Segregation and Wiring in the Brain

See allHide authors and affiliations

Science  30 Mar 2012:
Vol. 335, Issue 6076, pp. 1582-1584
DOI: 10.1126/science.1221366

A mosaic of hundreds of interconnected and microscopically identifiable areas in the human cerebral cortex controls cognition, perception, and behavior. Each area covers up to 40 cm2 of the cortical surface and consists of up to 750 million nerve cells (1). The architecture—the spatial distribution, density, size, and shape of nerve cells and their processes—varies between different cortical areas. Nerve cells are interconnected within each area and with other brain regions and the spinal cord via fiber tracts, synapses, transmitters, modulators, and receptors. This incredible structural complexity underlies the functional segregation in the cerebral cortex. The ultimate goal—to understand the driving forces and organizational principles of the human brain beyond the cellular and functional details—remains a challenge. Reports by Chen et al. (2) and Wedeen et al. (3) on pages 1634 and 1628 of this issue, respectively, accept this challenge by analyzing the genetic topography of the cortex and the spatial course of fiber pathways in the brain. The studies find unifying hierarchical and geometric rules behind the organizational details.

In addition to classical cytoarchitectonics (the spatial organization of cellular composition) of the brain (4), the fields of cell biology, neuroimaging, neuroinformatics, and genetics have provided myriads of valuable data about the structure and function of the cerebral cortex. Only after the advent of modern genetics was it possible to identify the driving forces behind segmentation as an organizational principle of the vertebrate brain. However, a segmentation process in the cortex, similar to that seen in other brain regions and the spinal cord, does not seem likely because an overt segmental or metameric partitioning of the neocortex is not visible during embryogenesis (5). Instead, Chen et al. demonstrate a genetically controlled and hierarchically organized structural segregation in the human cortex.

Chen et al. investigated human brains by cortical surface reconstruction, advanced atlas mapping, and genetic analyses. By comparing mono- and dizygotic twins, the authors calculated how shared genetics influence area expansion between cortical regions. A cluster analysis of this genetic topography was performed in the absence of predefined anatomical information. The authors could thus parcel the cortical surface area into 12 genetic regions. A possible correlation between this genetic topography and the cytoarchitecture of the cortex (6) may reveal rules behind the mosaic of cortical area. However, some genetically determined regions are very large, even crossing the borders of cytoarchitectonic areas and expanding into other lobes. For example, the occipital and inferior parietal clusters of Chen et al. comprise numerous cytoarchitectonic areas (7, 8).

Genetic topography, gradation, and cytoarchitecture.

Regional organization of the human frontal lobe is shown according to the gradation hypothesis (A), genetic topography (B), and cytoarchitecture (C). Corresponding regions are highlighted with colors. Dark black lines represent sulci; thin gray lines mark boundaries of the zones. Blue, red, and green arrows label the architectonic gradation streams from the cingulate cortex, the central sulcus, and the insular cortex, respectively. Regions in (A): FmZ, frontomotoric zone; FoZ, fronto-opercular zone; FpZ, frontopolar zone; PmZ, paramotoric zone; PoZ, paraopercular zone. Regions in (B): 1, motor-premotor cluster; 2, dorsolateral-prefrontal cluster; 4, orbitofrontal cluster; 5, opercular-subcentral cluster. In (C), numerals indicate Brodmann areas (6) (regions defined according to the structure and organization of cells).


Although there is mismatch in some details, this should not obscure an important commonality revealed between cytoarchitecture and genetic topography–hierarchical organization. Cytoarchitectonic maps reflect functional and molecular aspects of brain organization. The regional distributions of transmitter receptors (9), cell types (10), and connectivity (8), as well as the genetic topography determined by Chen et al., are hierarchically organized. That is, cortical subdivisions are nested in a common superstructure and interrelated in a tree-like organization. They show different degrees of functional, molecular, or structural similarities. Hierarchical organization of the cortex is thus the unifying rule, which encompasses all scales from the molecular to the systems level.

The concept of hierarchy is reminiscent of an old but unfortunately forgotten finding (1113): the gradation of architectonic features. This concept explains differences between cortical areas by a stepwise and directed architectonic differentiation of the neocortex. These developmental gradation streams follow non-Euclidean axes. The rostrocaudal axis is bent around the end of the lateral fissure; the other axes are orthogonal to it. In the frontal lobe (see the figure), gradation streams originate in the border regions of the insula, anterior cingulate gyrus, olfactory area, and central sulcus (13). These regions are phylogenetically old or are located in primary sulci (depressions or fissures in the surface of the brain). Along the gradation streams, the cellular composition and organization change stepwise according to the changing relationship between the sizes of layer V to layer III pyramidal cells in the cortex, along with the development of an inner granular cell layer IV. Thus, the gradation hypothesis describes a hierarchical rule behind the details of architectonic phenotypes. The results of Chen et al. suggest that the gradation topography principally corresponds to the genetic topography.

Previous studies of fiber tracts—bundles of nerve fibers—provided a wealth of detail about their spatial orientation in a highly descriptive, but not analytical way. Wedeen et al. propose a rectilinear, grid-like geometric organization of fiber pathways in human and nonhuman primate brains. Starting from major paths, the authors identified adjacent paths that cross the initial ones orthogonally. The crossings form well-defined curved, two-dimensional sheets. The major longitudinal paths provide the long-range anatomical connectivity. The short U-fibers connect adjacent gyri (ridges on the cerebral cortex). As in the case of gradation streams, the three-dimensional grid of fiber tracts follows the three principal axes of growth direction during brain development (longitudinal, mediolateral, and dorsoventral). This organization may be achieved by chemotactic mechanisms of neuronal path finding and incremental rewiring.

Wedeen et al. also used the Frobenius theorem from differential geometry to support their argument for the formation of geometric spatial organization of fiber paths. At present it is difficult to decide whether three chemotactic gradients are the basis for the three families of vector fields observed by the authors, and whether other constraints of this theorem are fulfilled. If this “three-families hypothesis” as geometric principle holds true, procedures to map neural fiber tracts can be constrained, which might make it easier to avoid the limitations of current algorithms and to validate different tracking techniques (across scales, across subjects, across species, etc.).

A common view that arises from the studies of both Chen et al. and Wedeen et al. is the idea of the brain as a regionally highly differentiated, but hierarchically and geometrically organized, spatial structure. Detailed aspects of this “canonical brain organization” can be modified by environmental conditions including pathology and genetic diversity. Mathematical methods such as hierarchical clustering and differential geometry can help us to understand the principles behind variable phenotypes and to guide the development of a realistic brain model.


View Abstract

Navigate This Article