A stress-activated neuronal ensemble in the supramammillary nucleus produces anxiety-like behavior in male mice.

A stress-activated neuronal ensemble in the supramammillary nucleus produces anxiety-like behavior in male mice - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice eLife . 2026 Apr 15;14:RP108593. doi: 10.7554/eLife.108593 Search in PMC Search in PubMed View in NLM Catalog Add to search A stress-activated neuronal ensemble in the supramammillary nucleus produces anxiety-like behavior in male mice Jinming Zhang Jinming Zhang 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China Find articles by Jinming Zhang 1 , Kexin Yu Kexin Yu 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China Find articles by Kexin Yu 1 , Junmin Zhang Junmin Zhang 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China Find articles by Junmin Zhang 1 , Yuan Chang Yuan Chang 2 Department of Histology and Embryology, School of Basic Medical Science, Xi'an Medical University, Xi'an, China Find articles by Yuan Chang 2 , Xiao Sun Xiao Sun 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China Find articles by Xiao Sun 1 , Zhaoqiang Qian Zhaoqiang Qian 3 Laboratory Animal Center, Shaanxi Normal University, Xi’an, China Find articles by Zhaoqiang Qian 3 , Zongpeng Sun Zongpeng Sun 4 School of Psychology, Shaanxi Normal University, Xi'an, China Find articles by Zongpeng Sun 4 , Yanning Qiao Yanning Qiao 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China Find articles by Yanning Qiao 1 , Zhiqiang Liu Zhiqiang Liu 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China Find articles by Zhiqiang Liu 1 , Wei Ren Wei Ren 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China 5 Faculty of Education, Shaanxi Normal University, Xi’an, China Find articles by Wei Ren 1, 5 , Jing Han Jing Han 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China Find articles by Jing Han 1, ✉ Editors: Kate M Wassum 6 , Kate M Wassum 7 Author information Article notes Copyright and License information 1 Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, China 2 Department of Histology and Embryology, School of Basic Medical Science, Xi'an Medical University, Xi'an, China 3 Laboratory Animal Center, Shaanxi Normal University, Xi’an, China 4 School of Psychology, Shaanxi Normal University, Xi'an, China 5 Faculty of Education, Shaanxi Normal University, Xi’an, China 6 University of California, Los Angeles, United States 7 University of California, Los Angeles, United States ✉ Corresponding author. Roles Kate M Wassum : Reviewing Editor Kate M Wassum : Senior Editor Collection date 2026. © 2025, Zhang et al This article is distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use and redistribution provided that the original author and source are credited. PMC Copyright notice PMCID: PMC13082789  PMID: 41983536 Previous version available: This article is based on a previously available preprint with doi: https://doi.org/10.1101/2025.01.15.633288 . Previous version available: This article is based on a previously available preprint with doi: https://doi.org/10.7554/eLife.108593.1 . Previous version available: This article is based on a previously available preprint with doi: https://doi.org/10.7554/eLife.108593.2 . Abstract Anxiety is a prevalent negative emotional state induced by stress; however, the neural mechanism underlying anxiety is still largely unknown. We used acute and chronic stress to induce anxiety and test anxiety-like behavior; immunostaining, multichannel extracellular electrophysiological recording, and Ca 2+ imaging to evaluate neuronal activity; and virus-based neuronal tracing to label circuits and manipulate circuitry activity. Here, we identified a hypothalamic region, the supramammillary nucleus (SuM), that plays an important role in anxiety-like behavior. We then characterized a small ensemble of stress-activated neurons (SANs) that are recruited by stress. These SANs respond specifically to stress, and their activation robustly increases anxiety-like behavior in male mice. We also found that ventral subiculum (vSub)-SuM projections, but not dorsal subiculum (dSub)-SuM projections, encode anxiety-like behavior and that inhibition of these vSub-SuM projections has an antianxiety effect. These results indicate that the reactivation of stress-activated supramammillary cells and relevant neural circuits is an important neural process underlying anxiety-like behavior. Research organism: Mouse Introduction Anxiety is a fundamental negative emotion observed in almost all mammal species. Long-lasting and uncontrollable anxiety often leads to several mental disorders, anxiety disorders, and even depression ( Kesner et al., 2021 ). Recent studies have shown that the supramammillary nucleus (SuM), a part of the hypothalamus, regulates sleep ( Qin et al., 2022 ), memory ( Li et al., 2022b ; Li et al., 2022a ), novelty exploration ( Li et al., 2022b ; Pedersen et al., 2017 ), social memory ( Farrell et al., 2021 ; Pan and McNaughton, 2002 ), neurogenesis ( López-Ferreras et al., 2020 ), consciousness ( López-Ferreras et al., 2019 ; Liang et al., 2023 ), locomotor activity ( Tonegawa et al., 2015 ), and theta oscillations in the hippocampus ( Josselyn and Tonegawa, 2020 ). Projections from the SuM to the hippocampus have been largely studied and were found to modulate either episodic memory ( Li et al., 2022b ) or social memory ( Farrell et al., 2021 ; Pan and McNaughton, 2002 ) depending on the subregion of the hippocampus targeted ( Chen et al., 2020 ). Although the SuM is located near the mammillary nucleus, a key region implicated in emotion regulation via the Papez circuit, its role in regulating emotion has been explored only superficially, without in-depth investigation. Despite some discussions of this role of the SuM, no consistent conclusion has been reached thus far ( López-Ferreras et al., 2020 ; Liu et al., 2012 ; Guenthner et al., 2013 ). Activity-dependent activation of cells has been studied in many brain areas ( Sun et al., 2023 ; Azevedo et al., 2019 ). Tagged cells react to specific stimuli, such as conditional stimuli ( Koren et al., 2021 ; Ryan et al., 2015 ), pain ( Yan et al., 2022 ), food ( Jimenez et al., 2018 ), and even peripheral inflammation ( Forro et al., 2022 ), and mediate the storage and retrieval of relevant memories. The manipulation of those believed memory-associated cells can alleviate neurodegenerative diseases ( Strange et al., 2014 ) or inflammation ( Forro et al., 2022 ). Naturally, this has led us to consider whether there is a special neuronal ensemble that plays roles in regulating anxiety or anxiety-like behaviors. Recent studies have focused on the role of the hippocampus and related neuronal afferents and efferents ( Kesner et al., 2023 ; Cumbers et al., 2007 ; Silveira et al., 1993 ). The dorsal part of the hippocampus mainly contributes to cognition, whereas the ventral hippocampus is often associated with emotion ( Escobedo et al., 2023 ; LeDuke et al., 2023 ). Although the ventral hippocampus-hypothalamus circuit was reported to modulate anxiety ( Kesner et al., 2023 ; Cumbers et al., 2007 ), it is still unknown whether the SuM is part of this regulatory circuit. Although the SuM sends and receives dense neuronal projections, few studies have focused on its afferents or its ability to modulate behavior and emotion ( Aranda et al., 2006 ). In this study, we hypothesize that stress can recruit a special neuronal ensemble that exclusively encodes anxiety. To test this hypothesis, we first used multiple methods to assess whether the SuM responds to acute or chronic stress. The activity of the SuM was chemogenetically manipulated, and anxiety-like behavior in rodents was tested. We subsequently employed the targeted recombination in active populations (TRAP) strategy to label and manipulate stress-activated neurons (SANs) in SuM in terms of anxiety-like behavior. After demonstrating how the SuM modulates anxiety, we sought to identify upstream brain areas that may contribute to supramammillary function. We also examined the functions of neuronal projections from the ventral subiculum (vSub) to the SuM using fiber photometry to measure calcium dynamics, as well as chemogenetic manipulation. These results allowed us to characterize a previously unreported role of SuM in regulating anxiety-like behavior. In addition, we showed that projections from the vSub, but not the dorsal subiculum (dSub), to the SuM govern chronic stress-induced anxiety-like behaviors. Results Stress increases neuronal activity in the SuM c-Fos protein expression was assessed after acute stress exposure to test whether the SuM was activated ( Figure 1A ). The number of c-Fos + cells was significantly increased by foot shock exposure ( Figure 1B and C ). To investigate if SuM would be responsive to diverse stressors, we next examined whether chronic stress, which has different mechanisms underlying, affects neuronal activity in the SuM ( Figure 1D and E ). We chose in vivo electrophysiological extracellular recordings to reveal the neuronal activity before and after chronic social defeat stress. The data shows that the firing rate of regular-spiking neurons (RNs) ( Figure 1F ) but not fast-spiking neurons (FNs) ( Figure 1—figure supplement 1A and B ) increased after CSDS. Regarding local field potentials, there were no noticeable differences between the naïve and CSDS groups according to power spectrum analysis ( Figure 1—figure supplement 1C and D ). These results indicate that acute and chronic stress can strongly activate the SuM. Figure 1. Stress activates the supramammillary nucleus (SuM). ( A ) Workflow of the c-Fos staining. ( B ) Representative images of c-Fos staining (DAPI: blue; c-Fos: white; scale bar: 100 µm). ( C ) Statistical analysis of the number of c-Fos-positive cells displayed in panel B. n=10–13 per group; unpaired t test. ( D ) Workflow of CSDS exposure and in vivo recording. ( E ) Representative spikes acquired by multichannel recording. ( F ) Statistical analysis of the firing rates of regular-spiking neurons (RNs) at baseline and after CSDS exposure. n=16‒23 per group; two-way ANOVA followed by Sidak’s post hoc test. The bars in C and F indicate the Min to Max of all data points, and the ‘+’ indicates mean value of all data points. ‘**’, p<0.01; ‘***’, p<0.001. CSDS: chronic social stress; FS: foot shock. Figure 1—source data 1. The number of c-Fos+ cells in panel C and normalized firing rate in panel F, and corresponding statistical results. elife-108593-fig1-data1.xlsx (14.7KB, xlsx) Figure 1—figure supplement 1. The effect of CSDS on the local field potential (LFP) in the supramammillary nucleus (SuM). ( A ) Representative images of the location of electrodes. ( B ) Statistical comparison of the firing rate of fast-spiking neurons (FNs) between baseline and after CSDS. n=4–12 neurons from 2 to 4 mice per group, two-way ANOVA, Sidak’s post hoc test. ( C–D ) Power spectrum of LFPs. Data in B is presented as mean ± SEM. ‘ns’, p>0.05. CSDS: chronic social stress. Open in a new tab Activation of SuM produces anxiety-like behavior After confirming the activation of SuM caused different types of stress, we then further investigate whether the SuM regulates anxiety behavior in mice. Chemogenetic manipulations were conducted to activate SuM neurons. The experiments were performed as shown in the workflow ( Figure 2A and B ). The mice were subjected to the open field (OF) and elevated zero maze (EZM) tests at least 2 weeks after virus injection, followed by a reward-seeking test ( Figure 2E–H , Figure 2—figure supplement 1A ). Clozapine N-oxide (CNO) was administered intraperitoneally 30 min before the test. Chemogenetic activation of the SuM did not affect the performance of mice in the OF test ( Figure 2E and F , Figure 2—figure supplement 1B ). Compared with control mice, mice in which the SuM was activated explored the open arms of the EZM less ( Figure 2G ), despite no change in distance traveled ( Figure 2—figure supplement 1C ). Moreover, mice in which the SuM was activated consumed less food than control mice did ( Figure 2F ). These data suggest that there are neuronal ensembles that control the expression of anxiety behavior. Figure 2. Activation of the supramammillary nucleus (SuM) produces anxiety-like behavior. ( A–B ) Virus injection information ( A ) and workflow for chemogenetic manipulation ( B ). ( C ) Representative c-Fos images. ( D ) Statistical analysis of the number of c-Fos-positive cells displayed in panel C. n=4 per group; unpaired t test. ( E–F ) Statistical analysis of the distance traveled in the central area ( E ) and the time that the mice spent in the central area ( F ) in the OF test. n=10 per group; unpaired t test for the data in ( E ) and the Mann-Whitney test for the data in ( F ). ( G ) Statistical analysis of the time that the mice spent in the open arms of the elevated zero maze (EZM). n=8 per group; unpaired t test. ( H ) Statistical analysis of sucrose pellets consumed. n=8–10 per group; Mann-Whitney test. The bars in D–H indicate the Min to Max of all data points, and the ‘+’ indicates mean value of all data points. ‘ns’, p>0.05; ‘**’, p<0.01. Figure 2—source data 1. The number of c-Fos+ cells in panel D, and behavioral test data in panel E-H, and corresponding statistical results. elife-108593-fig2-data1.xlsx (13.1KB, xlsx) Figure 2—figure supplement 1. The chemogenetic manipulation of supramammillary nucleus (SuM) and stress-activated neurons (SANs) has effects on the performance of mice in the open field (OF) and the elevated zero maze (EZM). ( A ) Representative images of virus expression. ( B ) Statistical comparison of the total distance that wild-type (WT) mice traveled in OF. ( C ) Statistical comparison of the total distance that WT mice traveled in EZM. ( D ) Statistical comparison of the total distance that TRAP2 mice traveled in OF. ( E ) Statistical comparison of the total distance that TRAP2 mice traveled in EZM. Data in B–I are presented as mean ± SEM. ‘ns’, p>0.05, ‘**’, p<0.01, ‘***’, p<0.001. Open in a new tab Identification of SANs in the SuM We next investigated whether an ensemble that encodes stress and controls the expression of anxiety exists. By crossbreeding Fos 2A-iCreERT2(TRAP2) and Rosa26-LSL-tdTomato (Ai14) mice, we generated TRAP2;Ai14 mice in which activated cells were genetically tagged for visualization ( Figure 3A ). Foot shock exposure strongly activated neurons in the SuM but not adjacent areas ( Figure 3B and C ). We hypothesized that these SANs respond exclusively to stress, but not to other stimuli such as reward. To validate the activity-dependent labeling in TRAP2;Ai14 mice, we labeled SANs in two cohorts of mice exposed either to the home cage condition or to foot shock. Several days after labeling, the mice were exposed to either sucrose pellets or social stress to induce reward-related or stress-related c-Fos expression, respectively. The reactivation of SANs under reward stimulation and stress was then compared ( Figure 3D–H ). Foot shocks dramatically activated and labeled neurons in the SuM ( Figure 3F ). Social stress but not reward (presentation of sucrose pellets) induced neuronal activation ( Figure 3G ) and led to a much greater chance of reactivation of SANs ( Figure 3H ). These data suggest the specific regulatory effect of the SuM on the effects of stress but not reward. Figure 3. Stress-activated neurons (SANs) in the supramammillary nucleus (SuM) selectively respond to social stress but not reward. Open in a new tab ( A ) Workflow of neuronal tagging. ( B ) Representative image of SANs in the SuM (DAPI: blue, tagged cells: red). ( C ) Quantitative statistics of stress-tagged cells in several brain areas. n=3 per area; one-way ANOVA followed by Tukey’s post hoc test. ( D ) Workflow of neuronal tagging and c-Fos staining. ( E ) Representative images of stress-tagged cells and c-Fos expression induced by sucrose and social stress (DAPI: blue, tdTomato: red, c-Fos: green). ( F ) Statistical analysis of the number of stress-tagged cells in the SuM. n=3 per group; one-way ANOVA followed by Tukey’s post hoc test. ( G ) Statistical analysis of the number of c-Fos + cells after sucrose or social stress exposure. n=3 per group; one-way ANOVA followed by Tukey’s post hoc test. ( H ) Statistical analysis of the reactivation of stress-tagged cells. n=3 per group; one-way ANOVA followed by Tukey’s post hoc test. The data in C and F–H are presented as the means ± SEMs. ‘ns’, p>0.05; ‘*’, p<0.05; ‘**’, p<0.01; ‘***’, p<0.001. Figure 3—source data 1. The number of TRAPed cells in panel C and panel F, c-Fos+ cells in panel G and the reactivation ratio in panel H, and corresponding statistical results. elife-108593-fig3-data1.xlsx (11.7KB, xlsx) Reactivation of SuM SANs promotes anxiety-like behavior To make sure mice are on similar basal conditions while applying chemogenetic manipulation, we subjected mice to an acute stress protocol involving foot shocks and then performed the elevated plus maze (EPM) and EZM tests to evaluate anxiety on days 2 and 7 ( Figure 4A ). The mice that experienced foot shocks showed decreases in the exploration time in the open arms on day 2. However, acute stress-induced anxiety was not detected on day 7 ( Figure 4B ), which allowed us to compare the reactivation of SANs produced anxiety-like behavior between groups at the same baseline. Seven days after SANs tagging, specific activation of SANs significantly increased the concentration of corticosterone (a peripheral indicator of stress) in the mouse serum ( Figure 4C and D ). Experiments involving chemogenetic manipulation also revealed the sound-selective activation of SANs in the SuM ( Figure 4E and F ). We then tested whether manipulating SANs in the SuM influences the anxiety-like behavior of the mice. The mice were subjected to the OF and EZM tests at least 1 week after SANs were tagged, followed by reward-seeking tests ( Figure 4G and H ). CNO was administered intraperitoneally 30 min before the test. Chemogenetic activation of SANs decreased the total distance traveled by the mice in the OF and EZM tests ( Figure 2—figure supplement 1D and E ). The mice also presented decreases in the distance traveled in the central area ( Figure 4I ) and time spent in the central area in the OF test ( Figure 4J ), time spent in the open arms in the EZM test ( Figure 4K ) and food consumption ( Figure 4L ). These data suggest that SANs in the SuM encode anxiety-like behavior. Figure 4. Selective chemogenetic activation of stress-activated neurons (SANs) elevates corticosterone level and produces anxiety-like behavior. Open in a new tab ( A ) Workflow of the acute stress and anxiety tests. ( B ) Statistical analysis of the time that the mice spent in the open arms of the elevated zero maze (EZM). n=10‒11 per group; two-way ANOVA followed by Sidak’s post hoc test. ( C ) Workflow of the CORT assay and c-Fos staining. ( D ) Statistical analysis of the serum concentration of corticosterone after the application of clozapine N-oxide (CNO). n=4–5 per group; unpaired t test. ( E ) Representative images of stress-tagged cells and c-Fos expression induced by chemogenetic manipulation (DAPI: blue, EGFP: green, c-Fos: violet). ( F ) Statistical analysis of the percentage of costained cells relative to EGPF + cells in the supramammillary nucleus (SuM). n=4–5 per group; unpaired t test. ( G–H ) Virus injection information and workflow of chemogenetic manipulation. ( I–J ) Statistical analysis of the distance traveled in the central area ( I ) and the time that the mice spent in the central area ( J ) in the open field (OF) test. n=14–15 per group; unpaired t test. ( K ) Statistical analysis of the time that the mice spent in the open arms of the EZM. n=14–15 per group; Mann-Whitney test. ( L ) Statistical analysis of sucrose pellets consumed. n=13–15 per group; unpaired t test. The data in D and F are presented as the means ± SEMs. The bars in B and I–L indicate the Min to Max of all data points, and the ‘+’ indicates mean value of all data points. ‘*’, p<0.05; ‘**’, p<0.01; ‘***’, p<0.001. Figure 4—source data 1. The behavioral test data in panel B and I-L, corticosterone level in panel D, co-stain ratio in panel F and corresponding statistical results. elife-108593-fig4-data1.xlsx (13.1KB, xlsx) vSub-SuM projections encode anxiety-like behavior The SuM receives afferents from various brain areas, and we identified projections to the SuM by using a nonvirus- and virus-based retrograde tracing strategy ( Figure 5A , Figure 5—figure supplement 1A–C ). Afferents from the dSub and vSub were identified using CTB-647 and adeno-associated virus (AAV) ( Figure 5—figure supplement 2A and B ). These projection neurons expressed Vglut1 RNA but not Vgat RNA ( Figure 5B ), suggesting that Sub-SuM projections are excitatory neuronal projections, as Vglut1 is a crucial marker of glutamatergic neurons. We then performed an electrophysiological experiment. Optogenetics-evoked postsynaptic currents (PSCs) in SuM neurons were blocked by perfusion with DNQX, which indicates the existence of glutamatergic projections from the Sub to the SuM ( Figure 5C–E ). To investigate how Sub-SuM projections modulate stress and anxiety-like behavior, we then used fiber photometry to measure the calcium concentration to assess the activity patterns of the projection neurons ( Figure 5F–H ). The projection neurons in the vSub, but not those in the dSub, were more strongly activated when the mice moved into the open arms from the closed arms of the EZM ( Figure 5I–M ). On the other hand, vSub, but not dSub, projection neurons show lower calcium activity while mice back to the closed arms ( Figure 5—figure supplement 3A–D ). Following exposure to acute stress, both dSub-SuM and vSub-SuM projection neurons presented increased calcium activity ( Figure 5N–R ). These data suggest that vSub-SuM projections, but not dSub-SuM projections, may participate in regulating anxiety-like behavior. Figure 5. Ventral subiculum (vSub)-supramammillary nucleus (SuM) projections encoding anxiety-like behavior. ( A ) Workflow of virus-based retrograde neuronal tracing. ( B ) Representative images of in situ RNA staining (DAPI: blue, Slc32a1: green, Slc17a7: red, EGFP: white). ( C ) Workflow of ex vivo electrophysiological recording. ( D ) Schematic of optically induced postsynaptic currents (oPSCs) in the SuM. ( E ) Representative traces of oPSCs. ( F ) Workflow of Ca 2+ imaging. ( G ) Schematic of Ca 2+ imaging of dorsal subiculum (dSub) and vSub projection neurons. ( H ) Representative images of GCaMP7b expression in the dSub, vSub, and SuM (DAPI: blue, GCaMP7b: green). ( I ) Heatmap of the Ca 2+ fluorescence intensity during the transition from the closed to the open arms. ( J ) Representative Ca 2+ activity during the transition from the closed arms to the open arms. ( K ) Average ∆F/F of Ca 2+ recorded in the dSub and vSub. ( L ) Statistical analysis of the peak Ca 2+ activity. n=5 per group; unpaired t test. ( M ) Statistical analysis of the area under the curve of Ca 2+ activity. n=5 per group; unpaired t test. ( N ) Heatmap of the Ca 2+ fluorescence intensity during exposure to foot shocks. ( O ) Representative Ca 2+ activity during exposure to foot shocks. ( P ) Average ∆F/F of Ca 2+ recorded in the dSub and vSub. ( Q ) Statistical analysis of the peak Ca 2+ activity. n=5‒6 per group; unpaired t test. ( R ) Statistical analysis of the area under the curve of Ca 2+ activity. n=5‒6 per group; unpaired t test. The bars in L–M and Q–R indicate the Min to Max of all data points, and the ‘+’ indicates mean value of all data points. ‘ns’, p>0.05; ‘*’, p<0.05. Figure 5—source data 1. The Ca 2+ peak of ΔF/F in panel L and Q, the area under curve of the Ca 2+ trace in panel M and R, and corresponding statistical results. elife-108593-fig5-data1.xlsx (12.3KB, xlsx) Figure 5—figure supplement 1. Specific retrograde neuronal tracing of the upstream of the stress-activated neuron (SAN) in the supramammillary nucleus (SuM). ( A ) Workflow of rabies virus (RV)-based retrograde neuronal tracing. ( B ) Representative image of virus expression in SuM. ( C ) Representative images of traced upstreaming brain area of SuM. Figure 5—figure supplement 2. Non-virus and virus-based retrograde neuronal tracing of the upstream of the supramammillary nucleus (SuM). ( A ) Representative images of retrograde neuronal tracing using CTB-647, injection site, dorsal subiculum (dSub), and ventral subiculum (vSub). ( B ) Representative images of retrograde neuronal tracing using AAV2/Retro, injection site, dSub, and vSub. Figure 5—figure supplement 3. Calcium fiber photometry during elevated plus maze (EPM) test. ( A ) Heatmap of the Ca 2+ fluorescence intensity during the transition from the open arms to the closed arms. ( B ) Representative Ca 2+ activity during the transition from the open to the closed arms. ( C ) Average ∆F/F of Ca 2+ recorded in the dorsal subiculum (dSub) and ventral subiculum (vSub). ( D ) Statistical analysis of the peak Ca 2+ activity. n=5 per group; unpaired t test. Open in a new tab Chronic inhibition of vSub-SuM projections alleviates anxiety-like behavior After confirming the regulatory role of vSub-SuM projections in anxiety, we hypothesized that inhibition of this projection would alleviate chronic stress-induced anxiety. In the following experiments, the activity of these projections was chronically inhibited via a chemogenetic strategy. The mice were exposed to CSDS after the expression of the Gi protein was induced specifically on vSub-SuM projection neurons and their axons ( Figure 6A, B, and D ). The body weights of the mice were monitored throughout the entire procedure to assess their health ( Figure 6C ). The mice showed no change in social interaction test scores after CSDS exposure ( Figure 6E ). In the EZM test, the mice in which vSub-SuM projections were inhibited presented less anxiety-like behavior, as indicated by a longer time spent in the open arms of the EZM but no significant change in the distance traveled ( Figure 6F and G ). Taken together, these data suggest that vSub-SuM projections are the essential neuronal projections for regulating chronic stress-induced anxiety-like behavior. Figure 6. Selective inhibition of ventral subiculum (vSub)-supramammillary nucleus (SuM) projections alleviated CSDS-induced anxiety. Open in a new tab ( A ) Workflow of CSDS and chemogenetic manipulation. ( B ) Schematic of chemogenetic manipulation of specific projections. ( C ) Body weight during CSDS exposure. ( D ) Representative images of virus expression. ( E ) Statistical analysis of the social interaction ratio after CSDS exposure. n=6–10 per group; two-way ANOVA followed by Sidak’s post hoc test. ( F ) Statistical analysis of the distance that the mice traveled in the elevated zero maze (EZM). n=6–10 per group; two-way ANOVA followed by Sidak’s post hoc test. ( G ) Statistical analysis of the time that the mice spent in the open arms of the EZM; two-way ANOVA followed by Sidak post hoc test. The bars in E–G indicate the Min to Max of all data points, and the ‘+’ indicates mean value of all data points. ‘ns’, p>0.05; ‘*’, p<0.05; ‘**’, p<0.01; ‘***’, p<0.001. CSDS: chronic social stress. Figure 6—source data 1. The body weight in panel C, and behavioral test data in panel E-G, and corresponding statistical results. elife-108593-fig6-data1.xlsx (15.9KB, xlsx) Discussion In this study, we combined multiple methods to determine whether SuM is a brain region that is involved in modulating anxiety. SuM neurons strongly respond to acute and chronic stress, and their activation results in robust increases in anxiety-like behavior in mice. We then defined a small ensemble of neurons that are activated by stress, called SANs. These SANs specifically respond to stressful stimuli but not reward. Selective activation of SANs in the SuM increases the serum concentration of corticosterone and anxiety-like behavior in mice. The neuronal circuits that may underlie the regulation of anxiety were also determined in this study. The subiculum sends glutamatergic projections to the SuM and can be activated by stress, whereas only the vSub has a potential effect on the transition to anxiety. We finally determined that inhibition of SANs in the vSub project to the SuM is sufficient to alleviate anxiety in mice after CSDS exposure. Regulation of anxiety avoidance by the SuM The SuM has been demonstrated to respond to novel environments, social stimulation ( Li et al., 2022b ; Beck and Fibiger, 1995 ), and stress ( Tonegawa et al., 2018 ; Sun et al., 2020 ). Given these findings, we assume that the SuM may be activated by foot shocks, a quantifiable acute stressor used in animal studies. Consistent with this hypothesis, we found that the SuM robustly positively responds to both acute and chronic stress through observable increases in c-Fos expression and increases in the neuronal firing rate. The activation of the SuM was also demonstrated to be essential for maintaining arousal ( Qin et al., 2022 ; López-Ferreras et al., 2019 ). Sensitization to stressful events and high arousal are often associated with anxiety ( Lacagnina et al., 2019 ). Thus, our data strongly suggests that the SuM potentially modulates anxiety. To further confirm whether the SuM participates in anxiety regulation, we recorded neuronal action potentials via multichannel extracellular recording while the mice were moving in the EPM, a traditional type of maze used to test anxiety in rodents. The change in the neuronal firing frequency when mice transitioned from the open arms to the closed arms supports the idea that the SuM may somewhat modulate anxiety. We then manipulated neuronal activity in the SuM via a chemogenetic method and subjected the mice to the EZM test, an improved test for assessing anxiety in rodents. The decrease in exploration time in the open arms by mice in which the SuM was activated by hM3Dq indicated increased anxiety. We noted that these results are inconsistent with those of some previous reports. Some studies have reported that lesions in the SuM and adjacent areas decrease anxiogenic behavior in rats ( Josselyn and Tonegawa, 2020 ; Zheng et al., 2024 ; Zhang et al., 2019 ). López-Ferreras et al. performed the OF test, a complicated test, and reported that the chemogenetic activation of neurons in the SuM results in increased anxiety-like behaviors in rats ( Liu et al., 2012 ; Guenthner et al., 2013 ). However, further experiments involving specific tests (e.g. the EZM test) are needed to confirm whether there is a potential difference across species. The role of SANs in regulating anxiety avoidance Recent studies have highlighted the importance of activity-tagged neuronal ensembles in regulating various behaviors, particularly memory ( Koren et al., 2021 ; Báez et al., 1996 ; Chen et al., 2024 ; Handa et al., 1994 ), food consumption ( Yan et al., 2022 ), the inflammatory response ( Forro et al., 2022 ), and emotion ( Fanselow and Dong, 2010 ; Dos Santos Corrêa et al., 2019 ). A negative experience-related neuronal ensemble in the hippocampus was found to increase susceptibility to chronic stress ( Fanselow and Dong, 2010 ). A recent study reported that the lateral habenula contains a small population of neurons that are recruited in response to stress and mediate the development of depression in mice ( Dos Santos Corrêa et al., 2019 ). These studies suggest that SANs may be important for emotional regulation. In this study, we found that the SuM was more strongly activated by acute stress than were adjacent areas. SANs were more likely to be reactivated by social stress than by sucrose reward, indicating their potential to specifically encode anxiety. The serum corticosterone concentration can be used as a marker of stress-induced change in the peripheral blood. Previous studies showed serum corticosterone can be increased by various stress stimulation ( Báez et al., 1996 ; Chen et al., 2024 ; Handa et al., 1994 ; Dos Santos Corrêa et al., 2019 ); meanwhile, intentionally supplementing the diet with corticosterone can induce anxiety-like behaviors in rodents ( Peng et al., 2021 ). Our data showed that the chemogenetic activation of SANs in the SuM increased the serum corticosterone concentration, whereas the inactivation of SANs had no effect, suggesting the chemogenetic manipulation of SANs may cause similar anxiety effects like real stressors. These findings, in combination with the results of the OF and EZM, suggest that SANs in the SuM are more likely involved in modulating anxiety-like avoidance. However, the reactivation rate of SANs caused by different stressors was relatively lower than the initial activation rate caused by foot shock ( Figure 3 ). This suggests that stress-activated neuronal clusters may have more flexible recruitment principles, with only a small number of neurons potentially encoding emotional information, while most other neurons remain involved in encoding other neural activities. Studies in other fields, particularly studies of memory engram, have shown that the sets of neurons activated during learning are dynamic and exhibit high flexibility ( Zaki and Cai, 2024 ; Sweis et al., 2021 ). While the activation of SANs produced anxiety-like behavior, the future study will examine whether silencing SuM SANs, either during stress exposure or during anxiety testing, can prevent or reduce stress-induced anxiety. We also found that both the nonselective activation of SuM neurons and the selective activation of SANs in the SuM significantly suppressed the consumption of sucrose pellets. This result may be attributed to the anxiety-induced suppression of reward seeking ( Peng et al., 2021 ). However, further experiments are still needed to confirm whether this effect is anxiety dependent and whether basal food consumption is affected. A relevant neural circuit that regulates anxiety avoidance The SuM recruits and is targeted by neuronal projections in the hippocampus, medial septum, and cortex ( Aranda et al., 2006 ). To further understand the circuitry through which the SuM regulates anxiety, we identified projections from the dSub and the vSub to the SuM ( Tang et al., 2016 ). Fiber photometry was used to measure the Ca 2+ concentration in projection neurons in the dSub and vSub, and the results revealed increased Ca 2+ activity in vSub-SuM projection neurons but not dSub-SuM projection neurons when the mice transited from the closed arms to the open arms, indicating that vSub-SuM projections encode anxiety. To confirm the regulation of anxiety by vSub-SuM projections, we exposed mice to CSDS and found that constant inhibition of vSub-SuM activity significantly abolished CSDS-induced anxiety in mice. Unlike the dorsal hippocampus, which is involved in the regulation of cognition, the ventral hippocampus is often involved in regulating emotion ( Shi et al., 2023 ). The ventral CA1 area and its projections to the lateral hypothalamic area were found to mediate innate anxiety, and its activation increases anxiety-like behavior in mice ( Cumbers et al., 2007 ). Although very close spatially, neurons in the subiculum are somewhat different from those in the CA1 region ( Aggleton and Christiansen, 2015 ; Ding et al., 2020 ). The vSub and its downstream brain areas were found to regulate anxiety ( Mueller et al., 2004 ; Ghasemi et al., 2022 ). Jing-Jing et al. reported that the vSub and its projections to the anterior hypothalamic nucleus are essential for anxiety because the inhibition of these projections decreases anxiety-like behavior ( Kesner et al., 2023 ). Our data is consistent with these findings and suggests the mediating role of the vSub and its projections to different subareas in the hypothalamus. In summary, the activation of SuM increases anxiety-like behavior. A stressful event recruits a neuronal ensemble in SuM. The activation of SANs also significantly increases anxiety-like behavior and suppresses reward seeking. SuM receives glutamatergic projections from the vSub, and inhibition of these projections can diminish CSDS-induced anxiety-like behavior. These results suggest that SuM plays an important role in regulating anxiety-like behavior, and furthermore, studies are worth performing. Materials and methods Key resources table. Reagent type (species) or resource Designation Source or reference Identifiers Additional information Strain, strain background ( ‌Mus musculus ) C57BL/6J Key Laboratory of Modern Teaching Technology, Ministry of Education RRID: IMSR_JAX:000664 Strain, strain background ( ‌Mus musculus ) Fos 2A-iCreERT2 JAX 030323, RRID: IMSR_JAX:030323 Strain, strain background ( ‌Mus musculus ) Rosa26-CAG-LSL-tdTomato (Ai14) Shanghai Model Organisms Center NM-KI-225042 Strain, strain background ( ‌Mus musculus ) CD-1 Charles River #201 Recombinant DNA reagent AAV2/9-hSyn-hM3Dq-EGFP BrainVTA PT-0891 Recombinant DNA reagent AAV2/9-hSyn-EGFP BrainVTA PT-1990 Recombinant DNA reagent AAV2/9-hSyn-DIO-hM3Dq -EGFP BrainVTA PT-0891 Recombinant DNA reagent AAV2/9-hSyn-DIO-EGFP BrainVTA PT-1103 Recombinant DNA reagent AAV2/Retro-hSyn-Cre Taitool S0278 Recombinant DNA reagent AAV2/9-hSyn-DIO-hM4Di-mCherry Taitool S0193 Recombinant DNA reagent AAV2/9-hSyn-DIO-mCherry Braincase BC-0025 Recombinant DNA reagent AAV2/9-hSyn-DIO-GCaMP7b BrainVTA PT-2892 Recombinant DNA reagent AAV2/9-hSyn-ChR2-mCherry BrainVTA PT-0150 Recombinant DNA reagent AAV2/Retro-hSyn-EGFP Taitool S0237 Recombinant DNA reagent AAV2/9-Ef1α-DIO-RVG BrainVTA PT-0061 Recombinant DNA reagent AAV2/9-Ef1α-DIO-mCherry-F2A-TVA BrainVTA PT-0207 Recombinant DNA reagent RV-ENVA-ΔG-EGFP BrainVTA R01001 Antibody Anti-c-Fos (Rabbit Monoclonal) Cell Signaling Technology Cat#2250, RRID: AB_2247211 1:500 Antibody Donkey anti-rabbit conjugated to AF647 Jackson ImmunoResearch Cat#706-605-148, RRID: AB_2340476 1:500 Commercial assay or kit RNAscope ACDbio Cat#323100 Chemical compound, drug 4-Hydroxytamoxifen Sigma-Aldrich Cat#H6278 Chemical compound, drug 4-Hydroxytamoxifen Bidepharm Cat# BD00958757 Chemical compound, drug Clozapine N-oxide Cayman Cat#25780 Software, algorithm SpikeInterface SpikeInterface RRID: SCR_021150 Other Antifade reagent Invitrogen P36981 Open in a new tab Animals Male C57BL/6J mice aged 12–20 weeks were used. Fos 2A-iCreERT2 (TRAP2) mice were a gift from Wenting Wang (JAX, Cat. No. 030323). Rosa26-CAG-LSL-tdTomato (Ai14) mice were purchased from the Shanghai Model Organisms Center (Cat. No. NM-KI-225042). Male CD-1 mice aged 8–10 months were purchased from Charles River (Cat. No. 201). To construct TRAP2;Ai14 mice, homozygous male TRAP2 mice and homozygous female Ai14 mice were bred. Homozygous TRAP2;Ai14 mice were maintained and used in the experiments. We performed genotyping for TRAP2 and Ai14 via PCR with the following primers: TRAP2 (wild-type: 357 bp, mutant: 232 bp): wild-type forward: GTCCGGTTCCTTCTATGCAG , mutant forward: CCTTGCAAAAGTATTACATCACG , common: GAACCTTCGAGGGAAGACG ; Ai14 (wild-type: 297 bp, mutant: 196 bp): wild-type forward: AAGGGAGCTGCAGTGGAGTA , wild-type reverse: CCGAAAATCTGTGGGAAGTC , mutant forward: GGCATTAAAGCAGCGTATCC , mutant reverse: CTGTTCCTGTACGGCATGG . The mice were housed 4–5 per cage at a constant temperature and humidity (22 ± 1°C, 30–40% RH) on a day-night cycle (lights on from 08:00-20:00) with a fixed-intensity light source. Each mouse was acclimated to the testing environment for 1–2 min. Acclimation was performed for 3 days before the behavioral experiments. The mice were fed ad libitum and euthanized with CO 2 after all the tests were finished. Sample size was determined according to our previous experience. All experiments and analysis described in this study were conducted double-blindly. All laboratory procedures were conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals in China and the regulations of the Animal Care and Use Committee of Shaanxi Normal University. This study protocol was reviewed and approved by the Academic Committee of the Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University (Approval No. L20230102-01). Behavioral procedure Acute stress exposure The mice were exposed to acute stress according to a previously reported procedure ( Marcus et al., 2020 ); specifically, they were exposed to 20 foot shocks with an intensity of 0.5 mA that were randomly delivered across 10 min. The foot shocks were delivered in a fear conditioning box (Med Associates). Chronic stress exposure We used a CSDS protocol to induce anxiety and depression in the mice ( Kim et al., 2017 ). Each C57BL6/J mouse was housed with one CD-1 mouse, and the mice were separated by a transparent plexiglass board with several small holes. The mice were allowed to contact each other directly for 10 min every day for 10 days. Body weight was measured and recorded every day before contact. OF test The OF test was carried out in a 50×50×35 cm 3 arena made of white plexiglass. The mice were allowed to move freely in the arena for 10 min, and the distance the mice traveled and the time the mice spent in the central area were recorded and analyzed. EPM test The EPM consisted of two open arms (30×7 cm 2 ), two closed arms (30×7×14 cm 3 ), and a central area (7×7 cm 2 ). The mice were allowed to move freely in the arena for 10 min, and the time the mice spent in the open arms was recorded and analyzed. EZM test The EZM was used to test whether the mice were anxious. The EZM used in this study was made of organic glass (height of 60 cm), with an inner diameter of 51.8 cm and an outer diameter of 65 cm. The closed arms of the EZM were separated by two 15-cm-high pieces of organic glass, the outer one of which was opaque. After a 15 min habituation period, the mice were placed into the EZM and allowed to move freely for 10 min. Videos were recorded and analyzed using EthoVisionXT software. The time that the mice spent in the open arms was compared between the groups to evaluate anxiety-like behavior. Reward seeking On the first day, sucrose pellets were provided for habituation. The mice were then deprived of food on the second day. On the third day, the mice were placed in a new home cage without bedding and allowed to eat freely for 2 hr. The pellets were weighed to evaluate whether the experimental manipulation influenced reward seeking by the mice. Social interaction test The SIT was conducted as previously described ( Kim et al., 2017 ). The SIT involved two 2.5 min phases. In the first phase (no-target phase), we placed each C57BL6/J mouse in the periphery of the arena opposite the social interaction area (SIA). We allowed the animal to explore the arena freely. In the second phase (with-target phase), each C57BL6/J mouse was placed in the arena again, with a new CD-1 mouse in the SIA. The social interaction ratio (SIR) was calculated using the following formula: S o c i a l i n t e r a c t i o n r a t i o ( S I R ) = T i m e i n S I A W i t h − t a r g e t − T i m e i n S I A N o - t a r g e t T i m e i n S I A W i t h − t a r g e t + T i m e i n S I A N o − t a r g e t Neuronal tagging of SANs To specifically label SANs, TRAP2 or TRAP2;Ai14 mice were intraperitoneally (i.p.) injected with 4-hydroxytamoxifen (4-OHT, 50 mg/kg) immediately after acute stress exposure or the learning phase of the CFC test. The mice were subjected to the next experiment or test after 7 days to allow Cre-dependent recombination. 4-OHT (CAS No. 68392-35-8. Sigma, Cat. No. H6278 or Bidepharm, Cat. No. BD00958757) was dissolved in DMSO at a concentration of 62.5 mg/mL and diluted with vehicle (containing 10% Tween 80 and 80% saline) on the day of neuronal tagging. The final concentration of DMSO was kept below 10% to avoid toxicity. Observation of the reactivation of SANs One week after neuronal tagging, whether previously tagged SANs were reactivated in TRAP2;Ai14 mice when they were subjected to social stress was assessed. The mice in the first group underwent neuronal tagging in their home cages, and c-Fos expression was induced by sucrose pellets. The mice in the second group were subjected to neuronal tagging in response to foot shock exposure, and c-Fos expression was induced by sucrose pellets. The mice in the third group were subjected to neuronal tagging in response to foot shock exposure, and c-Fos expression was induced by social stress (one CD-1 mouse was placed in the home cage). The mice were then sacrificed 90 min after sucrose pellet feeding or social stress exposure, and c-Fos immunofluorescence staining was performed. The number of c-Fos-positive neurons was counted to determine whether reward and cross-strain social stress could activate neurons in the SuM. Viral vectors An AAV vector was used to label and manipulate specific neurons or determine the calcium concentration. To manipulate the neuronal activity in the SuM, AAV2/9-hSyn-hM3Dq-EGFP (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-0891) or its control vector AAV2/9-hSyn-EGFP (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-1990) was injected into the SuM of mice. To manipulate the activity of SANs in the SuM, AAV2/9-hSyn-DIO-hM3Dq-EGFP (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-0891) or its control vector AAV2/9-hSyn-DIO-EGFP (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-1103) was injected into the SuM of TRAP2 mice. To chronically inhibit vSub-SuM circuitry activity, AAV2/Retro-hSyn-Cre (titer: 2.00E+12 GC/mL, Taitool, Cat. No. S0278) was injected into the SuM, and AAV2/9-hSyn-DIO-hM4Di-mCherry (titer: 2.00E+12 GC/mL, Taitool, Cat. No. S0193) or its control vector AAV2/9-hSyn-DIO-mCherry (titer: 2.00E+12 GC/mL, Braincase, Cat. No. BC-0025) was injected into the vSub of wild-type mice. To determine the calcium concentration in dSub/vSub-SuM projection neurons, AAV2/Retro-hSyn-Cre (titer: 2.00E+12 GC/mL, Taitool, Cat. No. S0278) was injected into the SuM, and AAV2/9-hSyn-DIO-GCaMP7b (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-2892) was injected into the dSub/vSub of wild-type mice. For the ex vivo electrophysiological experiment, AAV2/9-hSyn-ChR2-mCherry (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-0150) was injected into the vSub of wild-type mice. Neuronal tracing Retrograde neuronal tracing was initially performed via injection of a serotype-2 AAV vector (AAV2/Retro-hSyn-EGFP, titer: 5.00E+12 GC/mL, Taitool, Cat. No. S0237) and CTB-647 (1 µg/µL, Thermo Fisher, Cat. No. C34778 ) into the SuM of wild-type mice. The mice were then sacrificed after 2 weeks, and the brains were cut into coronal slices for imaging. To precisely trace neuronal afferents projecting to SuM SANs , AAV2/9-Ef1α-DIO-RVG (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-0061) and AAV2/9-Ef1α-DIO-mCherry-F2A-TVA (titer: 5.00E+12 GC/mL, BrainVTA, Cat. No. PT-0207) were injected into the SuM of TRAP2 mice simultaneously. The rabies virus (RV) vector RV-ENVA-ΔG-EGFP (titer: 2.00E+08 IFU/mL, BrainVTA, Cat. No. R01001 ) was injected into the SuM 2 weeks after neuronal tagging. The mice were then sacrificed after 2 weeks, and the brains were cut into coronal slices for imaging. Stereotaxic surgery The mice were anesthetized using isoflurane at a concentration of 1.5–2.0%. A virus was injected into the SuM (AP: –2.8, ML: 0, DV: –4.5 mm), dSub (AP: –2.8, ML: ±0.7, DV: –1.7 mm) or vSub (AP: –3.5, ML: ±3.0, DV: –4.6 mm) according to the experimental design. If only one type of virus needed to be injected into a single brain area, the final volume was typically 150 nL. Otherwise, the final volume of the virus mixture was 200 nL. The viruses were injected at a rate of 50 nL/min. The syringe was held in place for at least 5 min and carefully removed from the brain. The mice were then returned to their home cages, and their health was monitored on the following days. All the mice that underwent surgery were subjected to the subsequent experiment after 2 weeks or more to allow virus expression. For fiber photometry, ceramic ferrules (outer diameter: 2.5 mm, core diameter: 0.2 mm, NA: 0.50) were inserted into the dSub (AP: –2.8, ML: ±0.7, DV: –1.5 mm) or vSub (AP: –3.5, ML: ±3.0, DV: –4.4 mm) 2 weeks after virus injection under the guidance of a laser (wavelength: 470 nm). Calcium imaging was conducted at least 1 week after ferrule implementation. Fiber photometry Commercially available equipment (Thinker Tech) was used to determine the calcium concentration. The fluorescence signal was activated by a laser at 470 nm, and the signal was transmitted through a low-autofluorescence fiber-optic patch cord and rotary (doric lenses) and collected. The final activation intensity was set to ~40 µW. The sampling rate was 50 Hz for all the recordings. The mice were habituated to the fiber-optic patch cord for 3 consecutive days before recording. A TTL lasting 0.1 s was delivered by the software to mark the timepoint when the mouse moved from a closed arm to an open arm in the EPM (USB-IO box, Noldus). Continuous data were stored as *.tdms files and analyzed using custom-made software in MATLAB. Chemogenetic manipulation To manipulate neuronal activity in the SuM in wild-type and TRAP2 mice, CNO (5 mg/kg; Cayman, Cat. No. 25780) was injected i.p. 30 min before behavioral tests were performed. For chronic inhibition of circuit activity, CNO was administered orally (25 mg/L). For acute and chronic experiments, CNO was dissolved in DMSO at a concentration of 10 mg/mL and stored at –20°C or in saline at a concentration of 1 mg/mL and stored at –80°C. The storage solution was diluted with saline to a concentration of 0.75 mg/mL to prepare a working solution for acute manipulation or to a concentration of 25 mg/L to prepare a working solution for chronic inhibition on the day of the experiment. Immunofluorescence The mice were anesthetized with 20% urethane and perfused with PBS or saline. The mouse brain was dissected and immersed in 4% paraformaldehyde (PFA) at 4°C overnight. The PFA solution was then replaced with a 30% sucrose solution. After the brain sank to the bottom, it was embedded in optimal cutting temperature (OCT) compound and frozen in a cryostat (CM1950, Leica). Coronal slices (40 µm) were cut and collected in a 24-well plate. After the residual OCT was removed with PBS, the slices were blocked with 0.3% Triton X-100 and 10% normal donkey serum at room temperature (RT) for 2 hr. The slices were then incubated with diluted primary antibody (rabbit anti-c-Fos, 1:500, Cell Signaling Technology, Cat. No. 2250, RRID: AB_2247211 ) at 4°C overnight. The next day, the slices were washed and incubated with secondary antibody dilutions (donkey anti-rabbit conjugated to AF647, 1:500, Jackson ImmunoResearch, Cat. No. 706-605-148, RRID: AB_2340476 ) at RT for 2 hr. After washing, the slices were transferred to slides and mounted with an antifade reagent (Thermo Fisher, Cat. No. P36981 ). Images of the slices were collected using a Zeiss M2 microscope and then analyzed. RNA fluorescence in situ hybridization The samples were processed as described in the Immunofluorescence section. Slices (10 µm thick) were cut and dried at RT for ~15 min and then heated at 37°C for 30 min in a hybridization oven. The baked slides were then moved to precooled 4% PFA solution for fixation (~15 min). The slices were dehydrated in 100% ethanol at RT for 5 min. The dehydration step was then repeated. The following steps were performed as recommended by the manufacturer (ACDbio, Cat. No. 323100). To label vglut1, vglut2, and vgat RNA, the slices were hybridized with Mm- Slc17a7 (ACDbio, Cat. No. 416631-C1), Mm- Slc17a6 (ACDbio, Cat. No. 319171-C1) and Mm- Slc32a1 (ACDbio, Cat. No. 319191-C3), respectively. The samples were then stained with Opal dye. Costaining of protein and RNA After confirming RNA staining, the slices were blocked in 10% normal goat serum for 1 hr. The blocking solution was removed, and the slices were incubated with diluted primary antibody (mouse anti-GFP, 1:500, Thermo Fisher, Cat. No. MA5-16256; rabbit anti-tdTomato, 1:500, Oasis BioFarm, Cat. No. OB-PRB013) at 4°C overnight. The slides were washed with PBS and incubated with secondary antibody solution (goat anti-rabbit/mouse conjugated to HRP, Proteintech, Cat. No. PR30009) at RT for 1 hr (in the dark). After washing, the slices were stained with Opal dye at RT for 30 min. The slides were then mounted and imaged. Corticosterone assay Mouse whole blood was collected 90 min after CNO injection (5 mg/kg, i.p.). The samples were subsequently centrifuged at 2000× g for 10 min at 4°C after being left to stand at RT for 30–60 min. The supernatant was then carefully collected as the serum. Corticosterone levels were then measured using a commercial ELISA kit (Beyotime, Cat. No. PC100) according to the manufacturer’s instructions. Ex vivo electrophysiology The mice were anesthetized with urethane and then decapitated. The brain was quickly removed from the skull and immersed in precooled sucrose-based cutting solution (in mM, 225 sucrose, 2.5 KCl, 1.25 NaH 2 PO 4 , 26 NaHCO 3 , 11 D-glucose, 5 L-ascorbic acid, 3 sodium pyruvate, 7 MgSO 4 ·7H 2 O, 0.5 CaCl 2 ). After being fixed on a metal plate, the brain was cut into 300 µm slices. The slices were then collected and incubated in artificial cerebrospinal fluid (ACSF) containing (in mM): 122 NaCl, 2.5 KCl, 1.25 NaH 2 PO 4 , 26 NaHCO 3 , 11 D-glucose, 2 MgSO 4 ·7H 2 O, and 2 CaCl 2 equilibrated with 95% O 2 -5% CO 2 at 28°C for at least 1 hr before recording. To evoke PSCs using light, whole-cell recording of global SuM neurons near axons illuminated by ChR2-mCherry injected into the vSub was performed. The final light intensity at the end of the optical fiber was set to ~5 mW/mm 2 . Then, blue light (470 nm, width: 10 ms, frequency: 0.05 Hz) was used to evoke optically induced PSCs (oPSCs). DNQX (20 µM) was perfused into the ACSF to isolate AMPA-dependent currents from the oPSCs after establishing a 5 min baseline. In vivo electrophysiology The mice were anesthetized with 2% isoflurane and fixed to a stereotaxic device. A 16-channel microwire electrode array (KD-MWA, KedouBC), a 4×4 array of 25 µm NiTi wires spaced 200 µm apart, was slowly inserted into the mouse brain. Four small nails were first inserted into the skull, with a ground wire presoldered onto one of them. The electrode array was left in the SuM (AP: –2.8, ML: 0, DV: –4.55 mm), and then dental cement was used to fix it onto the skull. The mice were introduced to the recording area at least 1 week after surgery. During the day, the electrode array attached to the mouse skull was connected to the OpenEphys acquisition board through an Intan head stage. An OpenEphys GUI was used to visualize and save electrical signals. The mice were allowed to move freely inside a home cage-like arena for at least 20 min. Only data acquired during the last 5 min were saved and then analyzed via Python-based software. Spikes were detected and divided into single units using SpikeInterface ( Buccino et al., 2020 ; https://github.com/SpikeInterface/spikeinterface ; Buccino et al., 2026 ). Continuous binary raw data (sampling rate: 30 kHz) were imported and filtered using a bandpass butter filter at a cutoff value of 300 Hz. Movement artifacts were removed by subtracting medians across all channels. The templates were then extracted and fitted using SpyKING CIRCUS 2 inside the SpikeInterface frame. Neurons meeting the following criteria were excluded from the subsequent analysis: (1) spikes with refracting period violations smaller than 1 ms, accounting for more than 2% of total spikes, and (2) a total frequency lower than 0.2 Hz. Neurons with spike frequencies ≥10 Hz were considered RNs, whereas those with spike frequencies <10 Hz were considered FNs, as reported in a previous study ( Li et al., 2022b ). The local field potential was extracted and analyzed using the power spectrum analysis tool in MATLAB. Statistical analysis The data are presented as the means ± SEMs in all the figures in this manuscript. For normally distributed data with equal standard deviations, independent t tests for unpaired data and dependent t tests for paired data were performed in GraphPad software to compare mean values between two groups. Otherwise, the Mann-Whitney test for unpaired data and the Wilcoxon test for paired data were performed instead. One-way ANOVA followed by Tukey’s post hoc test and two-way ANOVA followed by Sidak’s post hoc test were performed to compare mean values among more than three groups. A p value less than 0.05 was considered to indicate a statistically significant difference between groups. ‘*’ represents p<0.05, ‘**’ represents p<0.01, and ‘***’ represents p<0.001. Acknowledgements We thank Dr. Wenting Wang for his generous gift of the transgenic mice. We thank all the members of MTT for their valuable comments. This research was funded by the National Natural Science Foundation of China (No. 82371518, 82071516, and 82441060), STI 2030—Major Projects 2021ZD0200500, the Humanities and Social Science Fund of the Ministry of Education of China (No. 22XJC880005), the Innovation Capability Support Program of Shaanxi (Program No. 2021PT-055), the Natural Science Basic Research Plan in Shaanxi Province of China (Program No. 2024JC-YBQN-0902 and 2023-JC-YB-189), and the Scientific and Technological Innovation Team of Shaanxi Innovation Capability Support Plan (No. 2022TD-47). Funding Statement The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. Contributor Information Jing Han, Email: jhan2012@snnu.edu.cn. Kate M Wassum, University of California, Los Angeles, United States. Kate M Wassum, University of California, Los Angeles, United States. Funding Information This paper was supported by the following grants: National Natural Science Foundation of China

82371518 to Jing Han. National Natural Science Foundation of China

82071516 to Zhiqiang Liu. National Natural Science Foundation of China

82441060 to Jing Han. STI 2030-Major Projects 2021ZD0200500 to Zongpeng Sun. Humanities and Social Science Fund of Ministry of Education of China

22XJC880005 to Zongpeng Sun. Innovation Capability Support Program of Shaanxi

2021PT-055 to Zhaoqiang Qian. Natural Science Basic Research Program of Shaanxi Province

2024JC-YBQN-0902 to Yuan Chang. Natural Science Basic Research Program of Shaanxi Province

2023-JC-YB-189 to Yanning Qiao. Scientific and Technological Innovation Team of Shaanxi Innovation Capability Support Plan

2022TD-47 to Jing Han. Additional information Competing interests No competing interests declared. Author contributions Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing. Investigation, Visualization, Writing – original draft. Investigation, Visualization, Writing – review and editing. Resources, Investigation, Visualization, Writing – review and editing. Formal analysis, Investigation. Resources, Funding acquisition, Project administration. Resources, Funding acquisition. Resources, Investigation. Resources, Funding acquisition. Resources, Supervision, Funding acquisition, Writing – review and editing. Conceptualization, Resources, Supervision, Funding acquisition, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing. Ethics All laboratory procedures were conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals in China and the regulations of the Animal Care and Use Committee of Shaanxi Normal University. This study protocol was reviewed and approved by the Academic Committee of the Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University (Approval No. L20230102-01). Additional files MDAR checklist elife-108593-mdarchecklist1.docx (88.8KB, docx) Data availability All data generated or analyzed during this study are included in the manuscript and source data files. Customized code can be accessed through GitHub ( https://github.com/zjm199502/elife2026 , copy archived at Zhang, 2026 ). References Aggleton JP, Christiansen K. The subiculum: the heart of the extended hippocampal system. Progress in Brain Research. 2015;219:65–82. doi: 10.1016/bs.pbr.2015.03.003. [ DOI ] [ PubMed ] [ Google Scholar ] Aranda L, Santín LJ, Begega A, Aguirre JA, Arias JL. Supramammillary and adjacent nuclei lesions impair spatial working memory and induce anxiolitic-like behavior. Behavioural Brain Research. 2006;167:156–164. doi: 10.1016/j.bbr.2005.09.002. [ DOI ] [ PubMed ] [ Google Scholar ] Azevedo EP, Pomeranz L, Cheng J, Schneeberger M, Vaughan R, Stern SA, Tan B, Doerig K, Greengard P, Friedman JM. A role of Drd2 hippocampal neurons in context-dependent food intake. Neuron. 2019;102:873–886. doi: 10.1016/j.neuron.2019.03.011. [ DOI ] [ PubMed ] [ Google Scholar ] Báez M, Siriczman I, Volosin M. Corticosterone is involved in foot shock-induced inactivity in rats. Physiology & Behavior. 1996;60:795–801. doi: 10.1016/0031-9384(96)00025-x. [ DOI ] [ PubMed ] [ Google Scholar ] Beck CH, Fibiger HC. Conditioned fear-induced changes in behavior and in the expression of the immediate early gene c-fos: with and without diazepam pretreatment. The Journal of Neuroscience. 1995;15:709–720. doi: 10.1523/JNEUROSCI.15-01-00709.1995. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Buccino AP, Hurwitz CL, Garcia S, Magland J, Siegle JH, Hurwitz R, Hennig MH. SpikeInterface, a unified framework for spike sorting. eLife. 2020;9:1834. doi: 10.7554/eLife.61834. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Buccino A, Garcia S, Mayorquin H, Yger P. Spikeinterface. 89381feGithub. 2026 https://github.com/SpikeInterface/spikeinterface Chen S, He L, Huang AJY, Boehringer R, Robert V, Wintzer ME, Polygalov D, Weitemier AZ, Tao Y, Gu M, Middleton SJ, Namiki K, Hama H, Therreau L, Chevaleyre V, Hioki H, Miyawaki A, Piskorowski RA, McHugh TJ. A hypothalamic novelty signal modulates hippocampal memory. Nature. 2020;586:270–274. doi: 10.1038/s41586-020-2771-1. [ DOI ] [ PubMed ] [ Google Scholar ] Chen Y, Zhou X, Chu B, Xie Q, Liu Z, Luo D, Zhang J. Restraint stress, foot shock and corticosterone differentially alter autophagy in the rat hippocampus, basolateral amygdala and prefrontal cortex. Neurochemical Research. 2024;49:492–506. doi: 10.1007/s11064-023-04048-x. [ DOI ] [ PubMed ] [ Google Scholar ] Cumbers MR, Chung ST, Wakerley JB. A neuromodulatory role for oxytocin within the supramammillary nucleus. Neuropeptides. 2007;41:217–226. doi: 10.1016/j.npep.2007.04.004. [ DOI ] [ PubMed ] [ Google Scholar ] Ding S-L, Yao Z, Hirokawa KE, Nguyen TN, Graybuck LT, Fong O, Bohn P, Ngo K, Smith KA, Koch C, Phillips JW, Lein ES, Harris JA, Tasic B, Zeng H. Distinct transcriptomic cell types and neural circuits of the subiculum and prosubiculum along the dorsal-ventral axis. Cell Reports. 2020;31:107648. doi: 10.1016/j.celrep.2020.107648. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Dos Santos Corrêa M, Vaz BDS, Grisanti GDV, de Paiva JPQ, Tiba PA, Fornari RV. Relationship between footshock intensity, post-training corticosterone release and contextual fear memory specificity over time. Psychoneuroendocrinology. 2019;110:104447. doi: 10.1016/j.psyneuen.2019.104447. [ DOI ] [ PubMed ] [ Google Scholar ] Escobedo A, Holloway SA, Votoupal M, Cone AL, Skelton H, Legaria AA, Ndiokho I, Floyd T, Kravitz AV, Bruchas MR, Norris AJ. Glutamatergic supramammillary nucleus neurons respond to threatening stressors and promote active coping. eLife. 2023;12:0972. doi: 10.7554/eLife.90972. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Fanselow MS, Dong HW. Are the dorsal and ventral hippocampus functionally distinct structures? Neuron. 2010;65:7–19. doi: 10.1016/j.neuron.2009.11.031. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Farrell JS, Lovett-Barron M, Klein PM, Sparks FT, Gschwind T, Ortiz AL, Ahanonu B, Bradbury S, Terada S, Oijala M, Hwaun E, Dudok B, Szabo G, Schnitzer MJ, Deisseroth K, Losonczy A, Soltesz I. Supramammillary regulation of locomotion and hippocampal activity. Science. 2021;374:1492–1496. doi: 10.1126/science.abh4272. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Forro T, Volitaki E, Malagon-Vina H, Klausberger T, Nevian T, Ciocchi S. Anxiety-related activity of ventral hippocampal interneurons. Progress in Neurobiology. 2022;219:102368. doi: 10.1016/j.pneurobio.2022.102368. [ DOI ] [ PubMed ] [ Google Scholar ] Ghasemi M, Navidhamidi M, Rezaei F, Azizikia A, Mehranfard N. Anxiety and hippocampal neuronal activity: Relationship and potential mechanisms. Cognitive, Affective & Behavioral Neuroscience. 2022;22:431–449. doi: 10.3758/s13415-021-00973-y. [ DOI ] [ PubMed ] [ Google Scholar ] Guenthner CJ, Miyamichi K, Yang HH, Heller HC, Luo L. Permanent genetic access to transiently active neurons via TRAP: targeted recombination in active populations. Neuron. 2013;78:773–784. doi: 10.1016/j.neuron.2013.03.025. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Handa RJ, Nunley KM, Lorens SA, Louie JP, McGivern RF, Bollnow MR. Androgen regulation of adrenocorticotropin and corticosterone secretion in the male rat following novelty and foot shock stressors. Physiology & Behavior. 1994;55:117–124. doi: 10.1016/0031-9384(94)90018-3. [ DOI ] [ PubMed ] [ Google Scholar ] Jimenez JC, Su K, Goldberg AR, Luna VM, Biane JS, Ordek G, Zhou P, Ong SK, Wright MA, Zweifel L, Paninski L, Hen R, Kheirbek MA. Anxiety cells in a hippocampal-hypothalamic circuit. Neuron. 2018;97:670–683. doi: 10.1016/j.neuron.2018.01.016. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Josselyn SA, Tonegawa S. Memory engrams: Recalling the past and imagining the future. Science. 2020;367:4325. doi: 10.1126/science.aaw4325. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kesner AJ, Shin R, Calva CB, Don RF, Junn S, Potter CT, Ramsey LA, Abou-Elnaga AF, Cover CG, Wang DV, Lu H, Yang Y, Ikemoto S. Supramammillary neurons projecting to the septum regulate dopamine and motivation for environmental interaction in mice. Nature Communications. 2021;12:2811. doi: 10.1038/s41467-021-23040-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kesner AJ, Mozaffarilegha M, Thirtamara Rajamani K, Arima Y, Harony-Nicolas H, Hashimotodani Y, Ito HT, Song J, Ikemoto S. Hypothalamic supramammillary control of cognition and motivation. The Journal of Neuroscience. 2023;43:7538–7546. doi: 10.1523/JNEUROSCI.1320-23.2023. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kim HD, Call T, Carotenuto S, Johnson R, Ferguson D. Testing depression in mice: A chronic social defeat stress model. Bio-Protocol. 2017;7:e2203. doi: 10.21769/BioProtoc.2203. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Koren T, Yifa R, Amer M, Krot M, Boshnak N, Ben-Shaanan TL, Azulay-Debby H, Zalayat I, Avishai E, Hajjo H, Schiller M, Haykin H, Korin B, Farfara D, Hakim F, Kobiler O, Rosenblum K, Rolls A. Insular cortex neurons encode and retrieve specific immune responses. Cell. 2021;184:5902–5915. doi: 10.1016/j.cell.2021.10.013. [ DOI ] [ PubMed ] [ Google Scholar ] Lacagnina AF, Brockway ET, Crovetti CR, Shue F, McCarty MJ, Sattler KP, Lim SC, Santos SL, Denny CA, Drew MR. Distinct hippocampal engrams control extinction and relapse of fear memory. Nature Neuroscience. 2019;22:753–761. doi: 10.1038/s41593-019-0361-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] LeDuke DO, Borio M, Miranda R, Tye KM. Anxiety and depression: A top-down, bottom-up model of circuit function. Annals of the New York Academy of Sciences. 2023;1525:70–87. doi: 10.1111/nyas.14997. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Li YD, Luo YJ, Chen ZK, Quintanilla L, Cherasse Y, Zhang L, Lazarus M, Huang ZL, Song J. Hypothalamic modulation of adult hippocampal neurogenesis in mice confers activity-dependent regulation of memory and anxiety-like behavior. Nature Neuroscience. 2022a;25:630–645. doi: 10.1038/s41593-022-01065-x. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Li J, Sun X, You Y, Li Q, Wei C, Zhao L, Sun M, Meng H, Zhang T, Yue W, Wang L, Zhang D. Auts2 deletion involves in DG hypoplasia and social recognition deficit: The developmental and neural circuit mechanisms. Science Advances. 2022b;8:1238. doi: 10.1126/sciadv.abk1238. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Liang M, Jian T, Tao J, Wang X, Wang R, Jin W, Chen Q, Yao J, Zhao Z, Yang X, Xiao J, Yang Z, Liao X, Chen X, Wang L, Qin H. Hypothalamic supramammillary neurons that project to the medial septum modulate wakefulness in mice. Communications Biology. 2023;6:1255. doi: 10.1038/s42003-023-05637-w. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Liu X, Ramirez S, Pang PT, Puryear CB, Govindarajan A, Deisseroth K, Tonegawa S. Optogenetic stimulation of a hippocampal engram activates fear memory recall. Nature. 2012;484:381–385. doi: 10.1038/nature11028. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] López-Ferreras L, Eerola K, Mishra D, Shevchouk OT, Richard JE, Nilsson FH, Hayes MR, Skibicka KP. GLP-1 modulates the supramammillary nucleus-lateral hypothalamic neurocircuit to control ingestive and motivated behavior in a sex divergent manner. Molecular Metabolism. 2019;20:178–193. doi: 10.1016/j.molmet.2018.11.005. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] López-Ferreras L, Eerola K, Shevchouk OT, Richard JE, Nilsson FH, Jansson LE, Hayes MR, Skibicka KP. The supramammillary nucleus controls anxiety-like behavior; key role of GLP-1R. Psychoneuroendocrinology. 2020;119:104720. doi: 10.1016/j.psyneuen.2020.104720. [ DOI ] [ PubMed ] [ Google Scholar ] Marcus DJ, Bedse G, Gaulden AD, Ryan JD, Kondev V, Winters ND, Rosas-Vidal LE, Altemus M, Mackie K, Lee FS, Delpire E, Patel S. Endocannabinoid signaling collapse mediates stress-induced amygdalo-cortical strengthening. Neuron. 2020;105:1062–1076. doi: 10.1016/j.neuron.2019.12.024. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Mueller NK, Dolgas CM, Herman JP. Stressor-selective role of the ventral subiculum in regulation of neuroendocrine stress responses. Endocrinology. 2004;145:3763–3768. doi: 10.1210/en.2004-0097. [ DOI ] [ PubMed ] [ Google Scholar ] Pan WX, McNaughton N. The role of the medial supramammillary nucleus in the control of hippocampal theta activity and behaviour in rats. The European Journal of Neuroscience. 2002;16:1797–1809. doi: 10.1046/j.1460-9568.2002.02267.x. [ DOI ] [ PubMed ] [ Google Scholar ] Pedersen NP, Ferrari L, Venner A, Wang JL, Abbott SBG, Vujovic N, Arrigoni E, Saper CB, Fuller PM. Supramammillary glutamate neurons are a key node of the arousal system. Nature Communications. 2017;8:1405. doi: 10.1038/s41467-017-01004-6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Peng B, Xu Q, Liu J, Guo S, Borgland SL, Liu S. Corticosterone attenuates reward-seeking behavior and increases anxiety via d2 receptor signaling in ventral tegmental area dopamine neurons. The Journal of Neuroscience. 2021;41:1566–1581. doi: 10.1523/JNEUROSCI.2533-20.2020. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Qin H, Fu L, Jian T, Jin W, Liang M, Li J, Chen Q, Yang X, Du H, Liao X, Zhang K, Wang R, Liang S, Yao J, Hu B, Ren S, Zhang C, Wang Y, Hu Z, Jia H, Konnerth A, Chen X. REM sleep-active hypothalamic neurons may contribute to hippocampal social-memory consolidation. Neuron. 2022;110:4000–4014. doi: 10.1016/j.neuron.2022.09.004. [ DOI ] [ PubMed ] [ Google Scholar ] Ryan TJ, Roy DS, Pignatelli M, Arons A, Tonegawa S. Memory Engram cells retain memory under retrograde amnesia. Science. 2015;348:1007–1013. doi: 10.1126/science.aaa5542. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Shi HJ, Wang S, Wang XP, Zhang RX, Zhu LJ. Hippocampus: Molecular, cellular, and circuit features in anxiety. Neuroscience Bulletin. 2023;39:1009–1026. doi: 10.1007/s12264-023-01020-1. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Silveira MCL, Sandner G, Graeff FG. Induction of Fos immunoreactivity in the brain by exposure to the elevated plus-maze. Behavioural Brain Research. 1993;56:115–118. doi: 10.1016/0166-4328(93)90028-O. [ DOI ] [ PubMed ] [ Google Scholar ] Strange BA, Witter MP, Lein ES, Moser EI. Functional organization of the hippocampal longitudinal axis. Nature Reviews. Neuroscience. 2014;15:655–669. doi: 10.1038/nrn3785. [ DOI ] [ PubMed ] [ Google Scholar ] Sun X, Bernstein MJ, Meng M, Rao S, Sørensen AT, Yao L, Zhang X, Anikeeva PO, Lin Y. Functionally distinct neuronal ensembles within the memory engram. Cell. 2020;181:410–423. doi: 10.1016/j.cell.2020.02.055. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Sun H, Li Z, Qiu Z, Shen Y, Guo Q, Hu SW, Ding HL, An S, Cao JL. A common neuronal ensemble in nucleus accumbens regulates pain-like behaviour and sleep. Nature Communications. 2023;14:450-3. doi: 10.1038/s41467-023-40450-3. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Sweis BM, Mau W, Rabinowitz S, Cai DJ. Dynamic and heterogeneous neural ensembles contribute to a memory engram. Current Opinion in Neurobiology. 2021;67:199–206. doi: 10.1016/j.conb.2020.11.017. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Tang H, Wu GS, Xie J, He X, Deng K, Wang H, Xu F, Luo HR. Brain-wide map of projections from mice ventral subiculum. Neuroscience Letters. 2016;629:171–179. doi: 10.1016/j.neulet.2016.07.014. [ DOI ] [ PubMed ] [ Google Scholar ] Tonegawa S, Liu X, Ramirez S, Redondo R. Memory engram cells have come of age. Neuron. 2015;87:918–931. doi: 10.1016/j.neuron.2015.08.002. [ DOI ] [ PubMed ] [ Google Scholar ] Tonegawa S, Morrissey MD, Kitamura T. The role of engram cells in the systems consolidation of memory. Nature Reviews. Neuroscience. 2018;19:485–498. doi: 10.1038/s41583-018-0031-2. [ DOI ] [ PubMed ] [ Google Scholar ] Yan JJ, Ding XJ, He T, Chen AX, Zhang W, Yu ZX, Cheng XY, Wei CY, Hu QD, Liu XY, Zhang YL, He M, Xie ZY, Zha X, Xu C, Cao P, Li H, Xu XH. A circuit from the ventral subiculum to anterior hypothalamic nucleus GABAergic neurons essential for anxiety-like behavioral avoidance. Nature Communications. 2022;13:7464. doi: 10.1038/s41467-022-35211-7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Zaki Y, Cai DJ. Memory engram stability and flexibility. Neuropsychopharmacology. 2024;50:285–293. doi: 10.1038/s41386-024-01979-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Zhang TR, Larosa A, Di Raddo ME, Wong V, Wong AS, Wong TP. Negative memory engrams in the hippocampus enhance the susceptibility to chronic social defeat stress. The Journal of Neuroscience. 2019;39:7576–7590. doi: 10.1523/JNEUROSCI.1958-18.2019. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Zhang J. Software Heritage; 2026. https://archive.softwareheritage.org/swh:1:dir:80ceeb41eb7fc353b3036e752113088fea7d95cd;origin=https://github.com/zjm199502/elife2026;visit=swh:1:snp:7afc301f8557e170381b66034c129f170c74ab47;anchor=swh:1:rev:ef9ba23ce6fb22d88b1a557e39662bc4fe949aef [ Google Scholar ] Zheng Z, Liu Y, Mu R, Guo X, Feng Y, Guo C, Yang L, Qiu W, Zhang Q, Yang W, Dong Z, Qiu S, Dong Y, Cui Y. A small population of stress-responsive neurons in the hypothalamus-habenula circuit mediates development of depression-like behavior in mice. Neuron. 2024;112:3924–3939. doi: 10.1016/j.neuron.2024.09.012. [ DOI ] [ PubMed ] [ Google Scholar ] eLife. doi: 10.7554/eLife.108593.3.sa0 eLife Assessment Kate M Wassum Kate M Wassum 1 University of California, Los Angeles, United States Reviewing Editor Find articles by Kate M Wassum 1 Author information Article notes Copyright and License information 1 University of California, Los Angeles, United States Roles Kate M Wassum : Reviewing Editor Keywords: Convincing Keywords: Valuable PMC Copyright notice This manuscript provides a valuable contribution by identifying a stress-responsive circuit and its regulation of anxiety-related behaviors. The evidence is convincing that the supramammillary nucleus contains stress-responsive neurons that increase anxiety-like behaviors when activated, and that ventral subiculum projections to the supramammillary are also activated by stress and their inhibition alleviates some effects of stress. Evidence that this pathway encodes and is functionally specific to anxiety is, at present, not sufficiently support and will require future studies. This work offers new insights into how distinct circuits are activated by stress and can regulate emotional behaviors and will be of interest to those interested in brain systems of aversive emotional and behavioral states. eLife. doi: 10.7554/eLife.108593.3.sa1 Reviewer #1 (Public review): Anonymous Anonymous Reviewer Find articles by Anonymous Author information Copyright and License information Roles Anonymous : Reviewer PMC Copyright notice In the revised manuscript, the authors refine their conclusions, narrow their interpretation, and add limited new analyses but have not added additional new data or made fundamental changes in the analyses of their data. The central findings are that the SuM contains neurons that are activated by stressors (foot shock and social defeat). Chemogenetic activation of SuM and the neurons genetically tagged as active during foot shocks, which the authors define as Stress Activated Neurons, increases classic anxiety-like behaviors. The subiculum projects to the SuM, and terminals in the SuM from the ventral versus dorsal subiculum are differentially active during elevated plus-maze transitions. Chronic inhibition of vSub neurons that project to SuM mitigates CSDS-induced anxiety-like behaviors. Due to limitations in the data and experimental design the findings are felt to remain incomplete. A central limitation is the discordance between the temporal resolution of the behavioral assays and the neural interventions used. This weakens support for the conclusions drawn about the causal roles the SuM and specific vSub projections to SuM (vSub→SuM) may play in anxiety and anxiety-like behaviors. The authors acknowledge this limitation but do not address it experimentally in the revised manuscript. Furthermore, the connection between chronic inhibition of vSub→SuM neurons for 10 days and the alleviation of CSDS-induced anxiety is incomplete. Separately, the use of foot shock and social defeat stressors in connection with SuM neurons, with limited exploration of the potential (or lack thereof) relation between the two groups, further limits the ability to draw conclusions from the data. Although a number of interesting points are raised through the experiments the weakness noted will reduce the impact of the work in the field. eLife. doi: 10.7554/eLife.108593.3.sa2 Reviewer #2 (Public review): Anonymous Anonymous Reviewer Find articles by Anonymous Author information Copyright and License information Roles Anonymous : Reviewer PMC Copyright notice This manuscript investigates the neural mechanisms of anxiety and identifies the supramammillary nucleus (SuM) as a critical hub in mediating anxiety-related behaviors. The authors describe a population of neurons in the SuM that are activated by acute and chronic stress. While their activity is not required for fear memory recall, reactivation of these neurons after chronic stress robustly increases anxiety-like behaviors as well as physiological stress markers. Circuit analysis further shows that these stress-activated neurons are driven by inputs from the ventral, but not dorsal, subiculum, and inhibition of this pathway exerts an anxiolytic effect. The study provides an elegant integration of techniques linking stress, neuronal ensembles, and circuit function, advancing our understanding of the neural substrates of anxiety. A particularly notable point is the selective role of these stress-activated neurons in anxiety, but not in associative fear memory, highlighting functional distinctions between neural circuits underlying anxiety and fear. The recruited neuronal population is activated by acute and chronic stress, though the overlap across stress exposures is partial, suggesting that further studies will be important to define how these neurons respond under other stressors and conditions. Overall, this work identifies SuM stress-activated neurons and their ventral subiculum inputs as central elements of anxiety circuitry, providing a valuable framework for future studies and potential targeted interventions for stress-related disorders. eLife. doi: 10.7554/eLife.108593.3.sa3 Reviewer #3 (Public review): Anonymous Anonymous Reviewer Find articles by Anonymous Author information Copyright and License information Roles Anonymous : Reviewer PMC Copyright notice Summary: The authors aim to investigate the mechanisms of anxiety. The paper focuses on supramammillary nucleus (SuM) based on a fos screen and recordings showing that footshock and social defeat stress increases activity in this region. Using activity-dependent tagging, they show that reactivation of stress-activated neurons in SuM has an anxiety-like effect, reducing open-arm exploration in the elevated zero task. They then investigate the ventral subiculum as a potential source of anxiety-related information for SuM. They show that ventral subiculum (vSub) inputs to SuM are more strongly activated than dSub when mice explore open arms of the elevated zero. Finally, they show that DREADD-mediated inhibition of vSub-SuM projections alleviates stress-enhanced anxiety. Overall the results provide good evidence that SuM contains a stress-activated neuronal population whose later activity increases anxiety-like behavior. It further provides evidence that vSub projects to SuM are activated by stress and their inhibition alleviates some effects of stress. Strengths: Strengths of this paper include the use of convergent methods (e.g., fos plus electrode recordings, footshock and social defeat) to demonstrate that the SuM is activated by different forms of stress. The activity-dependent tagging experiment shows that footshock-activated SuM neurons are reactivated by social defeat but not sucrose is also compelling because it provides evidence that SuM neurons are driven by some integrative aspect of stress rather than by a simple sensory stimulus. Weaknesses: The strength of some evidence is judged to be incomplete. The paper provides good evidence that SuM contains stress-responsive neurons, and the activity of these neurons increases some measure of anxiety-like behavior. However, the evidence that the vSub-SuM projection "encodes anxiety" and that the SuM is a key regulator of anxiety is judged to be incomplete. I am not convinced that the identified SuM cells have a specific anxiety function. As the authors mention in the introduction, SuM regulates exploration and theta activity. Since theta potently regulates hippocampal function, there is the concern that SuM manipulations could have broad effects beyond anxiety-like behavior. eLife. 2026 Apr 15;14:RP108593. doi: 10.7554/eLife.108593.3.sa4 Author response Jinming Zhang Jinming Zhang 1 Shaanxi Normal University, Xi'an, China Author Find articles by Jinming Zhang 1 , Kexin Yu Kexin Yu 2 Shaanxi Normal University, Xi'an, China Author Find articles by Kexin Yu 2 , Junmin Zhang Junmin Zhang 3 Shaanxi Normal University, Xi'an, China Author Find articles by Junmin Zhang 3 , Yuan Chang Yuan Chang 4 Xi&apos;an Medical University, Xi'an, China Author Find articles by Yuan Chang 4 , Xiao Sun Xiao Sun 5 Shaanxi Normal University, Xi'an, China Author Find articles by Xiao Sun 5 , Zhaoqiang Qian Zhaoqiang Qian 6 Shaanxi Normal University, Xi'an, China Author Find articles by Zhaoqiang Qian 6 , Zongpeng Sun Zongpeng Sun 7 Shaanxi Normal University, Xi'an, China Author Find articles by Zongpeng Sun 7 , Yanning Qiao Yanning Qiao 8 Shaanxi Normal University, Xi'an, China Author Find articles by Yanning Qiao 8 , Zhiqiang Liu Zhiqiang Liu 9 Shaanxi Normal University, Xi'an, China Author Find articles by Zhiqiang Liu 9 , Wei Ren Wei Ren 10 Shaanxi Normal University, Xi'an, China Author Find articles by Wei Ren 10 , Jing Han Jing Han 11 Shaanxi Normal University, Xi'an, China Author Find articles by Jing Han 11 Author information Article notes Copyright and License information 1 Shaanxi Normal University, Xi'an, China 2 Shaanxi Normal University, Xi'an, China 3 Shaanxi Normal University, Xi'an, China 4 Xi&apos;an Medical University, Xi'an, China 5 Shaanxi Normal University, Xi'an, China 6 Shaanxi Normal University, Xi'an, China 7 Shaanxi Normal University, Xi'an, China 8 Shaanxi Normal University, Xi'an, China 9 Shaanxi Normal University, Xi'an, China 10 Shaanxi Normal University, Xi'an, China 11 Shaanxi Normal University, Xi'an, China Roles Jinming Zhang : Author Kexin Yu : Author Junmin Zhang : Author Yuan Chang : Author Xiao Sun : Author Zhaoqiang Qian : Author Zongpeng Sun : Author Yanning Qiao : Author Zhiqiang Liu : Author Wei Ren : Author Jing Han : Author Collection date 2026. PMC Copyright notice The following is the authors’ response to the original reviews. Public Reviews: Reviewer #1 (Public review): Weaknesses As presented, the manuscript has limitations that weaken support for the central conclusions drawn by the authors. Many of the findings align with prior work on this topic, but do not extend those findings substantially. An overarching limitation is the lack of temporal resolution in the manipulations relative to the behavioral assays. This is particularly important for anxiety-like behaviors, as antecedent exposures can alter performance. In the open field and elevated zero maze assays, testing occurred 30 minutes after CNO injection. During much of this interval, the targeted neurons were likely active, making it difficult to determine whether observed behavioral changes were primary - resulting directly from SuM neuronal activity - or secondary, reflecting a stress-like state induced by prolonged activation of SuM and related circuits. This concern also applies to the chronic inhibition of ventral subiculum (vSub) neurons during 10 days of CSDS. We appreciate the reviewer's concern regarding the timing of CNO administration relative to behavioral testing. The 30-minute interval was selected according to some previous studies[1, 2]. This window ensures stable and specific neuronal manipulation while minimizing off-target effects and was strictly performed through all experiments. We acknowledge that shorter interval (~15 mins) can be efficient to produce biological effect in vivo[3, 4]. We repeated chemogenetic tests 2-3 times to make sure to get reliable data for statistical analysis. However, we cannot exclude potential side-effects caused by chemogenetically prolonged activation of SuM because of its poor temporal resolution compared to optogenetic manipulation. We agree that employing techniques with higher temporal resolution, such as optogenetics, in future studies would provide an excellent complement to these findings. The combination of stressors (foot shock and CSDS) and behavioral assays further complicates interpretation. The precise role of SuM neurons, including SANs, remains unclear. Both vSub and dSub neurons responded to foot shock, but only vSub neurons showed activity differences associated with open-arm transitions in the EZM. We agree that the use of multiple stressors (foot shock and CSDS) adds complexity to the interpretation. Our rationale was to test the generality of the SuM response and the role of SANs across different stress modalities (acute vs. chronic). The key finding is that while both vSub and dSub projections to the SuM were activated by the acute stressor of foot shock (Figure 5N-R), only the vSub-SuM pathway showed a significant increase in calcium activity specifically during the anxiety-provoking transition from the closed to the open arms of the EZM (Figure 5I-M). This dissociation suggests a selective role for the vSub-SuM circuit in encoding anxiety-related information, beyond a general response to stress. In light of prior studies linking SuM to locomotion (Farrell et al., Science 2021; Escobedo et al., eLife 2024), the absence of analyses connecting subpopulations to locomotor changes weakens the claim that vSub neurons selectively encode anxiety. Because open- and closed-arm transitions are inherently tied to locomotor activity, locomotion must be carefully controlled to avoid confounding interpretations. We thank the reviewer for highlighting the important studies linking the SuM to locomotion. We acknowledge this known function and carefully considered it in our analyses. Non-selective activation of the entire SuM didn’t affect total distance traveled in open field and elevated zero maze (Supplemental Figure 2 B-C). Although the locomotion of mice in OF and EZM was affected while targeting SANs, we also compared the travel distance in the central area of OF, to some extent, to minimize the influence of locomotion on the estimation of anxiety produced avoidance to the central area (Figure 4 I). We agree that future work delineating the specific subpopulations within the SuM that regulate locomotion versus anxiety would be highly valuable. Another limitation is the narrow behavioral scope. Beyond open field and EZM, no additional assays were used to assess how SAN reactivation affects other behaviors. Without richer behavioral analyses, interpretations about fear engrams, freezing, or broader stress-related functions of SuM remain incomplete. In addition, small n values across several datasets reduce confidence in the strength of the conclusions. We acknowledge that the primary focus on OF and EZM tests is a limitation in fully characterizing the behavioral profile of SAN manipulation. These tests were selected as they are well-validated, standard assays for anxiety-like behavior in rodents[5–10]. However, we also included the reward-seeking test, where activation of SANs significantly suppressed sucrose consumption (Figure 4L), suggesting a broader impact on motivational state that is often linked to anxiety. We fully agree with the reviewer that employing a richer behavioral battery—such as tests for social avoidance, conditioned place aversion, or Pavlovian fear conditioning—in future studies will be essential to comprehensively define the functional scope of SuM SANs and to conclusively dissect their role from fear memory engrams. Figure level concerns: (1) Figure 1: In Figure 1, the acute recruitment of SuM neurons by for shock is paired with changes in neural activity induced by social defeat stress. Although interesting, the connections of changes induced by a chronic stressor to Fos induction following acute foot shock are unclear and do not establish a baseline for the studies in Figure 3 on activation of SANs by social stressors. Thank you for this important comment. We agree that directly linking acute foot shock-induced cFos expression with chronic social defeat stress (CSDS) electrophysiological changes may create an interpretive gap. In Figure 1, we aimed to demonstrate that both acute (foot shock) and chronic (CSDS) stressors can activate SuM neurons, using complementary methods (cFos for acute, in vivo recording for chronic). We did not intend to imply that the same neuronal population responds identically to both stressors. To address this, we have clarified in the text that the purpose of Figure 1 is to show that SuM is responsive to diverse stressors, rather than to establish a direct mechanistic link between acute and chronic activation patterns. The baseline for SAN studies in Figure 3 is established through the TRAP2 tagging protocol following foot shock, independent of the CSDS model. We acknowledge that future studies should compare SAN recruitment across acute vs. chronic stressors to better define their functional overlap. (2) Figure 2: The chemogenetic experiments using AAV-hSyn-Gq-DREADDs lack data or images, or hit maps showing viral spread across animals. This omission is critical given the small size of SuM, where viral spread directly determines which neurons are manipulated. Without this, it is difficult to interpret findings in the context of prior studies on SuM circuits involved in threats and rewards. Please see Supplemental Figure 2 for the infection area of AAV. (3) Figure 3: The TRAP experiments show that the number of labeled neurons following foot shock (Figure 3F) is approximately double that of baseline home-cage animals, though y-axis scaling complicates interpretation. It is unclear whether this reflects true Fos induction, low TRAP efficiency, or baseline recombination. We thank the reviewer for pointing out the axis scaling issue. We have modified the y-axis to start from 0. The SuM nucleus has been reported to play role in the awake of rodents, it’s reasonable to have some basal neuronal activation after 4-OHT i.p. injection. Overlap analyses are also limited. For example, it is not shown what proportion of foot shock SANs are reactivated by subsequent foot shock. Comparisons of Fos induction after sucrose reward are also weakened by the very low Fos signal observed. If sucrose reward does not robustly induce Fos in SuM, its utility in distinguishing reward- versus stress-activated neurons is questionable. Thus, conclusions about overlap between SANs and socially stressed neurons remain uncertain due to the missing quantification of Fos+ populations. Thank you for the question. We have replaced the reactivation chance graph with a new reactivation percent analysis graph to show the proportion of SANs that reactivated by subsequent sucrose reward or stress. The rationale we use social stress other than foot shock is to show the potential generality of foot-shock tagged neurons. The lower expression of cFos after sucrose exposure suggest first, the SuM may not involve in reward regulation, which we agree with you; second, those SANs are more likely to modulate anxiety-like behavior but not reward. (4) Supplemental Figure 3: The claim that "SANs in the SuM encode anxiety but not fear memory" is not well supported. Inhibition of SANs (Gi-DREADDs) did not alter freezing behavior, but the absence of change could reflect technical issues (e.g., insufficient TRAP efficiency, low expression of Gi-DREADDs). Moreover, the manuscript does not provide a positive control showing that SuM SANs inhibition alters anxiety-like behavior, making it difficult to interpret the negative result. Prior work (Escobedo et al., eLife 2024) suggests SuM neurons drive active responses, not freezing, raising further interpretive questions. We agree that here we didn’t provide enough data to confirm there is no regulation effect of SuM-SANs on fear memory. Relevant statement has been removed to avoid any further misunderstanding. (5) Figure 4: The statement that corticosterone concentration is "usually used to estimate whether an individual is anxious" (line 236) is an overstatement. Corticosterone fluctuates dynamically across the day and responds to a broad range of stimuli beyond anxiety. Thank you for your kind reminder. Corticosterone/cortisol, the primary stress hormone, is a well-established biomarker whose levels are elevated in response to stress and in anxiety states.[11, 12]. Some studies also reported that supplying corticosterone can produce anxiety-like behaviors in rodents[13–16]. We collect the blood sample at the same timepoint in Figure 4 C-D. We agree that line 236 is a kind of overstatement and has modified. (6) Figures 5-6: The conclusion that vSub neurons encode anxiety-like behavior is not firmly supported. Data from photo-activating terminals in SuM is shown for ex vivo recording, but not in vivo behavior, which would strengthen support for this conclusion. Both vSub and dSub neurons responded to foot shock. The key evidence comes from apparent differential recruitment during open-arm exploration. However, the timing appears to lag arm entry, no data are provided for closed-arm entry, and there is heterogeneity across animals. These limitations reduce confidence in the authors' central claim regarding vSub-specific encoding of anxiety. We thank the reviewer for this important point. To address the concern regarding the in vivo behavioral encoding specificity of the vSub-SuM pathway, we further analyzed the in vivo fiber photometry data. The new analysis revealed that calcium activity in vSub-SuM projection neurons exhibited bidirectional, instantaneous, and specific changes during transitions between the open and closed arms of the elevated plus maze: their activity significantly and immediately decreased when mice moved from the open arm to the closed arm (new results shown in Supplemental Figure 5), and conversely, significantly and immediately increased upon transitioning from the closed to the open arm. However, under the same behavioral events, dSub-SuM projection neurons showed no significant change in activity. We hope this finding could strengthens the role of the vSub-SuM pathway in encoding anxiety-like behavior. An appraisal of whether the authors achieved their aims, and whether the results support their conclusions: (1) From the data presented, the authors conclude that "the SuM is the critical brain region that regulates anxiety" (line 190). This interpretation appears overstated, as it downplays well-established contributions of other brain regions and does not place SuM's role within a broader network context. The data support that SuM neurons are recruited by foot shock and, to a lesser extent, by acute social stress. However, the alterations in activity of SuM subpopulations following chronic stress reported in Figure 1 remain largely unexplored, limiting insight into their functional relevance. Thank you for the suggestion. We have modified the line 190 with cautious “In this study, we combined multiple methods to determine whether the SuM is a brain region that involve in modulating anxiety.” (2) The limited temporal resolution of DREADD-based manipulations leaves alternative explanations untested. For example, if SANs encode signals of threat, generalized stress, or nociception, then prolonged activation could indirectly alter behavior in the open field and EZM assays, rather than reflecting direct anxiety regulation. We discussed the DREADD method in the first part in our response. (3) The conclusion that "SuM store information about stress but not memory" (line 240) is not fully supported, particularly with respect to possible roles in memory. The lack of a role in memory of events, as opposed to the output of threat or stress memory, may be true, but is functionally untested in presented experiments. The data do indicate activation of the SuM neuron by foot shock, which has been previously reported (Escobedo et al eLife 2024). The changes in SuM activity following chronic stress (Figure 1) are intriguing, but their relationship to "stress information storage" is not clearly established. Thank you for your valuable comments. Foot-shock-activated neurons may play role in modulate any of the following anxiety-like behaviors and emotional memory (fear memory). We realized that we didn’t fully test all aspects of anxiety and memory, thus resulting in some overstatements in the manuscript. It is more proper to focus on “anxiety avoidance” according to the reduced open-arm exploration in EZM/EPM. Reviewer #2 (Public review): This manuscript investigates the neural mechanisms of anxiety and identifies the supramammillary nucleus (SuM) as a critical hub in mediating anxiety-related behaviors. The authors describe a population of neurons in the SuM that are activated by acute and chronic stress. While their activity is not required for fear memory recall, reactivation of these neurons after chronic stress robustly increases anxiety-like behaviors as well as physiological stress markers. Circuit analysis further shows that these stress-activated neurons are driven by inputs from the ventral, but not dorsal, subiculum, and inhibition of this pathway exerts an anxiolytic effect. The study provides an elegant integration of techniques to link stress, neuronal ensembles, and circuit function, thereby advancing our understanding of the neural substrates of anxiety. A particularly notable point is the selective role of these stress-activated neurons in anxiety, but not in associative fear memory, which highlights functional distinctions between neural circuits underlying anxiety and fear. Some aspects would benefit from clarification. For example, how selective is the recruitment of this population to stress compared with other aversive states, and how should one best interpret their definition as "stress-activated neurons" given the relatively modest overlap across stress exposures? In addition, the use of the term "engram" in this context raises conceptual questions. Is it appropriate to describe a neuronal ensemble encoding an emotional state as an engram, a term usually tied to specific memory recall? Overall, this work makes a valuable contribution by identifying SuM stress-activated neurons and their ventral subiculum inputs as central elements of the circuitry underlying anxiety. These findings provide a valuable framework for future studies investigating anxiety circuitry and may inform the development of targeted interventions for stress-related disorders. We thank the reviewer for raising these important points. We agree that further clarification is warranted. In our study, we compared SAN reactivation across different stimuli: foot shock (acute physical stress), social stress (chronic psychosocial stress), and sucrose reward (non-aversive positive stimulus). As shown in Figure 3, SANs in the supramammillary nucleus (SuM) were significantly reactivated by social stress but not by sucrose reward. Moreover, the c-Fos response in SuM was markedly higher after foot shock compared to home cage controls (Figure 1). While we did not test all possible aversive states (e.g., pain, sickness), our data support that SuM SANs are preferentially recruited by stressors rather than by reward or neutral conditions. We acknowledge that the overlap across stress modalities is not complete, which may reflect differences in stress intensity, duration, or circuit engagement. Future work will systematically compare SAN recruitment across diverse aversive and non-aversive states to further define their selectivity. The term “stress-activated neurons” (SANs) here refers to neurons that are reliably activated by at least one type of stressor and can be reactivated by subsequent stress exposure. The partial overlap across stressors likely reflects the diversity of stress responses and the possibility that distinct subpopulations within SuM may encode different aspects of aversive experience. Importantly, chemogenetic activation of SANs was sufficient to induce anxiety-like behavior and elevate corticosterone (Figure 4), supporting their functional role in stress-related behavioral and physiological outputs. We have revised the manuscript to clarify that SANs represent a stress-responsive ensemble rather than a uniform population activated identically by all stressors. We appreciate the reviewer’s conceptual caution. In the revised manuscript, we intentionally avoided using the term “engram” to describe SANs. Our focus is on a stress-activated neuronal ensemble that drives anxiety-like behavior, not on memory recall per se. We refer to SANs as an “ensemble” or “population” rather than an engram, consistent with the TRAP-based labeling approach used to capture neurons activated during a specific experience. We agree that “engram” is best reserved for memory-encoding cells and will ensure this distinction remains clear throughout the text. Reviewer #3 (Public review): Weaknesses: The strength of some of the evidence is judged to be incomplete. The paper provides good evidence that SuM contains stress-responsive neurons, and the activity of these neurons increases some measure of anxiety-like behavior. However, the evidence that the vSub-SuM projection "encodes anxiety" and that the SuM is a key regulator of anxiety is judged to be incomplete. The claim that SuM generates an "anxiety engram" is also judged to be incompletely supported by the evidence. Namely, what is unclear is whether these cells/regions encode anxiety per se versus modulate behaviors (like exploration) that tend to correlate with anxiety. Since many brain regions respond to footshock and other stressors, the response of SuM to these stimuli is not strong evidence for a role in anxiety. I am not convinced that the identified SuM cells have a specific anxiety function. As the authors mention in the introduction, SuM regulates exploration and theta activity. Since theta potently regulates hippocampal function, there is the concern that SuM manipulations could have broad effects. As shown in Supplementary Figure 2, stimulating stress-responsive cells in SuM potently reduces general locomotor exploration. This raises concerns that the manipulation could have broader effects that go beyond just changes in anxiety-like behavior. Furthermore, the meaning of an "anxiety engram" is unclear. Would this engram encode stress, the sense of a potential threat, or the behavioral response? A more developed analysis of the behavioral correlates of SuM activity and the behavioral effects of SuM manipulations could give insight into these questions. We appreciate the reviewer’s thoughtful critique regarding the specificity of SuM’s role in anxiety and the interpretation of our findings. We acknowledge that SuM has broad functions, including regulating exploration and hippocampal theta. However, our data show that general SuM activation increases anxiety-like measures (reduced open-arm time in EZM, decreased center exploration in OF) without altering total locomotion (Fig. 2, Suppl. Fig. 2). The locomotor reduction in SAN activation experiments (Suppl. Fig. 2F–G) was observed alongside clear anxiety-like behavioral changes (e.g. suppressed reward seeking), suggesting that the effects are not solely due to motor suppression. We agree that the methods we used to estimate anxiety-like behaviors base on mice movement when testing, and this could be a shortage of this research when trying to link the data to anxiety. Therefore it will be more proper to interpret the results as modulation of anxiety-like behavior (anxiety related avoidance) but not anxiety itself. We have modified the manuscript to describe more precise to avoid overstatement. Our fiber photometry data (Fig. 5) show that vSub–SuM projection neurons increase activity specifically when mice enter open arms of the EZM—a behavioral transition associated with anxiety—whereas dSub–SuM projections do not. This activity correlates with anxiety-related behavior, not merely with movement or stress per se. We also agree that the term “engram” may be misleading in this context. In the manuscript, we refer to SANs as a “stress-activated neuronal ensemble” rather than an anxiety engram. Our data indicate that these neurons are recruited by stress and their reactivation produces more anxiety related avoidance to open arms. We have revised the text to avoid conceptual overreach and to clarify that SuM SANs likely contribute to a state of sustained anxiety/avoidance. Recommendations for the authors: Reviewing Editor Comments: Should you choose to revise your manuscript, if you have not already done so, please include full statistical reporting, including exact p-values wherever possible alongside the summary statistics (test statistic and df) and, where appropriate, 95% confidence intervals. These should be reported for all key questions and not only when the p-value is less than 0.05 in the main manuscript. Readers would also benefit from noting that the subjects were male in the abstract and discussion of the limitations of the exclusion of females. Thank you for the suggestion. We have included the full statistical detail in a separate sheet as Table 1. Also, we have modified the title of the manuscript to reflect the sex of the mice. Reviewer #1 (Recommendations for the authors): (1) In line 211, the authors state, "we recorded neuronal action potentials via multichannel extracellular recording while the mice were moving in the EPM, a traditional type of maze used to test anxiety in rodents,". However, it is unclear what data is presented in the paper, that is, extracellular recordings from SuM in mice on the elevated plus maze. We have deleted the description of multichannel recording data in EPM as the data was removed earlier. Minor corrections to the text and figures. (2) For bar plots, perhaps clarify how the data is presented. For example, in Figure 4, "The data in B, D, E and I-L are presented as the means {plus minus} SEMs," but this does not appear to be plotted as a mean with SEM error bars because the error bars cover all the values. Corrected. (3) In Figure 5, the white text for EGFP in panel B is very difficult to see. Corrected. (4) For Figure 5D, it would be helpful to more clearly specify which neurons in SuM were recorded from. Was it SANs or all SuM neurons? We did whole-cell recording on all SuM neurons. (5) Fos2A-iCreERT2 is mislabeled as "Fos2A-iCreERT" in the methods. Corrected. (6) The sentence at line 139 "To make sure foot shock induced anxiety won't last until manipulation, we subjected139mice to an acute stress protocol involving foot shocks and then performed the elevated plus140maze (EPM) and elevated zero maze (EZM) tests to evaluate anxiety on days 2 and 7," is unclear as written. Thank you for pointing this. We have modified the sentence to make it more clear. “To make sure mice are on similar basal condition while applying chemo-genetic manipulation, we subjected mice to an acute stress protocol involving foot shocks and then performed the elevated plus maze (EPM) and elevated zero maze (EZM) tests to evaluate anxiety on days 2 and 7 (Figure 4 A). The mice that experienced foot shocks showed decreases in the exploration time in the open arms on day 2. However, acute stress-induced anxiety was not detected on day 7 (Figure 4 B), which allow us to compare the reactivation of SANs produced anxiety-like behavior between groups at the same baseline.” (7) The details of the viral injections used for ex vivo electrophysiology are not sufficient to understand the experiment and the implications of the data. Which neurons (SANs?) are recorded from, what percent of those had inputs, were the sub-neurons globally labeled or just SANs? We performed whole-cell recording on global SuM neurons to show if the projection is innervated by glutamergic neurons in Sub as shown in Figure 5-B that the projection neurons in Sub are exclusively vglut1 expressed. Based on this aim of the experiment, we didn’t keep any neurons that were not response to the light stimulation, therefore can’t calculate the input percent in this case. We have added words to clearly show that we did global SuM neurons in Methods. (8) The scale used in Figure 6C renders that data unreadable. 120 to 40% changes in body weight are well beyond the variability in the data. We have modified the axis (90 to 110%) to show the body weight change clearer. (9) The dose of CNO used, 5 mg/kg, is high, and using lower doses or other DREADD ligands is worth considering. Thank you for your valuable comment. We have noticed that people are using relatively lower dose of CNO or other DREADD ligands that are reported much higher affinity and less side-effect. The dose of 5mg/kg was adapted from earlier papers that using DREADD and show no obvious side-effect in mice[17], e.g locomotion (S Figure 2B), in our experiments, so we keep using this dose in this project to make it consistent across different cohorts of experiments. We are switching to DCZ to avoid any potential side-effect of CNO in the following experiments based on this project. Reviewer #2 (Recommendations for the authors): This is a strong manuscript that provides important insights into the role of the supramammillary nucleus (SuM) and its inputs from the ventral subiculum in regulating anxiety. The combination of behavioral, imaging, electrophysiological, and circuit manipulation approaches is impressive, and the distinction the authors propose between anxiety-related and fear-related circuits is conceptually important. There are, however, some points that I think need clarification. The authors emphasize that the hippocampus is essential for fear memory recall, yet they do not directly evaluate whether the SuM-hippocampal pathway might contribute differentially to anxiety versus fear memory. Addressing this would help to explain where the dissociation between the two processes arises. Thank you for the suggestion. We realized that we didn’t collect enough data to exclude the role of those SANs on memory, especially fear memory, a memory formation bases on strong emotional training as aforementioned. The data and relevant discussion have been removed to avoid misunderstanding and overstatement. I am also not fully convinced about the definition of the "stress-activated neurons" (SANs). The overlap across repeated stress exposures is quite modest (around 20%), which suggests that this population may not be strictly stress-specific but rather a dynamic subset that is preferentially, though not exclusively, engaged by stress. Related to this, the use of the term "engram" raises conceptual questions. Since the classic engram refers to an ensemble encoding and recalling a specific memory, it is not obvious whether it is appropriate to apply the term to a neuronal population that appears to represent a persistent emotional state. The authors should consider justifying this choice of terminology more carefully or adopting a different term. Thank you for your important comments. Yes we agree that the SANs in this manuscript are more likely dynamic subset other than exclusive foot-stress engaged “engram”. That’s why we use “stress-activated neurons” but not “engram” to describe this neuronal ensemble. To avoid further misleading, we have made some modification to reduce the use of “engram” across the manuscript. Some parts of the text also need more precision. For example, the statement in lines 63-65 that "few studies have explored emotion-related engram cells" is potentially misleading, as most engram studies focus on memories with a strong emotional component. The rationale for this claim should be clarified. This sentence has been deleted since it is not necessary to link the text and misleading. In Figure 1, the choice of methods is also puzzling: cFos immunostaining is used after shock delivery, while electrophysiology is used for the CSDS paradigm. It would be helpful to explain why different readouts were chosen for different stress models, and whether this may affect the comparability of the results. Thank you for this important comment. In Figure 1, we aimed to demonstrate that both acute (foot shock) and chronic (CSDS) stressors can activate SuM neurons, using complementary methods (cFos for acute, in vivo recording for chronic). The reason we chose different method is that acute stress produces transit effect while chronic stress produces long-lasting effect. To our knowledge, cFos is a well-established marker for strong neuronal activation, but with short lifespan (~4-6 hours) and suits acute paradigm better. In vivo recording allows us to compare the neuronal activity before and after chronic experiments within subjects and has ability to reveal cumulative effect which cFos cannot. To address this, we have clarified in the text that the purpose of Figure 1 in Line 112-113: “To investigate if SuM would be responsive to diverse stressors, we next examined whether chronic stress, which different mechanism underlying…” Finally, some additional details would strengthen the presentation. The discussion of corticosterone and other physiological markers could be expanded to indicate whether these effects were robust across stress paradigms. Similarly, the relatively modest overlap between SANs activated by different stressors could be framed more explicitly as part of a broader principle of flexible ensemble recruitment in anxiety-related circuits. Thank you for your suggestion. We have added more discussion about the corticosterone and the flexibility of SANs in the manuscript. See Line 267-270: “The serum corticosterone concentration can be used as a marker of stress-induced change in the peripheral blood. Previous studies showed serum corticosterone can be increased by various stress stimulation [39–42]; meanwhile, intentionally supplementing the diet with corticosterone can induce anxiety-like behaviors in rodents[43].” and Line 275-281: “However, the reactivation rate of SANs caused by different stressor was relatively lower than the initial activation rate caused by foot shock (Figure 3). This suggests that stress-activated neuronal clusters may have more flexible recruitment principles, with only a small number of neurons potentially encoding emotional information, while most other neurons remain involved in encoding other neural activities. Studies in other field, particularly studies of memory engram, has shown that the sets of neurons activated during learning are dynamic and exhibit high flexibility [44, 45].” Overall, the work is of high quality and provides a valuable contribution to the field, but addressing these points would help sharpen the mechanistic claims and ensure that the conceptual framework is as clear and precise as the experimental data. Reviewer #3 (Recommendations for the authors): (1) Since increased SuM activity is hypothesized to mediate the effects of stress on anxiety-like behavior, a logical step would be to test for necessity by silencing the stress-activated SuM cells. We agree this is a logical and valuable experiment. While our current study focused primarily on the sufficiency of SuM/SAN activation to induce anxiety-like behavior, we acknowledge that inhibition experiments would provide critical complementary evidence for necessity. We have added a statement in the Discussion noting that “future studies should examine whether silencing SuM SANs, either during stress exposure or during anxiety testing, can prevent or reduce stress-induced anxiety”. This will help establish a more complete causal role. (2) Discuss what is meant by "anxiety engram" and what features of anxiety the labeled cells might encode. We concur that “stress-activated neuron (SAN)” is a more precise descriptor than “engram” in this context. We have revised the text to avoid the potentially misleading term “engram” and instead refer to a “stress-activated neuron”. The labeled cells are preferentially reactivated by stress (not reward), and their activation promotes both behavioral avoidance and physiological stress markers (corticosterone). They likely contribute to the maintenance of an anxious state under perceived threat, rather than encoding discrete threat cues or memories. (3) A more nuanced analysis of behavioral correlates of SuM activity and/or the behavioral effects of SuM manipulations would strengthen this paper. To provide a more nuanced understanding of the behavioral correlates, we have performed additional analyses on our fiber photometry data (now presented in Supplemental Figure 6). and have also planned additional experiments for the future study to deepen our understanding. References: (1) Jendryka M, Palchaudhuri M, Ursu D, van der Veen B, Liss B, Kätzel D, et al. Pharmacokinetic and pharmacodynamic actions of clozapine-N-oxide, clozapine, and compound 21 in DREADD-based chemogenetics in mice. Sci Rep. 2019;9. (2) Koike H, Demars MP, Short JA, Nabel EM, Akbarian S, Baxter MG, et al. Chemogenetic Inactivation of Dorsal Anterior Cingulate Cortex Neurons Disrupts Attentional Behavior in Mouse. Neuropsychopharmacology. 2016;41:1014–1023. (3) Guettier J-M, Gautam D, Scarselli M, Ruiz De Azua I, Li JH, Rosemond E, et al. A chemical-genetic approach to study G protein regulation of cell function in vivo. Proceedings of the National Academy of Sciences. 2009;106:19197–19202. (4) Wess J, Nakajima K, Jain S. Novel designer receptors to probe GPCR signaling and physiology. Trends Pharmacol Sci. 2013;34:385–392. (5) Kraeuter AK, Guest PC, Sarnyai Z. The Elevated Plus Maze Test for Measuring Anxiety-Like Behavior in Rodents. Methods in Molecular Biology, vol. 1916, Humana Press Inc.; 2019. p. 69–74. (6) Kraeuter AK, Guest PC, Sarnyai Z. The Open Field Test for Measuring Locomotor Activity and Anxiety-Like Behavior. Methods in Molecular Biology, vol. 1916, Humana Press Inc.; 2019. p. 99–103. (7) Wall PM, Messier C. Methodological and conceptual issues in the use of the elevated plus-maze as a psychological measurement instrument of animal anxiety-like behavior. Neurosci Biobehav Rev. 2001;25:275–286. (8) Carobrez AP, Bertoglio LJ. Ethological and temporal analyses of anxiety-like behavior: The elevated plus-maze model 20 years on. Neurosci Biobehav Rev. 2005;29:1193–1205. (9) Seibenhener ML, Wooten MC. Use of the open field maze to measure locomotor and anxiety-like behavior in mice. Journal of Visualized Experiments. 2015. 6 February 2015. https://doi.org/10.3791/52434 . (10) Prut L, Belzung C. The open field as a paradigm to measure the effects of drugs on anxiety-like behaviors: A review. Eur J Pharmacol. 2003;463:3–33. (11) Chen Y, Zhou X, Chu B, Xie Q, Liu Z, Luo D, et al. Restraint Stress, Foot Shock and Corticosterone Differentially Alter Autophagy in the Rat Hippocampus, Basolateral Amygdala and Prefrontal Cortex. Neurochem Res. 2024;49:492–506. (12) Hassell JE, Nguyen KT, Gates CA, Lowry CA. The Impact of Stressor Exposure and Glucocorticoids on Anxiety and Fear. Curr. Top. Behav. Neurosci., vol. 43, Springer; 2019. p. 271–321. (13) Peng B, Xu Q, Liu J, Guo S, Borgland SL, Liu S. Corticosterone attenuates reward-seeking behavior and increases anxiety via D2 receptor signaling in ventral tegmental area dopamine neurons. Journal of Neuroscience. 2021;41:1566–1581. (14) Myers B, Greenwood-Van Meerveld B. Elevated corticosterone in the amygdala leads to persistant increases in anxiety-like behavior and pain sensitivity. Behavioural Brain Research. 2010;214:465–469. (15) Demuyser T, Deneyer L, Bentea E, Albertini G, Van Liefferinge J, Merckx E, et al. In-depth behavioral characterization of the corticosterone mouse model and the critical involvement of housing conditions. Physiol Behav. 2016;156:199–207. (16) Shoji H, Maeda Y, Miyakawa T. Chronic corticosterone exposure causes anxiety- and depression-related behaviors with altered gut microbial and brain metabolomic profiles in adult male C57BL/6J mice. Molecular Brain . 2024;17. (17) Manvich DF, Webster KA, Foster SL, Farrell MS, Ritchie JC, Porter JH, et al. The DREADD agonist clozapine N-oxide (CNO) is reverse-metabolized to clozapine and produces clozapine-like interoceptive stimulus effects in rats and mice. Sci Rep. 2018;8. Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Figure 1—source data 1. The number of c-Fos+ cells in panel C and normalized firing rate in panel F, and corresponding statistical results. elife-108593-fig1-data1.xlsx (14.7KB, xlsx) Figure 2—source data 1. The number of c-Fos+ cells in panel D, and behavioral test data in panel E-H, and corresponding statistical results. elife-108593-fig2-data1.xlsx (13.1KB, xlsx) Figure 3—source data 1. The number of TRAPed cells in panel C and panel F, c-Fos+ cells in panel G and the reactivation ratio in panel H, and corresponding statistical results. elife-108593-fig3-data1.xlsx (11.7KB, xlsx) Figure 4—source data 1. The behavioral test data in panel B and I-L, corticosterone level in panel D, co-stain ratio in panel F and corresponding statistical results. elife-108593-fig4-data1.xlsx (13.1KB, xlsx) Figure 5—source data 1. The Ca 2+ peak of ΔF/F in panel L and Q, the area under curve of the Ca 2+ trace in panel M and R, and corresponding statistical results. elife-108593-fig5-data1.xlsx (12.3KB, xlsx) Figure 6—source data 1. The body weight in panel C, and behavioral test data in panel E-G, and corresponding statistical results. elife-108593-fig6-data1.xlsx (15.9KB, xlsx) MDAR checklist elife-108593-mdarchecklist1.docx (88.8KB, docx) Data Availability Statement All data generated or analyzed during this study are included in the manuscript and source data files. Customized code can be accessed through GitHub ( https://github.com/zjm199502/elife2026 , copy archived at Zhang, 2026 ). Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd ACTIONS View on publisher site PDF (6.2 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• State and trait anxiety modulate hemodynamic amygdala responses to innocuous auditory stimulation#442448
• Impact of sex chromosomes and gonad type in stress susceptibility in corticostriatal brain regions.#255172
• Gendering the Menstrual Cycle in Behavioral Neuroendocrinology#125790
• Gendering the Menstrual Cycle in Behavioral Neuroendocrinology.#125272
• Stress and alcohol impact network states involved in emotional processing: relevance of comorbid AUD and psychiatric illnesses.#269491