, 2002, Miller and Bloomfield, 1983 and Vaney, 1990). In contrast to most retinal neurons, SACs display an extensive dendritic overlap (Tauchi and Masland, 1984), Everolimus which enables them to provide independent neuronal hardware for the different functional subtypes of DS cells. It has been

proposed earlier that SACs are importantly involved in the DS computation (Borg-Graham and Grzywacz, 1992, Masland et al., 1984a and Vaney et al., 1989), but experimental proof for this notion came less than 10 years ago, when it was shown that massive ablation of SACs results in a selective loss of retinal direction selectivity (Amthor et al., 2002 and Yoshida et al., 2001; but see He and Masland, 1997). Optical measurements of light-stimulus-evoked

Ca2+ concentration changes (Denk and Detwiler, 1999) in the dendrites of SACs demonstrated that stimuli moving this website from the soma to the dendritic tips, i.e., centrifugal motion, evoked larger Ca2+ responses in the distal dendrites than motion in the opposite direction, i.e., centripetal motion (Figure 5B2) (Euler et al., 2002). Because SAC output synapses are located in the distal dendrites (Famiglietti, 1991) and transmitter release from SACs is Ca2+-dependent (O’Malley et al., 1992 and Zheng et al., 2004), this indicated that SACs are able to provide DS ganglion cells with directionally tuned input. In fact, since the dendritic sectors are electrically isolated from each other, each sector can be thought of as an independent detector for centrifugal motion (Figure 5B3). Around the same time, the related long-standing question as to whether retinal direction selectivity is computed in the ganglion cells themselves or presynaptically by interneurons (reviewed in Masland, 2004) was successfully addressed. Patch-clamp studies revealed that the synaptic input to ON/OFF DS cells

is already DS TCL (Borg-Graham, 2001, Fried et al., 2002 and Taylor and Vaney, 2002): Preferred direction motion elicits more excitation and less inhibition in the ganglion cells, whereas null direction motion elicits more inhibition and less excitation. This suggested (1) that both inhibitory and excitatory inputs are DS, (2) that the Barlow-Levick model does not fully capture retinal DS computations, and (3) that the latter are indeed already performed presynaptically by interneurons. Note that the latter point does not exclude that postsynaptic, i.e., ganglion cell-intrinsic mechanisms contribute to the overall direction selectivity observed in ganglion cells (see Mechanisms at the Ganglion Cell Level). It was already known for long that blocking GABAA receptors abolishes DS responses in the ganglion cells but leaves their responsiveness intact (Caldwell et al., 1978 and Massey et al., 1997).