Complex microstructures of sensory cortical connections

General Anesthesia Decouples Cortical Pyramidal Neurons

The mystery of general anesthesia is that it specifically suppresses consciousness by disrupting feedback signaling in the brain, even when feedforward signaling and basic neuronal function are left relatively unchanged. The mechanism for such selectiveness is unknown. Here we show that three different anesthetics have the same disruptive influence on signaling along apical dendrites in cortical layer 5 pyramidal neurons in mice. We found that optogenetic depolarization of the distal apical dendrites caused robust spiking at the cell body under awake conditions that was blocked by anesthesia. Moreover, we found that blocking metabotropic glutamate and cholinergic receptors had the same effect on apical dendrite decoupling as anesthesia or inactivation of the higher-order thalamus. If feedback signaling occurs predominantly through apical dendrites, the cellular mechanism we found would explain not only how anesthesia selectively blocks this signaling but also why conscious perception depends on both cortico-cortical and thalamo-cortical connectivity.

Anisotropy of ongoing neural activity in the primate visual cortex

The mammalian neocortex features distinct anatomical variation in its tangential and radial extents. This review consolidates previously published findings from our group in order to compare and contrast the spatial profile of neural activity coherence across these distinct cortical dimensions. We focus on studies of ongoing local field potential (LFP) data obtained simultaneously from multiple sites in the primary visual cortex in two types of experiments in which electrode contacts were spaced either along the cortical surface or at different laminar positions. These studies demonstrate that across both dimensions the coherence of ongoing LFP fluctuations diminishes as a function of interelectrode distance, although the nature and spatial scale of this falloff is very different. Along the cortical surface, the overall LFP coherence declines gradually and continuously away from a given position. In contrast, across the cortical layers, LFP coherence is discontinuous and compartmentalized as a function of depth. Specifically, regions of high LFP coherence fall into discrete superficial and deep laminar zones, with an abrupt discontinuity between the granular and infragranular layers. This spatial pattern of ongoing LFP coherence is similar when animals are at rest and when they are engaged in a behavioral task. These results point to the existence of partially segregated laminar zones of cortical processing that extend tangentially within the laminar compartments and are thus oriented orthogonal to the cortical columns. We interpret these electrophysiological observations in light of the known anatomical organization of the cortical microcircuit.

Enhancement of neural representation capacity by modular architecture in networks of cortical neurons

Biological networks are ubiquitously modular, a feature that is believed to be essential for the enhancement of their functional capacities. Here, we have used a simple modular in vitro design to examine the possibility that modularity enhances network functionality in the context of input representation. We cultured networks of cortical neurons obtained from newborn rats in vitro on substrate-integrated multi-electrode arrays, forcing the network to develop two well-defined modules of neural populations that are coupled by a narrow canal. We measured the neural activity, and examined the capacity of each module to individually classify (i.e. represent) spatially distinct electrical stimuli and propagate input-specific activity features to their downstream coupled counterpart. We show that, although each of the coupled modules maintains its autonomous functionality, a significant enhancement of representational capacity is achieved when the system is observed as a whole. We interpret our results in terms of a relative decorrelation effect imposed by weak coupling between modules.

Cortical region interactions and the functional role of apical dendrites

The basal and distal apical dendrites of pyramidal cells occupy distinct cortical layers and are targeted by axons originating in different cortical regions. Hence, apical and basal dendrites receive information from distinct sources. Physiological evidence suggests that this anatomically observed segregation of input sources may have functional significance. This possibility has been explored in various connectionist models that employ neurons with functionally distinct apical and basal compartments. A neuron in which separate sets of inputs can be integrated independently has the potential to operate in a variety of ways not possible for the conventional neuron model, in which all inputs are treated equally. This article thus considers how functionally distinct apical and basal dendrites can contribute to the information-processing capacities of single neurons and, in particular, how information from different cortical regions could have disparate effects on neural activity and learning.

Differential dynamics of transient neuronal assemblies in visual compared to auditory cortex

Large-scale, coherent, but highly transient networks of neurons, 'neuronal assemblies', operate over a sub-second time frame. Such assemblies of brain cells need not necessarily respect well-defined anatomical compartmentalisation, but represent an intermediate level of brain organisation between identified brain regions and individual neurons dependent on the activity status of the synaptic connections and axonal projections. To study neuronal assemblies both in slices and in the living brain, optical imaging using voltage-sensitive dyes (VSDI) offers the highest spatial and temporal resolution in real-time. Applying VSDI technique to compare assemblies in visual versus auditory cortices under standardised experimental protocols, we observed no significant variations in the basic parameters of fluorescence signal and assembly size: such results might be predicted from the canonical invariance of cortical structures across modalities. However, further analysis revealed less obvious yet significant differences in the assembly dynamics of the two regions. The neural assemblies spread widely across layers in the two cortices following paired-pulse stimulation of putative layer 4. The respective patterns of activity started to differentiate within a specific time frame (250-300 ms). The signal was predominant near the point of stimulation in the visual cortex, whereas in the auditory cortex the signal was stronger in the superficial layers. This modality-specific divergence in assembly dynamics highlights a previously under-appreciated level of neuronal processing. Additionally, these findings could prompt a new approach to the understanding of how information from different senses, transmitted as action potentials with identical electrochemical characteristics across different cortices, be it visual or auditory, can eventually yield, nonetheless, the qualitatively distinct experiences of seeing or hearing.

Neocortical areas, layers, connections, and gene expression

Cortical patterns of gene expression provide a new approach to long standing issues of lamination, and area identity and formation. In this review, we summarize recent findings where molecular biological techniques have revealed a small number of area-specific genes in the nonhuman primate cortex. One of these (occ1) is strongly expressed in primary visual cortex and is associated with thalamocortical connections. Another gene, RBP, is more strongly expressed in association areas. It is not clear whether RBP might be linked with any particular connectional system, but several possibilities are raised. We also discuss possible roles of area-specific genes in postnatal development, and conclude with a brief sketch of future directions.

Inter- and intra-laminar connections of pyramidal cells in the neocortex

The flow of excitation through cortical columns has long since been predicted by studying the axonal projection patterns of excitatory neurones situated within different laminae. In grossly simplified terms and assuming random connectivity, such studies predict that input from the thalamus terminates primarily in layer 4, is relayed 'forward' to layer 3, then to layers 5 and 6 from where the modified signal may exit the cortex. Projection patterns also indicate 'back' projections from layer 5 to 3 and layer 6 to 4. More recently it has become clear that the interconnections between these layers are not random; forward projections primarily contact specific pyramidal subclasses and intracortical back projections innervate interneurones. This indicates that presynaptic axons or postsynaptic dendrites are capable of selecting their synaptic partners and that this selectivity is layer dependent. For the past decade, we and others have studied pyramidal cell targeting in circuits both within, and between laminae using paired intracellular recordings with biocytin filling and have begun to identify further levels of selectivity through the preferential targeting of electrophysiologically and/or morphologically distinct pyramidal subtypes. Presented here, therefore, is a brief overview of current thinking on the layer and subclass specific connectivity of neocortical principle excitatory cells.

Zinc-enriched amygdalo- and hippocampo-cortical connections to the inferotemporal cortices in macaque monkey

Synaptic zinc (Zn), a co-factor in some glutamatergic synapses, has been implicated in plasticity effects, as well as in several excitotoxic and other pathophysiological conditions. In this study, we provide information about the distribution of Zn in inferotemporal cortex, a region at the interface of the visual and hippocampal networks. In brief, we found a lateral to medial increase in Zn, where TEad, a unimodal visual area, showed low levels of Zn; TEav, intermediate levels; and perirhinal cortex, a multimodal limbic area, high levels. The distribution of parvalbumin, a calcium binding protein, showed a reverse gradient to that of Zn. The neurons of origin of the Zn+ termination were identified by making intracortical injections sodium selenite (Na2SeO3). This substance interacts with Zn to form precipitates of ZnSe and in this form is transported retrogradely to the soma. A mixed population of labeled neurons was visualized, which included Zn+ neurons in CA1 of the hippocampus and in several amygdala subnuclei. In CA1, Zn+ neurons were restricted to the upper part of stratum pyramidale. Zn is thought to contribute to activity-dependent synaptic plasticity. The specifically high level in perirhinal cortex, and its origin from neurons in CA1 and the amygdala, may relate to cellular events involved in visual long-term memory formation.

Disruption of layer 4 development alters laminar processing in ferret somatosensory cortex

Treatment with the anti-mitotic agent methylazoxymethanol (MAM) on embryonic day 33 (E33) in ferrets changes features of somatosensory cortex. These include dramatic reduction of cells in layer 4, and altered distributions of thalamocortical afferent terminations and GABA(A) receptors. To determine the effect of the relative absence of layer 4 on processing of sensory stimuli we used current source-density profiles to assess laminar activity patterns. Nearly synchronous activation occurs across all layers in treated animals, which contrasts with the normal cortical activation pattern of initial sinks in layer 4. This change after MAM treatment is consistent with the absence of layer 4 cells and widespread termination of thalamocortical afferents. Using periodic stimulation at 'flutter' frequency, layer 4 neurons in normal somatosensory cortex fire reproducibly to the stimulus rate; the capacity for entrainment is best for layer 4 and weaker in the extragranular layers. The capacity to encode periodic sensory stimuli is disrupted in MAM-treated somatosensory cortex; after an initial response to the onset of periodic stimuli, neurons in all cortical layers show weak entrainment. Neural responses to sensory drive in E33 MAM-treated cortex are also embedded in levels of neural activity substantially above those in normal somatosensory cortex. Sustained stimulation additionally reveals different capacities in each layer for improved signal-to-noise ratios, with layer 4 neurons in normal animals exhibiting the most improved signaling over time. We conclude that normal thalamic terminations, an intact layer 4 and subsequent intracortical processing are integral to proper encoding of stimulus features.

A feedback model of visual attention

Feedback connections are a prominent feature of cortical anatomy and are likely to have a significant functional role in neural information processing. We present a neural network model of cortical feedback that successfully simulates neurophysiological data associated with attention. In this domain, our model can be considered a more detailed, and biologically plausible, implementation of the biased competition model of attention. However, our model is more general as it can also explain a variety of other top-down processes in vision, such as figure/ground segmentation and contextual cueing. This model thus suggests that a common mechanism, involving cortical feedback pathways, is responsible for a range of phenomena and provides a unified account of currently disparate areas of research.

Topology and graph theory applied to cortical anatomy may help explain working memory capacity for three or four simultaneous items

Cognitive experimentation suggests that at any single instant only three or four items ("chunks") are simultaneously prominent as a working memory (WM) trace, if we disregard the rehearsal component of WM. The reason for small WM capacity may concern combinatorial manageability. How might the neural representations of these few coactive chunks occupy a spatially distributed set of areas of the sheet-like cortex, while providing both order and flexibility to associate items in WM? Each attribute of each simultaneously active WM item must have broad access to the representational facilities of the cortical sheet, comprising tens of thousands of modular "cortical columns." The two hypothesized neural levels of WM during any moment of cognition comprise (a) "binding" together of many distributed attribute representations within each respective WM chunk, and (b) combinatorial play among three or four WM chunk-representations. Anatomical and functional evidence of cortical unity through its depth suggests that cortex may be viewed as essentially planar in its distribution of activations. Thus, a moment's WM is hypothesized here to reside in myriad activated cortical planar "patches," each subdivided into up to four amoeboid "subpatches." Two different lines of topological reasoning suggest orderly associations of such representations. (1) The four-color principle of map topology, and the related K(4) is planar theorem of graph theory, imply that if a small cortical area is dynamically subdivided into no more than four, discretely bounded planar subareas, then each such segment has ample free access to each of the others. (2) A hypothetical alternative to such associative adjacency of simultaneously active cortical representations of chunk-attributes is associative overlap, whereby, in dense cortical neuropil, activated subpatches behave like Venn diagrams of intersecting sets. As the number of Venn-like coactive subpatches within a patch increases, maintaining ad hoc associativity among all combinations requires exponentially proliferating intersections. Beyond four, serpentine subpatch shapes are required, which could easily lead to pathologies of omission or commission. As hypothesized by many researchers, the binding of the widely distributed cortical modules that represent a given chunk may involve synchrony or coherence of a single EEG frequency. Elsewhere, I have conjectured that such a binding frequency for a single chunk may bear a harmonic relationship with the additional EEG frequencies that are simultaneously binding the other WM chunks. Other possible mechanisms of binding have also been hypothesized. Whatever the mechanism, the many attributes of a moment's complement of three or four WM chunks must generally have an accidental relationship with the spatial distribution of the cortical feature analyzers that must be activated to represent those attributes. Therefore, the cortex may need, and have, comprehensive anatomical connections of each of its modules for representing an attribute (or of small redundant module groupings) with every other. If such whole-part cortico-cortical connections are somehow exploited not only to fully represent each cognitive chunk in its bound-together attributes, but also to bring the major business of intensive WM information processing down to the level of local circuits, in the sorts of topological patterning hypothesized here, there may be two adaptive results: (1) Time and other economies would be achieved in the reduction of activity in distant cortico-cortical connections to lower-energy global orchestration, or binding processes. (2) The piecemeal local topological limit to four subpatches would be writ large, across the entire cortex, preventing an unconstrained combinatorial explosion of associations among all attributes of all three or four simultaneously active chunks. Such hypothetical convergence to foci in local subpatch interactions might take place primarily in association cortex, and/or it might involve temporary shifts in response properties in some cortical subpat might involve temporary shifts in response properties in some cortical subpatches. Quantitative studies of the densely packed cortical fine structure, by Braitenberg and Schüz, and others, seem potentially consistent with this vision of cortical function in cognition.

Honeycomb-like mosaic at the border of layers 1 and 2 in the cerebral cortex

In this report, we present evidence of a small-scale modularity (<100 microm) at the border of layers 1 and 2 in neocortical areas. The modularity is best seen in tangential sections, with double-labeling immunohistochemistry to reveal overlapping or complementary relationships of different markers. The pattern is overall like a reticulum or mosaic but is described as a "honeycomb," in which the walls and hollows are composed of distinct afferent and dendritic systems. We demonstrate the main components of the honeycomb in rat visual cortex. These are as follows: (1) zinc-enriched, corticocortical terminations in the walls, and in the hollows, thalamocortical terminations (labeled by antibody against vesicular glutamate transporter 2 and by cytochrome oxidase); (2) parvalbumin-dense neuropil in the walls that partly colocalizes with elevated levels of glutamate receptors 2/3, NMDAR receptor 1, and calbindin; and (3) dendritic subpopulations preferentially situated within the walls (dendrites of layer 2 neurons) or hollows (dendrites of deeper neurons in layers 3 and 5). Because the micromodularity is restricted to layers 2 and 1b, without extending into layer 3, this may be another indication of a laminar-specific substructure at different spatial scales within cortical columns. The suggestion is that corticocortical and thalamocortical terminations constitute parallel circuits at the level of layer 2, where they are segregated in association with distinct dendritic systems. Results from parvalbumin staining show that the honeycomb mosaic is not limited to rat visual cortex but can be recognized at the layer 1-2 border in other areas and species.

Category-sensitive excitatory and inhibitory processes in human extrastriate cortex

Single-cell recordings from the temporal lobe of monkeys viewing stimuli show that cells may be highly selective, responding for example to particular objects such as faces. However, stimulus-selective cells may be inhibited by nonpreferred stimuli. Can such inhibitory mechanisms be detected in human visual cortex? In previous recordings from the surface of human ventral extrastriate cortex, we found that specific categories of stimuli such as faces and words generate category-specific negative event-related potentials (ERPs) with a peak latency of about 200 ms (N200). Laminar recordings in animal cortex suggest that the human N200 reflects excitatory depolarizing potentials in apical dendrites of pyramidal cells. In this study we found that, at about half of word-specific N200 sites, faces generated a positive ERP (P200); conversely, at about half of face-specific sites, words generated P200s. The electrogenesis of N200 implies that P200 ERPs reflect hyperpolarizing inhibition of apical dendrites. These recordings, together with the prior animal recordings, provide strong circumstantial evidence that in human cortex populations of cells responsive to one stimulus category (such as faces) inhibit cells responsive to another category (such as words), probably by a type of lateral inhibition. Of the stimulus categories studied quantitatively, face-specific cells are maximally inhibited by words and vice versa, but other categories of stimuli may generate smaller P200s, suggesting that inhibition of category-specific cells by nonpreferred stimuli is a general feature of human extrastriate cortex involved in object recognition.

The modular organization of brain systems. Basal forebrain: the last frontier

Computational anatomical studies suggest that specific clusters of projection neurons in the basal forebrain together with specific prefrontal and posterior cortical associational regions constitute distributed parts of functional parallel circuits. The predictable sequence of cell clusters consisting of various types of noncholinergic cell populations in the basal forebrain suggests further subdivisions within these circuits. It is possible that similar to the parallel basal ganglia circuits (Alexander and Crutcher, 1990), large number of specialized channels and sub-channels exist within this triangular circuitry that permit parallel, multilevel processing concurrently. The location and size of the active modules may temporarily vary according to the prevalence of state-related diffuse ascending brain stem and specific telencephalic inputs. From this latter group of afferents, the prefrontal input may function as an external threshold control which allocates attentional resources via the basal forebrain to distributed cortical processes in a selective, self-regulatory fashion.

Preintegration lateral inhibition enhances unsupervised learning

A large and influential class of neural network architectures uses postintegration lateral inhibition as a mechanism for competition. We argue that these algorithms are computationally deficient in that they fail to generate, or learn, appropriate perceptual representations under certain circumstances. An alternative neural network architecture is presented here in which nodes compete for the right to receive inputs rather than for the right to generate outputs. This form of competition, implemented through preintegration lateral inhibition, does provide appropriate coding properties and can be used to learn such representations efficiently. Furthermore, this architecture is consistent with both neuroanatomical and neurophysiological data. We thus argue that preintegration lateral inhibition has computational advantages over conventional neural network architectures while remaining equally biologically plausible.

Synaptic connections between layer 4 spiny neurone-layer 2/3 pyramidal cell pairs in juvenile rat barrel cortex: physiology and anatomy of interlaminar signalling within a cortical column

Whole-cell voltage recordings were obtained from 64 synaptically coupled excitatory layer 4 (L4) spiny neurones and L2/3 pyramidal cells in acute slices of the somatosensory cortex ('barrel' cortex) of 17- to 23-days-old rats. Single action potentials (APs) in the L4 spiny neurone evoked single unitary EPSPs in the L2/3 pyramidal cell with a peak amplitude of 0.7 +/- 0.6 mV. The average latency was 2.1 +/- 0.6 ms, the rise time was 0.8 +/- 0.3 ms and the decay time constant was 12.7 +/- 3.5 ms. The percentage of failures of an AP in a L4 spiny neurone to evoke a unitary EPSP in the L2/3 pyramidal cell was 4.9 +/- 8.8 % and the coefficient of variation (c.v.) of the unitary EPSP amplitude was 0.27 +/- 0.13. Both c.v. and percentage of failures decreased with increased average EPSP amplitude. Postsynaptic glutamate receptors (GluRs) in L2/3 pyramidal cells were of the N-methyl-D-aspartate (NMDA) receptor (NMDAR) and the non-NMDAR type. At -60 mV in the presence of extracellular Mg2+ (1 mM), 29 +/- 15 % of the EPSP voltage-time integral was blocked by NMDAR antagonists. In 0 Mg2+, the NMDAR/AMPAR ratio of the EPSC was 0.50 +/- 0.29, about half the value obtained for L4 spiny neurone connections. Burst stimulation of L4 spiny neurones showed that EPSPs in L2/3 pyramidal cells depressed over a wide range of frequencies (1-100 s(-1) ). However, at higher frequencies (30 s(-1)) EPSP summation overcame synaptic depression so that the summed EPSP was larger than the first EPSP amplitude in the train. The number of putative synaptic contacts established by the axonal collaterals of the L4 projection neurone with the target neurone in layer 2/3 varied between 4 and 5, with an average of 4.5 +/- 0.5 (n = 13 pairs). Synapses were established on basal dendrites of the pyramidal cell. Their mean geometric distance from the pyramidal cell soma was 67 +/- 34 microm (range, 16-196 microm). The results suggest that each connected L4 spiny neurone produces a weak but reliable EPSP in the pyramidal cell. Therefore transmission of signals to layer 2/3 is likely to have a high threshold requiring simultaneous activation of many L4 neurons, implying that L4 spiny neurone to L2/3 pyramidal cell synapses act as a gate for the lateral spread of excitation in layer 2/3.

Columnar organization of dendrites and axons of single and synaptically coupled excitatory spiny neurons in layer 4 of the rat barrel cortex

Cortical columns are the functional units of the neocortex that are particularly prominent in the "barrel" field of the somatosensory cortex. Here we describe the morphology of two classes of synaptically coupled excitatory neurons in layer 4 of the barrel cortex, spiny stellate, and star pyramidal cells, respectively. Within a single barrel, their somata tend to be organized in clusters. The dendritic arbors are largely confined to layer 4, except for the distal part of the apical dendrite of star pyramidal neurons that extends into layer 2/3. In contrast, the axon of both types of neurons spans the cortex from layer 1 to layer 6. The most prominent axonal projections are those to layers 4 and 2/3 where they are largely restricted to a single cortical column. In layers 5 and 6, a small fraction of axon collaterals projects also across cortical columns. Consistent with the dense axonal projection to layers 4 and 2/3, the total number and density of boutons per unit axonal length was also highest there. Electron microscopy combined with GABA postimmunogold labeling revealed that most (>90%) of the synaptic contacts were established on dendritic spines and shafts of excitatory neurons in layers 4 and 2/3. The largely columnar organization of dendrites and axons of both cell types, combined with the preferential and dense projections within cortical layers 4 and 2/3, suggests that spiny stellate and star pyramidal neurons of layer 4 serve to amplify thalamic input and relay excitation vertically within a single cortical column.

Laminar sources of synaptic input to cortical inhibitory interneurons and pyramidal neurons

The functional role of an individual neuron within a cortical circuit is largely determined by that neuron's synaptic input. We examined the laminar sources of local input to subtypes of cortical neurons in layer 2/3 of rat visual cortex using laser scanning photostimulation. We identified three distinct laminar patterns of excitatory input that correspond to physiological and morphological subtypes of neurons. Fast-spiking inhibitory basket cells and excitatory pyramidal neurons received strong excitatory input from middle cortical layers. In contrast, adapting inhibitory interneurons received their strongest excitatory input either from deep layers or laterally from within layer 2/3. Thus, differential laminar sources of excitatory inputs contribute to the functional diversity of cortical inhibitory interneurons.

Cross-modal reorganization of horizontal connectivity in auditory cortex without altering thalamocortical projections

The development of the different, highly specialized regions of the mammalian cerebral cortex depends in part on neural activity, either intrinsic spontaneous activity or externally driven sensory activity. To determine whether patterned sensory activity instructs the development of intrinsic cortical circuitry, we have experimentally altered the modality of sensory inputs to cerebral cortex. Neonatal diversion of retinal axons to the auditory thalamus (cross-modal rewiring) results in a primary auditory cortex (AI) that resembles visual cortex in its response properties and topography (Roe et al., 1990, 1992). To test the hypothesis that the visual response properties are created by a visually driven reorganization of auditory cortical circuitry, we investigated the effect of early visual experience on the development of intrinsic, horizontal connections within AI. Horizontal connections are likely to play an important role in the construction of visual response properties in AI as they do in visual cortex. Here we show that early visual inputs to auditory thalamus can reorganize horizontal connections in AI, causing both an increase in their extent and a change in pattern, so that projections are not restricted to the isofrequency axis, but extend in a more isotropic pattern around the injection site. Thus, changing afferent modality, without altering the source of the thalamocortical axons, can profoundly alter cortical circuitry. Similar changes may underlie cortical compensatory processes in deaf or blind humans and may also have played a role in the parcellation of neocortex during mammalian evolution.