RE innervates SO and SLM of dCA1.

(A-B) Experimental schema and virus expression. (A) Mice (Nanimals = 7) were stereotaxically injected into nucleus reuniens (RE) with AAV2.1 encoding mCherry. (B) A microphotography of mCherry expression in RE, virus penetrance and extent of viral expression (pink, minimal; red, maximal).

(C-D) Microphotographs showing RE axonal fibers (red), PSD-95 immunostaining (blue) and axon colocalization with PSD-95 (white) in dorsal hippocampus (C) and dCA1 strata (D): stratum oriens (SO), radiatum (SR) and lacunosum-moleculare (SLM). RE boutons (red) were identified as axon thickenings colocalizing with PSD-95-positive puncta (gray). In insets, identified boutons outlined (yellow) in ImageJ.

(E-G) Summary of results showing RE axon length (Mann-Whitney test, U = 11.5, p <0.001; effect size, SR-SO: −0.0974 [95CI 0.0152; −0.155]; SLM-SO: −0.0518 [95CI 0.190, −0.057]; SLM-SO: 0.131 [95CI 0.207; 0.0732], bouton density (Mann-Whitney test, U = 108, p = 0.060; Nanimals = 7; effect size, SLM-SO: −0.137 [95CI 0.004; −0.244]) and bouton area in dCA1 strata (Mann-Whitney test, U = 12550. p < 0.001; SO Nboutons = 266; SLM Nboutons = 282; effect size: SLM-SO, 0.367 [95CI 0.419; 0.321]). E, each dot represents one tissue brick. F, each dot represents one axon. G, violin plots are based on data for individual boutons.

(H-L) The analysis of synaptic responses in dCA1 in mice after chemogenetic inhibition of RE→dCA1.

(H) Experimental schema. Mice were stereotaxically injected with CAV2 encoding Cre-recombinase with green fluorescent protein (CAV2:Cre_GFP) into dCA1 and with AAV2.1 encoding Cre-dependent inhibitory DREADD with mCherry (AAV:DIO_hM4Di _mCherry) into RE, resulting in expression of Cre in neurons projecting to dCA1, and hM4Di in dCA1-projecting neurons of RE (RE→dCA1). Mice were i.p. injected with saline or CNO (3 mg/kg) and sacrificed 30 minutes later. Their brains were sliced and used for electrophysiological analysis of dCA1 synaptic responses.

(I) Microphotographs showing hM4Di expression in RE→dCA1 neurons, extent of viral expression (pink, minimal; red, maximal) and virus penetrance in RE.

(J-L) Microphotographs of the experimental setups (stim., stimulating electrode; rec., recording electrode), representative fEPSPs traces, summary of data for input–output plots of fEPSP slopes and fiber volley (FV) amplitude recorded in SO (J), SR (K) and SLM (L) in response to increasing intensities of stimulation. CNO decreased fEPSP in SLM (RM ANOVA, effect of CNO, F(1, 24) = 4.95, p = 0.036) and SO (F(1, 16) = 5.906, p = 0.027) but not in SR (F(1, 20) = 0.681, p = 0.419). CNO slightly decreased FV amplitudes in SLM (RM two-way ANOVA, F(1, 19) = 3.752, p = 0.068) but not SO (F(1, 10) = 0.020, p = 0.888) or SR (F(1, 25) = 0.042, p = 0.839).

Extinction of contextual fear remodels RE boutons in SLM.

(A) Experimental timeline and summary of data showing freezing levels during CFC and CFE session. After injection of AAV2.1 encoding mCherry into RE, Thy1-GFP(M) mice underwent CFC and were sacrificed 24 hours later (5US) or underwent CFE session and were sacrificed after the session (Ext) (RM ANOVA with post-hoc LSD test, effect of time: F(2.74, 10.9) = 8.25, p = 0.004). Naive animals were used as a control group.

(B) Summary of mCherry expression analysis in RE. Confocal scan of RE with local expression of mCherry, Penetrance of mCherry, and extent of viral expression (pink, minimal; red, maximal).

(C-D) RE axons analysis in dCA1.

(C) Representative images of RE axons, immunostaining for PSD-95 protein and their colocalisation in SO (Left) and summary of data showing bouton density (Mann Whiteny test, U = 1.73; p = 0.443; effect size, Naive − 5US: −0.0836 [95CI 0.0644; −0.2231]; Naive - Ext: −0.077 [95CI 0.0643; −0.2185]; 5US - Ext: 0.0065 [95CI 0.1457; −0.1326]), and bouton area (Mann Whiteny test, U = 0.096; p = 0.960; effect size, 5US - Naive: 0.0 [95CI 0.018; −0.018]; Ext - Naive: −0.0152 [95CI 0.0; −0.034]; Ext - 5US: −0.0040 [95CI 0.0; −0.035]; Naive Nboutons= 257, Nanimals= 6; 5US Nboutons= 203, Nanimals= 6; Extinction Nboutons= 400, Nanimals= 7) (Right).

(D) Representative images of RE axons, immunostaining for PSD-95 protein and their colocalisation in SLM (Left) and data analyzing of bouton density (unpaired t-test, t(46) = 0.642; p = 0.523; effect size, Naive - 5US: −0.0331 [95CI 0.1245; −0.1908]; Naive - Ext: −0.0054 [95CI 0.1631; −0.1522]; 5US - Ext: 0.0386 [95CI 0.1889; −0.1117]) and bouton area (Mann Whiteny test, U = 29409; p < 0.001; effect size, Naive - 5US: 0.1854 [95CI 0.2585; 0.1137]; Naive − Ext: 0.0614 [95CI 0.1346; −0.0097]; 5US - Ext: −0.132 [95CI −0.069; −0.192]; Naive Nboutons= 278, Nanimals= 7; 5US Nboutons= 272, Nanimals= 7; Extinction Nboutons= 291, Nanimals= 7) (Right).

Extinction of contextual fear remodels SLM dendritic spines on RE+ dendrites.

Thy1-GFP mice were injected with AAV2.1:camk2a_mCherry into RE and 21 days later they underwent CFC and were sacrificed 24 hours later (5US, n = 6), or underwent contextual fear extinction session and were sacrificed after the session (Extinction, n = 6) (see Figure 2A). Naive animals (n = 5) were used as a control. For the estimation statistics see Supplementary Figure 1 and Supplementary Tables 1-9.

(A) Representative high-magnification scans in SLM showing RE axons (mCherry), dendritic spines (GFP), immunostaining for PSD-95 protein (blue), axon x PSD-95 colocalizations (gray) and dendrite x PSD-95 colocalizations (gray). Colocalizations were identified on individual confocal planes using the Colocalization Highlighter function (ImageJ), and next transformed into max projections.

(B) Analysis of dendritic spines in SO. (left) Representative high-resolution images of RE axons (red), RE+ and RE- dendritic fragments (green), colocalization of dendrites with PSD-95 immunostaining (blue), and axons with PSD-95 (gray) in SO. Insets show magnification of RE+ spines. (right) Summary of data showing density of dendritic spines (two-way ANOVA with Šídák’s multiple comparisons test, effect of training: F(2, 14) = 0.526, p = 0.602; effect of spine type: F(1,39, 19,1) = 33.3, p < 0.001; spine type × training interaction: F(6, 41) = 2.54, p = 0.035; density of RE+/+ spines is plotted on the right Y axis), dendritic spine area (Kruskal-Wallis test with Dunn’s multiple comparisons test, All: U = 14.8; p < 0.001; RE-: U = 11.1; p = 0.004; RE+/−: U = 9.54; p = 0.008; RE+/+: U = 3.28; p = 0.191) and PSD-95 expression per dendritic spine (Kruskal-Wallis test with Dunn’s post hoc tests for planned comparisons, All: U = 116; p < 0.001; RE-: U = 92.03; p < 0.001; RE+/−: U = 57.33; p < 0.001; RE+/+: U = 1.44; p = 0.487; Naive, RE+/−: Nspines= 172; RE+/+: Nspines= 15; RE-: Nspines= 407; 5US, RE+/−: Nspines= 149; RE+/+: Nspines= 12; RE-: Nspines= 581; Extinction, RE+/−: Nspines= 302; RE+/+: Nspines= 16; RE-: Nspines= 949).

(C) Analysis of dendritic spines in SR. Representative high-resolution images of RE- dendritic fragments (green), and colocalization of GFP with PSD-95 immunostaining (blue). Summary of data showing density of dendritic spines (one-way ANOVA, F(2, 15) = 0,0085, p = 0.991), dendritic spine area (Kruskal-Wallis test, U = 3.51; p = 0.173) and PSD-95 expression per dendritic spine (Kruskal-Wallis test, U = 154; p < 0.001). Naive Nspines= 648; 5US Nspines= 643; Extinction Nspines= 715.

(D) Analysis of dendritic spines in SLM. (left) Representative high-resolution images of RE axons (red), RE+ and RE- dendritic fragments (green), colocalization of dendrites with PSD-95 immunostaining (blue), and axons with PSD-95 (gray). (right) Summary of data showing density of dendritic spines (two-way ANOVA with LSD post hoc test, effect of training: F(2, 13) = 2.60, p = 0.113; effect of spine type: F(3, 37) = 58.3, p < 0.001, spine type × training interaction: F(6, 37) = 1.34, p = 0.265; density of RE+/+ spines is plotted on the right Y axis), dendritic spine area (Kruskal-Wallis test with Dunn’s multiple comparisons test, All: U = 28; p < 0.001; RE-: U = 8.34; p = 0.015; RE+/−: U = 33; p < 0.001; RE+/+: U = 11.3; p = 0.004) and PSD-95 expression per dendritic spine (Kruskal-Wallis test with Dunn’s multiple comparisons test, All: U = 72.3, p < 0.001; RE-: U = 5.92; p = 0.052; RE+/−: U = 140; p < 0.001; RE+/+: U = 6.50; p = 0.039). Naive, RE+/−: Nspines= 182; RE+/+: Nspines= 16; RE-: Nspines= 524; 5US, RE+/−: Nspines= 225; RE+/+: Nspines= 15; RE-: Nspines= 424; Extinction, RE+/−: Nspines= 357; RE+/+: Nspines= 27; RE-: Nspines= 541.

Repeated exposure to novel context exposure neither axonal boutons nor dendritic spines in SLM.

(A) Experimental timeline and summary of data showing mice activity levels during novel context exposure. Thy1-GFP mice were stereotaxically injected with AAV2.1:camk2a_mCherry into RE and 21 days later they were exposed to a novel context. Mice were sacrificed 24 hours after novel context exposure (Ctx, Nanimals = 6) or they were re-exposed to the context and sacrificed immediately afterwards (Ctx-Ctx, Nanimals = 6).

(B-C) Morphological analysis of RE axons in SLM. (B) Summary of data showing bouton density (unpaired t-test, t(10) = 0.755, p = 0.468; effect size for Ctx-Ctx - Ctx: −0.069 [95CI 0.0351; −0.139]) and (C) bouton area (Mann Whiteny test, U = 2745; p = 0.48; effect size, Ctx-Ctx - Ctx: −0.028 [95CI 0.054; −0.110]). Ctx: Nboutons = 230, Nanimals= 6; Ctx-Ctx: Nboutons= 239, Nanimals= 6.

(D-E) Analysis of dendritic spines in SLM. (D) Summary of data showing spine density (two-way ANOVA with Šídák’s multiple comparisons test, effect of training: F(1, 20) = 1.29, p = 0.270, effect of spine type: F(1.91, 12.8) = 53.1, p < 0.001; effect size for Ctx-Ctx − Ctx, All:−0.071 [95CI 0.044; −0.186]; RE-:−0.062 [95CI 0.099; −0.223]; RE+/−:−0.182 [95CI −0.033; −0.330]; RE+/+: 0.007 [95CI 0.211; −0.197];) and (E) spine area (Kruskal-Wallis tests Dunn’s multiple comparisons test; U = 19.9, p = 0.004; effect size for Ctx-Ctx − Ctx, All: 0.027 [95CI 0.056; −0.001]; RE-: 0.030 [95CI 0.077; −0.015; RE+/−: 0.035 [95CI 0.072; −0.002]; RE+/+:−0.003 [95CI 0.216; −0.222]; Ctx RE-: Nspines= 458, RE+/−: Nspines= 590, RE+/+: Nspines= 17; Ctx-Ctx RE-: Nspines= 338, RE+/−: Nspines= 589, RE+/+: Nspines= 18. For A, B and D, data are presented as means and SEM. For C and E, data are presented as medians and IQR.

Contextual fear extinction remodels postsynaptic densities of the excitatory synapses in SLM.

Mice underwent CFC and were sacrificed 24 hours later (5US, n = 5) or after CFE session (Ext, n = 6). Naive animals were used as a control (n = 4). For the estimation statistics see Supplementary Tables 10-17.

(A-B) The principles for SBFSEM analysis of dendritic spines and PSDs in SLM. (A) Tracing of a dendritic spine and PSD. A representative trace of a dendritic spine (blue), PSD surface area (red) and volume (yellow).

(B) Exemplary reconstructions of dendritic spines and their PSDs from SBFSEM scans in SLM for the 5US and Ext groups. Dendritic spines and PSDs were reconstructed and analyzed in tissue bricks (6 × 6 × 2.5 μm). The gray background square is x = 6 × y = 6 μm.

(C-G) Summary of SBFSEM data:

(C) density of dendritic spines (F(2, 13) = 0.2, p =0.829; effect size for 5US-Naive: −9.08 [95CI 23.48; −41.64]; Ext-Naive: −6.19 [95CI 24.981; −37.377]; Ext-5US: 2.88 [95CI 34.06; −28.29]) and volume of dendritic spines (Kruskal-Wallis test with Dunn’s post hoc tests, U = 13.2, p = 0.001; effect size for 5US-Naive: −0.0019 [95CI 0.0017; −0.0055]; Ext-Naive: −0079.19 [95CI −0.0035; −0.013]; Ext-5US: 0.0055 [95CI 0.010; 0.0007]; Naive N=315, 5US N = 238, Ext N = 259;

(D) surface area of PSDs (Kruskal-Wallis test with Dunn’s post hoc tests, U = 17.9, p < 0.001; effect size for 5US-Naive: −0.005 [95CI 0.0012; −0.012]; Ext-Naive: −0.017 [95CI 0.0096; −0.026]; Ext-5US: 0.011 [95CI 0.021; 0.0023]) and correlation of dendritic spine volume and PSD surface area (rs, Spearman correlation r), each dot represents one dendritic spine;

(E) volume of PSDs (Kruskal-Wallis test with Dunn’s post hoc tests, U = 17.2, p < 0.001; effect size for 5US-Naive: −0.00001 [95CI 0.0002; −0.0002]; Ext-Naive: −0.0005 [95CI −0.0002; −0.0008]; Ext-5US: 0.0005 [95CI 0.0008; 0.0002]) and correlation of dendritic spine volume and PSD volume (rs, Spearman correlation r), each dot represents one dendritic spine;

(F) exemplary reconstruction of spine with macular PSD and data analysis for density of macular PSDs (one-way ANOVA, F(2, 13) = 1.047, p = 0.3788; effect size for 5US-Naive: −10.57 [95CI 41.83; −20.679; Ext-Naive: 20.04 [95CI 49.96; −9.88; Ext-5US: −9.46 [95CI 39.389; −20.45]) and volume of macular PSDs (Kruskal-Wallis test with Dunn’s post hoc tests, U = 12.2, p = 0.002; e; effect size for 5US-Naive: 0.00003 [95CI 0.0002; −0.00017]; Ext-Naive: −0.0003 [95CI −0.0001; −0.0006]; Ext-5US: 0.0003 [95CI 0.0006; 0.0001]; Naive N = 212, 5US N = 163, Ext N = 168);

(G) exemplary reconstruction of spine with complex PSD and data analysis for density of complex PSDs (one-way ANOVA, F(2, 13) = 1.57, p = 0.246; effect size for 5US-Naive: 18.86 [95CI 58.9; −21.18]; Ext-Naive: 25.03 [95CI 63.37; −13.30]; Ext-5US: 6.17 [95CI 44.51; −31.16]) and volume of complex PSDs (Kruskal-Wallis test with Dunn’s post hoc tests, U = 12.6, p = 0.002; effect size for 5US-Naive: −0.0007 [95CI 0.000012; −0.0013]; Ext-Naive: −0.001 [95CI −0.0006; −0.0019]; Ext-5US: 0.0005 [95CI 0.0013; −0.0002]; Naive N = 103, 5US N = 75, Ext N = 91).

Chemogenetic inhibition of RE→dCA1 impairs extinction of contextual fear memory.

(A) Experimental design. Mice received stereotaxic injection of AAV2.1:hSyn_DIO_hM4Di_mCherry (hM4Di) or control virus (AAV2.1:hSyn_mCherry) into RE, and CAV2:Cre_GFP into dCA1.

(B) Representative microphotographs of hM4Di expression in RE and the extent of viral expression (blue, minimal; dark blue, maximal).

(C) Experimental timeline.

(D) Summary of data for freezing levels of the hM4Di group during the training (Training: RM ANOVA with Fisher’s LSD post hoc test, effect of time: F(1, 21) = 446, p < 0.001; effect of CNO: F(1, 21) = 0.165, p = 0.689; time × CNO interaction: F(1, 21) = 6.64, p = 0.018), fear extinction session (Extinction: RM ANOVA with Šidak’s post hoc test, effect of time: F(1, 21) = 95.55, p < 0.001; effect of CNO: F(1, 21) = 0.259, p = 0.616; time × CNO interaction: F(1, 21) = 0.1740, p = 0.681) and fear extinction test (Test: unpaired t-test: t(20) = 3.26, p = 0.002; effect size: 25.442 [95CI 9.492; 41.392]).

(E) Summary of data for freezing levels of the Control group during the training (Training: RM ANOVA with Fisher’s LSD post hoc test, effect of time: F (1, 10) = 119 P<,001; effect of CNO: F (1, 10) = 2,86 P=,122; time × CNO interaction: F (1, 10) = 3,13 P=,107), fear extinction session (Extinction: RM ANOVA with Šidak’s post hoc test, effect of time: F (2,897, 31,86) = 21,32, p < 0.001; effect of CNO: F (1, 11) = 2,320, p = 0.156; time × CNO interaction: F (5, 55) = 1,472, p = 0.240) and fear extinction test (Test: unpaired t-test: t(10) = 0.149, p = 0.884; effect size: 1.362 [95CI −19.016; 21.749]).

(F) Experimental timeline and summary of data for freezing levels during the training. After stereotactic injection of DIO_hM4Di into RE and CAV:Cre_GFP into dCA1 mice underwent CFC (Training: RM ANOVA with Šidak’s post hoc test, effect of time: F(1, 14) = 218, p < 0.001; effect of CNO: F(1, 14) = 0.001, p = 0.974; time × CNO interaction: F(1, 14) = 0.010, p = 0.919), CFE session (Extinction 1: RM ANOVA with Šidak’s post hoc test, effect of time: F(1, 14) = 45.5, p < 0.001; effect of CNO: F(1, 14) = 0.121, p = 0.73; time × CNO interaction: F(1, 14) = 0.020, p = 0.89), short-term CFE memory test (Test 1: t-test: t(14) = 2.45, p = 0.028; effect size: 23.935 [95CI 8.310; 39.561]), long-term CFE memory test (Extinction 2: RM ANOVA with Šidak’s post hoc test, effect of time: F(1, 14) = 23.1, p < 0.001; effect of CNO: F(1, 14) = 10.5, p = 0.006; time × CNO interaction: F(1, 14) = 0.028, p = 0.87) and exposure to the context B (CtxB: unpaired t-test: t(14) = 0.057, p = 0.955; effect size: −3.306 [95CI −22.850; 16.239]). Due to technical problems only 6 mice out of 12 from the CNO group were trained on Day 4 (Test 2 and CtxB)

(G) Experimental timeline and summary of data for freezing levels during the training. Mice were stereotaxically injected into RE with AAV2.1 encoding hM4Di and cannulae placed into dCA1. Next, mice underwent CFC (Training: RM ANOVA with Šidak’s post hoc tests, effect of time: F(1, 12) = 146, p < 0.001; effect of CNO: F(1, 12) = 1.60. p = 0.23; time × CNO interaction: F(1, 12) = 0.12, p = 0.72), CFE session (Extinction: RM ANOVA with Šidak’s post hoc test, effect of time: F(1, 15) = 36, p < 0.001; effect of CNO: F(1, 15) = 2.14, p = 0.16; time × CNO interaction: F(1, 15) = 4.59, p = 0.49) and long-term fear extinction memory test (unpaired t-test, t(15) = 2.30. p = 0.036; effect size: 22.803 [95CI 2.383; 43.223]).

(H) Experimental timeline and summary of data for freezing levels during the training. Mice were stereotaxically injected into RE with AAV2.1 encoding mCherry and cannulae were placed into dCA1. After surgery they underwent CFC (Training: RM ANOVA with Šidak’s post hoc tests, effect of time: F(1, 10) = 84.3, p < 0.001; effect of CNO: F(1, 10) = 2.18, p = 0.17; time × CNO interaction: F(1, 10) = 1.61, p = 0.23), CFE session (RM ANOVA with Šidak’s post hoc test, effect of time: F(1, 11) = 25.8, p < 0.001; effect of CNO: F(1, 11) = 0.006, p = 0.93; time × CNO interaction: F(1, 11) = 0.12, p = 0.72) and CFE memory test (unpaired t-test, t(11) = 1.12, p = 0.28; effect size: 13.718 [95CI 13.256; 40.696]).

(I) Microphotography and schematic representations of cannulae placement.

RE regulates PSD-95 expression in dCA1 during contextual fear extinction.

(A) Experimental timeline and summary of data showing freezing levels during CFC and CFE. After injection of AAV2.1 encoding mCherry (Control) or hM4Di into RE, mice underwent CFC (two-way ANOVA, effect of training, F(1, 26) = 322, p < 0.001; effect of group, F(3, 26) = 0.094, p = 0.962) and fear extinction session (effect of training, F(1, 26) = 144, p < 0.001; effect of group, F(3, 26) = 1.12, p = 0.358). Mice were injected with CNO or Saline 30 minutes before Extinction and were sacrificed immediately after the session. The dCA1 brain slices with visible RE axons were used for PSD-95 immunostaining.

(B-C) AAVs expression analysis in RE. (Left) Confocal scan of RE with local expression of mCherry. (Right) Penetrance of mCherry in experimental groups. (C) The extent of viral expression.

(D) Representative images of PSD-95 immunofluorescent staining (green) and RE axons (red) in 3 strata of dCA1 (Left) and data analysis (three-way RM ANOVA with post hoc LSD test for planned comparisons, effect of virus: F(1, 21) = 8.58, p = 0.008; effect of CNO: F(1, 21) = 0.566, p = 0.460; virus × CNO interaction: F(1, 21) = 7.72, p = 0.011; effect size for Control CNO − Control sal, SO: −1.190 [95CI 8.703; −11.085]; SR: 4.557 [95CI 18.149; −8.494]; SLM: −2.197 [95CI 10.410; −14.806]; effect size for hM4Di CNO − hM4Di sal, SO: −4.891 [95CI 3.427; −13.411]; SR: −8.757 [95CI 1.587; −19.697]; SLM: −9.481 [95CI −1.725; −17.037]).

The numbers of the analyzed sections and animals are indicated in the legend.