Large-scale automated histology in the pursuit of connectomes

An algorithm to predict the connectome of neural microcircuits

Experimentally mapping synaptic connections, in terms of the numbers and locations of their synapses and estimating connection probabilities, is still not a tractable task, even for small volumes of tissue. In fact, the six layers of the neocortex contain thousands of unique types of synaptic connections between the many different types of neurons, of which only a handful have been characterized experimentally. Here we present a theoretical framework and a data-driven algorithmic strategy to digitally reconstruct the complete synaptic connectivity between the different types of neurons in a small well-defined volume of tissue—the micro-scale connectome of a neural microcircuit. By enforcing a set of established principles of synaptic connectivity, and leveraging interdependencies between fundamental properties of neural microcircuits to constrain the reconstructed connectivity, the algorithm yields three parameters per connection type that predict the anatomy of all types of biologically viable synaptic connections. The predictions reproduce a spectrum of experimental data on synaptic connectivity not used by the algorithm. We conclude that an algorithmic approach to the connectome can serve as a tool to accelerate experimental mapping, indicating the minimal dataset required to make useful predictions, identifying the datasets required to improve their accuracy, testing the feasibility of experimental measurements, and making it possible to test hypotheses of synaptic connectivity.

Mapping Synaptic Input Fields of Neurons with Super-Resolution Imaging

As a basic functional unit in neural circuits, each neuron integrates input signals from hundreds to thousands of synapses. Knowledge of the synaptic input fields of individual neurons, including the identity, strength, and location of each synapse, is essential for understanding how neurons compute. Here, we developed a volumetric super-resolution reconstruction platform for large-volume imaging and automated segmentation of neurons and synapses with molecular identity information. We used this platform to map inhibitory synaptic input fields of On-Off direction-selective ganglion cells (On-Off DSGCs), which are important for computing visual motion direction in the mouse retina. The reconstructions of On-Off DSGCs showed a GABAergic, receptor subtype-specific input field for generating direction selective responses without significant glycinergic inputs for mediating monosynaptic crossover inhibition. These results demonstrate unique capabilities of this super-resolution platform for interrogating neural circuitry.

Facing the challenge of mammalian neural microcircuits: taking a few breaths may help

Breathing in mammals is a seemingly straightforward behaviour controlled by the brain. A brainstem nucleus called the preBötzinger Complex sits at the core of the neural circuit generating respiratory rhythm. Despite the discovery of this microcircuit almost 25 years ago, the mechanisms controlling breathing remain elusive. Given the apparent simplicity and well-defined nature of regulatory breathing behaviour, the identification of much of the circuitry, and the ability to study breathing in vitro as well as in vivo, many neuroscientists and physiologists are surprised that respiratory rhythm generation is still not well understood. Our view is that conventional rhythmogenic mechanisms involving pacemakers, inhibition or bursting are problematic and that simplifying assumptions commonly made for many vertebrate neural circuits ignore consequential detail. We propose that novel emergent mechanisms govern the generation of respiratory rhythm. That a mammalian function as basic as rhythm generation arises from complex and dynamic molecular, synaptic and neuronal interactions within a diverse neural microcircuit highlights the challenges in understanding neural control of mammalian behaviours, many (considerably) more elaborate than breathing. We suggest that the neural circuit controlling breathing is inimitably tractable and may inspire general strategies for elucidating other neural microcircuits.

The anatomical problem posed by brain complexity and size: a potential solution

Over the years the field of neuroanatomy has evolved considerably but unraveling the extraordinary structural and functional complexity of the brain seems to be an unattainable goal, partly due to the fact that it is only possible to obtain an imprecise connection matrix of the brain. The reasons why reaching such a goal appears almost impossible to date is discussed here, together with suggestions of how we could overcome this anatomical problem by establishing new methodologies to study the brain and by promoting interdisciplinary collaboration. Generating a realistic computational model seems to be the solution rather than attempting to fully reconstruct the whole brain or a particular brain region.

Glial Regulation of the Neuronal Connectome through Local and Long-Distant Communication

If "the connectome" represents a complete map of anatomical and functional connectivity in the brain, it should also include glia. Glia define and regulate both the brain's anatomical and functional connectivity over a broad range of length scales, spanning the whole brain to subcellular domains of synaptic interactions. This Perspective article examines glial interactions with the neuronal connectome (including long-range networks, local circuits, and individual synaptic connections) and highlights opportunities for future research. Our understanding of the structure and function of the neuronal connectome would be incomplete without an understanding of how all types of glia contribute to neuronal connectivity and function, from single synapses to circuits.

Different neuropeptides are expressed in different functional subsets of cholinergic excitatory motorneurons in the nematode Ascaris suum

Neuropeptides are known to have dramatic effects on neurons and synapses; however, despite extensive studies of the motorneurons in the parasitic nematode Ascaris suum, their peptide content had not yet been described. We determined the peptide content of single excitatory motorneurons by mass spectrometry and tandem mass spectrometry. There are two subsets of ventral cord excitatory motorneurons, each with neuromuscular output either anterior or posterior to their cell body, mediating forward or backward locomotion, respectively. Strikingly, the two sets of neurons contain different neuropeptides, with AF9 and six novel peptides (As-NLP-21.1-6) in anterior projectors, and the six afp-1 peptides in addition to AF2 in posterior projectors. In situ hybridization confirmed the expression of these peptides, validating the integrity of the dissection technique. This work identifies new components of the functional behavioral circuit, as well as potential targets for antiparasitic drug development.

Anatomical identification of extracellularly recorded cells in large-scale multielectrode recordings

This study combines for the first time two major approaches to understanding the function and structure of neural circuits: large-scale multielectrode recordings, and confocal imaging of labeled neurons. To achieve this end, we develop a novel approach to the central problem of anatomically identifying recorded cells, based on the electrical image: the spatiotemporal pattern of voltage deflections induced by spikes on a large-scale, high-density multielectrode array. Recordings were performed from identified ganglion cell types in the macaque retina. Anatomical images of cells in the same preparation were obtained using virally transfected fluorescent labeling or by immunolabeling after fixation. The electrical image was then used to locate recorded cell somas, axon initial segments, and axon trajectories, and these signatures were used to identify recorded cells. Comparison of anatomical and physiological measurements permitted visualization and physiological characterization of numerically dominant ganglion cell types with high efficiency in a single preparation.

From a meso- to micro-scale connectome: array tomography and mGRASP

Mapping mammalian synaptic connectivity has long been an important goal of neuroscience because knowing how neurons and brain areas are connected underpins an understanding of brain function. Meeting this goal requires advanced techniques with single synapse resolution and large-scale capacity, especially at multiple scales tethering the meso- and micro-scale connectome. Among several advanced LM-based connectome technologies, Array Tomography (AT) and mammalian GFP-Reconstitution Across Synaptic Partners (mGRASP) can provide relatively high-throughput mapping synaptic connectivity at multiple scales. AT- and mGRASP-assisted circuit mapping (ATing and mGRASPing), combined with techniques such as retrograde virus, brain clearing techniques, and activity indicators will help unlock the secrets of complex neural circuits. Here, we discuss these useful new tools to enable mapping of brain circuits at multiple scales, some functional implications of spatial synaptic distribution, and future challenges and directions of these endeavors.

X-ray microscopy as an approach to increasing accuracy and efficiency of serial block-face imaging for correlated light and electron microscopy of biological specimens

The recently developed three-dimensional electron microscopic (EM) method of serial block-face scanning electron microscopy (SBEM) has rapidly established itself as a powerful imaging approach. Volume EM imaging with this scanning electron microscopy (SEM) method requires intense staining of biological specimens with heavy metals to allow sufficient back-scatter electron signal and also to render specimens sufficiently conductive to control charging artifacts. These more extreme heavy metal staining protocols render specimens light opaque and make it much more difficult to track and identify regions of interest (ROIs) for the SBEM imaging process than for a typical thin section transmission electron microscopy correlative light and electron microscopy study. We present a strategy employing X-ray microscopy (XRM) both for tracking ROIs and for increasing the efficiency of the workflow used for typical projects undertaken with SBEM. XRM was found to reveal an impressive level of detail in tissue heavily stained for SBEM imaging, allowing for the identification of tissue landmarks that can be subsequently used to guide data collection in the SEM. Furthermore, specific labeling of individual cells using diaminobenzidine is detectable in XRM volumes. We demonstrate that tungsten carbide particles or upconverting nanophosphor particles can be used as fiducial markers to further increase the precision and efficiency of SBEM imaging.

The big data problem: turning maps into knowledge

In this NeuroView, Engert discusses the challenges for the connectomics field in making insights about brain function from big data.

VolRoverN: enhancing surface and volumetric reconstruction for realistic dynamical simulation of cellular and subcellular function

Establishing meaningful relationships between cellular structure and function requires accurate morphological reconstructions. In particular, there is an unmet need for high quality surface reconstructions to model subcellular and synaptic interactions among neurons and glia at nanometer resolution. We address this need with VolRoverN, a software package that produces accurate, efficient, and automated 3D surface reconstructions from stacked 2D contour tracings. While many techniques and tools have been developed in the past for 3D visualization of cellular structure, the reconstructions from VolRoverN meet specific quality criteria that are important for dynamical simulations. These criteria include manifoldness, water-tightness, lack of self- and object-object-intersections, and geometric accuracy. These enhanced surface reconstructions are readily extensible to any cell type and are used here on spiny dendrites with complex morphology and axons from mature rat hippocampal area CA1. Both spatially realistic surface reconstructions and reduced skeletonizations are produced and formatted by VolRoverN for easy input into analysis software packages for neurophysiological simulations at multiple spatial and temporal scales ranging from ion electro-diffusion to electrical cable models.

A workflow for the automatic segmentation of organelles in electron microscopy image stacks

Electron microscopy (EM) facilitates analysis of the form, distribution, and functional status of key organelle systems in various pathological processes, including those associated with neurodegenerative disease. Such EM data often provide important new insights into the underlying disease mechanisms. The development of more accurate and efficient methods to quantify changes in subcellular microanatomy has already proven key to understanding the pathogenesis of Parkinson's and Alzheimer's diseases, as well as glaucoma. While our ability to acquire large volumes of 3D EM data is progressing rapidly, more advanced analysis tools are needed to assist in measuring precise three-dimensional morphologies of organelles within data sets that can include hundreds to thousands of whole cells. Although new imaging instrument throughputs can exceed teravoxels of data per day, image segmentation and analysis remain significant bottlenecks to achieving quantitative descriptions of whole cell structural organellomes. Here, we present a novel method for the automatic segmentation of organelles in 3D EM image stacks. Segmentations are generated using only 2D image information, making the method suitable for anisotropic imaging techniques such as serial block-face scanning electron microscopy (SBEM). Additionally, no assumptions about 3D organelle morphology are made, ensuring the method can be easily expanded to any number of structurally and functionally diverse organelles. Following the presentation of our algorithm, we validate its performance by assessing the segmentation accuracy of different organelle targets in an example SBEM dataset and demonstrate that it can be efficiently parallelized on supercomputing resources, resulting in a dramatic reduction in runtime.

Slicing, sampling, and distance-dependent effects affect network measures in simulated cortical circuit structures

The neuroanatomical connectivity of cortical circuits is believed to follow certain rules, the exact origins of which are still poorly understood. In particular, numerous nonrandom features, such as common neighbor clustering, overrepresentation of reciprocal connectivity, and overrepresentation of certain triadic graph motifs have been experimentally observed in cortical slice data. Some of these data, particularly regarding bidirectional connectivity are seemingly contradictory, and the reasons for this are unclear. Here we present a simple static geometric network model with distance-dependent connectivity on a realistic scale that naturally gives rise to certain elements of these observed behaviors, and may provide plausible explanations for some of the conflicting findings. Specifically, investigation of the model shows that experimentally measured nonrandom effects, especially bidirectional connectivity, may depend sensitively on experimental parameters such as slice thickness and sampling area, suggesting potential explanations for the seemingly conflicting experimental results.

Development of MRI-based atlases of non-human brains

Brain atlases are a fundamental resource for neuroscience research. In the past few decades they have undergone a transition from traditional printed histological atlases to digital atlases made up of multiple data sets from multiple modalities, and atlases based on magnetic resonance imaging (MRI) have become widespread. Here we discuss the methods involved in making an MRI brain atlas, including registration of multiple data sets into a model, ontological classification, segmentation of a minimum deformation model, dissemination strategies, and applications of these atlases. Finally, we discuss possible future directions in the development of brain atlases.

Using mammalian GFP reconstitution across synaptic partners (mGRASP) to map synaptic connectivity in the mouse brain

Many types of questions in neuroscience require the detection and mapping of synapses in the complex mammalian brain. A tool, mammalian GFP reconstitution across synaptic partners (mGRASP), offers a relatively easy, quick and economical approach to this technically challenging task. Here we describe in step-by-step detail the protocols for virus production, gene delivery, brain specimen preparation, fluorescence imaging and image analysis, calibrated substantially and specifically to make mGRASP-assisted circuit mapping (mGRASPing) practical in the mouse brain. The protocol includes troubleshooting suggestions and solutions to common problems. The mGRASP method is suitable for mapping mammalian synaptic connectivity at multiple scales: microscale for synapse-by-synapse or neuron-by-neuron analysis, and mesoscale for revealing local and long-range circuits. The entire protocol takes 5-6 weeks, including time for incubation and virus expression.

Input clustering and the microscale structure of local circuits

The recent development of powerful tools for high-throughput mapping of synaptic networks promises major advances in understanding brain function. One open question is how circuits integrate and store information. Competing models based on random vs. structured connectivity make distinct predictions regarding the dendritic addressing of synaptic inputs. In this article we review recent experimental tests of one of these models, the input clustering hypothesis. Across circuits, brain regions and species, there is growing evidence of a link between synaptic co-activation and dendritic location, although this finding is not universal. The functional implications of input clustering and future challenges are discussed.

Large-scale automated identification of mouse brain cells in confocal light sheet microscopy images

MOTIVATION

Recently, confocal light sheet microscopy has enabled high-throughput acquisition of whole mouse brain 3D images at the micron scale resolution. This poses the unprecedented challenge of creating accurate digital maps of the whole set of cells in a brain.

RESULTS

We introduce a fast and scalable algorithm for fully automated cell identification. We obtained the whole digital map of Purkinje cells in mouse cerebellum consisting of a set of 3D cell center coordinates. The method is accurate and we estimated an F1 measure of 0.96 using 56 representative volumes, totaling 1.09 GVoxel and containing 4138 manually annotated soma centers.

AVAILABILITY AND IMPLEMENTATION

Source code and its documentation are available at http://bcfind.dinfo.unifi.it/. The whole pipeline of methods is implemented in Python and makes use of Pylearn2 and modified parts of Scikit-learn. Brain images are available on request.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Connectomic constraints on computation in feedforward networks of spiking neurons

Several efforts are currently underway to decipher the connectome or parts thereof in a variety of organisms. Ascertaining the detailed physiological properties of all the neurons in these connectomes, however, is out of the scope of such projects. It is therefore unclear to what extent knowledge of the connectome alone will advance a mechanistic understanding of computation occurring in these neural circuits, especially when the high-level function of the said circuit is unknown. We consider, here, the question of how the wiring diagram of neurons imposes constraints on what neural circuits can compute, when we cannot assume detailed information on the physiological response properties of the neurons. We call such constraints-that arise by virtue of the connectome-connectomic constraints on computation. For feedforward networks equipped with neurons that obey a deterministic spiking neuron model which satisfies a small number of properties, we ask if just by knowing the architecture of a network, we can rule out computations that it could be doing, no matter what response properties each of its neurons may have. We show results of this form, for certain classes of network architectures. On the other hand, we also prove that with the limited set of properties assumed for our model neurons, there are fundamental limits to the constraints imposed by network structure. Thus, our theory suggests that while connectomic constraints might restrict the computational ability of certain classes of network architectures, we may require more elaborate information on the properties of neurons in the network, before we can discern such results for other classes of networks.

Cartography of neurexin alternative splicing mapped by single-molecule long-read mRNA sequencing

Neurexins are evolutionarily conserved presynaptic cell-adhesion molecules that are essential for normal synapse formation and synaptic transmission. Indirect evidence has indicated that extensive alternative splicing of neurexin mRNAs may produce hundreds if not thousands of neurexin isoforms, but no direct evidence for such diversity has been available. Here we use unbiased long-read sequencing of full-length neurexin (Nrxn)1α, Nrxn1β, Nrxn2β, Nrxn3α, and Nrxn3β mRNAs to systematically assess how many sites of alternative splicing are used in neurexins with a significant frequency, and whether alternative splicing events at these sites are independent of each other. In sequencing more than 25,000 full-length mRNAs, we identified a novel, abundantly used alternatively spliced exon of Nrxn1α and Nrxn3α (referred to as alternatively spliced sequence 6) that encodes a 9-residue insertion in the flexible hinge region between the fifth LNS (laminin-α, neurexin, sex hormone-binding globulin) domain and the third EGF-like sequence. In addition, we observed several larger-scale events of alternative splicing that deleted multiple domains and were much less frequent than the canonical six sites of alternative splicing in neurexins. All of the six canonical events of alternative splicing appear to be independent of each other, suggesting that neurexins may exhibit an even larger isoform diversity than previously envisioned and comprise thousands of variants. Our data are consistent with the notion that α-neurexins represent extracellular protein-interaction scaffolds in which different LNS and EGF domains mediate distinct interactions that affect diverse functions and are independently regulated by independent events of alternative splicing.

Molecular neuroanatomy: a generation of progress

The neuroscience research landscape has changed dramatically over the past decade. Specifically, an impressive array of new tools and technologies have been generated, including but not limited to: brain gene expression atlases, genetically encoded proteins to monitor and manipulate neuronal activity, and new methods for imaging and mapping circuits. However, despite these technological advances, several significant challenges must be overcome to enable a better understanding of brain function and to develop cell type-targeted therapeutics to treat brain disorders. This review provides an overview of some of the tools and technologies currently being used to advance the field of molecular neuroanatomy, and also discusses emerging technologies that may enable neuroscientists to address these crucial scientific challenges over the coming decade.

Mapping mammalian synaptic connectivity

Mapping mammalian synaptic connectivity has long been an important goal of neuroscientists since it is considered crucial for explaining human perception and behavior. Yet, despite enormous efforts, the overwhelming complexity of the neural circuitry and the lack of appropriate techniques to unravel it have limited the success of efforts to map connectivity. However, recent technological advances designed to overcome the limitations of conventional methods for connectivity mapping may bring about a turning point. Here, we address the promises and pitfalls of these new mapping technologies.

Electron Microscopy Reconstruction of Brain Structure Using Sparse Representations Over Learned Dictionaries

A central problem in neuroscience is reconstructing neuronal circuits on the synapse level. Due to a wide range of scales in brain architecture such reconstruction requires imaging that is both high-resolution and high-throughput. Existing electron microscopy (EM) techniques possess required resolution in the lateral plane and either high-throughput or high depth resolution but not both. Here, we exploit recent advances in unsupervised learning and signal processing to obtain high depth-resolution EM images computationally without sacrificing throughput. First, we show that the brain tissue can be represented as a sparse linear combination of localized basis functions that are learned using high-resolution datasets. We then develop compressive sensing-inspired techniques that can reconstruct the brain tissue from very few (typically five) tomographic views of each section. This enables tracing of neuronal processes and, hence, high throughput reconstruction of neural circuits on the level of individual synapses.

Synaptic inputs compete during rapid formation of the calyx of Held: a new model system for neural development

Hallmark features of neural circuit development include early exuberant innervation followed by competition and pruning to mature innervation topography. Several neural systems, including the neuromuscular junction and climbing fiber innervation of Purkinje cells, are models to study neural development in part because they establish a recognizable endpoint of monoinnervation of their targets and because the presynaptic terminals are large and easily monitored. We demonstrate here that calyx of Held (CH) innervation of its target, which forms a key element of auditory brainstem binaural circuitry, exhibits all of these characteristics. To investigate CH development, we made the first application of serial block-face scanning electron microscopy to neural development with fine temporal resolution and thereby accomplished the first time series for 3D ultrastructural analysis of neural circuit formation. This approach revealed a growth spurt of added apposed surface area (ASA)>200 μm2/d centered on a single age at postnatal day 3 in mice and an initial rapid phase of growth and competition that resolved to monoinnervation in two-thirds of cells within 3 d. This rapid growth occurred in parallel with an increase in action potential threshold, which may mediate selection of the strongest input as the winning competitor. ASAs of competing inputs were segregated on the cell body surface. These data suggest mechanisms to select "winning" inputs by regional reinforcement of postsynaptic membrane to mediate size and strength of competing synaptic inputs.

Retinal connectomics: towards complete, accurate networks

Connectomics is a strategy for mapping complex neural networks based on high-speed automated electron optical imaging, computational assembly of neural data volumes, web-based navigational tools to explore 10(12)-10(15) byte (terabyte to petabyte) image volumes, and annotation and markup tools to convert images into rich networks with cellular metadata. These collections of network data and associated metadata, analyzed using tools from graph theory and classification theory, can be merged with classical systems theory, giving a more completely parameterized view of how biologic information processing systems are implemented in retina and brain. Networks have two separable features: topology and connection attributes. The first findings from connectomics strongly validate the idea that the topologies of complete retinal networks are far more complex than the simple schematics that emerged from classical anatomy. In particular, connectomics has permitted an aggressive refactoring of the retinal inner plexiform layer, demonstrating that network function cannot be simply inferred from stratification; exposing the complex geometric rules for inserting different cells into a shared network; revealing unexpected bidirectional signaling pathways between mammalian rod and cone systems; documenting selective feedforward systems, novel candidate signaling architectures, new coupling motifs, and the highly complex architecture of the mammalian AII amacrine cell. This is but the beginning, as the underlying principles of connectomics are readily transferrable to non-neural cell complexes and provide new contexts for assessing intercellular communication.

3D BrainCV: simultaneous visualization and analysis of cells and capillaries in a whole mouse brain with one-micron voxel resolution

Systematic cellular and vascular configurations are essential for understanding fundamental brain anatomy and metabolism. We demonstrated a 3D brainwide cellular and vascular (called 3D BrainCV) visualization and quantitative protocol for a whole mouse brain. We developed a modified Nissl staining method that quickly labeled the cells and blood vessels simultaneously in an entire mouse brain. Terabytes 3D datasets of the whole mouse brains, with unprecedented details of both individual cells and blood vessels, including capillaries, were simultaneously imaged at 1-μm voxel resolution using micro-optical sectioning tomography (MOST). For quantitative analysis, we proposed an automatic image-processing pipeline to perform brainwide vectorization and analysis of cells and blood vessels. Six representative brain regions from the cortex to the deep, including FrA, M1, PMBSF, V1, striatum, and amygdala, and six parameters, including cell number density, vascular length density, fractional vascular volume, distance from the cells to the nearest microvessel, microvascular length density, and fractional microvascular volume, had been quantitatively analyzed. The results showed that the proximity of cells to blood vessels was linearly correlated with vascular length density, rather than the cell number density. The 3D BrainCV made overall snapshots of the detailed picture of the whole brain architecture, which could be beneficial for the state comparison of the developing and diseased brain.

Beyond counts and shapes: studying pathology of dendritic spines in the context of the surrounding neuropil through serial section electron microscopy

Because dendritic spines are the sites of excitatory synapses, pathological changes in spine morphology should be considered as part of pathological changes in neuronal circuitry in the forms of synaptic connections and connectivity strength. In the past, spine pathology has usually been measured by changes in their number or shape. A more complete understanding of spine pathology requires visualization at the nanometer level to analyze how the changes in number and size affect their presynaptic partners and associated astrocytic processes, as well as organelles and other intracellular structures. Currently, serial section electron microscopy (ssEM) offers the best approach to address this issue because of its ability to image the volume of brain tissue at the nanometer resolution. Renewed interest in ssEM has led to recent technological advances in imaging techniques and improvements in computational tools indispensable for three-dimensional analyses of brain tissue volumes. Here we consider the small but growing literature that has used ssEM analysis to unravel ultrastructural changes in neuropil including dendritic spines. These findings have implications in altered synaptic connectivity and cell biological processes involved in neuropathology, and serve as anatomical substrates for understanding changes in network activity that may underlie clinical symptoms.

Cellular organization of cortical barrel columns is whisker-specific

The cellular organization of the cortex is of fundamental importance for elucidating the structural principles that underlie its functions. It has been suggested that reconstructing the structure and synaptic wiring of the elementary functional building block of mammalian cortices, the cortical column, might suffice to reverse engineer and simulate the functions of entire cortices. In the vibrissal area of rodent somatosensory cortex, whisker-related "barrel" columns have been referred to as potential cytoarchitectonic equivalents of functional cortical columns. Here, we investigated the structural stereotypy of cortical barrel columns by measuring the 3D neuronal composition of the entire vibrissal area in rat somatosensory cortex and thalamus. We found that the number of neurons per cortical barrel column and thalamic "barreloid" varied substantially within individual animals, increasing by ∼2.5-fold from dorsal to ventral whiskers. As a result, the ratio between whisker-specific thalamic and cortical neurons was remarkably constant. Thus, we hypothesize that the cellular architecture of sensory cortices reflects the degree of similarity in sensory input and not columnar and/or cortical uniformity principles.

Automated image computing reshapes computational neuroscience

We briefly identify several critical issues in current computational neuroscience, and present our opinions on potential solutions based on bioimage informatics, especially automated image computing.

Array tomography

In array tomography ordered, ribbon-like assemblies of ultrathin serial sections are deposited on a solid substrate and imaged afterwards. The resulting images are then aligned and reconstructed into a three-dimensional representation of the object. Depending on the preparation and labelling regime, different imaging modalities can be applied. When using light microscopy, the labelling with fluorescent markers would be the obvious choice, whereas the imaging in a scanning electron microscope would require impregnation with heavy metals. Depending on preparative constraints, the combination of diverse imaging modalities or truly correlative imaging is possible.

Spontaneous cortical activity alternates between motifs defined by regional axonal projections

Using millisecond-timescale voltage-sensitive dye imaging in lightly anesthetized or awake adult mice, we show that a palette of sensory-evoked and hemisphere-wide activity motifs are represented in spontaneous activity. These motifs can reflect multiple modes of sensory processing, including vision, audition and touch. We found similar cortical networks with direct cortical activation using channelrhodopsin-2. Regional analysis of activity spread indicated modality-specific sources, such as primary sensory areas, a common posterior-medial cortical sink where sensory activity was extinguished within the parietal association area and a secondary anterior medial sink within the cingulate and secondary motor cortices for visual stimuli. Correlation analysis between functional circuits and intracortical axonal projections indicated a common framework corresponding to long-range monosynaptic connections between cortical regions. Maps of intracortical monosynaptic structural connections predicted hemisphere-wide patterns of spontaneous and sensory-evoked depolarization. We suggest that an intracortical monosynaptic connectome shapes the ebb and flow of spontaneous cortical activity.

Segmentation of the mouse fourth deep lumbrical muscle connectome reveals concentric organisation of motor units

Connectomic analysis of the nervous system aims to discover and establish principles that underpin normal and abnormal neural connectivity and function. Here we performed image analysis of motor unit connectivity in the fourth deep lumbrical muscle (4DL) of mice, using transgenic expression of fluorescent protein in motor neurones as a morphological reporter. We developed a method that accelerated segmentation of confocal image projections of 4DL motor units, by applying high resolution (63×, 1.4 NA objective) imaging or deconvolution only where either proved necessary, in order to resolve axon crossings that produced ambiguities in the correct assignment of axon terminals to identified motor units imaged at lower optical resolution (40×, 1.3 NA). The 4DL muscles contained between 4 and 9 motor units and motor unit sizes ranged in distribution from 3 to 111 motor nerve terminals per unit. Several structural properties of the motor units were consistent with those reported in other muscles, including suboptimal wiring length and distribution of motor unit size. Surprisingly, however, small motor units were confined to a region of the muscle near the nerve entry point, whereas their larger counterparts were progressively more widely dispersed, suggesting a previously unrecognised form of segregated motor innervation in this muscle. We also found small but significant differences in variance of motor endplate length in motor units, which correlated weakly with their motor unit size. Thus, our connectomic analysis has revealed a pattern of concentric innervation that may perhaps also exist in other, cylindrical muscles that have not previously been thought to show segregated motor unit organisation. This organisation may be the outcome of competition during postnatal development based on intrinsic neuronal differences in synaptic size or synaptic strength that generates a territorial hierarchy in motor unit size and disposition.

CLARITY for mapping the nervous system

With potential relevance for brain-mapping work, hydrogel-based structures can now be built from within biological tissue to allow subsequent removal of lipids without mechanical disassembly of the tissue. This process creates a tissue-hydrogel hybrid that is physically stable, that preserves fine structure, proteins and nucleic acids, and that is permeable to both visible-spectrum photons and exogenous macromolecules. Here we highlight relevant challenges and opportunities of this approach, especially with regard to integration with complementary methodologies for brain-mapping studies.

The mind-brain relationship as a mathematical problem

This paper aims to frame certain fundamental aspects of the human mind (content and meaning of mental states) and foundational elements of brain computation (spatial and temporal patterns of neural activity) so as to enable at least in principle their integration within one and the same quantitative representation. Through the history of science, similar approaches have been instrumental to bridge other seemingly mysterious scientific phenomena, such as thermodynamics and statistical mechanics, optics and electromagnetism, or chemistry and quantum physics, among several other examples. Identifying the relevant levels of analysis is important to define proper mathematical formalisms for describing the brain and the mind, such that they could be mapped onto each other in order to explain their equivalence. Based on these premises, we overview the potential of neural connectivity to provide highly informative constraints on brain computational process. Moreover, we outline approaches for representing cognitive and emotional states geometrically with semantic maps. Next, we summarize leading theoretical framework that might serve as an explanatory bridge between neural connectivity and mental space. Furthermore, we discuss the implications of this framework for human communication and our view of reality. We conclude by analyzing the practical requirements to manage the necessary data for solving the mind-brain problem from this perspective.

Optogenetic approaches for functional mouse brain mapping

To better understand the connectivity of the brain, it is important to map both structural and functional connections between neurons and cortical regions. In recent years, a set of optogenetic tools have been developed that permit selective manipulation and investigation of neural systems. These tools have enabled the mapping of functional connections between stimulated cortical targets and other brain regions. Advantages of the approach include the ability to arbitrarily stimulate brain regions that express opsins, allowing for brain mapping independent of behavior or sensory processing. The ability of opsins to be rapidly and locally activated allows for investigation of connectivity with spatial resolution on the order of single neurons and temporal resolution on the order of milliseconds. Optogenetic methods for functional mapping have been applied in experiments ranging from in vitro investigation of microcircuits, to in vivo probing of inter-regional cortical connections, to examination of global connections within the whole brain. We review recently developed functional mapping methods that use optogenetic single-point stimulation in the rodent brain and employ cellular electrophysiology, evoked motor movements, voltage sensitive dyes (VSDs), calcium indicators, or functional magnetic resonance imaging (fMRI) to assess activity. In particular we highlight results using red-shifted organic VSDs that permit high temporal resolution imaging in a manner spectrally separated from Channelrhodopsin-2 (ChR2) activation. VSD maps stimulated by ChR2 were dependent on intracortical synaptic activity and were able to reflect circuits used for sensory processing. Although the methods reviewed are powerful, challenges remain with respect to finding approaches that permit selective high temporal resolution assessment of stimulated activity in animals that can be followed longitudinally.

Neurons with GABAergic phenotype in the visual system of Drosophila

The visual system of Drosophila contains ~60,000 neurons per hemisphere that are organized in parallel, retinotopically arranged columns. The neuroanatomy of these neurons has been mapped in considerable detail at both the light and ultrastructural level. However, studies providing direct evidence for synaptic signaling and the neurotransmitter used by individual neurons are relatively sparse. Here we characterize those neurons in the Drosophila optic lobes that possibly release gamma aminobutyric acid (GABA), the major inhibitory neurotransmitter of the insect central nervous system. We identified 26 different types of neurons of the lamina, medulla, lobula, and lobula plate. Based on the previous Golgi-staining analysis (Fischbach and Dittrich [1989] Cell Tissue Res 258:441-475), the identified neurons are further classified into 11 major subgroups representing lamina monopolar (L), medulla intrinsic (Mi, Mt), bushy T (T), transmedullary (Tm), transmedullary Y (TmY), Y, lobula-complex intrinsic (Lccn), lobula columnar (Lcn), lobula plate intrinsic (Lpi), and lobula tangential (Lt) cell types. This detailed map of neurons with GABAergic phenotype will contribute to the future neurogenetic dissection of information processing circuits in the fly visual system.

Connecting a connectome to behavior: an ensemble of neuroanatomical models of C. elegans klinotaxis

Increased efforts in the assembly and analysis of connectome data are providing new insights into the principles underlying the connectivity of neural circuits. However, despite these considerable advances in connectomics, neuroanatomical data must be integrated with neurophysiological and behavioral data in order to obtain a complete picture of neural function. Due to its nearly complete wiring diagram and large behavioral repertoire, the nematode worm Caenorhaditis elegans is an ideal organism in which to explore in detail this link between neural connectivity and behavior. In this paper, we develop a neuroanatomically-grounded model of salt klinotaxis, a form of chemotaxis in which changes in orientation are directed towards the source through gradual continual adjustments. We identify a minimal klinotaxis circuit by systematically searching the C. elegans connectome for pathways linking chemosensory neurons to neck motor neurons, and prune the resulting network based on both experimental considerations and several simplifying assumptions. We then use an evolutionary algorithm to find possible values for the unknown electrophsyiological parameters in the network such that the behavioral performance of the entire model is optimized to match that of the animal. Multiple runs of the evolutionary algorithm produce an ensemble of such models. We analyze in some detail the mechanisms by which one of the best evolved circuits operates and characterize the similarities and differences between this mechanism and other solutions in the ensemble. Finally, we propose a series of experiments to determine which of these alternatives the worm may be using.

Quantification of anisotropy and fiber orientation in human brain histological sections

Diffusion weighted imaging (DWI) has provided unparalleled insight into the microscopic structure and organization of the central nervous system. Diffusion tensor imaging (DTI) and other models of the diffusion MRI signal extract microstructural properties of tissues with relevance to the normal and injured brain. Despite the prevalence of such techniques and applications, accurate and large-scale validation has proven difficult, particularly in the human brain. In this report, human brain sections obtained from a digital public brain bank were employed to quantify anisotropy and fiber orientation using structure tensor analysis. The derived maps depict the intricate complexity of white matter fibers at a resolution not attainable with current DWI experiments. Moreover, the effects of multiple fiber bundles (i.e., crossing fibers) and intravoxel fiber dispersion were demonstrated. Examination of the cortex and hippocampal regions validated-specific features of previous in vivo and ex vivo DTI studies of the human brain. Despite the limitation to two dimensions, the resulting images provide a unique depiction of white matter organization at resolutions currently unattainable with DWI. The method of analysis may be used to validate tissue properties derived from DTI and alternative models of the diffusion signal.

Estimating the structure of small dynamical networks from the state time evolution of one node

We consider small dynamical networks of coupled oscillators for which the network topology is unknown and try to use partial knowledge of the oscillators' dynamics to estimate both the network couplings and the states of the nodes. We focus on the case where the state time evolution from only one oscillator is available. We propose an adaptive strategy that uses synchronization between the true network and a replica network in order to estimate both the couplings and the states. The adaptive scheme is tested with several modules of coupled oscillators. We consider the effects of small mismatches in the parameters of the individual oscillators and we propose an alternative version of the strategy that is suitable to handle noise in the received signal.

Non-uniformity of cell density and networks in the monkey brain

The brain is a very complex structure. Over the past several decades, many studies have aimed to understand how various non-uniform variables relate to each other. The current study compared the whole-brain network organization and global spatial distribution of cell densities in the monkey brain. Wide comparisons between 27 graph theoretical measures and cell densities revealed that only participation coefficients (PCs) significantly correlated with cell densities. Interestingly, PCs did not show a significant correlation with spatial coordinates. Furthermore, the significance of the correlation between cell densities and spatial coordinates disappeared only with the removal of the visual module, while the significance of the correlation between cell densities and PCs disappeared with the removal of any one module. Taken together, these results suggested the presence of a combinatorial effect of modular architectures in the network organization related to the non-uniformity of cell densities additional to the spatially monotonic change.

The neuronal organization of the retina

The mammalian retina consists of neurons of >60 distinct types, each playing a specific role in processing visual images. They are arranged in three main stages. The first decomposes the outputs of the rod and cone photoreceptors into ∼12 parallel information streams. The second connects these streams to specific types of retinal ganglion cells. The third combines bipolar and amacrine cell activity to create the diverse encodings of the visual world--roughly 20 of them--that the retina transmits to the brain. New transformations of the visual input continue to be found: at least half of the encodings sent to the brain (ganglion cell response selectivities) remain to be discovered. This diversity of the retina's outputs has yet to be incorporated into our understanding of higher visual function.

Building retinal connectomes

Understanding vertebrate vision depends on knowing, in part, the complete network graph of at least one representative retina. Acquiring such graphs is the business of synaptic connectomics, emerging as a practical technology due to improvements in electron imaging platform control, management software for large-scale datasets, and availability of data storage. The optimal strategy for building complete connectomes uses transmission electron imaging with 2 nm or better resolution, molecular tags for cell identification, open-access data volumes for navigation, and annotation with open-source tools to build 3D cell libraries, complete network diagrams and connectivity databases. The first forays into retinal connectomics have shown that even nominally well-studied cells have much richer connection graphs than expected.

Vectorization of optically sectioned brain microvasculature: learning aids completion of vascular graphs by connecting gaps and deleting open-ended segments

A graph of tissue vasculature is an essential requirement to model the exchange of gasses and nutriments between the blood and cells in the brain. Such a graph is derived from a vectorized representation of anatomical data, provides a map of all vessels as vertices and segments, and may include the location of nonvascular components, such as neuronal and glial somata. Yet vectorized data sets typically contain erroneous gaps, spurious endpoints, and spuriously merged strands. Current methods to correct such defects only address the issue of connecting gaps and further require manual tuning of parameters in a high dimensional algorithm. To address these shortcomings, we introduce a supervised machine learning method that (1) connects vessel gaps by "learned threshold relaxation"; (2) removes spurious segments by "learning to eliminate deletion candidate strands"; and (3) enforces consistency in the joint space of learned vascular graph corrections through "consistency learning." Human operators are only required to label individual objects they recognize in a training set and are not burdened with tuning parameters. The supervised learning procedure examines the geometry and topology of features in the neighborhood of each vessel segment under consideration. We demonstrate the effectiveness of these methods on four sets of microvascular data, each with >800(3) voxels, obtained with all optical histology of mouse tissue and vectorization by state-of-the-art techniques in image segmentation. Through statistically validated sampling and analysis in terms of precision recall curves, we find that learning with bagged boosted decision trees reduces equal-error error rates for threshold relaxation by 5-21% and strand elimination performance by 18-57%. We benchmark generalization performance across datasets; while improvements vary between data sets, learning always leads to a useful reduction in error rates. Overall, learning is shown to more than halve the total error rate, and therefore, human time spent manually correcting such vectorizations.

Three-dimensional brain reconstruction of in vivo electrode tracks for neuroscience and neural prosthetic applications

The brain is a densely interconnected network that relies on populations of neurons within and across multiple nuclei to code for features leading to perception and action. However, the neurophysiology field is still dominated by the characterization of individual neurons, rather than simultaneous recordings across multiple regions, without consistent spatial reconstruction of their locations for comparisons across studies. There are sophisticated histological and imaging techniques for performing brain reconstructions. However, what is needed is a method that is relatively easy and inexpensive to implement in a typical neurophysiology lab and provides consistent identification of electrode locations to make it widely used for pooling data across studies and research groups. This paper presents our initial development of such an approach for reconstructing electrode tracks and site locations within the guinea pig inferior colliculus (IC) to identify its functional organization for frequency coding relevant for a new auditory midbrain implant (AMI). Encouragingly, the spatial error associated with different individuals reconstructing electrode tracks for the same midbrain was less than 65 μm, corresponding to an error of ~1.5% relative to the entire IC structure (~4–5 mm diameter sphere). Furthermore, the reconstructed frequency laminae of the IC were consistently aligned across three sampled midbrains, demonstrating the ability to use our method to combine location data across animals. Hopefully, through further improvements in our reconstruction method, it can be used as a standard protocol across neurophysiology labs to characterize neural data not only within the IC but also within other brain regions to help bridge the gap between cellular activity and network function. Clinically, correlating function with location within and across multiple brain regions can guide optimal placement of electrodes for the growing field of neural prosthetics.

Inhibitory Regulation of Dendritic Activity in vivo

The spatiotemporal control of neuronal excitability is fundamental to the inhibitory process. We now have a wealth of information about the active dendritic properties of cortical neurons including axonally generated sodium action potentials as well as local sodium spikelets generated in the dendrites, calcium plateau spikes, and NMDA spikes. All of these events have been shown to be highly modified by the spatiotemporal pattern of nearby inhibitory input which can drastically change the output firing mode of the neuron. This means that particular populations of interneurons embedded in the neocortical microcircuitry can more precisely control pyramidal cell output than has previously been thought. Furthermore, the output of any given neuron tends to feed back onto inhibitory circuits making the resultant network activity further dependent on inhibition. Network activity is therefore ultimately governed by the subcellular microcircuitry of the cortex and it is impossible to ignore the subcompartmentalization of inhibitory influence at the neuronal level in order to understand its effects at the network level. In this article, we summarize the inhibitory circuits that have been shown so far to act on specific dendritic compartments in vivo.