Scientific Understanding of Consciousness
Consciousness as an Emergent Property of Thalamocortical Activity

Connectivity of the Neural Network


Science 26 November 2010:   Vol. 330 no. 6008 pp. 1198-1201

From the Connectome to the Synaptome

Javier DeFelipe

Instituto Cajal (Consejo Superior de Investigaciones Científicas) and Laboratorio de Circuitos Corticales (Centro de Tecnologí Biomédica, Universidad Politécnica de Madrid), Spain.


A major challenge in neuroscience is to decipher the structural layout of the brain. The term “connectome” has recently been proposed to refer to the highly organized connection matrix of the human brain. However, defining how information flows through such a complex system represents so difficult a task that it seems unlikely it could be achieved in the near future or, for the most pessimistic, perhaps ever. Circuit diagrams of the nervous system can be considered at different levels, although they are surely impossible to complete at the synaptic level. Nevertheless, advances in our capacity to marry macro- and microscopic data may help establish a realistic statistical model that could describe connectivity at the ultrastructural level, the “synaptome,” giving us cause for optimism.

In the 1930s, it had been shown histologically that the terminal axonal bouton or synaptic specialization was separated by a “membranous synaptic barrier.” The large number of studies demonstrating this phenomenon resolved this issue for the vast majority of scientists; therefore, they turned their interest to other aspects of synaptic structure. At the areas of contact between the axon terminal and the soma or dendrite, only one membrane was visible (the synaptic membrane), presumably because the membrane of the pre- and postsynaptic elements were so close together that only a single membrane could be distinguished. Thanks to the introduction of transmission electron microscopy (TEM) in the 1950s, along with the development of methods to prepare nervous tissue for ultrastructural analysis, the nature of synapses was examined, confirming a critical aspect of the neuron theory: The presynaptic and the postsynaptic elements in both invertebrate and vertebrate nervous tissues are physically separated by a space of approximately 10 to 20 nm, the synaptic cleft. This essentially confirmed the observations and theories of Cajal.

Connectivity of the nervous system can be analyzed at three quite distinct levels: (i) macroscopically, focusing on major tract connectivity, for example, by examining images of the whole brain (or of large brain regions), which can even be performed in vivo by magnetic resonance imaging (MRI) or other techniques; (ii) at an intermediate resolution, as can be achieved by light microscopy, which in addition allows putative synaptic contacts to be mapped; and (iii) at the ultrastructural level, which can only be studied using electron microscopy (EM) and serves to map true synaptic contacts. Thus, we propose using the terms “connectome” to refer to the map of connections at the macroscopic and intermediate levels and “synaptome” for the map at the ultrastructural level. Only by combining studies at all three levels can we fully understand the structural plan of the brain as a whole.

At present, there are powerful methods to trace the connectome in macro- (long distance connections) and microcircuits (intrinsic or local connections). These methods include classical tracing methods, as well as molecular/genetic/physiological approaches and imaging techniques, such as two-photon imaging and ontogenetic techniques. Among the most sophisticated technologies are the genetically modified mice that express fluorescent proteins in subsets of cells [e.g., the Brainbow mouse]. In these animals, multicolored labeling of the axons and dendrites of particular neurons may be useful to define large-scale principles of connectivity.

However, it is important to bear in mind that connectivity at the light microscope level is in general rather coarse (e.g., connections between brain regions), and, in most cases, point-to-point connections between local neurons and between neurons or afferent fibers cannot be accurately determined. Indeed, axonal boutons are embedded in a complex neuropil adjacent to several possible synaptic targets. In addition, not all axonal boutons establish synaptic contacts, and a given bouton may establish several synapses. Thus, the presence of a labeled terminal in close apposition with a given neuronal element can only be considered as a putative synaptic contact.

To define the synaptome, serial reconstruction is the method of choice. Serial sectioning TEM is a well-established technology to obtain three-dimensional (3D) data from ultrathin sections, and it is based on imaging ribbons of consecutive sections by TEM. However, obtaining long series of ultrathin sections is extremely time-consuming and difficult, often making it impossible to reconstruct large volumes of tissue. However, the recent development of automated EM techniques will represent an important advance in the study of the synaptome.

Despite the technical difficulties, by adopting appropriate strategies with the tools now available coupled with the beginning of huge international projects like the Human Connectome Project or the Blue Brain Project, it should be possible to make spectacular advances in unraveling brain organization, even in humans.

We will not need to reconstruct the whole layer within a given area to determine the absolute number and types of synapses, but rather the range of variability can be ascertained by multiple sampling of relatively small regions within that area. Combining these data with those obtained by light microscopy, such as the thickness of the gray matter, the volume fraction of cortical elements (neuropil, neurons, glia, and blood vessels), the density of neurons and glia per volume, the microanatomy of neurons (i.e., patterns of dendritic arbors, distribution and density of dendritic spines, and dendritic length), together with the patterns of intrinsic (intralaminar and translaminar) and long-range (corticocortical, thalamocortical, corticothalamic, and subcortical extrathalamic) connections, would seem to be the best and most feasible strategy to unravel the complex organization of the brain. For example, to determine the synaptic contribution of pyramidal cells in a given cortical layer, it is impractical to try to reconstruct all these cells at the electron microscope level. However, this parameter could be inferred by combining quantitative light microscopy data on the total number and microanatomical characteristics of these cells on the one hand, with the average density of axospinous and axodendritic synapses obtained by FIB-SEM microscopy on the other. Another approach might be to identify the nature of the axon terminals and their spatial distribution. In other words, the most appropriate route to follow now appears to be to link detailed anatomical structural data with the incomplete light and EM wiring diagrams, thereby generating a realistic statistical model, rather than attempting to fully reconstruct the nervous system.

At the macroscopic level, techniques such as structural and functional MRI, diffusion tensor imaging, magnetoencephalography, and electroencephalography represent powerful methods to trace brain connections. These techniques also serve to define general wiring principles in the brain (structural and functional),

Light microscopy methods can now be applied to human tissue (biopsy or autopsy) that offer images of extraordinary quality. This is the case for intracellular injection in fixed material using markers like lucifer yellow. The morphology of these neurons can be visualized in such detail that they allow fine morphological details of labeled neurons to be visualized (e.g., dendritic spines) and 3D reconstructions of the dendritic arbors to be performed. In combination with immunocytochemistry for a variety of neurotransmitters, this technique can be applied to generate 3D maps of neurotransmitter-labeled appositions with the dendritic arbor of injected neurons.

At the ultrastructural level, the analysis of specimens removed during the course of neurosurgery in patients with tumors or intractable epilepsy represents an excellent opportunity to study human brain. This is partly because the resected tissue can be immediately immersed in the fixative so that postmortem factors are mainly eliminated. Undoubtedly, this is why the quality of the EM images of this human biopsy material is comparable to that obtained in experimental animals. Furthermore, it is inevitable that surgical excisions pass through normal cortical regions, and, hence, this material can be exploited to analyze various ultrastructural aspects of the neocortex in detail. In this way, we can better understand the microorganization of the human brain, which would otherwise be virtually impossible to define.

Computational models of neuronal networks based on real circuits have become useful tools to study aspects of the functional organization of the brain. Thus, why shouldn’t it be possible to construct a “siliconcortex,” a computer with an artificial cerebral cortex based on a realistic model of the complete anatomical, physiological, and molecular design of the cortical circuit?

(end of paraphrase)


Return to — Neural Network

Link to — Consciousness Subject Outline

Further discussion — Covington Theory of Consciousness