AA is postulated as a postsynaptic NMDAR-dependent retrograde signal (Dumuis et al., 1988; Leu and Schmidt, 2008; Sanfeliu et al., 1990) and required for the presynaptic expression of hippocampal LTP (Clements et al., 1991; GSK-3 signaling pathway Williams et al., 1989). In zebrafish larvae, postsynaptically released AA can regulate developing refinement of presynaptic axon arborization at retinotectal synapses (Leu and Schmidt, 2008). To examine the role of AA in the LTP expression at BC-RGC synapses, we first bath applied arachidonic tri-fluoromethyl ketone (AACOCF3, 100 μM), an inhibitor of cytosolic phospholipase A2 that is required for AA release from neurons (Leu and Schmidt, 2008;

Sanfeliu et al., 1990), to suppress AA release. Pretreatment of AACOCF3 for 30 min efficiently prevented TBS-induced enhancement of RGC e-EPSCs (Figure 5A) and changes in the PPR and CV of RGC e-EPSCs (Figures 5B, 5C, and S7A). Furthermore,

AACOCF3 pretreatment also abolished TBS-induced increases of calcium responses in BC axon terminals (n = 34, obtained from four different retinae; Figures selleckchem 5D and 5E). These results suggest that AA is required for the presynaptic expression of LTP at BC-RGC synapses. Next, we found that perfusion of AA (100 μM) itself for 5 min led to a persistent increase in the amplitude of RGC e-EPSCs for more than 30 min (130% ± 7% of the control, n = 5; p = 0.001; Figure 5F). Similar to TBS-induced LTP, this AA-induced enhancement also involves presynaptic however changes because both the PPR and CV

of RGC e-EPSCs were also significantly reduced after AA perfusion (compare Figures 5G and 5H with Figures 4E and 4F). The PPR was 0.99 ± 0.09 and 0.81 ± 0.07 before and 10–30 min after AA perfusion (n = 5; p = 0.01; Figures 5G and S7B), respectively, meanwhile the CV was 0.31 ± 0.05 and 0.26 ± 0.06 before and after AA perfusion (n = 5; p = 0.03; Figure 5H), respectively. Taken together, these results indicate that AA is necessary and sufficient for the expression of LTP at BC-RGC synapses. We next examined whether natural visual stimulation can induce LTP at BC-RGC synapses. We found that RFS (100 whole-field flashes with 2 s duration at 0.33 Hz) caused a persistent increase in the efficacy of BC-RGC synapses, as assayed by the amplitude of e-EPSCs in RGCs (Figure 6A). The mean e-EPSC amplitude during 10–35 min after RFS was 140% ± 11% of the control value observed before RFS (“No MK-801,” n = 9; p = 0.005; Figure 6A). This synaptic enhancement persisted for as long as stable recording could be made. Similar to TBS-induced LTP, this RFS-induced LTP at BC-RGC synapses required the activation of postsynaptic NMDARs because intracellular loading of MK-801 (1 mM) into the RGC prevented the LTP induction by RFS (“MK-801,” 93% ± 9% of the control, n = 6; p = 0.74; Figure 6A). In addition both the PPR and CV of the e-EPSC amplitude also showed significant decrease after RFS (Figure 6B).