Figure 2. LA neurons with increased CREB function during CPP training become a critical component of a cocaine-cue memory engram. A, B, Neurons overexpressing CREB (but not those expressing GFP only) during training are more likely to be allocated to the cocaine-cue memory trace (arc+) than their noninfected neighbors. Mice were microinjected with CREB or GFP vector 2 d before training and catFISH for nuclear localized arc conducted 5 min after drug-free re-exposure to the chamber previously paired with cocaine. Active neurons (part of memory trace) = neurons with nuclear localized arc RNA. B, Example images from results quantified in A. In mice microinjected with CREB vector, arc RNA is preferentially localized in GFP+ (infected) neurons (rather than neighboring uninfected GFP− neurons), whereas in mice microinjected with GFP vector, arc is observed equally in infected and uninfected neurons. CREB vector n = 31 slices (5 mice), GFP n = 21 slices (3 mice). C, D, Post-training ablation of neurons overexpressing CREB (but not a similar number of random neurons expressing GFP alone) disrupts subsequent expression of cocaine-cue memory. C, iDTR transgenic or WT littermate mice microinjected with CREB-cre vector before training and administered DT (or PBS) after training to ablate CREB-overexpressing neurons. WT mice-PBS (n = 12), WT mice-DT (n = 13), iDTR mice-PBS (n = 11), and iDTR mice-DT (n = 12). D, iDTR mice microinjected with CREB-cre or GFP-cre vector before training and administered DT (or PBS) after training. No effect of ablating neurons expressing GFP alone, whereas ablating neurons overexpressing CREB disrupts cocaine-cue memory expression. GFP vector-PBS (n = 9), GFP vector-DT (n = 10), CREB vector-PBS (n = 12), and CREB vector-DT (n = 11). E, In contrast to behavioral extinction, COC administration before a memory test fails to reinstate a cocaine-cue memory after ablation of neurons overexpressing CREB. iDTR mice were microinjected with CREB-cre or GFP-cre vector before training and administered DT after training (as above). Ablation of CREB-overexpressing (but not GFP-expressing) neurons disrupted expression of cocaine-cue memory (Test 1, as above). Mice with GFP-cre vector were given extinction training (both chambers paired with SAL) until these mice no longer preferred the chamber originally paired with COC (extinction group). During this time, CREB-cre mice (neuronal ablation group) remained drug-free in homecage. Immediately before a final memory test (Test 4), all mice were administered COC. Cocaine-cue memory was reinstated in the extinction group only (not in the neuron ablation group), suggesting that ablating CREB-overexpressing neurons effectively degrades the cocaine-cue memory trace. GFP-cre+DT (n = 13), CREB-cre+DT (n = 11). F, Similar to permanent ablation, temporarily silencing neurons overexpressing CREB (using hM4Di DREADD+CNO) before a memory test disrupted cocaine-cue memory expression. GFP vector: CNO (n = 9), VEH (n = 7); CREB vector: CNO (n = 8), VEH (n = 9); GFP-hM4Di vector: CNO (n = 9), VEH (n = 8); CREB-hM4Di vector: CNO (n = 8), VEH (n = 7). Data presented are mean ± SEM. n.s. denotes note significantly different, * denotes p < 0.05, *** denotes p < 0.001.