After NMDAR activation, BAPTA blocked changes in rectification, implicating the involvement of elevated Ca2+ and supporting a role for NMDARs in this process (n = 8; RI, 0.54 ± 0.06 to 0.57 ± 0.07; p = 0.39). Our hypothesis is that the NMDAR-induced changes in rectification we observe are due to a loss of CI-AMPARs with a possible replacement by CP-AMPARs. There are several different mechanisms by which the loss of CI-AMPARs could occur. One is mTOR inhibitor through lateral diffusion of AMPARs from the synaptic to the extrasynaptic membrane (Borgdorff and Choquet, 2002). However, the best-characterized mechanism

of AMPAR removal is dynamin-dependent endocytosis, triggered by an NMDAR-induced rise in postsynaptic Ca2+ (Carroll et al., 2001). We tested whether the CI-AMPARs are internalized due to dynamin activity by dialyzing RGCs with 10 mM PCI-32765 mouse dynamin-inhibitory peptide (DIP), which blocks endocytosis of AMPARs by interfering with the binding of amphiphysin with dynamin (Lüscher et al., 1999). In RGCs, DIP causes a run up of the extrasynaptic, not synaptic,

AMPAR-mediated response due to unbalanced insertion of AMPARs undergoing rapid cycling (Xia et al., 2007). To ensure that this separate effect of DIP would not confound our results, we first recorded a 10 min baseline of light responses during DIP dialysis before recording the control I-V. The light responses of all ten cells remained stable during this period (3.2% ± 1.3% change over Sodium butyrate 10 min; data not shown), indicating that synaptic AMPARs under our recording condition are stable and that DIP does not affect the initial AMPAR ratio. While the mean baseline RI was higher in DIP-loaded cells than that of the control cells (Figure 4F; RI = 0.74), this effect was not significant (p = 0.15, t test) and probably reflects the variability of RIs as seen in Figure 1D. We find that inclusion of DIP in the pipette solution consistently

blocked the induction of synaptic plasticity with NMDA. The average rectification was 0.74 ± 0.04 before and 0.73 ± 0.07 after application of NMDA (Figure 4; n = 10, p = 0.64). Although, on average, there was no change in RI, in three out of ten cells, there was an increase in response amplitude at −60mV and no change in amplitude at +40mV. This result suggests that new CP-AMPARs were inserted into the membrane, presumably through persistent exocytosis or membrane diffusion, and supports the hypothesis that NMDAR activation induces an exchange of CI-AMPARs for CP-AMPARs. Our findings suggest that direct pharmacological activation of NMDARs on ON and ON-OFF RGCs drives AMPAR plasticity, but they do not establish whether endogenous transmitter release from presynaptic ON bipolar cells can similarly drive NMDAR-dependent plasticity.