Proprioception, the sense of body position and movement, is essential for effective motor control. Because proprioceptive sensory neurons are embedded in complex and dynamic tissues, it has been challenging to understand how they sense and encode mechanical stimuli. Here, we find that proprioceptor neurons in the Drosophila femur are organized into functional groups that are biomechanically specialized to detect features of tibia joint kinematics. The dendrites of position and vibration-tuned proprioceptors receive distinct mechanical signals via the arculum, an elegant mechanical structure that decomposes movement of the tibia joint into orthogonal components. The cell bodies of position-tuned proprioceptors form a goniotopic map of joint angle, whereas the dendrites of vibration-tuned proprioceptors form a tonotopic map of tibia vibration frequency. Our findings reveal biomechanical mechanisms that underlie proprioceptor feature selectivity and identify common organizational principles between proprioception and other topographically organized sensory systems.

The posterior parietal cortex (PPC) exhibits choice-selective activity during perceptual decision-making tasks. However, it is not known how this selective activity arises from the underlying synaptic connectivity. Here, we combined virtual reality behavior, two-photon calcium imaging, high throughput electron microscopy, and circuit modeling to analyze how synaptic connectivity between neurons in PPC relates to their selective activity. We found that excitatory pyramidal neurons preferentially target inhibitory interneurons with the same selectivity. In turn, inhibitory interneurons preferentially target pyramidal neurons with opposite selectivity, forming an opponent inhibition motif. Using circuit models, we show that opponent inhibition amplifies selective inputs and induces competition between neural populations with opposite selectivity, thereby improving the encoding of trial-type information. These results provide evidence for how synaptic connectivity in cortical circuits supports a learned decision-making task.

Neuron-glia interactions play a critical role in the regulation of synapse formation and circuit assembly. Here we demonstrate that canonical Sonic hedgehog (Shh) pathway signaling in cortical astrocytes acts to coordinate layer-specific synaptic connectivity. We show that the Shh receptor Ptch1 is expressed by cortical astrocytes during development and that Shh signaling is necessary and sufficient to promote the expression of genes involved in regulating synaptic development and layer-enriched astrocyte molecular identity. Loss of Shh in layer V neurons reduces astrocyte complexity and coverage by astrocytic processes in tripartite synapses; conversely, cell-autonomous activation of Shh signaling in astrocytes promotes cortical excitatory synapse formation. Furthermore, Shh-dependent genes Lrig1 and Sparc distinctively contribute to astrocyte morphology and synapse formation. Together, these results suggest that Shh secreted from deep-layer cortical neurons acts to specialize the molecular and functional features of astrocytes during development to shape circuit assembly and function.

To understand how the cerebellar cortex transforms mossy fiber (MF) inputs into Purkinje cell (PC) outputs, it is vital to delineate the elements of this circuit. Candelabrum cells (CCs) are enigmatic interneurons of the cerebellar cortex that have been identified based on their morphology, but their electrophysiological properties, synaptic connections and function remain unknown. Here, we clarify these properties using electrophysiology, single-nucleus RNA sequencing, in situ hybridization and serial electron microscopy in mice. We find that CCs are the most abundant PC layer interneuron. They are GABAergic, molecularly distinct and present in all cerebellar lobules. Their high resistance renders CC firing highly sensitive to synaptic inputs. CCs are excited by MFs and granule cells and are strongly inhibited by PCs. CCs in turn primarily inhibit molecular layer interneurons, which leads to PC disinhibition. Thus, inputs, outputs and local signals converge onto CCs to allow them to assume a unique role in controlling cerebellar output.

The cerebellum is thought to help detect and correct errors between intended and executed commands1,2 and is critical for social behaviours, cognition and emotion3-6. Computations for motor control must be performed quickly to correct errors in real time and should be sensitive to small differences between patterns for fine error correction while being resilient to noise7. Influential theories of cerebellar information processing have largely assumed random network connectivity, which increases the encoding capacity of the network's first layer8-13. However, maximizing encoding capacity reduces the resilience to noise7. To understand how neuronal circuits address this fundamental trade-off, we mapped the feedforward connectivity in the mouse cerebellar cortex using automated large-scale transmission electron microscopy and convolutional neural network-based image segmentation. We found that both the input and output layers of the circuit exhibit redundant and selective connectivity motifs, which contrast with prevailing models. Numerical simulations suggest that these redundant, non-random connectivity motifs increase the resilience to noise at a negligible cost to the overall encoding capacity. This work reveals how neuronal network structure can support a trade-off between encoding capacity and redundancy, unveiling principles of biological network architecture with implications for the design of artificial neural networks.

When an animal moves through the world, its brain receives a stream of information about the body's translational velocity from motor commands and sensory feedback signals. These incoming signals are referenced to the body, but ultimately, they must be transformed into world-centric coordinates for navigation1,2. Here we show that this computation occurs in the fan-shaped body in the brain of Drosophila melanogaster. We identify two cell types, PFNd and PFNv3-5, that conjunctively encode translational velocity and heading as a fly walks. In these cells, velocity signals are acquired from locomotor brain regions6 and are multiplied with heading signals from the compass system. PFNd neurons prefer forward-ipsilateral movement, whereas PFNv neurons prefer backward-contralateral movement, and perturbing PFNd neurons disrupts idiothetic path integration in walking flies7. Downstream, PFNd and PFNv neurons converge onto hΔB neurons, with a connectivity pattern that pools together heading and translation direction combinations corresponding to the same movement in world-centric space. This network motif effectively performs a rotation of the brain's representation of body-centric translational velocity according to the current heading direction. Consistent with our predictions, we observe that hΔB neurons form a representation of translational velocity in world-centric coordinates. By integrating this representation over time, it should be possible for the brain to form a working memory of the path travelled through the environment8-10.

The study of neural circuits requires the reconstruction of neurons and the identification of synaptic connections between them. To scale the reconstruction to the size of whole-brain datasets, semi-automatic methods are needed to solve those tasks. Here, we present an automatic method for synaptic partner identification in insect brains, which uses convolutional neural networks to identify post-synaptic sites and their pre-synaptic partners. The networks can be trained from human generated point annotations alone and require only simple post-processing to obtain final predictions. We used our method to extract 244 million putative synaptic partners in the fifty-teravoxel full adult fly brain (FAFB) electron microscopy (EM) dataset and evaluated its accuracy on 146,643 synapses from 702 neurons with a total cable length of 312 mm in four different brain regions. The predicted synaptic connections can be used together with a neuron segmentation to infer a connectivity graph with high accuracy: between 92% and 96% of edges linking connected neurons are correctly classified as weakly connected (less than five synapses) and strongly connected (at least five synapses). Our synaptic partner predictions for the FAFB dataset are publicly available, together with a query library allowing automatic retrieval of up- and downstream neurons.

The blood-brain barrier (BBB) is critical for protecting the brain and maintaining neuronal homeostasis. Although the BBB is a unique feature of the central nervous system (CNS) vasculature, not all brain regions have the same degree of impermeability. Differences in BBB permeability are important for controlling the local extracellular environment of specific brain regions to regulate the function and plasticity of particular neural circuits. However, how BBB heterogeneity occurs is poorly understood. Here, we demonstrate how regional specialization of the BBB is achieved. With unbiased cell profiling in small, defined brain regions, we compare the median eminence, which has a naturally leaky BBB, with the cortex, which has an impermeable BBB. We identify hundreds of molecular differences in endothelial cells (ECs) and demonstrate the existence of differences in perivascular astrocytes and pericytes in these regions, finding 3 previously unknown subtypes of astrocytes and several key differences in pericytes. By serial electron microscopy reconstruction and a novel, aqueous-based tissue clearing imaging method, we further reveal previously unknown anatomical specializations of these perivascular cells and their unique physical interactions with neighboring ECs. Finally, we identify ligand-receptor pairs between ECs and perivascular cells that may regulate regional BBB integrity in ECs. Using a bioinformatic approach we identified 26 and 26 ligand-receptor pairs underlying EC-pericyte and EC-astrocyte interactions, respectively. Our results demonstrate that differences in ECs, together with region-specific physical and molecular interactions with local perivascular cells, contribute to BBB functional heterogeneity. These regional cell inventories serve as a platform for further investigation of the dynamic and heterogeneous nature of the BBB in other brain regions. Identification of local BBB specializations provides insight into the function of different brain regions and will permit the development of region-specific drug delivery in the CNS.

To investigate circuit mechanisms underlying locomotor behavior, we used serial-section electron microscopy (EM) to acquire a synapse-resolution dataset containing the ventral nerve cord (VNC) of an adult female Drosophila melanogaster. To generate this dataset, we developed GridTape, a technology that combines automated serial-section collection with automated high-throughput transmission EM. Using this dataset, we studied neuronal networks that control leg and wing movements by reconstructing all 507 motor neurons that control the limbs. We show that a specific class of leg sensory neurons synapses directly onto motor neurons with the largest-caliber axons on both sides of the body, representing a unique pathway for fast limb control. We provide open access to the dataset and reconstructions registered to a standard atlas to permit matching of cells between EM and light microscopy data. We also provide GridTape instrumentation designs and software to make large-scale EM more accessible and affordable to the scientific community.

We present a simple, yet effective, auxiliary learning task for the problem of neuron segmentation in electron microscopy volumes. The auxiliary task consists of the prediction of Local Shape Descriptors (LSDs), which we combine with conventional voxel-wise direct neighbor affinities for neuron boundary detection. The shape descriptors are designed to capture local statistics about the neuron to be segmented, such as diameter, elongation, and direction. On a large study comparing several existing methods across various specimen, imaging techniques, and resolutions, we find that auxiliary learning of LSDs consistently increases segmentation accuracy of affinity-based methods over a range of metrics. Furthermore, the addition of LSDs promotes affinitybased segmentation methods to be on par with the current state of the art for neuron segmentation (Flood-Filling Networks, FFN), while being two orders of magnitudes more efficient—a critical requirement for the processing of future petabyte-sized datasets. Implementations of the new auxiliary learning task, network architectures, training, prediction, and evaluation code, as well as the datasets used in this study are publicly available as a benchmark for future method contributions.

To effectively control their bodies, animals rely on feedback from proprioceptive mechanosensory neurons. In the Drosophila leg, different proprioceptor subtypes monitor joint position, movement direction, and vibration. Here, we investigate how these diverse sensory signals are integrated by central proprioceptive circuits. We find that signals for leg joint position and directional movement converge in second-order neurons, revealing pathways for local feedback control of leg posture. Distinct populations of second-order neurons integrate tibia vibration signals across pairs of legs, suggesting a role in detecting external substrate vibration. In each pathway, the flow of sensory information is dynamically gated and sculpted by inhibition. Overall, our results reveal parallel pathways for processing of internal and external mechanosensory signals, which we propose mediate feedback control of leg movement and vibration sensing, respectively. The existence of a functional connectivity map also provides a resource for interpreting connectomic reconstruction of neural circuits for leg proprioception.

Imaging neuronal networks provides a foundation for understanding the nervous system, but resolving dense nanometer-scale structures over large volumes remains challenging for light microscopy (LM) and electron microscopy (EM). Here we show that X-ray holographic nano-tomography (XNH) can image millimeter-scale volumes with sub-100-nm resolution, enabling reconstruction of dense wiring in Drosophila melanogaster and mouse nervous tissue. We performed correlative XNH and EM to reconstruct hundreds of cortical pyramidal cells and show that more superficial cells receive stronger synaptic inhibition on their apical dendrites. By combining multiple XNH scans, we imaged an adult Drosophila leg with sufficient resolution to comprehensively catalog mechanosensory neurons and trace individual motor axons from muscles to the central nervous system. To accelerate neuronal reconstructions, we trained a convolutional neural network to automatically segment neurons from XNH volumes. Thus, XNH bridges a key gap between LM and EM, providing a new avenue for neural circuit discovery.

Sensory experience remodels neural circuits in the early postnatal brain through mechanisms that remain to be elucidated. Applying a new method of ultrastructural analysis to the retinogeniculate circuit, we find that visual experience alters the number and structure of synapses between the retina and the thalamus. These changes require vision-dependent transcription of the receptor Fn14 in thalamic relay neurons and the induction of its ligand TWEAK in microglia. Fn14 functions to increase the number of bulbous spine-associated synapses at retinogeniculate connections, likely contributing to the strengthening of the circuit that occurs in response to visual experience. However, at retinogeniculate connections near TWEAK-expressing microglia, TWEAK signals via Fn14 to restrict the number of bulbous spines on relay neurons, leading to the elimination of a subset of connections. Thus, TWEAK and Fn14 represent an intercellular signaling axis through which microglia shape retinogeniculate connectivity in response to sensory experience.

Navigation can be directed toward distant targets represented within the brain’s spatial maps; alternatively, navigation can be directed toward objects in the local environment. Here we identify neurons in the Drosophila brain that integrate these two types of navigation drives. These neurons send axonal projections to the ventral nerve cord, and their activity predicts and influences steering during walking. Meanwhile, their dendrites integrate steering signals from the compass in the brain’s spatial memory center, as well as stimulus-directed steering signals from multimodal sensory pathways that bypass the compass. Using a computational model, we show how the specific connectivity of this network can generate steering behavior directed toward internal (remembered) goals, and we show how environmental cues can dynamically alter the balance of stimulus- and memory-directed steering. Our results suggest a framework where motor dynamics emerge from the integration of parallel feedback loops that drive steering toward internal versus external goals.

Electron microscopy (EM) is a powerful tool for circuit mapping, but identifying specific cell types in EM datasets remains a major challenge. Here we describe a technique enabling simultaneous visualization of multiple genetically identified neuronal populations so that synaptic interactions between them can be unequivocally defined. We present 15 adeno-associated virus constructs and 6 mouse reporter lines for multiplexed EM labeling in the mammalian nervous system. These reporters feature dAPEX2, which exhibits dramatically improved signal compared with previously described ascorbate peroxidases. By targeting this enhanced peroxidase to different subcellular compartments, multiple orthogonal reporters can be simultaneously visualized and distinguished under EM using a protocol compatible with existing EM pipelines. Proof-of-principle double and triple EM labeling experiments demonstrated synaptic connections between primary afferents, descending cortical inputs, and inhibitory interneurons in the spinal cord dorsal horn. Our multiplexed peroxidase-based EM labeling system should therefore greatly facilitate analysis of connectivity in the nervous system.

Endosomal sorting complex required for transport (ESCRT) complex proteins regulate biogenesis and release of extracellular vesicles (EVs), which enable cell-to-cell communication in the nervous system essential for development and adult function. We recently showed human loss-of-function (LOF) mutations in ESCRT-III member CHMP1A cause autosomal recessive microcephaly with pontocerebellar hypoplasia, but its mechanism was unclear. Here, we show Chmp1a is required for progenitor proliferation in mouse cortex and cerebellum and progenitor maintenance in human cerebral organoids. In Chmp1a null mice, this defect is associated with impaired sonic hedgehog (Shh) secretion and intraluminal vesicle (ILV) formation in multivesicular bodies (MVBs). Furthermore, we show CHMP1A is important for release of an EV subtype that contains AXL, RAB18, and TMED10 (ART) and SHH. Our findings show CHMP1A loss impairs secretion of SHH on ART-EVs, providing molecular mechanistic insights into the role of ESCRT proteins and EVs in the brain.

Neural network function can be shaped by varying the strength of synaptic connections. One way to achieve this is to vary connection structure. To investigate how structural variation among synaptic connections might affect neural computation, we examined primary afferent connections in the Drosophila olfactory system. We used large-scale serial section electron microscopy to reconstruct all the olfactory receptor neuron (ORN) axons that target a left-right pair of glomeruli, as well as all the projection neurons (PNs) postsynaptic to these ORNs. We found three variations in ORN→PN connectivity. First, we found a systematic co-variation in synapse number and PN dendrite size, suggesting total synaptic conductance is tuned to postsynaptic excitability. Second, we discovered that PNs receive more synapses from ipsilateral than contralateral ORNs, providing a structural basis for odor lateralization behavior. Finally, we found evidence of imprecision in ORN→PN connections that can diminish network performance.

High-resolution serial-section electron microscopy (ssEM) makes it possible to investigate the dense meshwork of axons, dendrites, and synapses that form neuronal circuits1. However, the imaging scale required to comprehensively reconstruct these structures is more than ten orders of magnitude smaller than the spatial extents occupied by networks of interconnected neurons2, some of which span nearly the entire brain. Difficulties in generating and handling data for large volumes at nanoscale resolution have thus restricted vertebrate studies to fragments of circuits. These efforts were recently transformed by advances in computing, sample handling, and imaging techniques1, but high-resolution examination of entire brains remains a challenge. Here, we present ssEM data for the complete brain of a larval zebrafish (Danio rerio) at 5.5 days post-fertilization. Our approach utilizes multiple rounds of targeted imaging at different scales to reduce acquisition time and data management requirements. The resulting dataset can be analysed to reconstruct neuronal processes, permitting us to survey all myelinated axons (the projectome). These reconstructions enable precise investigations of neuronal morphology, which reveal remarkable bilateral symmetry in myelinated reticulospinal and lateral line afferent axons. We further set the stage for whole-brain structure–function comparisons by co-registering functional reference atlases and in vivo two-photon fluorescence microscopy data from the same specimen. All obtained images and reconstructions are provided as an open-access resource.