For understanding the connectional architecture of the brain it would be very helpful if neuronal and regional connectivity could be related to each other. Unfortunately, their relationship is not straightforward.
When studying neurons in brain sections the dendritic tree and even most of the local axon arborizations can be assessed (Staiger et al. 1999). It is not known, by contrast how many of the distant interregional projections have been severed. It is exceedingly rare that individual axons have been followed and reconstructed beyond the direct vicinity into other brain regions (e.g. Rockland et al. 1999; Zhong et al. 2006). From the available studies, however, important general rules have been gleaned, which are of great significance for our understanding of cortical organization. Such general rules state that interregional projections originate from neurons in supra- and infragranular layers, that the projection neurons (emanating intrahemispheric association fibres, contralateral commissural fibres and extracortical projection fibres) are pyramidal cells, and that the size of the pyramid-shaped somata is roughly related to the length of their axons. Since pyramidal cells invariably use excitatory amino acids (mainly glutamate) as their transmitter, it is reasonable to infer that all interregional projections are excitatory. The effect of one cortical area on another, however, depends not only on the type of projection neuron but also on the type of neuronal targets and their dynamics: Functionally, association fibres appear to have predominantly excitatory effects on most cortical regions, whereas commissural fibres tend to evoke contralateral inhibition (e.g., Ferbert et al. 1992).
There remains the question of what it means when we say that one brain region projects to another? Here, brain regions, in particular cortical areas, are conceptualized as homogeneous units whose characteristics are invariant along major spatial dimensions. Nevertheless, the criterion of homogeneity depends on the particular feature studied with the result that there are many different definitions of brain regions, which vary in their level of detail and precision. This variability is comparable to the variability in neuron classifications at the cellular level, and the uncertainties resulting from the two sets of variability are independent and thus not constrained by mutual comparisons. As a result of this enormous variability it is difficult to obtain precise statements on the neuronal underpinnings of interregional connectivity. For example, along the vertical dimension of the cerebral cortex different layers of the same cortical region have long been known to reach different intra- and subcortical targets; the horizontal dimension along neurons in the centre and in, the periphery of cortical columns have different projection preferences among neighbouring columns with different functional consequences. Such intraregional differences need to be taken into account at the level of interregional projections.
Current interregional tract tracing experiments provide some information about vertical intraregional differentiation referring to the laminar origin and termination of interregional fibres in the cerebral cortex: In analogy to the ascending thalamocortical fibres, which terminate predominantly in layer IV of their cortical target regions, corticocortical fibres that target preferentially layer IV of another region are also referred to as ascending or feedforward connections. By contrast, the axon terminals of reciprocal pathways usually avoid layer IV and are consequently named descending or feedback connections (Rockland and Pandya 1979; Felleman and Van Essen 1991). Although the clarity of laminar termination preferences varies with the presumed hierarchical distance between two connected areas, the most obvious ascending termination patterns arise from supragranular neurons, whereas infragranular neurons contribute predominantly to descending projection patterns (Barone et al. 2000). A third type of columnar termination pattern across all layers is attributed to so-called 'lateral' projections between cortical regions that are supposed to belong to similar levels in the processing hierarchy.
In contrast to the laminar preferences, the neuronal preferences of axon terminals are largely unknown. For example, it is not known whether interregional projections contact excitatory or inhibitory interneurons or projection neurons in their target regions. Knowing the identity of those target cells would be very important for understanding the functional impact of interregional projections (see difference in functional impact between association fibres and commissural fibres mentioned above). One may even go as far as measuring the volume of axonal boutons of region-specific projections to obtain an estimate of their functional impact on distinct cortical layers (Ger-muska et al. 2006).
It is also relevant to obtain an estimate of the specific processing of extrinsic versus intrinsic signals, and of the likelihood and timing of further transmission to subsequent brain regions. It has been estimated that even in primary sensory regions thalamocortical afferents form only a small proportion of the total number of synapses, even in the dominant target layer IV in the cerebral cortex. Whether networks of intrinsic granule cells act to boost thalamocortical signals is a matter of speculation although they certainly possess the required connectional features (e.g. Schubert et al. 2003). It is likely that thalamocortical inputs contribute only a fraction of the input required to activate a cortical brain region. Thus an input-driven mean-field model of cortical activity spread (e.g. Kotter et al. 2002) is a gross simplification applying to the situation where the cortical system is ready to respond to such inputs. To improve our understanding of whether certain inputs lead to a significant change in cortical activity we depend crucially on more detailed data on their cellular targets. This may serve as an example that the characteristics of local microcircuits will influence the models of global activity patterns in the cerebral cortex.
Was this article helpful?