Mitochondria play several crucial roles in the life of oligodendrocytes. During development of the myelin sheath they are essential providers of carbon skeletons and energy for lipid synthesis. During normal brain function their consumption of pyruvate will be a key determinant of how much lactate is available for oligodendrocytes to export to power axonal function. Finally, during calcium-overload induced pathology, as occurs in ischemia, mitochondria may buffer calcium or induce apoptosis. Despite their important functions, very little is known of the properties of oligodendrocyte mitochondria, and mitochondria have never been observed in the myelin sheaths. We have now used targeted expression of fluorescent mitochondrial markers to characterize the location and movement of mitochondria within oligodendrocytes. We show for the first time that mitochondria are able to enter and move within the myelin sheath. Within the myelin sheath the highest number of mitochondria was in the cytoplasmic ridges along the sheath. Mitochondria moved more slowly than in neurons and, in contrast to their behavior in neurons and astrocytes, their movement was increased rather than inhibited by glutamate activating NMDA receptors. By electron microscopy we show that myelin sheath mitochondria have a low surface area of cristae, which suggests a low ATP production. These data specify fundamental properties of the oxidative phosphorylation system in oligodendrocytes, the glial cells that enhance cognition by speeding action potential propagation and provide metabolic support to axons. GLIA 2016;64:810-825.

The formation of large, well-ordered crystals for crystallographic experiments remains a crucial bottleneck to the structural understanding of many important biological systems. To help alleviate this problem in crystallography, we have developed the MicroED method for the collection of electron diffraction data from 3D microcrystals and nanocrystals of radiation-sensitive biological material. In this approach, liquid solutions containing protein microcrystals are deposited on carbon-coated electron microscopy grids and are vitrified by plunging them into liquid ethane. MicroED data are collected for each selected crystal using cryo-electron microscopy, in which the crystal is diffracted using very few electrons as the stage is continuously rotated. This protocol gives advice on how to identify microcrystals by light microscopy or by negative-stain electron microscopy in samples obtained from standard protein crystallization experiments. The protocol also includes information about custom-designed equipment for controlling crystal rotation and software for recording experimental parameters in diffraction image metadata. Identifying microcrystals, preparing samples and setting up the microscope for diffraction data collection take approximately half an hour for each step. Screening microcrystals for quality diffraction takes roughly an hour, and the collection of a single data set is ∼10 min in duration. Complete data sets and resulting high-resolution structures can be obtained from a single crystal or by merging data from multiple crystals.

The life of an mRNA is dynamic within a cell. The development of quantitative fluorescent microscopy techniques to image single molecules of RNA has allowed many aspects of the mRNA lifecycle to be directly observed in living cells. Recent advances in live-cell multicolor RNA imaging, however, have now made it possible to investigate RNA metabolism in greater detail. In this chapter, we present an overview of the design and implementation of the translating RNA imaging by coat protein knockoff RNA biosensor, which allows untranslated mRNAs to be distinguished from ones that have undergone a round of translation. The methods required for establishing this system in mammalian cell lines and Drosophila melanogaster oocytes are described here, but the principles may be applied to any experimental system.

Clarifying gene expression in narrowly defined neuronal populations can provide insight into cellular identity, computation, and functionality. Here, we used next-generation RNA sequencing (RNA-seq) to produce a quantitative, whole genome characterization of gene expression for the major excitatory neuronal classes of the hippocampus; namely, granule cells and mossy cells of the dentate gyrus, and pyramidal cells of areas CA3, CA2, and CA1. Moreover, for the canonical cell classes of the trisynaptic loop, we profiled transcriptomes at both dorsal and ventral poles, producing a cell-class- and region-specific transcriptional description for these populations. This dataset clarifies the transcriptional properties and identities of lesser-known cell classes, and moreover reveals unexpected variation in the trisynaptic loop across the dorsal-ventral axis. We have created a public resource, Hipposeq (http://hipposeq.janelia.org), which provides analysis and visualization of these data and will act as a roadmap relating molecules to cells, circuits, and computation in the hippocampus.

In this review, we discuss the emerging field of computational behavioral analysis-the use of modern methods from computer science and engineering to quantitatively measure animal behavior. We discuss aspects of experiment design important to both obtaining biologically relevant behavioral data and enabling the use of machine vision and learning techniques for automation. These two goals are often in conflict. Restraining or restricting the environment of the animal can simplify automatic behavior quantification, but it can also degrade the quality or alter important aspects of behavior. To enable biologists to design experiments to obtain better behavioral measurements, and computer scientists to pinpoint fruitful directions for algorithm improvement, we review known effects of artificial manipulation of the animal on behavior. We also review machine vision and learning techniques for tracking, feature extraction, automated behavior classification, and automated behavior discovery, the assumptions they make, and the types of data they work best with. Expected final online publication date for the Annual Review of Neuroscience Volume 39 is July 08, 2016. Please see http://www.annualreviews.org/catalog/pubdates.aspx for revised estimates.

The precise positioning of organ progenitor cells constitutes an essential, yet poorly understood step during organogenesis. Using primordial germ cells that participate in gonad formation, we present the developmental mechanisms maintaining a motile progenitor cell population at the site where the organ develops. Employing high-resolution live-cell microscopy, we find that repulsive cues coupled with physical barriers confine the cells to the correct bilateral positions. This analysis revealed that cell polarity changes on interaction with the physical barrier and that the establishment of compact clusters involves increased cell-cell interaction time. Using particle-based simulations, we demonstrate the role of reflecting barriers, from which cells turn away on contact, and the importance of proper cell-cell adhesion level for maintaining the tight cell clusters and their correct positioning at the target region. The combination of these developmental and cellular mechanisms prevents organ fusion, controls organ positioning and is thus critical for its proper function.

Previously, we identified that visual and olfactory associative memories of Drosophila share the mushroom body (MB) circuits (Vogt et al. 2014). Despite well-characterized odor representations in the Drosophila MB, the MB circuit for visual information is totally unknown. Here we show that a small subset of MB Kenyon cells (KCs) selectively responds to visual but not olfactory stimulation. The dendrites of these atypical KCs form a ventral accessory calyx (vAC), distinct from the main calyx that receives olfactory input. We identified two types of visual projection neurons (VPNs) directly connecting the optic lobes and the vAC. Strikingly, these VPNs are differentially required for visual memories of color and brightness. The segregation of visual and olfactory domains in the MB allows independent processing of distinct sensory memories and may be a conserved form of sensory representations among insects.

Diverse structures facilitate direct exchange of proteins between cells, including plasmadesmata in plants and tunnelling nanotubes in bacteria and higher eukaryotes.  Here we describe a new mechanism of protein transfer, flagellar membrane fusion, in the unicellular parasite Trypanosoma brucei. When fluorescently tagged trypanosomes were co-cultured, a small proportion of double-positive cells were observed. The formation of double-positive cells was dependent on the presence of extracellular calcium and was enhanced by placing cells in medium supplemented with fresh bovine serum. Time-lapse microscopy revealed that double-positive cells arose by bidirectional protein exchange in the absence of nuclear transfer.  Furthermore, super-resolution microscopy showed that this process occurred in ≤1 minute, the limit of temporal resolution in these experiments. Both cytoplasmic and membrane proteins could be transferred provided they gained access to the flagellum. Intriguingly, a component of the RNAi machinery (Argonaute) was able to move between cells, raising the possibility that small interfering RNAs are transported as cargo. Transmission electron microscopy showed that shared flagella contained two axonemes and two paraflagellar rods bounded by a single membrane. In some cases flagellar fusion was partial and interactions between cells were transient. In other cases fusion occurred along the entire length of the flagellum, was stable for several hours and might be irreversible. Fusion did not appear to be deleterious for cell function: paired cells were motile and could give rise to progeny while fused. The motile flagella of unicellular organisms are related to the sensory cilia of higher eukaryotes, raising the possibility that protein transfer between cells via cilia or flagella occurs more widely in nature.

Neural activity maintains representations that bridge past and future events, often over many seconds. Network models can produce persistent and ramping activity, but the positive feedback that is critical for these slow dynamics can cause sensitivity to perturbations. Here we use electrophysiology and optogenetic perturbations in the mouse premotor cortex to probe the robustness of persistent neural representations during motor planning. We show that preparatory activity is remarkably robust to large-scale unilateral silencing: detailed neural dynamics that drive specific future movements were quickly and selectively restored by the network. Selectivity did not recover after bilateral silencing of the premotor cortex. Perturbations to one hemisphere are thus corrected by information from the other hemisphere. Corpus callosum bisections demonstrated that premotor cortex hemispheres can maintain preparatory activity independently. Redundancy across selectively coupled modules, as we observed in the premotor cortex, is a hallmark of robust control systems. Network models incorporating these principles show robustness that is consistent with data.

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.