Recent advances in optical microscopy have enabled biological imaging beyond the diffraction limit at nanometer resolution. A general feature of most of the techniques based on photoactivated localization microscopy (PALM) or stochastic optical reconstruction microscopy (STORM) has been the use of thin biological samples in combination with total internal reflection, thus limiting the imaging depth to a fraction of an optical wavelength. However, to study whole cells or organelles that are typically up to 15 microm deep into the cell, the extension of these methods to a three-dimensional (3D) super resolution technique is required. Here, we report an advance in optical microscopy that enables imaging of protein distributions in cells with a lateral localization precision better than 50 nm at multiple imaging planes deep in biological samples. The approach is based on combining the lateral super resolution provided by PALM with two-photon temporal focusing that provides optical sectioning. We have generated super-resolution images over an axial range of approximately 10 microm in both mitochondrially labeled fixed cells, and in the membranes of living S2 Drosophila cells.

The representation of acoustic stimuli in the brainstem forms the basis for higher auditory processing. While some characteristics of this representation (e.g. tuning curve) are widely accepted, it remains a challenge to predict the firing rate at high temporal resolution in response to complex stimuli. In this study we explore models for in vivo, single cell responses in the medial nucleus of the trapezoid body (MNTB) under complex sound stimulation. We estimate a family of models, the multilinear models, encompassing the classical spectrotemporal receptive field and allowing arbitrary input-nonlinearities and certain multiplicative interactions between sound energy and its short-term auditory context. We compare these to models of more traditional type, and also evaluate their performance under various stimulus representations. Using the context model, 75% of the explainable variance could be predicted based on a cochlear-like, gamma-tone stimulus representation. The presence of multiplicative contextual interactions strongly reduces certain inhibitory/suppressive regions of the linear kernels, suggesting an underlying nonlinear mechanism, e.g. cochlear or synaptic suppression, as the source of the suppression in MNTB neuronal responses. In conclusion, the context model provides a rich and still interpretable extension over many previous phenomenological models for modeling responses in the auditory brainstem at submillisecond resolution.

Electrophysiology is the most used approach for the collection of functional data in basic and translational neuroscience, but it is typically limited to either intracellular or extracellular recordings. The integration of multiple physiological modalities for the routine acquisition of multimodal data with microelectrodes could be useful for biomedical applications, yet this has been challenging owing to incompatibilities of fabrication methods. Here, we present a suite of glass pipettes with integrated microelectrodes for the simultaneous acquisition of multimodal intracellular and extracellular information in vivo, electrochemistry assessments, and optogenetic perturbations of neural activity. We used the integrated devices to acquire multimodal signals from the CA1 region of the hippocampus in mice and rats, and show that these data can serve as ground-truth validation for the performance of spike-sorting algorithms. The microdevices are applicable for basic and translational neurobiology, and for the development of next-generation brain-machine interfaces.

The central nucleus of the amygdala (CEA) is a brain region that integrates external and internal sensory information and executes innate and adaptive behaviors through distinct output pathways. Despite its complex functions, the diversity of molecularly defined neuronal types in the CEA and their contributions to major axonal projection targets have not been examined systematically. Here, we performed single-cell RNA-sequencing (scRNA-Seq) to classify molecularly defined cell types in the CEA and identified marker-genes to map the location of these neuronal types using expansion assisted iterative fluorescence in situ hybridization (EASI-FISH). We developed new methods to integrate EASI-FISH with 5-plex retrograde axonal labeling to determine the spatial, morphological, and connectivity properties of ∼30,000 molecularly defined CEA neurons. Our study revealed spatio-molecular organization of the CEA, with medial and lateral CEA associated with distinct cell families. We also found a long-range axon projection network from the CEA, where target regions receive inputs from multiple molecularly defined cell types. Axon collateralization was found primarily among projections to hindbrain targets, which are distinct from forebrain projections. This resource reports marker-gene combinations for molecularly defined cell types and axon-projection types, which will be useful for selective interrogation of these neuronal populations to study their contributions to the diverse functions of the CEA.

In this issue of Neuron, Gurnani and Silver (2021) report that activity across Golgi cells, a major type of inhibitory interneuron in the cerebellar cortex, is multidimensional and modulated by behavior. These results suggest multiple functions for inhibition in cerebellar computations.

In this work, we address the problem of pose detection and tracking of multiple individuals for the study of behaviour in insects and animals. Using a Deep Neural Network architecture, precise detection and association of the body parts can be performed. The models are learned based on user-annotated training videos, which gives flexibility to the approach. This is illustrated on two different animals: honeybees and mice, where very good performance in part recognition and association are observed despite the presence of multiple interacting individuals.

The mechanisms linking sensation and action during learning are poorly understood. Layer 2/3 neurons in the motor cortex might participate in sensorimotor integration and learning; they receive input from sensory cortex and excite deep layer neurons, which control movement. Here we imaged activity in the same set of layer 2/3 neurons in the motor cortex over weeks, while mice learned to detect objects with their whiskers and report detection with licking. Spatially intermingled neurons represented sensory (touch) and motor behaviours (whisker movements and licking). With learning, the population-level representation of task-related licking strengthened. In trained mice, population-level representations were redundant and stable, despite dynamism of single-neuron representations. The activity of a subpopulation of neurons was consistent with touch driving licking behaviour. Our results suggest that ensembles of motor cortex neurons couple sensory input to multiple, related motor programs during learning.

In many machine learning applications, labeling every instance of data is burdensome. Multiple Instance Learning (MIL), in which training data is provided in the form of labeled bags rather than labeled instances, is one approach for a more relaxed form of supervised learning. Though much progress has been made in analyzing MIL problems, existing work considers bags that have a finite number of instances. In this paper we argue that in many applications of MIL (e.g. image, audio, etc.) the bags are better modeled as low dimensional manifolds in high dimensional feature space. We show that the geometric structure of such manifold bags affects PAC-learnability. We discuss how a learning algorithm that is designed for finite sized bags can be adapted to learn from manifold bags. Furthermore, we propose a simple heuristic that reduces the memory requirements of such algorithms. Our experiments on real-world data validate our analysis and show that our approach works well.

How neurons form synapses within specific layers remains poorly understood. In the Drosophila medulla, neurons target to discrete layers in a precise fashion. Here we demonstrate that the targeting of L3 neurons to a specific layer occurs in two steps. Initially, L3 growth cones project to a common domain in the outer medulla, overlapping with the growth cones of other neurons destined for a different layer through the redundant functions of N-Cadherin (CadN) and Semaphorin-1a (Sema-1a). CadN mediates adhesion within the domain and Sema-1a mediates repulsion through Plexin A (PlexA) expressed in an adjacent region. Subsequently, L3 growth cones segregate from the domain into their target layer in part through Sema-1a/PlexA-dependent remodeling. Together, our results and recent studies argue that the early medulla is organized into common domains, comprising processes bound for different layers, and that discrete layers later emerge through successive interactions between processes within domains and developing layers.

Physical traces underlying simple memories can be confined to a single group of cells in the brain. In the fly Drosophila melanogaster, the Kenyon cells of the mushroom bodies house traces for both appetitive and aversive odor memories. The adenylate cyclase protein, Rutabaga, has been shown to mediate both traces. Here, we show that, for appetitive learning, another group of cells can additionally accommodate a Rutabaga-dependent memory trace. Localized expression of rutabaga in either projection neurons, the first-order olfactory interneurons, or in Kenyon cells, the second-order interneurons, is sufficient for rescuing the mutant defect in appetitive short-term memory. Thus, appetitive learning may induce multiple memory traces in the first- and second-order olfactory interneurons using the same plasticity mechanism. In contrast, aversive odor memory of rutabaga is rescued selectively in the Kenyon cells, but not in the projection neurons. This difference in the organization of memory traces is consistent with the internal representation of reward and punishment.