9% of baseline, n = 12, p = 0.005; Figures 2A and 2B). LTPGABA was accompanied by a decrease in PPR (baseline: 1.061 ± 0.045; post-HFS: 0.879 ± 0.065; p = 0.003; Figure 2C) and CV (baseline:

0.379 ± 0.032; post-HFS: 0.278 ± 0.028; p = 0.004; Figure 2D), and STI571 supplier an increase in the frequency of sIPSCs (177% ± 32.9% of baseline, p = 0.048), with no change in sIPSC amplitude (97% ± 12.0% of baseline, p = 0.793). In agreement with an essential role for CB1Rs in gating LTPGABA, HFS also elicited LTPGABA in CB1−/− mice (138% ± 6.9% of baseline, n = 5, p = 0.039; Figure S2). These results suggest that eCBs produced during HFS act as a retrograde signal to induce LTDGABA through their actions at presynaptic CB1Rs. In the absence of CB1Rs, LTD does not manifest and the same stimulus induces LTP. We next asked what impact the duration of stimulation had on the ability of these synapses to undergo plasticity. We examined the effects

of recruiting fibers at the same frequency but in shorter stimulation epochs (1 and 2 s) in control and AM251. Following stimulation with 1s epochs (100 Hz for 1 s × 2, 0.05 Hz interval), GABA synapses exhibited heterogeneous responses that see more were biased toward LTPGABA in control conditions (142% ± 18.7% of baseline, n = 5, p = 0.085). Synaptic potentiation was more reliable in the presence of AM251 (161% ± 23.6% of baseline, n = 7, p = 0.041, Figures 2E and 2F). When the duration of each stimulus epoch was increased to 2 s, we failed to observe any reliable changes in synaptic strength (100% ± 4.0% of baseline, n = 5, p = 0.960; Figures 2E and 2F). Once again, in the presence of AM251, we observed a robust potentiation (160% ± 16.0% of baseline, n = 5, p = 0.024, Figures 2E and 2F). At 4 s, there is clear evidence of LTDGABA that shifts to LTPGABA in the presence

of AM251. Overall, these data indicate that increasing the duration of the presynaptic burst shifts GABA synapses from those that are unreliable, but favor potentiation, to ones that exhibit reliable depression. In the absence of CB1R signaling, stable LTPGABA is observed regardless of burst duration, suggesting that CB1Rs cause LTDGABA and gate LTPGABA at these synapses. To delve more deeply into the mechanisms responsible for LTDGABA versus LTPGABA, the remaining experiments were all conducted using 4 s stimulus epochs. The LTPGABA observed here out is reminiscent of NO-dependent LTPGABA described in the ventral tegmental area (Nugent et al., 2007). To test the hypothesis that retrograde NO signaling mediates LTPGABA, we first blocked NO production with the NO synthase inhibitor, Nω-nitro-L-arginine methyl ester (L-NAME; 200 μM) and repeated the HFS in the presence of AM251. This abolished LTPGABA (77% ± 14.1% of baseline, n = 6, p = 0.175; Figure 3A) and prevented the change in PPR (baseline: 0.884 ± 0.131; post-HFS: 0.856 ± 0.103; p = 0.928) and CV (baseline: 0.133 ± 0.023; post-HFS: 0.