Typical patterned movements in animals are achieved through combinations of contraction and delayed relaxation of groups of muscles. However, how intersegmentally coordinated patterns of muscular relaxation are regulated by the neural circuits remains poorly understood. Here, we identify Canon, a class of higher-order premotor interneurons, that regulates muscular relaxation during backward locomotion of Drosophila larvae. Canon neurons are cholinergic interneurons present in each abdominal neuromere and show wave-like activity during fictive backward locomotion. Optogenetic activation of Canon neurons induces relaxation of body wall muscles, whereas inhibition of these neurons disrupts timely muscle relaxation. Canon neurons provide excitatory outputs to inhibitory premotor interneurons. Canon neurons also connect with each other to form an intersegmental circuit and regulate their own wave-like activities. Thus, our results demonstrate how coordinated muscle relaxation can be realized by an intersegmental circuit that regulates its own patterned activity and sequentially terminates motor activities along the anterior-posterior axis.

Animal locomotion requires spatiotemporally coordinated contraction of muscles throughout the body. Here, we investigate how contractions of antagonistic groups of muscles are intersegmentally coordinated during bidirectional crawling of Drosophila larvae. We identify two pairs of higher-order premotor excitatory interneurons present in each abdominal neuromere that intersegmentally provide feedback to the adjacent neuromere during motor propagation. The two feedback neuron pairs are differentially active during either forward or backward locomotion but commonly target a group of premotor interneurons that together provide excitatory inputs to transverse muscles and inhibitory inputs to the antagonistic longitudinal muscles. Inhibition of either feedback neuron pair compromises contraction of transverse muscles in a direction-specific manner. Our results suggest that the intersegmental feedback neurons coordinate contraction of synergistic muscles by acting as delay circuits representing the phase lag between segments. The identified circuit architecture also shows how bidirectional motor networks could be economically embedded in the nervous system.

The generation of four-dimensional (4D) confocal datasets; consisting of 3D image sequences over time; provides an excellent methodology to capture cellular behaviors involved in developmental processes.  The ability to track and follow cell movements is limited by sample movements that occur due to drift of the sample or, in some cases, growth during image acquisition. Tracking cells in datasets affected by drift and/or growth will incorporate these movements into any analysis of cell position. This may result in the apparent movement of static structures within the sample. Therefore prior to cell tracking, any sample drift should be corrected. Using the open source Fiji distribution (1)  of ImageJ (2,3) and the incorporated LOCI tools (4), we developed the Correct 3D drift plug-in to remove erroneous sample movement in confocal datasets. This protocol effectively compensates for sample translation or alterations in focal position by utilizing phase correlation to register each time-point of a four-dimensional confocal datasets while maintaining the ability to visualize and measure cell movements over extended time-lapse experiments.

Locomotor systems generate diverse motor patterns to produce the movements underlying behavior, requiring that motor neurons be recruited at various phases of the locomotor cycle. Reciprocal inhibition produces alternating motor patterns; however, the mechanisms that generate other phasic relationships between intrasegmental motor pools are unknown. Here, we investigate one such motor pattern in the Drosophila larva, using a multidisciplinary approach including electrophysiology and ssTEM-based circuit reconstruction. We find that two motor pools that are sequentially recruited during locomotion have identical excitable properties. In contrast, they receive input from divergent premotor circuits. We find that this motor pattern is not orchestrated by differential excitatory input but by a GABAergic interneuron acting as a delay line to the later-recruited motor pool. Our findings show how a motor pattern is generated as a function of the modular organization of locomotor networks through segregation of inhibition, a potentially general mechanism for sequential motor patterns.

We present semiparametric spectral modeling of the complete larval Drosophila mushroom body connectome. Motivated by a thorough exploratory data analysis of the network via Gaussian mixture modeling (GMM) in the adjacency spectral embedding (ASE) representation space, we introduce the latent structure model (LSM) for network modeling and inference. LSM is a generalization of the stochastic block model (SBM) and a special case of the random dot product graph (RDPG) latent position model, and is amenable to semiparametric GMM in the ASE representation space. The resulting connectome code derived via semiparametric GMM composed with ASE captures latent connectome structure and elucidates biologically relevant neuronal properties.

Sensory navigation results from coordinated transitions between distinct behavioral programs. During chemotaxis in the larva, the detection of positive odor gradients extends runs while negative gradients promote stops and turns. This algorithm represents a foundation for the control of sensory navigation across phyla. In the present work, we identified an olfactory descending neuron, PDM-DN, which plays a pivotal role in the organization of stops and turns in response to the detection of graded changes in odor concentrations. Artificial activation of this descending neuron induces deterministic stops followed by the initiation of turning maneuvers through head casts. Using electron microscopy, we reconstructed the main pathway that connects the PDM-DN neuron to the peripheral olfactory system and to the pre-motor circuit responsible for the actuation of forward peristalsis. Our results set the stage for a detailed mechanistic analysis of the sensorimotor conversion of graded olfactory inputs into action selection to perform goal-oriented navigation.

Complex animal behaviors are built from dynamical relationships between sensory inputs, neuronal activity, and motor outputs in patterns with strategic value. Connecting these patterns illuminates how nervous systems compute behavior. Here, we study Drosophila larva navigation up temperature gradients toward preferred temperatures (positive thermotaxis). By tracking the movements of animals responding to fixed spatial temperature gradients or random temperature fluctuations, we calculate the sensitivity and dynamics of the conversion of thermosensory inputs into motor responses. We discover three thermosensory neurons in each dorsal organ ganglion (DOG) that are required for positive thermotaxis. Random optogenetic stimulation of the DOG thermosensory neurons evokes behavioral patterns that mimic the response to temperature variations. In vivo calcium and voltage imaging reveals that the DOG thermosensory neurons exhibit activity patterns with sensitivity and dynamics matched to the behavioral response. Temporal processing of temperature variations carried out by the DOG thermosensory neurons emerges in distinct motor responses during thermotaxis.

The neuroarchitecture of Acoela has been at the center of morphological debates. Some authors, using immunochemical tools, suggest that the nervous system in Acoela is organized as a commissural brain that bears little resemblance to the central, ganglionic type brain of other flatworms, and bilaterians in general. Others, who used histological staining on paraffin sections, conclude that it is a compact structure (an endonal brain; e.g., Raikova 2004; von Graff 1891; Delage Arch Zool Exp Gén 4:109-144, 1886). To address this question with modern tools, we have obtained images from serial transmission electron microscopic sections of the entire hatchling of Symsagittifera roscoffensis. In addition, we obtained data from wholemounts of hatchlings labeled with markers for serotonin and tyrosinated tubulin. Our data show that the central nervous system of a juvenile S. roscoffensis consists of an anterior compact brain, formed by a dense, bilobed mass of neuronal cell bodies surrounding a central neuropile. The neuropile flanks the median statocyst and contains several types of neurites, classified according to their types of synaptic vesicles. The neuropile issues three pairs of nerve cords that run at different dorso-ventral positions along the whole length of the body. Neuronal cell bodies flank the cords, and neuromuscular synapses are abundant. The TEM analysis also reveals different classes of peripheral sensory neurons and provides valuable information about the spatial relationships between neurites and other cell types within the brain and nerve cords. We conclude that the acoel S. roscoffensis has a central brain that is comparable in size and architecture to the brain of other (rhabditophoran) flatworms.

Structured learning provides a powerful framework for empirical risk minimization on the predictions of structured models. It allows end-to-end learning of model parameters to minimize an application specific loss function. This framework is particularly well suited for discrete optimization models that are used for neuron reconstruction from anisotropic electron microscopy (EM) volumes. However, current methods are still learning unary potentials by training a classifier that is agnostic about the model it is used in. We believe the reason for that lies in the difficulties of (1) finding a representative training sample, and (2) designing an application specific loss function that captures the quality of a proposed solution. In this paper, we show how to find a representative training sample from human generated ground truth, and propose a loss function that is suitable to minimize topological errors in the reconstruction. We compare different training methods on two challenging EM-datasets. Our structured learning approach shows consistently higher reconstruction accuracy than other current learning methods.

NeuromedinU is a potent regulator of food intake and activity in mammals. In Drosophila, neurons producing the homologous neuropeptide hugin regulate feeding and locomotion in a similar manner. Here, we use EM-based reconstruction to generate the entire connectome of hugin-producing neurons in the Drosophila larval CNS. We demonstrate that hugin neurons use synaptic transmission in addition to peptidergic neuromodulation and identify acetylcholine as a key transmitter. Hugin neuropeptide and acetylcholine are both necessary for the regulatory effect on feeding. We further show that subtypes of hugin neurons connect chemosensory to endocrine system by combinations of synaptic and peptide-receptor connections. Targets include endocrine neurons producing DH44, a CRH-like peptide, and insulin-like peptides. Homologs of these peptides are likewise downstream of neuromedinU, revealing striking parallels in flies and mammals. We propose that hugin neurons are part of an ancient physiological control system that has been conserved at functional and molecular level.