Double-bouquet cells in the monkey and human cerebral cortex with special reference to areas 17 and 18

Variation and convergence in the morpho-functional properties of the mammalian neocortex

Man's natural inclination to classify and hierarchize the living world has prompted neurophysiologists to explore possible differences in brain organisation between mammals, with the aim of understanding the diversity of their behavioural repertoires. But what really distinguishes the human brain from that of a platypus, an opossum or a rodent? In this review, we compare the structural and electrical properties of neocortical neurons in the main mammalian radiations and examine their impact on the functioning of the networks they form. We discuss variations in overall brain size, number of neurons, length of their dendritic trees and density of spines, acknowledging their increase in humans as in most large-brained species. Our comparative analysis also highlights a remarkable consistency, particularly pronounced in marsupial and placental mammals, in the cell typology, intrinsic and synaptic electrical properties of pyramidal neuron subtypes, and in their organisation into functional circuits. These shared cellular and network characteristics contribute to the emergence of strikingly similar large-scale physiological and pathological brain dynamics across a wide range of species. These findings support the existence of a core set of neural principles and processes conserved throughout mammalian evolution, from which a number of species-specific adaptations appear, likely allowing distinct functional needs to be met in a variety of environmental contexts.

Signature morphoelectric properties of diverse GABAergic interneurons in the human neocortex

Human cortex transcriptomic studies have revealed a hierarchical organization of γ-aminobutyric acid-producing (GABAergic) neurons from subclasses to a high diversity of more granular types. Rapid GABAergic neuron viral genetic labeling plus Patch-seq (patch-clamp electrophysiology plus single-cell RNA sequencing) sampling in human brain slices was used to reliably target and analyze GABAergic neuron subclasses and individual transcriptomic types. This characterization elucidated transitions between PVALB and SST subclasses, revealed morphological heterogeneity within an abundant transcriptomic type, identified multiple spatially distinct types of the primate-specialized double bouquet cells (DBCs), and shed light on cellular differences between homologous mouse and human neocortical GABAergic neuron types. These results highlight the importance of multimodal phenotypic characterization for refinement of emerging transcriptomic cell type taxonomies and for understanding conserved and specialized cellular properties of human brain cell types.

Differential effects of group III metabotropic glutamate receptors on spontaneous inhibitory synaptic currents in spine-innervating double bouquet and parvalbumin-expressing dendrite-targeting GABAergic interneurons in human neocortex

Diverse neocortical GABAergic neurons specialize in synaptic targeting and their effects are modulated by presynaptic metabotropic glutamate receptors (mGluRs) suppressing neurotransmitter release in rodents, but their effects in human neocortex are unknown. We tested whether activation of group III mGluRs by L-AP4 changes GABAA receptor-mediated spontaneous inhibitory postsynaptic currents (sIPSCs) in 2 distinct dendritic spine-innervating GABAergic interneurons recorded in vitro in human neocortex. Calbindin-positive double bouquet cells (DBCs) had columnar "horsetail" axons descending through layers II-V innervating dendritic spines (48%) and shafts, but not somata of pyramidal and nonpyramidal neurons. Parvalbumin-expressing dendrite-targeting cell (PV-DTC) axons extended in all directions innervating dendritic spines (22%), shafts (65%), and somata (13%). As measured, 20% of GABAergic neuropil synapses innervate spines, hence DBCs, but not PV-DTCs, preferentially select spine targets. Group III mGluR activation paradoxically increased the frequency of sIPSCs in DBCs (to median 137% of baseline) but suppressed it in PV-DTCs (median 92%), leaving the amplitude unchanged. The facilitation of sIPSCs in DBCs may result from their unique GABAergic input being disinhibited via network effect. We conclude that dendritic spines receive specialized, diverse GABAergic inputs, and group III mGluRs differentially regulate GABAergic synaptic transmission to distinct GABAergic cell types in human cortex.

The large numbers of minicolumns in the primary visual cortex of humans, chimpanzees and gorillas are related to high visual acuity

Minicolumns are thought to be a fundamental neural unit in the neocortex and their replication may have formed the basis of the rapid cortical expansion that occurred during primate evolution. We sought evidence of minicolumns in the primary visual cortex (V-1) of three great apes, three rodents and representatives from three other mammalian orders: Eulipotyphla (European hedgehog), Artiodactyla (domestic pig) and Carnivora (ferret). Minicolumns, identified by the presence of a long bundle of radial, myelinated fibers stretching from layer III to the white matter of silver-stained sections, were found in the human, chimpanzee, gorilla and guinea pig V-1. Shorter bundles confined to one or two layers were found in the other species but represent modules rather than minicolumns. The inter-bundle distance, and hence density of minicolumns, varied systematically both within a local area that might represent a hypercolumn but also across the whole visual field. The distance between all bundles had a similar range for human, chimpanzee, gorilla, ferret and guinea pig: most bundles were 20-45 μm apart. By contrast, the space between bundles was greater for the hedgehog and pig (20-140 μm). The mean density of minicolumns was greater in tangential sections of the gorilla and chimpanzee (1,243-1,287 bundles/mm2) than in human (314-422 bundles/mm2) or guinea pig (643 bundles/mm2). The minicolumnar bundles did not form a hexagonal lattice but were arranged in thin curving and branched bands separated by thicker bands of neuropil/somata. Estimates of the total number of modules/minicolumns within V-1 were strongly correlated with visual acuity.

Classical-Contextual Interactions in V1 May Rely on Dendritic Computations

A signature feature of the neocortex is the dense network of horizontal connections (HCs) through which pyramidal neurons (PNs) exchange "contextual" information. In primary visual cortex (V1), HCs are thought to facilitate boundary detection, a crucial operation for object recognition, but how HCs modulate PN responses to boundary cues within their classical receptive fields (CRF) remains unknown. We began by "asking" natural images, through a structured data collection and ground truth labeling process, what function a V1 cell should use to compute boundary probability from aligned edge cues within and outside its CRF. The "answer" was an asymmetric 2-D sigmoidal function, whose nonlinear form provides the first normative account for the "multiplicative" center-flanker interactions previously reported in V1 neurons (Kapadia et al., 1995, 2000; Polat et al., 1998). Using a detailed compartmental model, we then show that this boundary-detecting classical-contextual interaction function can be computed by NMDAR-dependent spatial synaptic interactions within PN dendrites - the site where classical and contextual inputs first converge in the cortex. In additional simulations, we show that local interneuron circuitry activated by HCs can powerfully leverage the nonlinear spatial computing capabilities of PN dendrites, providing the cortex with a highly flexible substrate for integration of classical and contextual information.

Mapping the BCPNN Learning Rule to a Memristor Model

The Bayesian Confidence Propagation Neural Network (BCPNN) has been implemented in a way that allows mapping to neural and synaptic processes in the human cortexandhas been used extensively in detailed spiking models of cortical associative memory function and recently also for machine learning applications. In conventional digital implementations of BCPNN, the von Neumann bottleneck is a major challenge with synaptic storage and access to it as the dominant cost. The memristor is a non-volatile device ideal for artificial synapses that fuses computation and storage and thus fundamentally overcomes the von Neumann bottleneck. While the implementation of other neural networks like Spiking Neural Network (SNN) and even Convolutional Neural Network (CNN) on memristor has been studied, the implementation of BCPNN has not. In this paper, the BCPNN learning rule is mapped to a memristor model and implemented with a memristor-based architecture. The implementation of the BCPNN learning rule is a mixed-signal design with the main computation and storage happening in the analog domain. In particular, the nonlinear dopant drift phenomenon of the memristor is exploited to simulate the exponential decay of the synaptic state variables in the BCPNN learning rule. The consistency between the memristor-based solution and the BCPNN learning rule is simulated and verified in Matlab, with a correlation coefficient as high as 0.99. The analog circuit is designed and implemented in the SPICE simulation environment, demonstrating a good emulation effect for the BCPNN learning rule with a correlation coefficient as high as 0.98. This work focuses on demonstrating the feasibility of mapping the BCPNN learning rule to in-circuit computation in memristor. The feasibility of the memristor-based implementation is evaluated and validated in the paper, to pave the way for a more efficient BCPNN implementation, toward a real-time brain emulation engine.

Ultrastructural, Molecular and Functional Mapping of GABAergic Synapses on Dendritic Spines and Shafts of Neocortical Pyramidal Neurons

The location of GABAergic synapses on dendrites is likely key for neuronal integration. In particular, inhibitory inputs on dendritic spines could serve to selectively veto or modulate individual excitatory inputs, greatly expanding the computational power of individual neurons. To investigate this, we have undertaken a combined functional, molecular, and ultrastructural mapping of the location of GABAergic inputs onto dendrites of pyramidal neurons from upper layers of juvenile mouse somatosensory cortex. Using two-photon uncaging of GABA, intracellular labeling with gerphyrin intrabodies, and focused ion beam milling with scanning electron microscopy, we find that most (96-98%) spines lack GABAergic synapses, although they still display GABAergic responses, potentially due to extrasynaptic GABA receptors. We conclude that GABAergic inputs, in practice, contact dendritic shafts and likely control clusters of excitatory inputs, defining functional zones on dendrites.

Urokinase-Type Plasminogen Activator Protects Cerebral Cortical Neurons from Soluble Aβ-Induced Synaptic Damage

Soluble amyloid β (Aβ)-induced synaptic dysfunction is an early event in the pathogenesis of Alzheimer's disease (AD) that precedes the deposition of insoluble Aβ and correlates with the development of cognitive deficits better than the number of plaques. The mammalian plasminogen activation (PA) system catalyzes the generation of plasmin via two activators: tissue-type (tPA) and urokinase-type (uPA). A dysfunctional tPA-plasmin system causes defective proteolytic degradation of Aβ plaques in advanced stages of AD. In contrast, it is unknown whether uPA and its receptor (uPAR) contribute to the pathogenesis of this disease. Neuronal cadherin (NCAD) plays a pivotal role in the formation of synapses and dendritic branches, and Aβ decreases its expression in cerebral cortical neurons. Here we show that neuronal uPA protects the synapse from the harmful effects of soluble Aβ. However, Aβ-induced inactivation of the eukaryotic initiation factor 2α halts the transcription of uPA mRNA, leaving unopposed the deleterious effects of Aβ on the synapse. In line with these observations, the synaptic abundance of uPA, but not uPAR, is decreased in the frontal cortex of AD patients and 5xFAD mice, and in cerebral cortical neurons incubated with soluble Aβ. We found that uPA treatment increases the synaptic expression of NCAD by a uPAR-mediated plasmin-independent mechanism, and that uPA-induced formation of NCAD dimers protects the synapse from the harmful effects of soluble Aβ oligomers. These data indicate that Aβ-induced decrease in the synaptic abundance of uPA contributes to the development of synaptic damage in the early stages of AD.SIGNIFICANCE STATEMENT Soluble amyloid β (Aβ)-induced synaptic dysfunction is an early event in the pathogenesis of cognitive deficits in Alzheimer's disease (AD). We found that neuronal urokinase-type (uPA) protects the synapse from the deleterious effects of soluble Aβ. However, Aβ-induced inactivation of the eukaryotic initiation factor 2α decreases the synaptic abundance of uPA, leaving unopposed the harmful effects of Aβ on the synapse. In line with these observations, the synaptic expression of uPA is decreased in the frontal cortex of AD brains and 5xFAD mice, and uPA treatment abrogates the deleterious effects of Aβ on the synapse. These results unveil a novel mechanism of Aβ-induced synaptic dysfunction in AD patients, and indicate that recombinant uPA is a potential therapeutic strategy to protect the synapse before the development of irreversible brain damage.

An Indexing Theory for Working Memory Based on Fast Hebbian Plasticity

Working memory (WM) is a key component of human memory and cognition. Computational models have been used to study the underlying neural mechanisms, but neglected the important role of short-term memory (STM) and long-term memory (LTM) interactions for WM. Here, we investigate these using a novel multiarea spiking neural network model of prefrontal cortex (PFC) and two parietotemporal cortical areas based on macaque data. We propose a WM indexing theory that explains how PFC could associate, maintain, and update multimodal LTM representations. Our simulations demonstrate how simultaneous, brief multimodal memory cues could build a temporary joint memory representation as an “index” in PFC by means of fast Hebbian synaptic plasticity. This index can then reactivate spontaneously and thereby also the associated LTM representations. Cueing one LTM item rapidly pattern completes the associated uncued item via PFC. The PFC–STM network updates flexibly as new stimuli arrive, thereby gradually overwriting older representations.

Introducing double bouquet cells into a modular cortical associative memory model

We present an electrophysiological model of double bouquet cells and integrate them into an established cortical columnar microcircuit model that has previously been used as a spiking attractor model for memory. Learning in that model relies on a Hebbian-Bayesian learning rule to condition recurrent connectivity between pyramidal cells. We here demonstrate that the inclusion of a biophysically plausible double bouquet cell model can solve earlier concerns about learning rules that simultaneously learn excitation and inhibition and might thus violate Dale’s principle. We show that learning ability and resulting effective connectivity between functional columns of previous network models is preserved when pyramidal synapses onto double bouquet cells are plastic under the same Hebbian-Bayesian learning rule. The proposed architecture draws on experimental evidence on double bouquet cells and effectively solves the problem of duplexed learning of inhibition and excitation by replacing recurrent inhibition between pyramidal cells in functional columns of different stimulus selectivity with a plastic disynaptic pathway. We thus show that the resulting change to the microcircuit architecture improves the model’s biological plausibility without otherwise impacting the model’s spiking activity, basic operation, and learning abilities.

The Pediatric Cell Atlas: Defining the Growth Phase of Human Development at Single-Cell Resolution

Single-cell gene expression analyses of mammalian tissues have uncovered profound stage-specific molecular regulatory phenomena that have changed the understanding of unique cell types and signaling pathways critical for lineage determination, morphogenesis, and growth. We discuss here the case for a Pediatric Cell Atlas as part of the Human Cell Atlas consortium to provide single-cell profiles and spatial characterization of gene expression across human tissues and organs. Such data will complement adult and developmentally focused HCA projects to provide a rich cytogenomic framework for understanding not only pediatric health and disease but also environmental and genetic impacts across the human lifespan.

Secretagogin is Expressed by Developing Neocortical GABAergic Neurons in Humans but not Mice and Increases Neurite Arbor Size and Complexity

The neocortex of primates, including humans, contains more abundant and diverse inhibitory neurons compared with rodents, but the molecular foundations of these observations are unknown. Through integrative gene coexpression analysis, we determined a consensus transcriptional profile of GABAergic neurons in mid-gestation human neocortex. By comparing this profile to genes expressed in GABAergic neurons purified from neonatal mouse neocortex, we identified conserved and distinct aspects of gene expression in these cells between the species. We show here that the calcium-binding protein secretagogin (SCGN) is robustly expressed by neocortical GABAergic neurons derived from caudal ganglionic eminences (CGE) and lateral ganglionic eminences during human but not mouse brain development. Through electrophysiological and morphometric analyses, we examined the effects of SCGN expression on GABAergic neuron function and form. Forced expression of SCGN in CGE-derived mouse GABAergic neurons significantly increased total neurite length and arbor complexity following transplantation into mouse neocortex, revealing a molecular pathway that contributes to morphological differences in these cells between rodents and primates.

Combining biophysical modeling and deep learning for multielectrode array neuron localization and classification

Neural circuits typically consist of many different types of neurons, and one faces a challenge in disentangling their individual contributions in measured neural activity. Classification of cells into inhibitory and excitatory neurons and localization of neurons on the basis of extracellular recordings are frequently employed procedures. Current approaches, however, need a lot of human intervention, which makes them slow, biased, and unreliable. In light of recent advances in deep learning techniques and exploiting the availability of neuron models with quasi-realistic three-dimensional morphology and physiological properties, we present a framework for automatized and objective classification and localization of cells based on the spatiotemporal profiles of the extracellular action potentials recorded by multielectrode arrays. We train convolutional neural networks on simulated signals from a large set of cell models and show that our framework can predict the position of neurons with high accuracy, more precisely than current state-of-the-art methods. Our method is also able to classify whether a neuron is excitatory or inhibitory with very high accuracy, substantially improving on commonly used clustering techniques. Furthermore, our new method seems to have the potential to separate certain subtypes of excitatory and inhibitory neurons. The possibility of automatically localizing and classifying all neurons recorded with large high-density extracellular electrodes contributes to a more accurate and more reliable mapping of neural circuits. NEW & NOTEWORTHY We propose a novel approach to localize and classify neurons from their extracellularly recorded action potentials with a combination of biophysically detailed neuron models and deep learning techniques. Applied to simulated data, this new combination of forward modeling and machine learning yields higher performance compared with state-of-the-art localization and classification methods.

Cornu Ammonis Regions-Antecedents of Cortical Layers?

Studying neocortex and hippocampus in parallel, we are struck by the similarities. All three to four layered allocortices and the six layered mammalian neocortex arise in the pallium. All receive and integrate multiple cortical and subcortical inputs, provide multiple outputs and include an array of neuronal classes. During development, each cell positions itself to sample appropriate local and distant inputs and to innervate appropriate targets. Simpler cortices had already solved the need to transform multiple coincident inputs into serviceable outputs before neocortex appeared in mammals. Why then do phylogenetically more recent cortices need multiple pyramidal cell layers? A simple answer is that more neurones can compute more complex functions. The dentate gyrus and hippocampal CA regions-which might be seen as hippocampal antecedents of neocortical layers-lie side by side, albeit around a tight bend. Were the millions of cells of rat neocortex arranged in like fashion, the surface area of the CA pyramidal cell layers would be some 40 times larger. Even if evolution had managed to fold this immense sheet into the space available, the distances between neurones that needed to be synaptically connected would be huge and to maintain the speed of information transfer, massive, myelinated fiber tracts would be needed. How much more practical to stack the "cells that fire and wire together" into narrow columns, while retaining the mechanisms underlying the extraordinary precision with which circuits form. This demonstrably efficient arrangement presents us with challenges, however, not the least being to categorize the baffling array of neuronal subtypes in each of five "pyramidal layers." If we imagine the puzzle posed by this bewildering jumble of apical dendrites, basal dendrites and axons, from many different pyramidal and interneuronal classes, that is encountered by a late-arriving interneurone insinuating itself into a functional circuit, we can perhaps begin to understand why definitive classification, covering every aspect of each neurone's structure and function, is such a challenge. Here, we summarize and compare the development of these two cortices, the properties of their neurones, the circuits they form and the ordered, unidirectional flow of information from one hippocampal region, or one neocortical layer, to another.

GABAergic Interneurons in the Neocortex: From Cellular Properties to Circuits

Cortical networks are composed of glutamatergic excitatory projection neurons and local GABAergic inhibitory interneurons that gate signal flow and sculpt network dynamics. Although they represent a minority of the total neocortical neuronal population, GABAergic interneurons are highly heterogeneous, forming functional classes based on their morphological, electrophysiological, and molecular features, as well as connectivity and in vivo patterns of activity. Here we review our current understanding of neocortical interneuron diversity and the properties that distinguish cell types. We then discuss how the involvement of multiple cell types, each with a specific set of cellular properties, plays a crucial role in diversifying and increasing the computational power of a relatively small number of simple circuit motifs forming cortical networks. We illustrate how recent advances in the field have shed light onto the mechanisms by which GABAergic inhibition contributes to network operations.

Synaptic plasticity in dendrites: complications and coping strategies

The elaborate morphology, nonlinear membrane mechanisms and spatiotemporally varying synaptic activation patterns of dendrites complicate the expression, compartmentalization and modulation of synaptic plasticity. To grapple with this complexity, we start with the observation that neurons in different brain areas face markedly different learning problems, and dendrites of different neuron types contribute to the cell's input-output function in markedly different ways. By committing to specific assumptions regarding a neuron's learning problem and its input-output function, specific inferences can be drawn regarding the synaptic plasticity mechanisms and outcomes that we 'ought' to expect for that neuron. Exploiting this assumption-driven approach can help both in interpreting existing experimental data and designing future experiments aimed at understanding the brain's myriad learning processes.

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.

The complexity of the calretinin-expressing progenitors in the human cerebral cortex

The complex structure and function of the cerebral cortex critically depend on the balance of excitation and inhibition provided by the pyramidal projection neurons and GABAergic interneurons, respectively. The calretinin-expressing (CalR⁺) cell is a subtype of GABAergic cortical interneurons that is more prevalent in humans than in rodents. In rodents, CalR⁺ interneurons originate in the caudal ganglionic eminence (CGE) from Gsx2⁺ progenitors, but in humans it has been suggested that a subpopulation of CalR⁺ cells can also be generated in the cortical ventricular/subventricular zone (VZ/SVZ). The progenitors for cortically generated CalR⁺ subpopulation in primates are not yet characterized. Hence, the aim of this study was to identify patterns of expression of the transcription factors (TFs) that commit cortical stem cells to the CalR fate, with a focus on Gsx2. First, we studied the expression of Gsx2 and its downstream effectors, Ascl1 and Sp8 in the cortical regions of the fetal human forebrain at midgestation. Next, we established that a subpopulation of cells expressing these TFs are proliferating in the cortical SVZ, and can be co-labeled with CalR. The presence and proliferation of Gsx2⁺ cells, not only in the ventral telencephalon (GE) as previously reported, but also in the cerebral cortex suggests cortical origin of a subpopulation of CalR⁺ neurons in humans. In vitro treatment of human cortical progenitors with Sonic hedgehog (Shh), an important morphogen in the specification of interneurons, decreased levels of Ascl1 and Sp8 proteins, but did not affect Gsx2 levels. Taken together, our ex-vivo and in vitro results on human fetal brain suggest complex endogenous and exogenous regulation of TFs implied in the specification of different subtypes of CalR⁺ cortical interneurons.

Reactivation in working memory: an attractor network model of free recall

The dynamic nature of human working memory, the general-purpose system for processing continuous input, while keeping no longer externally available information active in the background, is well captured in immediate free recall of supraspan word-lists. Free recall tasks produce several benchmark memory phenomena, like the U-shaped serial position curve, reflecting enhanced memory for early and late list items. To account for empirical data, including primacy and recency as well as contiguity effects, we propose here a neurobiologically based neural network model that unifies short- and long-term forms of memory and challenges both the standard view of working memory as persistent activity and dual-store accounts of free recall. Rapidly expressed and volatile synaptic plasticity, modulated intrinsic excitability, and spike-frequency adaptation are suggested as key cellular mechanisms underlying working memory encoding, reactivation and recall. Recent findings on the synaptic and molecular mechanisms behind early LTP and on spiking activity during delayed-match-to-sample tasks support this view.

Distribution of vesicular glutamate transporter 2 (VGluT2) in the primary visual cortex of the macaque and human

The majority of thalamic terminals in V1 arise from lateral geniculate nucleus (LGN) afferents. Thalamic afferent terminals are preferentially labeled by an isoform of the vesicular glutamate transporter, VGluT2. The goal of our study was to determine the distribution of VGluT2-ir puncta in macaque and human visual cortex. First, we investigated the distribution of VGluT2-ir puncta in all layers of macaque monkey primary visual cortex (V1), and found a very close correspondence between the known distribution of LGN afferents from previous studies and the distribution of VGluT2-immunoreactive (-ir) puncta. There was also a close correspondence between cytochrome oxidase density and VGluT2-ir puncta distribution. After validating the correspondence in macaque, we made a comparative study in human V1. In many aspects, the distribution of VGluT2-ir puncta in human was qualitatively similar to that of the macaque: high densities in layer 4C, patches of VGluT2-ir puncta in the supragranular layer (2/3), lower but clear distribution in layers 1 and 6, and very few puncta in layers 5 and 4B. However, there were also important differences between macaques and humans. In layer 4A of human, there was a sparse distribution of VGluT2-ir puncta, whereas in macaque, there was a dense distribution with the characteristic honeycomb organization. The results suggest important changes in the pattern of cortical VGluT2 immunostaining that may be related to evolutionary differences in the cortical organization of LGN afferents between Old World monkeys and humans.

COUP-TFII expressing interneurons in human fetal forebrain

Transcription factor COUP-TFII in rodents is important for migration of cortical interneurons from caudal ganglionic eminence (CGE) to the neocortex. Since in human, unlike in rodents, cortical interneurons have both ganglionic eminence (GE) and dorsal cortical origin, we studied the distribution of COUP-TFII in the human developing neocortex from 9 to 22 gestational weeks. COUP-TFII is expressed at all stages studied in the GE and in various cortical zones, from the proliferative ventricular/subventricular zone (VZ/SVZ) to layer I. Gradients of COUP-TFII expression are present in the GE, with peak expression in the CGE, and in the neocortex, from high expression in the temporal and occipital cortex to moderate in the frontal and dorsal cortex. Double immunofluorescence with γ-aminobutyric acid (GABA), calretinin, or calbindin, established that subpopulations of interneurons express COUP-TFII. A small fraction of COUP-TFII(+) cells are progenitor cells that proliferate in the CGE (3.4 ± 0.3%) and in the cortical VZ/SVZ (1.7 ± 0.1%). In summary, COUP-TFII is expressed in the human fetal forebrain in GABAergic cells, according to its possible role in migration of cortical interneurons. The source of these cells seems to be the CGE and, to a smaller extent, the cortical VZ/SVZ.

Interneurons in the developing human neocortex

Cortical interneurons play a crucial role in the functioning of cortical microcircuitry as they provide inhibitory input to projection (pyramidal) neurons. Despite their involvement in various neurological and psychiatric disorders, our knowledge about their development in human cerebral cortex is still incomplete. Here we demonstrate that at the beginning of corticogenesis, at embryonic 5 gestation weeks (gw, Carnegie stage 16) in human, early neurons could be labeled with calretinin, calbindin, and GABA antibodies. These immunolabeled cells show a gradient from the ganglionic eminences (GE) toward the neocortex, suggesting that GE is a well conserved source of early born cortical interneurons from rodents to human. At mid-term (20 gw), however, a subset of calretinin(+) cells proliferates in the cortical subventricular zone (SVZ), suggesting a second set of interneuron progenitors that have neocortical origin. Neuropeptide Y, somatostatin, or parvalbumin cells are sparse in mid-term cerebral cortex. In addition to the early source of cortical interneurons in the GE and later in the neocortical SVZ, other regions, such as the subpial granular layer, may also contribute to the population of human cortical interneurons. In conclusion, our findings from cryosections and previous in vitro results suggest that cortical interneuron progenitor population is more complex in humans relative to rodents. The increased complexity of progenitors is probably evolutionary adaptation necessary for development of the higher brain functions characteristic to humans.

GABAergic complex basket formations in the human neocortex

Certain GABAergic interneurons in the cerebral cortex, basket cells, establish multiple connections with cell bodies that typically outline the somata and proximal dendrites of pyramidal cells. During studies into the distribution of the vesicular GABA transporter (VGAT) in the human cerebral cortex, we were struck by the presence of a very dense, pericellular arrangement of multiple VGAT-immunoreactive (-ir) terminals in certain cortical areas. We called these terminals "Complex basket formations" (Cbk-formations) to distinguish them from the simpler and more typical pericellular GABAergic innervations of most cortical neurons. Here we examined the distribution of these VGAT-ir Cbk-formations in various cortical areas, including the somatosensory (area 3b), visual (areas 17 and 18), motor (area 4), associative frontal (dorsolateral areas 9, 10, 45, 46, and orbital areas 11, 12, 13, 14, 47), associative temporal (areas 20, 21, 22, and 38), and limbic cingulate areas (areas 24, 32). Furthermore, we used dual or triple staining techniques to study the chemical nature of the innervated cells. We found that VGAT-ir Cbk-formations were most frequently found in area 4 followed by areas 3b, 13, and 18. In addition, they were mostly observed in layer III, except in area 17, where they were most dense in layer IV. We also found that 70% of the innervated neurons were pyramidal cells, while the remaining 30% were multipolar cells. Most of these multipolar cells expressed the calcium-binding protein parvalbumin and the lectin Vicia villosa agglutinin.

Differential distribution of neurons in the gyral white matter of the human cerebral cortex

The neurons in the cortical white matter (WM neurons) originate from the first set of postmitotic neurons that migrates from the ventricular zone. In particular, they arise in the subplate that contains the earliest cells generated in the telencephalon, prior to the appearance of neurons in gray matter cortical layers. These cortical WM neurons are very numerous during development, when they are thought to participate in transient synaptic networks, although many of these cells later die, and relatively few cells survive as WM neurons in the adult. We used light and electron microscopy to analyze the distribution and density of WM neurons in various areas of the adult human cerebral cortex. Furthermore, we examined the perisomatic innervation of these neurons and estimated the density of synapses in the white matter. Finally, we examined the distribution and neurochemical nature of interneurons that putatively innervate the somata of WM neurons. From the data obtained, we can draw three main conclusions: first, the density of WM neurons varies depending on the cortical areas; second, calretinin-immunoreactive neurons represent the major subpopulation of GABAergic WM neurons; and, third, the somata of WM neurons are surrounded by both glutamatergic and GABAergic axon terminals, although only symmetric axosomatic synapses were found. By contrast, both symmetric and asymmetric axodendritic synapses were observed in the neuropil. We discuss the possible functional implications of these findings in terms of cortical circuits.

Dorsal radial glial cells have the potential to generate cortical interneurons in human but not in mouse brain

Radial glial (RG) cells, in the neocortical ventricular/subventricular zone (VZ/SVZ), generate cortical projection neurons both in rodents and humans, but whether they can also generate cortical interneurons is not clear. We demonstrated both on cryosections and in cell cultures that in the human VZ/SVZ, cells can be double labeled with RG markers and calretinin (CalR) and GABA, markers that suggest interneuronal lineage. We examined in more detail the cell fate of human RG cells isolated from the VZ/SVZ at midterm. After 24 h, no CalR(+) or GABA(+) cells were seen in cultures, whereas 5-10% cells expressed Nkx2.1 and Dlx, two ventral transcription factors. CalR(+) and GABA(+) cells were apparent for the first time after 3 d in vitro, and their number increased in subsequent days, consistent with the gradual transition of RG cells into CalR(+) or GABA(+) cells. Indeed, the progeny of genetically labeled RG cells could be immunolabeled with antibodies to CalR and GABA or ventral transcription factors (Nkx2.1(+), Dlx(+)). In contrast to humans, in the embryonic mouse, similar experiments showed that only RG cells isolated from the subpallium (ganglionic eminence) generate CalR(+) or GABA(+) cells, whereas this was not the case with RG cells isolated from the pallium. These findings support the idea that human, but not mouse, dorsal RG cells have the potential to generate various subtypes of neocortical interneurons. Multiple progenitors and sites of cortical interneuron origin in human might be an evolutionary adaptation underlying brain expansion and the increased complexity of cortical circuitry in humans.

A cortical sparse distributed coding model linking mini- and macrocolumn-scale functionality

No generic function for the minicolumn – i.e., one that would apply equally well to all cortical areas and species – has yet been proposed. I propose that the minicolumn does have a generic functionality, which only becomes clear when seen in the context of the function of the higher-level, subsuming unit, the macrocolumn. I propose that: (a) a macrocolumn's function is to store sparse distributed representations of its inputs and to be a recognizer of those inputs; and (b) the generic function of the minicolumn is to enforce macrocolumnar code sparseness. The minicolumn, defined here as a physically localized pool of ∼20 L2/3 pyramidals, does this by acting as a winner-take-all (WTA) competitive module, implying that macrocolumnar codes consist of ∼70 active L2/3 cells, assuming ∼70 minicolumns per macrocolumn. I describe an algorithm for activating these codes during both learning and retrievals, which causes more similar inputs to map to more highly intersecting codes, a property which yields ultra-fast (immediate, first-shot) storage and retrieval. The algorithm achieves this by adding an amount of randomness (noise) into the code selection process, which is inversely proportional to an input's familiarity. I propose a possible mapping of the algorithm onto cortical circuitry, and adduce evidence for a neuromodulatory implementation of this familiarity-contingent noise mechanism. The model is distinguished from other recent columnar cortical circuit models in proposing a generic minicolumnar function in which a group of cells within the minicolumn, the L2/3 pyramidals, compete (WTA) to be part of the sparse distributed macrocolumnar code.

A comparative perspective on minicolumns and inhibitory GABAergic interneurons in the neocortex

Neocortical columns are functional and morphological units whose architecture may have been under selective evolutionary pressure in different mammalian lineages in response to encephalization and specializations of cognitive abilities. Inhibitory interneurons make a substantial contribution to the morphology and distribution of minicolumns within the cortex. In this context, we review differences in minicolumns and GABAergic interneurons among species and discuss possible implications for signaling among and within minicolumns. Furthermore, we discuss how abnormalities of both minicolumn disposition and inhibitory interneurons might be associated with neuropathological processes, such as Alzheimer's disease, autism, and schizophrenia. Specifically, we explore the possibility that phylogenetic variability in calcium-binding protein-expressing interneuron subtypes is directly related to differences in minicolumn morphology among species and might contribute to neuropathological susceptibility in humans.

Primate-specific origins and migration of cortical GABAergic neurons

Gamma-aminobutyric-acidergic (GABAergic) cells form a very heterogeneous population of neurons that play a crucial role in the coordination and integration of cortical functions. Their number and diversity increase through mammalian brain evolution. Does evolution use the same or different developmental rules to provide the increased population of cortical GABAergic neurons? In rodents, these neurons are not generated in the pallial proliferative zones as glutamatergic principal neurons. They are produced almost exclusively by the subpallial proliferative zones, the ganglionic eminence (GE) and migrate tangentially to reach their target cortical layers. The GE is organized in molecularly different subdomains that produce different subpopulations of cortical GABAergic neurons. In humans and non-human primates, in addition to the GE, cortical GABAergic neurons are also abundantly generated by the proliferative zones of the dorsal telencephalon. Neurogenesis in ventral and dorsal telencephalon occurs with distinct temporal profiles. These dorsal and ventral lineages give rise to different populations of GABAergic neurons. Early-generated GABAergic neurons originate from the GE and mostly migrate to the marginal zone and the subplate. Later-generated GABAergic neurons, originating from both proliferative sites, populate the cortical plate. Interestingly, the pool of GABAergic progenitors in dorsal telencephalon produces mainly calretinin neurons, a population known to be significantly increased and to display specific features in primates. We conclude that the development of cortical GABAergic neurons have exclusive features in primates that need to be considered in order to understand pathological mechanisms leading to some neurological and psychiatric diseases.

Selective depletion of molecularly defined cortical interneurons in human holoprosencephaly with severe striatal hypoplasia

Cortical excitatory glutamatergic projection neurons and inhibitory GABAergic interneurons follow substantially different developmental programs. In rodents, projection neurons originate from progenitors within the dorsal forebrain, whereas interneurons arise from progenitors in the ventral forebrain. In contrast, it has been proposed that in humans, the majority of cortical interneurons arise from progenitors within the dorsal forebrain, suggesting that their origin and migration is complex and evolutionarily divergent. However, whether molecularly defined human cortical interneuron subtypes originate from distinct progenitors, including those in the ventral forebrain, remains unknown. Furthermore, abnormalities in cortical interneurons have been linked to human disorders, yet no distinct cell population selective loss has been reported. Here we show that cortical interneurons expressing nitric oxide synthase 1, neuropeptide Y, and somatostatin, are either absent or substantially reduced in fetal and infant cases of human holoprosencephaly (HPE) with severe ventral forebrain hypoplasia. Notably, another interneuron subtype normally abundant from the early fetal period, marked by calretinin expression, and different subtypes of projection neuron were present in the cortex of control and HPE brains. These findings have important implications for the understanding of neuronal pathogenesis underlying the clinical manifestations associated with HPE and the developmental origins of human cortical interneuron diversity.

Associative memory models: from the cell-assembly theory to biophysically detailed cortex simulations

The second half of the past century saw the emergence of a theory of cortical associative memory function originating in Donald Hebb's hypotheses on activity-dependent synaptic plasticity and cell-assembly formation and dynamics. This conceptual framework has today developed into a theory of attractor memory that brings together many experimental observations from different sources and levels of investigation into computational models displaying information-processing capabilities such as efficient associative memory and holistic perception. Here, we outline a development that might eventually lead to a neurobiologically grounded theory of cortical associative memory.

Role of interneuron diversity in the cortical microcircuit for attention

Receptive fields of neurons in cortical area V4 are large enough to fit multiple stimuli, making V4 the ideal place to study the effects of selective attention at the single-neuron level. Experiments have revealed evidence for stimulus competition and have characterized the effect thereon of spatial and feature-based attention. We developed a biophysical model with spiking neurons and conductance-based synapses. To account for the comprehensive set of experimental results, it was necessary to include in the model, in addition to regular spiking excitatory (E) cells, two types of interneurons: feedforward interneurons (FFI) and top-down interneurons (TDI). Feature-based attention was mediated by a projection of the TDI to the FFI, stimulus competition was mediated by a cross-columnar excitatory connection to the FFI, whereas spatial attention was mediated by an increase in activity of the feedforward inputs from cortical area V2. The model predicts that spatial attention increases the FFI firing rate, whereas feature-based attention decreases the FFI firing rate and increases the TDI firing rate. During strong stimulus competition, the E cells were synchronous in the beta frequency range (15-35 Hz), but with feature-based attention, they became synchronous in the gamma frequency range (35-50 Hz). We propose that the FFI correspond to fast-spiking, parvalbumin-positive basket cells and that the TDI correspond to cells with a double-bouquet morphology that are immunoreactive to calbindin or calretinin. Taken together, the model results provide an experimentally testable hypothesis for the behavior of two interneuron types under attentional modulation.

Recursive trace line method for detecting myelinated bundles: a comparison study with pyramidal cell arrays

Minicolumns are thought to be the smallest cortical modules within the hierarchical organization of the isocortex. Several reports suggest alterations in minicolumnar morphometry may be involved in psychiatric disorders such as autism, dyslexia, and schizophrenia. Thus far anatomical studies of minicolumns have primarily relied on measurements of pyramidal cell arrays. This study expands on a recursive trace line segmentation method used to define morphometric measures for myelinated axon bundles. The results were compared against those of pyramidal cell arrays derived from immediately adjacent serial sections. Width estimates based on cell somas and myelinated axon bundles were highly correlated (r=0.9888). Histograms of signal intensity using the recursive trace line method produced expected features of myeloarchitectonics; that is, bundles of Meynert and intervening interradiary plexus. The close correspondence of derived values for myelinated axon bundles and pyramidal cell arrays suggests their participation and interaction within the same modular arrangement of the isocortex.