The mammalian transcriptome and the cellular complexity of the brain

摘要

The complexity of neural networks in the central nervous system is enormous, resulting not only from the vast combinatorial possibilities for neuronal connections, but also from the diversity of properties and functions of neuronal cells themselves. Systematic analysis of neuronal diversity over the last century, using techniques such as the Golgi method, has proved the existence of a large variety of morphological types of neurons. Because shapes and arborization patterns are the visible expression of neuronal connectivity, it is not surprising that different morphological cell types have so far without exception turned out to have distinct physiological functions. Only a few morphologically defined cell types, though, express usefully distinctive cytochemical markers, and those that share common markers do not necessarily display the same patterns of connections or electrophysiological properties. A high level of heterogeneity of function within morphological types is becoming increasingly apparent. Furthermore, in the cortex, there are probably many subtle variations in cellular properties and local circuitry, according to the specific function and development of different cortical areas. Ultimately, the function of these neurons and networks may become clear as a result of extracellular recordings in vivo (i.e. a ‘top-down’ approach), but the lack of knowledge about their properties and organization limits the ability to design appropriate experiments. Therefore, an increased understanding of the construction and organization of the local circuits of the brain is required, i.e. a ‘bottom-up’ approach. This entails a description of the cell types that are components of a neural centre, the identification of their chemical mediators, channels and receptors, and an analysis of their synaptic connections. Critical genetic and physiological experiments are then designed to understand cell functions. The size of this task, however, is staggering: the Golgi method has revealed 50 anatomical types of local circuit neurons in the monkey striate cortex, which has led to the suggestion that there could be as many as 100 different types of interneurons in each layer of the neocortex. Such extreme cellular heterogeneity has also significantly impaired gene expression analysis. To confront this complexity, specific populations must first be distinguished within the highly heterogeneous in vivo network/tissue. Over the last few years, high throughput techniques such as microarray have begun to offer the possibility for physiologists to bridge the gap in understanding between top-down and bottom-up knowledge of the brain, by systematic, and to an increasing extent automated, approaches to collecting genome-wide data for characterizing cells. These approaches are rapidly becoming almost standard tools in cellular neurophysiology labs, and bring with them a new set of issues relating to data analysis and experimental design. It is an opportune time therefore for The Journal of Physiology to review recent progress in applying functional genomics techniques to the brain. The articles in this special issue are distinguished not just by their focus on high throughput techniques, but also by a physiological emphasis on understanding how this new kind of data illuminates the dynamical function of neural systems. A common theme of these papers is a note of caution: high-throughput techniques often entail an increased level of noise, and the large-scale approach brings with it increased difficulty in scrutinizing the sense of each individual piece of data. These issues are particularly important to consider as we begin to construct open databases of functional genomic data. The community as a whole is only beginning to compare results from different groups on the same or similar systems, and to reach a consensus on the meaning of results. Yet, properly applied, these techniques offer unprecedented power for uncovering new interactions, connections,