In the retina, all-trans retinoic acid (at-RA) could function as a light signal because its production increases with the level of illumination. Given the well-established effects of retinoic acid on cell coupling in other tissues, it is possible that the changing levels of at-RA modulate the gap junctional permeability between retinal neurons. This study examines the effects of retinoic acid on horizontal cell coupling, which is known to be modulated by the ambient light level. Single horizontal cells were injected under visual control with either Neurobiotin (mouse retina) or Lucifer yellow (rabbit retina) and the extent of tracer coupling or dye coupling was used to monitor the gap junctional permeability. In the mouse retina, the injection of Neurobiotin revealed a network of approximately 150-250 tracer-coupled horizontal cells. The tracer coupling was completely abolished by incubating the retina in 150 microM at-RA for 35 min. In the rabbit retina, the injection of Lucifer yellow into A-type horizontal cells revealed networks of approximately 15-30 dye-coupled horizontal cells. Incubation in 150 microM at-RA reduced the dye coupling within 12 min and complete uncoupling was achieved after 35 min. The uncoupling effects of at-RA in the mouse and rabbit retinas were concentration- and time-dependent and they were reversible after washout. The coupling was not affected by either the 9-cis form of retinoic acid or by at-RA that had been isomerized by intensive light. The uncoupling effect of at-RA persisted following treatment with a D1 receptor antagonist and thus was dopamine-independent. This study has established that at-RA is able to modulate the gap junctional permeability between horizontal cells in the mammalian retina, where its light-dependent release has already been demonstrated.