The generation of directionally selective responses in the retina.

Abstract

Although the phenomenon of directional selectivity has been of great interest to sensory physiologists for more than three decades, the cellular mechanisms that underlie it are still unknown. Directionally selective (DS) neurons, which have been described in the retina, in various cortical areas and in the peripheral nervous system, respond well to stimulus motion in one (preferred) direction, but respond little or not at all to motion in the opposite (null) direction. In the vertebrate retina, the neurotransmitter γ-aminobutyric acid (GABA) mediates null-direction inhibition of On-Off DS ganglion cells (Caldwell et al. 1978). These DS ganglion cells also receive substantial synaptic contact from acetylcholine (ACh)-containing amacrine cells, called ‘starburst’ amacrine cells due to their regularly spaced, evenly radiating dendrites. Although it has been proposed that the excitatory transmitter ACh mediates preferred-direction facilitation (Vaney, 1990; Borg-Graham & Grzywacz, 1991), clear evidence that this is so has been lacking. For example, nicotinic ACh receptor blockade with curare reduces preferred-direction responses without eliminating directional selectivity (Ariel & Daw, 1982). Furthermore, targeted laser ablation of starburst amacrine cells does not reveal any asymmetric contribution to the DS ganglion cell light response (He & Masland, 1997). The study of Grzywacz et al. (1998) in this issue of The Journal of Physiology provides the first clear evidence that nicotinic ACh input mediates the DS light responses of rabbit On-Off DS ganglion cells. The authors show that the contribution of ACh to DS responses depends on the stimulus configuration. Nicotinic blockade eliminates directional selectivity to drifting sine- and square-wave gratings, but as reported previously (Ariel & Daw, 1982), does not eliminate directional selectivity to isolated, moving bar stimuli. Thus, an asymmetric nicotinic input to DS ganglion cells extends the range of stimuli that can elicit directional responses to include moving textures, that is to those stimuli with multiple peaks in their spatial luminance profile. In fact, the directional responses of simple cells in cat striate cortex also depend on whether bar or grating stimuli are used (Casanova et al. 1992). Directionally selective responses in the rabbit retina are thus constructed piecemeal from an asymmetric nicotinic input that provides preferred-direction facilitation and from an asymmetric GABA input that provides null-direction inhibition (see Fig. 1). The ACh input to DS ganglion cells also depends on a separate GABA input from an unidentified amacrine cell (Massey et al. 1997). The two asymmetric inputs appear to act in parallel, providing On-Off DS ganglion cells with the ability to respond directionally to both low and high spatiotemporal frequency stimuli via GABA and ACh inputs, respectively. Although the actual functional role of On-Off DS retinal ganglion cells is at present unclear, the broad stimulus range over which these cells respond and the fact that their axons synapse onto cells in the lateral geniculate nucleus and superior colliculus suggest that the computation of directionality to many aspects of the visual scene is important for visual processing and behaviour. Figure 1 A simplified model of the generation of directional responses by rabbit On-Off directionally selective (DS) ganglion cells Grzywacz et al. (1998) were able to show that a nicotinic ACh input mediates directional selectivity presumably because GABA-mediated null-direction inhibition exhibits a slow decay time and is spatially wide (Amthor & Grzywacz, 1993). As a result, null-direction inhibition is ineffective at higher spatial and temporal frequencies and the blockade of nicotinic ACh receptors when grating stimuli were used revealed the presence of the asymmetric ACh input. The findings of Grzywacz et al. (1998) thus also suggest that the GABA input that regulates the asymmetric ACh input to DS cells (Massey et al. 1997) is separate from and operates on a faster time scale than GABA-mediated null-direction inhibition. Previous studies (Ariel & Daw, 1982; He & Masland, 1997) may not have found evidence of an asymmetric ACh input to DS cells simply because only bar stimuli were used. Consequently, these studies may have been examining primarily the slower/spatially wide GABA-mediated null-direction inhibition. The findings of Grzywacz et al. (1998) are thus a reminder that, in spite of over three decades of experimentation, the light response repertoire of DS ganglion cells had been insufficiently characterized. Several questions still remain unanswered. First, the exact nature of the ACh- and GABA-mediated asymmetries is not understood at the subcellular/anatomical level. Second, the roles of NMDA receptors (Tjepkes & Amthor, 1998) and calcium channels (Jensen, 1995) in the generation of directional responses also remain unclear. Finally, directional responses can be induced from non-directional ganglion cells in the amphibian retina by GABAB receptor activation (Pan & Slaughter, 1991). This finding raises the possibility that directionality may be a relatively common attribute of ganglion cells, but one that is usually non-functional. The generation of DS responses in the vertebrate retina may thus be more complicated than previously appreciated.