(test 1) that followed the RCS (Fig. 1G). Reten-
tion of the CTA memory was disrupted in the
anisomycin group compared with control (Fig. 1H).
This disruption was accompanied by an impair-
ment in the original AFC memory, although the
AFC memory had not been directly reactivated
during the anisomycin infusion (Fig. 1I). A sec-
ond nonreactivated group of mice that had not
received the RCS (no RCS, home cage group) also
showed the CTA memory impairment, but their
AFC memory recall was intact (Fig. 1, J and K).
Cellular compartment analysis of temporal activity by use of fluorescence in situ hybridiza-tion with Arc and Homer 1a (Arc/H1a catFISH)
(23, 24) revealed a dynamic interaction between
CTA and AFC at the neuronal ensemble level in
all the brain areas analyzed (Fig. 2 and fig. S4).
The population of the Arc/H1a–double-positive
neurons in the basolateral amygdala of the co-
retrieval group was significantly greater than
that in the control and no-reactivation groups,
despite the same-sized population of Arc/H1a–
single-positive cells among the groups (Fig. 2, B
to E, and fig. S4). The percentage of the coshared
neuronal population (overlapping ensemble) in
the amygdala between CTA and AFC to the total
ensembles activated during retrieval was 15.2%
Fig. 4. Overlapping ensemble links two distinct emotional memories. (A) Strategy to label the overlapping ensemble in double-transgenic
mice (c-Fos::tTA/R26R::H2B-mCherry) with the LVs TRE3G::CreERT2,
E-SARE::DIO-Arch T-EYFP (top). Schematics showing labeling and manipulating intersectional subpopulations of repeatedly activated ensembles
during CTA and AFC retrieval tests (bottom). (B) Time-dependent distribution of intracellular CreERT2 signal after AFC retrieval test (top). The white
dotted lines outline nuclei (bottom). Scale bar, 10 mm (C) Quantitative
classification of CreERT2+ cells by intracellular signal location after the
AFC retrieval test. (D) Procedure for targeting the overlapping ensemble
with Arch T-EYFP. (E) Overlapping ensembles expressing the Arch T-EYFP in coretrieval and control groups. Control group received only CTA-CS presentation
during RCS. The yellow arrowheads indicate an Arch T-EYFP+, H2B-mCherry+ (double-positive) cell in the overlapping ensemble (insets). Scale bar, 100 mm;
inset, 20 mm. (F and G) Quantitative population analysis of Arch T-EYFP+ cells and mCherry+ cells (F) and Arch T-EYFP+, H2B-mCherry+ (double-positive)
cells (G). Coretrieval, control, test 2(−), n = 5, 4, 5 mice, respectively; H2B-mCherry+, P > 0.36, F2,13 = 1.13; Arch T-EYFP+, P = 7.44E-5, F2,13 = 25.47; Arch T-EYFP+, H2B-mCherry+, P = 0.001, F2,13 = 13.36. Red line indicates chance. (H) Procedure for suppressing the neuronal activity of the overlapping ensemble
through yellow-light illumination. Mice received the same treatment until test 2 in (D). (I) Saccharin-induced freezing (%) during test 2, test 3 with light
illumination, and test 5 (CTA-paired and CTA-unpaired, n = 13 and 12 mice, respectively; two-way ANOVA, Tukey-Kramer post hoc test, P = 0.00051, F2,74 = 8.49).
(J) Aversion index in test 2, test 3, and test 5 of CTA-paired and CTA-unpaired groups (two-way ANOVA, P > 0.64, F2, 74 = 0.45). (K) Tone-induced freezing (%) of
AFC-paired and AFC-unpaired groups with light illumination in the AFC retrieval test (test 3) (AFC-paired, n = 8 mice; AFC-unpaired, n = 7 mice; t test, P > 0.78,
t13 = 0.28). Data are means ± SEM. Significance for multiple comparisons, *P < 0.05, **P < 0.01.