve as the central hub of this geographically distributed U19 team. It will monitor progress, promote exchange of results and technology by organizing monthly and quarterly team video- conferences and annual face-to-face meetings. To provide oversight and critical feedback, it will organize annual meetings with the External Advisory Board. The Core will establish a pipeline for shipping sequencing samples from Projects 1, 3, and 4 to the Genome Technology Center of NYU School of Medicine. The Administrative Core will also coordinate the transfer of sequencing data from NYU School of Medicine back to the projects and Data Science Core. The Core will provide financial and regulatory management to ensure compliance with federal and institutional rules and budgetary oversight. In addition, it will provide support for scientific communications and implement and support a program web site to disseminate information, tools, and data to the broader neuroscience community.

Project Summary (Overall: Cracking the Olfactory Code) Sensation drives perception, which informs decisions and actions. Olfaction is the main sense used by most animals to interact with the environment. However, olfaction remains shrouded in mystery ? we do not know which molecular odorant features matter to the olfactory system and which do not, how information about these features is recombined to create holistic odor representations within the brain, or how those representations relate to perception. As a consequence, we lack an empirical understanding of the core transformations taking place at each stage of olfactory processing, which ultimately lead to perception and behavior. In addition, we lack a clear theoretical framework for understanding how a stimulus space that is both discrete and high-dimensional yields a perceptual space that is continuous and low dimensional. Because the olfactory system is ?shallow? ? meaning that within two synapses information about complete odor objects is abstracted and generalized ? understanding this specific circuit will also afford general insight both into architecturally-related allocortical brain regions critical to behavior (e.g., cerebellum, hippocampus), and into cortical centers that play a key role in integrating diverse sources of information (e.g., prefrontal cortex, posterior parietal cortex). Here we propose to reveal the computational logic of olfaction by collecting the first system-wide dataset of neural and perceptual responses to a large, principled set of odorants, and by applying a unified statistical and theoretical approach to its interpretation. This project will convene research groups with expertise that spans neurobiology, and will leverage recent technical advances in molecular genetics, neural imaging, electrophysiology, opto- and chemogenetics, human psychophysics, and machine learning to interrogate all levels (from peripheral receptors to cortex to perceptual and behavioral output) of the olfactory system. Taken together, these experiments will establish a reference dataset that reveals the key transformations performed by the olfactory system, test a key unifying theory for olfaction, and create a community-wide resource that will prompt new theory and experiment. This work will also have wide-ranging implications for our general understanding of how sensory information is organized in the brain to facilitate adaptive action.

Project Summary Two of the most fundamental questions of sensory neuroscience are: 1) how is stimulus information represented by the activity of populations of neurons at different levels of information processing? and 2) what features of this activity are read at the next levels of neural processing to guide behavior? The first question has been the subject of a large body of work across different sensory stimuli. To answer the second question, one needs to establish a causal link between neuronal activity and behavior. In many systems sensory information is represented by complex spatiotemporal patterns of neuronal activity. Novel recording and stimulating technology will soon allow the precise temporal control of hundreds and thousands of individual neurons, however, conceptual approaches of finding relevance of different spatiotemporal features of neural code still lag behind. To develop a new approach we chose the mammalian olfaction as a model system, because odor stimuli evoke complex patterns of glomerular activity with spatial and temporal scales fully compatible with existing imaging and pattern stimulation technologies. In addition, the accessibility of the cells in the next processing level, the mitral/tufted cells which get input from olfactory glomeruli and transmit the signal to higher brain areas, allows a systematic study of encoding different features of neural activity with known behavioral relevance. We propose a novel approach to map spatiotemporal code features to neural and perceptual spaces. First, we substitute sensory-driven neural activation by artificial and fully parametrized optogenetic pattern stimulation. By varying the parameters of such stimulation and recording the behavioral outcomes of the stimulation, we will build a detailed empirically-validated mathematical model of the relevance of different features of neural activity. Then we will test this model for natural odor stimuli, and explore how these features are processed and encoded by the next level of processing. Successful execution of the project will produce the first (to our knowledge) causally validated model for behavioral relevance of a distributed neural code. It will shine light on long standing questions in olfactory processing, approaching the olfactory code from the perspective of its behavioral relevance. The proposed approach can be further applied to different neural systems using multi-neuronal recording and stimulation techniques.

Summary Neuroscience has an essential requirement for large-scale neural recording and perturbation technologies for the understanding of brain function, as well as in the diagnosis and treatment of neurological disorders. At present, a large gap exists between the localized optical microscopy studies looking at fast neuronal activities at single cell resolution level and the whole-brain observations of slow hemodynamics and brain metabolism provided by the macroscopic imaging modalities. The proposed three- year project is aimed at developing a highly synergistic triple-modality platform combining acoustic stimulation with volumetric optoacoustic and planar fluorescence imaging to volumetrically monitor and perturb the activity of large, distributed neuronal populations with unprecedented spatiotemporal resolution. This goal will be accomplished by constructing a bi-directional interface based on a spherical matrix array transducer capable of both recording real-time three-dimensional optoacoustic tomographic data and acoustic phased array beam steering and holography for ultrasonic neural stimulation. The high temporal resolution in these volumetric recordings will make it possible to directly and indirectly track neural activity, with novel near- infrared calcium (Ca2+) sensors and intrinsic hemodynamic contrast, respectively. The resulting scanner will simultaneously record activity from large fields of view in scattering brains, including deep subcortical structures inaccessible by any light microscope. The plan of action includes screening of several potential candidates for Ca2+ imaging, including genetic and chemigenetic sensors. System validation will be performed in vivo in mice, aiming at establishing sensitivity and spatiotemporal resolution metrics in detecting Ca2+ relevant for sensory-based decision making. Finally, the complete system will be used to probe the link between neural activity and behavior by systematically characterizing the effects of image-targeted US perturbation in mice performing olfactory-guided tasks. In contrast to purely optical techniques, the proposed method is tailored for non-invasive deep brain observations and manipulations and is ideal for large fields of view and columnar-scale mesoscopic resolutions.