New reconstruction techniques are generating connectomes of unprecedented size. These must be analyzed to generate human comprehensible results. The analyses being used fall into three general categories. The first is interactive tools used during reconstruction, to help guide the effort, look for possible errors, identify potential cell classes, and answer other preliminary questions. The second type of analysis is support for formal documents such as papers and theses. Scientific norms here require that the data be archived and accessible, and the analysis reproducible. In contrast to some other "omic" fields such as genomics, where a few specific analyses dominate usage, connectomics is rapidly evolving and the analyses used are often specific to the connectome being analyzed. These analyses are typically performed in a variety of conventional programming language, such as Matlab, R, Python, or C++, and read the connectomic data either from a file or through database queries, neither of which are standardized. In the short term we see no alternative to the use of specific analyses, so the best that can be done is to publish the analysis code, and the interface by which it reads connectomic data. A similar situation exists for archiving connectome data. Each group independently makes their data available, but there is no standardized format and long-term accessibility is neither enforced nor funded. In the long term, as connectomics becomes more common, a natural evolution would be a central facility for storing and querying connectomic data, playing a role similar to the National Center for Biotechnology Information for genomes. The final form of analysis is the import of connectome data into downstream tools such as neural simulation or machine learning. In this process, there are two main problems that need to be addressed. First, the reconstructed circuits contain huge amounts of detail, which must be intelligently reduced to a form the downstream tools can use. Second, much of the data needed for these downstream operations must be obtained by other methods (such as genetic or optical) and must be merged with the extracted connectome.

Automatic image segmentation is critical to scale up electron microscope (EM) connectome reconstruction. To this end, segmentation competitions, such as CREMI and SNEMI, exist to help researchers evaluate segmentation algorithms with the goal of improving them. Because generating ground truth is time-consuming, these competitions often fail to capture the challenges in segmenting larger datasets required in connectomics. More generally, the common metrics for EM image segmentation do not emphasize impact on downstream analysis and are often not very useful for isolating problem areas in the segmentation. For example, they do not capture connectivity information and often over-rate the quality of a segmentation as we demonstrate later. To address these issues, we introduce a novel strategy to enable evaluation of segmentation at large scales both in a supervised setting, where ground truth is available, or an unsupervised setting. To achieve this, we first introduce new metrics more closely aligned with the use of segmentation in downstream analysis and reconstruction. In particular, these include synapse connectivity and completeness metrics that provide both meaningful and intuitive interpretations of segmentation quality as it relates to the preservation of neuron connectivity. Also, we propose measures of segmentation correctness and completeness with respect to the percentage of "orphan" fragments and the concentrations of self-loops formed by segmentation failures, which are helpful in analysis and can be computed without ground truth. The introduction of new metrics intended to be used for practical applications involving large datasets necessitates a scalable software ecosystem, which is a critical contribution of this paper. To this end, we introduce a scalable, flexible software framework that enables integration of several different metrics and provides mechanisms to evaluate and debug differences between segmentations. We also introduce visualization software to help users to consume the various metrics collected. We evaluate our framework on two relatively large public groundtruth datasets providing novel insights on example segmentations.

The field of connectomics has recently produced neuron wiring diagrams from relatively large brain regions from multiple animals. Most of these neural reconstructions were computed from isotropic (e.g., FIBSEM) or near isotropic (e.g., SBEM) data. In spite of the remarkable progress on algorithms in recent years, automatic dense reconstruction from anisotropic data remains a challenge for the connectomics community. One significant hurdle in the segmentation of anisotropic data is the difficulty in generating a suitable initial over-segmentation. In this study, we present a segmentation method for anisotropic EM data that agglomerates a 3D over-segmentation computed from the 3D affinity prediction. A 3D U-net is trained to predict 3D affinities by the MALIS approach. Experiments on multiple datasets demonstrates the strength and robustness of the proposed method for anisotropic EM segmentation.

The anterolateral motor cortex (ALM) and ventromedial (VM) thalamus are functionally linked to support persistent activity during motor planning. We analyzed the underlying synaptic interconnections using optogenetics and electrophysiology in mice (♀/♂). In cortex, thalamocortical (TC) axons from VM excited VM-projecting pyramidal-tract (PT) neurons in layer 5B of ALM. These axons also strongly excited layer 2/3 neurons (which strongly excite PT neurons, as previously shown) but not VM-projecting corticothalamic (CT) neurons in layer 6. The strongest connections in the VM→PT circuit were localized to apical-tuft dendrites of PT neurons, in layer 1. These tuft inputs were selectively augmented after blocking hyperpolarization-activated cyclic nucleotide-gated (HCN) channels. In thalamus, axons from ALM PT neurons excited ALM-projecting VM neurons, located medially in VM. These axons provided weak input to neurons in mediodorsal nucleus, and little or no input either to neurons in the GABAergic reticular thalamic nucleus or to neurons in VM projecting to primary motor cortex (M1). Conversely, M1 PT axons excited M1- but not ALM-projecting VM neurons. Our findings indicate, first, a set of cell-type-specific connections forming an excitatory thalamo-cortico-thalamic (T-C-T) loop for ALM↔VM communication and a circuit-level substrate for supporting reverberant activity in this system. Second, a key feature of this loop is the prominent involvement of layer 1 synapses onto apical dendrites, a subcellular compartment with distinct signaling properties, including HCN-mediated gain control. Third, the segregation of the ALM↔VM loop from M1-related circuits of VM adds cellular-level support for the concept of parallel pathway organization in the motor system.Anterolateral motor cortex (ALM), a higher-order motor area in the mouse, and ventromedial thalamus (VM) are anatomically and functionally linked, but their synaptic interconnections at the cellular level are unknown. Our results show that ALM pyramidal tract neurons monosynaptically excite ALM-projecting thalamocortical neurons in a medial subdivision of VM, and vice versa. The thalamo-cortico-thalamic loop formed by these recurrent connections constitutes a circuit-level substrate for supporting reverberant activity in this system.

Epithelial tissues can elongate in two dimensions by polarized cell intercalation, oriented cell division, or cell shape change, owing to local or global actomyosin contractile forces acting in the plane of the tissue. In addition, epithelia can undergo morphogenetic change in three dimensions. We show that elongation of the wings and legs of Drosophila involves a columnar-to-cuboidal cell shape change that reduces cell height and expands cell width. Remodeling of the apical extracellular matrix by the Stubble protease and basal matrix by MMP1/2 proteases induces wing and leg elongation. Matrix remodeling does not occur in the haltere, a limb that fails to elongate. Limb elongation is made anisotropic by planar polarized Myosin-II, which drives convergent extension along the proximal-distal axis. Subsequently, Myosin-II relocalizes to lateral membranes to accelerate columnar-to-cuboidal transition and isotropic tissue expansion. Thus, matrix remodeling induces dynamic changes in actomyosin contractility to drive epithelial morphogenesis in three dimensions.

Nutrients, such as amino acids and glucose, signal through the Rag GTPases to activate mTORC1. The GATOR1 protein complex-comprising DEPDC5, NPRL2 and NPRL3-regulates the Rag GTPases as a GTPase-activating protein (GAP) for RAGA; loss of GATOR1 desensitizes mTORC1 signalling to nutrient starvation. GATOR1 components have no sequence homology to other proteins, so the function of GATOR1 at the molecular level is currently unknown. Here we used cryo-electron microscopy to solve structures of GATOR1 and GATOR1-Rag GTPases complexes. GATOR1 adopts an extended architecture with a cavity in the middle; NPRL2 links DEPDC5 and NPRL3, and DEPDC5 contacts the Rag GTPase heterodimer. Biochemical analyses reveal that our GATOR1-Rag GTPases structure is inhibitory, and that at least two binding modes must exist between the Rag GTPases and GATOR1. Direct interaction of DEPDC5 with RAGA inhibits GATOR1-mediated stimulation of GTP hydrolysis by RAGA, whereas weaker interactions between the NPRL2-NPRL3 heterodimer and RAGA execute GAP activity. These data reveal the structure of a component of the nutrient-sensing mTORC1 pathway and a non-canonical interaction between a GAP and its substrate GTPase.

Many brain functions depend on the ability of neural networks to temporally integrate transient inputs to produce sustained discharges. This can occur through cell-autonomous mechanisms in individual neurons and through reverberating activity in recurrently connected neural networks. We report a third mechanism involving temporal integration of neural activity by a network of astrocytes. Previously, we showed that some types of interneurons can generate long-lasting trains of action potentials (barrage firing) following repeated depolarizing stimuli. Here we show that calcium signaling in an astrocytic network correlates with barrage firing; that active depolarization of astrocyte networks by chemical or optogenetic stimulation enhances; and that chelating internal calcium, inhibiting release from internal stores, or inhibiting GABA transporters or metabotropic glutamate receptors inhibits barrage firing. Thus, networks of astrocytes influence the spatiotemporal dynamics of neural networks by directly integrating neural activity and driving barrages of action potentials in some populations of inhibitory interneurons.

The advent of direct electron detectors has enabled the routine use of single-particle cryo-electron microscopy (EM) approaches to determine structures of a variety of protein complexes at near-atomic resolution. Here, we report the development of methods to account for local variations in defocus and beam-induced drift, and the implementation of a data-driven dose compensation scheme that significantly improves the extraction of high-resolution information recorded during exposure of the specimen to the electron beam. These advances enable determination of a cryo-EM density map for β-galactosidase bound to the inhibitor phenylethyl β-D-thiogalactopyranoside where the ordered regions are resolved at a level of detail seen in X-ray maps at ∼ 1.5 Å resolution. Using this density map in conjunction with constrained molecular dynamics simulations provides a measure of the local flexibility of the non-covalently bound inhibitor and offers further opportunities for structure-guided inhibitor design.

Proteins in the fibrous amyloid state are a major hallmark of neurodegenerative disease. Understanding the multiple conformations, or polymorphs, of amyloid proteins at the molecular level is a challenge of amyloid research. Here, we detail the wide range of polymorphs formed by a segment of human TAR DNA-binding protein 43 (TDP-43) as a model for the polymorphic capabilities of pathological amyloid aggregation. Using X-ray diffraction, microelectron diffraction (MicroED) and single-particle cryo-EM, we show that theDLIIKGISVHIsegment from the second RNA-recognition motif (RRM2) forms an array of amyloid polymorphs. These associations include seven distinct interfaces displaying five different symmetry classes of steric zippers. Additionally, we find that this segment can adopt three different backbone conformations that contribute to its polymorphic capabilities. The polymorphic nature of this segment illustrates at the molecular level how amyloid proteins can form diverse fibril structures.

Mitochondrial apoptosis is mediated by BAK and BAX, two proteins that induce mitochondrial outer membrane permeabilization, leading to cytochrome c release and activation of apoptotic caspases. In the absence of active caspases, mitochondrial DNA (mtDNA) triggers the innate immune cGAS/STING pathway, causing dying cells to secrete type I interferon. How cGAS gains access to mtDNA remains unclear. We used live-cell lattice light-sheet microscopy to examine the mitochondrial network in mouse embryonic fibroblasts. We found that after BAK/BAX activation and cytochrome c loss, the mitochondrial network broke down and large BAK/BAX pores appeared in the outer membrane. These BAK/BAX macropores allowed the inner mitochondrial membrane to herniate into the cytosol, carrying with it mitochondrial matrix components, including the mitochondrial genome. Apoptotic caspases did not prevent herniation but dismantled the dying cell to suppress mtDNA-induced innate immune signaling.