The anatomical problem posed by brain complexity and size: a potential solution

Abstract

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.

Keywords: choice of species for studying the brain; connectome; electron microscopy; interdisciplinary approaches; neuron doctrine; synaptome.

Figure 1

The central nervous system works as a whole. Schematic drawing by Barker (1899) to illustrate some of the multiple relationships between different parts of the central nervous system. Taken from DeFelipe (2014).

Figure 2

The complexity of the brain. Artistic composition showing a coronal histological section of the human brain and a hand holding a pin with a pinhead (approximately 1 mm³) to graphically illustrate the complexity of the brain. In a volume of human cerebral cortex similar to the pinhead in this figure, there are about 27,000 neurons and 1000 million synapses (Alonso-Nanclares et al., 2008). The diameter of the pin (0.5 mm) is equivalent to the thickness of a cortical column. Since a human pyramidal neuron typically has a dendritic tree with a minimum total length of several mm, in this volume there would be several thousand mm of dendrites. Taking a medium-sized pyramidal neuron with a dendritic length of 10 mm as an example, and considering that pyramidal cells represent approximately 80% of the total population (see text for further details) there would be approximately 216 m of pyramidal cell dendrites in this 1 mm³. Furthermore, the brain is one of the organs of the body with the highest metabolic demands and thus, there is a very dense network of blood vessels in association with the neurons and glia (see e.g., Blinder et al., ; Magistretti and Allaman, ; Yuan et al., 2015). Taken from DeFelipe (2014).

Figure 3

Reconstruction of a minicolumn. (A) Schematic representation of a minicolumn in which only the soma and proximal dendrites of pyramidal cells (black) and the main axon (blue) are represented. Note that the axons form bundles due to the vertical arrangement of pyramidal cells. (B) The apical dendrites also form vertical bundles and, although variability exists both between cortical areas and species in the size and number of dendrites that form the bundles as well as in the layer where the terminal dendritic tufts terminate, in general, the vertical dendritic organization is as follows. As reviewed in DeFelipe (2005), distinct bundles of pyramidal neuron apical dendrites are formed in different levels of layer V, and ascend towards the pial surface. Apical dendrites originating from pyramidal cells in layers II–III mainly join the bundle periphery. At this level, the apical dendrites of layer V and layer II–III pyramidal neurons begin to form terminal tufts which end in layer I. By contrast, the apical dendrites originating from Layer VI pyramidal cells do not join the layer V bundles, but are arranged as separate bundles which ascend to layer IV and form terminal tufts there. The core of the long dendritic bundles that extend from layer V to layers II–III is therefore principally composed of the apical dendrites pertaining to layer V pyramidal neurons. (C) Image captured by focused ion beam milling and scanning electron microscopy (FIB/SEM) to show the relatively high density of synapses in the neuropil and the ultrastructural appearance of asymmetric and symmetric synapses in the rat cerebral cortex. Four asymmetric synapses (arrows) and one symmetric synapse (arrowhead) can be identified on four dendritic spines (d1 to d4). Asymmetric synapses show a thick post-synaptic density. The symmetric synapse has a thin post-synaptic density, which is similar to the pre-synaptic density, and is located on the neck of a dendritic spine (d1). Scale bar, 500 nm. (D–G) Three-dimensional representation of a stack of serial sections and the synaptic profiles that appear in the corresponding counting brick. (D,E) show a stack of serial sections, slightly rotated counter-clockwise through the vertical axis in (E). Only 12 sections are shown out of the 115 that compose the complete stack. An unbiased counting frame was drawn on each section, taking the green and the red lines as the acceptance and exclusion boundaries, respectively. To extend the counting frame to three dimensions, the front section was considered as an acceptance plane and the last section as an exclusion plane. Thus, synaptic profiles (contours of the synaptic membrane densities) were counted inside an unbiased counting brick bound by three acceptance planes (top, left and front) and three exclusion planes (right, bottom and back). As an example, the 10 synaptic profiles that appeared in the first section (acceptance plane), without intersecting any of the exclusion planes, have been numbered from 1 to 10 in (D,E). The counting frame measured 6.86 × 5.28 μm after correction for tissue shrinkage. In (F,G) the counting brick and the three dimensional reconstructions of synaptic profiles have been rendered. Green objects represent asymmetric synaptic profiles and red objects symmetric synaptic profiles. All the objects shown were inside the counting brick or intersected one of the acceptance boundaries, without intersecting any of the exclusion planes. Numbered objects correspond to the same synaptic profiles shown in (D,E). Note that every object can be individually identified and localized in the 3D space. Panels (A,B) have been adapted from DeFelipe (2005), and (C–G) and legend have been taken from Merchán-Pérez et al. (2009).

Figure 4

Long-distance axonal projections of individual pyramidal neurons. Images obtained from an adult Thy1-eGFP mouse brain using a fluorescence micro-optical sectioning tomography (fMOST) method. In this figure is shown the long-distance projectionpattern of eight layer V pyramidal neurons located in different cortical areas. 3D reconstruction results were merged with the direct volume rendering of a whole brain image stack in sagittal, coronal and horizontal views. The image stack had been resampled from a voxel size of 0.32 × 0.32 × 2 μm³ to 4 × 4 × 4 μm³. Courtesy of Hui Gong. Unpublished material taken from Gong et al. (2013).

Figure 5

Long-range corticofugal axons. Tracing of single axons labeled with small injections of biotinylated dextran amine in the rat motor cortex. (A) Axon of a lateral agranular cortex (AGl) pyramidal tract neuron that emits multiple collaterals (shown with different colors) including subthalamic nucleus (STN) collaterals. No cortical collateral was found, though this neuron had multiple collaterals innervating striatum (Str), thalamic, mesencephalic, pontine, and medullary nuclei. The STN collaterals of the neurons had thin branches entering zona incerta (ZI). One of the cerebral peduncle collaterals of the neuron emitted ZI branch forming boutons. (B) Axon of a medial agranular cortex (AGm) pyramidal tract neuron that emits multiple collaterals including STN, Str, thalamic, and pontine nuclei. The neuron had cortical collaterals innervating AGm, granular cortex (Gr), and Str. The thalamic collateral of the neuron travelled through the middle of the thalamus. One of the cerebral peduncle collaterals of the neuron B traversed STN and then to ZI without forming boutons. Other abbreviations: APT, anterior pretectal nucleus; cp, cerebral peduncle; DpMe, deep mesencephalic nuclei; Gi, gigantocellular reticular nucleus; GPe, Globus pallidus external segment; ic, internal capsule; IO, inferior olive: lfp, longitudinal fasciculus of the pons; ot, optic tract; Pn, pontine nucleus; PnO, pontine reticular nucleus, oral part; Po, posterior thalamic nuclei; py, medullary pyramid; pyd, pyramidal decussation; Rt, reticular thalamic nucleus; SC, superior colliculus; SN, substantia nigra; VL, ventrolateral thalamic nucleus; VM, ventromedial thalamic nucleus. Courtesy of Hitoshi Kita. Figure and legend taken from Kita and Kita (2012).

Figure 6

Camera lucida reconstruction of two motor thalamocortical axons in the rat labeled with viral vectors. Axon fibers of IZ neurons (inhibitory afferent-dominant zone of the ventral anterior-ventral lateral motor thalamic nuclei [VA-VL complex]) were widely distributed in motor-associated areas and neostriatum (A). Of cerebral cortical layers, layer I was most intensely innervated by the axon fibers of IZ neurons (B–D). In contrast, axon fibers of EZ neurons (excitatory subcortical afferent-dominant zone of the VA-VL complex) were found only in motor-associated areas (E) and distributed mainly in cortical layers II–V (F,G). Panels (D,G) are representative planes, in which the results of 10 serial sections were superimposed onto a parasagittal plane of the fifth section. Other abbreviations: FL, forelimb region of primary somatosensory-motor area; HL, hindlimb region of primary somatosensory-motor area; M1, primary motor area; M2, secondary motor area; S1, primary somatosensory area. Courtesy of Takeshi Kaneko. Figure and legend taken from Kuramoto et al. (2009).

Figure 7

Axonal arborizations of cortico-cortical cells in monkey sensory-motor cortex. These neurons were labeled after small extracellular injections of horseradish peroxidase into a stratum of corticocortical axons situated in the white matter immediately deep to area 3b (asterisks). (A) Retrogradely labeled corticocortical cell with soma (arrow) in area 1, a minor collateral to area 3b, dense boutonal clusters in areas 1 and 2, and major collaterals apparently continuing on toward area 5. (B) Retrogradely labeled corticocortical cells with somata (arrows) in areas 3b and 3a and focused concentrations of boutons in each area. The boutonal plots were produced from high-magnification drawings of the full collateral ramifications. Each dot indicates one bouton. Bar, 500 μm. Taken from DeFelipe et al. (1986).