Connectome-driven neural inventory of a complete visual system.

Vision provides animals with detailed information about their surroundings, conveying diverse features such as color, form, and movement across the visual scene. Computing these parallel spatial features requires a large and diverse network of neurons, such that in animals as distant as flies and humans, visual regions comprise half the brain's volume. These visual brain regions often reveal remarkable structure-function relationships, with neurons organized along spatial maps with shapes that directly relate to their roles in visual processing. To unravel the stunning diversity of a complex visual system, a careful mapping of the neural architecture matched to tools for targeted exploration of that circuitry is essential. Here, we report a new connectome of the right optic lobe from a male Drosophila central nervous system FIB-SEM volume and a comprehensive inventory of the fly's visual neurons. We developed a computational framework to quantify the anatomy of visual neurons, establishing a basis for interpreting how their shapes relate to spatial vision. By integrating this analysis with connectivity information, neurotransmitter identity, and expert curation, we classified the ~53,000 neurons into 727 types, about half of which are systematically described and named for the first time. Finally, we share an extensive collection of split-GAL4 lines matched to our neuron type catalog. Together, this comprehensive set of tools and data unlock new possibilities for systematic investigations of vision in Drosophila, a foundation for a deeper understanding of sensory processing.

En bloc preparation of Drosophila brains enables high-throughput FIB-SEM connectomics.

Deriving the detailed synaptic connections of an entire nervous system is the unrealized goal of the nascent field of connectomics. For the fruit fly Drosophila, in particular, we need to dissect the brain, connectives, and ventral nerve cord as a single continuous unit, fix and stain it, and undertake automated segmentation of neuron membranes. To achieve this, we designed a protocol using progressive lowering of temperature dehydration (PLT), a technique routinely used to preserve cellular structure and antigenicity. We combined PLT with low temperature en bloc staining (LTS) and recover fixed neurons as round profiles with darkly stained synapses, suitable for machine segmentation and automatic synapse detection. Here we report three different PLT-LTS methods designed to meet the requirements for FIB-SEM imaging of the Drosophila brain. These requirements include: good preservation of ultrastructural detail, high level of en bloc staining, artifact-free microdissection, and smooth hot-knife cutting to reduce the brain to dimensions suited to FIB-SEM. In addition to PLT-LTS, we designed a jig to microdissect and pre-fix the fly's delicate brain and central nervous system. Collectively these methods optimize morphological preservation, allow us to image the brain usually at 8 nm per voxel, and simultaneously speed the formerly slow rate of FIB-SEM imaging.

A connectome and analysis of the adult Drosophila central brain.

The neural circuits responsible for animal behavior remain largely unknown. We summarize new methods and present the circuitry of a large fraction of the brain of the fruit fly Drosophila melanogaster. Improved methods include new procedures to prepare, image, align, segment, find synapses in, and proofread such large data sets. We define cell types, refine computational compartments, and provide an exhaustive atlas of cell examples and types, many of them novel. We provide detailed circuits consisting of neurons and their chemical synapses for most of the central brain. We make the data public and simplify access, reducing the effort needed to answer circuit questions, and provide procedures linking the neurons defined by our analysis with genetic reagents. Biologically, we examine distributions of connection strengths, neural motifs on different scales, electrical consequences of compartmentalization, and evidence that maximizing packing density is an important criterion in the evolution of the fly's brain.

Animal brains of all sizes, from the smallest to the largest, work in broadly similar ways. Studying the brain of any one animal in depth can thus reveal the general principles behind the workings of all brains. The fruit fly Drosophila is a popular choice for such research. With about 100,000 neurons ­ compared to some 86 billion in humans ­ the fly brain is small enough to study at the level of individual cells. But it nevertheless supports a range of complex behaviors, including navigation, courtship and learning. Thanks to decades of research, scientists now have a good understanding of which parts of the fruit fly brain support particular behaviors. But exactly how they do this is often unclear. This is because previous studies showing the connections between cells only covered small areas of the brain. This is like trying to understand a novel when all you can see is a few isolated paragraphs. To solve this problem, Scheffer, Xu, Januszewski, Lu, Takemura, Hayworth, Huang, Shinomiya et al. prepared the first complete map of the entire central region of the fruit fly brain. The central brain consists of approximately 25,000 neurons and around 20 million connections. To prepare the map ­ or connectome ­ the brain was cut into very thin 8nm slices and photographed with an electron microscope. A three-dimensional map of the neurons and connections in the brain was then reconstructed from these images using machine learning algorithms. Finally, Scheffer et al. used the new connectome to obtain further insights into the circuits that support specific fruit fly behaviors. The central brain connectome is freely available online for anyone to access. When used in combination with existing methods, the map will make it easier to understand how the fly brain works, and how and why it can fail to work correctly. Many of these findings will likely apply to larger brains, including our own. In the long run, studying the fly connectome may therefore lead to a better understanding of the human brain and its disorders. Performing a similar analysis on the brain of a small mammal, by scaling up the methods here, will be a likely next step along this path.

Publisher Correction: Gas cluster ion beam SEM for imaging of large tissue samples with 10 nm isotropic resolution.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Gas cluster ion beam SEM for imaging of large tissue samples with 10 nm isotropic resolution.

We demonstrate gas cluster ion beam scanning electron microscopy (SEM), in which wide-area ion milling is performed on a series of thick tissue sections. This three-dimensional electron microscopy technique acquires datasets with <10 nm isotropic resolution of each section, and these can then be stitched together to span the sectioned volume. Incorporating gas cluster ion beam SEM into existing single-beam and multibeam SEM workflows should be straightforward, increasing reliability while improving z resolution by a factor of three or more.

Contacts between the endoplasmic reticulum and other membranes in neurons.

Close appositions between the membrane of the endoplasmic reticulum (ER) and other intracellular membranes have important functions in cell physiology. These include lipid homeostasis, regulation of Ca2+ dynamics, and control of organelle biogenesis and dynamics. Although these membrane contacts have previously been observed in neurons, their distribution and abundance have not been systematically analyzed. Here, we have used focused ion beam-scanning electron microscopy to generate 3D reconstructions of intracellular organelles and their membrane appositions involving the ER (distance ≤30 nm) in different neuronal compartments. ER-plasma membrane (PM) contacts were particularly abundant in cell bodies, with large, flat ER cisternae apposed to the PM, sometimes with a notably narrow lumen (thin ER). Smaller ER-PM contacts occurred throughout dendrites, axons, and in axon terminals. ER contacts with mitochondria were abundant in all compartments, with the ER often forming a network that embraced mitochondria. Small focal contacts were also observed with tubulovesicular structures, likely to be endosomes, and with sparse multivesicular bodies and lysosomes found in our reconstructions. Our study provides an anatomical reference for interpreting information about interorganelle communication in neurons emerging from functional and biochemical studies.

Enhanced FIB-SEM systems for large-volume 3D imaging.

Focused Ion Beam Scanning Electron Microscopy (FIB-SEM) can automatically generate 3D images with superior z-axis resolution, yielding data that needs minimal image registration and related post-processing. Obstacles blocking wider adoption of FIB-SEM include slow imaging speed and lack of long-term system stability, which caps the maximum possible acquisition volume. Here, we present techniques that accelerate image acquisition while greatly improving FIB-SEM reliability, allowing the system to operate for months and generating continuously imaged volumes > 106 µm3. These volumes are large enough for connectomics, where the excellent z resolution can help in tracing of small neuronal processes and accelerate the tedious and time-consuming human proofreading effort. Even higher resolution can be achieved on smaller volumes. We present example data sets from mammalian neural tissue, Drosophila brain, and Chlamydomonas reinhardtii to illustrate the power of this novel high-resolution technique to address questions in both connectomics and cell biology.

Ultrastructurally smooth thick partitioning and volume stitching for large-scale connectomics.

Focused-ion-beam scanning electron microscopy (FIB-SEM) has become an essential tool for studying neural tissue at resolutions below 10 nm × 10 nm × 10 nm, producing data sets optimized for automatic connectome tracing. We present a technical advance, ultrathick sectioning, which reliably subdivides embedded tissue samples into chunks (20 µm thick) optimally sized and mounted for efficient, parallel FIB-SEM imaging. These chunks are imaged separately and then 'volume stitched' back together, producing a final three-dimensional data set suitable for connectome tracing.

Imaging ATUM ultrathin section libraries with WaferMapper: a multi-scale approach to EM reconstruction of neural circuits.

The automated tape-collecting ultramicrotome (ATUM) makes it possible to collect large numbers of ultrathin sections quickly-the equivalent of a petabyte of high resolution images each day. However, even high throughput image acquisition strategies generate images far more slowly (at present ~1 terabyte per day). We therefore developed WaferMapper, a software package that takes a multi-resolution approach to mapping and imaging select regions within a library of ultrathin sections. This automated method selects and directs imaging of corresponding regions within each section of an ultrathin section library (UTSL) that may contain many thousands of sections. Using WaferMapper, it is possible to map thousands of tissue sections at low resolution and target multiple points of interest for high resolution imaging based on anatomical landmarks. The program can also be used to expand previously imaged regions, acquire data under different imaging conditions, or re-image after additional tissue treatments.

Distinct profiles of myelin distribution along single axons of pyramidal neurons in the neocortex.

Myelin is a defining feature of the vertebrate nervous system. Variability in the thickness of the myelin envelope is a structural feature affecting the conduction of neuronal signals. Conversely, the distribution of myelinated tracts along the length of axons has been assumed to be uniform. Here, we traced high-throughput electron microscopy reconstructions of single axons of pyramidal neurons in the mouse neocortex and built high-resolution maps of myelination. We find that individual neurons have distinct longitudinal distribution of myelin. Neurons in the superficial layers displayed the most diversified profiles, including a new pattern where myelinated segments are interspersed with long, unmyelinated tracts. Our data indicate that the profile of longitudinal distribution of myelin is an integral feature of neuronal identity and may have evolved as a strategy to modulate long-distance communication in the neocortex.

Stacked endoplasmic reticulum sheets are connected by helicoidal membrane motifs.

The endoplasmic reticulum (ER) often forms stacked membrane sheets, an arrangement that is likely required to accommodate a maximum of membrane-bound polysomes for secretory protein synthesis. How sheets are stacked is unknown. Here, we used improved staining and automated ultrathin sectioning electron microscopy methods to analyze stacked ER sheets in neuronal cells and secretory salivary gland cells of mice. Our results show that stacked ER sheets form a continuous membrane system in which the sheets are connected by twisted membrane surfaces with helical edges of left- or right-handedness. The three-dimensional structure of tightly stacked ER sheets resembles a parking garage, in which the different levels are connected by helicoidal ramps. A theoretical model explains the experimental observations and indicates that the structure corresponds to a minimum of elastic energy of sheet edges and surfaces. The structure allows the dense packing of ER sheets in the restricted space of a cell.

Dynamically partitionable autoassociative networks as a solution to the neural binding problem.

An outstanding question in theoretical neuroscience is how the brain solves the neural binding problem. In vision, binding can be summarized as the ability to represent that certain properties belong to one object while other properties belong to a different object. I review the binding problem in visual and other domains, and review its simplest proposed solution - the anatomical binding hypothesis. This hypothesis has traditionally been rejected as a true solution because it seems to require a type of one-to-one wiring of neurons that would be impossible in a biological system (as opposed to an engineered system like a computer). I show that this requirement for one-to-one wiring can be loosened by carefully considering how the neural representation is actually put to use by the rest of the brain. This leads to a solution where a symbol is represented not as a particular pattern of neural activation but instead as a piece of a global stable attractor state. I introduce the Dynamically Partitionable AutoAssociative Network (DPAAN) as an implementation of this solution and show how DPANNs can be used in systems which perform perceptual binding and in systems that implement syntax-sensitive rules. Finally I show how the core parts of the cognitive architecture ACT-R can be neurally implemented using a DPAAN as ACT-R's global workspace. Because the DPAAN solution to the binding problem requires only "flat" neural representations (as opposed to the phase encoded representation hypothesized in neural synchrony solutions) it is directly compatible with the most well developed neural models of learning, memory, and pattern recognition.

Pervasive synaptic branch removal in the mammalian neuromuscular system at birth.

VIDEO ABSTRACT: Using light and serial electron microscopy, we show profound refinements in motor axonal branching and synaptic connectivity before and after birth. Embryonic axons become maximally connected just before birth when they innervate ∼10-fold more muscle fibers than in maturity. In some developing muscles, axons innervate almost every muscle fiber. At birth, each neuromuscular junction is coinnervated by approximately ten highly intermingled axons (versus one in adults). Extensive die off of terminal branches occurs during the first several postnatal days, leading to much sparser arbors that still span the same territory. Despite the extensive pruning, total axoplasm per neuron increases as axons elongate, thicken, and add more synaptic release sites on their remaining targets. Motor axons therefore initially establish weak connections with nearly all available postsynaptic targets but, beginning at birth, massively redistribute synaptic resources, concentrating many more synaptic sites on many fewer muscle fibers. Analogous changes in connectivity may occur in the CNS.

Sensitivity to nonaccidental properties across various shape dimensions.

Nonaccidental properties (NAPs) are image properties that are invariant over orientation in depth and are distinguished from metric properties (MPs) that can change continuously with variations over depth orientation. To a large extent NAPs allow facile recognition of objects at novel viewpoints. Two match-to-sample experiments with 2D or 3D appearing geons assessed sensitivity to NAP vs. MP differences. A matching geon was always identical to the sample and the distractor differed from the matching geon in either a NAP or an MP on a single generalized cone dimension. For example, if the sample was a cylinder with a slightly curved axis, the NAP distractor would have a straight axis and the MP distractor would have an axis of greater curvature than the sample. Critically, the NAP and MP differences were scaled so that the MP differences were slightly greater according to pixel energy and Gabor wavelet measures of dissimilarity. Exp. 1 used a staircase procedure to determine the threshold presentation time required to achieve 75% accuracy. Exp. 2 used a constant, brief display presentation time with reaction times and error rates as dependent measures. Both experiments revealed markedly greater sensitivity to NAP over MP differences, and this was generally true for the individual dimensions. The NAP advantage was not reflected in the similarity computations of the C2 stage of HMAX, a widely cited model of later stage cortical ventral stream processing.

High-contrast en bloc staining of neuronal tissue for field emission scanning electron microscopy.

Conventional heavy metal poststaining methods on thin sections lend contrast but often cause contamination. To avoid this problem, we tested several en bloc staining techniques to contrast tissue in serial sections mounted on solid substrates for examination by field emission scanning electron microscopy (FESEM). Because FESEM section imaging requires that specimens have higher contrast and greater electrical conductivity than transmission electron microscopy (TEM) samples, our technique uses osmium impregnation (OTO) to make the samples conductive while heavily staining membranes for segmentation studies. Combining this step with other classic heavy metal en bloc stains, including uranyl acetate (UA), lead aspartate, copper sulfate and lead citrate, produced clean, highly contrasted TEM and scanning electron microscopy (SEM) samples of insect, fish and mammalian nervous systems. This protocol takes 7-15 d to prepare resin-embedded tissue, cut sections and produce serial section images.

The neural basis for shape preferences.

Several dimensions of shape, such as curvature or taper, can be regarded as extending from a singular (S) or 0 value (e.g., a straight contour with 0 curvature or parallel contours with a 0 angle of convergence) to an infinity of non-singular (NS) values (e.g., curves and non-parallel contours). As orientation in depth is varied, an S value remains S, and a NS value will vary but will remain NS. Infant and adult human participants viewed pairs of geons where one member had an S and the other had a NS value on a given shape dimension, e.g., a cylinder vs. a cone. Both adults and infants looked first, and adults looked longer at the NS geons. The NS geons also produced greater fMRI activation in shape selective cortex (LOC), a result consistent with the greater single unit activity in macaque IT produced by those geons (Kayaert et al., 2005). That NS stimuli elicit higher neural activity and attract eye movements may account for search asymmetries in that these stimuli pop out from their S distractors but not the reverse. A positive association between greater activation in higher-level areas of the ventral pathway and visual preference has been demonstrated previously for real world scenes (Yue, Vessel, & Biederman, 2007) and may reflect the workings of a motivational system that leads humans to seek novel but richly interpretable information.

Neural encoding of relative position.

Late ventral visual areas generally consist of cells having a significant degree of translation invariance. Such a "bag of features" representation is useful for the recognition of individual objects; however, it seems unable to explain our ability to parse a scene into multiple objects and to understand their spatial relationships. We review several schemes (e.g., global features and serial attention) for how to reconcile bag-of-features representation with our ability to understand relationships; we review structural description theories that, in contrast, suggest that a neural binding mechanism assigns the features of each object in a scene to a separate "slot" to which relational information for that object is explicitly bound. Four functional magnetic resonance imaging-adaptation experiments assessed how ventral stream regions respond to rearrangements of two objects in a minimal scene that depict scene translations and relational changes. Changes of relative position (e.g., elephant above bus changing to bus above elephant) produced larger releases of adaptation in the anterior lateral occipital complex (LOC) than physically equivalent translations, providing evidence that spatial relations are explicitly encoded in the anterior LOC in agreement with structural description theories.

Adaptation to objects in the lateral occipital complex (LOC): shape or semantics?

A change in the basic-level class when viewing a sequence of two objects produces a large release from adaptation in LOC compared to when the images are identical. Is this due to a change in semantics or shape? In an fMRI-adaptation experiment, subjects viewed a sequence of two objects and judged whether the stimuli were identical in shape. Different-shaped stimuli could be from the same or different basic-level classes, where the physical similarities of the pairs in the two conditions were equated by a model of simple cell similarity. BOLD responses in LOC for the two conditions were equivalent, and higher than that of the identical condition, indicating that LOC is sensitive to shape rather than to basic-level semantics.

Neural evidence for intermediate representations in object recognition.

The lateral occipital complex (LOC), a cortical region critical for human object recognition, has been shown to primarily code the shape, rather than the surface properties, of an object. But what aspects of shape? Using an fMRI-adaptation (fMRI-a) paradigm in which subjects judged whether two contour-deleted images of objects were the same or different exemplars, virtually all the adaptation in LOC [especially in LOC's most anterior portion (pFs)] could be attributed to repetition of the parts, almost none to the repetition of local image features, such as lines or vertices, templates, or basic- or subordinate-level concepts of the object. These results support the hypothesis that the neural representation of shape in LOC is an intermediate one, encoding the parts of an object.