Endopiriform projects to the ventral CA1

(A) Schematic diagram of a retrograde tracer injection performed. CAV-2-Cre or CTB647 was used in combination with Ai14 or wild-type mice, respectively. (B) An example fluorescent image of injection site. (C) Distribution of retrogradely labeled vCA1-projecting neurons in entire brain as a result of injection in panel B. (D-G) Fluorescent images showing retrogradely labeled vCA1-projecitng neurons in septum (SP), entorhinal cortex (ENT), basolateral amygdala (BLA), and endopiriform (EN). Str: Striatum, CLA: Claustrum. (H) Box plots showing relative number of vCA1-projecting neurons in different brain areas. Cell number in each brain area were normalized by summed cell number of EN, BLA, SP (lateral), and SP (medial). Cell numbers in each brain areas of individual brains are presented in Table 1. Median value for EN, BLA, SP (lateral), and SP (medial) is 0.45, 0.24, 0.08, and 0.21, respectively. Statistics: EN vs BLA, SP (lateral), or SP (medial), p < 0.001 for all; BLA vs SP (lateral) or SP (medial): p = 0.002 and 0.694, respectively; SP (lateral) vs SP (medial) p = 0.023. One-way ANOVA, repeated comparison with Tukey-Kramer test. N = 5. (I) An example fluorescent image of horizontal section showing retrogradely labeled vCA1-projecting neurons in EN. A: anterior, P: posterior. (J) Top: Example map showing distribution of vCA1-projecting neurons in entire EN in one mouse. Bottom left: A plot showing a quantified distribution at bin size of 1 mm. Each bin was normalized to total number of labeled neurons in EN. Data from each mouse is shown in grey. Median is indicated in black. Bottom right: Box plots comparing a number of vCA1-projecting neurons in anterior and posterior EN. Data were normalized to a total number of labeled neurons in EN. Median value for anterior vs posterior was 0.39 vs 0.54. Statistics: p = 0.008, Rank-sum.

Related to Fig. 1H

Axon branching of vCA1-projecting EN neurons

(A) Schematic diagram of injection performed and two hypothesized outcomes. Yellow color indicates double-labeled neurons. PFC: Prefrontal cortex, ACC: Anterior cingulate cortex, MC: Motor cortex, ENT(dm): Dorsomedial entorhinal cortex, CLA: claustrum (B-F) Example fluorescent images of injection site for each region indicated. (G-J) Example fluorescent images of retrogradely labeled neurons in CLA complex. vCA1-projecting neurons were colored in red and cortex-projecting neurons were colored in green. (K) Box plots showing percentage of double-labeled vCA1-projecting neurons for each cortical injection. Medial value for PFC, ACC, MC, and ENT(dm) was 28.6, 4.6, 0, and 1.2, respectively. Statistics: PFC vs ACC, MC or ENT(dm), p = 0.405, 0.015 and 0.047, respectively. ACC vs MC or ENT (dm), p = 0.422 and 0.64, respectively. MC vs ENT (dm), p = 0.993. One-way ANOVA with post-hoc Kruskal-Wallis test. Slice and animal number: 16 and 4 for PFC, 20 and 5 for ACC, 10 and 3 for MC, and 18 and 4 for ENT(dm). (L) Schematic diagram of projection pattern found in the experiment shown in A. (M) Schematic diagram of the intersectional approach performed to label vCA1-projecting EN neurons. (N-O) Example fluorescent images showing vCA1-projectig axons in different brain areas. All images were from the same mouse. (P) Magnified images showing vCA1-projecting axons in cortical layers. Magnified areas are indicated in O and fig. S2A-C as dashed squares. PIR: Piriform cortex

EN projection pattern in vCA1

(A) Schematic diagram of injection performed. (B) An example fluorescent image of horizontal brain section containing injection site in EN. (C) An example fluorescent image of ventral hippocampal slice with EN axons resulted from the injection in panel B. Dashed lines indicate the distal and proximal vCA1 regions where GFP signals were measured for the plot in panel D. (D) Normalized GFP signals across the layer. All signals were normalized by maximum GFP signal in the distal subregion. Median of area under the curve: distal; 11.12, proximal; 5.99. distal vs proximal, p = 0.016. Wilcoxon signed-rank test. n = 8 slices (N = 4 mice). SR: stratum radiatum, Py: pyramidal layer.

ENvCA1-proj. neurons target vCA1 interneurons

(A) Schematic diagram of injection performed for monosynaptic rabies tracing. Flox-TVA-GFP (60 nl) was injected into the vCA1. 4 weeks later, Rabies-mCherry (60 nl) was injected. (B) Location of starter cells in the hippocampus. Data from 5 brains (each brain represented by different color) were overlaid. (C) Example images showing presynaptic cells in EN (white arrow). (D) Top: Location of presynaptic EN neurons. Data from 5 brains were overlaid (different brains were coded by different colors). Bottom: Plots indicate distribution of presynaptic cells in EN along AP axis. Cell counts in a given bin (1 mm) were normalized by total cell number in EN. Thin gray lines indicate individual data. A thick black line indicates median. Inset showing plot comparing cell number in anterior and posterior location as in Fig. 1J. Median: anterior vs posterior, 0.25 vs 0.72, p = 0.031; Wilcoxon signed-rank test. (E) Top: Schematic diagram of injection performed. Bottom: An example fluorescent image of injection site in EN. (F) Schematic diagram of recordings performed. Recordings were performed from identified GABAergic neurons sequentially in a random order. (G) An example confocal image of an acute vCA1 slice used for recordings. Recorded GABAergic neurons were filled with biocytin and post-labeled with Alexa488. White dashed lines indicate border of Py-SR and SR-SLM. (H) Median traces of photo-evoked EPSCs recorded from neurons at different laminar position. (I) Connection probability (CP) of EN axons and recorded neurons at different laminar positions. Numbers in bars indicate number of response positive neurons /total recorded neurons. (J) Left: Peak amplitudes of evoked EPSCs (normalized to total input from recorded slice). Right: the same as left but for mean amplitude. Gray circles indicate each neuron. Black lines indicate median. Median (peak amp): Py-SR vs SR, SR-SLM or SLM: 1.1 vs 1.0 vs 1.1 vs 0.0. [Py-SR vs SR, SR/SLM vs SLM], p = 0.999, 0.988, 0.076; [SR vs SR-SLM or SLM], p = 0.972, 0.111; [SR-SLM vs SLM], p = 0.039; one-way ANOVA, repeated comparison with Tukey-Kramer test. Median (mean amp): Py-SR vs SR, SR-SLM, or SLM: 1.0 vs 0.8 vs 1.4 vs 0.0. [Py-SR vs SR, SR-SLM or SLM], p = 0.991, 0.999, and 0.141; [SR vs SR-SLM or SLM], p = 0.978, and 0.251; [SR-SLM vs SLM], p = 0.114; one-way ANOVA, repeated comparison with Tukey-Kramer test. N = 5 mice.

EN disynaptically inhibits vCA1 pyramidal neurons

(A) A confocal image of recorded pyramidal neurons in vCA1. Recorded pyramidal neurons were filled with biocytin and post-labelled with Alexa488. Ori.: oriens, Super.: superficial layer pyramidal neuron. Deep: deep layer pyramidal neuron. (B) Example traces of photo-evoked EPSCs and IPSCs in superficial (black trace) and deep (gray trace) pyramidal neurons. Red trace: after TTX and 4AP. Blue trace: after gabazine. (C) Scatter plots of peak amplitudes of EPSCs and IPSCs for each neuron. Dashed line indicates a unitary line. Superficial neurons: empty Black circles. Deep neurons: Gray circles. N = 4 mice. (D) A summary diagram of EN◊vCA1 circuit. In vCA1, EN axons innervate GABAergic neurons (IN), which in turn inhibits pyramidal neurons (both superficial and deep).

ENvCA1-proj. neurons receive inputs from the piriform cortex and within EN

(A) Schematic diagram of performed. After CAV2-Cre (60 nl) injection into the vCA1, flox-TVA-mCherry (100 nl) was injected into the EN. 4 weeks later, Rabies-GFP (100 nl) was injected into the EN. (B) Location of starter cells in EN. Data from 6 brains were overlaid. (C) An example map of presynaptic cells in one brain. Top: Horizontal view. Injections were made in left hemisphere. Bottom: Front view. (D) Number of presynaptic cells in different brain areas (normalized to total presynaptic cell number). Median for AIA, AON, PIR, ENT(lateral), CLA, EN, Pallidum, and SI (in fraction): 0.7, 0.3, 3.8, 0.3, 0.5, 1.7, 0.3, and 0.2. [AON, ENT(lateral), CLA, Pallidum, SI] vs PIR: p = 0.016, 0.008, 0.048, 0.006, and 0.003. EN vs SI, p = 0.043. one-way ANOVA with post-hoc Kruskal-Wallis test. Full list of statistical tests and p values are in Table 2. (E) An example fluorescent image of coronal slice containing EN and PIR and labeled presynaptic cells. (F) Example maps of presynaptic cells in PIR (top) and EN (bottom). (G) Number of presynaptic cells in PIR in different layers (normalized to the total number of labeled neurons in PIR). Median: L1 vs L2 vs L3, 0.11 vs 0.61 vs 0.27. L1 vs L2, p < 0.001; L1 vs L3, p = 0.005; L2 vs L3, p < 0.001; one-way ANOVA, repeated comparison with Tukey-Kramer test. (H) Summary diagram showing input-output circuit of ENvCA1-proj. neurons (red circle).

Related to Fig.6D

Activity of ENvCA1-proj. neurons correlates social/object discrimination performance

(A) Right and Middle: Schematic diagrams of injection and optic cannula implantation performed. CAV2-Cre (60 nl) was injected into the vCA1, and dlox-jGCaMP8s (100 nl) was injected into the EN. Left: An example image showing jGCaMP8s expressing ENvCA1-proj. neurons and a tract of optic cannula. (B) The diagram of social recognition memory test. A test consisted of pretest (5 min), sociability test (10 min), and discrimination test (5 min). (C) Left: An example image of a mouse with nose point marker tracked by DLC. Right: An example map of nose points and calcium events in the arena during sociability test. (D) Nose distance vector (distance between subject’s nose point and a center point of novel/familiar mouse chambers) and simultaneously acquired calcium signals during discrimination test. Gray dashed lines indicate the border of interaction zones. Open red circles indicate the calcium event detected (threshold: Z > 2.58). A part of calcium signal trace was expanded at the bottom for clarity. (E) Left: Correlation between calcium event ratios (calcium event number in the mouse interaction zone/ calcium event number in the object interaction zone) and sociability index. p = 0.29, N = 8 mice. Right: same as in left panel, but for calcium event ratios (calcium event number in the unfamiliar mouse interaction zone/ calcium event number in the familiar mouse interaction zone) and discrimination index. p = 0.003, N = 8. (F) The diagram of novel object recognition memory test. A test consisted of habituation (5 min), familiarization (10 min), and discrimination test (5 min). (G) Same as in (E), but for calcium event number ratios (calcium event number in the unfamiliar object interaction zone/calcium event number in the familiar object interaction zone and discrimination index. p = 0.01. N = 9 mice.

Chemogenetic inhibition of ENvCA1-proj. neurons specifically impairs recognition memory

(A) Schematic diagram of injection performed. (B) Left: An example fluorescent image showing expression of hM4Di in ENCA1-proj. neurons. Right: Expansion of areas marked by white rectangle in image on the left. (C) The diagram of social recognition memory test. A session consisted of habituation (5 min), Pretest (5 min), Sociability test (10 min), and discrimination test (5 min). The group treated with a saline in session 1 was treated with CNO (1mg/ kg) in session 2 the following day. (D) Example cumulative times heat map in arena during sociability tests for the same mouse under two different treatments. (E) Plots comparing sociability index for hM4Di and tdTomato mice under two different treatments. Gray empty circles indicate each mouse. Gray thin lines show paired data from one mouse. Thick black bars indicate median. Median sociability index: hM4Di mice (Saline vs CNO), 0.75 vs 0.76, p = 0.391; signed-rank test; tdTomato mice (Saline vs CNO), 0.77 vs 0.75, p = 0.383; signed-rank test. (F) Example cumulative time heat maps in arena during discrimination tests for the same mouse under two different treatments. (G) Comparison of discrimination index for hM4Di and tdTomato mice under two different treatments. Gray empty circles indicate each mouse. Gray thin lines show paired data from one mouse. Thick black bars indicate median. Median discrimination index: hM4Di mice (Saline vs CNO), 0.66 vs 0.57, p = 0.042; signed-rank test; tdTomato mice (Saline vs CNO), 0.65 vs 0.63, p = 0.742; signed-rank test. (H) The diagram of novel object recognition memory test. A session consisted of habituation (5 min), familiarization (10 min), and discrimination test (5 min). CNO treatment was the same as in C. (I) Same as in G, but for novel object discrimination index. Median discrimination index: hM4Di mice (Saline vs CNO), 0.64 vs 0.5, p = 0.003; signed-rank test.

The role of ENvCA1-proj. neurons in recognition memory.

(A) Summary of findings from in vivo experiments. (B) The model of ENvCA1-proj. circuit function in recognition memory. During memory-guided exploration, recurrent circuits in ENvCA1-proj. neurons maintain the attentional response to novel stimuli while regulating the degree of response to familiar stimuli in vCA1 via feedforward inhibition.

Whole brain distribution of vCA1-projecing neurons

(A) Fluorescent images of the injection site (vCA1) of CAV2-Cre (60 nl) of Ai14 mice for analysis of presynaptic neuron distribution using AMaSiNe. The presynaptic cell numbers counted in individual brain are presented in Table 1. (B) An example distribution map of vCA1-projecting neurons in SP (lateral and medial), BLA and ENT (lateral and medial) shown from front and top view. Neurons in different brain areas are color-coded. (C) The side view of the brain shown in Fig. 1J. Plots indicate distribution of vCA1-projecting neurons in EN along dorsal-ventral axis (DV). Data from different mice are shown in gray. Median is indicated in black.

Axon branching of vCA1-projecting EN neurons

(A-D) Example fluorescent images of branching of vCA1-projecting neurons. All images were from the same mouse shown in Fig. 2N-O. DG: Dentate gyrus

Projections of ENvCA-proj. neurons to the hippocampus

(A) Top: An example fluorescent images of intermediate hippocampal sections from the same brain as in Fig. 3B and C. Bottom: GFP signal profile at the distal and proximal region indicated in the top image. (B) The same as in (A) but for dorsal hippocampus. Median area under the curve (AUC): distal vs proximal (intermediate), 6.3 vs 3.5, p = 0.007; distal vs proximal (dorsal), 4.3 vs 3.0, p = 0.055; Wilcoxon signed-rank test. n = 8 slices (N = 4 mice). (C) Comparison of GFP signal (in AUC) between different hippocampal axis for distal or proximal vCA1. Distal; ventral vs intermediate, p = 0.086; ventral vs dorsal, p < 0.001; intermediate vs dorsal, p = 0.136. Proximal: ventral vs intermediate, p = 0.018; ventral vs dorsal, p = 0.001; intermediate vs dorsal, p = 0.695; Kruskal-Wallis test. n = 8 slices (N = 4 mice).

Presynaptic neurons of ventral hippocampal GABAergic neurons

(A) Example fluorescent images of starter cells in vCA1. TVA_oG expressing neurons were labeled by GFP and rabies expressing neurons were labeled by mCherry. GFP and mCherry double expressing neurons were counted as starter cells. White allows in eight panel indicate starter cells. White dashed lines indicate the Py-SR and SR-SLM border. (B) An example map of presynaptic cells in a whole brain. Injection was made in the left hemisphere. (C-F) Fluorescent images of presynaptic cells in coronal sections from the same mouse brain. Presynaptic neurons were in EN (C), septum(SP) (D), nucleus reunion (RE) (E), contralateral vCA3 (vCA3) (F), ventral tegmental area (VTA) (F), and lateral ENT (F).

EN-mediated excitatory inputs to ventral hippocampal neurons

(A) Plots comparing peak amplitude of photo-evoked EPSCs in superficial and depp pyramidal neurons. Thin gray lines indicate paired data. Median peak amplitude; superficial vs deep (in pA): 11.5 vs 5.9, p = 0.625; signed-rank test. (B) Plots comparing peak amplitude of photo-evoked EPSCs in pyramidal and GABAergic neurons. The data from superficial and deep pyramidal neurons were pooled. n = 10 for pyramidal neurons and 11 for GABAergic neurons. N = 4 mice. Median EPSC peak amplitude: Pyramidal vs GABAergic (in pA): 7.4 vs 8.0, p = 0.86. Rank-sum test.

Presynaptic neurons of ENvCA1-proj. neurons

(A) An example fluorescent image of starter cells in EN. (B-D) Expanded images of the area marked by white square in panel A. White allows indicate starter cells. (E) Example fluorescent images of coronal sections in series. Labeled neurons (in green) indicate presynaptic cells of ENvCA1-proj. neurons. TTd: Taenia tecta, AON: Anterior olfactory nucleus, PIR: Piriform cortex, NAc: Nucleus accumbens, EN: Endopiriform, CLA: Claustrum, BLA: Basolateral amygdala, ENT: Entorhinal cortex.

Nose point location and calcium signal data during social memory test

Maps of nose points (filled gray circles) and calcium events (filled red circles) in arena. Heat maps of the same data (but cumulative time and calcium event are separated) were also shown for each mouse (N = 8). Peason’s correlation coefficients with p < 0.05 is indicated by asterisk. All p values were presented in table S1.

Calcium signals during exploratory behavior in open field, pretest and familiarization test.

(A) Example of maps of nose points and calcium events in open field arena. All p values in a correlation map were presented in table S2. (B) Correlation plots of event number and cumulative time in each pixels shown in A. p < 0.001, N = 7. (C) Box plots comparing a calcium event number when the mouse was in the interaction zones. In pretest, the object was aligned by longer interaction time (preferred object is plotted in right side). Median event muber: Pre (left vs right), 20 vs 63.5; Sociability (mouse vs object), 92.5 vs 77.5; Discrimination (familiar vs novel), 22.5 vs 63.5. p = 0.001, 0.385, and 0.006. Signed-rank test. N = 6. (D and E) Correlation plots of calcium event number ratios and preference index in pretest (D) and familiarization session (E). Pretest, p < 0.001, N = 8; Familiarization, p = 0.001, N = 9.

The effect of chemoegentic inhibition of ENvCA1-proj. neurons on the pretest and locomotion

(A) Frequency-current (FI) curve showing the effect of CNO on action potential number generated by current injection. Recording was made from hM4Di expressing ENvCA1-proj. neurons before and after bath application of CNO (100 nM). Black line indicates before and red line indicates after CNO application. Inset: Traces showing spike response to 100 pA (left) and 200 pA (right) current injection. Saline vs CNO; F (1, 4.77) = 0.38, p = 0.0329, two-way ANOVA. n = 4 neurons. N = 2 mice. (B) Same as in Fig. 8E, but for preference index in pretest. Median preference index: hM4Di mice (Saline vs CNO), 0.5 vs 0.5, p = 0.92; signed-rank test; tdTomato mice (Saline vs CNO), 0.48 vs 0.53, p = 0.227; signed-rank test. (C) Absolute change of preference index before and after CNO treatment in hM4Di and tdTomato mice. Median change of preference index: hM4Di mice 0.11 (N = 14); tdTomato 0.11 ( N = 8). hM4Di vs tdTomato, p = 0.517, ranksum test. (D) Same as in B, but for preference index in familiarization session of novel object recognition test. Median preference index: hM4Di mice (Saline vs CNO), 0.49 vs 0.5, p = 0.28; signed-rank test (E) Plots comparing locomotor activity (distance moved and velocity) for hM4Di and tdTomato mice under two different treatments. Median total distance: Pretest, hM4Di mice (Saline vs CNO), 17.17 vs 19.10, p = 0.426; tdTomato mice (Saline vs CNO), 20.60 vs 17.98, p = 0.25; Sociability test, hM4Di mice (Saline vs CNO), 34.72 vs 33.51, p = 0.855; tdTomato mice (Saline vs CNO), 38.59 vs 35.76, p = 0.547, Discrimination test, hM4Di mice (Saline vs CNO), 16.55 vs 18.58, p = 0.0676; tdTomato mice (Saline vs CNO), 18.73 vs 20.60, p = 0.945, all signed-rank test. Median velocity: Pretest, hM4Di mice (Saline vs CNO), 5.98 vs 6.55, p = 0.391; tdTomato mice (Saline vs CNO), 7.14 vs 6.21, p = 0.383; Sociability test, hM4Di mice (Saline vs CNO), 6.01 vs 5.76, p = 0.808; tdTomato mice (Saline vs CMO), 6.78 vs 6.13, p = 0.313; Discrimination test, hM4Di mice (Saline vs CNO), 5.88 vs 6.57, p = 0.104; tdTomato mice (Saline vs CNO), 6.79 vs 6.99, p = 0.945. All signed-rank test. (F) Total time in interaction zones during discrimination test (familiar mouse interaction zone + novel mouse interaction zone). Median total time: (Saline vs CNO), 248.32 vs 294.37; p = 0.67, signed-rank test.

The effect of chemoegentic inhibition of ENvCA1-proj. neurons on trace fear conditioning

(A) Schematic of trace fear conditioning protocol. (B) The effect of CNO on freezing behavior during conditioning. Black indicates saline treated group and red indicates CNO treated group. Data points of individual mice are indicated by empty circles. Thick horizontal bars indicate median values. Saline vs CNO, F(1, 9) = 2.52, p = 0.115; two-way ANOVA. (C) The effect of CNO on freezing behavior during recall to context “A”. Median values of freezing %: Saline = 14.91, CNO = 30.13. Saline vs CNO, p = 0.576; one-way ANOVA. (D) The effect of CNO on freezing behavior during conditional stimulus recall in context “B”. Saline vs CNO, F (1, 9) = 0.38, p = 0.54; two-way ANOVA