Neuroscience 2020

Working memory (the ability to hold some information in mind for a few seconds, and to manipulate that information) and decision-making (committing to one out of multiple possible choices) are fundamental to our daily lives. But how brain circuits and dynamics create those computations is still unknown. Using behaving rodents and deploying the latest research advances in neuroscience, we are a team of 7 different labs, working together in an attempt to crack the problem. We aim to understand how populations of neurons in multiple regions of the brain interact with each other in order to jointly produce what we experience as working memory and decision-making.

Learning Objectives:

Identifying the diversity of neuronal cell types of the nervous system is one of the main objectives of the BRAIN Initiative, with the vision that distinct neuronal identities will allow for their selective manipulation and reveal their functional contributions in health and disease. However, identifying all the cell types of the mammalian brain is not a trivial undertaking and is hindered by the fact that a satisfactory definition of neuronal cell type is nonexistent, with terms like “class”, “subclass”, “type”, and “subtype” often used interchangeably without proper definition. As part of the BICCN (BRAIN Initiative Cell Census Network), we have developed a practical, robust, and multi-pronged workflow to systematically classify projection neuron types based on their precise anatomical location, long-range connectivity, input/output organization, and detailed neuronal morphology. These connectivity-based neuronal classification strategies can be integrated or validated with other cell type specific information, such as molecular identities (mRNA expression, epigenomics, or genetic labeling), electrophysiological properties, and functional specificities. In this panel presentation, Dr. Hong-Wei Dong will present our recent progress of constructing a whole mouse brain wiring diagram, which lays an anatomical frame for characterizing neuronal cell types. Dr. Giorgio Ascoli will present our progress of developing informatics tools to reconstruct, analyze, and present cell type-specific morphology and a novel shotgun projectome strategy. Dr. Byungkook Lim will present our effort of developing cutting edge tracing technologies to dissect cell type- and projection- specific neural circuits.

Learning Objectives:

1. Constructing a whole-brain wiring diagram

2. Anatomical characterization of neuron cell types based on anatomical locations, projection targets, and neuronal morphology

3. Genetically dissect cell type-specific neural circuits

Learning is often an emotional process. Emotional stimuli with different valences, such as threat and reward, can transform an otherwise neutral sensory input into one that can trigger distinct behavioral outcomes, such as defensive versus reward-seeking behaviors. The amygdala, a structure deep in the brain, plays an essential role in such emotional learning. By integrating sensory information with signals of negative or positive valences, the neuronal circuits within the amygdala are modified during emotional learning. However, how individual emotional stimuli differentially modify the amygdala to mediate the defined, sometimes opposite behavioral outcomes is largely unknown. The overarching goal of our collaborative project is to elucidate the logics and mechanisms underlying differential emotional learning at the level of neuronal circuits. A crucial step forward is to determine the activities of individual neurons within discrete amygdala circuits before, during, and after learning tasks. However, this goal is challenging to achieve. First, the amygdala is buried deep within the brain, making it difficult to access by imaging methods, such as calcium imaging, which is the technique of choice for interrogating neuronal activities with cellular resolution over large neuronal populations. Second, stress and reward signals are in part encoded as neuromodulatory activities, which have been difficult to measure especially in behaving animals. Adding to the difficulty, the identity of individual amygdala circuits, as well as where each circuit receives inputs and where it sends outputs, are only partially understood.

In on-going progress, the team of three co-PIs meets these challenges by integrating the cutting-edge, complementary technological advances from each laboratory. In defined emotional learning behavioral paradigms we can now image calcium as a proxy for neuronal activity in the amygdala by performing two-photon imaging via a tiny GRIN lens (Φ~0.5 mm), which offers optical access to deep brain structures even in behaving animals. To establish a readout for stress/reward-induced neuromodulatory signals, we have successfully imaged the activity dynamics of the cAMP/protein kinase A (PKA) signaling pathway, which is a common downstream pathway for many neuromodulators, including norepinephrine and dopamine, by combining novel genetically-encoded sensors with two-photon fluorescence lifetime imaging microscopy and GRIN lenses. In conjunction, we are performing whole-brain high-resolution imaging and computation-based anatomical circuitry analyses to dissect novel functional subdivisions of the amygdala and identify the input-output of each subdivision with cell-type specificity. Looking forward, we will integrate the advances in these three aspects to systematically map circuits and determine how neurons in each amygdala circuit are recruited by and contribute to the generation of valence-specific behaviors.

Learning Objectives:

1. Introducing the important roles of amygdala in emotional learning and the challenges in understanding it.

2. Presenting the ongoing progresses in meeting the challenges towards dissecting the circuit logics of amygdala-based emotional learning.

Learning is often an emotional process. Emotional stimuli with different valences, such as threat and reward, can transform an otherwise neutral sensory input into one that can trigger distinct behavioral outcomes, such as defensive versus reward-seeking behaviors. The amygdala, a structure deep in the brain, plays an essential role in such emotional learning. By integrating sensory information with signals of negative or positive valences, the neuronal circuits within the amygdala are modified during emotional learning. However, how individual emotional stimuli differentially modify the amygdala to mediate the defined, sometimes opposite behavioral outcomes is largely unknown. The overarching goal of our collaborative project is to elucidate the logics and mechanisms underlying differential emotional learning at the level of neuronal circuits. A crucial step forward is to determine the activities of individual neurons within discrete amygdala circuits before, during, and after learning tasks. However, this goal is challenging to achieve. First, the amygdala is buried deep within the brain, making it difficult to access by imaging methods, such as calcium imaging, which is the technique of choice for interrogating neuronal activities with cellular resolution over large neuronal populations. Second, stress and reward signals are in part encoded as neuromodulatory activities, which have been difficult to measure especially in behaving animals. Adding to the difficulty, the identity of individual amygdala circuits, as well as where each circuit receives inputs and where it sends outputs, are only partially understood.

In on-going progress, the team of three co-PIs meets these challenges by integrating the cutting-edge, complementary technological advances from each laboratory. In defined emotional learning behavioral paradigms we can now image calcium as a proxy for neuronal activity in the amygdala by performing two-photon imaging via a tiny GRIN lens (Φ~0.5 mm), which offers optical access to deep brain structures even in behaving animals. To establish a readout for stress/reward-induced neuromodulatory signals, we have successfully imaged the activity dynamics of the cAMP/protein kinase A (PKA) signaling pathway, which is a common downstream pathway for many neuromodulators, including norepinephrine and dopamine, by combining novel genetically-encoded sensors with two-photon fluorescence lifetime imaging microscopy and GRIN lenses. In conjunction, we are performing whole-brain high-resolution imaging and computation-based anatomical circuitry analyses to dissect novel functional subdivisions of the amygdala and identify the input-output of each subdivision with cell-type specificity. Looking forward, we will integrate the advances in these three aspects to systematically map circuits and determine how neurons in each amygdala circuit are recruited by and contribute to the generation of valence-specific behaviors.

Learning Objectives:

1. Introducing the important roles of amygdala in emotional learning and the challenges in understanding it.

2. Presenting the ongoing progresses in meeting the challenges towards dissecting the circuit logics of amygdala-based emotional learning.

Episodic memories are essential for human cognition, but the underlying neural mechanisms remain poorly understood. We utilize the opportunity to record in-vivo from human single neurons simultaneously in multiple brain areas in patients undergoing treatment for drug resistant epilepsy to study the underlying mechanisms. Supported by the BRAIN initiative, we formed a consortium among four institutions (Cedars-Sinai/Caltech, Johns Hopkins, U Toronto, and Children’s/Harvard) to maximize the use of these rare and precious opportunities to further our understanding of human memory. In this panel talk, we will highlight different aspects of our work that has become possible due to this highly interdisciplinary group of experimental and computational scientists, neurologists, neurosurgeons, and ethics experts.

We are developing a circuit-level understanding of human memory by utilizing invasive in-vivo recordings together with behavior, focal electrical stimulation, and computational modeling. We are investigating a putative circuit for human recognition memory that is composed of specific functional types of cells in the medial temporal lobe (MTL) and the medial frontal cortex (MFC). In the MTL, we are investigating the role of visually selective and memory selective cells, whereas in the MFC we are studying memory-choice cells that represent a putative readout of memories to support memory-based recognition and confidence decisions. Together, these results begin to provide a circuit-level understanding of human memory at a level of detail that is needed for the development of new treatments for memory disorders.

Approximately 300,000 people in the United States have a spinal cord injury with many of these individuals experiencing permanent motor and sensory deficits. For individuals with cervical spinal cord injury, restoration of arm and hand function is a top priority.  Our team is developing a bidirectional brain-computer interface (BCI) to address this priority.  Neural activity is recorded from the motor cortex and movement intention can be decoded and used as a control signal for a robotic arm and hand.  The relationship between neural activity and movement intention derived during attempted movements in someone with spinal cord injury is similar to that measured during overt movements in intact subjects.  The use of biomimetic decoding approaches enables robust performance approaching that of an able-bodied person without significant training or learning.  We have also shown that a biomimetic approach can be used to restore the sense of touch through microstimulation of electrodes implanted in the somatosensory cortex. Tactile sensations restored with the BCI feel like they originate from the participant’s own hand and can be graded in intensity. Sensations are evoked in a spatially organized manner as would be expected from neurophysiologic studies in able-bodied subjects.  Finally, we show that BCI performance on functional tasks involving object transport is improved substantially with the addition of sensory feedback, even for tasks that were performed well with only vision.

Learning Objectives:

1. Introduce the audience to the impact of spinal cord injury

2. Describe biomimetic principles that guide the development of bidirectional brain-computer interfaces

3. Present data to demonstrate significantly improved brain-computer interface performance when visual feedback is supplemented with restored tactile sensations

Mechanistic understanding of neural systems is daunting to achieve in large part due to the heterogeneity of the neuronal elements in both form and function and the complexity of the circuits formed by these elements. In this talk, we provide an update on our efforts to understand a neural system with a multi-faceted approach combining large-scale imaging of neuronal activity, reconstruction and analysis of network connectivity, and computational models of network function. We focus our work on the oculomotor neural integrator of the zebrafish, a circuit involved in the control of eye position that has been demonstrated to perform a mathematical integral of its inputs.  We identified through calcium imaging and targeted ablations neurons involved in planning, initiating, and maintaining changes in eye position. We used serial-section electron microscopy and crowd-sourced imaged analysis to reconstruct the circuit formed by these neurons and many of their synaptic partners. Computational analysis revealed a strongly-recurrent module in the circuit, consistent with theoretical predictions for the circuit mechanism of neural integration.  A whole-circuit, neural network model built from the underlying connectome reproduced the encoding of eye position seen experimentally across multiple species. We conclude with thoughts on how this approach can be extended to other domains.

Learning Objectives:

1. Understand how functional imaging, connectomic, and computational modeling approaches can be combined to investigate circuit mechanisms

2. Understand the role of recurrent excitation and mutual inhibition in generating and coordinating persistent neural activity

Brain machine interfaces (BMIs) aim to help patients with paralysis to use their recorded brain activity to control assistive devices.  BMI research requires the collaboration of neuroscientists, engineers, computational neuroscientists, and clinicians.  Generally motor cortex has been used as a source of control signals.  With our U01 BRAIN Initiative grant and other NIH grant support, we have examined a different area of cortex for BMI control, the posterior parietal cortex (PPC).  The PPC forms the intentions and plans to make movements.  Richard Andersen will discuss the finding that PPC represents many intended movement variables over essentially the entire body.  Cognitive variables include goals, cognitive strategies, action observation, and action semantics.  So many variables can be decoded from just a few hundred PPC neurons because they are represented in a clever, partially mixed structure.  Spencer Kellis will describe how intracortical microstimulation of primary somatosensory cortex produces cutaneous and proprioceptive sensations in the limb of a tetraplegic subject.  Being able to provide this naturalistic feedback is important for dexterous robotic control.  Tyson Aflalo will discuss a study in which microelectrode arrays are implanted in both PPC and primary motor cortex.  This study establishes different motor functions for these two areas that can complement brain control and maintain their basic functional differences even after many months of BMI sessions.  Charles Liu will describe the clinical aspects that guide every step of human BMI studies from recruitment, to surgery, to long term care of the participants.  He will highlight the importance of surgical techniques and the support of rehabilitation hospitals.

Learning Objectives:

1. Explain how mixed selectivity allows the posterior parietal cortex to represent so many action variables with a small number of neurons

2. Identify why restoring somatosensory feedback is important for brain-machine interfaces that control robotic hands.

3. Account for why the lack of functional restructuring of the posterior parietal cortex and motor cortex with brain-machine interface usage makes it important to select the right cortical areas for recording brain signals.

4. Enumerate ways that human brain-machine interface research benefits from having multidisciplinary research teams.

The public health burden of Treatment Resistant Depression (TRD) has prompted clinical trials of deep brain stimulation (DBS) that have, unfortunately, produced inconsistent outcomes. Potential gaps and opportunities include a need: (1) to better understand the neurocircuitry of the disease; (2) for precision DBS devices that can target brain networks in a clinically and physiologically validated manner; and (3) for greater insight into stimulation dose-response relationships. These needs are based on our overarching hypothesis that network-guided neuromodulation is critical for the efficacy of DBS in TRD. This project aims to address the unmet need of TRD patients by identifying brain networks critical for treating depression and to use next generation precision DBS with steering capability to engage these targeted networks and develop a new therapy for TRD. We use the Boston Scientific (BS) Vercise DBS system, which offers a segmented steerable lead with multiple independent current sources that allows true directional steering. Moreover, this system integrates stimulation field modeling (SFM) with MR tractography to predict network engagement. We use an innovative approach of targeting both subgenual cingulate (SGC) and ventral capsule/ventral striatum (VC/VS), which we term corticomesolimbic DBS. These targets are hubs in distinct yet partially overlapping depression networks and emerging basic science literature implicates them in bidirectional modulation of depression circuits. We also apply a paradigm-shifting approach using intracranial stereo-EEG (sEEG) subacutely after DBS implant to evaluate the clinical reliability of steering, SFMs, and tractography and to define and then target the networks mediating symptoms of depression. The impact of this proposal includes physiological validation of current “steering” DBS technology to target specific networks, insights into effects of stimulation parameters on network physiology, an improved understanding of the pathophysiology of depression, and, perhaps most importantly, a novel approach for treating TRD. This research will also pioneer a novel and high-yield test bed for DBS therapy development consistent with BRAIN priorities.

The neural basis of simple rhythmic and reflexive behaviors such as swimming and gill withdrawal have been successfully studied in nudibranchs and other gastropod molluscs because the brains of these animals contain large, individually identifiable neurons. However, it is a painstakingly slow process to identify neurons and determine their functions one neuron at a time, making it impossible to scale up to the whole brain. Furthermore, although nudibranchs are advantageous for electrophysiology, they are generally difficult to obtain because they need to be caught from the ocean. Unlike other nudibranchs, Berghia stephanieae, which was largely unstudied, can be bred in the lab, making it available in large numbers and amenable to developmental studies. In September 2018, we initiated our collaboration to use large scale transcriptomic, connectomic, and live imaging techniques to obtain a whole-brain view of the generation of complex, motivated behaviors. In a short time, our team has delineated behavioral actions in Berghia that are motivated by feeding or reproductive state, and by the presence of food or shelter. A cellular-level atlas of the brain is being constructed that represents the locations and phenotypic properties of all the neurons in the brain. This atlas will be aligned with a synaptic connectivity diagram obtained from reconstructed serial electron micrographic images. Finally, the locations of neuronal activity recorded with voltage-sensitive dyes will be aligned to the brain atlas to test models of how behavioral responses are produced. New mathematical models are being devised to help understand the complexities of such network activity. Plans are underway to use gene-editing technology to express genetically-encoded sensors and actuators in neurons to more precisely observe and control the activity of particular neuron types. The ultimate goal is to move from a small circuit approach to a whole-brain view of the neural control of behavior.

Learning Objectives:

1. Introduce the audience to identified neurons and molluscan neuroscience

2. Explain the approach to introducing a new species for laboratory research

3. Present results from team approach to studying the neural basis of behavior.

Although neuroscience has provided a great deal of information about how neurons work, the fundamental question of how neurons function together in a network to produce cognition has been difficult to address. Our BRAIN team, consisting of four collaborating laboratories at NYU, Columbia and Stanford, aims to make the first attempt to fully understand a cognitively important event, called memory replay, in terms of the detailed properties of the brain cells involved. We use cutting-edge largescale recording technologies to study and manipulate identified brain cell types in behaving animals, and we will construct the first full-scale computational of model of the brain area that produces the memory replay in which every cell is explicitly simulated. These powerful new approaches are likely to yield major insights into the principles by which the interactions of neurons give rise to cognitive function, with important implications for memory disorders.

Learning Objectives:

1. Define hippocampal ripple-related memory replay

2. Identify novel large-scale recording and modeling techniques used to study the cells and circuits generating ripples

3. Discuss advances in our understanding of the mechanisms and routing of ripple content

Brain function is remarkably reliable despite the imprecise performance of neurons and the continuous perturbations caused by aging, disease or injury. How does the brain succeed in producing stereotypic behaviors over long periods of time despite these perturbations? We are interested in studying the cellular and system mechanisms by which neuronal circuits are able to "self-tune" and adapt to perturbations.  We are using genetic manipulations to perturb brain circuits to ask two types of questions: (i) how do brain circuits adapt when a large percentage of their neurons are deleted or silenced?, (ii) how do brain circuits adapt when electrical noise is injected into the system?.  To study these questions we are focusing on the song circuits of birds because song is an extremely stereotypical behavior that can be rigorously measured.

Learning Objectives:

1. Define the concepts of brain resilience and behavioral stereotypy

2. Explain the experimental methods that can be used to investigate brain resilience

3. Explain how investigating mechanisms of brain resilience in animals could be useful to improve outcomes in humans after diseases such as stroke

We present here a framework to generate a realistic multiscale circuit model of the larval zebrafish brain – the multiscale virtual fish (MVF). The model will be based on algorithms inferred from behavioral assays and it will span spatial ranges across three levels: from the nanoscale at the synaptic level, to the microscale describing local circuits, to the macroscale brain-wide activity patterns distributed across many regions. The model will be constrained and validated by functional imaging and sparse connectomics of identified circuit elements.

Specifically, we focus on five ethologically relevant behaviors: the opto-motor response, phototaxis, rheotaxis, escape, and hunting. First, we extract the precise algorithms underlying each behavior and develop a version of the circuit model to understand their neural implementation. Second, we refine the model to account for multimodal integration and decision making, events that naturally happen when conflicting stimuli driving different behaviors are presented simultaneously. Third, we will examine how internal brain states, such as hunger or stress, influence and modulate the specific behaviors or behavioral interactions. Implementation of neurochemical modulation into the framework of the MVF will be achieved through simulation of highly conserved neuromodulatory neurotransmitter systems such as serotonin, acetylcholine, epinephrine, dopamine and oxytocin.

Learning Objectives:

1. Describe the advantages (and disadvantages) of the larval zebrafish as a model system for Neuroscience.
2. Discuss what specific questions in neuroscience can be answered by combining brain wide recordings of neural activity with brain wide connectomics in the same animal.

The locomotion of humans and other animals requires a seamless flow of information from sensory modalities all the way to the motor periphery. As such, locomotion is an excellent system for understanding how the brain modifies neural signals from one processing stage to the next, ultimately transforming sensory signals into a code that controls motor output and behavior. The presentations in this panel represent a sample of the work being conducted by a consortium of seven laboratories supported by a U19 grant through the Brain Initiative. Our research exploits a single, experimentally tractable model system, the fruit fly Drosophila melanogaster, in which we can easily study the functions of genetically identified cell classes in ethologically relevant behaviors. The team’s long-term goal is to develop a comprehensive theory of animal behavior that explicitly incorporates neural processes operating across hierarchical levels — from circuits that regulate the action of individual muscles to those that regulate behavioral sequences and decisions. The talks in this panel – all delivered by graduate students and post-docs - will highlight a range of experimental approaches that collectively constitutes a systematic attack on the neural basis of behavior that integrates vertically across phenomenological tiers.

Learning Objectives:

1. Introduce the audience to the concept of integrative approaches in neuroscience

2. Provide research examples spanning a range of phenomenological tiers, from local sensory-motor circuits to behavior

3. Illustrate the utility of Drosophila as a model system in neuroscience