Neuronal cell types

Abstract

For the first time in any substantially complex CNS structure, the inventory of cell types in the retina appears to be virtually complete. Evidence for this comes partly from a new, unbiased, morphological sampling technique; from the sheer volume of well-visualized cells that is easily generated by modern cell-filling methods; from the three-dimensional resolving power of digital microscopy; and from the law of coverage, which dictates that a true cell type be evenly spaced. Many types are now terminally identified: the combination of morphology, gene expression, and mosaic make further subdivision of the type extremely unlikely. Fine-tuning of the classification will still occur, and a few orphan types remain among the wide-field cells, which can cover the retina using small absolute numbers of cells. But the survey probably encompasses 98% of all of the retina's neurons and it is sure that no major players have been missed.If there are ∼60 cell types in the retina, how many are there in the cortex? From plausible assumptions of spacing, cell number and dendritic field diameter, one observer estimated the number at ∼1,000. At the least, this estimate serves the rhetorical purpose of pointing out how important is the number of types for any understanding of this structure. If there are 1000 types of neuron in the cortex (or even 100), there is a huge gulf between that number and our understanding of the cortex's physiological diversity. A resolution, one way or the other, would tell us which conceptual models of the cortex are plausible.New probes originating from genome screening promise to help sort the cells, although few instances of truly cell-type-specific gene expression have been discovered so far. The common occurrence is partial specificity, in which several types of neuron express the gene in question. Although it is disappointing that more precise specificity is not widespread, mixed expression is nonetheless extremely valuable. The ambiguity adds only a need for a marker that reveals the cell's arborizations, because selective expression of the protein in combination with neuronal structure is unambiguous. This combined approach is less cumbersome and more robust than strictly morphological methods, and more reliable than depending solely on the type-specificity of expression of a newly identified gene. It will probably be the main way of identifying new cell types for the foreseeable future.