Search

BigNeuron is an open community bench-testing platform with the goal of setting open standards for accurate and fast automatic neuron tracing. We gathered a diverse set of image volumes across several species that is representative of the data obtained in many neuroscience laboratories interested in neuron tracing. Here, we report generated gold standard manual annotations for a subset of the available imaging datasets and quantified tracing quality for 35 automatic tracing algorithms. The goal of generating such a hand-curated diverse dataset is to advance the development of tracing algorithms and enable generalizable benchmarking. Together with image quality features, we pooled the data in an interactive web application that enables users and developers to perform principal component analysis, t-distributed stochastic neighbor embedding, correlation and clustering, visualization of imaging and tracing data, and benchmarking of automatic tracing algorithms in user-defined data subsets. The image quality metrics explain most of the variance in the data, followed by neuromorphological features related to neuron size. We observed that diverse algorithms can provide complementary information to obtain accurate results and developed a method to iteratively combine methods and generate consensus reconstructions. The consensus trees obtained provide estimates of the neuron structure ground truth that typically outperform single algorithms in noisy datasets. However, specific algorithms may outperform the consensus tree strategy in specific imaging conditions. Finally, to aid users in predicting the most accurate automatic tracing results without manual annotations for comparison, we used support vector machine regression to predict reconstruction quality given an image volume and a set of automatic tracings. This resource describes a collection of neurons from a variety of light microscopy-based datasets, which can serve as a gold standard for testing automated tracing algorithms, as shown by comparison of the performance of 35 algorithms.

BigNeuron is an open community bench-testing platform combining the expertise of neuroscientists and computer scientists toward the goal of setting open standards for accurate and fast automatic neuron reconstruction. The project gathered a diverse set of image volumes across several species representative of the data obtained in most neuroscience laboratories interested in neuron reconstruction. Here we report generated gold standard manual annotations for a selected subset of the available imaging datasets and quantified reconstruction quality for 35 automatic reconstruction algorithms. Together with image quality features, the data were pooled in an interactive web application that allows users and developers to perform principal component analysis, t-distributed stochastic neighbor embedding, correlation and clustering, visualization of imaging and reconstruction data, and benchmarking of automatic reconstruction algorithms in user-defined data subsets. Our results show that image quality metrics explain most of the variance in the data, followed by neuromorphological features related to neuron size. By benchmarking automatic reconstruction algorithms, we observed that diverse algorithms can provide complementary information toward obtaining accurate results and developed a novel algorithm to iteratively combine methods and generate consensus reconstructions. The consensus trees obtained provide estimates of the neuron structure ground truth that typically outperform single algorithms. Finally, to aid users in predicting the most accurate automatic reconstruction results without manual annotations for comparison, we used support vector machine regression to predict reconstruction quality given an image volume and a set of automatic reconstructions.

It has become increasingly clear that the neurons in the cerebral cortex are not randomly interconnected. This wiring specificity can result from synapse formation mechanisms that interconnect neurons depending on their activity or genetically defined identity. Here we report that in addition to these synapse formation mechanisms, the structural composition of the neuropil provides a third prominent source by which wiring specificity emerges in cortical networks. This structurally determined wiring specificity reflects the packing density, morphological diversity and similarity of the dendritic and axonal processes. The higher these three factors are, the more recurrent the topology of the networks. Conversely, low density, diversity and similarity yield feedforward networks. These principles predict connectivity patterns from subcellular to network scales that are remarkably consistent with empirical observations from a rich body of literature. Thus, cortical network architectures reflect the specific morphological properties of their constituents to much larger degrees than previously thought.

The mammalian brain has an enormous demand for energy, which is thought to impose strong selective pressure by which the neurons evolve in ways that ensure robust function at minimal energy cost. However, which principles drive the ion channel distributions in the dendrites to implement different neuronal functions is yet unclear. Here we found that an energy-efficient generation of dendritic calcium action potentials in cortical pyramidal neurons requires a low expression of slow inactivating potassium channels. We demonstrate that this relationship between energy cost and neuronal function is independent of the dendritic morphology and the expression patterns of other ion channels that implement additional perisomatic and dendritic functions. Moreover, we found that calcium action potentials can arise from a wide spectrum of ion channel expression patterns, including configurations with high potassium channel densities in the dendrites. These configurations can account equally well for the characteristic intrinsic physiology of the pyramidal neurons. However, only configurations with low potassium channel densities in the distal dendrites are observed empirically. Thus, our findings indicate that cortical neurons do not utilize all theoretically possible ways to implement their functions, but instead select those optimized for energy-efficient active dendritic computations.

Maintaining an appropriate balance between excitation and inhibition is critical for neuronal information processing. Cortical neurons can cell-autonomously adjust the inhibition they receive to individual levels of excitatory input, but the underlying mechanisms are unclear. We describe that Ste20-like kinase (SLK) mediates cell-autonomous regulation of excitation-inhibition balance in the thalamocortical feedforward circuit, but not in the feedback circuit. This effect is due to regulation of inhibition originating from parvalbumin-expressing interneurons, while inhibition via somatostatin-expressing interneurons is unaffected. Computational modeling shows that this mechanism promotes stable excitatory-inhibitory ratios across pyramidal cells and ensures robust and sparse coding. Patch-clamp RNA sequencing yields genes differentially regulated by SLK knockdown, as well as genes associated with excitation-inhibition balance participating in transsynaptic communication and cytoskeletal dynamics. These data identify a mechanism for cell-autonomous regulation of a specific inhibitory circuit that is critical to ensure that a majority of cortical pyramidal cells participate in information coding.

Perception is causally linked to a calcium-dependent spiking mechanism that is built into the distal dendrites of layer 5 pyramidal tract neurons – the major output cell type of the cerebral cortex. It is yet unclear which circuits activate this cellular mechanism upon sensory stimulation. Here we found that the same thalamocortical axons that relay sensory signals to layer 4 also densely target the dendritic domain by which pyramidal tract neurons initiate calcium spikes. Distal dendritic inputs, which normally appear greatly attenuated at the cell body, thereby generate bursts of action potentials in cortical output during sensory processing. Our findings indicate that thalamus gates an active dendritic mechanism to facilitate the combination of sensory signals with top-down information streams into cortical output. Thus, in addition to being the central hub for sensory signals, thalamus is also likely to ensure that the signals it relays to cortex are perceived by the animal.

Pyramidal tract neurons (PTs) represent the major output cell type of the neocortex. To investigate principles of how the results of cortical processing are broadcasted to different downstream targets thus requires experimental approaches, which provide access to the in vivo electrophysiology of PTs, whose subcortical target regions are identified. On the example of rat barrel cortex (vS1), we illustrate that retrograde tracer injections into multiple subcortical structures allow identifying the long-range axonal targets of individual in vivo recorded PTs. Here we report that soma depth and dendritic path lengths within each cortical layer of vS1, as well as spiking patterns during both periods of ongoing activity and during sensory stimulation, reflect the respective subcortical target regions of PTs. We show that these cellular properties result in a structure–function parameter space that allows predicting a PT’s subcortical target region, without the need to inject multiple retrograde tracers., The major output cell type of the neocortex – pyramidal tract neurons (PTs) – send axonal projections to various subcortical areas. Here the authors combined in vivo recordings, retrograde tracings, and reconstructions of PTs in rat somatosensory cortex to show that PT structure and activity can predict specific subcortical targets.

Summary Pyramidal tract neurons (PTs) represent the major output cell type of the mammalian neocortex. Here, we report the origins of the PTs’ ability to respond to a broad range of stimuli with onset latencies that rival or even precede those of their intracortical input neurons. We find that neurons with extensive horizontally projecting axons cluster around the deep-layer terminal fields of primary thalamocortical axons. The strategic location of these corticocortical neurons results in high convergence of thalamocortical inputs, which drive reliable sensory-evoked responses that precede those in other excitatory cell types. The resultant fast and horizontal stream of excitation provides PTs throughout the cortical area with input that acts to amplify additional inputs from thalamocortical and other intracortical populations. The fast onsets and broadly tuned characteristics of PT responses hence reflect a gating mechanism in the deep layers, which assures that sensory-evoked input can be reliably transformed into cortical output., Highlights • Simulations predict in vivo responses for major output cell type of the neocortex • Simulations reveal strategy how to test the origins of cortical output empirically • Manipulations confirm that deep-layer corticocortical neurons gate cortical output • Gating of cortical output originates from deep-layer thalamocortical input stratum, Egger, Narayanan, et al. describe the cellular and circuit mechanisms underlying the transformation of sensory-evoked thalamocortical input into fast and broadly tuned cortical output. The study provides a comprehensive multi-scale cortex model for studying streams of sensory-evoked excitation in silico.

Recent research in the neurosciences has revealed a wealth of new information about the structural organization and physiological operation of the cerebral cortex. These details span vast spatial scales and range from the expression, arrangement, and interac- tion of molecular gene products at the synapse to the organization of computational net- works across the whole brain. This chapter highlights recent discoveries that have laid bare important aspects of the brain’s functional architecture. It begins by describing the dynamic and contingent arrangement of subcellular elements in synaptic connections. Amid this complexity, several common neural circuit motifs, identified across multiple species and preparations, shape the electrophysiological signaling in the cortex. It then turns to the topic of network organization, spurred by routine capacity for noninvasive MRI in humans, where interdisciplinary tools are lending new insights into large-scale principles of brain organization. Discussion follows on one of the most important aspects of brain architecture; namely, the plasticity that affords an animal flexible behav- ior. In closing, reflections are put forth on the nature of the brain’s complexity, and how its biological details might be best captured in computational models in the future.

Sensory information enters the neocortex via thalamocortical axons that define the major ‘input’ layer 4. The same thalamocortical axons, however, additionally innervate the deep ‘output’ layers 5/6. How such bistratification impacts cortical processing remains unknown. Here, we find a class of neurons that cluster specifically around thalamocortical axons at the layer 5/6 border. We show that these border stratum cells are characterized by extensive horizontal axons, that they receive strong convergent input from the thalamus, and that this input is sufficient to drive reliable sensory-evoked responses, which precede those in layer 4. These cells are hence strategically placed to amplify and relay thalamocortical inputs across the cortical area, for example to drive the fast onsets of cortical output patterns. Layer 4 is therefore not the sole starting point of cortical processing. Instead, parallel activation of layer 4 and the border stratum is necessary to broadcast information out of the neocortex.

Singing behavior in the adult male zebra finch is dependent upon the activity of a cortical region known as HVC (proper name). The vast majority of HVC projection neurons send primary axons to either the downstream premotor nucleus RA (robust nucleus of the arcopallium, or primary motor cortex) or Area X (basal ganglia), which play important roles in song production or song learning, respectively. In addition to these long-range outputs, HVC neurons also send local axon collaterals throughout that nucleus. Despite their implications for a range of circuit models, these local processes have never been completely reconstructed. Here, we use in vivo single-neuron Neurobiotin fills to examine 40 projection neurons across 31 birds with somatic positions distributed across HVC. We show that HVC(RA) and HVC(X) neurons have categorically distinct dendritic fields. Additionally, these cell classes send axon collaterals that are either restricted to a small portion of HVC ("local neurons") or broadly distributed throughout the entire nucleus ("broadcast neurons"). Overall, these processes within HVC offer a structural basis for significant local processing underlying behaviorally relevant population activity.

Highlights • Digital model of the rat facial nucleus anatomy. • Precise registration of structural data – obtained across animals – into a digital facial nucleus reference frame. • 3D reconstruction and registration of vibrissal motoneurons with identified target muscle. • Quantification of morphological cell-to-cell variably within a population of motoneurons that innervate the same muscle. • Mapping the presynaptic populations of reconstructed vibrissal motoneurons, whose target muscle is known., The rodent facial nucleus (FN) comprises motoneurons (MNs) that control the facial musculature. In the lateral part of the FN, populations of vibrissal motoneurons (vMNs) innervate two groups of muscles that generate movements of the whiskers. Vibrissal MNs thus represent the terminal point of the neuronal networks that generate rhythmic whisking during exploratory behaviors and that modify whisker movements based on sensory–motor feedback during tactile-based perception. Here, we combined retrograde tracer injections into whisker-specific muscles, with large-scale immunohistochemistry and digital reconstructions to generate an average model of the rat FN. The model incorporates measurements of the FN geometry, its cellular organization and a whisker row-specific map formed by vMNs. Furthermore, the model provides a digital 3D reference frame that allows registering structural data – obtained across scales and animals – into a common coordinate system with a precision of ∼60 µm. We illustrate the registration method by injecting replication competent rabies virus into the muscle of a single whisker. Retrograde transport of the virus to vMNs enabled reconstruction of their dendrites. Subsequent trans-synaptic transport enabled mapping the presynaptic neurons of the reconstructed vMNs. Registration of these data to the FN reference frame provides a first account of the morphological and synaptic input variability within a population of vMNs that innervate the same muscle.

The sequential activation of neurons has been observed in various areas of the brain, but in no case is the underlying network structure well understood. Here we examined the circuit anatomy of zebra finch HVC, a cortical region that generates sequences underlying the temporal progression of the song. We combined serial block-face electron microscopy with light microscopy to determine the cell types targeted by HVC(RA) neurons, which control song timing. Close to their soma, axons almost exclusively targeted inhibitory interneurons, consistent with what had been found with electrical recordings from pairs of cells. Conversely, far from the soma the targets were mostly other excitatory neurons, about half of these being other HVC(RA) cells. Both observations are consistent with the notion that the neural sequences that pace the song are generated by global synaptic chains in HVC embedded within local inhibitory networks. DOI: http://dx.doi.org/10.7554/eLife.24364.001, eLife digest For us to interact with the world around us, our brains must plan and execute our movements and behaviors. For instance, although speaking is often quite effortless, it is also remarkably complex; all of the muscles in our vocal cords have to be activated at just the right moments to create words. It remains poorly understood how exactly the brain generates such precise timing signals that enable these movements. A specific portion of the songbird brain allows the bird to sing its song, a process that has clear parallels with human speech. Previous work had demonstrated that this region in the bird’s brain acts as a ‘clock’ for singing behavior, with individual brain cells active at just a single moment, or ‘tick’. Little consensus had been reached concerning how this might be achieved. Kornfeld, Benezra et al. have now used new anatomical methods to better understand how the songbird clock works. A technique called 3D electron microscopy allowed the connections between the neurons in the clock brain region to be seen directly. This revealed that these neurons form direct connections with each other, which is consistent with the idea that one ‘tick’ can lead to the next and so on, like a series of falling dominoes. Several mysteries remain to be resolved by future research. First, the connections that Kornfeld, Benezra et al. found are between cells that are quite distant from each other. This arrangement is fundamentally different from many other brain areas where neighboring cells are thought to work together. Second, although these key brain cells form appropriate connections to act as a clock, it is still not clear whether and how the network uses these connections during singing. By resolving these mysteries, we will establish a new framework for understanding how the brain encodes learned motor gestures that may help to spur innovative new approaches for combatting motor-related deficits due to injury or disease. DOI: http://dx.doi.org/10.7554/eLife.24364.002

Neuroanatomical analysis, such as classification of cell types, depends on reliable reconstruction of large numbers of complete 3D dendrite and axon morphologies. At present, the majority of neuron reconstructions are obtained from preparations in a single tissue slice in vitro, thus suffering from cut off dendrites and, more dramatically, cut off axons. In general, axons can innervate volumes of several cubic millimeters and may reach path lengths of tens of centimeters. Thus, their complete reconstruction requires in vivo labeling, histological sectioning and imaging of large fields of view. Unfortunately, anisotropic background conditions across such large tissue volumes, as well as faintly labeled thin neurites, result in incomplete or erroneous automated tracings and even lead experts to make annotation errors during manual reconstructions. Consequently, tracing reliability renders the major bottleneck for reconstructing complete 3D neuron morphologies. Here, we present a novel set of tools, integrated into a software environment named ‘Filament Editor’, for creating reliable neuron tracings from sparsely labeled in vivo datasets. The Filament Editor allows for simultaneous visualization of complex neuronal tracings and image data in a 3D viewer, proof-editing of neuronal tracings, alignment and interconnection across sections, and morphometric analysis in relation to 3D anatomical reference structures. We illustrate the functionality of the Filament Editor on the example of in vivo labeled axons and demonstrate that for the exemplary dataset the final tracing results after proof-editing are independent of the expertise of the human operator.

Soma location, dendrite morphology, and synaptic innervation may represent key determinants of functional responses of individual neurons, such as sensory-evoked spiking. Here, we reconstruct the 3D circuits formed by thalamocortical afferents from the lemniscal pathway and excitatory neurons of an anatomically defined cortical column in rat vibrissal cortex. We objectively classify 9 cortical cell types and estimate the number and distribution of their somata, dendrites, and thalamocortical synapses. Somata and dendrites of most cell types intermingle, while thalamocortical connectivity depends strongly upon the cell type and the 3D soma location of the postsynaptic neuron. Correlating dendrite morphology and thalamocortical connectivity to functional responses revealed that the lemniscal afferents can account for some of the cell type- and location-specific subthreshold and spiking responses after passive whisker touch (e.g., in layer 4, but not for other cell types, e.g., in layer 5). Our data provides a quantitative 3D prediction of the cell type-specific lemniscal synaptic wiring diagram and elucidates structure-function relationships of this physiologically releVant pathway at single-cell resolution. © The Author 2011. Published by Oxford University Press. All rights reserved.

This is the second article in a series of three studies that investigate the anatomical determinants of thalamocortical (TC) input to excitatory neurons in a cortical column of rat primary somatosensory cortex (S1). Here, we report the number and distribution of NeuN-positive neurons within the C2, D2, and D3 TC projection columns in P27 rat somatosensory barrel cortex based on an exhaustive identification of 89,834 somata in a 1.15 mm(3) volume of cortex. A single column contained 19,109 ± 444 neurons (17,560 ± 399 when normalized to a standard-size projection column). Neuron density differences along the vertical column axis delineated "cytoarchitectonic" layers. The resulting neuron numbers per layer in the average column were 63 ± 10 (L1), 2039 ± 524 (L2), 3735 ± 905 (L3), 4447 ± 439 (L4), 1737 ± 251 (L5A), 2235 ± 99 (L5B), 3786 ± 168 (L6A), and 1066 ± 170 (L6B). These data were then used to derive the layer-specific action potential (AP) output of a projection column. The estimates confirmed previous reports suggesting that the ensembles of spiny L4 and thick-tufted pyramidal neurons emit the major fraction of APs of a column. The number of APs evoked in a column by a sensory stimulus (principal whisker deflection) was estimated as 4441 within 100 ms post-stimulus.

One of the major challenges in neuroscience is to provide an answer to the question: Is it possible to define a set of neuronal cell types that is sufficient to describe the organizational principles of cortex and relate structure to function? In this chapter, we will provide an overview of some of the most common current cell type classification schemes and discuss the relationships between them and their present limits. Then, we will review some of our own work that aims to elucidate the relationships between soma location, dendrite morphology, axonal projections, and in vivo spiking of individual excitatory cortical neurons. Using these classification criteria, we introduce a set of 10 input–response–output excitatory cell types in the vibrissal part of rat primary somatosensory cortex, and discuss whether these cell types represent a “minimal set” that is sufficient to describe structural and functional organizational principles of whisker-evoked information processing.

Jiang et al . (Research Article, 27 November 2015, aac9462) describe detailed experiments that substantially add to the knowledge of cortical microcircuitry and are unique in the number of connections reported and the quality of interneuron reconstruction. The work appeals to experts and laypersons because of the notion that it unveils new principles and provides a complete description of cortical circuits. We provide a counterbalance to the authors’ claims to give those less familiar with the minutiae of cortical circuits a better sense of the contributions and the limitations of this study.

Summary Models of cortical dynamics often assume a homogeneous connectivity structure. However, we show that heterogeneous input connectivity can prevent the dynamic balance between excitation and inhibition, a hallmark of cortical dynamics, and yield unrealistically sparse and temporally regular firing. Anatomically based estimates of the connectivity of layer 4 (L4) rat barrel cortex and numerical simulations of this circuit indicate that the local network possesses substantial heterogeneity in input connectivity, sufficient to disrupt excitation-inhibition balance. We show that homeostatic plasticity in inhibitory synapses can align the functional connectivity to compensate for structural heterogeneity. Alternatively, spike-frequency adaptation can give rise to a novel state in which local firing rates adjust dynamically so that adaptation currents and synaptic inputs are balanced. This theory is supported by simulations of L4 barrel cortex during spontaneous and stimulus-evoked conditions. Our study shows how synaptic and cellular mechanisms yield fluctuation-driven dynamics despite structural heterogeneity in cortical circuits., Highlights • Structural heterogeneity threatens the dynamic balance of excitation and inhibition • Reconstruction of cortical networks reveals significant structural heterogeneity • Spike-frequency adaptation can act locally to facilitate global balance • Inhibitory homeostatic plasticity can compensate for structural imbalance, Landau et al. show that anatomical variability is expected to generate substantial heterogeneity in the input connectivity of local cortical networks. This heterogeneity threatens the balance of excitation and inhibition. Balance can be recovered by spike-frequency adaptation or homeostatic plasticity.

Vertical thalamocortical afferents give rise to the elementary functional units of sensory cortex, cortical columns. Principles that underlie communication between columns remain however unknown. Here we unravel these by reconstructing in vivo-labeled neurons from all excitatory cell types in the vibrissal part of rat primary somatosensory cortex (vS1). Integrating the morphologies into an exact 3D model of vS1 revealed that the majority of intracortical (IC) axons project far beyond the borders of the principal column. We defined the corresponding innervation volume as the IC-unit. Deconstructing this structural cortical unit into its cell type-specific components, we found asymmetric projections that innervate columns of either the same whisker row or arc, and which subdivide vS1 into 2 orthogonal [supra-]granular and infragranular strata. We show that such organization could be most effective for encoding multi whisker inputs. Communication between columns is thus organized by multiple highly specific horizontal projection patterns, rendering IC-units as the primary structural entities for processing complex sensory stimuli.

Cortical inhibitory interneurons (INs) are subdivided into a variety of morphologically and functionally specialized cell types. How the respective specific properties translate into mechanisms that regulate sensory-evoked responses of pyramidal neurons (PNs) remains unknown. Here, we investigated how INs located in cortical layer 1 (L1) of rat barrel cortex affect whisker-evoked responses of L2 PNs. To do so we combined in vivo electrophysiology and morphological reconstructions with computational modeling. We show that whisker-evoked membrane depolarization in L2 PNs arises from highly specialized spatiotemporal synaptic input patterns. Temporally L1 INs and L2–5 PNs provide near synchronous synaptic input. Spatially synaptic contacts from L1 INs target distal apical tuft dendrites, whereas PNs primarily innervate basal and proximal apical dendrites. Simulations of such constrained synaptic input patterns predicted that inactivation of L1 INs increases trial-to-trial variability of whisker-evoked responses in L2 PNs. The in silico predictions were confirmed in vivo by L1-specific pharmacological manipulations. We present a mechanism—consistent with the theory of distal dendritic shunting—that can regulate the robustness of sensory-evoked responses in PNs without affecting response amplitude or latency.

In the cerebral cortex, individual neurons have been assigned to cell types based on different gene expression patterns, biophysical and physiological properties, and morphological features of somata, dendrites, and axons. The goal of grouping neurons into cell types is to understand common principles in development and function of the cerebral cortex at the cellular level. The somatodendritic cell-type classification presented here demonstrates how synaptic innervation, sensory-evoked functional responses, and axonal output pathways can be correlated at the levels of single neurons and cell types and how they relate to cell type-specific computations during different behavioral states of the animal. This definition of cell types lays the anatomical foundation for articles describing detailed brain mapping approaches of excitatory cortical pathways.

Soma location, dendrite morphology, and synaptic innervation are key determinants of neuronal function. Unfortunately, conventional functional measurements of sensory-evoked activity in vivo yield limited structural information. In particular, when trying to infer mechanistic principles that underlie perception and behavior, interpretations from functional recordings of individual or small groups of neurons often remain ambiguous without detailed knowledge of the underlying network structures. Here we review a novel reverse engineering approach that allows investigating sensory-evoked signal flow through individual and ensembles of neurons within the context of their surrounding neural networks. To do so, spontaneous and sensory-evoked activity patterns are recorded from individual neurons in vivo. In addition, the complete 3D dendrite and axon projection patterns of such in vivo-characterized neurons are reconstructed and integrated into an anatomically realistic model of the rat vibrissal cortex. This model allows estimating the number and cell type-specific subcellular distribution of synapses on these neurons with 50 μm precision. As a result, each neuron can be described by a rich set of parameters that allows investigating structure–function relationships and simulation experiments at single-neuron and network levels.

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.

Recent progress in large-volume microscopy, tissue-staining, as well as in image processing methods and 3D anatomy reconstruction allow neuroscientists to extract previously inaccessible anatomical data with high precision. For instance, determination of neuron numbers, 3D distributions and 3D axonal and dendritic branching patterns support recently started efforts to reconstruct anatomically realistic network models of many thousand neurons. Such models aid in understanding neural network structure, and, by numerically simulating electro-physiological signaling, also to reveal their function. We illustrate the impact of visual computing on neurobiology at the example of important steps that are required for the reconstruction of large neural networks. In our case, the network to be reconstructed represents a single cortical column in the rat brain, which processes sensory information from its associated facial whisker hair. We demonstrate how analysis and reconstruction tasks, such as neuron somata counting and tracing of neuronal branches, have been incrementally accelerated – finally leading to efficiency gains of orders of magnitude. We also show how steps that are difficult to automatize can now be solved interactively with visual support. Additionally, we illustrate how visualization techniques have aided computer scientists during algorithm development. Finally, we present visual analysis techniques allowing neuroscientists to explore morphology and function of 3D neural networks. Altogether, we demonstrate that visual computing techniques make an essential difference in terms of scientific output, both qualitatively, i.e., whether particular goals can be achieved at all, and quantitatively in terms of higher accuracy, faster work-flow and larger scale processing. Such techniques have therefore become essential in the daily work of neuroscientists.

The three-dimensional (3D) structure of neural circuits represents an essential constraint for information flow in the brain. Methods to directly monitor streams of excitation, at subcellular and millisecond resolution, are at present lacking. Here, we describe a pipeline of tools that allow investigating information flow by simulating electrical signals that propagate through anatomically realistic models of average neural networks. The pipeline comprises three blocks. First, we review tools that allow fast and automated acquisition of 3D anatomical data, such as neuron soma distributions or reconstructions of dendrites and axons from in vivo labeled cells. Second, we introduce NeuroNet, a tool for assembling the 3D structure and wiring of average neural networks. Finally, we introduce a simulation framework, NeuroDUNE, to investigate structure-function relationships within networks of full-compartmental neuron models at subcellular, cellular and network levels. We illustrate the pipeline by simulations of a reconstructed excitatory network formed between the thalamus and spiny stellate neurons in layer 4 (L4ss) of a cortical barrel column in rat vibrissal cortex. Exciting the ensemble of L4ss neurons with realistic input from an ensemble of thalamic neurons revealed that the location-specific thalamocortical connectivity may result in location-specific spiking of cortical cells. Specifically, a radial decay in spiking probability toward the column borders could be a general feature of signal flow in a barrel column. Our simulations provide insights of how anatomical parameters, such as the subcellular organization of synapses, may constrain spiking responses at the cellular and network levels.

How does the brain compute? Answering this question necessitates neuronal connectomes, annotated graphs of all synaptic connections within defined brain areas. Further, understanding the energetics of the brain's computations requires vascular graphs. The assembly of a connectome requires sensitive hardware tools to measure neuronal and neurovascular features in all three dimensions, as well as software and machine learning for data analysis and visualization. We present the state of the art on the reconstruction of circuits and vasculature that link brain anatomy and function. Analysis at the scale of tens of nanometers yields connections between identified neurons, while analysis at the micrometer scale yields probabilistic rules of connection between neurons and exact vascular connectivity.

Despite a long history of anatomical mapping of neuronal networks, we are only beginning to understand the detailed three-dimensional (3D) organization of the cortical micro-circuitry. This is in part due to the lack of complete reconstructions of individual cortical neurons. Morphological studies are either performed on incomplete cells in vitro, or when performed in vivo, lack the necessary cellular resolution. We recently reconstructed the in vivo axonal and dendritic morphology of two types of L(ayer) 5 neurons from vibrissal cortex. The 3D profiles of short-range as well as longrange projections indicate that L5 slender-tufted and L5 thick-tufted neurons represent very different building blocks of the cortical circuitry. In this addendum to Oberlaender et al. (PNAS 2011), we motivate our technical approach and the advancements this may give in reconstructing the cortical micro-circuitry.

The cortical output layer 5 contains two excitatory cell types, slender- and thick-tufted neurons. In rat vibrissal cortex, slender-tufted neurons carry motion and phase information during active whisking, but remain inactive after passive whisker touch. In contrast, thick-tufted neurons reliably increase spiking preferably after passive touch. By reconstructing the 3D patterns of intracortical axon projections from individual slender- and thick-tufted neurons, filled in vivo with biocytin, we were able to identify cell type-specific intracortical circuits that may encode whisker motion and touch. Individual slender-tufted neurons showed elaborate and dense innervation of supragranular layers of large portions of the vibrissal area (total length, 86.8 ± 5.5 mm). During active whisking, these long-range projections may modulate and phase-lock the membrane potential of dendrites in layers 2 and 3 to the whisking cycle. Thick-tufted neurons with soma locations intermingling with those of slender-tufted ones display less dense intracortical axon projections (total length, 31.6 ± 14.3 mm) that are primarily confined to infragranular layers. Based on anatomical reconstructions and previous measurements of spiking, we put forward the hypothesis that thick-tufted neurons in rat vibrissal cortex receive input of whisker motion from slender-tufted neurons onto their apical tuft dendrites and input of whisker touch from thalamic neurons onto their basal dendrites. During tactile-driven behavior, such as object location, near-coincident input from these two pathways may result in increased spiking activity of thick-tufted neurons and thus enhanced signaling to their subcortical targets.

We present a fast and robust method for the alignment of image stacks containing filamentous structures. Such stacks are usually obtained by physical sectioning a specimen, followed by an optical sectioning of each slice. For reconstruction, the filaments have to be traced and the sub-volumes aligned. Our algorithm takes traced filaments as input and matches their endpoints to find the optimal transform. We show that our method is able to quickly and accurately align sub-volumes containing neuronal processes, acquired using brightfield microscopy. Our method also makes it possible to align traced microtubuli, obtained from electron tomography data, which are extremely difficult to align manually.

We present a novel approach for automated detection of neuron somata. A three-step processing pipeline is described on the example of confocal image stacks of NeuN-stained neurons from rat somato-sensory cortex. It results in a set of position landmarks, representing the midpoints of all neuron somata. In the first step, foreground and background pixels are identified, resulting in a binary image. It is based on local thresholding and compensates for imaging and staining artifacts. Once this pre-processing guarantees a standard image quality, clusters of touching neurons are separated in the second step, using a marker-based watershed approach. A model-based algorithm completes the pipeline. It assumes a dominant neuron population with Gaussian distributed volumes within one microscopic field of view. Remaining larger objects are hence split or treated as a second neuron type. A variation of the processing pipeline is presented, showing that our method can also be used for co-localization of neurons in multi-channel images. As an example, we process 2-channel stacks of NeuN-stained somata, labeling all neurons, counterstained with GAD67, labeling GABAergic interneurons, using an adapted pre-processing step for the second channel. The automatically generated landmark sets are compared to manually placed counterparts. A comparison yields that the deviation in landmark position is negligible and that the difference between the numbers of manually and automatically counted neurons is less than 4%. In consequence, this novel approach for neuron counting is a reliable and objective alternative to manual detection.

A fundamental challenge in neuroscience is the determination of the three-dimensional (3D) morphology of neurons in the cortex. Here we describe a semiautomated method to trace single biocytin-filled neurons using a transmitted light brightfield microscope. The method includes 3D tracing of dendritic trees and axonal arbors from image stacks of serial 100-microm-thick tangential brain sections. Key functionalities include mosaic scanning and optical sectioning, high-resolution image restoration, and fast, parallel computing for neuron tracing. The mosaic technique compensates for the limited field of view at high magnification, allowing the acquisition of high-resolution image stacks on a scale of millimeters. The image restoration by deconvolution is based on experimentally verified assumptions about the optical system. Restoration yields a significant improvement of signal-to-noise ratio and resolution of neuronal structures in the image stack. Application of local threshold and thinning filters result in a 3D graph representation of dendrites and axons in a section. The reconstructed branches are then manually edited and aligned. Branches from adjacent sections are spliced, resulting in a complete 3D reconstruction of a neuron. A comparison with 3D reconstructions from manually traced neurons shows that the semiautomated system is a fast and reliable alternative to the manual tracing systems currently available.