Mechanosensory corpuscles detect transient touch and vibratory signals in the skin of vertebrates, enabling navigation, foraging, and precise manipulation of objects1. The corpuscle core comprises a terminal neurite of a mechanoreceptor afferent, the only known touch-sensing element within corpuscles, surrounded by terminal Schwann cells called lamellar cells (LCs)2–4. However, the precise corpuscular ultrastructure, and the role of LCs in touch detection are unknown. Here we used enhanced focused ion beam scanning electron microscopy and electron tomography to reveal the three-dimensional architecture of avian Meissner (Grandry) corpuscle5. We show that corpuscles contain a stack of LCs innervated by two afferents, which form large-area contacts with LCs. LCs form tether-like connections with the afferent membrane and contain dense core vesicles which release their content onto the afferent. Furthermore, by performing simultaneous electrophysiological recordings from both cell types, we show that mechanosensitive LCs use calcium influx to trigger action potential firing in the afferent and thus serve as physiological touch sensors in the skin. Our findings suggest a bi-cellular mechanism of touch detection, which comprises the afferent and LCs, likely enables corpuscles to encode the nuances of tactile stimuli.

Many animals rely on optic flow for navigation, using differences in eye image velocity to detect deviations from their intended direction of travel. However, asymmetries in image velocity between the eyes are often overshadowed by strong, symmetric translational optic flow during navigation. Yet, the brain efficiently extracts these asymmetries for course control. While optic flow sensitive-neurons have been found in many animal species, far less is known about the postsynaptic circuits that support such robust optic flow processing. In the fly Drosophila melanogaster, a group of neurons called the horizontal system (HS) are involved in course control during high-speed translation. To understand how HS cells facilitate robust optic flow processing, we identified central networks that connect to HS cells using full brain electron microscopy datasets. These networks comprise three layers: convergent inputs from different, optic flow-sensitive cells, a middle layer with reciprocal, and lateral inhibitory interactions among different interneuron classes, and divergent output projecting to both the ventral nerve cord (equivalent to the vertebrate spinal cord), and to deeper regions of the fly brain. By combining two-photon optical imaging to monitor free calcium dynamics, manipulating GABA receptors and modeling, we found that lateral disinhibition between brain hemispheres enhance the selectivity to rotational visual flow at the output layer of the network. Moreover, asymmetric manipulations of interneurons and their descending outputs induce drifts during high-speed walking, confirming their contribution to steering control. Together, these findings highlight the importance of competitive disinhibition as a critical circuit mechanism for robust processing of optic flow, which likely influences course control and heading perception, both critical functions supporting navigation.

For most model organisms in neuroscience, research into visual processing in the brain is difficult because of a lack of high-resolution maps that capture complex neuronal circuitry. The microinsect Megaphragma viggianii, because of its small size and non-trivial behavior, provides a unique opportunity for tractable whole-organism connectomics. We image its whole head using serial electron microscopy. We reconstruct its compound eye and analyze the optical properties of the ommatidia as well as the connectome of the first visual neuropil-the lamina. Compared with the fruit fly and the honeybee, Megaphragma visual system is highly simplified: it has 29 ommatidia per eye and 6 lamina neuron types. We report features that are both stereotypical among most ommatidia and specialized to some. By identifying the "barebones" circuits critical for flying insects, our results will facilitate constructing computational models of visual processing in insects.

Flying insects exhibit remarkable navigational abilities controlled by their compact nervous systems. Optic flow, the pattern of changes in the visual scene induced by locomotion, is a crucial sensory cue for robust self-motion estimation, especially during rapid flight. Neurons that respond to specific, large-field optic flow patterns have been studied for decades, primarily in large flies, such as houseflies, blowflies, and hover flies. The best-known optic-flow sensitive neurons are the large tangential cells of the dipteran lobula plate, whose visual-motion responses, and to a lesser extent, their morphology, have been explored using single-neuron neurophysiology. Most of these studies have focused on the large, Horizontal and Vertical System neurons, yet the lobula plate houses a much larger set of 'optic-flow' sensitive neurons, many of which have been challenging to unambiguously identify or to reliably target for functional studies. Here we report the comprehensive reconstruction and identification of the Lobula Plate Tangential Neurons in an Electron Microscopy (EM) volume of a whole Drosophila brain. This catalog of 58 LPT neurons (per brain hemisphere) contains many neurons that are described here for the first time and provides a basis for systematic investigation of the circuitry linking self-motion to locomotion control. Leveraging computational anatomy methods, we estimated the visual motion receptive fields of these neurons and compared their tuning to the visual consequence of body rotations and translational movements. We also matched these neurons, in most cases on a one-for-one basis, to stochastically labeled cells in genetic driver lines, to the mirror-symmetric neurons in the same EM brain volume, and to neurons in an additional EM data set. Using cell matches across data sets, we analyzed the integration of optic flow patterns by neurons downstream of the LPTs and find that most central brain neurons establish sharper selectivity for global optic flow patterns than their input neurons. Furthermore, we found that self-motion information extracted from optic flow is processed in distinct regions of the central brain, pointing to diverse foci for the generation of visual behaviors.

Animal behavior is principally expressed through neural control of muscles. Therefore understanding how the brain controls behavior requires mapping neuronal circuits all the way to motor neurons. We have previously established technology to collect large-volume electron microscopy data sets of neural tissue and fully reconstruct the morphology of the neurons and their chemical synaptic connections throughout the volume. Using these tools we generated a dense wiring diagram, or connectome, for a large portion of the Drosophila central brain. However, in most animals, including the fly, the majority of motor neurons are located outside the brain in a neural center closer to the body, i.e. the mammalian spinal cord or insect ventral nerve cord (VNC). In this paper, we extend our effort to map full neural circuits for behavior by generating a connectome of the VNC of a male fly.

The forthcoming assembly of the adult Drosophila melanogaster central brain connectome, containing over 125,000 neurons and 50 million synaptic connections, provides a template for examining sensory processing throughout the brain. Here, we create a leaky integrate-and-fire computational model of the entire Drosophila brain, based on neural connectivity and neurotransmitter identity, to study circuit properties of feeding and grooming behaviors. We show that activation of sugar-sensing or water-sensing gustatory neurons in the computational model accurately predicts neurons that respond to tastes and are required for feeding initiation. Computational activation of neurons in the feeding region of the Drosophila brain predicts those that elicit motor neuron firing, a testable hypothesis that we validate by optogenetic activation and behavioral studies. Moreover, computational activation of different classes of gustatory neurons makes accurate predictions of how multiple taste modalities interact, providing circuit-level insight into aversive and appetitive taste processing. Our computational model predicts that the sugar and water pathways form a partially shared appetitive feeding initiation pathway, which our calcium imaging and behavioral experiments confirm. Additionally, we applied this model to mechanosensory circuits and found that computational activation of mechanosensory neurons predicts activation of a small set of neurons comprising the antennal grooming circuit that do not overlap with gustatory circuits, and accurately describes the circuit response upon activation of different mechanosensory subtypes. Our results demonstrate that modeling brain circuits purely from connectivity and predicted neurotransmitter identity generates experimentally testable hypotheses and can accurately describe complete sensorimotor transformations.

Whereas progress has been made in the identification of neural signals related to rapid, cued decisions, less is known about how brains guide and terminate more ethologically relevant decisions in which an animal's own behaviour governs the options experienced over minutes. Drosophila search for many seconds to minutes for egg-laying sites with high relative value and have neurons, called oviDNs, whose activity fulfills necessity and sufficiency criteria for initiating the egg-deposition motor programme. Here we show that oviDNs express a calcium signal that (1) dips when an egg is internally prepared (ovulated), (2) drifts up and down over seconds to minutes-in a manner influenced by the relative value of substrates-as a fly determines whether to lay an egg and (3) reaches a consistent peak level just before the abdomen bend for egg deposition. This signal is apparent in the cell bodies of oviDNs in the brain and it probably reflects a behaviourally relevant rise-to-threshold process in the ventral nerve cord, where the synaptic terminals of oviDNs are located and where their output can influence behaviour. We provide perturbational evidence that the egg-deposition motor programme is initiated once this process hits a threshold and that subthreshold variation in this process regulates the time spent considering options and, ultimately, the choice taken. Finally, we identify a small recurrent circuit that feeds into oviDNs and show that activity in each of its constituent cell types is required for laying an egg. These results argue that a rise-to-threshold process regulates a relative-value, self-paced decision and provide initial insight into the underlying circuit mechanism for building this process.

Navigating animals continuously integrate velocity signals to update internal representations of their directional heading and spatial location in the environment. How neural circuits combine sensory and motor information to construct these velocity estimates and how these self-motion signals, in turn, update internal representations that support navigational computations are not well understood. Recent work in Drosophila has identified a neural circuit that performs angular path integration to compute the fly's head direction, but the nature of the velocity signal is unknown. Here we identify a pair of neurons necessary for angular path integration that encode the fly's rotational velocity with high accuracy using both visual optic flow and motor information. This estimate of rotational velocity does not rely on a moment-to-moment integration of sensory and motor information. Rather, when visual and motor signals are congruent, these neurons prioritize motor information over visual information, and when the two signals are in conflict, reciprocal inhibition selects either the motor or visual signal. Together, our results suggest that flies update their head direction representation by constructing an estimate of rotational velocity that relies primarily on motor information and only incorporates optic flow signals in specific sensorimotor contexts, such as when the motor signal is absent.

Training spiking recurrent neural networks on neuronal recordings or behavioral tasks has become a prominent tool to study computations in the brain. With an increasing size and complexity of neural recordings, there is a need for fast algorithms that can scale to large datasets. We present optimized CPU and GPU implementations of the recursive least-squares algorithm in spiking neural networks. The GPU implementation allows training networks to reproduce neural activity of an order of millions neurons at order of magnitude times faster than the CPU implementation. We demonstrate this by applying our algorithm to reproduce the activity of > 66, 000 recorded neurons of a mouse performing a decision-making task. The fast implementation enables efficient training of large-scale spiking models, thus allowing for in-silico study of the dynamics and connectivity underlying multi-area computations.