a, VGAT-Cre mice were micro-infused with AAV-DIO-iC++-EYFP or AAV-DIO-EYFP virus in CA3 and optical fibers were implanted in the ADn, to optogenetically inhibit the CA3–ADn projection during fear memory test, at 1 or 28 d post-training. At 90 min after the test, mice were killed and their brains processed to analyze c-Fos expression in ADn. b, Representative images showing viral infection in CA3 and projections in ADn (repeated in n = 14 (iC++); n = 11 (EYFP); scale bar, 100 μm). c, At 1-d test, iC++- and EYFP-infused mice froze equally during light-OFF and light-ON epochs (iC++ n = 14; EYFP n = 11; two-way repeated-measures ANOVA iC++ versus EYFP × light-OFF versus light-ON; interaction, F1,46 = 0.13, P = 0.72; light-OFF versus light-ON, F1,46 = 5.71, P = 0.02; iC++ versus EYFP, F1,46 = 1.35, P = 0.25). d, At 28-d test, iC++- and EYFP-infused mice froze equally during light-OFF epoch (3-min duration), but iC++-infused mice showed reduced freezing in comparison with EYFP-infused mice during light-ON epoch (3-min duration) (iC++ n = 18; EYFP n = 17; two-sided Mann–Whitney test iC++ versus EYFP, P = 0.03; OFF: two-sided Wilcoxon signed-rank test iC++ versus EYFP, P = 0.68; ON: two-sided Wilcoxon signed-rank test iC++ versus EYFP, P = 0.014). e, At 1 d, iC++-infused mice showed an equivalent level of c-Fos expression in the ADn (normalized to mean expression level in home-cage control mice (n = 16)), in comparison with EYFP-infused control mice. But at 28 d, iC++-infused mice showed an elevated level of c-Fos expression in the ADn, in comparison with controls (1 d: iC++ n = 6; EYFP n = 5; 28 d: iC++ n = 9; EYFP n = 9; two-way ANOVA iC++ versus EYFP × 1 d versus 28 d; interaction, F1,25 = 4.77, P = 0.039; 1 d versus 28 d, F1,25 = 3.27, P = 0.083; iC++ versus EYFP, F1,25 = 12.62, P = 0.0015; post hoc Bonferroni’s test, 1-d iC++ versus EYFP P = 0.79, 28-d iC++ versus EYFP P = 0.0002). Data are individual mouse, or mean ± s.e.m. (*P < 0.05, ***P < 0.001).