From single cells and simple circuits to large neuronal networks

Many of the different types of neocortical neurons have already been described more than 100 years ago by Ramón y Cajal (1904). However, the morphological and functional properties of the neuronal microcircuitry in which they are embedded are still rather unknown. In a first attempt, Szentágothai (1978) proposed a hypothetical model of neuronal networks within a cortical column purely based on morphological data that included both principal (excitatory) neurons and GABAergic (inhibitory) interneurons known at that time (Fig. 1). A major drawback of this model circuitry was the limited information concerning the functional properties of individual neurons and their synaptic connections.

            In the barrel cortex, a cortical column contains approximately 10,000 neurons distributed over six cortical layers. They fall into two major classes: excitatory principal neurons, pyramidal and spiny stellate cells that constitute 80-85% of the neuronal population providing most of the local, cortico-cortical and extra-cortical projections and a heterogeneous population of GABAergic interneurons (Somogyi et al., 1998;Cauli et al., 2000;Gupta et al., 2000;Ascoli et al., 2008) distributed throughout cortical layers 1-6. Due to the superposition of many neuronal microcircuits within the columnar network it is necessary to determine the structure and function of individual synaptic connections and the functional role of distinct inhibitory interneurons to understand the signal processing within a cortical column.

            Neuronal connectivity, the reliability and efficacy of synaptic connections together with morphological characteristics of the excitatory neurons such as the extent of their dendritic arbor, their axonal projection pattern and the number and location of synaptic contacts determine the flow of excitatory signals in the cortical column. Within the barrel column, specific synaptic connections have been shown to exist that are either local (intracolumnar, i.e. largely within a barrel column) or long-range (transcolumnar; projecting across several barrel columns), intralaminar (within a respective cortical layer) or translaminar (between cortical layers) in nature. Furthermore, the direction of synaptic signaling can be either unidirectional as in cortical relay circuits, or reciprocal, i.e. both neurons in a pair act as pre- and postsynaptic neurons.

            Most synaptic connections studied so far in the neocortex and specifically in the barrel cortex are local and intralaminar (Lübke and Feldmeyer, 2007;Schubert et al., 2007;Thomson and Lamy, 2007). The majority of intralaminar connections exhibit a remarkably high degree of reciprocal, bi-directional coupling. This may allow recurrent excitation (and possibly signal amplification), feedback inhibition or - in case of connections between two inhibitory neurons - disinhibition. In contrast to intralaminar connections, translaminar connections are apparently characterized by a unidirectional vertical signal flow (for review (Helmstaedter et al., 2007;Petersen, 2007).

            Excitatory neurons in layer 4 of sensory cortices receive direct, monosynaptic input from the VPM nucleus of the thalamus. This layer is targeted by the majority of these lemniscal thalamic afferents and it represent therefore the starting point of the so-called ‘canonical’ signal-processing pathway in the neocortex (reviewed by Douglas and Martin, 2007). From there incoming signals are relayed to pyramidal cells in layer 2/3 and subsequently to thick-tufted pyramidal neurons in layer 5B. Both layer 2/3 and 5B pyramidal cells project outside the barrel cortex to other cortical areas such as the motor cortex or to subcortical target regions thus integrating and distributing incoming sensory information.

            Paralemniscal thalamic afferents from the POm thalamic nucleus target pyramidal cells in layer 5A which in turn project to layer 2/3. It has been suggested that this forms a separate paralemniscal network (Bureau et al., 2006;Ahissar et al., 2000) and that the activity of POm neurons primarily encodes information about whisking, while VPM neurons encode combined information about whisking and whisker-object contact (Yu et al., 2006). However, an early convergence of lemniscal and paralemniscal cortical microcircuits has been demonstrated to exist in form of a layer 4 (lemniscal targets) synaptic input to layer 5A pyramidal neurons (paralemniscal targets) (Feldmeyer et al., 2005;Schubert et al., 2006). In addition to these microcircuits there are also excitatory neurons involved in cortico-thalamic feedback (Zhang and Deschênes, 1997). This feedback may be direct or via input from layer 4, however the properties of these neuronal microcircuits are largely unknown due to their low synaptic connectivity (Lefort et al., 2009).

            Beside a structural and functional analysis of the mirocircuitry in a barrel-related neocortical column by single cell and paired recordings followed by detailed 3D reconstruction of the cell´s complete morphology, novel imaging techniques in genetically modified mice offer a new perspective for analyses of neocortical network activity (for review Petersen, 2009;Aronoff and Petersen, 2008;Göbel and Helmchen, 2007). In conventional brain slice preparations, this approach offers all the advantages of the functional in vitro analyses, i.e. control of the extracellular milieu, application of drugs in well defined concentrations and excellent visual identification of (labeled) neurons. Under in vivo conditions, the combined approach of genetic techniques and novel fast imaging methods (see SP 8, Helmchen) will allow the functional analysis of large-scale network activity at a cellular resolution of hitherto unknown precision. For the first time, the activation pattern of single neurons and local neuronal ensembles can be studied in well-defined behavioral paradigms.