Introduction

Synapses are the core of the retina as a computational machine, so much so that a proper review of retinal synapses would, in effect, be a description of how the retina works. We are still a long way from a complete description of how the retina works, but the last few years have seen great advances in our understanding of retinal synapses, and it seems likely that the retina will be the first neural machine of any complexity that we will truly understand at a cellular level.

An understanding of retinal synapses requires answers to two related but distinct questions. The first is the question of connectivity: which types of retinal neuron synapse with which other types of neuron? The second question relates to how these synapses behave. As has become increasingly clear over the last decades, this is not simply a question of whether a synapse is excitatory or inhibitory, but requires knowledge of the kinetics of transmitter release and clearance, details of postsynaptic channel gating and modulation, and more. The two issues—connectivity and function—have very different histories, and it is worth reflecting that, whereas the outline of connectivity was appreciated by Cajal in the early 1890s, the mechanisms of synaptic function were a complete mystery as recently as the middle of the twentieth century. To gauge how far the field has developed in the last 50 years, we might consider the state of contemporary understanding of retinal synaptic function, as summarized by Steven Polyak writing in 1955: “Speaking figuratively, it seems that there is not so much an actual passage of a material substratum or a direct transmission of nervous excitation as rather an emanation of a certain imponderable ‘influence’ as yet to be precisely defined in terms of microdynamics” (Polyak, 1957).

What has happened with retinal synapses since Polyak wrote this unenlightening synopsis? Several advances spring to mind. First, the work begun by Cajal in providing anatomical descriptions of synapses between the cell types in the retina has been enormously extended, largely through the electron microscope, to the point that there are now quantitative descriptions of the synapses between some retinal neurons. Second, the identity of transmitter substances in the retina has been almost completely uncovered. This has not been a straight path, and even in the case of the photoreceptor transmitter, there have been numerous wrong turns. The simple story, now firmly established, is that glutamate is the transmitter in the direct pathway, photoreceptors to bipolar cells, bipolar cells to ganglion cells; and that gamma-aminobutyric acid (GABA) is the main transmitter in the lateral and feedback pathways mediated by horizontal cells in the outer retina and amacrine cells in the inner retina. Third, something is now known about the molecular identity of the transmitter receptors found on many of the cells in the retina. This molecular characterization is still at an early stage, but has already indicated a level of complexity in retinal organization that goes well beyond our present understanding of how the retina works, leaving the unsettling impression that the retina is a much more subtle machine than we presently acknowledge. Lastly, the way in which transmitters are released from presynaptic terminals in the retina is beginning to be understood, although clearly there are major advances still to be made in this area.

Fifty years have produced a vast increase in the understanding of retinal synapses, but even so, it would be quite wrong to imagine that, but for a few details, retinal synapses are understood. Certainly, the outer retina is much better understood than the inner retina, largely because it has fewer types of neurons and has a simpler organization. Even in the outer retina though, there are important, unanswered questions, and the patterns of connectivity, although closely investigated for several decades, continue to reveal unexpected novelty. Two recent examples of this are provided by two studies describing an uncommon, unsuspected but nevertheless important synapse between mammalian rods and OFF cone bipolar cells (Hack et al., 1999; Tsukamoto et al., 2001), as well as the finding that there may be a layer of anatomically strange and unexplored synapses lying beneath the terminals of mammalian cones (Haverkamp et al., 2000).

The connections within the inner retina, with the exception of a few special pathways, are not known in detail, and our understanding of the physiology of defined synapses is scant. A major task in the inner retina will be the elucidation of the many modulation pathways that are suspected to exist there. This is a particularly challenging area because not only are the mechanisms of modulation largely unknown, but it is also not really clear what modulation contributes to retinal processing.

My intentions in this review are to set out the basic pattern of synaptic organization in the retina, to draw attention to recent advances, and to point out the unsolved problems of retinal synapses. The direct pathway of photoreceptors to bipolar cells to ganglion cells is presented first; then, lateral pathways and feedback loops involving horizontal cells and amacrine cells are described. Lastly, a brief description of spinules highlights an example of plasticity in the retina. Synaptic function, rather than connectivity, is the focus of this chapter, but even with this restriction, a great amount of important work has to be left out including, among other things, any discussion of efferent input, interplexiform cells, and glycine or dopamine as transmitters.