Cellular Neurobiology Laboratory
Information Processing in the Retina

Dr Masland | Recent Publications | Links

Retinas sense light and send a coded signal about it to the brain. Light is detected by photoreceptor cells, an array of about 100 million densely packed cells that line the back of the eye. The photoreceptor cells change a photic signal into an electrical one, which is transmitted chemically across synapses to intermediate cells within the retina. These ultimately converge on the retinal ganglion cells, whose axons join together to form the optic nerve.

Along the way, information falling on the photoreceptors is shaped and recoded. The output of many photoreceptors is collected by a smaller number of bipolar cells, and these in turn converge upon a still smaller number of retinal ganglion cells. In addition, the flow of information along this pathway is shaped and reshaped by two levels of laterally conducting pathways, those involving horizontal cells and amacrine cells. Particularly in the inner retina, these pathways are remarkably diversethere are about 20 identifiably different kinds of amacrine cell, each playing a different role in the processing of visual information.

We have two main lines of research: (1) an attempt to account for all of the retina’s neurons anatomically and (2) studies of the means by which the response properties of retinal ganglion cells are generated. (Some of this work has been supported by the National Institutes of Health.)

Missing Neurons in the Retina and Elsewhere

The building blocks of the nervous system are the individual classes of neuron. Remarkably, all the neurons in any nervous system structure are accounted for in only a few cold-blooded animals. The reason is simple: to identify subclasses of neurons in mammals, one must have a method for distinguishing one class of cells from others i.e., a specific stain. This allows confident identification of the cells that are stained. What it does not do is show the cells that are not stained. In principle, one could simply learn the total number of neurons and add up the stained cell classes until a 100 percent census was reached. However, technical problems make this much easier said than done. Even in the retina, unaccounted neurons remain.

We began by creating a base of reference. We used a combination of light and electron microscopy to identify every cell in samples of rabbit retina. From these samples we learned that bipolar cells are 42 percent of the total, horizontal cells 2 percent, amacrine cells 37 percent, and Müller glia (supporting cells) 19 percent. With these values in hand, we could learn how many unidentified amacrine cells exist. The largest known populations of amacrine cells were stained with fluorescent probes and counted. Knowing the total fraction of amacrine cells, it was possible to learn what part of the total is accounted for by the cells that have previously been identified. We found that 78 percent of all amacrine cells in the rabbit’s retina are known only from isolated examples, if at all.

This finding indicates a major gap in our understanding of the retina. Amacrine cells are important to the retina’s function: they control and modify the flow of information through the tissue. If more than three-quarters of them are "missing," our understanding of the retina’s function is far from complete. A large proportion of missing neurons is not a peculiarity of the rabbit’s retina. In comparative studies, we showed that the overall proportions of cells in the mouse, rat, cat, and monkey are similar to those in the rabbit. In none of those retinas have any larger populations of amacrine cells been stainable. The conclusion is that similar numbers of cells are missing from our understanding of most mammalian retinas.

The situation is analogous in other regions of the central nervous system. For example, Golgi studies of the cortex reveal a diversity of neurons equal to or greater than that found in the retina. And yet, our concept of the cortex (derived primarily from single-unit recording) allows for only three or four physiological types of cells. The meaning of this huge mismatch is uncertain; among other things, it points to the need for a more rigorous accounting of cortical neurons.

Direction Selectivity After Targeted Ablation of Amacrine Cells

A type of retinal ganglion cell is termed directionally selective because the cells fire a burst of action potentials when the stimulus moves in one direction and fire none or are inhibited when the stimulus moves in the other direction. This selectivity represents a classic paradigm of computation by neural microcircuits because the computation is simple and clear and is somehow created by a very few cells. However, its cellular mechanism remains obscure.

A cell that figures prominently in theories of direction selectivity is the starburst amacrine cell, so named because of its regularly spaced, evenly radiating dendrites. Work from our laboratory showed that the starburst cells cofasciculate with the dendrites of the directionally selective ganglion cells and that they excite those ganglion cells directly. The starburst cells are very numerous, and several models of direction selectivity propose that the starburst cells are key in the directional discrimination. To test this theory, we ablated starburst cells while recording from the directionally selective ganglion cells.

We labeled starburst cells in living retinas by using a fluorescent probe that binds to the cells without damaging them. While recording electrophysiologically from a directionally selective retinal ganglion cell, we used a laser focused to a 1- to 2-m m spot to kill starburst cells overlapping the dendritic arbor of the directionally selective retinal ganglion cell.

We found that the starburst cells do not participate in the directional discrimination itself. However, they appear to have another, equally critical action on the ganglion cell: they provide a movement facilitation that primes the ganglion cell to respond to other inputs. This facilitation happens in all directions of movement; it is a generic movement-sensitive signal. This finding focuses the search for the directional discriminator on amacrine cells (presumably among the large "missing" population) that could provide an asymmetric inhibition in the direction of movement to which the ganglion cell is unresponsive.

Neurosurgery Clinical Units MGH Neurosurgical Service Spine Tumor Center at MGH Brain Tumor Center at MGH Research at MGH Neurosurgery Neuroscience Care Units at MGH Residents at Neurosurgery