A thalamo-hippocampal circuit regulating memory precision during contextual learning

A thalamo-hippocampal circuit regulating memory precision during contextual learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article A thalamo-hippocampal circuit regulating memory precision during contextual learning Chun Xu, Qingge Wu, Yumian Li, Shishuo Chen, Zhou Sun, Bingqing Zhao, and 12 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7828569/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract The balance between memory specificity and generalization is essential for animals’ adaptive behaviours in diverse contexts, and is controlled by thalamus and hippocampus. Yet, whether and how thalamo-hippocampal circuitry regulates this balance remains elusive. Using single-neuron projectome analysis, we found that dorsal and ventral subregions of ventral CA1 (vCA1d and vCA1v) received inputs from two distinct neuronal populations in the nucleus of reuniens (NRe vCA1d and NRe vCA1v neurons), respectively. Cell type-specific trans-synaptic retrograde tracing and circuit-barcoded single-cell RNA sequencing of NRe vCA1d and NRe vCA1v neurons uncovered distinct presynaptic inputs as well as gene expression patterns underlying their different intrinsic excitabilities and spike waveforms. Using single-unit recording, circuit specific chemogenetic inhibitions, and miniscope Ca 2+ imaging, we found that NRe vCA1d (but not NRe vCA1v ) neurons and downstream vCA1d neurons exhibited reduced contextual discriminability, promoting memory generalization after threat learning. These results demonstrate that the NRe-vCA1d circuit regulates memory precision by promoting memory generalization over specificity. Biological sciences/Neuroscience/Learning and memory/Hippocampus Biological sciences/Molecular biology/Transcriptomics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Introduction Memory enables animals to express adaptive behaviors in a volatile world. Generalized threat memories can protect animals from potential dangers by responding appropriately to a novel environment that resembles a previously harmful experience, but overgeneralized threat memories may lead to maladaptive behaviors and have been linked to anxiety- and stress-related disorders, as evidenced by post-traumatic stress disorders (PTSD) in humans 1 – 6 and PTSD-like animal models 7 , 8 . Memory precision control is thus essential for the balance between memory specificity and generalization 9 , thereby supporting animals’ survival and well-being. The hippocampus (HIP) and medial prefrontal cortex (mPFC) are crucial for memory precision control 9 – 16 . Recent studies showed that NRe, a thalamic nucleus and key bridge linking HIP and mPFC 17 – 20 , played key roles in memory precision control 9 , 21 , in addition to a variety of memory-related processes including working memory 22 , spatial memory and navigation 23 – 28 , memory consolidation 29 , and threat memory extinction 30 – 32 . Such functional diversity of NRe neurons may have to do with their diverse circuit connectivity 17 and cellular makers 33 and firing properties 34 , requiring a systematic understanding of brain-wide connectivity of NRe neurons at the single-neuron level. Prior studies suggest that NRe neurons control contextual memory precision 11 , possibly by regulating neuronal activity in HIP 21 . Yet, it remains elusive how NRe neurons regulate neural activity in HIP during memory precision control. Here we leveraged single-neuron projectome reconstruction and analysis to delineate brain-wide axon projections of NRe neurons and identified two subpopulations of NRe neurons preferentially projecting to vCA1d and vCA1v, respectively. By combining retrograde tracing with chemogenetic manipulations, miniscope Ca 2+ imaging, in vitro slice recording, in vivo single-unit recording and cell-type specific rabies transsynaptic tracing, we found systematic distinctions between NRe vCA1d and NRe vCA1v neurons at molecular, cellular, circuit, and behavioral levels. We further showed that NRe vCA1d neurons promoted threat generalization, via the downregulation of contextual discriminability of both themselves and their downstream vCA1d neurons following contextual threat learning, revealing a regulatory role of a thalamo-hippocampal circuit in memory precision control. Results Diversity of whole-brain axon projections from NRe neurons To understand the diversity and spatial organization of axon projections from NRe neurons, we sparsely labeled NRe neurons by a mixture of adeno-associated virus (AAV)–expressing Cre-dependent enhanced green fluorescent protein (EGFP) and low-titer AAV–expressing Cre under a human synapsin promoter 35 , 36 , and then imaged their whole-brain axon projections by fluorescence Micro-Optical Sectioning Tomography 35 – 38 (Fig. 1 A, see Methods). We reconstructed axon projections and soma of 303 NRe neurons from 8 mice 39 , and identified 10 downstream target areas, including striatum (STR), olfactory area (OLF), thalamus (TH), hypothalamus (HY), cortical subplate-related areas (CTXsp), palladium (PAL), isocortex (ISO), entorhinal cortex (ENT), and HIP (Fig. 1 B). All neurons were registered to the Allen Mouse Brain Common Coordinate Framework 40 , based on the propidium iodide (PI) staining 35 , 36 , 38 . To classify projectome cell types of NRe neurons, we first digitalized the axon routes of all neurons and reconstructed each neuron and projection pattern by axon route clusters ( Figure S1A-C , see Methods), and then performed unsupervised hierarchical clustering of projection patterns based on their axon route clusters ( Figure S1D-F , see Methods). In total, 6 projectome cell types were classified, and further grouped into 2 major classes: 4 projectome types in the first class exhibited strong axonal projections to ISO, ENT and HIP with different axon routes and strengths; 2 projectome types in the second class projected preferentially to ISO and STR, respectively (Fig. 1 B and 1 C). Moreover, projectome cell type 1 sent axon projections to HIP and ISO with similar strength. Types 3 and 4 sent preferential axon projections to HIP whereas type 5 preferentially projected to ISO, including anterior cingulate cortex (ACC) and mPFC (Fig. 1 B). The NRe is believed to be the key bridge to connect with HIP and medial prefrontal cortex (mPFC), a part of ISO 41 . Interestingly, projectome cell type 1–4 projected to HIP and ISO with differential strengths. To determine whether single NRe neurons send differential projections to mPFC and HIP as well as other targets. We first investigated this by measuring the correlation coefficients and selectivity index for the axon projection strength in target areas. We identified 10 common projection patterns of NRe neurons based on the unsupervised clustering analysis of the correlation coefficients (Fig. 1 D). The HIP subregions such as dentate gyrus (DG) and CA1 belonged to a target pattern of hippocampal formation subregions (pattern 2), whereas mPFC subregions belonged to other patterns: infralimbic cortex (ILA) in pattern 4 (together with retrosplenial cortex and anterior cingulate cortex) and prelimbic cortex (PL) in pattern 8 (together with olfactory bulb). We then measured the selectivity index for individual NRe neurons between mPFC and HIP, and found that many NRe neurons showed high selectivity for either mPFC or HIP (e.g., index − 1 is exclusively for mPFC, Fig. 1 E, lower panel). In contrast, NRe neurons projected to ENT and HIP with balanced strengths (i.e., selectivity index near 0, Fig. 1 E, middle panel). Thus, NRe neurons could be grouped by their axon projection strengths to brain-wide target areas, as manifested by the preferential axon projections to mPFC and HIP, respectively. Distinct projections from NRe and NRe neurons to CA1 subregions Prior studies showed that NRe neurons modulated contextual memory generalization and cFos signals in HIP 9 . To understand axon projections from NRe neurons to HIP subregions, we performed unsupervised clustering analysis of the projection strength in the HIP subregions including CA1, CA2, CA3, DG, and subicular complex. We identified 6 subgroups of 161 HIP-projecting NRe neurons with distinct target patterns: subgroup 2, 4 and 6 co-projecting to multiple subregions, whereas subgroup 1,3 and 5 selectively projecting to HIP subregions of DG, CA1 and subicular complex, respectively (Fig. 2 A). Of note, NRe neurons in all 6 subgroups sent collateral projections to ENT (Fig. 2 B), consistent with above results that most NRe neurons sent comparable axon projections to both HIP and ENT. Among all the HIP subregions, NRe neurons sent the strongest projections to CA1 (Fig. 2 C). Interestingly, axon arbors in CA1 were enriched in two subregions of ventral CA1 (vCA1) centered around 4 mm and 7 mm along the dorsal-ventral (D-V) axis, respectively (Fig. 2 C), raising the possibility of separable axon projections into CA1 subregions. Next, we selected the HIP-projecting NRe neurons and performed K-means clustering of principal components based on the soma and axon terminals’ center of mass in CA1. Three clusters of NRe neurons were classified by distinct spatial preferences of axon terminals but not soma location (Fig. 2 D and 2 E), and were preferentially enriched in projectome cell type 3 and 4 (Fig. 2 E, lower panel). Interestingly, their axon arbors were spatially enriched in the stratum molecular-lacunosum of CA1 (Fig. 2 F). Notably, the axon projection of cluster1 and 2 (preferring vCA1) were much stronger than that of cluster 3 (preferring dorsal CA1), as reflected by the quantity of cells and axon tips (Fig. 2 E, upper panel). Furthermore, axon arbors of cluster 1 were primarily located in vCA1d, whereas those of cluster 2 mainly in vCA1v (Fig. 2 F and 2 G). These two clusters of NRe neurons sent axon collaterals to HY, ISO, and TH with different projection strengths but to ENT and OLF with similar strengths (Fig. 2 H), revealing a systematic difference in axon projections between vCA1d-projecting and vCA1v-projecting (NRe vCA1d and NRe vCA1v ) neurons. Consistently, dual-color fluorescent retrograde tracing from vCA1d and vCA1v labeled largely separate subpopulations of NRe neurons (Fig. 2 I- 2 K). Thus, the dissociable axon projections from NRe neurons, particularly NRe vCA1d and NRe vCA1v neurons, could provide the circuit basis for specific memory functions. Differential regulations of contextual memory by NRe and NRe neurons To probe the behavioral function of NRe vCA1d and NRe vCA1v neurons in memory precision control, we performed selective chemogenetic inhibition of them when animals were subjected to a contextual threat conditioning (CTC) paradigm (Fig. 3 A). We first validated the in vivo chemogenetic inhibition 42 by tetrodes-based single-unit recording from NRe neurons after injecting AAV-hSyn-hM4Di-mCherry into NRe ( Figure S2A-C ). Upon the administration of clozapine N-oxide (CNO), a significantly higher proportion of NRe neurons in the CNO group (~ 60%) exhibited prolonged (at least one hour) inhibition compared to those in the saline group (CNO 57/98 vs. saline 40/95; Chi-square test, p = 0.026; Figure S2D-F ), demonstrating a mild and long lasting chemogenetic inhibition of NRe’s neuronal activity. We went on to express hM4Di-mCherry (hM4Di group) or mCherry (control group) specifically in NRe vCA1d and NRe vCA1v neurons by injecting Cre-dependent hM4Di into NRe and AAVretro-Cre into vCA1d and vCA1v, respectively (Fig. 3 B and 3 D, Figure S3 ). Mice were administered with CNO 30 min before the CTC in context A in day 1 (see Methods) and were subjected to contextual threat retrieval in context B (neutral context) and context A (conditioned context) in day 2 and day 3, respectively. For animals with NRe vCA1d neurons labeled, animals in the hM4Di group showed similar freezing in context A, but significantly lower freezing in context B, compared to those in the control group (Fig. 3 C). Consequently, the generalization index (ratio of freezing in B to freezing in A) was significantly reduced in the hM4Di group, suggesting that NRe vCA1d neurons facilitate the contextual memory generalization. For animals with NRe vCA1v neurons labeled, animals in the hM4Di group showed significantly lower freezing in context A but similar generalization index, compared to those in the control group (Fig. 3 E). These results suggest that NRe vCA1d neurons are promoting contextual memory generalization whereas NRe vCA1v neurons are critical for the strength of associative memory formation. Finally, we inquired whether two subpopulations of NRe neurons contribute to the contextual memory retrieval. After two weeks of rest, these animals were subjected to CTC again in context A in day 1 and then were administered with CNO 30 min before the memory retrieval in contexts B and A in day 2 and day 3, respectively ( Figure S2G-2K ). For animals with NRe vCA1d neurons labeled, context B evoked lower freezing in hM4Di group than in control group, leading to a significant reduction of generalization index ( Figure S2I ). For animals with NRe vCA1v neurons labeled, contexts A and B both evoked similar freezing in hM4Di group and control group ( Figure S2K ). Taken together, these results suggest that NRe vCA1v neurons regulate the strength of context-dependent associative memory formation, whereas NRe vCA1d neurons control the contextual memory precision during both memory formation and retrieval. Molecular, cellular and circuit differences between NRe and NRe neurons To compare NRe vCA1d and NRe vCA1v neurons at molecular, cellular and circuit levels, we employed multiple approaches to compare their gene expression patterns, electrophysiological characteristics, and presynaptic connections. To compare the transcriptomic profiles of NRe vCA1d and NRe vCA1v neurons, we performed barcoded scRNA-seq 43 of NRe neurons by injecting two distinct barcoded retrograde AAV-GFP into vCA1d and vCA1v, respectively (Fig. 4 A). After 6 weeks, we collected 37894 cells from NRe-containing tissues (N = 3 mice) and prepared single-cell transcriptional libraries and barcode expression libraries for scRNA-seq afterwards (see Methods). Then we performed unsupervised clustering analysis and annotated the class of NRe neurons (n = 15495 cells) based on 3 markers ( Calb2 , Stmn1 and Gap43 ) that we identified from a spatial transcriptome database 44 and a previously reported markers Nox4 45 ( Figure S4A-G ). Further clustering analysis revealed 7 clusters of NRe neurons (Fig. 4 B and 4 C). We found that vCA1-projecting cells (EGFP positive cells) were significantly enriched in cluster 1 (23.6% vs. 6.5% in other clusters, Fisher test, p < 0.001, Fig. 4 D and 4 E). As the EGFP expression and barcode number were highly correlated ( Figure S5A ), we utilized the barcode number in each cell as a proxy for the axon projection strength and classified NRe vCA1d and NRe vCA1v neurons in the scRNA-seq data based on their barcode preferences ( Figure S5B and S 5C ). We found that NRe vCA1d neurons were preferentially enriched in cluster 3 (Fig. 4 E, lower panel). Analyses of differentially expressed gene (DEG) and gene ontology (GO) revealed distinct scores for many pathways including monoatomic ion transport, fear response and response to stimulus ( Fig. S5D and S 5E ). Further analysis showed that NRe vCA1d neurons, compared to NRe vCA1v neurons, exhibited lower expressions of Na + channel subunits but higher expression levels of K + channel subunits (Fig. 4 F). Thus, NRe vCA1d and NRe vCA1v neurons may show different depolarization and hyperpolarization components of the spike waveform. To test this possibility and further compare electrophysiological characteristics of NRe vCA1d and NRe vCA1v neurons, we performed whole-cell recordings from NRe vCA1d and NRe vCA1v neurons in acute brain slices prepared from animals with red retrobeads locally injected into vCA1d and vCA1v, respectively (Figs. 5 A and 5 B). We measured firing frequency-versus-current curves as a proxy of neuronal excitability, and found that membrane excitability of NRe vCA1d neurons was lower than that of NRe vCA1v neurons which may result from lower expressions of Na + channel in NRe vCA1d neurons (Figs. 5 C-G). Interestingly, the spike waveform of NRe vCA1d neurons was significantly different from that of NRe vCA1v neurons, as reflected by the half-width ratio of hyperpolarization to depolarization mediated by K + and Na + channels (Figs. 5 H). These results suggest that NRe vCA1d and NRe vCA1v neurons exhibit distinct intrinsic electrophysiological properties including the membrane excitability and spike waveform. To validate the electrophysiological heterogeneity of NRe neurons in vivo , we performed tetrode-based 32-channel recordings from NRe neurons in behaving animals (Fig. 5 I). Single units (n = 92) were sorted based on feature spaces of the spike waveform such as peak-valley amplitudes (Fig. 5 J and 5 K). We measured 4 features of the spike waveform and classified NRe neurons into two clusters (C1 and C2) using t-SNE dimension reduction and DBSCAN clustering 46 (Figs. 5 L and 5 M). Interestingly, C1 neurons exhibited significantly higher half-width ratio of hyperpolarization to depolarization than C2 neurons (Figs. 5 N and 5 O), consistent with the difference between NRe vCA1d and NRe vCA1v neurons during slice recording (Fig. 5 H). Thus, C1 and C2 neurons were defined as putative NRe vCA1d and NRe vCA1v neurons, respectively. Notably, the C1 (NRe vCA1d ) neurons exhibited higher frequencies of spontaneous firing activity in the range above 4 Hz than C2 (NRe vCA1v ) neurons, whereas they exhibited comparable bursting index and durations (Figs. 5 P-R), suggesting that NRe vCA1d neurons are likely to operate in a phasic mode that has been shown to promote memory generalization 11 . Finally, we profiled presynaptic inputs of NRe vCA1d and NRe vCA1v neurons using a cell-type specific rabies retrograde tracing strategy. We injected Cre-dependent rabies helper AAV into NRe and AAVretro-hSyn-Cre into vCA1d and vCA1v, respectively, followed by the injection of rabies vectors into NRe 21 days later (Fig. 6 A). The starter cells for retrograde transsynaptic tracing were identified by co-labeling of rabies-GFP (green) and TVA-mCherry (red) in NRe neurons (Fig. 6 B). We quantified the rabies-labeled cells in the whole brain after retrograde transsynaptic tracing from NRe vCA1d neurons (N = 4 mice) and NRe vCA1v neurons (N = 4 mice). Analysis showed that NRe vCA1d neurons received more inputs from nucleus of the lateral olfactory tract (NLOT2), infralimbic area (ILA), agranular insular area (AI), paraventricular nucleus of the thalamus (PVT), zona incerta (ZI), posterior hypothalamic nucleus (PH), superior colliculus motor related deep gray layer (SCdg), and medial pretectal area (MPT), whereas NRe vCA1v neurons received more inputs from nucleus accumbens (ACB), caudoputamen (CP), diagonal band nucleus (NDB), bed nuclei of the stria terminalis (BST), substantia innominate (SI), ventrolateral preoptic nucleus (VLPO), periventricular hypothalamic nucleus posterior part (PVp) (Fig. 6 C and 6 D). Thus, the distinct presynaptic inputs onto NRe vCA1d and NRe vCA1v neurons could provide the circuit bases for their distinct memory functions. Taken together, these results have revealed molecular, cellular and circuit bases that may underlie distinct memory functions of NRe vCA1d and NRe vCA1v neurons. Learning decreased contextual discrimination of putative NRe neurons The fact that NRe vCA1d neurons promoted contextual threat generalization after threat learning prompted us to examine the contextual discrimination of NRe vCA1d neurons themselves. To examine the contextual discrimination of putative NRe vCA1d neurons and their neural plasticity, we recorded context-evoked neuronal activity of NRe neurons from animals subjected to the CTC paradigm in which context A and C exhibited a greater difference than A and B (Fig. 7 A). After CTC, animals exhibited generalized and lower freezing in neutral context B and C compared to that in conditioned context A (Fig. 7 B). Using the same unsupervised clustering method as Fig. 5 N, NRe neurons were grouped into two clusters (Fig. 7 C). The putative NRe vCA1d (n = 33) and NRe vCA1v neurons (n = 57) were identified by the differences in the half-width ratio of hyperpolarization to depolarization (Fig. 7 D). At the single-cell level, we measured the discrimination index (DI) for each cell and found insignificant changes in the discrimination index (Fig. 7 E). At the population level, CTC significantly reduced the Mahalanobis distance in the coding space between conditioned and neutral contexts in putative NRe vCA1d neurons, but not in putative NRe vCA1v neurons (Fig. 7 G). Thus, NRe vCA1d neurons may promote threat generalization because of the learning-downregulated contextual discrimination. NRe vCA1d neurons suppressed the contextual discrimination of vCA1d neurons How do NRe vCA1d neurons promote threat generalization? A plausible explanation lies in their regulation of neural activity in downstream areas, particularly the vCA1d. We thus investigated how NRe vCA1d neurons regulate the context-evoked activity of vCA1d neurons by combining deep-brain Ca 2+ imaging of vCA1d neurons with chemogenetic inhibition of NRe vCA1d neurons in behaving animals with head-mounted miniscope 47 , 48 . We first expressed CaMKII-dependent GCaMP6f 49 in pyramidal cells in vCA1d and Cre-dependent hM4Di in NRe vCA1d neurons using the same retrograde strategy as chemogenetic behavioral experiments described above (Fig. 8 A and B ). Then we implanted a gradient-index (GRIN) lens above vCA1d to monitor the Ca 2+ activity of vCA1d neurons during a 3-day behavioral paradigm for contextual memory generalization. To evaluate the contextual discrimination of vCA1d neurons before and after CTC, animals were exposed to context A/B/C, in which context A was conditioning context and context B and C were neutral contexts with different similarities compared to context A (Fig. 8 C). Two groups of animals were subjected to the CTC and were administered with saline or CNO 30 min before the CTC in day 2, respectively. Animals in both saline and CNO groups showed comparable freezing to conditioned context A. The saline (control) group exhibited highly generalized freezing in B and some in C, whereas the CNO group showed lower generalized freezing in B and very little in C, consistent with the notion that NRe neurons promote contextual memory generalization (Fig. 8 D). We identified the context-preferring neurons by a bootstrap method comparing their context-evoked Ca 2+ activity between A and B, or between A and C (see Methods). These context discriminative neurons, including non-context preferring neurons, were spatially intermingled (Fig. 8 E and 8 F). Overall, the proportions of context discriminative neurons for A vs. B and A vs. C were slightly but significantly decreased after CTC in the saline group (Fig. 8 G and 8 H, upper panels). In the CNO group, the proportion of context discriminative neurons for A vs. B was maintained and that for A vs. C was even increased (Fig. 8 G and 8 H, upper panels). Consistently, the DI for A vs. B and A vs. C was largely decreased after CTC in saline group, whereas the DI for A vs. B was only slightly decreased and the DI for A vs. C was even increased after CTC in the CNO group (Fig. 8 G and 8 H, lower panels). Furthermore, we took a finer approach to analyze the temporal dynamics of context-evoked Ca 2+ activity by correlation analysis 50 and determine the impact of NRe vCA1d neurons on the context discrimination of vCA1d neurons (Fig. 8 I, low correlation represents good discrimination). We found a robust increase of the correlation in context-evoked Ca 2+ signals after CTC in saline group, but a dramatic decrease of the correlation in CNO group (Fig. 8 J). These results demonstrate that NRe vCA1d neurons impose an inhibitory effect on the contextual discrimination of individual vCA1d neurons. Finally, we performed population analysis to evaluate the context representation by vCA1d neurons in the saline and CNO groups. As exemplified in Fig. 9 A, the principal component analysis (PCA) showed that neural representations of three contexts in the coding space shrank such that their Mahalanobis distances became smaller after CTC in the saline group (Fig. 9 B and 9 C), which correlated with different levels of fear generalization in contexts B and C (Fig. 8 D). In contrast, as exemplified in Fig. 9 D, the neural representations of three contexts expanded in the coding space such that their Mahalanobis distances became larger after CTC in the CNO group (Fig. 9 E and 9 F). In summary, chemogenetic inhibition of NRe vCA1d neurons in the CNO group resulted in an enlarged distance between conditioned and neutral contexts in the coding space (Fig. 9 G and 9 H), providing a neural mechanism that NRe vCA1d neurons promote contextual memory generalization at the population level. This notion was further supported by a decoding analysis using a random forest classifier 51 . The context decoding accuracy was decreased after CTC in the saline group, but was reversed to an increase in the CNO group (Fig. 9 I). Of note, chemogenetic inhibition of NRe vCA1d neurons during baseline conditions (without learning) did not affect the contextual discrimination of vCA1d neurons in general ( Figure S6A-F ), and resulted in a marginal fluctuation of Mahalanobis distance in the coding space of contexts ( Figure S6G ). Taken together, these results suggest that NRe vCA1d neurons decrease the memory specificity of context information in vCA1d neurons at both single-cell and population levels, thereby promoting contextual memory generalization. Discussion In this study, we have systematically mapped the single-neuron projectomes of NRe neurons and characterized two distinct NRe subpopulations using multidisciplinary approaches – NRe vCA1d and NRe vCA1v neurons preferentially targeted different vCA1 subregions and exhibited distinct ion channel gene expressions, membrane excitabilities, and spike waveforms. Our work demonstrated that NRe vCA1d neurons and downstream vCA1d neurons exhibited down-regulated neural coding of contextual discrimination upon learning, suggesting a specific role of NRe vCA1d neurons in memory precision control, which may have to do with their phasic mode of firing activity and presynaptic inputs from brain areas promoting memory generalization. These findings reveal circuit-specific functional diversities among NRe neurons and an unprecedent NRe-vCA1d circuit that favors memory generalization over memory specificity. The NRe has long been regarded as a key bridge for prefrontal-hippocampal interactions 17 – 20 , 23 , 52 . This thalamic nucleus in the ventral midline is involved in various memory functions including working memory 22 , spatial memory 23 – 28 and memory consolidation 29 , generalization 9 , 21 and extinction 30 – 32 , as well as feeding 53 , 54 , circadian regulation 55 , 56 , impulse inhibition 57 , stress and anxiety 58 . These diverse brain functions may be mediated by specific downstream areas preferentially - though not exclusively - targeted by distinct NRe neuron subpopulations. For instance, while NRe neurons of projectome types 5 and 6 both projected to the mPFC and striatum as previously known 59 , they showed very distinct target area preferences (Fig. 1 B and 1 D). This raises a possibility that NRe neurons of projectome types 5 and 6 contribute to different functions linked to the mPFC and striatum 60 , 61 , respectively. Many HIP-projecting NRe neurons send axon collaterals to ENT with similar projection strength. While ENT could be an additional relay from mPFC to HIP 41 , the ENT- and HIP-co-projecting NRe neurons could exert coordinated modulation of direct and indirect information flow from mPFC to HIP, and even interact with other NRe neuron subpopulations such as mPFC-projecting neurons to regulate the prefrontal-hippocampal communication. It remains to be studied in future that how ENT neurons relay the neural coding of context discrimination from NRe to DG or CA1, or both. Our study has revealed evident differences in axon projections between NRe vCA1d and NRe vCA1v neurons including distinctions in their transcriptomic profiles, intrinsic electrophysiological properties, spontaneous neural activities, and presynaptic inputomes. Although putative NRe vCA1d and NRe vCA1v neurons identified in vivo may contain non-HIP-projecting neurons, the finding that putative NRe vCA1d neurons exhibited higher spontaneous firing rate than NRe vCA1v neurons suggests that NRe vCA1d neurons are more likely to discharge in a phasic mode, which was previously shown to promote contextual memory generalization 11 . The gene expression patterns of Na + and K + channels could account for distinct intrinsic and spontaneous electrophysiological properties of NRe vCA1d and NRe vCA1v neurons and may provide therapeutic targets for treating overgeneralization of threat memory in PTSD patients. Compared to NRe vCA1v neurons, NRe vCA1d neurons preferentially receive presynaptic inputs from ZI that has been implicated in the tone threat generalization 62 , 63 . In summary, the connectivity-defined NRe vCA1d neurons may leverage their distinctive ion channel-related gene expressions and intrinsic excitability to promote memory generalization by integrating presynaptic inputs from upstream areas that contribute to memory generalization. In contrast, NRe vCA1v neurons exhibited higher input-output excitability than NRe vCA1d neurons, and received synaptic inputs from brain areas such as the striatum that are deeply involved in aversive learning 64 – 67 , thereby playing a permissive role in the formation of contextual threat memory. The contextual memory depends on the integration of multimodal information driven by various context elements 68 – 70 , thus, the overlap between neural representations of the attributes and/or features of contextual memory may lead to the memory generalization 71 , 72 . Numerous neural mechanisms for memory specificity have been identified as minimizing the overlap between memory attributes/features during memory formation. For instance, the pattern separation of the neural coding of context elements in the circuit network of dentate gyrus, CA3 and septum is crucial for the memory specificity during contextual memory formation 10 , 14 , 73 – 75 . Although neural mechanisms that promote memory generalization were less studied, our work has identified a crucial role of thalamo-hippocampal circuitry, particularly the NRe-vCA1d circuit, in promoting memory generalization. Together with previous findings on the critical role of a Fos-activated DG ensemble 76 and amygdala-ACC projection 77 in promoting memory generalization, the neural network of NRe, amygdala, hippocampus and ACC are emerging as key regulators of memory precision. In conclusion, our study has revealed the diversity of NRe neurons and identified subpopulations differing at molecular, cellular and circuit levels. The learning-induced plasticity in specific NRe subpopulations, along with plasticity in their specific downstream areas in hippocampus, plays pivotal roles in the control of contextual memory precision. METHODS Animals All mice (over 8 weeks) were housed under a 12h light/dark cycle with food and water ad libitum and were socially housed in numbers of two to six littermates in the animal facility of Institute of Neuroscience, Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences. Wild-type C57BL/6J (Slac Laboratory Animal, Shanghai) were used in the study. All animal procedures were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the Institute of Neuroscience. Viral vectors The following AAV vectors (titers in genome copies/ml) were acquired from Taitool Biological, Shanghai and injected into the brain. AAV2/8-hSyn-Cre-WPRE-pA (1.98×10 13 ), AAV2/8-CAG-FLEX-EGFP-WPRE-pA (1.71×10 13 ), scAAV2/retro-hSyn-Cre (1.65×10 13 ), AAV2/9-hSyn-DIO-hM4D(Gi)-mCherry-WPRE-pA (1.03×10 13 ), AAV-EF1ɑ-DIO-mCherry (2.42 ~ 2.9×10 13 ), AAV2/9-mCaMKIIα-GCaMP6f-WPRE-pA (1.03×10 13 ) and AAV2/9-hSyn-hM4D(Gi)-mCherry-WPRE-pA (1.03×10 13 ). The following AAV vectors were produced at gene editing facility at the Institute of Neuroscience. AAV2/DJ-EF1a-GFP (1.09×10 13 ), AAV2/9-CMV-FLEX-TVAmCherry-2A-oG (5×10 13 ), AAVretro-CAG-EGFP-barcode(d) (1.68×10 13 , CCTGTATGCGTGGAG), AAVretro-CAG-EGFP-barcode(v) (2.57×10 13 , GCGTAAGTCTCCTTG). Surgeries Mice were anesthetized under isoflurane (~ 2%) and fixed in a stereotactic fame (RWD R510IP, China). The body temperature was maintained at 35°C by a feedback-controlled heating pad (FHC, USA). Eyes were protected from desiccation and surgery light using erythromycin eye ointment (Fuyuan medicine). Glass pipettes filled with solutions (tip diameter 10–20 µm) were connected to a Picospritzer III (Parker Hannifin Corporation) unless otherwise stated, and stayed for 5 min after injection at following coordinates (posterior to Bregma, AP; lateral to the midline, LAT; below the brain surface, DV; in mm): vCA1d stratum pyramidale , -3.3 AP, -3.2 ML, -1.5 DV; vCA1v stratum pyramidale , AP -3.2, ML -3.5, DV -3.4; vCA1d stratum lacunosum-moleculare , AP -3.35, ML -2.8, DV -1.77; vCA1v stratum lacunosum-moleculare , AP -3.2, ML -3.25, DV -3.4; NRe, AP -0.4, ML -1.35, DV -4.35 with angle 18°. After injection, the skin was sutured by surgical needles and tissue adhesive (Vetbond, 3M) was applied to the wound for protection. Animals remained on a heating pad until full recovery from anesthesia. For sparse labeling, the mixed viral solutions (ratio 1:1, AAV2/8-hSyn-Cre-WPRE-pA diluted to 10 8 , AAV2/8-CAG-FLEX-EGFP-WPRE-pA 10 12 ) were loaded into a mineral oil-filled glass micropipette attached to an oil hydraulic micromanipulator (MO-10B, Narishige), and were unilaterally injected into the NRe at a speed of 80 nL/min. For retrograde dye labeling, 300 nL CTB-647 (#C34778, Invitrogen) and fluorogold (FC10001, Fluorochrome, LLC; diluted at 1:8 in distilled water or saline) were unilaterally injected into vCA1d and vCA1v, in a counter-balanced fashion. For chemogenetic manipulation, 200 nL of AAV2/9-hSyn-DIO-hM4D(Gi)-mCherry-WPRE-Pa was unilaterally injected into left NRe. scAAV2/retro-hSyn-Cre was bilaterally injected into SLM layer of vCA1d or vCA1v. For barcode-based retrograde tracing, 150nL AAVretro-CAG-EGFP-barcode(d) and AAVretro-CAG-EGFP-barcode(v) were unilaterally injected into vCA1d and vCA1v in the same animal, respectively. For Ca 2+ imaging, 250 nL of AAV2/9-mCaMKIIα-GCaMP6f-WPRE-pA was unilaterally injected into pyramidal layer of right vCA1d. After the injection, animals were allowed to recover for 3–4 weeks before the start of behavioral experiment. For slice recording, 275 µl red retrobeads (1:50 dilution) was injected into the stratum lacunosum-moleculare of vCA1d or vCA1v. For tetrodes recording, tetrodes were custom made or acquired from Kedou Brain Computer Technology at Suzhou. The skin was cut open and skull was cleaned by 1/10 hydrogen peroxide. The cranial nails were screwed in skull and the ground wire was welded to them. The skull surface and the cross-section of the skin were coated with 3M tissue adhesive for better adhesion of dental cement and protection, respectively. The skull was opened by cranial drill (78001, RWD) and the dura was removed gently by forceps. The electrode was implanted slowly into the brain. For tetrode recording in NRe, electrodes were implanted at coordinates: AP -0.4, ML -1.35, DV -4.15 with angle 18°. The skull window was covered by Vaseline (BODI) for sealing. The ground wire from cranial nails was welded to the electrode and dental cement were applied to the skull, cranial nails and the electrode for fixation. Lincomycin and lidocaine gel (Decheng Pharmaceutical, Jiangxi) was applied to the cross-section of the skin for postoperative local anaesthetization and anti-infection. After a 2-week recovery period, tetrodes were gradually lowered to reach the target before the start of behavioral experiment. Behavioral paradigm Contextual fear conditioning. Two contexts were used as conditioning context (context A) and neutral context (context B), respectively. Context A was a square chamber (in cm, 30 [L] × 30 [W] × 35 [H]) with mental grid floor, transparent walls, and cleaned with 75% ethanol. Context B was a triangle chamber with flat blue acrylic floor and patterned walls, and was cleaned with 1% acetic acid. Context A and B shared one wall. The behavioral contexts were placed into a sound-attenuating chamber. Before behavioral training, animals were habituated to the training room for 30 min on three consecutive days. On day 1, animals were habituated in the Context A for 5 min to explore the context freely. On day 2 for threat conditioning, 3 foot-shocks (0.75 mA, 2s, shocker H13-15, Coulbourn Instruments) were delivered after 5 min re-habituation. Animals remained in the context A for 30s after the last foot-shock and then returned to their home cage. On day 3, animals were exposed to context B for 5 min. On day 4, animals were exposed to context A for 5 min. Behavior videos (640 x 480 pixels at 30 Hz) were recorded with Cinelyzer (Plexon Inc). The freezing level was scored based on video-frame pixel changes, as previously described 78 . The immobile episodes longer than 1 s were marked as freezing episodes and the freezing was expressed as a percentage of time spent freezing. All parameters were optimized based on the behavioral video and kept the same for the same contexts. Behavior with chemogenetic manipulation. Before fear conditioning, animals were systemically administered with CNO (5 mg/kg, i.p. injection, Sigma) or saline. CNO was first dissolved in 20 µL dimethyl sulfoxide (DMSO, Sigma), then mixed into 380 µL saline before usage. Behavior with miniscope recording. Three contexts were used: Context A and B were the same as described above. Context C was another neutral context in a home-cage shape with no common elements of conditioning context at all. Animals were habituated to the head-mounted miniscope for 2 days before the behavioral training. Calcium imaging was captured by UCLA miniscope (V2, LabMaker) and synchronized with the behavior protocol via a TTL signal from Anilab system. On day 1, animals were exposed to context C, B and A sequentially (5 min in each context with 5 min interval). On day 2 (conditioning day), animals went through same fear conditioning in context A as described above. On day 3, animals were exposed to context C, B and A as same as day 1. Behavior with NRe tetrode recording. For animals that went through threat conditioning paradigm: On day 1, animals were exposed to context B and A sequentially with a 5-min interval. On day 2, animals were intraperitoneally injected with CNO (5 mg/kg). 40 min after injection, animals went through same fear conditioning in context A as described above. On day 3, animals were exposed to context B and A again (same as Day 1). For animals with local chemogenetic inhibition in NRe: On day 1 and 2, animals were exposed to a neutral context for 1 h right after the intraperitoneal injection of CNO and saline, respectively. The injection order of CNO and saline on day 1 and day 2 were counterbalanced among individual animals. On day 3 and day 4, same behavioral procedures were repeated for each animal after tetrodes were moved 40 µm deeper. Generalization index . The contextual generalization index was calculated as the percentage of freezing in the neutral context, divided by the percentage of freezing in the training context. Single-neuron projectome reconstruction and analysis fMOST imaging. Two animals in each batch of injected animals were used to check sparse labeling 2–3 weeks after the AAV injection 36 , 39 . Rest animals in the qualified batches were sacrificed 2–3 months after the AAV injection. Brains were carefully extracted, post-fixed in 4% PFA, rinsed, dehydrated through graded ethanol, and embedded in Lowicryl HM20 resin. Dual-channel high-resolution images were acquired with a voxel resolution of 0.32×0.32×1 µm³, utilizing a 20× water-immersion objective (NA 1.0) 79 . Propidium iodide staining was applied during imaging to provide cytoarchitectural landmarks 38 , with the GFP channel used for neurite tracing and the PI channel for image registration. Each brain yielded approximately 10,000 16-bit TIFF images at 30,000×20,000 pixel resolution. Raw data were segmented into ~ 1,000,000 data cubes (256×256×100 voxels) using the slice2cube command in Fast Neurite Tracer (FNT) and subsequently compressed using High-Efficiency Video Coding (HEVC). For whole-brain overviews, the raw data were down-sampled to a resolution of 256×256×100 µm. Single-cell reconstruction. The reconstruction process involved the following steps 39 : (1) Samples with tissue damage, overly dense labeling (> 200 cells), weak axon terminal signals, or high background fluorescence were excluded. (2) Each neuron was independently reconstructed by two human tracers, and a third tracer randomly reviewed 10–20% of cells per brain by quantifying false branches and misconnections. Samples scoring below 90/100 were returned for correction. (3) A trained expert conducted lint analysis using FNT software to eliminate overlapping traces between neurons and finalized the dataset for analysis. In total, 8 brain samples were used in the reconstruction. Single-cell registration. Reconstructed neurons were spatially aligned to the Allen Mouse Brain Common Coordinate Framework V3 as previously described 39 , 40 . An initial affine transformation including translation, rotation, scaling, and shearing was followed by non-rigid registration to achieve precise alignment 80 . For each brain, a transformation matrix was computed based on PI-stained cytoarchitectural landmarks and applied to the corresponding neurons. The following brain regions were manually segmented in the images and assigned with uniform artificial grayscale values: HIP, anterior commissure, fourth ventricle, caudate putamen, dentate gyrus, genu of the corpus callosum, fasciculus retroflexus, columns of the fornix, granular layer, medial habenula, mammillothalamic tract, pontine gray, paraventricular hypothalamic nucleus, facial nerve, and lateral ventricle. The final transformation matrix was then applied to all reconstructed neurons within the brain. Axon digitalization and projectome cell typing . To effectively classify projectome cell types of NRe neurons, we first classified various types of axon routes, then digitized each axon by a combination of axon routes, finally classified projectome cell types of NRe neurons by unsupervised clustering analysis of digitized axons. The procedures were as follows: (1) The classification of axon route types. Axon trajectories from the soma to each axon terminal were extracted as axon routes. The 3D coordinates of each axon route were compressed to 32-dimensional embeddings by a Deep Neuron Network algorithm 81 . We then performed unsupervised hierarchical fuzzy C-means clustering of these embeddings. The number of clusters (axon route types) were finalized when the clustering dendrogram’s decisive membership was close to 0.5 and its second-highest was close to 0.1. The classification of axon routes was validated and refined using a Gaussian Mixture Model. (2) Axon digitalization. Each axon was reconstructed by a combination of various axon routes using a support vector machine algorithm. (3) Projectome cell typing. After the determination of NRe targeted brain areas, NRe neurons were grouped by their projection patterns. Projectome cell types were determined by the unsupervised clustering of these projection patterns based on their digitalized axons. Slice electrophysiology and data analysis Animals were anaesthetized by 5% isoflurane and then deeply anaesthetized by high-dose isoflurane (~ 150 µL in custom nose masks). Animals were transcardially perfused with ice-cold NMDG-based artificial cerebral spinal fluid (ACSF, 93 mM NMDG, 2.5 mM KCl, 1.2 mM NaH 2 PO 4 , 30 mM NaHCO 3 , 20 mM HEPES, 5 mM sodium ascorbate, 2 mM thiourea, 3 mM sodium pyruvate, 25 mM D-glucose,12 mM N-Acetyl-L-cysteine, 10 mM MgCl 2 , 0.5 mM CaCl 2 , oxygenated with 95% O 2 /5% CO 2 ). The brain was immediately removed after the perfusion, and transferred to ice-cold NMDG-based ACSF. Coronal brain slices (300 µm thick) containing hippocampus were prepared using a vibratome (VT-1200S, Leica) in an ice-cold NMDG-based ACSF. Slices were maintained for 12 min at 32°C in NMDG-based ACSF and were subsequently transferred into HEPES-based ACSF containing (in mM: 92 NaCl, 2.5 KCl,1.2 NaH 2 PO 4 , 30 NaHCO 3 , 20 HEPES, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 25 D-glucose, 2 MgCl 2 , 2 CaCl 2 , bubbled with 95% O 2 /5% CO 2 ) at room temperature and incubated more than 1 h before recording, and then were kept at room temperature (20–22°C) until start of recordings. Slices were transferred to a recording chamber and infused with ~ 30°C recording ACSF (in mM: 119 NaCl, 5 KCl, 1.25 NaH 2 PO 4 , 26 NaHCO 3 , 10 D-glucose, 1 MgCl 2 , 2 CaCl 2 , bubbled with 95% O 2 /5% CO 2 ). All chemicals were purchased from Sigma-Aldrich. Patch pipettes (4–7 MΩ) pulled from borosilicate glass (Sutter instrument, BF150-86-10) were filled with a K-gluconate based internal solution (in mM: 126 K-gluconate, 2 KCl, 2 MgCl 2 ,10 HEPES, 0.2 EGTA, 4 MgATP 2 , 0.4 Na 3 GTP, 10 Na-phosphate creatine, 290 mOsm, adjusted to pH 7.2 ~ 7.3 with KOH). Whole-cell recording was performed with a Multiclamp 700B amplifier and a Digidata 1440A (Molecular Device). To measure I-V curves and assess the intrinsic excitability, NRe neurons were held at -60 mV under current clamp mode, and injected with 1-s current steps starting from 0 pA (10 pA increments, 10 s intervals). The Ra and Rm were assessed by a -10-mV step (repeated 5 times) with the membrane potential held at -70 mV under voltage clamp mode. To measure the parameters of spike waveforms, the first current-evoked spike in each cell was analyzed by Clampfit and MATLAB scripts. The baseline for each spike was determined by the averaged membrane potential during the 5 ms period before the spike threshold (20 mV/ms). Barcode- and scRNA-seq data acquisition and analysis The NRe region was micro-dissected from acute brain slices that were prepared using the same protocol as described above. The isolated NRe-containing tissues were pooled into a single tube and subjected to enzymatic dissociation in a choline chloride-based solution containing 20 U/mL papain and 100 U/mL DNase I (Cat# LK003176, Worthington) at 37°C for 23 minutes. The choline chloride-based dissociation buffer contained (in mM): 92 choline chloride, 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 glucose, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 10 MgSO₄·7H₂O, 0.5 CaCl₂·2H₂O, and 12 N-acetyl-L-cysteine. A second digestion step was performed using 1 mg/mL protease (Cat# P5147, Sigma) and 1 mg/mL dispase (Cat# LS02104, Worthington) at 25°C for 18 minutes. Following enzymatic digestion, the tissues were gently triturated with fire-polished glass Pasteur pipettes to obtain single-cell suspensions. The suspensions were filtered through a 30 µm cell strainer (Cat# 130-098-458, Miltenyi) to remove aggregates and cleaned using a debris removal solution (Cat# 130-109-398, Miltenyi). All steps were completed within 2.5 hours post euthanasia. Finally, approximately 15,000 to 20,000 cells were loaded into the well of a 10× Chromium chip for single-cell RNA sequencing. Single cells were captured using the Chromium Single Cell 4’ Reagent Kits and single-cell transcriptional libraries were obtained by the 10× Genomics protocol. Barcode expression libraries were obtained by PCR using 100 ng of cDNA fragments derived from post-cDNA-amplification cleanup material (containing barcodes and unique molecular identifier), NEBNext Ultra II Q5 Master Mix (NEB, M0544S) and following primers (200 nM): P5-Read1 (TGATACGGCGACCACCGAGATCTACACTCTTTCCCT ACACGACGCTC) and P7-index-Read2-EGFP (CAAGCAGAAGACGGCATACGAGA TAGGATTCGGTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGgCATGGACGAGCTGTACAAG). The PCR protocol was following: initial denaturation, 98°C for 60 s; denaturation, 20–24 cycles of 98°C for 10 s; annealing/extension 65°C for 75 s; final hold, 75°C for 5 min. Finally, the barcodes libraries were harvested from 482–488 bp (including 120 bp adapters) of pooled PCR reactions using a 0.8 × SPRI selection cleanup (Beckman, B23318), and sequenced on a Novaseq Xplus-PE150 Xplus platform. Transcriptome analysis. A custom mouse reference genome was built using Cell Ranger (v8.0.1) based on the GRCm39 genome assembly. Raw sequencing data were aligned and quantified using Cell Ranger with default parameters 82 , and were utilized to generate digital gene expression matrices for further analyses in the Seurat R package (v5.3.0) 83,84 . Cells with abnormally high or low gene counts (representing doublets or background noise) were filtered out if the number of genes (nFeatureRNA) was less than 1000, or over 5% of their UMI counts were derived from mitochondrial genes. After sample integration (“SelectIntegrationFeatures” for features integration, “FindIntegrationAnchors” for anchors, "SCT" for normalization method), dimensionality reduction and clustering were performed through principal component analysis (PCA) using “RunPCA,” followed by clustering via the “FindClusters” function. Cluster-specific marker genes among all clusters, and differentially expressed genes (DEGs) between specific cell subtypes were identified with “FindAllMarkers” and “FindMarkers”, respectively, using thresholds of p 1. To identify barcoded NRe neurons that project to vCA1, we utilized the following criteria. (1) The EGFP mRNA was detectable (nUMI > 0). (2) Barcode count in each neuron was higher than the 99th percentile of barcode counts in non-neuronal cells. (3) The top 1% neurons with highest barcode counts were considered as potential doublets or PCR amplification artifacts and exclude from further analysis. To identify NRe vCA1d and NRe vCA1v neurons in the scRNA-seq dataset, we first quantify the projection preference of each neuron – the selectivity index (SI) – based on the count of two barcode types (d and v) using the equation: SI = (d –v) / (d + v). Then the NRe vCA1d and NRe vCA1v neurons were identified based on the bimodal distribution of SI. Neurons with SI close to 0 (10% of the bimodal distribution, corresponding to absolute SI < 0.2) were considered as co-projecting neurons. Expression profiles of ion channels and transmitter receptors. SCN family: Scn1a, Scn2a, Scn3a, Scn4a, Scn5a, Scn7a, Scn8a, Scn9a, Scn10a, Scn11a, Scn1b, Scn2b, Scn3b, Scn4b. KCN family: Kcnc1, Kcnc3, Kcnn1, Kcnn2, Kcnn3, Kcnq2, Kcnq3, Kcnq5, Kcnma1 . CACNA family: Cacna1c, Cacna1d, Cacna1a, Cacna1b, Cacna1e, Cacna1g, Cacna1h, Cacna1i . HCN family: Hcn1, Hcn2, Hcn3, Hcn4. ACH family: Chrm1, Chrm2, Chrm3, Chrna1, Chrna2, Chrna3, Chrna4, Chrna5, Chrna6, Chrna7, Chrna9, Chrnb1, Chrnb2, Chrnb3, Chrnb4 . GABA receptor family: Gabra1, Gabra2, Gabra3, Gabra4, Gabra5, Gabrb1, Gabrb2, Gabrb3, Gabrg1, Gabrg2, Gabrg3, Gabrd, Gabre, Gabrp, Gabrq, Gabrr1, Gabrr2, Gabrr3, Gabbr1, Gabbr2 . Glutamate receptor family: Grik1, Grik2, Grik3, Grik4, Grik5, Gria1, Gria2, Gria3, Gria4, Grin1, Grin2a, Grin2b, Grin2c, Grin2d, Grin3a, Grin3b . Neurexin family: Nrxn1, Nrxn2, Nrxn3. Neuroligin family: Lrrtm1, Lrrtm2, Lrrtm3, Lrrtm4 . Transporter family: Slc6a1, Slc6a13, Slc6a11, Slc6a12, Slc32a1, Slc1a3, Slc1a2, Slc1a1, Slc1a6, Slc17a7, Slc17a6, Slc17a8 . Spatial transcriptome analysis The spatial gene expression matrix for the thalamus and hypothalamus (TH/HY) was obtained from a published dataset 44 and binned by aggregating transcripts of the same gene within each bin (50 × 50 DNA nanoball grid, 25 µm in diameter per bin). Data processing and visualization were conducted using the Seurat package (v4.0.2) and the Tidyverse package (v1.3.1) in R (v4.1.2). The number of detected genes and mitochondrial gene were used for quality control. Bins were excluded from analysis if nFeature 5%. The dataset was log-normalized and variance-stabilized using “SCTransform” with default parameters. Batch effects among brain slices were corrected using the Harmony package (v0.1.0). Clusters and subclusters were identified using “FindClusters” in Seurat, applying shared nearest neighbor modularity optimization with a clustering resolution of 0.1. The neighborhood graph was projected into two dimensions using Uniform Manifold Approximation and Projection (UMAP) for visualization. The marker genes in each cluster/subclusters were identified by differential expression analysis using “FindAllMarkers” in Seurat. Single-unit electrophysiology and analysis Tetrodes implanted in NRe was connected with a digital headstage, which was connected to CerePlex Direct (Blackrock Neurotech) via a lightweight data cable. Neuronal activity was digitized at 30 kHz and low-cut Bessel filtered at 300 Hz. The tetrodes were moved in small daily increments (40 µm) to record different single units. At the end of the experiment, recording sites were marked with electrolytic lesions (7 µA, 5 s) and reconstructed with standard histological procedures. Single-unit spike sorting was conducted using Off-Line Spike Sorter (OFSS, Plexon). Raw signals were low-cut filtered at 300Hz by a 4-pole Bessel filtering. After thresholds adjustment and waveform detection, single units were manually defined based on 3D peak-valley features and principal component features. Spikes within the refractory period (1.5 ms) were removed. If two single units share 50% spikes at 1 ms temporal resolution, the unit with lower firing rates was defined as duplicated spikes and removed. The high quality of spike sorting was reflected by a high J3 value (ratio of inter-cluster distances to intra-cluster distances), a low DB index (ratio of intra-cluster distances to inter-cluster distances) and significant multivariate ANOVA (p < 0.05). Burst spikes were defined as a cluster of densely fired spikes occurring within a specified time window and satisfying criteria including a minimum duration of 10 ms, a minimum spike count of 3, and a maximum inter-spike interval of 300 ms. The burst index was calculated as the ratio of the total number of spikes within bursts to the total spike count, serving as a quantitative measure of the proportion of burst-related activity. The burst duration was determined as the difference between the maximum and minimum timestamps of spikes within each burst, providing the temporal extent of the burst. Context discriminative cells were identified by a bootstrap method 85 . For each cell, all the spikes during context exposures were used to create a surrogate distribution of expected spikes for each context by shuffling the inter-spike intervals from the original spike timestamps (1000 iterations). Cells were considered to be context discriminative if their average firing rate in the context fell outside of the surrogate distribution (P < 0.05). Contextual discrimination index (DI) for a single unit was computed as the absolute difference between averaged firing rates in two contexts divided by the sum of averaged firing rates in two contexts. The higher DI, the better contextual discrimination. Mahalanobis distance in the populational coding space was determined by the following steps: (1) Raw data of each neuron was z-scored within each context. (2) For two groups of single units with different numbers, only a random subset of neurons (same number as the smaller group) in the larger group were used for the calculation of Mahalanobis distance 85 . (3) The previous step was repeated for 50 times to determine the significance of Mahalanobis distance (Wilcoxon rank-sum test). Miniscope Ca 2+ imaging and data analysis Three weeks following AAV injection, a GRIN lens (1.8 mm Φ, 0.25 pitch, 0.55 NA, Edmunt 64519) was implanted above vCA1d. Briefly, a cranial window was made above the vCA1, and the brain tissue above the target was aspirated with syringe blunt needle attached to a Peristaltic Pump (BT100-1F, LongerPump). Saline was repeatedly applied to the exposed tissue to prevent desiccation and clean the blood. Aspiration was stopped once a thin layer of corpus callosum was seen. After the clear of blood, the GRIN lens was slowly lowered above the vCA1d with a custom holder. The lens was fixed to the skull using ultraviolet-light curing dental resin (CharmFil Flow, DentKist Inc). Finally, acrylic dental cement (Super Bond C&B, Jakarta) was used to seal the skull. One week after lens implantation, GCaMP6f fluorescence was checked regularly using a miniature microscope (UCLA V2, LabMaker). Once a clear field of view with neurons was observed, a baseplate attached with a miniscope was fixed to the skull using acrylic dental cement. The miniscope was detached after the fixation. The baseplate was protected by a plastic cover. The animal was returned to the home cage and remained undisturbed for at least one week. During the habituation of head-mounted miniscope, the acquisition settings including exposure time, gain and LED power were set properly to get a clear vision for single neurons in the FOV. The same miniscope was used across all behavioral sessions in the same animal with same acquisition settings to perform repetitive imaging. Before each imaging session, the miniscope was attached to the animal head and habituated for 5 min in home cage. A snapshot of imaging FOV was used to confirm the consistency across multiple sessions. The behavioral video recording (Cinelyzer, Plexon Inc.) and the miniscope imaging were synchronized by TTL signals from a Anilab system (Anilab, China). The miniscope imaging data was recorded to uncompressed AVI files at 30 Hz by MiniScopeControl software (UCLA V2). The imaging data across all sessions were concatenated into a single video. The field of view was cropped and then 2× spatially and 3× temporally down-sampled using moviepy python package to reduce the computation load afterwards. The preprocessed data was processed by CaImAn (1.8.5) toolbox using constrained nonnegative matrix factorization extension (CNMFe) method 86 . The motion correction and source extraction were all done with default parameter setting except the following: down-sampling factor in time for initialization (tsub) 4; neuron diameter (gSiz) 13 pixels; 2D Gaussian kernel smoothing (gSig) 3 pixels; minimum peak to noise ratio (min_pnr) 10; spatial consistency (rval_thr) 0.85. The denoised temporal activity trace (estimates.C) of each neuron extracted by CNMF-E for further analysis with MATLAB scripts (MathWorks). Context discriminative cells and discriminative index (DI) were determined in the same way as single-unit analysis. Mahalanobis distance. For each neuron, the calcium activity was z-scored and temporally binned with a 5-s time window. Neurons recorded in all mice were pooled for population analysis. We randomly subsampled 50 neurons to calculate the Mahalanobis distance with 5000 repetitions. Context decoding. A classifier was trained by the neural activity evoked by context A and B, using a Treebagger algorithm in MATLAB. A total of 100 classification trees were used for each classifier. The 5 min of memory retrieval was split into 60 bins (5 s/bin). The classifier was trained by a 15 s of neural activity and tested by the rest of neural activity. The decoding accuracy was determined from ~ 30 repetitive decoding with a sliding window of 10 s. Histology and co-localization analysis Mice were administered a lethal dose of isoflurane and intraperitoneal perfused with phosphate buffered saline (PBS) followed by 4% w/v paraformaldehyde (PFA). Following dissection, brains were post-fixed in 4% PFA overnight at 4°C and coronal sections were cut at 80 µm thickness using a vibratome (Leica, Germany, VT1000S). Slices were mounted on microscope slides, dried, and imaged with a FV3000 confocal microscope (Olympus). The fluorescent neurons were labeled by CTB647 and Fluorogold, and were quantified by the CellCounter plugin in Fiji. Quantification and statistical analysis Data are summarized as mean values with standard error of the mean (SEM). The cell number is indicated with n and the animal number is indicated with N. No statistical methods were used to predetermine sample sizes. The sample sizes were chosen based on published studies and current standards in the field. Statistical analysis was performed in Prism 8 (GraphPad) and MATLAB (MathWorks). For all datasets normality was tested using the Kolmogorov–Smirnov test (a < 0.05) and homogeneity of variance with Levene’s test (a < 0.05) to determine whether parametric or non-parametric analyses were required. Parametric analyses included unpaired t -tests. If either homogeneity of variance or normality assumptions were not met, non-parametric analyses such as non-parametric Wilcoxon rank sum, or Wilcoxon signed rank were used. For data analyzed in the bootstrap resampling procedure to identify context cell and non-context cell, 95% confident interval were calculated by approximating to a normal distribution, owing to the large sample size. All tests were two-tailed. Statistical parameters and significance are reported in the text or in the figure legends. The significance threshold was placed at 𝑎 = 0.05 (n.s., P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001). Lead Contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Chun XU ( [email protected] ). Data Availability Source data have been deposited at Brain Data Center, Chinese Academy of Sciences and are publicly available at https://doi.org/10.12412/BSDC.1757059765.20001 . The scRNA-seq data would be freely available from the National Genomics Data Center (NGDC) and associated repositories Genome Sequence Archive (GSA), upon publication (accession number: CRA030346; URL: https://ngdc.cncb.ac.cn/gsa/s/x5u9X6JU ). Code Availability This paper does not report original code. Declarations Acknowledgments We thank all members of the Xu lab for helpful discussions and comments. We thank Enpeng Liang, the animal facility, gene editing facility and the imaging facility for technical support. We thank the UCLA miniscope team for sharing the design of the miniscope system. This study was supported by National Science and Technology Innovation 2030 Major Program (2022ZD0205000, 2021ZD0201001), CAS Project for Young Scientists in Basic Research (YSBR-116), National Key R&D Program of China (2020YFE0205900), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB32010105), and Lingang Lab (Grant LG202104-01-08). Author contributions Conceptualization: C.X., Q.W., J.L., Y.S., H.H, Y.C, H.G., X.L. Investigation (behaviors and neural recordings): Q.W., Y.L., Z.S., G.W., Y.Z. Investigation (single-neuron projectome): Q.W., X.L., H.G. Investigation (transcriptome data acquisition): Q.W., R.Z., E.L., Y.C. Formal analysis (behaviors and neural recordings): Q.W., Y.L., G.W., Z.S., Y.G. Formal analysis (transcriptome data analysis): Q.W., M.Y. Formal analysis (single-neuron projectome): S.C., B.Z. S.C. Funding acquisition and resources: C.X., H.H., Y.G., H.G. Supervision: C.X. Writing (original draft): C.X., Q.W., S.C., Y.L., Y.G. Writing (review & editing): C.X., Q.W. Declaration of interests The authors declare no competing interests. References Shmuel Lissek, S. R., Randi E. Heller, D. L., Marilla Geraci, & Daniel S. Pine, C. G. Overgeneralization of Conditioned Fear as a Pathogenic Marker of Panic Disorder. Am J Psychiatry , doi:10.1176/appi.ajp.2009.0903041 (2010). Dunsmoor, J. E. & Paz, R. Fear Generalization and Anxiety: Behavioral and Neural Mechanisms. Biol Psychiatry 78 , 336-343, doi:10.1016/j.biopsych.2015.04.010 (2015). Elzinga, B. M. & Bremner, J. D. Are the neural substrates of memory the final common pathway in posttraumatic stress disorder (PTSD)? Journal of Affective Disorders , doi:10.1016/s0165-0327(01)00351-2 (2002). Kaczkurkin, A. N. et al. Neural Substrates of Overgeneralized Conditioned Fear in PTSD. Am J Psychiatry 174 , 125-134, doi:10.1176/appi.ajp.2016.15121549 (2017). Lissek, S. et al. Generalized anxiety disorder is associated with overgeneralization of classically conditioned fear. Biol Psychiatry 75 , 909-915, doi:10.1016/j.biopsych.2013.07.025 (2014). Lissek, S. et al. Generalization of conditioned fear-potentiated startle in humans: experimental validation and clinical relevance. Behav Res Ther 46 , 678-687, doi:10.1016/j.brat.2008.02.005 (2008). Tovote, P., Fadok, J. P. & Luthi, A. Neuronal circuits for fear and anxiety. Nat Rev Neurosci 16 , 317-331, doi:10.1038/nrn3945 (2015). Botta, P. et al. Regulating anxiety with extrasynaptic inhibition. Nat Neurosci 18 , 1493-1500, doi:10.1038/nn.4102 (2015). Xu, W. & Sudhof, T. C. A neural circuit for memory specificity and generalization. Science 339 , 1290-1295, doi:10.1126/science.1229534 (2013). McHugh, T. J. et al. Dentate gyrus NMDA receptors mediate rapid pattern separation in the hippocampal network. Science 317 , 94-99, doi:10.1126/science.1140263 (2007). Xu, W. & Südhof, T. C. A Neural Circuit for Memory Specificity and Generalization. Science 339 , 1290-1295, doi:10.1126/science.1229534 (2013). Cullen, P. K., Gilman, T. L., Winiecki, P., Riccio, D. C. & Jasnow, A. M. Activity of the anterior cingulate cortex and ventral hippocampus underlie increases in contextual fear generalization. Neurobiol Learn Mem 124 , 19-27, doi:10.1016/j.nlm.2015.07.001 (2015). Xu, W. et al. Distinct neuronal coding schemes in memory revealed by selective erasure of fast synchronous synaptic transmission. Neuron 73 , 990-1001, doi:10.1016/j.neuron.2011.12.036 (2012). Besnard, A., Miller, S. M. & Sahay, A. Distinct Dorsal and Ventral Hippocampal CA3 Outputs Govern Contextual Fear Discrimination. Cell Rep 30 , 2360-2373 e2365, doi:10.1016/j.celrep.2020.01.055 (2020). Hyman, J. M., Ma, L., Balaguer-Ballester, E., Durstewitz, D. & Seamans, J. K. Contextual encoding by ensembles of medial prefrontal cortex neurons. Proc Natl Acad Sci U S A 109 , 5086-5091, doi:10.1073/pnas.1114415109 (2012). Frankland, P. W., Cestari, V., Filipkowski, R. K. & McDonald, R. J. The Dorsal Hippocampus Is Essential for Context Discrimination but Not for Contextual Conditioning. Behavioral Neuroscience (1998). Vertes, R. P., Hoover, W. B., Do Valle, A. C., Sherman, A. & Rodriguez, J. J. Efferent projections of reuniens and rhomboid nuclei of the thalamus in the rat. J Comp Neurol 499 , 768-796, doi:10.1002/cne.21135 (2006). Weel, M. J. D.-v. d., Griffin, A. L., Ito, H. T. & Shapiro, M. L. The nucleus reuniens of the thalamus sits at the nexus of a hippocampus and medial prefrontal cortex circuit enabling memory and behavior. Learning & Memory , doi:10.1101/lm.048389.118 (2020). Cassel, J. C. et al. The reuniens and rhomboid nuclei: neuroanatomy, electrophysiological characteristics and behavioral implications. Prog Neurobiol 111 , 34-52, doi:10.1016/j.pneurobio.2013.08.006 (2013). Varela, C., Kumar, S., Yang, J. Y. & Wilson, M. A. Anatomical substrates for direct interactions between hippocampus, medial prefrontal cortex, and the thalamic nucleus reuniens. Brain Struct Funct 219 , 911-929, doi:10.1007/s00429-013-0543-5 (2014). Ratigan, H. C., Krishnan, S., Smith, S. & Sheffield, M. E. J. A thalamic-hippocampal CA1 signal for contextual fear memory suppression, extinction, and discrimination. Nat Commun 14 , 6758, doi:10.1038/s41467-023-42429-6 (2023). Hallock, H. L., Wang, A., Shaw, C. L. & Griffin, A. L. Transient Inactivation of the Thalamic Nucleus Reuniens and Rhomboid Nucleus Produces Deficits of a Working-Memory Dependent Tactile-Visual Conditional Discrimination Task. doi:10.l037/a0034653 (2013). Hallock, H. L., Wang, A. & Griffin, A. L. Ventral Midline Thalamus Is Critical for Hippocampal-Prefrontal Synchrony and Spatial Working Memory. J Neurosci 36 , 8372-8389, doi:10.1523/JNEUROSCI.0991-16.2016 (2016). Huang, X. et al. A Visual Circuit Related to the Nucleus Reuniens for the Spatial-Memory-Promoting Effects of Light Treatment. Neuron 109 , 347-362 e347, doi:10.1016/j.neuron.2020.10.023 (2021). Loureiro, M. et al. The ventral midline thalamus (reuniens and rhomboid nuclei) contributes to the persistence of spatial memory in rats. J Neurosci 32 , 9947-9959, doi:10.1523/JNEUROSCI.0410-12.2012 (2012). Jones, M. W. & Wilson, M. A. Theta rhythms coordinate hippocampal-prefrontal interactions in a spatial memory task. PLoS Biol 3 , e402, doi:10.1371/journal.pbio.0030402 (2005). Ito, H. T., Zhang, S. J., Witter, M. P., Moser, E. I. & Moser, M. B. A prefrontal-thalamo-hippocampal circuit for goal-directed spatial navigation. Nature 522 , 50-55, doi:10.1038/nature14396 (2015). Jankowski, M. M. et al. Nucleus reuniens of the thalamus contains head direction cells. Elife 3 , doi:10.7554/eLife.03075 (2014). Ferraris, M., Cassel, J. C., Pereira de Vasconcelos, A., Stephan, A. & Quilichini, P. P. The nucleus reuniens, a thalamic relay for cortico-hippocampal interaction in recent and remote memory consolidation. Neurosci Biobehav Rev 125 , 339-354, doi:10.1016/j.neubiorev.2021.02.025 (2021). Ramanathan, K. R., Jin, J., Giustino, T. F., Payne, M. R. & Maren, S. Prefrontal projections to the thalamic nucleus reuniens mediate fear extinction. Nat Commun 9 , 4527, doi:10.1038/s41467-018-06970-z (2018). Silva, B. A. et al. A thalamo-amygdalar circuit underlying the extinction of remote fear memories. Nat Neurosci 24 , 964-974, doi:10.1038/s41593-021-00856-y (2021). Tomaszewski, K. F. et al. , doi:10.1101/2024.05.20.594930 (2024). Bokor, H., Csaki, A., Kocsis, K. & Kiss, J. Cellular architecture of the nucleus reuniens thalami and its putative aspartatergic/glutamatergic projection to the hippocampus and medial septum in the rat. Eur J Neurosci 16 , 1227-1239, doi:10.1046/j.1460-9568.2002.02189.x (2002). Walsh, D. A., Brown, J. T. & Randall, A. D. In vitro characterization of cell-level neurophysiological diversity in the rostral nucleus reuniens of adult mice. J Physiol 595 , 3549-3572, doi:10.1113/JP273915 (2017). Qiu, S. et al. Whole-brain spatial organization of hippocampal single-neuron projectomes. Science 383 , doi:10.1126/science.adj9198 (2024). Qiu, S. et al. Whole-brain spatial organization of hippocampal single-neuron projectomes. Science 383 , eadj9198, doi:10.1126/science.adj9198 (2024). Gong, H. et al. High-throughput dual-colour precision imaging for brain-wide connectome with cytoarchitectonic landmarks at the cellular level. Nat Commun 7 , 12142, doi:10.1038/ncomms12142 (2016). Gong, H. et al. High-throughput dual-colour precision imaging for brain-wide connectome with cytoarchitectonic landmarks at the cellular level. Nature Communications 7 , doi:10.1038/ncomms12142 (2016). Ren, B., Shi, X., Zhao, B., Xu, C. & Wang, X. Reconstruction of Single-Neuron Projectomes in Mice. Bio Protoc 15 , e5300, doi:10.21769/BioProtoc.5300 (2025). Wang, Q. et al. The Allen Mouse Brain Common Coordinate Framework: A 3D Reference Atlas. Cell 181 , 936-953 e920, doi:10.1016/j.cell.2020.04.007 (2020). Dolleman-van der Weel, M. J. et al. The nucleus reuniens of the thalamus sits at the nexus of a hippocampus and medial prefrontal cortex circuit enabling memory and behavior. Learn Mem 26 , 191-205, doi:10.1101/lm.048389.118 (2019). Roth, B. L. DREADDs for Neuroscientists. Neuron 89 , 683-694, doi:10.1016/j.neuron.2016.01.040 (2016). Xu, P. et al. High-throughput mapping of singleneuron projection and molecular features by retrograde barcoded labeling. eLife , doi:https://doi.org/10.7554/eLife.85419 (2024). Han, L. et al. Single-cell spatial transcriptomic atlas of the whole mouse brain. Neuron 113 , 2141-2160 e2149, doi:10.1016/j.neuron.2025.02.015 (2025). Yao, Z. et al. A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature 624 , 317-332, doi:10.1038/s41586-023-06812-z (2023). Niell, C. M. & Stryker, M. P. Highly selective receptive fields in mouse visual cortex. J Neurosci 28 , 7520-7536, doi:10.1523/JNEUROSCI.0623-08.2008 (2008). Cai, D. J. et al. A shared neural ensemble links distinct contextual memories encoded close in time. Nature 534 , 115-118, doi:10.1038/nature17955 (2016). Ghosh, K. K. et al. Miniaturized integration of a fluorescence microscope. Nature Methods 8 , 871-878, doi:10.1038/nmeth.1694 (2011). Chen, T.-W. et al. Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499 , 295-300, doi:10.1038/nature12354 (2013). Jimenez, J. C. et al. Contextual fear memory retrieval by correlated ensembles of ventral CA1 neurons. Nat Commun 11 , 3492, doi:10.1038/s41467-020-17270-w (2020). Fiser, A. et al. Experience-dependent spatial expectations in mouse visual cortex. Nat Neurosci 19 , 1658-1664, doi:10.1038/nn.4385 (2016). Ferraris, M. et al. The Nucleus Reuniens Controls Long-Range Hippocampo-Prefrontal Gamma Synchronization during Slow Oscillations. J Neurosci 38 , 3026-3038, doi:10.1523/JNEUROSCI.3058-17.2018 (2018). PURVIS, C. C. & DUNCAN, M. J. Discrete thalamic lesions attenuate winter adaptations and increase body weight. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology , doi:https://doi.org/10.1152/ajpregu.1997.273.1.R226 (1997). Wilmot, C. A., Sullivan, A. C. & Levin, B. E. Effects of diet and obesity on brain a 1- and a2-noradrenergic receptors in the rat Brain Research , doi:https://doi.org/10.1016/0006-8993(88)90154-0 (1988). Freeman, D. A. & Zucker, I. Refractoriness to melatonin occurs independently atmultiple brain sites in Siberian hamsters. PNAS , doi:https://doi.org/10.1073/pnas.111140398 (2001). Giardino, L. et al. DAILY MODIFICATIONS OF 3H-NALOXONE BINDING SITES IN THE RAT BRAIN: A QUANTITATIVE AUTORADIOGRAPHIC STUDY. Chronobiology International , doi:https://doi.org/10.3109/07420528909056920 (1989). Prasad, J. A., Macgregor, E. M. & Chudasama, Y. Lesions of the thalamic reuniens cause impulsive but not compulsive responses. Brain Struct Funct 218 , 85-96, doi:10.1007/s00429-012-0378-5 (2013). CULLINAN, W. E., HERMAN, J. P., BATTAGLIA, D. F., AKIL, H. & WATSON, S. J. PATTERN AND TIME COURSE OF IMMEDIATE EARLY GENE EXPRESSION IN RAT BRAIN FOLLOWING ACUTE STRESS Neuroscience , doi:https://doi.org/10.1016/0306-4522(94)00355-9 (1995). Otake, K. & Nakamura, Y. Single midline thalamic neurons projecting to both the ventral striatum and the prefrontal cortex in the rat. Neuroscience 86 , 635-649, doi:10.1016/s0306-4522(98)00062-1 (1998). Otake, K. & Nakamura, Y. Single midline thalamic neurons projecting to both the ventral striatum and the prefrontal cortex in the rat. Neuroscience , doi:https://doi.org/10.1016/S0306-4522(98)00062-1 (1998). Berendse, H. W. & Groenewegen, H. J. Organization of the thalamostriatal projections in the rat, with special emphasis on the ventral striatum. THE JOURNAL OF COMPARATIVE NEUROLOGY , doi:https://doi.org/10.1002/cne.902990206 (1990). Venkataraman, A. et al. Modulation of fear generalization by the zona incerta. Proc Natl Acad Sci U S A 116 , 9072-9077, doi:10.1073/pnas.1820541116 (2019). Venkataraman, A. et al. Incerto-thalamic modulation of fear via GABA and dopamine. Neuropsychopharmacology 46 , 1658-1668, doi:10.1038/s41386-021-01006-5 (2021). Correia, S. S., McGrath, A. G., Lee, A., Graybiel, A. M. & Goosens, K. A. Amygdala-ventral striatum circuit activation decreases long-term fear. Elife 5 , doi:10.7554/eLife.12669 (2016). Chen, A. P. F. et al. Nigrostriatal dopamine modulates the striatal-amygdala pathway in auditory fear conditioning. Nat Commun 14 , 7231, doi:10.1038/s41467-023-43066-9 (2023). Belilos, A. et al. Nucleus accumbens local circuit for cue-dependent aversive learning. Cell Rep 42 , 113488, doi:10.1016/j.celrep.2023.113488 (2023). Hikida, T. et al. Pathway-specific modulation of nucleus accumbens in reward and aversive behavior via selective transmitter receptors. Proc Natl Acad Sci U S A 110 , 342-347, doi:10.1073/pnas.1220358110 (2013). Yadav, N. et al. Prefrontal feature representations drive memory recall. Nature 608 , 153-160, doi:10.1038/s41586-022-04936-2 (2022). O'Reilly, R. C. & Rudy, J. W. Conjunctive representations in learning and memory: principles of cortical and hippocampal function. Psychol Rev 108 , 311-345, doi:10.1037/0033-295x.108.2.311 (2001). Huckleberry, K. A., Ferguson, L. B. & Drew, M. R. Behavioral mechanisms of context fear generalization in mice. Learn Mem 23 , 703-709, doi:10.1101/lm.042374.116 (2016). Howard, M. W., Fotedar, M. S., Datey, A. V. & Hasselmo, M. E. The temporal context model in spatial navigation and relational learning: toward a common explanation of medial temporal lobe function across domains. Psychol Rev 112 , 75-116, doi:10.1037/0033-295X.112.1.75 (2005). McTighe, S. M., Cowell, R. A., Winters, B. D., Bussey, T. J. & Saksida, L. M. Paradoxical false memory for objects after brain damage. Science 330 , 1408-1410, doi:10.1126/science.1194780 (2010). Besnard, A. et al. Dorsolateral septum somatostatin interneurons gate mobility to calibrate context-specific behavioral fear responses. Nat Neurosci 22 , 436-446, doi:10.1038/s41593-018-0330-y (2019). McClelland, J. L., McNaughton, B. L. & O'Reilly, R. C. Why There Are Complementary Learning Systems in the hippocampus and Neocortex: Insights From the Successes and Failures of Connectionist Models of Learning and Memory. Psychological Review , doi:https://doi.org/10.1037/0033-295X.102.3.419 (1995). Willshaw, D. J., Dayan, P. & Morris, R. G. Memory, modelling and Marr: a commentary on Marr (1971) 'Simple memory: a theory of archicortex'. Philos Trans R Soc Lond B Biol Sci 370 , doi:10.1098/rstb.2014.0383 (2015). Sun, X. et al. Functionally Distinct Neuronal Ensembles within the Memory Engram. Cell 181 , 410-423 e417, doi:10.1016/j.cell.2020.02.055 (2020). Vincent, C. J., Aguilar-Alvarez, R., Vanderhoof, S. O., Mott, D. D. & Jasnow, A. M. An amygdala-cortical circuit for encoding generalized fear memories. Mol Psychiatry , doi:10.1038/s41380-025-03140-8 (2025). Bai, T. et al. Learning-prolonged maintenance of stimulus information in CA1 and subiculum during trace fear conditioning. Cell Rep 42 , 112853, doi:10.1016/j.celrep.2023.112853 (2023). Zhong, Q. et al. High-definition imaging using line-illumination modulation microscopy. Nature Methods 18 , 309-315, doi:10.1038/s41592-021-01074-x (2021). Rohlfing, T. & Maurer, C. R., Jr. Nonrigid image registration in shared-memory multiprocessor environments with application to brains, breasts, and bees. IEEE Trans Inf Technol Biomed 7 , 16-25, doi:10.1109/titb.2003.808506 (2003). Hinton, G. E. & Salakhutdinov, R. R. Reducing the dimensionality of data with neural networks. Science 313 , 504-507, doi:10.1126/science.1127647 (2006). Zheng, G. X. Y. et al. Massively parallel digital transcriptional profiling of single cells. Nature Communications 8 , doi:10.1038/ncomms14049 (2017). Stuart, T. et al. Comprehensive Integration of Single-Cell Data. Cell 177 , 1888-1902 e1821, doi:10.1016/j.cell.2019.05.031 (2019). Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184 , 3573-3587 e3529, doi:10.1016/j.cell.2021.04.048 (2021). Rozeske, R. R. et al. Prefrontal-Periaqueductal Gray-Projecting Neurons Mediate Context Fear Discrimination. Neuron 97 , 898-910 e896, doi:10.1016/j.neuron.2017.12.044 (2018). Zhou, P. et al. Efficient and accurate extraction of in vivo calcium signals from microendoscopic video data. Elife 7 , doi:10.7554/eLife.28728 (2018). Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryFigures.docx Supplementary Figures Figure S1. Classification of axon routes and projectome cell types of NRe neurons (A) Schematic diagram showing the procedure of axon routes classification: reconstruction, embedding, clustering, and refinement. (B) Examples showing axon trajectories of 8 clusters of axon routes. (C) The heatmap showing the axon projection strength of 8 axon route clusters in various brain areas. The strength was normalized by the max strength among all brain areas and coded by the color bar on the right. (D) Schematic diagram showing the procedure of projectome cell typing: categorizing neurons based on their projection patterns, digitalizing a total of 150 projection patterns using digitalized axon route clusters, and clustering digitalized projection patterns. (E) Summary of silhouette score (upper) and Davies Bouldin score (lower) as a function of the number of projection cell types. The number of projectome cell types (N = 6) corresponded to the red line reflecting the local maxima of the silhouette score curve and the local minimal of the Davies Bouldin score curve. (F) Upper, heatmap showing clustered projection patterns (separated by lines) based on the representation of each projection pattern using counts of each axon route cluster. Lower, heatmap showing the axon projection strength in target brain regions. Heatmaps were normalized by the min-max and color coded with scale bars on the right. Figure S2. NRe vCA1d neurons promote the expression of threat generalization. (A) Schematic diagram illustrating virus injection and tetrodes recording in NRe. (B) Experimental protocol showing that recording started right after saline or CNO injection. On the next day, same recording was repeated again with the other injection. (C) Example pictures showing fluorescent expression of hM4Di-mCherry and the recording site marked by electrical lesion. Scale bar, 200 μm. (D) Histogram showing firing rates of an uninhibited cell (upper) and an inhibited cell (lower) upon the injection of CNO. (E) Summary of cell numbers of inhibited and uninhibited neurons upon saline or CNO injections (8 sessions for saline group, 8 sessions for CNO group, N = 4 mice). The inhibited neuron was determined by a significant reduction of firing rate from 0-10 min to 30-40 min post saline or CNO injection. The proportion of inhibited cells was significantly higher in the CNO group (57/98 vs. 40/95, Chi-square test, p = 0.0257). (F) Summary of the firing rate of all recorded NRe cells during the 60-min period post injection of saline (n = 95) or CNO (n = 98). Kruskal-Wallis test, p < 0.0001. Dunn's multiple comparisons test at various time periods. (G) Schematic diagram illustrating behavioral protocol with CNO treatment during threat memory retrieval in context B and A. (H) Reproduction of the viral expression strategies shown in Figure 3B. (I) Summary of time spent in freezing during threat memory retrieval in context A and B (left), and generalization index (right) for animals with NRe vCA1d neurons labeled. Unpaired t -test, control group (N = 11) vs. hM4Di group (N = 9). (J) Reproduction of the viral expression strategies shown in Figure 3D. (K) The same plots as (I) for animals with NRe vCA1v neurons labeled. Unpaired t -test, control group (N = 18) vs. hM4Di group (N = 13). Data summary: mean ± SEM. Statistics: two-tailed. P values: ***p < 0.001, **p < 0.01, *p 0.05. Figure S3. Summary of viral injection sites and tetrode recording sites. (A and B) Fluorescent signals of blue beads in vCA1d (A) and vCA1v (B) and reconstructed positions of co-injected blue beads in animals used for chemogenetic experiments. Scale bars in histology picture and zoom in: 0.5 mm and 0.1 mm. (C) Positions of tetrode recording sites. Figure S4. Spatial and single-cell transcriptome analysis in NRe. (A) Volcano plot showing DEGs between NRe (area defined by histology) and brain areas surrounding NRe in thalamus and hypothalamus. Dots were colored if |Log2 fold change| > 1 and p value < 0.05. (B) Dot plot showing the expression of DEGs in NRe compared to thalamic areas surrounding NRe and hypothalamus. (C) Spatial expression patterns of NRe marker genes. Scale bars, 0.5 mm. (D) Violin plots showing the number of detected genes, UMI counts, and percent of mitochondrial genes in each snRNA-seq-defined cell cluster. (E) The UMAP embeddings showing single-gene expression patterns. (F) Color-coded UMAP clusters of cell types in the NRe-containing brain samples. (G) Stacked violin plot showing the expression of gene markers for each cluster. Figure S5. Transcriptomic analysis of barcoded NRe neurons. (A) Correlation of EGFP counts and barcode counts in EGFP positive NRe cells. (B) Histogram showing distribution of selectivity index between barcode(d) and barcode(v) in individual NRe cells. Red line indicates the kernel density estimation. (C) Scatter plot showing the count of two types of barcodes in each cell. NRe vCA1d and NRe vCA1v cells as well as vCA1d/vCA1v co-projecting NRe cells (NRe co ) were shown in magenta, cyan, and grey, respectively. (D) Volcano plot showing DEGs between NRe vCA1d and NRe vCA1v cells. (E) GO terms enriched in NRe vCA1d (left) and NRe vCA1v (right) neurons. Figure S6. The impact of NRe vCA1d neurons on context discriminability of vCA1d neurons in baseline condition. (A) Schematic diagram showing the behavioral protocol: animals were exposed to context C, B and A with CNO or saline injection. (B) Pie chart showing proportion of discriminative and non-discriminative cells between context A and B. Saline group (n = 1942 cells) vs. CNO group (n = 1501 cells), Chi-square test. (C) Same Pie chart as (B) showing discriminative and non-discriminative cells between context A and C. (D) Summary of the discriminative index for cells analyzed in (B) and (C). Saline vs. CNO group, unpaired t -test. (E) Summary of correlation index reflecting temporal correlation between context A- and B-evoked neural activity, or between context A- and C-evoked neural activity. Saline group (n = 1942 cells) vs. CNO group (n = 1501 cells), Wilcoxon rank-sum test, n.s. p > 0.05. (F) Summary of decoding accuracy of classifying context identity based on neural activity in saline group (grey) and CNO group (orange). Wilcoxon rank-sum test, n.s. p > 0.05. (G) Box plots showing Mahalanobis distance between context A and B, or between context A and C. Wilcoxon rank-sum test, ****p < 0.0001. Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7828569","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":530170640,"identity":"be4a163a-2ecd-45c9-9f4d-62e2e6de5514","order_by":0,"name":"Chun Xu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAr0lEQVRIiWNgGAWjYDACCQglx0ayFmPStSQ2EK2Df3aP2QPGtsPpfey9Bxh+1DDImxO05M4ZcwOGM4dz23jOJTD2HGMw3EnIPgOJHDMJhgqgFokcAwbeBoYEgwNEaTE4nM4G1ML4l3gtFYcTQFqYibJF4kZamUTCmXTDNp4zBodljkkYbiCkhX9G8jaJj23W8vLtPYYP39TYyBO0BQwSoPQBeDSNglEwCkbBKKAMAACBbjUYG+267gAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0001-5149-8257","institution":"Institute of Neuroscience, State Key Laboratory of Neuroscience, Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences","correspondingAuthor":true,"prefix":"","firstName":"Chun","middleName":"","lastName":"Xu","suffix":""},{"id":530170641,"identity":"b078e322-a772-405e-9fb6-999963b1e86b","order_by":1,"name":"Qingge Wu","email":"","orcid":"","institution":"Institute of Neuroscience, CEBSIT, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Qingge","middleName":"","lastName":"Wu","suffix":""},{"id":530170642,"identity":"f8504948-00bb-4947-9d7e-ea1086e5f160","order_by":2,"name":"Yumian Li","email":"","orcid":"","institution":"Institute of Neuroscience, CEBSIT, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Yumian","middleName":"","lastName":"Li","suffix":""},{"id":530170643,"identity":"9121439c-c98f-443e-97c7-2dfd3f6c3a24","order_by":3,"name":"Shishuo Chen","email":"","orcid":"","institution":"1Institute of Neuroscience, Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences, Shanghai 200031, China","correspondingAuthor":false,"prefix":"","firstName":"Shishuo","middleName":"","lastName":"Chen","suffix":""},{"id":530170644,"identity":"08600844-8ff7-4fdf-856d-288df047334f","order_by":4,"name":"Zhou Sun","email":"","orcid":"","institution":"Institutes of Brain Science, Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Zhou","middleName":"","lastName":"Sun","suffix":""},{"id":530170645,"identity":"ed50aee3-d27b-451e-aac3-b6055cf72bcd","order_by":5,"name":"Bingqing Zhao","email":"","orcid":"","institution":"Institute of Neuroscience, CEBSIT, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Bingqing","middleName":"","lastName":"Zhao","suffix":""},{"id":530170646,"identity":"4e71ffc1-1bdb-445b-a36d-348b222b1468","order_by":6,"name":"Yiming Huang","email":"","orcid":"","institution":"Institute of Neuroscience, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Yiming","middleName":"","lastName":"Huang","suffix":""},{"id":530170647,"identity":"16a277b1-0492-4ff2-9198-f2ee5a0a87cc","order_by":7,"name":"Guangling Wang","email":"","orcid":"https://orcid.org/0000-0001-7120-3110","institution":"Institute of Neuroscience, Chinese Academy of Sciences; University of Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Guangling","middleName":"","lastName":"Wang","suffix":""},{"id":530170648,"identity":"06c34565-b941-49ea-8ed8-98e300e15714","order_by":8,"name":"Mingpo Yang","email":"","orcid":"","institution":"Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Mingpo","middleName":"","lastName":"Yang","suffix":""},{"id":530170649,"identity":"d2d67c9d-3d61-40c3-a88d-ec5c54d32a03","order_by":9,"name":"Runjiao Zhang","email":"","orcid":"","institution":"Institute of Neuroscience, CEBSIT, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Runjiao","middleName":"","lastName":"Zhang","suffix":""},{"id":530170650,"identity":"918a39b3-a982-4184-8c48-cd08535ec3ab","order_by":10,"name":"Yimu Zhang","email":"","orcid":"","institution":"Institutes of Brain Science, Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Yimu","middleName":"","lastName":"Zhang","suffix":""},{"id":530170651,"identity":"2e8b5956-fcfc-4281-a001-fb321c096c37","order_by":11,"name":"Yuejun Chen","email":"","orcid":"https://orcid.org/0000-0002-4625-2604","institution":"Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Yuejun","middleName":"","lastName":"Chen","suffix":""},{"id":530170652,"identity":"24de978c-b898-4ea6-ad72-72df5a69f468","order_by":12,"name":"Xiangning Li","email":"","orcid":"https://orcid.org/0000-0002-3747-2824","institution":"Hainan University","correspondingAuthor":false,"prefix":"","firstName":"Xiangning","middleName":"","lastName":"Li","suffix":""},{"id":530170653,"identity":"b13af92e-293d-42d3-95b0-6ee477a0820a","order_by":13,"name":"Hui Gong","email":"","orcid":"https://orcid.org/0000-0001-5519-6248","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Hui","middleName":"","lastName":"Gong","suffix":""},{"id":530170654,"identity":"35cfd19a-bd55-4f19-ab89-a0f4d3581748","order_by":14,"name":"Yidi Sun","email":"","orcid":"https://orcid.org/0000-0002-4191-2917","institution":"Institute of Neuroscience, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Yidi","middleName":"","lastName":"Sun","suffix":""},{"id":530170655,"identity":"249a1265-0431-47bd-ab49-8f31ba972386","order_by":15,"name":"Yu Gu","email":"","orcid":"https://orcid.org/0000-0001-7882-5422","institution":"Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Gu","suffix":""},{"id":530170656,"identity":"a396e89c-7cb4-4e73-84f0-1ae2f8f531f2","order_by":16,"name":"Jing Liu","email":"","orcid":"","institution":"Institute of Neuroscience, CEBSIT, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Liu","suffix":""},{"id":530170657,"identity":"1bfe3be1-add3-4fa8-950e-c8d22f57dd7b","order_by":17,"name":"Hua He","email":"","orcid":"","institution":"Third Affiliated Hospital of Navy Military Medical University","correspondingAuthor":false,"prefix":"","firstName":"Hua","middleName":"","lastName":"He","suffix":""}],"badges":[],"createdAt":"2025-10-10 15:31:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7828569/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7828569/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97701948,"identity":"83c5e607-c191-4f07-9440-c2fe25887798","added_by":"auto","created_at":"2025-12-08 12:29:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1662396,"visible":true,"origin":"","legend":"\u003cp\u003eBrain-wide single-neuron projectomes of NRe neurons\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Left, the fluorescence images of a NRe section expressing EGFP sparsely in soma and axon arbors. Right images were higher-resolution images for square boxes in the left. Scale bars: left 100 μm, right 10 μm. Right, visualization of single-neuron projectomes of randomly color-coded NRe neurons (n = 303).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Top, hierarchical clustering of NRe neurons based on digitalized axon routes for each projection pattern (see Methods). STR, stratum; OLF, olfaction areas; TH, thalamus; BS, brain stem; HY, hypothalamus; CTXsp, cortical subplate; PAL, pallidum; ISO, isocortex; ENT, entorhinal cortex; HIP, hippocampus. Middle, the total length of axon arbors in each target brain regions (normalized by maximum, color-coded with scale bar on right). Bottom, the cell number for each projectome subtype.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Example neurons for six projectome cell types of NRe neurons. White arrows indicated the location of cell bodies. Axon arbors in text-annotated brain areas were shown in same colors as texts.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Hierarchical clustering of Pearson correlation coefficients (color-coded, scale bar on top) between axonal arbor lengths (projection strength) in different target pairs for NRe neurons. Downstream target areas were clustered into 10 correlated projection patterns (dashed boxes, numbered at far left). Major brain areas were annotated with colors on left (legend on top).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Histograms showing the number of neurons grouped by the selectivity index including ILA vs. PL (top), ENT vs. HIP (middle), and PFC vs. HIP (bottom), respectively (±1 indicate maximal selectivity, see Methods for definition).\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/7dc9c0147a21f9eb8256f0ab.png"},{"id":97701950,"identity":"4a7e68a4-dc7b-4c12-a0f2-4090b1ed75e2","added_by":"auto","created_at":"2025-12-08 12:29:25","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1726203,"visible":true,"origin":"","legend":"\u003cp\u003eDistinct axon projections from NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons to vCA1 subregions.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Hierarchical clustering of HIP-projecting NRe neurons based on their axon arborization patterns. Each vertical line represents one neuron. The projection strength (total length of axon arbors) of each neuron in HIP subregions was normalized by the maximal length of axon arbors within the 5 HIP subregions (color-coded scale bar on left, bar graphs summarizing projection strengths in each subregion on right). The 6 HIP projection subgroups with different target patterns were represented by colors and numbers shown below.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Dot plot showing the intra- and extra projection patterns of 6 HIP projection subgroups. The projection strength in each target brain area was normalized by the max strength in each subgroup and color-coded (scale bar on right).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) The mean projection strength in HIP subregions along hippocampal D-V axis (filtered by Fourier transform \u0026lt; 15Hz) from HIP-projecting NRe neurons (n = 161 cells).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) The K-Means clustering of CA1-projecting NRe neurons based on the main principal component of 3D coordinates of their soma locations in NRe and the center-of-mass of axonal terminals in CA1. Cross error bars indicate the standard deviation within three clusters (cluster 1 in red, 44 cells; cluster 2 in blue, 71 cells; cluster 3 in green, 26 cells).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Top, histogram plot of the number of axon tips along the hippocampal D-V axis from NRe neurons in three clusters. Bottom, stacked bar plot showing the total number of CA1 projectors and the percentage of three clusters in each NRe projectome cell types.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eF\u003c/strong\u003e) Examples of axon arbors from three NRe neurons in cluster 1, 2 and 3, respectively. Each panel displayed axonal arbors within a 200-µm block of coronal hippocampal sections (Bregma coordinates on top) in the Allen mouse common coordinate Framework.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG\u003c/strong\u003e) Top, mean projection strength along hippocampal D-V axis from NRe neurons in cluster 1 (red, n = 44 cells) and cluster 2 (blue, n = 71 cells). Bottom, heat map showing the projection strength (normalized by the row max) along D-V axis for axon arbors of each neuron (in one row) in cluster 1 and 2. Neurons were sorted by the D-V coordinate of their maximal projection strength.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eH\u003c/strong\u003e) Bar graphs summarizing the projection strength (mean ± SEM) in various target areas by NRe neurons in cluster 1 and 2, respectively. Unpaired \u003cem\u003et\u003c/em\u003e-test between cluster 1 and 2 in each target, *p \u0026lt; 0.05, **p\u0026lt;0.01, n.s. p \u0026gt; 0.05.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eI\u003c/strong\u003e) Scheme showing stereotaxic injections of CTB647 and Fluorogold into vCA1d and vCA1v and the example picture showing their fluorescence in a hippocampal section. Scale bar, 500 μm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eJ\u003c/strong\u003e) Examples of coronal sections at three Bregma coordinates showing the fluorescence of CTB647 and Fluorogold-labeled NRe neurons. Scale bar: 100 μm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eK\u003c/strong\u003e) Pie chart summarizing the percentage of NRe neurons projecting to vCA1d, vCA1v, or both.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/b39a432a5909316806913d01.png"},{"id":97894787,"identity":"0bbd159a-8f7c-4add-b230-b49d279967fe","added_by":"auto","created_at":"2025-12-10 15:33:01","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":727218,"visible":true,"origin":"","legend":"\u003cp\u003eDistinct behavioral roles of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons in contextual memory.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Scheme illustrating the behavioral paradigm, in which CNO was injected before the conditioning.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e and \u003cstrong\u003eD\u003c/strong\u003e) Left, scheme illustrating experimental design for retrograde labeling of NRe\u003csup\u003evCA1d\u003c/sup\u003e (B) and NRe\u003csup\u003evCA1v\u003c/sup\u003e (D) neurons with hM4Di-mCherry or mCherry. Right, example pictures showing Cre-dependent fluorescent expression of hM4Di-mCherry and mCherry (Bottom) in NRe, and fluorescent signals of blue beads (arrows) co-injected into vCA1d. Scale bar, 100 μm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e and \u003cstrong\u003eE\u003c/strong\u003e) Left, summary of post-CTC freezing in context A (conditioned) and context B (neutral) for animals with labeling of NRe\u003csup\u003evCA1d\u003c/sup\u003e (C) and NRe\u003csup\u003evCA1v\u003c/sup\u003e (E) neurons, respectively. Right, summary of generalization index based on the freezing in conditioned and neutral contexts. Panel C: control group, N = 11 mice; hM4Di group, N = 9 mice. Panel E: control group, N = 18 mice; hM4Di group, N = 13 mice. Unpaired \u003cem\u003et\u003c/em\u003e-test, *p \u0026lt; 0.05.**p \u0026lt; 0.01. n.s. p \u0026gt; 0.05.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/15c748946bc96783aba9d08c.png"},{"id":97894852,"identity":"930e9170-3428-481b-be52-9ea737a0371c","added_by":"auto","created_at":"2025-12-10 15:33:09","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":940552,"visible":true,"origin":"","legend":"\u003cp\u003eDistinct ion channel expressions in NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Schematic diagram showing the workflow for microdissection of NRe from brain slices in AAV-barcode injected mice, single-cell isolation and RNA sequencing by 10x Genomics.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Uniform manifold approximation and projection (UMAP) embedding showing 7 subclusters of NRe neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Violin plots showing the expression of markers for 7 NRe clusters.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) UMAP embedding of GFP positive (blue, GFP counts \u0026gt; 0) and negative cells (grey).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Upper panel, bar plot describing the cell count for GFP positive and negative NRe cells. Lower panel, stacked bar plot showing the ratio of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons as well as vCA1d/vCA1v co-projecting (vCA1\u003csup\u003eco\u003c/sup\u003e) neurons in each cluster of NRe neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eF\u003c/strong\u003e) Dot plot showing the average expression of Scn, Kcn, Cacna, HCN family in NRe\u003csup\u003evCA1d \u003c/sup\u003eand NRe\u003csup\u003evCA1v \u003c/sup\u003esubpopulations.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG\u003c/strong\u003e) Dot plot showing the average expression of ACH, GABA, Cacna, Hcn family in NRe\u003csup\u003evCA1d \u003c/sup\u003eand NRe\u003csup\u003evCA1v \u003c/sup\u003esubpopulations.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/8a60624ab3b98e5b2bc744e2.png"},{"id":97894302,"identity":"0550f596-fb64-4595-b296-89fdec4d1f12","added_by":"auto","created_at":"2025-12-10 15:32:16","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1362902,"visible":true,"origin":"","legend":"\u003cp\u003eDistinct electrophysiological properties of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e and \u003cstrong\u003eB\u003c/strong\u003e) From left to right, the four panels represent the injection site of red retrobeads in the vCA1d (A) and vCA1v (B), the patch clamp recording site, the recorded cells and red retrobeads labeled NRe cells in the boxed area of the 2\u003csup\u003end\u003c/sup\u003e panel. Scale bars:1000 μm, 200 μm, 10 μm, 10 μm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e and \u003cstrong\u003eD\u003c/strong\u003e) Example traces of a NRe\u003csup\u003evCA1d\u003c/sup\u003e neuron (C) and a NRe\u003csup\u003evCA1v\u003c/sup\u003e neuron (D) upon injected currents (square pulses) of 0 pA, 30 pA, and 70 pA, respectively. The example waveform of a single spike was shown in the right. Dash lines indicated the baseline.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Summary of spike numbers evoked by various injected currents in NRe\u003csup\u003evCA1d\u003c/sup\u003e (pink, n = 17 cells from 5 mice) and NRe\u003csup\u003evCA1v\u003c/sup\u003e (cyan, n = 8 cells from 2 mice) neurons. 2way-ANOVA, ****p \u0026lt; 0.0001.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eF\u003c/strong\u003e) Summary of spike thresholds of same neurons recorded in (E). Unpaired \u003cem\u003et\u003c/em\u003e-test, p = 0.0863.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG\u003c/strong\u003e) Summary of spike amplitude of same neurons recorded in (E). Unpaired \u003cem\u003et\u003c/em\u003e-test, p = 0.0338.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eH\u003c/strong\u003e) Left panel, scheme illustrating the hyperpolarization and depolarization components of the spike waveform and the quantification of these two components (W2:W1) in whole-cell recording. Dash line indicated the baseline. Right panel, summary of the ratio of W2 to W1 in same neurons shown in (E). Unpaired \u003cem\u003et\u003c/em\u003e-test, p = 0.0297.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eI\u003c/strong\u003e) Upper, scheme showing the tetrode implantation strategy. Lower, the histology picture showing the tetrode recording site marked by electric lesion in NRe. Scale bar, 0.5 mm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eJ\u003c/strong\u003e) Scatter plot showing the peak-valley value in tetrode 3 and 4 for spikes of 4 sorted units and unsorted units.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eK\u003c/strong\u003e) Summary of unit isolation in 26 tetrodes from which single units were recorded (n = 92 cells, N = 4 mice). High J3 and low DB indicate good isolation.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eL\u003c/strong\u003e) Schematic diagram depicting 4 waveform features including end slop, peak2/peak1 ratio, peak1-to-peak2 amplitude, and half width of peak 1. The negative-going peak 1 corresponded to the depolarizing component of the in vivo spike waveform, whereas positive-going peak 2 corresponded to the hyperpolarizing component.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eM\u003c/strong\u003e) The t-SNE plot of 2 clusters after DBSCAN clustering of 4 waveform features shown in (L). Cluster 1, 52 neurons. Cluster 2, 40 neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eN\u003c/strong\u003e) Scheme illustrating the hyperpolarization and depolarization components of the spike waveform and the quantification of these two components (W2:W1) in single-unit recording.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eO\u003c/strong\u003e) Left, summary of the ratio of W2 to W1 in cluster 1 and 2 as classified in (M). C1 vs. C2, Wilcoxon rank-sum test, ****p \u0026lt; 0.0001. Right, spike waveform examples for 2 neurons in C1 and C2, respectively.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eP\u003c/strong\u003e) The cumulative distribution of spontaneous firing frequencies in C1 and C2 neurons. C1 vs. C2, K-S test: bellow 4Hz, n.s.; above 4Hz, *p \u0026lt; 0.05.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eQ\u003c/strong\u003e and \u003cstrong\u003eR\u003c/strong\u003e) Summary of the burst index (Q) and burst duration (R) of C1 and C2 neurons. C1 vs. C2, unpaired \u003cem\u003et\u003c/em\u003e-test, n.s.\u003c/p\u003e\n\u003cp\u003eData summary: mean ± SEM.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/2486ae7b27f54ad05dd2c769.png"},{"id":97701953,"identity":"e18a09ad-0bc9-46fb-a5b0-e0eeb84b92c3","added_by":"auto","created_at":"2025-12-08 12:29:26","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1392575,"visible":true,"origin":"","legend":"\u003cp\u003eDistinct inputome to NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Scheme illustrating experimental protocols for cell-type specific retrograde trans-synaptic tracing from NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons. Retrograde Cre-expressing AAV was injected into vCA1d or vCA1v and Cre-dependent helper AAV was injected into NRe, and then EnVA-pseudotyped rabies vectors were injected into NRe.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Example fluorescent images of NRe with rabies-based tracing from NRe\u003csup\u003evCA1d\u003c/sup\u003e (upper panels) and NRe\u003csup\u003evCA1v\u003c/sup\u003e (lower panels) neurons. Left, overlaid fluorescence of TVAmCherry and rabies-GFP. Right, higher-resolution images (TVAmCherry in red, Rabies in green) of square boxes on the left. Arrows indicate representative starter cells co-expressing mCherry and GFP. Scale bars: left, 100 μm; right, 20 μm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Summary of the number of rabies-labeled presynaptic cells to NRe neurons in various brain areas. Abbreviations were defined in Allen Mouse Brain Common Coordinate Framework (CCF). \u0026nbsp;Red bars, presynaptic cells to NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons, N = 4 mice. Blue bars, presynaptic cells to NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, N = 4 mice. Red vs. Blue in each brain area, Wilcoxon rank-sum test: n.s., p \u0026gt; 0.05, *p \u0026lt; 0.05, **p \u0026lt; 0.01, ***p \u0026lt; 0.001.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Example fluorescent images of rabies-labeled cells in various brain regions based on tracing from NRe\u003csup\u003evCA1d\u003c/sup\u003e (upper panels) and NRe\u003csup\u003evCA1v\u003c/sup\u003e (lower panels) neurons. Scale bars: 100 μm.\u003c/p\u003e\n\u003cp\u003eData summary: mean ± SEM.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/4051b8e76204b064f4e47d16.png"},{"id":97895064,"identity":"12662508-20f8-4510-a214-03a31d03a5d7","added_by":"auto","created_at":"2025-12-10 15:33:28","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":427248,"visible":true,"origin":"","legend":"\u003cp\u003eLearning-reduced context discriminability in putative NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Schematic diagram showing CTC protocol: foot shocks were delivered in context A, which was more similar to B than C.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Summary of post-CTC freezing in conditioned context (A) and neutral context (B and C) for animals with single-unit recording from NRe. One-way ANOVA with Sidak's multiple comparisons test: A vs. B, *p = 0.0168; Ctx. A vs. C, p = 0.0523.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) The t-SNE plot showing two clusters using DBSCAN clustering of NRe neurons based on the same spike waveform features as Figure 5M using. Cluster 1 (C1, n = 33 neurons) were shown in red, and cluster 2 (C2, n = 57 neurons) were shown in blue.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Summary of the ratio of half spike peak 2 width (W2) to half spike peak 1 width (W1) of NRe neurons in cluster 1 and 2. Unpaired \u003cem\u003et\u003c/em\u003e-test with Welch's correction, ****p \u0026lt; 0.0001. Based on the Compared to Figure 5L. NRe neurons in cluster 1 (red) were putative NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons, and those in cluster 2 (blue) were putative NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Summary of the discrimination index between context A and B (left) and between context A and C (right). Neuron numbers were shown on the bar plot. Unpaired \u003cem\u003et\u003c/em\u003e-test was used for C1 (red). Wilcoxon rank-sum test was used for C2 (blue).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eF\u003c/strong\u003e) Summary of the Mahalanobis distance in the coding space between context A and B (left) and between context A and C (right). Neuron numbers were same as (E). shown on the bar plot. Wilcoxon rank-sum test was used for both C1 (red) and C2 (blue). Circles represented the mean of the data.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/b7c87766560db5f1658bd75f.png"},{"id":97894627,"identity":"de4e3dac-88ba-419c-bc64-e9eaa030b951","added_by":"auto","created_at":"2025-12-10 15:32:49","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1395238,"visible":true,"origin":"","legend":"\u003cp\u003eLearning-reduced context discriminability of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons at single-cell level.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Schematic diagrams illustrating head-mounted miniscope, GRIN lens implanted above vCA1d, and the viral injection strategy for Ca\u003csup\u003e2+\u003c/sup\u003e imaging of vCA1d neurons and chemogenetic inhibition of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Left, fluorescent image of hM4Di-mCherry expression in NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons. Scale bar, 100 μm. Middle, GCaMP6f fluorescence in vCA1d underneath the GRIN lens. Scale bars, 500 μm. Right, magnified view of boxed area in the middle panel. Scale bar, 50 μm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Scheme illustrating CTC protocols (left) and context designs (right) combined with miniscope Ca\u003csup\u003e2+\u003c/sup\u003e imaging.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Summary of post-CTC freezing in three contexts, in which A is the conditioned context. Animals with hM4Di-expressing NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons were separated into two groups. The control group was administrated with saline (N = 5 mice) and the other with CNO (N = 5 mice). Unpaired \u003cem\u003et\u003c/em\u003e-test for freezing, *p \u0026lt; 0.05, **p \u0026lt; 0.01, n.s. p \u0026gt; 0.05.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e and \u003cstrong\u003eF\u003c/strong\u003e) Left panels, example traces showing the Ca\u003csup\u003e2+\u003c/sup\u003e activity in context A (green), context B (red) and context C (orange) for three cells with different contextual preferences (legend on top). Right panels, the cell map illustrating the spatial locations of three cell types with different contextual preferences.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG\u003c/strong\u003e and \u003cstrong\u003eH\u003c/strong\u003e) Upper panels, pie charts showing the number of context discriminative cells versus non-discriminative cells based on their Ca\u003csup\u003e2+\u003c/sup\u003e activity, in contexts A and B (G), as well as in contexts A and C (H). The proportion of non-discriminative cells significantly increased after CTC in the saline group (Pre vs. Post, Chi-square test, *p \u0026lt; 0.05), but not in the CNO group (pre vs. post, Chi-square test, p \u0026gt; 0.05 based on Ctx A vs. B, ****P \u0026lt; 0.0001, significant decrease based on Ctx A vs. C). Lower left panels, summary of the discriminative index for vCA1d neurons during pre- and post-CTC context exposures (grey, saline group; orange, CNO group) based on their Ca\u003csup\u003e2+\u003c/sup\u003e activity, in contexts A and B as well as in contexts A and C. Lower right panels, summary of CTC-induced changes of the discriminative index in saline and CNO groups shown in left panels (Saline, n = 2420 cells, CNO, n = 1099 neurons).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eI\u003c/strong\u003e) Schematic diagram showing that context-evoked Ca\u003csup\u003e2+\u003c/sup\u003e signals showing high and low temporal dynamic correlation.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eJ\u003c/strong\u003e) The summary of temporal dynamic correlation index during pre- and post-CTC context exposures (grey, saline group; orange, CNO group) based on their Ca\u003csup\u003e2+\u003c/sup\u003e activity, in contexts A and B (left), as well as in contexts A and C (right). Wilcoxon rank-sum test, *p \u0026lt; 0.05, **p \u0026lt; 0.01, ***p \u0026lt; 0.001, n.s. p \u0026gt; 0.05.\u003c/p\u003e\n\u003cp\u003eData summary: mean ± SEM.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/43b58090f3c44daf5755ddd1.png"},{"id":97701957,"identity":"ebc778f4-d8ed-427e-8acc-9cf28bdc4a13","added_by":"auto","created_at":"2025-12-08 12:29:26","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":457217,"visible":true,"origin":"","legend":"\u003cp\u003eLearning-reduced context discriminability of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons at population level.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e and \u003cstrong\u003eD\u003c/strong\u003e) PCA-based 2D projection of population vectors for vCA1d neurons during pre- and post-CTC context exposures in saline (panel A) and CNO (panel D) groups.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e and \u003cstrong\u003eC\u003c/strong\u003e) The probability density (resampling n = 5000) of Mahalanobis distance in the coding space between context A and B (B) and between context A and C (C). Animals were treated with saline during CTC. Pre-CTC data (dashed line) vs. post-CTC data (solid line), Kolmogorov-Smirnov test, ****P \u0026lt; 0.0001.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e and \u003cstrong\u003eF\u003c/strong\u003e) The same plots for Mahalanobis distance as (B) and (C) for animal treated with CNO during CTC.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG \u003c/strong\u003eand \u003cstrong\u003eH\u003c/strong\u003e) Box plots showing CTC-induced changes of Mahalanobis distance between context A and B (G) and between context A and C (H), based on the results in panel B – F. Saline group vs. CNO group, Wilcoxon rank-sum test, ****p \u0026lt; 0.0001.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eI\u003c/strong\u003e) Summary of decoding accuracy for context identify using neural activity during pre- and post-CTC context exposures in saline (grey) and CNO (orange) groups. Wilcoxon rank-sum test was used for group comparisons.\u003c/p\u003e\n\u003cp\u003eData summary: mean ± SEM.\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/706d0e7741449423edd7b94b.png"},{"id":97902555,"identity":"4eb6aa9b-ba2b-494e-9f49-06e6b05505e6","added_by":"auto","created_at":"2025-12-10 15:52:50","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":11190796,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/336fd551-3186-42d4-8f57-ebbceabd530d.pdf"},{"id":97701951,"identity":"f57f27ef-dac0-40d2-b031-24e189a32133","added_by":"auto","created_at":"2025-12-08 12:29:25","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1964819,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Figures\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFigure S1. Classification of axon routes and projectome cell types of NRe neurons\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Schematic diagram showing the procedure of axon routes classification: reconstruction, embedding, clustering, and refinement.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Examples showing axon trajectories of 8 clusters of axon routes.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) The heatmap showing the axon projection strength of 8 axon route clusters in various brain areas. The strength was normalized by the max strength among all brain areas and coded by the color bar on the right.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Schematic diagram showing the procedure of projectome cell typing: categorizing neurons based on their projection patterns, digitalizing a total of 150 projection patterns using digitalized axon route clusters, and clustering digitalized projection patterns.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Summary of silhouette score (upper) and Davies Bouldin score (lower) as a function of the number of projection cell types. The number of projectome cell types (N = 6) corresponded to the red line reflecting the local maxima of the silhouette score curve and the local minimal of the Davies Bouldin score curve.\u003c/p\u003e\n\u003cp\u003e(F) Upper, heatmap showing clustered projection patterns (separated by lines) based on the representation of each projection pattern using counts of each axon route cluster. Lower, heatmap showing the axon projection strength in target brain regions. Heatmaps were normalized by the min-max and color coded with scale bars on the right.\u003c/p\u003e\n\u003cp\u003eFigure S2. NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons promote the expression of threat generalization.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Schematic diagram illustrating virus injection and tetrodes recording in NRe.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Experimental protocol showing that recording started right after saline or CNO injection. On the next day, same recording was repeated again with the other injection.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Example pictures showing fluorescent expression of hM4Di-mCherry and the recording site marked by electrical lesion. Scale bar, 200 μm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Histogram showing firing rates of an uninhibited cell (upper) and an inhibited cell (lower) upon the injection of CNO.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Summary of cell numbers of inhibited and uninhibited neurons upon saline or CNO injections (8 sessions for saline group, 8 sessions for CNO group, N = 4 mice). The inhibited neuron was determined by a significant reduction of firing rate from 0-10 min to 30-40 min post saline or CNO injection. The proportion of inhibited cells was significantly higher in the CNO group (57/98 vs. 40/95, Chi-square test, p = 0.0257).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eF\u003c/strong\u003e) Summary of the firing rate of all recorded NRe cells during the 60-min period post injection of saline (n = 95) or CNO (n = 98). Kruskal-Wallis test, p \u0026lt; 0.0001. Dunn's multiple comparisons test at various time periods.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG\u003c/strong\u003e) Schematic diagram illustrating behavioral protocol with CNO treatment during threat memory retrieval in context B and A.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eH\u003c/strong\u003e) Reproduction of the viral expression strategies shown in Figure 3B.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eI\u003c/strong\u003e) Summary of time spent in freezing during threat memory retrieval in context A and B (left), and generalization index (right) for animals with NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons labeled. Unpaired \u003cem\u003et\u003c/em\u003e-test, control group (N = 11) vs. hM4Di group (N = 9).\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eJ\u003c/strong\u003e) Reproduction of the viral expression strategies shown in Figure 3D.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eK\u003c/strong\u003e) The same plots as (I) for animals with NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons labeled. Unpaired \u003cem\u003et\u003c/em\u003e-test, control group (N = 18) vs. hM4Di group (N = 13).\u003c/p\u003e\n\u003cp\u003eData summary: mean ± SEM. Statistics: two-tailed. P values: ***p \u0026lt; 0.001, **p \u0026lt; 0.01, *p \u0026lt; 0.05, n.s., p \u0026gt; 0.05.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Figure S3. Summary of viral injection sites and tetrode recording sites.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e and \u003cstrong\u003eB\u003c/strong\u003e) Fluorescent signals of blue beads in vCA1d (A) and vCA1v (B) and reconstructed positions of co-injected blue beads in animals used for chemogenetic experiments. Scale bars in histology picture and zoom in: 0.5 mm and 0.1 mm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Positions of tetrode recording sites.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Figure S4. Spatial and single-cell transcriptome analysis in NRe.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Volcano plot showing DEGs between NRe (area defined by histology) and brain areas surrounding NRe in thalamus and hypothalamus. Dots were colored if |Log2 fold change| \u0026gt; 1 and p value \u0026lt; 0.05.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Dot plot showing the expression of DEGs in NRe compared to thalamic areas surrounding NRe and hypothalamus.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Spatial expression patterns of NRe marker genes. Scale bars, 0.5 mm.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Violin plots showing the number of detected genes, UMI counts, and percent of mitochondrial genes in each snRNA-seq-defined cell cluster.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) The UMAP embeddings showing single-gene expression patterns.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eF\u003c/strong\u003e) Color-coded UMAP clusters of cell types in the NRe-containing brain samples.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG\u003c/strong\u003e) Stacked violin plot showing the expression of gene markers for each cluster.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFigure S5. Transcriptomic analysis of barcoded NRe neurons.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Correlation of EGFP counts and barcode counts in EGFP positive NRe cells.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Histogram showing distribution of selectivity index between barcode(d) and barcode(v) in individual NRe cells. Red line indicates the kernel density estimation.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Scatter plot showing the count of two types of barcodes in each cell. NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e cells as well as vCA1d/vCA1v co-projecting NRe cells (NRe\u003csup\u003eco\u003c/sup\u003e) were shown in magenta, cyan, and grey, respectively.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Volcano plot showing DEGs between NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v \u003c/sup\u003ecells.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) GO terms enriched in NRe\u003csup\u003evCA1d\u003c/sup\u003e (left) and NRe\u003csup\u003evCA1v \u003c/sup\u003e(right) neurons.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Figure S6. The impact of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons on context discriminability of vCA1d neurons in baseline condition.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eA\u003c/strong\u003e) Schematic diagram showing the behavioral protocol: animals were exposed to context C, B and A with CNO or saline injection.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eB\u003c/strong\u003e) Pie chart showing proportion of discriminative and non-discriminative cells between context A and B. Saline group (n = 1942 cells) vs. CNO group (n = 1501 cells), Chi-square test.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eC\u003c/strong\u003e) Same Pie chart as (B) showing discriminative and non-discriminative cells between context A and C.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eD\u003c/strong\u003e) Summary of the discriminative index for cells analyzed in (B) and (C). Saline vs. CNO group, unpaired \u003cem\u003et\u003c/em\u003e-test.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eE\u003c/strong\u003e) Summary of correlation index reflecting temporal correlation between context A- and B-evoked neural activity, or between context A- and C-evoked neural activity. Saline group (n = 1942 cells) vs. CNO group (n = 1501 cells), Wilcoxon rank-sum test, n.s. p \u0026gt; 0.05.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eF\u003c/strong\u003e) Summary of decoding accuracy of classifying context identity based on neural activity in saline group (grey) and CNO group (orange). Wilcoxon rank-sum test, n.s. p \u0026gt; 0.05.\u003c/p\u003e\n\u003cp\u003e(\u003cstrong\u003eG\u003c/strong\u003e) Box plots showing Mahalanobis distance between context A and B, or between context A and C. Wilcoxon rank-sum test, ****p \u0026lt; 0.0001.\u003c/p\u003e","description":"","filename":"SupplementaryFigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-7828569/v1/c48785ef4f76c77d2267ef93.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"A thalamo-hippocampal circuit regulating memory precision during contextual learning","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMemory enables animals to express adaptive behaviors in a volatile world. Generalized threat memories can protect animals from potential dangers by responding appropriately to a novel environment that resembles a previously harmful experience, but overgeneralized threat memories may lead to maladaptive behaviors and have been linked to anxiety- and stress-related disorders, as evidenced by post-traumatic stress disorders (PTSD) in humans\u003csup\u003e\u003cspan additionalcitationids=\"CR2 CR3 CR4 CR5\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e and PTSD-like animal models\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Memory precision control is thus essential for the balance between memory specificity and generalization\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e, thereby supporting animals\u0026rsquo; survival and well-being.\u003c/p\u003e\u003cp\u003eThe hippocampus (HIP) and medial prefrontal cortex (mPFC) are crucial for memory precision control\u003csup\u003e\u003cspan additionalcitationids=\"CR10 CR11 CR12 CR13 CR14 CR15\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Recent studies showed that NRe, a thalamic nucleus and key bridge linking HIP and mPFC\u003csup\u003e\u003cspan additionalcitationids=\"CR18 CR19\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e, played key roles in memory precision control\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e, in addition to a variety of memory-related processes including working memory\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, spatial memory and navigation\u003csup\u003e\u003cspan additionalcitationids=\"CR24 CR25 CR26 CR27\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e, memory consolidation\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e, and threat memory extinction\u003csup\u003e\u003cspan additionalcitationids=\"CR31\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. Such functional diversity of NRe neurons may have to do with their diverse circuit connectivity\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e and cellular makers\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e and firing properties\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e, requiring a systematic understanding of brain-wide connectivity of NRe neurons at the single-neuron level. Prior studies suggest that NRe neurons control contextual memory precision\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e, possibly by regulating neuronal activity in HIP\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Yet, it remains elusive how NRe neurons regulate neural activity in HIP during memory precision control.\u003c/p\u003e\u003cp\u003eHere we leveraged single-neuron projectome reconstruction and analysis to delineate brain-wide axon projections of NRe neurons and identified two subpopulations of NRe neurons preferentially projecting to vCA1d and vCA1v, respectively. By combining retrograde tracing with chemogenetic manipulations, miniscope Ca\u003csup\u003e2+\u003c/sup\u003e imaging, \u003cem\u003ein vitro\u003c/em\u003e slice recording, \u003cem\u003ein vivo\u003c/em\u003e single-unit recording and cell-type specific rabies transsynaptic tracing, we found systematic distinctions between NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons at molecular, cellular, circuit, and behavioral levels. We further showed that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons promoted threat generalization, via the downregulation of contextual discriminability of both themselves and their downstream vCA1d neurons following contextual threat learning, revealing a regulatory role of a thalamo-hippocampal circuit in memory precision control.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eDiversity of whole-brain axon projections from NRe neurons\u003c/h2\u003e\u003cp\u003eTo understand the diversity and spatial organization of axon projections from NRe neurons, we sparsely labeled NRe neurons by a mixture of adeno-associated virus (AAV)\u0026ndash;expressing Cre-dependent enhanced green fluorescent protein (EGFP) and low-titer AAV\u0026ndash;expressing Cre under a human synapsin promoter\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e, and then imaged their whole-brain axon projections by fluorescence Micro-Optical Sectioning Tomography\u003csup\u003e\u003cspan additionalcitationids=\"CR36 CR37\" citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eA, see Methods). We reconstructed axon projections and soma of 303 NRe neurons from 8 mice\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e, and identified 10 downstream target areas, including striatum (STR), olfactory area (OLF), thalamus (TH), hypothalamus (HY), cortical subplate-related areas (CTXsp), palladium (PAL), isocortex (ISO), entorhinal cortex (ENT), and HIP (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). All neurons were registered to the Allen Mouse Brain Common Coordinate Framework\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e, based on the propidium iodide (PI) staining\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eTo classify projectome cell types of NRe neurons, we first digitalized the axon routes of all neurons and reconstructed each neuron and projection pattern by axon route clusters (\u003cb\u003eFigure S1A-C\u003c/b\u003e, see Methods), and then performed unsupervised hierarchical clustering of projection patterns based on their axon route clusters (\u003cb\u003eFigure S1D-F\u003c/b\u003e, see Methods). In total, 6 projectome cell types were classified, and further grouped into 2 major classes: 4 projectome types in the first class exhibited strong axonal projections to ISO, ENT and HIP with different axon routes and strengths; 2 projectome types in the second class projected preferentially to ISO and STR, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eB and \u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). Moreover, projectome cell type 1 sent axon projections to HIP and ISO with similar strength. Types 3 and 4 sent preferential axon projections to HIP whereas type 5 preferentially projected to ISO, including anterior cingulate cortex (ACC) and mPFC (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eB).\u003c/p\u003e\u003cp\u003eThe NRe is believed to be the key bridge to connect with HIP and medial prefrontal cortex (mPFC), a part of ISO\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. Interestingly, projectome cell type 1\u0026ndash;4 projected to HIP and ISO with differential strengths. To determine whether single NRe neurons send differential projections to mPFC and HIP as well as other targets. We first investigated this by measuring the correlation coefficients and selectivity index for the axon projection strength in target areas. We identified 10 common projection patterns of NRe neurons based on the unsupervised clustering analysis of the correlation coefficients (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). The HIP subregions such as dentate gyrus (DG) and CA1 belonged to a target pattern of hippocampal formation subregions (pattern 2), whereas mPFC subregions belonged to other patterns: infralimbic cortex (ILA) in pattern 4 (together with retrosplenial cortex and anterior cingulate cortex) and prelimbic cortex (PL) in pattern 8 (together with olfactory bulb). We then measured the selectivity index for individual NRe neurons between mPFC and HIP, and found that many NRe neurons showed high selectivity for either mPFC or HIP (e.g., index \u0026minus;\u0026thinsp;1 is exclusively for mPFC, Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eE, lower panel). In contrast, NRe neurons projected to ENT and HIP with balanced strengths (i.e., selectivity index near 0, Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eE, middle panel). Thus, NRe neurons could be grouped by their axon projection strengths to brain-wide target areas, as manifested by the preferential axon projections to mPFC and HIP, respectively.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eDistinct projections from NRe and NRe neurons to CA1 subregions\u003c/h3\u003e\n\u003cp\u003ePrior studies showed that NRe neurons modulated contextual memory generalization and cFos signals in HIP\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. To understand axon projections from NRe neurons to HIP subregions, we performed unsupervised clustering analysis of the projection strength in the HIP subregions including CA1, CA2, CA3, DG, and subicular complex. We identified 6 subgroups of 161 HIP-projecting NRe neurons with distinct target patterns: subgroup 2, 4 and 6 co-projecting to multiple subregions, whereas subgroup 1,3 and 5 selectively projecting to HIP subregions of DG, CA1 and subicular complex, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Of note, NRe neurons in all 6 subgroups sent collateral projections to ENT (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eB), consistent with above results that most NRe neurons sent comparable axon projections to both HIP and ENT. Among all the HIP subregions, NRe neurons sent the strongest projections to CA1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Interestingly, axon arbors in CA1 were enriched in two subregions of ventral CA1 (vCA1) centered around 4 mm and 7 mm along the dorsal-ventral (D-V) axis, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eC), raising the possibility of separable axon projections into CA1 subregions.\u003c/p\u003e\u003cp\u003eNext, we selected the HIP-projecting NRe neurons and performed K-means clustering of principal components based on the soma and axon terminals\u0026rsquo; center of mass in CA1. Three clusters of NRe neurons were classified by distinct spatial preferences of axon terminals but not soma location (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eD and \u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eE), and were preferentially enriched in projectome cell type 3 and 4 (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eE, lower panel). Interestingly, their axon arbors were spatially enriched in the \u003cem\u003estratum molecular-lacunosum\u003c/em\u003e of CA1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eF). Notably, the axon projection of cluster1 and 2 (preferring vCA1) were much stronger than that of cluster 3 (preferring dorsal CA1), as reflected by the quantity of cells and axon tips (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eE, upper panel). Furthermore, axon arbors of cluster 1 were primarily located in vCA1d, whereas those of cluster 2 mainly in vCA1v (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eF and \u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eG). These two clusters of NRe neurons sent axon collaterals to HY, ISO, and TH with different projection strengths but to ENT and OLF with similar strengths (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eH), revealing a systematic difference in axon projections between vCA1d-projecting and vCA1v-projecting (NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e) neurons. Consistently, dual-color fluorescent retrograde tracing from vCA1d and vCA1v labeled largely separate subpopulations of NRe neurons (Fig.\u0026nbsp;\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eI-\u003cspan refid=\"Fig17\" class=\"InternalRef\"\u003e2\u003c/span\u003eK). Thus, the dissociable axon projections from NRe neurons, particularly NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, could provide the circuit basis for specific memory functions.\u003c/p\u003e\n\u003ch3\u003eDifferential regulations of contextual memory by NRe and NRe neurons\u003c/h3\u003e\n\u003cp\u003eTo probe the behavioral function of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons in memory precision control, we performed selective chemogenetic inhibition of them when animals were subjected to a contextual threat conditioning (CTC) paradigm (Fig.\u0026nbsp;\u003cspan refid=\"Fig18\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). We first validated the \u003cem\u003ein vivo\u003c/em\u003e chemogenetic inhibition\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e by tetrodes-based single-unit recording from NRe neurons after injecting AAV-hSyn-hM4Di-mCherry into NRe (\u003cb\u003eFigure S2A-C\u003c/b\u003e). Upon the administration of clozapine N-oxide (CNO), a significantly higher proportion of NRe neurons in the CNO group (~\u0026thinsp;60%) exhibited prolonged (at least one hour) inhibition compared to those in the saline group (CNO 57/98 vs. saline 40/95; Chi-square test, p\u0026thinsp;=\u0026thinsp;0.026; \u003cb\u003eFigure S2D-F\u003c/b\u003e), demonstrating a mild and long lasting chemogenetic inhibition of NRe\u0026rsquo;s neuronal activity.\u003c/p\u003e\u003cp\u003eWe went on to express hM4Di-mCherry (hM4Di group) or mCherry (control group) specifically in NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons by injecting Cre-dependent hM4Di into NRe and AAVretro-Cre into vCA1d and vCA1v, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig18\" class=\"InternalRef\"\u003e3\u003c/span\u003eB and \u003cspan refid=\"Fig18\" class=\"InternalRef\"\u003e3\u003c/span\u003eD, \u003cb\u003eFigure S3\u003c/b\u003e). Mice were administered with CNO 30 min before the CTC in context A in day 1 (see Methods) and were subjected to contextual threat retrieval in context B (neutral context) and context A (conditioned context) in day 2 and day 3, respectively. For animals with NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons labeled, animals in the hM4Di group showed similar freezing in context A, but significantly lower freezing in context B, compared to those in the control group (Fig.\u0026nbsp;\u003cspan refid=\"Fig18\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). Consequently, the generalization index (ratio of freezing in B to freezing in A) was significantly reduced in the hM4Di group, suggesting that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons facilitate the contextual memory generalization. For animals with NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons labeled, animals in the hM4Di group showed significantly lower freezing in context A but similar generalization index, compared to those in the control group (Fig.\u0026nbsp;\u003cspan refid=\"Fig18\" class=\"InternalRef\"\u003e3\u003c/span\u003eE). These results suggest that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons are promoting contextual memory generalization whereas NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons are critical for the strength of associative memory formation.\u003c/p\u003e\u003cp\u003eFinally, we inquired whether two subpopulations of NRe neurons contribute to the contextual memory retrieval. After two weeks of rest, these animals were subjected to CTC again in context A in day 1 and then were administered with CNO 30 min before the memory retrieval in contexts B and A in day 2 and day 3, respectively (\u003cb\u003eFigure S2G-2K\u003c/b\u003e). For animals with NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons labeled, context B evoked lower freezing in hM4Di group than in control group, leading to a significant reduction of generalization index (\u003cb\u003eFigure S2I\u003c/b\u003e). For animals with NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons labeled, contexts A and B both evoked similar freezing in hM4Di group and control group (\u003cb\u003eFigure S2K\u003c/b\u003e). Taken together, these results suggest that NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons regulate the strength of context-dependent associative memory formation, whereas NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons control the contextual memory precision during both memory formation and retrieval.\u003c/p\u003e\n\u003ch3\u003eMolecular, cellular and circuit differences between NRe and NRe neurons\u003c/h3\u003e\n\u003cp\u003eTo compare NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons at molecular, cellular and circuit levels, we employed multiple approaches to compare their gene expression patterns, electrophysiological characteristics, and presynaptic connections. To compare the transcriptomic profiles of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, we performed barcoded scRNA-seq\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e of NRe neurons by injecting two distinct barcoded retrograde AAV-GFP into vCA1d and vCA1v, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig19\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). After 6 weeks, we collected 37894 cells from NRe-containing tissues (N\u0026thinsp;=\u0026thinsp;3 mice) and prepared single-cell transcriptional libraries and barcode expression libraries for scRNA-seq afterwards (see Methods). Then we performed unsupervised clustering analysis and annotated the class of NRe neurons (n\u0026thinsp;=\u0026thinsp;15495 cells) based on 3 markers (\u003cem\u003eCalb2\u003c/em\u003e, \u003cem\u003eStmn1\u003c/em\u003e and \u003cem\u003eGap43\u003c/em\u003e) that we identified from a spatial transcriptome database\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e and a previously reported markers \u003cem\u003eNox4\u003c/em\u003e\u003csup\u003e45\u003c/sup\u003e (\u003cb\u003eFigure S4A-G\u003c/b\u003e). Further clustering analysis revealed 7 clusters of NRe neurons (Fig.\u0026nbsp;\u003cspan refid=\"Fig19\" class=\"InternalRef\"\u003e4\u003c/span\u003eB and \u003cspan refid=\"Fig19\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). We found that vCA1-projecting cells (EGFP positive cells) were significantly enriched in cluster 1 (23.6% vs. 6.5% in other clusters, Fisher test, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, Fig.\u0026nbsp;\u003cspan refid=\"Fig19\" class=\"InternalRef\"\u003e4\u003c/span\u003eD and \u003cspan refid=\"Fig19\" class=\"InternalRef\"\u003e4\u003c/span\u003eE). As the EGFP expression and barcode number were highly correlated (\u003cb\u003eFigure S5A\u003c/b\u003e), we utilized the barcode number in each cell as a proxy for the axon projection strength and classified NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons in the scRNA-seq data based on their barcode preferences (\u003cb\u003eFigure S5B\u003c/b\u003e and S\u003cb\u003e5C\u003c/b\u003e). We found that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons were preferentially enriched in cluster 3 (Fig.\u0026nbsp;\u003cspan refid=\"Fig19\" class=\"InternalRef\"\u003e4\u003c/span\u003eE, lower panel). Analyses of differentially expressed gene (DEG) and gene ontology (GO) revealed distinct scores for many pathways including monoatomic ion transport, fear response and response to stimulus (\u003cb\u003eFig. S5D\u003c/b\u003e and S\u003cb\u003e5E\u003c/b\u003e). Further analysis showed that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons, compared to NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, exhibited lower expressions of Na\u003csup\u003e+\u003c/sup\u003e channel subunits but higher expression levels of K\u003csup\u003e+\u003c/sup\u003e channel subunits (Fig.\u0026nbsp;\u003cspan refid=\"Fig19\" class=\"InternalRef\"\u003e4\u003c/span\u003eF). Thus, NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons may show different depolarization and hyperpolarization components of the spike waveform.\u003c/p\u003e\u003cp\u003eTo test this possibility and further compare electrophysiological characteristics of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, we performed whole-cell recordings from NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons in acute brain slices prepared from animals with red retrobeads locally injected into vCA1d and vCA1v, respectively (Figs.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eA and \u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). We measured firing frequency-versus-current curves as a proxy of neuronal excitability, and found that membrane excitability of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons was lower than that of NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons which may result from lower expressions of Na\u003csup\u003e+\u003c/sup\u003e channel in NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons (Figs.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eC-G). Interestingly, the spike waveform of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons was significantly different from that of NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, as reflected by the half-width ratio of hyperpolarization to depolarization mediated by K\u003csup\u003e+\u003c/sup\u003e and Na\u003csup\u003e+\u003c/sup\u003e channels (Figs.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eH). These results suggest that NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons exhibit distinct intrinsic electrophysiological properties including the membrane excitability and spike waveform. To validate the electrophysiological heterogeneity of NRe neurons \u003cem\u003ein vivo\u003c/em\u003e, we performed tetrode-based 32-channel recordings from NRe neurons in behaving animals (Fig.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eI). Single units (n\u0026thinsp;=\u0026thinsp;92) were sorted based on feature spaces of the spike waveform such as peak-valley amplitudes (Fig.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eJ and \u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eK). We measured 4 features of the spike waveform and classified NRe neurons into two clusters (C1 and C2) using t-SNE dimension reduction and DBSCAN clustering\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e (Figs.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eL and \u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eM). Interestingly, C1 neurons exhibited significantly higher half-width ratio of hyperpolarization to depolarization than C2 neurons (Figs.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eN and \u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eO), consistent with the difference between NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons during slice recording (Fig.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eH). Thus, C1 and C2 neurons were defined as putative NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, respectively. Notably, the C1 (NRe\u003csup\u003evCA1d\u003c/sup\u003e) neurons exhibited higher frequencies of spontaneous firing activity in the range above 4 Hz than C2 (NRe\u003csup\u003evCA1v\u003c/sup\u003e) neurons, whereas they exhibited comparable bursting index and durations (Figs.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eP-R), suggesting that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons are likely to operate in a phasic mode that has been shown to promote memory generalization\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eFinally, we profiled presynaptic inputs of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons using a cell-type specific rabies retrograde tracing strategy. We injected Cre-dependent rabies helper AAV into NRe and AAVretro-hSyn-Cre into vCA1d and vCA1v, respectively, followed by the injection of rabies vectors into NRe 21 days later (Fig.\u0026nbsp;\u003cspan refid=\"Fig21\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). The starter cells for retrograde transsynaptic tracing were identified by co-labeling of rabies-GFP (green) and TVA-mCherry (red) in NRe neurons (Fig.\u0026nbsp;\u003cspan refid=\"Fig21\" class=\"InternalRef\"\u003e6\u003c/span\u003eB). We quantified the rabies-labeled cells in the whole brain after retrograde transsynaptic tracing from NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons (N\u0026thinsp;=\u0026thinsp;4 mice) and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons (N\u0026thinsp;=\u0026thinsp;4 mice). Analysis showed that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons received more inputs from nucleus of the lateral olfactory tract (NLOT2), infralimbic area (ILA), agranular insular area (AI), paraventricular nucleus of the thalamus (PVT), zona incerta (ZI), posterior hypothalamic nucleus (PH), superior colliculus motor related deep gray layer (SCdg), and medial pretectal area (MPT), whereas NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons received more inputs from nucleus accumbens (ACB), caudoputamen (CP), diagonal band nucleus (NDB), bed nuclei of the stria terminalis (BST), substantia innominate (SI), ventrolateral preoptic nucleus (VLPO), periventricular hypothalamic nucleus posterior part (PVp) (Fig.\u0026nbsp;\u003cspan refid=\"Fig21\" class=\"InternalRef\"\u003e6\u003c/span\u003eC and \u003cspan refid=\"Fig21\" class=\"InternalRef\"\u003e6\u003c/span\u003eD). Thus, the distinct presynaptic inputs onto NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons could provide the circuit bases for their distinct memory functions. Taken together, these results have revealed molecular, cellular and circuit bases that may underlie distinct memory functions of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons.\u003c/p\u003e\n\u003ch3\u003eLearning decreased contextual discrimination of putative NRe neurons\u003c/h3\u003e\n\u003cp\u003eThe fact that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons promoted contextual threat generalization after threat learning prompted us to examine the contextual discrimination of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons themselves. To examine the contextual discrimination of putative NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons and their neural plasticity, we recorded context-evoked neuronal activity of NRe neurons from animals subjected to the CTC paradigm in which context A and C exhibited a greater difference than A and B (Fig.\u0026nbsp;\u003cspan refid=\"Fig22\" class=\"InternalRef\"\u003e7\u003c/span\u003eA). After CTC, animals exhibited generalized and lower freezing in neutral context B and C compared to that in conditioned context A (Fig.\u0026nbsp;\u003cspan refid=\"Fig22\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). Using the same unsupervised clustering method as Fig.\u0026nbsp;\u003cspan refid=\"Fig20\" class=\"InternalRef\"\u003e5\u003c/span\u003eN, NRe neurons were grouped into two clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig22\" class=\"InternalRef\"\u003e7\u003c/span\u003eC). The putative NRe\u003csup\u003evCA1d\u003c/sup\u003e (n\u0026thinsp;=\u0026thinsp;33) and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons (n\u0026thinsp;=\u0026thinsp;57) were identified by the differences in the half-width ratio of hyperpolarization to depolarization (Fig.\u0026nbsp;\u003cspan refid=\"Fig22\" class=\"InternalRef\"\u003e7\u003c/span\u003eD). At the single-cell level, we measured the discrimination index (DI) for each cell and found insignificant changes in the discrimination index (Fig.\u0026nbsp;\u003cspan refid=\"Fig22\" class=\"InternalRef\"\u003e7\u003c/span\u003eE). At the population level, CTC significantly reduced the Mahalanobis distance in the coding space between conditioned and neutral contexts in putative NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons, but not in putative NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons (Fig.\u0026nbsp;\u003cspan refid=\"Fig22\" class=\"InternalRef\"\u003e7\u003c/span\u003eG). Thus, NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons may promote threat generalization because of the learning-downregulated contextual discrimination.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eNRe\u003csup\u003evCA1d\u003c/sup\u003e neurons suppressed the contextual discrimination of vCA1d neurons\u003c/h2\u003e\u003cp\u003eHow do NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons promote threat generalization? A plausible explanation lies in their regulation of neural activity in downstream areas, particularly the vCA1d. We thus investigated how NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons regulate the context-evoked activity of vCA1d neurons by combining deep-brain Ca\u003csup\u003e2+\u003c/sup\u003e imaging of vCA1d neurons with chemogenetic inhibition of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons in behaving animals with head-mounted miniscope\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e,\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. We first expressed CaMKII-dependent GCaMP6f\u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e in pyramidal cells in vCA1d and Cre-dependent hM4Di in NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons using the same retrograde strategy as chemogenetic behavioral experiments described above (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eA and \u003cb\u003eB\u003c/b\u003e). Then we implanted a gradient-index (GRIN) lens above vCA1d to monitor the Ca\u003csup\u003e2+\u003c/sup\u003e activity of vCA1d neurons during a 3-day behavioral paradigm for contextual memory generalization. To evaluate the contextual discrimination of vCA1d neurons before and after CTC, animals were exposed to context A/B/C, in which context A was conditioning context and context B and C were neutral contexts with different similarities compared to context A (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eC). Two groups of animals were subjected to the CTC and were administered with saline or CNO 30 min before the CTC in day 2, respectively. Animals in both saline and CNO groups showed comparable freezing to conditioned context A. The saline (control) group exhibited highly generalized freezing in B and some in C, whereas the CNO group showed lower generalized freezing in B and very little in C, consistent with the notion that NRe neurons promote contextual memory generalization (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eD).\u003c/p\u003e\u003cp\u003eWe identified the context-preferring neurons by a bootstrap method comparing their context-evoked Ca\u003csup\u003e2+\u003c/sup\u003e activity between A and B, or between A and C (see Methods). These context discriminative neurons, including non-context preferring neurons, were spatially intermingled (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eE and \u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eF). Overall, the proportions of context discriminative neurons for A vs. B and A vs. C were slightly but significantly decreased after CTC in the saline group (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eG and \u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eH, upper panels). In the CNO group, the proportion of context discriminative neurons for A vs. B was maintained and that for A vs. C was even increased (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eG and \u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eH, upper panels). Consistently, the DI for A vs. B and A vs. C was largely decreased after CTC in saline group, whereas the DI for A vs. B was only slightly decreased and the DI for A vs. C was even increased after CTC in the CNO group (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eG and \u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eH, lower panels). Furthermore, we took a finer approach to analyze the temporal dynamics of context-evoked Ca\u003csup\u003e2+\u003c/sup\u003e activity by correlation analysis\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e and determine the impact of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons on the context discrimination of vCA1d neurons (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eI, low correlation represents good discrimination). We found a robust increase of the correlation in context-evoked Ca\u003csup\u003e2+\u003c/sup\u003e signals after CTC in saline group, but a dramatic decrease of the correlation in CNO group (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eJ). These results demonstrate that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons impose an inhibitory effect on the contextual discrimination of individual vCA1d neurons.\u003c/p\u003e\u003cp\u003eFinally, we performed population analysis to evaluate the context representation by vCA1d neurons in the saline and CNO groups. As exemplified in Fig.\u0026nbsp;\u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eA, the principal component analysis (PCA) showed that neural representations of three contexts in the coding space shrank such that their Mahalanobis distances became smaller after CTC in the saline group (Fig.\u0026nbsp;\u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eB and \u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eC), which correlated with different levels of fear generalization in contexts B and C (Fig.\u0026nbsp;\u003cspan refid=\"Fig23\" class=\"InternalRef\"\u003e8\u003c/span\u003eD). In contrast, as exemplified in Fig.\u0026nbsp;\u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eD, the neural representations of three contexts expanded in the coding space such that their Mahalanobis distances became larger after CTC in the CNO group (Fig.\u0026nbsp;\u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eE and \u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eF). In summary, chemogenetic inhibition of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons in the CNO group resulted in an enlarged distance between conditioned and neutral contexts in the coding space (Fig.\u0026nbsp;\u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eG and \u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eH), providing a neural mechanism that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons promote contextual memory generalization at the population level. This notion was further supported by a decoding analysis using a random forest classifier\u003csup\u003e\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u003c/sup\u003e. The context decoding accuracy was decreased after CTC in the saline group, but was reversed to an increase in the CNO group (Fig.\u0026nbsp;\u003cspan refid=\"Fig24\" class=\"InternalRef\"\u003e9\u003c/span\u003eI). Of note, chemogenetic inhibition of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons during baseline conditions (without learning) did not affect the contextual discrimination of vCA1d neurons in general (\u003cb\u003eFigure S6A-F\u003c/b\u003e), and resulted in a marginal fluctuation of Mahalanobis distance in the coding space of contexts (\u003cb\u003eFigure S6G\u003c/b\u003e). Taken together, these results suggest that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons decrease the memory specificity of context information in vCA1d neurons at both single-cell and population levels, thereby promoting contextual memory generalization.\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we have systematically mapped the single-neuron projectomes of NRe neurons and characterized two distinct NRe subpopulations using multidisciplinary approaches \u0026ndash; NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons preferentially targeted different vCA1 subregions and exhibited distinct ion channel gene expressions, membrane excitabilities, and spike waveforms. Our work demonstrated that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons and downstream vCA1d neurons exhibited down-regulated neural coding of contextual discrimination upon learning, suggesting a specific role of NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons in memory precision control, which may have to do with their phasic mode of firing activity and presynaptic inputs from brain areas promoting memory generalization. These findings reveal circuit-specific functional diversities among NRe neurons and an unprecedent NRe-vCA1d circuit that favors memory generalization over memory specificity.\u003c/p\u003e\u003cp\u003eThe NRe has long been regarded as a key bridge for prefrontal-hippocampal interactions\u003csup\u003e\u003cspan additionalcitationids=\"CR18 CR19\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e,\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e. This thalamic nucleus in the ventral midline is involved in various memory functions including working memory\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, spatial memory\u003csup\u003e\u003cspan additionalcitationids=\"CR24 CR25 CR26 CR27\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e and memory consolidation\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e, generalization\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e and extinction\u003csup\u003e\u003cspan additionalcitationids=\"CR31\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e, as well as feeding\u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e,\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e, circadian regulation\u003csup\u003e\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e,\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e, impulse inhibition\u003csup\u003e\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e\u003c/sup\u003e, stress and anxiety\u003csup\u003e\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e\u003c/sup\u003e. These diverse brain functions may be mediated by specific downstream areas preferentially - though not exclusively - targeted by distinct NRe neuron subpopulations. For instance, while NRe neurons of projectome types 5 and 6 both projected to the mPFC and striatum as previously known\u003csup\u003e\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e, they showed very distinct target area preferences (Fig.\u0026nbsp;\u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eB and \u003cspan refid=\"Fig16\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). This raises a possibility that NRe neurons of projectome types 5 and 6 contribute to different functions linked to the mPFC and striatum\u003csup\u003e\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e,\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e, respectively. Many HIP-projecting NRe neurons send axon collaterals to ENT with similar projection strength. While ENT could be an additional relay from mPFC to HIP\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e, the ENT- and HIP-co-projecting NRe neurons could exert coordinated modulation of direct and indirect information flow from mPFC to HIP, and even interact with other NRe neuron subpopulations such as mPFC-projecting neurons to regulate the prefrontal-hippocampal communication. It remains to be studied in future that how ENT neurons relay the neural coding of context discrimination from NRe to DG or CA1, or both.\u003c/p\u003e\u003cp\u003eOur study has revealed evident differences in axon projections between NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons including distinctions in their transcriptomic profiles, intrinsic electrophysiological properties, spontaneous neural activities, and presynaptic inputomes. Although putative NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons identified \u003cem\u003ein vivo\u003c/em\u003e may contain non-HIP-projecting neurons, the finding that putative NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons exhibited higher spontaneous firing rate than NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons suggests that NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons are more likely to discharge in a phasic mode, which was previously shown to promote contextual memory generalization\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. The gene expression patterns of Na\u003csup\u003e+\u003c/sup\u003e and K\u003csup\u003e+\u003c/sup\u003e channels could account for distinct intrinsic and spontaneous electrophysiological properties of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons and may provide therapeutic targets for treating overgeneralization of threat memory in PTSD patients. Compared to NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons, NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons preferentially receive presynaptic inputs from ZI that has been implicated in the tone threat generalization\u003csup\u003e\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e,\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e\u003c/sup\u003e. In summary, the connectivity-defined NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons may leverage their distinctive ion channel-related gene expressions and intrinsic excitability to promote memory generalization by integrating presynaptic inputs from upstream areas that contribute to memory generalization. In contrast, NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons exhibited higher input-output excitability than NRe\u003csup\u003evCA1d\u003c/sup\u003e neurons, and received synaptic inputs from brain areas such as the striatum that are deeply involved in aversive learning\u003csup\u003e\u003cspan additionalcitationids=\"CR65 CR66\" citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e, thereby playing a permissive role in the formation of contextual threat memory.\u003c/p\u003e\u003cp\u003eThe contextual memory depends on the integration of multimodal information driven by various context elements\u003csup\u003e\u003cspan additionalcitationids=\"CR69\" citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e\u003c/sup\u003e, thus, the overlap between neural representations of the attributes and/or features of contextual memory may lead to the memory generalization\u003csup\u003e\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e,\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e\u003c/sup\u003e. Numerous neural mechanisms for memory specificity have been identified as minimizing the overlap between memory attributes/features during memory formation. For instance, the pattern separation of the neural coding of context elements in the circuit network of dentate gyrus, CA3 and septum is crucial for the memory specificity during contextual memory formation\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan additionalcitationids=\"CR74\" citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e\u003c/sup\u003e. Although neural mechanisms that promote memory generalization were less studied, our work has identified a crucial role of thalamo-hippocampal circuitry, particularly the NRe-vCA1d circuit, in promoting memory generalization. Together with previous findings on the critical role of a Fos-activated DG ensemble\u003csup\u003e\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e\u003c/sup\u003e and amygdala-ACC projection\u003csup\u003e\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e\u003c/sup\u003e in promoting memory generalization, the neural network of NRe, amygdala, hippocampus and ACC are emerging as key regulators of memory precision. In conclusion, our study has revealed the diversity of NRe neurons and identified subpopulations differing at molecular, cellular and circuit levels. The learning-induced plasticity in specific NRe subpopulations, along with plasticity in their specific downstream areas in hippocampus, plays pivotal roles in the control of contextual memory precision.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eAnimals\u003c/h2\u003e\u003cp\u003eAll mice (over 8 weeks) were housed under a 12h light/dark cycle with food and water ad libitum and were socially housed in numbers of two to six littermates in the animal facility of Institute of Neuroscience, Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences. Wild-type C57BL/6J (Slac Laboratory Animal, Shanghai) were used in the study. All animal procedures were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the Institute of Neuroscience.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eViral vectors\u003c/h2\u003e\u003cp\u003eThe following AAV vectors (titers in genome copies/ml) were acquired from Taitool Biological, Shanghai and injected into the brain. AAV2/8-hSyn-Cre-WPRE-pA (1.98\u0026times;10\u003csup\u003e13\u003c/sup\u003e), AAV2/8-CAG-FLEX-EGFP-WPRE-pA (1.71\u0026times;10\u003csup\u003e13\u003c/sup\u003e), scAAV2/retro-hSyn-Cre (1.65\u0026times;10\u003csup\u003e13\u003c/sup\u003e), AAV2/9-hSyn-DIO-hM4D(Gi)-mCherry-WPRE-pA (1.03\u0026times;10\u003csup\u003e13\u003c/sup\u003e), AAV-EF1ɑ-DIO-mCherry (2.42\u0026thinsp;~\u0026thinsp;2.9\u0026times;10\u003csup\u003e13\u003c/sup\u003e), AAV2/9-mCaMKIIα-GCaMP6f-WPRE-pA (1.03\u0026times;10\u003csup\u003e13\u003c/sup\u003e) and AAV2/9-hSyn-hM4D(Gi)-mCherry-WPRE-pA (1.03\u0026times;10\u003csup\u003e13\u003c/sup\u003e). The following AAV vectors were produced at gene editing facility at the Institute of Neuroscience. AAV2/DJ-EF1a-GFP (1.09\u0026times;10\u003csup\u003e13\u003c/sup\u003e), AAV2/9-CMV-FLEX-TVAmCherry-2A-oG (5\u0026times;10\u003csup\u003e13\u003c/sup\u003e), AAVretro-CAG-EGFP-barcode(d) (1.68\u0026times;10\u003csup\u003e13\u003c/sup\u003e, CCTGTATGCGTGGAG), AAVretro-CAG-EGFP-barcode(v) (2.57\u0026times;10\u003csup\u003e13\u003c/sup\u003e, GCGTAAGTCTCCTTG).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eSurgeries\u003c/h2\u003e\u003cp\u003eMice were anesthetized under isoflurane (~\u0026thinsp;2%) and fixed in a stereotactic fame (RWD R510IP, China). The body temperature was maintained at 35\u0026deg;C by a feedback-controlled heating pad (FHC, USA). Eyes were protected from desiccation and surgery light using erythromycin eye ointment (Fuyuan medicine). Glass pipettes filled with solutions (tip diameter 10\u0026ndash;20 \u0026micro;m) were connected to a Picospritzer III (Parker Hannifin Corporation) unless otherwise stated, and stayed for 5 min after injection at following coordinates (posterior to Bregma, AP; lateral to the midline, LAT; below the brain surface, DV; in mm): vCA1d \u003cem\u003estratum pyramidale\u003c/em\u003e, -3.3 AP, -3.2 ML, -1.5 DV; vCA1v \u003cem\u003estratum pyramidale\u003c/em\u003e, AP -3.2, ML -3.5, DV -3.4; vCA1d \u003cem\u003estratum lacunosum-moleculare\u003c/em\u003e, AP -3.35, ML -2.8, DV -1.77; vCA1v \u003cem\u003estratum lacunosum-moleculare\u003c/em\u003e, AP -3.2, ML -3.25, DV -3.4; NRe, AP -0.4, ML -1.35, DV -4.35 with angle 18\u0026deg;. After injection, the skin was sutured by surgical needles and tissue adhesive (Vetbond, 3M) was applied to the wound for protection. Animals remained on a heating pad until full recovery from anesthesia.\u003c/p\u003e\u003cp\u003eFor sparse labeling, the mixed viral solutions (ratio 1:1, AAV2/8-hSyn-Cre-WPRE-pA diluted to 10\u003csup\u003e8\u003c/sup\u003e, AAV2/8-CAG-FLEX-EGFP-WPRE-pA 10\u003csup\u003e12\u003c/sup\u003e) were loaded into a mineral oil-filled glass micropipette attached to an oil hydraulic micromanipulator (MO-10B, Narishige), and were unilaterally injected into the NRe at a speed of 80 nL/min.\u003c/p\u003e\u003cp\u003eFor retrograde dye labeling, 300 nL CTB-647 (#C34778, Invitrogen) and fluorogold (FC10001, Fluorochrome, LLC; diluted at 1:8 in distilled water or saline) were unilaterally injected into vCA1d and vCA1v, in a counter-balanced fashion.\u003c/p\u003e\u003cp\u003eFor chemogenetic manipulation, 200 nL of AAV2/9-hSyn-DIO-hM4D(Gi)-mCherry-WPRE-Pa was unilaterally injected into left NRe. scAAV2/retro-hSyn-Cre was bilaterally injected into SLM layer of vCA1d or vCA1v.\u003c/p\u003e\u003cp\u003eFor barcode-based retrograde tracing, 150nL AAVretro-CAG-EGFP-barcode(d) and AAVretro-CAG-EGFP-barcode(v) were unilaterally injected into vCA1d and vCA1v in the same animal, respectively.\u003c/p\u003e\u003cp\u003eFor Ca\u003csup\u003e2+\u003c/sup\u003e imaging, 250 nL of AAV2/9-mCaMKIIα-GCaMP6f-WPRE-pA was unilaterally injected into pyramidal layer of right vCA1d. After the injection, animals were allowed to recover for 3\u0026ndash;4 weeks before the start of behavioral experiment.\u003c/p\u003e\u003cp\u003eFor slice recording, 275 \u0026micro;l red retrobeads (1:50 dilution) was injected into the \u003cem\u003estratum lacunosum-moleculare\u003c/em\u003e of vCA1d or vCA1v.\u003c/p\u003e\u003cp\u003eFor tetrodes recording, tetrodes were custom made or acquired from Kedou Brain Computer Technology at Suzhou. The skin was cut open and skull was cleaned by 1/10 hydrogen peroxide. The cranial nails were screwed in skull and the ground wire was welded to them. The skull surface and the cross-section of the skin were coated with 3M tissue adhesive for better adhesion of dental cement and protection, respectively. The skull was opened by cranial drill (78001, RWD) and the dura was removed gently by forceps. The electrode was implanted slowly into the brain. For tetrode recording in NRe, electrodes were implanted at coordinates: AP -0.4, ML -1.35, DV -4.15 with angle 18\u0026deg;. The skull window was covered by Vaseline (BODI) for sealing. The ground wire from cranial nails was welded to the electrode and dental cement were applied to the skull, cranial nails and the electrode for fixation. Lincomycin and lidocaine gel (Decheng Pharmaceutical, Jiangxi) was applied to the cross-section of the skin for postoperative local anaesthetization and anti-infection. After a 2-week recovery period, tetrodes were gradually lowered to reach the target before the start of behavioral experiment.\u003c/p\u003e\u003cp\u003eBehavioral paradigm\u003c/p\u003e\u003cp\u003eContextual fear conditioning. Two contexts were used as conditioning context (context A) and neutral context (context B), respectively. Context A was a square chamber (in cm, 30 [L] \u0026times; 30 [W] \u0026times; 35 [H]) with mental grid floor, transparent walls, and cleaned with 75% ethanol. Context B was a triangle chamber with flat blue acrylic floor and patterned walls, and was cleaned with 1% acetic acid. Context A and B shared one wall. The behavioral contexts were placed into a sound-attenuating chamber. Before behavioral training, animals were habituated to the training room for 30 min on three consecutive days. On day 1, animals were habituated in the Context A for 5 min to explore the context freely. On day 2 for threat conditioning, 3 foot-shocks (0.75 mA, 2s, shocker H13-15, Coulbourn Instruments) were delivered after 5 min re-habituation. Animals remained in the context A for 30s after the last foot-shock and then returned to their home cage. On day 3, animals were exposed to context B for 5 min. On day 4, animals were exposed to context A for 5 min. Behavior videos (640 x 480 pixels at 30 Hz) were recorded with Cinelyzer (Plexon Inc). The freezing level was scored based on video-frame pixel changes, as previously described\u003csup\u003e\u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e78\u003c/span\u003e\u003c/sup\u003e. The immobile episodes longer than 1 s were marked as freezing episodes and the freezing was expressed as a percentage of time spent freezing. All parameters were optimized based on the behavioral video and kept the same for the same contexts.\u003c/p\u003e\u003cp\u003eBehavior with chemogenetic manipulation. Before fear conditioning, animals were systemically administered with CNO (5 mg/kg, i.p. injection, Sigma) or saline. CNO was first dissolved in 20 \u0026micro;L dimethyl sulfoxide (DMSO, Sigma), then mixed into 380 \u0026micro;L saline before usage.\u003c/p\u003e\u003cp\u003eBehavior with miniscope recording. Three contexts were used: Context A and B were the same as described above. Context C was another neutral context in a home-cage shape with no common elements of conditioning context at all. Animals were habituated to the head-mounted miniscope for 2 days before the behavioral training. Calcium imaging was captured by UCLA miniscope (V2, LabMaker) and synchronized with the behavior protocol via a TTL signal from Anilab system. On day 1, animals were exposed to context C, B and A sequentially (5 min in each context with 5 min interval). On day 2 (conditioning day), animals went through same fear conditioning in context A as described above. On day 3, animals were exposed to context C, B and A as same as day 1.\u003c/p\u003e\u003cp\u003eBehavior with NRe tetrode recording. For animals that went through threat conditioning paradigm: On day 1, animals were exposed to context B and A sequentially with a 5-min interval. On day 2, animals were intraperitoneally injected with CNO (5 mg/kg). 40 min after injection, animals went through same fear conditioning in context A as described above. On day 3, animals were exposed to context B and A again (same as Day 1). For animals with local chemogenetic inhibition in NRe: On day 1 and 2, animals were exposed to a neutral context for 1 h right after the intraperitoneal injection of CNO and saline, respectively. The injection order of CNO and saline on day 1 and day 2 were counterbalanced among individual animals. On day 3 and day 4, same behavioral procedures were repeated for each animal after tetrodes were moved 40 \u0026micro;m deeper.\u003c/p\u003e\u003cp\u003e\u003cem\u003eGeneralization index\u003c/em\u003e. The contextual generalization index was calculated as the percentage of freezing in the neutral context, divided by the percentage of freezing in the training context.\u003c/p\u003e\u003cp\u003eSingle-neuron projectome reconstruction and analysis\u003c/p\u003e\u003cp\u003e\u003cem\u003efMOST imaging.\u003c/em\u003e Two animals in each batch of injected animals were used to check sparse labeling 2\u0026ndash;3 weeks after the AAV injection\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. Rest animals in the qualified batches were sacrificed 2\u0026ndash;3 months after the AAV injection. Brains were carefully extracted, post-fixed in 4% PFA, rinsed, dehydrated through graded ethanol, and embedded in Lowicryl HM20 resin. Dual-channel high-resolution images were acquired with a voxel resolution of 0.32\u0026times;0.32\u0026times;1 \u0026micro;m\u0026sup3;, utilizing a 20\u0026times; water-immersion objective (NA 1.0)\u003csup\u003e79\u003c/sup\u003e. Propidium iodide staining was applied during imaging to provide cytoarchitectural landmarks\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e, with the GFP channel used for neurite tracing and the PI channel for image registration. Each brain yielded approximately 10,000 16-bit TIFF images at 30,000\u0026times;20,000 pixel resolution. Raw data were segmented into ~\u0026thinsp;1,000,000 data cubes (256\u0026times;256\u0026times;100 voxels) using the \u003cem\u003eslice2cube\u003c/em\u003e command in Fast Neurite Tracer (FNT) and subsequently compressed using High-Efficiency Video Coding (HEVC). For whole-brain overviews, the raw data were down-sampled to a resolution of 256\u0026times;256\u0026times;100 \u0026micro;m.\u003c/p\u003e\u003cp\u003e\u003cem\u003eSingle-cell reconstruction.\u003c/em\u003e The reconstruction process involved the following steps\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e: (1) Samples with tissue damage, overly dense labeling (\u0026gt;\u0026thinsp;200 cells), weak axon terminal signals, or high background fluorescence were excluded. (2) Each neuron was independently reconstructed by two human tracers, and a third tracer randomly reviewed 10\u0026ndash;20% of cells per brain by quantifying false branches and misconnections. Samples scoring below 90/100 were returned for correction. (3) A trained expert conducted lint analysis using FNT software to eliminate overlapping traces between neurons and finalized the dataset for analysis. In total, 8 brain samples were used in the reconstruction.\u003c/p\u003e\u003cp\u003e\u003cem\u003eSingle-cell registration.\u003c/em\u003e Reconstructed neurons were spatially aligned to the Allen Mouse Brain Common Coordinate Framework V3 as previously described\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e,\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. An initial affine transformation including translation, rotation, scaling, and shearing was followed by non-rigid registration to achieve precise alignment\u003csup\u003e\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e80\u003c/span\u003e\u003c/sup\u003e. For each brain, a transformation matrix was computed based on PI-stained cytoarchitectural landmarks and applied to the corresponding neurons. The following brain regions were manually segmented in the images and assigned with uniform artificial grayscale values: HIP, anterior commissure, fourth ventricle, caudate putamen, dentate gyrus, genu of the corpus callosum, fasciculus retroflexus, columns of the fornix, granular layer, medial habenula, mammillothalamic tract, pontine gray, paraventricular hypothalamic nucleus, facial nerve, and lateral ventricle. The final transformation matrix was then applied to all reconstructed neurons within the brain.\u003c/p\u003e\u003cp\u003e\u003cem\u003eAxon digitalization and projectome cell typing\u003c/em\u003e. To effectively classify projectome cell types of NRe neurons, we first classified various types of axon routes, then digitized each axon by a combination of axon routes, finally classified projectome cell types of NRe neurons by unsupervised clustering analysis of digitized axons. The procedures were as follows: (1) The classification of axon route types. Axon trajectories from the soma to each axon terminal were extracted as axon routes. The 3D coordinates of each axon route were compressed to 32-dimensional embeddings by a Deep Neuron Network algorithm\u003csup\u003e\u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e81\u003c/span\u003e\u003c/sup\u003e. We then performed unsupervised hierarchical fuzzy C-means clustering of these embeddings. The number of clusters (axon route types) were finalized when the clustering dendrogram\u0026rsquo;s decisive membership was close to 0.5 and its second-highest was close to 0.1. The classification of axon routes was validated and refined using a Gaussian Mixture Model. (2) Axon digitalization. Each axon was reconstructed by a combination of various axon routes using a support vector machine algorithm. (3) Projectome cell typing. After the determination of NRe targeted brain areas, NRe neurons were grouped by their projection patterns. Projectome cell types were determined by the unsupervised clustering of these projection patterns based on their digitalized axons.\u003c/p\u003e\u003cp\u003eSlice electrophysiology and data analysis\u003c/p\u003e\u003cp\u003eAnimals were anaesthetized by 5% isoflurane and then deeply anaesthetized by high-dose isoflurane (~\u0026thinsp;150 \u0026micro;L in custom nose masks). Animals were transcardially perfused with ice-cold NMDG-based artificial cerebral spinal fluid (ACSF, 93 mM NMDG, 2.5 mM KCl, 1.2 mM NaH\u003csub\u003e2\u003c/sub\u003ePO\u003csub\u003e4\u003c/sub\u003e, 30 mM NaHCO\u003csub\u003e3\u003c/sub\u003e, 20 mM HEPES, 5 mM sodium ascorbate, 2 mM thiourea, 3 mM sodium pyruvate, 25 mM D-glucose,12 mM N-Acetyl-L-cysteine, 10 mM MgCl\u003csub\u003e2\u003c/sub\u003e, 0.5 mM CaCl\u003csub\u003e2\u003c/sub\u003e, oxygenated with 95% O\u003csub\u003e2\u003c/sub\u003e/5% CO\u003csub\u003e2\u003c/sub\u003e). The brain was immediately removed after the perfusion, and transferred to ice-cold NMDG-based ACSF. Coronal brain slices (300 \u0026micro;m thick) containing hippocampus were prepared using a vibratome (VT-1200S, Leica) in an ice-cold NMDG-based ACSF. Slices were maintained for 12 min at 32\u0026deg;C in NMDG-based ACSF and were subsequently transferred into HEPES-based ACSF containing (in mM: 92 NaCl, 2.5 KCl,1.2 NaH\u003csub\u003e2\u003c/sub\u003ePO\u003csub\u003e4\u003c/sub\u003e, 30 NaHCO\u003csub\u003e3\u003c/sub\u003e, 20 HEPES, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 25 D-glucose, 2 MgCl\u003csub\u003e2\u003c/sub\u003e, 2 CaCl\u003csub\u003e2\u003c/sub\u003e, bubbled with 95% O\u003csub\u003e2\u003c/sub\u003e/5% CO\u003csub\u003e2\u003c/sub\u003e) at room temperature and incubated more than 1 h before recording, and then were kept at room temperature (20\u0026ndash;22\u0026deg;C) until start of recordings. Slices were transferred to a recording chamber and infused with ~\u0026thinsp;30\u0026deg;C recording ACSF (in mM: 119 NaCl, 5 KCl, 1.25 NaH\u003csub\u003e2\u003c/sub\u003ePO\u003csub\u003e4\u003c/sub\u003e, 26 NaHCO\u003csub\u003e3\u003c/sub\u003e, 10 D-glucose, 1 MgCl\u003csub\u003e2\u003c/sub\u003e, 2 CaCl\u003csub\u003e2\u003c/sub\u003e, bubbled with 95% O\u003csub\u003e2\u003c/sub\u003e/5% CO\u003csub\u003e2\u003c/sub\u003e). All chemicals were purchased from Sigma-Aldrich. Patch pipettes (4\u0026ndash;7 MΩ) pulled from borosilicate glass (Sutter instrument, BF150-86-10) were filled with a K-gluconate based internal solution (in mM: 126 K-gluconate, 2 KCl, 2 MgCl\u003csub\u003e2\u003c/sub\u003e,10 HEPES, 0.2 EGTA, 4 MgATP\u003csub\u003e2\u003c/sub\u003e, 0.4 Na\u003csub\u003e3\u003c/sub\u003eGTP, 10 Na-phosphate creatine, 290 mOsm, adjusted to pH 7.2\u0026thinsp;~\u0026thinsp;7.3 with KOH). Whole-cell recording was performed with a Multiclamp 700B amplifier and a Digidata 1440A (Molecular Device).\u003c/p\u003e\u003cp\u003eTo measure I-V curves and assess the intrinsic excitability, NRe neurons were held at -60 mV under current clamp mode, and injected with 1-s current steps starting from 0 pA (10 pA increments, 10 s intervals). The Ra and Rm were assessed by a -10-mV step (repeated 5 times) with the membrane potential held at -70 mV under voltage clamp mode. To measure the parameters of spike waveforms, the first current-evoked spike in each cell was analyzed by Clampfit and MATLAB scripts. The baseline for each spike was determined by the averaged membrane potential during the 5 ms period before the spike threshold (20 mV/ms).\u003c/p\u003e\u003cp\u003eBarcode- and scRNA-seq data acquisition and analysis\u003c/p\u003e\u003cp\u003eThe NRe region was micro-dissected from acute brain slices that were prepared using the same protocol as described above. The isolated NRe-containing tissues were pooled into a single tube and subjected to enzymatic dissociation in a choline chloride-based solution containing 20 U/mL papain and 100 U/mL DNase I (Cat# LK003176, Worthington) at 37\u0026deg;C for 23 minutes. The choline chloride-based dissociation buffer contained (in mM): 92 choline chloride, 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 glucose, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 10 MgSO₄\u0026middot;7H₂O, 0.5 CaCl₂\u0026middot;2H₂O, and 12 N-acetyl-L-cysteine.\u003c/p\u003e\u003cp\u003eA second digestion step was performed using 1 mg/mL protease (Cat# P5147, Sigma) and 1 mg/mL dispase (Cat# LS02104, Worthington) at 25\u0026deg;C for 18 minutes. Following enzymatic digestion, the tissues were gently triturated with fire-polished glass Pasteur pipettes to obtain single-cell suspensions. The suspensions were filtered through a 30 \u0026micro;m cell strainer (Cat# 130-098-458, Miltenyi) to remove aggregates and cleaned using a debris removal solution (Cat# 130-109-398, Miltenyi). All steps were completed within 2.5 hours post euthanasia. Finally, approximately 15,000 to 20,000 cells were loaded into the well of a 10\u0026times; Chromium chip for single-cell RNA sequencing. Single cells were captured using the Chromium Single Cell 4\u0026rsquo; Reagent Kits and single-cell transcriptional libraries were obtained by the 10\u0026times; Genomics protocol.\u003c/p\u003e\u003cp\u003eBarcode expression libraries were obtained by PCR using 100 ng of cDNA fragments derived from post-cDNA-amplification cleanup material (containing barcodes and unique molecular identifier), NEBNext Ultra II Q5 Master Mix (NEB, M0544S) and following primers (200 nM): P5-Read1 (TGATACGGCGACCACCGAGATCTACACTCTTTCCCT ACACGACGCTC) and P7-index-Read2-EGFP (CAAGCAGAAGACGGCATACGAGA TAGGATTCGGTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGgCATGGACGAGCTGTACAAG). The PCR protocol was following: initial denaturation, 98\u0026deg;C for 60 s; denaturation, 20\u0026ndash;24 cycles of 98\u0026deg;C for 10 s; annealing/extension 65\u0026deg;C for 75 s; final hold, 75\u0026deg;C for 5 min. Finally, the barcodes libraries were harvested from 482\u0026ndash;488 bp (including 120 bp adapters) of pooled PCR reactions using a 0.8 \u0026times; SPRI selection cleanup (Beckman, B23318), and sequenced on a Novaseq Xplus-PE150 Xplus platform.\u003c/p\u003e\u003cp\u003e\u003cem\u003eTranscriptome analysis.\u003c/em\u003e A custom mouse reference genome was built using Cell Ranger (v8.0.1) based on the GRCm39 genome assembly. Raw sequencing data were aligned and quantified using Cell Ranger with default parameters\u003csup\u003e\u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e82\u003c/span\u003e\u003c/sup\u003e, and were utilized to generate digital gene expression matrices for further analyses in the Seurat R package (v5.3.0)\u003csup\u003e83,84\u003c/sup\u003e. Cells with abnormally high or low gene counts (representing doublets or background noise) were filtered out if the number of genes (nFeatureRNA) was less than 1000, or over 5% of their UMI counts were derived from mitochondrial genes. After sample integration (\u0026ldquo;SelectIntegrationFeatures\u0026rdquo; for features integration, \u0026ldquo;FindIntegrationAnchors\u0026rdquo; for anchors, \"SCT\" for normalization method), dimensionality reduction and clustering were performed through principal component analysis (PCA) using \u0026ldquo;RunPCA,\u0026rdquo; followed by clustering via the \u0026ldquo;FindClusters\u0026rdquo; function.\u003c/p\u003e\u003cp\u003eCluster-specific marker genes among all clusters, and differentially expressed genes (DEGs) between specific cell subtypes were identified with \u0026ldquo;FindAllMarkers\u0026rdquo; and \u0026ldquo;FindMarkers\u0026rdquo;, respectively, using thresholds of p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and |log₂ fold change| \u0026gt;1.\u003c/p\u003e\u003cp\u003eTo identify barcoded NRe neurons that project to vCA1, we utilized the following criteria. (1) The EGFP mRNA was detectable (nUMI\u0026thinsp;\u0026gt;\u0026thinsp;0). (2) Barcode count in each neuron was higher than the 99th percentile of barcode counts in non-neuronal cells. (3) The top 1% neurons with highest barcode counts were considered as potential doublets or PCR amplification artifacts and exclude from further analysis.\u003c/p\u003e\u003cp\u003eTo identify NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons in the scRNA-seq dataset, we first quantify the projection preference of each neuron \u0026ndash; the selectivity index (SI) \u0026ndash; based on the count of two barcode types (d and v) using the equation: SI = (d \u0026ndash;v) / (d\u0026thinsp;+\u0026thinsp;v). Then the NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons were identified based on the bimodal distribution of SI. Neurons with SI close to 0 (10% of the bimodal distribution, corresponding to absolute SI\u0026thinsp;\u0026lt;\u0026thinsp;0.2) were considered as co-projecting neurons.\u003c/p\u003e\u003cp\u003e\u003cem\u003eExpression profiles of ion channels and transmitter receptors.\u003c/em\u003e SCN family: \u003cem\u003eScn1a, Scn2a, Scn3a, Scn4a, Scn5a, Scn7a, Scn8a, Scn9a, Scn10a, Scn11a, Scn1b, Scn2b, Scn3b, Scn4b.\u003c/em\u003e KCN family: \u003cem\u003eKcnc1, Kcnc3, Kcnn1, Kcnn2, Kcnn3, Kcnq2, Kcnq3, Kcnq5, Kcnma1\u003c/em\u003e. CACNA family: \u003cem\u003eCacna1c, Cacna1d, Cacna1a, Cacna1b, Cacna1e, Cacna1g, Cacna1h, Cacna1i\u003c/em\u003e. HCN family: \u003cem\u003eHcn1, Hcn2, Hcn3, Hcn4.\u003c/em\u003e ACH family: \u003cem\u003eChrm1, Chrm2, Chrm3, Chrna1, Chrna2, Chrna3, Chrna4, Chrna5, Chrna6, Chrna7, Chrna9, Chrnb1, Chrnb2, Chrnb3, Chrnb4\u003c/em\u003e. GABA receptor family: \u003cem\u003eGabra1, Gabra2, Gabra3, Gabra4, Gabra5, Gabrb1, Gabrb2, Gabrb3, Gabrg1, Gabrg2, Gabrg3, Gabrd, Gabre, Gabrp, Gabrq, Gabrr1, Gabrr2, Gabrr3, Gabbr1, Gabbr2\u003c/em\u003e. Glutamate receptor family: \u003cem\u003eGrik1, Grik2, Grik3, Grik4, Grik5, Gria1, Gria2, Gria3, Gria4, Grin1, Grin2a, Grin2b, Grin2c, Grin2d, Grin3a, Grin3b\u003c/em\u003e. Neurexin family: \u003cem\u003eNrxn1, Nrxn2, Nrxn3.\u003c/em\u003e Neuroligin family: \u003cem\u003eLrrtm1, Lrrtm2, Lrrtm3, Lrrtm4\u003c/em\u003e. Transporter family: \u003cem\u003eSlc6a1, Slc6a13, Slc6a11, Slc6a12, Slc32a1, Slc1a3, Slc1a2, Slc1a1, Slc1a6, Slc17a7, Slc17a6, Slc17a8\u003c/em\u003e.\u003c/p\u003e\u003cp\u003eSpatial transcriptome analysis\u003c/p\u003e\u003cp\u003eThe spatial gene expression matrix for the thalamus and hypothalamus (TH/HY) was obtained from a published dataset\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e and binned by aggregating transcripts of the same gene within each bin (50 \u0026times; 50 DNA nanoball grid, 25 \u0026micro;m in diameter per bin). Data processing and visualization were conducted using the Seurat package (v4.0.2) and the Tidyverse package (v1.3.1) in R (v4.1.2). The number of detected genes and mitochondrial gene were used for quality control. Bins were excluded from analysis if nFeature\u0026thinsp;\u0026lt;\u0026thinsp;200 or percent.mt\u0026thinsp;\u0026gt;\u0026thinsp;5%. The dataset was log-normalized and variance-stabilized using \u0026ldquo;SCTransform\u0026rdquo; with default parameters. Batch effects among brain slices were corrected using the Harmony package (v0.1.0).\u003c/p\u003e\u003cp\u003eClusters and subclusters were identified using \u0026ldquo;FindClusters\u0026rdquo; in Seurat, applying shared nearest neighbor modularity optimization with a clustering resolution of 0.1. The neighborhood graph was projected into two dimensions using Uniform Manifold Approximation and Projection (UMAP) for visualization. The marker genes in each cluster/subclusters were identified by differential expression analysis using \u0026ldquo;FindAllMarkers\u0026rdquo; in Seurat.\u003c/p\u003e\u003cp\u003eSingle-unit electrophysiology and analysis\u003c/p\u003e\u003cp\u003eTetrodes implanted in NRe was connected with a digital headstage, which was connected to CerePlex Direct (Blackrock Neurotech) via a lightweight data cable. Neuronal activity was digitized at 30 kHz and low-cut Bessel filtered at 300 Hz. The tetrodes were moved in small daily increments (40 \u0026micro;m) to record different single units. At the end of the experiment, recording sites were marked with electrolytic lesions (7 \u0026micro;A, 5 s) and reconstructed with standard histological procedures. Single-unit spike sorting was conducted using Off-Line Spike Sorter (OFSS, Plexon). Raw signals were low-cut filtered at 300Hz by a 4-pole Bessel filtering. After thresholds adjustment and waveform detection, single units were manually defined based on 3D peak-valley features and principal component features. Spikes within the refractory period (1.5 ms) were removed. If two single units share 50% spikes at 1 ms temporal resolution, the unit with lower firing rates was defined as duplicated spikes and removed. The high quality of spike sorting was reflected by a high J3 value (ratio of inter-cluster distances to intra-cluster distances), a low DB index (ratio of intra-cluster distances to inter-cluster distances) and significant multivariate ANOVA (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e\u003cp\u003e\u003cem\u003eBurst spikes\u003c/em\u003e were defined as a cluster of densely fired spikes occurring within a specified time window and satisfying criteria including a minimum duration of 10 ms, a minimum spike count of 3, and a maximum inter-spike interval of 300 ms. The burst index was calculated as the ratio of the total number of spikes within bursts to the total spike count, serving as a quantitative measure of the proportion of burst-related activity. The burst duration was determined as the difference between the maximum and minimum timestamps of spikes within each burst, providing the temporal extent of the burst.\u003c/p\u003e\u003cp\u003e\u003cem\u003eContext discriminative cells\u003c/em\u003e were identified by a bootstrap method\u003csup\u003e\u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e85\u003c/span\u003e\u003c/sup\u003e. For each cell, all the spikes during context exposures were used to create a surrogate distribution of expected spikes for each context by shuffling the inter-spike intervals from the original spike timestamps (1000 iterations). Cells were considered to be context discriminative if their average firing rate in the context fell outside of the surrogate distribution (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e\u003cp\u003e\u003cem\u003eContextual discrimination index\u003c/em\u003e (DI) for a single unit was computed as the absolute difference between averaged firing rates in two contexts divided by the sum of averaged firing rates in two contexts. The higher DI, the better contextual discrimination.\u003c/p\u003e\u003cp\u003e\u003cem\u003eMahalanobis distance\u003c/em\u003e in the populational coding space was determined by the following steps: (1) Raw data of each neuron was z-scored within each context. (2) For two groups of single units with different numbers, only a random subset of neurons (same number as the smaller group) in the larger group were used for the calculation of Mahalanobis distance\u003csup\u003e\u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e85\u003c/span\u003e\u003c/sup\u003e. (3) The previous step was repeated for 50 times to determine the significance of Mahalanobis distance (Wilcoxon rank-sum test).\u003c/p\u003e\u003cp\u003eMiniscope Ca\u003csup\u003e2+\u003c/sup\u003e imaging and data analysis\u003c/p\u003e\u003cp\u003eThree weeks following AAV injection, a GRIN lens (1.8 mm Φ, 0.25 pitch, 0.55 NA, Edmunt 64519) was implanted above vCA1d. Briefly, a cranial window was made above the vCA1, and the brain tissue above the target was aspirated with syringe blunt needle attached to a Peristaltic Pump (BT100-1F, LongerPump). Saline was repeatedly applied to the exposed tissue to prevent desiccation and clean the blood. Aspiration was stopped once a thin layer of corpus callosum was seen. After the clear of blood, the GRIN lens was slowly lowered above the vCA1d with a custom holder. The lens was fixed to the skull using ultraviolet-light curing dental resin (CharmFil Flow, DentKist Inc). Finally, acrylic dental cement (Super Bond C\u0026amp;B, Jakarta) was used to seal the skull. One week after lens implantation, GCaMP6f fluorescence was checked regularly using a miniature microscope (UCLA V2, LabMaker). Once a clear field of view with neurons was observed, a baseplate attached with a miniscope was fixed to the skull using acrylic dental cement. The miniscope was detached after the fixation. The baseplate was protected by a plastic cover. The animal was returned to the home cage and remained undisturbed for at least one week.\u003c/p\u003e\u003cp\u003eDuring the habituation of head-mounted miniscope, the acquisition settings including exposure time, gain and LED power were set properly to get a clear vision for single neurons in the FOV. The same miniscope was used across all behavioral sessions in the same animal with same acquisition settings to perform repetitive imaging. Before each imaging session, the miniscope was attached to the animal head and habituated for 5 min in home cage. A snapshot of imaging FOV was used to confirm the consistency across multiple sessions. The behavioral video recording (Cinelyzer, Plexon Inc.) and the miniscope imaging were synchronized by TTL signals from a Anilab system (Anilab, China). The miniscope imaging data was recorded to uncompressed AVI files at 30 Hz by MiniScopeControl software (UCLA V2).\u003c/p\u003e\u003cp\u003eThe imaging data across all sessions were concatenated into a single video. The field of view was cropped and then 2\u0026times; spatially and 3\u0026times; temporally down-sampled using \u003cem\u003emoviepy\u003c/em\u003e python package to reduce the computation load afterwards. The preprocessed data was processed by CaImAn (1.8.5) toolbox using constrained nonnegative matrix factorization extension (CNMFe) method\u003csup\u003e\u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e86\u003c/span\u003e\u003c/sup\u003e. The motion correction and source extraction were all done with default parameter setting except the following: down-sampling factor in time for initialization (tsub) 4; neuron diameter (gSiz) 13 pixels; 2D Gaussian kernel smoothing (gSig) 3 pixels; minimum peak to noise ratio (min_pnr) 10; spatial consistency (rval_thr) 0.85. The denoised temporal activity trace (estimates.C) of each neuron extracted by CNMF-E for further analysis with MATLAB scripts (MathWorks). Context discriminative cells and discriminative index (DI) were determined in the same way as single-unit analysis.\u003c/p\u003e\u003cp\u003e\u003cem\u003eMahalanobis distance.\u003c/em\u003e For each neuron, the calcium activity was z-scored and temporally binned with a 5-s time window. Neurons recorded in all mice were pooled for population analysis. We randomly subsampled 50 neurons to calculate the Mahalanobis distance with 5000 repetitions.\u003c/p\u003e\u003cp\u003e\u003cem\u003eContext decoding.\u003c/em\u003e A classifier was trained by the neural activity evoked by context A and B, using a Treebagger algorithm in MATLAB. A total of 100 classification trees were used for each classifier. The 5 min of memory retrieval was split into 60 bins (5 s/bin). The classifier was trained by a 15 s of neural activity and tested by the rest of neural activity. The decoding accuracy was determined from ~\u0026thinsp;30 repetitive decoding with a sliding window of 10 s.\u003c/p\u003e\u003cp\u003eHistology and co-localization analysis\u003c/p\u003e\u003cp\u003eMice were administered a lethal dose of isoflurane and intraperitoneal perfused with phosphate buffered saline (PBS) followed by 4% w/v paraformaldehyde (PFA). Following dissection, brains were post-fixed in 4% PFA overnight at 4\u0026deg;C and coronal sections were cut at 80 \u0026micro;m thickness using a vibratome (Leica, Germany, VT1000S). Slices were mounted on microscope slides, dried, and imaged with a FV3000 confocal microscope (Olympus). The fluorescent neurons were labeled by CTB647 and Fluorogold, and were quantified by the CellCounter plugin in Fiji.\u003c/p\u003e\u003cp\u003eQuantification and statistical analysis\u003c/p\u003e\u003cp\u003eData are summarized as mean values with standard error of the mean (SEM). The cell number is indicated with n and the animal number is indicated with N. No statistical methods were used to predetermine sample sizes. The sample sizes were chosen based on published studies and current standards in the field. Statistical analysis was performed in Prism 8 (GraphPad) and MATLAB (MathWorks). For all datasets normality was tested using the Kolmogorov\u0026ndash;Smirnov test (a\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and homogeneity of variance with Levene\u0026rsquo;s test (a\u0026thinsp;\u0026lt;\u0026thinsp;0.05) to determine whether parametric or non-parametric analyses were required. Parametric analyses included unpaired \u003cem\u003et\u003c/em\u003e-tests. If either homogeneity of variance or normality assumptions were not met, non-parametric analyses such as non-parametric Wilcoxon rank sum, or Wilcoxon signed rank were used. For data analyzed in the bootstrap resampling procedure to identify context cell and non-context cell, 95% confident interval were calculated by approximating to a normal distribution, owing to the large sample size. All tests were two-tailed. Statistical parameters and significance are reported in the text or in the figure legends. The significance threshold was placed at \u0026#119886; = 0.05 (n.s., P\u0026thinsp;\u0026gt;\u0026thinsp;0.05; *, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05; **, P\u0026thinsp;\u0026lt;\u0026thinsp;0.01; ***, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001; ****, P\u0026thinsp;\u0026lt;\u0026thinsp;0.0001).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eLead Contact\u003c/h2\u003e\u003cp\u003eFurther information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Chun XU ([email protected]).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eSource data have been deposited at Brain Data Center, Chinese Academy of Sciences and are publicly available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.12412/BSDC.1757059765.20001\u003c/span\u003e\u003cspan address=\"10.12412/BSDC.1757059765.20001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. The scRNA-seq data would be freely available from the National Genomics Data Center (NGDC) and associated repositories Genome Sequence Archive (GSA), upon publication (accession number: CRA030346; URL: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://ngdc.cncb.ac.cn/gsa/s/x5u9X6JU\u003c/span\u003e\u003cspan address=\"https://ngdc.cncb.ac.cn/gsa/s/x5u9X6JU\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003eCode Availability\u003c/h2\u003e\u003cp\u003eThis paper does not report original code.\u003c/p\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgments\u003c/p\u003e\n\u003cp\u003eWe thank all members of the Xu lab for helpful discussions and comments. We thank Enpeng Liang, the animal facility, gene editing facility and the imaging facility for technical support. We thank the UCLA\u0026nbsp;miniscope team for sharing the design of the miniscope system. This study was supported by National Science and Technology Innovation 2030 Major Program (2022ZD0205000, 2021ZD0201001), CAS Project for Young Scientists in Basic Research (YSBR-116), National Key R\u0026amp;D Program of China (2020YFE0205900), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB32010105), and Lingang Lab (Grant LG202104-01-08).\u003c/p\u003e\n\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eConceptualization: C.X., Q.W., J.L., Y.S., H.H, Y.C, H.G., X.L.\u003c/p\u003e\n\u003cp\u003eInvestigation (behaviors and neural recordings): Q.W., Y.L., Z.S., G.W., Y.Z.\u003c/p\u003e\n\u003cp\u003eInvestigation (single-neuron projectome): Q.W., X.L., H.G.\u003c/p\u003e\n\u003cp\u003eInvestigation (transcriptome data acquisition): Q.W., R.Z., E.L., Y.C.\u003c/p\u003e\n\u003cp\u003eFormal analysis (behaviors and neural recordings): Q.W., Y.L., G.W., Z.S., Y.G.\u003c/p\u003e\n\u003cp\u003eFormal analysis (transcriptome data analysis): Q.W., M.Y. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFormal analysis (single-neuron projectome): S.C., B.Z. S.C.\u003c/p\u003e\n\u003cp\u003eFunding acquisition and resources: C.X., H.H., Y.G., H.G.\u003c/p\u003e\n\u003cp\u003eSupervision: C.X.\u003c/p\u003e\n\u003cp\u003eWriting (original draft): C.X., Q.W., S.C., Y.L., Y.G.\u003c/p\u003e\n\u003cp\u003eWriting (review \u0026amp; editing): C.X., Q.W.\u003c/p\u003e\n\u003cp\u003eDeclaration of interests\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eShmuel Lissek, S. R., Randi E. Heller, D. L., Marilla Geraci, \u0026amp; Daniel S. Pine, C. G. Overgeneralization of Conditioned Fear as a Pathogenic Marker of Panic Disorder. \u003cem\u003eAm J Psychiatry\u003c/em\u003e, doi:10.1176/appi.ajp.2009.0903041 (2010).\u003c/li\u003e\n \u003cli\u003eDunsmoor, J. E. \u0026amp; Paz, R. Fear Generalization and Anxiety: Behavioral and Neural Mechanisms. \u003cem\u003eBiol Psychiatry\u003c/em\u003e \u003cstrong\u003e78\u003c/strong\u003e, 336-343, doi:10.1016/j.biopsych.2015.04.010 (2015).\u003c/li\u003e\n \u003cli\u003eElzinga, B. M. \u0026amp; Bremner, J. D. Are the neural substrates of memory the final common pathway in posttraumatic stress disorder (PTSD)? \u003cem\u003eJournal of Affective Disorders\u003c/em\u003e, doi:10.1016/s0165-0327(01)00351-2 (2002).\u003c/li\u003e\n \u003cli\u003eKaczkurkin, A. N.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Neural Substrates of Overgeneralized Conditioned Fear in PTSD. \u003cem\u003eAm J Psychiatry\u003c/em\u003e \u003cstrong\u003e174\u003c/strong\u003e, 125-134, doi:10.1176/appi.ajp.2016.15121549 (2017).\u003c/li\u003e\n \u003cli\u003eLissek, S.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Generalized anxiety disorder is associated with overgeneralization of classically conditioned fear. \u003cem\u003eBiol Psychiatry\u003c/em\u003e \u003cstrong\u003e75\u003c/strong\u003e, 909-915, doi:10.1016/j.biopsych.2013.07.025 (2014).\u003c/li\u003e\n \u003cli\u003eLissek, S.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Generalization of conditioned fear-potentiated startle in humans: experimental validation and clinical relevance. \u003cem\u003eBehav Res Ther\u003c/em\u003e \u003cstrong\u003e46\u003c/strong\u003e, 678-687, doi:10.1016/j.brat.2008.02.005 (2008).\u003c/li\u003e\n \u003cli\u003eTovote, P., Fadok, J. P. \u0026amp; Luthi, A. Neuronal circuits for fear and anxiety. \u003cem\u003eNat Rev Neurosci\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 317-331, doi:10.1038/nrn3945 (2015).\u003c/li\u003e\n \u003cli\u003eBotta, P.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Regulating anxiety with extrasynaptic inhibition. \u003cem\u003eNat Neurosci\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 1493-1500, doi:10.1038/nn.4102 (2015).\u003c/li\u003e\n \u003cli\u003eXu, W. \u0026amp; Sudhof, T. C. A neural circuit for memory specificity and generalization. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e339\u003c/strong\u003e, 1290-1295, doi:10.1126/science.1229534 (2013).\u003c/li\u003e\n \u003cli\u003eMcHugh, T. J.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Dentate gyrus NMDA receptors mediate rapid pattern separation in the hippocampal network. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e317\u003c/strong\u003e, 94-99, doi:10.1126/science.1140263 (2007).\u003c/li\u003e\n \u003cli\u003eXu, W. \u0026amp; S\u0026uuml;dhof, T. C. A Neural Circuit for Memory Specificity and Generalization. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e339\u003c/strong\u003e, 1290-1295, doi:10.1126/science.1229534 (2013).\u003c/li\u003e\n \u003cli\u003eCullen, P. K., Gilman, T. L., Winiecki, P., Riccio, D. C. \u0026amp; Jasnow, A. M. Activity of the anterior cingulate cortex and ventral hippocampus underlie increases in contextual fear generalization. \u003cem\u003eNeurobiol Learn Mem\u003c/em\u003e \u003cstrong\u003e124\u003c/strong\u003e, 19-27, doi:10.1016/j.nlm.2015.07.001 (2015).\u003c/li\u003e\n \u003cli\u003eXu, W.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Distinct neuronal coding schemes in memory revealed by selective erasure of fast synchronous synaptic transmission. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e73\u003c/strong\u003e, 990-1001, doi:10.1016/j.neuron.2011.12.036 (2012).\u003c/li\u003e\n \u003cli\u003eBesnard, A., Miller, S. M. \u0026amp; Sahay, A. Distinct Dorsal and Ventral Hippocampal CA3 Outputs Govern Contextual Fear Discrimination. \u003cem\u003eCell Rep\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 2360-2373 e2365, doi:10.1016/j.celrep.2020.01.055 (2020).\u003c/li\u003e\n \u003cli\u003eHyman, J. M., Ma, L., Balaguer-Ballester, E., Durstewitz, D. \u0026amp; Seamans, J. K. Contextual encoding by ensembles of medial prefrontal cortex neurons. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e \u003cstrong\u003e109\u003c/strong\u003e, 5086-5091, doi:10.1073/pnas.1114415109 (2012).\u003c/li\u003e\n \u003cli\u003eFrankland, P. W., Cestari, V., Filipkowski, R. K. \u0026amp; McDonald, R. J. The Dorsal Hippocampus Is Essential for Context Discrimination but Not for Contextual Conditioning. \u003cem\u003eBehavioral Neuroscience\u003c/em\u003e (1998).\u003c/li\u003e\n \u003cli\u003eVertes, R. P., Hoover, W. B., Do Valle, A. C., Sherman, A. \u0026amp; Rodriguez, J. J. Efferent projections of reuniens and rhomboid nuclei of the thalamus in the rat. \u003cem\u003eJ Comp Neurol\u003c/em\u003e \u003cstrong\u003e499\u003c/strong\u003e, 768-796, doi:10.1002/cne.21135 (2006).\u003c/li\u003e\n \u003cli\u003eWeel, M. J. D.-v. d., Griffin, A. L., Ito, H. T. \u0026amp; Shapiro, M. L. The nucleus reuniens of the thalamus sits at the nexus of a hippocampus and medial prefrontal cortex circuit enabling memory and behavior. \u003cem\u003eLearning \u0026amp; Memory\u003c/em\u003e, doi:10.1101/lm.048389.118 (2020).\u003c/li\u003e\n \u003cli\u003eCassel, J. C.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The reuniens and rhomboid nuclei: neuroanatomy, electrophysiological characteristics and behavioral implications. \u003cem\u003eProg Neurobiol\u003c/em\u003e \u003cstrong\u003e111\u003c/strong\u003e, 34-52, doi:10.1016/j.pneurobio.2013.08.006 (2013).\u003c/li\u003e\n \u003cli\u003eVarela, C., Kumar, S., Yang, J. Y. \u0026amp; Wilson, M. A. Anatomical substrates for direct interactions between hippocampus, medial prefrontal cortex, and the thalamic nucleus reuniens. \u003cem\u003eBrain Struct Funct\u003c/em\u003e \u003cstrong\u003e219\u003c/strong\u003e, 911-929, doi:10.1007/s00429-013-0543-5 (2014).\u003c/li\u003e\n \u003cli\u003eRatigan, H. C., Krishnan, S., Smith, S. \u0026amp; Sheffield, M. E. J. A thalamic-hippocampal CA1 signal for contextual fear memory suppression, extinction, and discrimination. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 6758, doi:10.1038/s41467-023-42429-6 (2023).\u003c/li\u003e\n \u003cli\u003eHallock, H. L., Wang, A., Shaw, C. L. \u0026amp; Griffin, A. L. Transient Inactivation of the Thalamic Nucleus Reuniens and Rhomboid Nucleus Produces Deficits of a Working-Memory Dependent Tactile-Visual Conditional Discrimination Task. doi:10.l037/a0034653 (2013).\u003c/li\u003e\n \u003cli\u003eHallock, H. L., Wang, A. \u0026amp; Griffin, A. L. Ventral Midline Thalamus Is Critical for Hippocampal-Prefrontal Synchrony and Spatial Working Memory. \u003cem\u003eJ Neurosci\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 8372-8389, doi:10.1523/JNEUROSCI.0991-16.2016 (2016).\u003c/li\u003e\n \u003cli\u003eHuang, X.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A Visual Circuit Related to the Nucleus Reuniens for the Spatial-Memory-Promoting Effects of Light Treatment. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e109\u003c/strong\u003e, 347-362 e347, doi:10.1016/j.neuron.2020.10.023 (2021).\u003c/li\u003e\n \u003cli\u003eLoureiro, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The ventral midline thalamus (reuniens and rhomboid nuclei) contributes to the persistence of spatial memory in rats. \u003cem\u003eJ Neurosci\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 9947-9959, doi:10.1523/JNEUROSCI.0410-12.2012 (2012).\u003c/li\u003e\n \u003cli\u003eJones, M. W. \u0026amp; Wilson, M. A. Theta rhythms coordinate hippocampal-prefrontal interactions in a spatial memory task. \u003cem\u003ePLoS Biol\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, e402, doi:10.1371/journal.pbio.0030402 (2005).\u003c/li\u003e\n \u003cli\u003eIto, H. T., Zhang, S. J., Witter, M. P., Moser, E. I. \u0026amp; Moser, M. B. A prefrontal-thalamo-hippocampal circuit for goal-directed spatial navigation. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e522\u003c/strong\u003e, 50-55, doi:10.1038/nature14396 (2015).\u003c/li\u003e\n \u003cli\u003eJankowski, M. M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Nucleus reuniens of the thalamus contains head direction cells. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, doi:10.7554/eLife.03075 (2014).\u003c/li\u003e\n \u003cli\u003eFerraris, M., Cassel, J. C., Pereira de Vasconcelos, A., Stephan, A. \u0026amp; Quilichini, P. P. The nucleus reuniens, a thalamic relay for cortico-hippocampal interaction in recent and remote memory consolidation. \u003cem\u003eNeurosci Biobehav Rev\u003c/em\u003e \u003cstrong\u003e125\u003c/strong\u003e, 339-354, doi:10.1016/j.neubiorev.2021.02.025 (2021).\u003c/li\u003e\n \u003cli\u003eRamanathan, K. R., Jin, J., Giustino, T. F., Payne, M. R. \u0026amp; Maren, S. Prefrontal projections to the thalamic nucleus reuniens mediate fear extinction. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 4527, doi:10.1038/s41467-018-06970-z (2018).\u003c/li\u003e\n \u003cli\u003eSilva, B. A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A thalamo-amygdalar circuit underlying the extinction of remote fear memories. \u003cem\u003eNat Neurosci\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 964-974, doi:10.1038/s41593-021-00856-y (2021).\u003c/li\u003e\n \u003cli\u003eTomaszewski, K. F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e, doi:10.1101/2024.05.20.594930 (2024).\u003c/li\u003e\n \u003cli\u003eBokor, H., Csaki, A., Kocsis, K. \u0026amp; Kiss, J. Cellular architecture of the nucleus reuniens thalami and its putative aspartatergic/glutamatergic projection to the hippocampus and medial septum in the rat. \u003cem\u003eEur J Neurosci\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 1227-1239, doi:10.1046/j.1460-9568.2002.02189.x (2002).\u003c/li\u003e\n \u003cli\u003eWalsh, D. A., Brown, J. T. \u0026amp; Randall, A. D. In vitro characterization of cell-level neurophysiological diversity in the rostral nucleus reuniens of adult mice. \u003cem\u003eJ Physiol\u003c/em\u003e \u003cstrong\u003e595\u003c/strong\u003e, 3549-3572, doi:10.1113/JP273915 (2017).\u003c/li\u003e\n \u003cli\u003eQiu, S.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Whole-brain spatial organization of hippocampal single-neuron projectomes. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e383\u003c/strong\u003e, doi:10.1126/science.adj9198 (2024).\u003c/li\u003e\n \u003cli\u003eQiu, S.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Whole-brain spatial organization of hippocampal single-neuron projectomes. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e383\u003c/strong\u003e, eadj9198, doi:10.1126/science.adj9198 (2024).\u003c/li\u003e\n \u003cli\u003eGong, H.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e High-throughput dual-colour precision imaging for brain-wide connectome with cytoarchitectonic landmarks at the cellular level. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 12142, doi:10.1038/ncomms12142 (2016).\u003c/li\u003e\n \u003cli\u003eGong, H.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e High-throughput dual-colour precision imaging for brain-wide connectome with cytoarchitectonic landmarks at the cellular level. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, doi:10.1038/ncomms12142 (2016).\u003c/li\u003e\n \u003cli\u003eRen, B., Shi, X., Zhao, B., Xu, C. \u0026amp; Wang, X. Reconstruction of Single-Neuron Projectomes in Mice. \u003cem\u003eBio Protoc\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, e5300, doi:10.21769/BioProtoc.5300 (2025).\u003c/li\u003e\n \u003cli\u003eWang, Q.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The Allen Mouse Brain Common Coordinate Framework: A 3D Reference Atlas. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e181\u003c/strong\u003e, 936-953 e920, doi:10.1016/j.cell.2020.04.007 (2020).\u003c/li\u003e\n \u003cli\u003eDolleman-van der Weel, M. J.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The nucleus reuniens of the thalamus sits at the nexus of a hippocampus and medial prefrontal cortex circuit enabling memory and behavior. \u003cem\u003eLearn Mem\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 191-205, doi:10.1101/lm.048389.118 (2019).\u003c/li\u003e\n \u003cli\u003eRoth, B. L. DREADDs for Neuroscientists. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e89\u003c/strong\u003e, 683-694, doi:10.1016/j.neuron.2016.01.040 (2016).\u003c/li\u003e\n \u003cli\u003eXu, P.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e High-throughput mapping of singleneuron projection and molecular features by retrograde barcoded labeling. \u003cem\u003eeLife\u003c/em\u003e, doi:https://doi.org/10.7554/eLife.85419 (2024).\u003c/li\u003e\n \u003cli\u003eHan, L.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Single-cell spatial transcriptomic atlas of the whole mouse brain. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e113\u003c/strong\u003e, 2141-2160 e2149, doi:10.1016/j.neuron.2025.02.015 (2025).\u003c/li\u003e\n \u003cli\u003eYao, Z.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e624\u003c/strong\u003e, 317-332, doi:10.1038/s41586-023-06812-z (2023).\u003c/li\u003e\n \u003cli\u003eNiell, C. M. \u0026amp; Stryker, M. P. Highly selective receptive fields in mouse visual cortex. \u003cem\u003eJ Neurosci\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 7520-7536, doi:10.1523/JNEUROSCI.0623-08.2008 (2008).\u003c/li\u003e\n \u003cli\u003eCai, D. J.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A shared neural ensemble links distinct contextual memories encoded close in time. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e534\u003c/strong\u003e, 115-118, doi:10.1038/nature17955 (2016).\u003c/li\u003e\n \u003cli\u003eGhosh, K. K.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Miniaturized integration of a fluorescence microscope. \u003cem\u003eNature Methods\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 871-878, doi:10.1038/nmeth.1694 (2011).\u003c/li\u003e\n \u003cli\u003eChen, T.-W.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Ultrasensitive fluorescent proteins for imaging neuronal activity. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e499\u003c/strong\u003e, 295-300, doi:10.1038/nature12354 (2013).\u003c/li\u003e\n \u003cli\u003eJimenez, J. C.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Contextual fear memory retrieval by correlated ensembles of ventral CA1 neurons. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 3492, doi:10.1038/s41467-020-17270-w (2020).\u003c/li\u003e\n \u003cli\u003eFiser, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Experience-dependent spatial expectations in mouse visual cortex. \u003cem\u003eNat Neurosci\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 1658-1664, doi:10.1038/nn.4385 (2016).\u003c/li\u003e\n \u003cli\u003eFerraris, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The Nucleus Reuniens Controls Long-Range Hippocampo-Prefrontal Gamma Synchronization during Slow Oscillations. \u003cem\u003eJ Neurosci\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 3026-3038, doi:10.1523/JNEUROSCI.3058-17.2018 (2018).\u003c/li\u003e\n \u003cli\u003ePURVIS, C. C. \u0026amp; DUNCAN, M. J. Discrete thalamic lesions attenuate winter adaptations and increase body weight. \u003cem\u003eAmerican Journal of Physiology-Regulatory, Integrative and Comparative Physiology\u003c/em\u003e, doi:https://doi.org/10.1152/ajpregu.1997.273.1.R226 (1997).\u003c/li\u003e\n \u003cli\u003eWilmot, C. A., Sullivan, A. C. \u0026amp; Levin, B. E. Effects of diet and obesity on brain a 1- and a2-noradrenergic receptors in the rat \u003cem\u003eBrain Research\u003c/em\u003e, doi:https://doi.org/10.1016/0006-8993(88)90154-0 (1988).\u003c/li\u003e\n \u003cli\u003eFreeman, D. A. \u0026amp; Zucker, I. Refractoriness to melatonin occurs independently atmultiple brain sites in Siberian hamsters. \u003cem\u003ePNAS\u003c/em\u003e, doi:https://doi.org/10.1073/pnas.111140398 (2001).\u003c/li\u003e\n \u003cli\u003eGiardino, L.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e DAILY MODIFICATIONS OF 3H-NALOXONE BINDING SITES IN THE RAT BRAIN: A QUANTITATIVE AUTORADIOGRAPHIC STUDY. \u003cem\u003eChronobiology International\u003c/em\u003e, doi:https://doi.org/10.3109/07420528909056920 (1989).\u003c/li\u003e\n \u003cli\u003ePrasad, J. A., Macgregor, E. M. \u0026amp; Chudasama, Y. Lesions of the thalamic reuniens cause impulsive but not compulsive responses. \u003cem\u003eBrain Struct Funct\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e, 85-96, doi:10.1007/s00429-012-0378-5 (2013).\u003c/li\u003e\n \u003cli\u003eCULLINAN, W. E., HERMAN, J. P., BATTAGLIA, D. F., AKIL, H. \u0026amp; WATSON, S. J. PATTERN AND TIME COURSE OF IMMEDIATE EARLY GENE EXPRESSION IN RAT BRAIN FOLLOWING ACUTE STRESS \u003cem\u003eNeuroscience\u003c/em\u003e, doi:https://doi.org/10.1016/0306-4522(94)00355-9 (1995).\u003c/li\u003e\n \u003cli\u003eOtake, K. \u0026amp; Nakamura, Y. Single midline thalamic neurons projecting to both the ventral striatum and the prefrontal cortex in the rat. \u003cem\u003eNeuroscience\u003c/em\u003e \u003cstrong\u003e86\u003c/strong\u003e, 635-649, doi:10.1016/s0306-4522(98)00062-1 (1998).\u003c/li\u003e\n \u003cli\u003eOtake, K. \u0026amp; Nakamura, Y. Single midline thalamic neurons projecting to both the ventral striatum and the prefrontal cortex in the rat. \u003cem\u003eNeuroscience\u003c/em\u003e, doi:https://doi.org/10.1016/S0306-4522(98)00062-1 (1998).\u003c/li\u003e\n \u003cli\u003eBerendse, H. W. \u0026amp; Groenewegen, H. J. Organization of the thalamostriatal projections in the rat, with special emphasis on the ventral striatum. \u003cem\u003eTHE JOURNAL OF COMPARATIVE NEUROLOGY\u003c/em\u003e, doi:https://doi.org/10.1002/cne.902990206 (1990).\u003c/li\u003e\n \u003cli\u003eVenkataraman, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Modulation of fear generalization by the zona incerta. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e \u003cstrong\u003e116\u003c/strong\u003e, 9072-9077, doi:10.1073/pnas.1820541116 (2019).\u003c/li\u003e\n \u003cli\u003eVenkataraman, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Incerto-thalamic modulation of fear via GABA and dopamine. \u003cem\u003eNeuropsychopharmacology\u003c/em\u003e \u003cstrong\u003e46\u003c/strong\u003e, 1658-1668, doi:10.1038/s41386-021-01006-5 (2021).\u003c/li\u003e\n \u003cli\u003eCorreia, S. S., McGrath, A. G., Lee, A., Graybiel, A. M. \u0026amp; Goosens, K. A. Amygdala-ventral striatum circuit activation decreases long-term fear. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, doi:10.7554/eLife.12669 (2016).\u003c/li\u003e\n \u003cli\u003eChen, A. P. F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Nigrostriatal dopamine modulates the striatal-amygdala pathway in auditory fear conditioning. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 7231, doi:10.1038/s41467-023-43066-9 (2023).\u003c/li\u003e\n \u003cli\u003eBelilos, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Nucleus accumbens local circuit for cue-dependent aversive learning. \u003cem\u003eCell Rep\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 113488, doi:10.1016/j.celrep.2023.113488 (2023).\u003c/li\u003e\n \u003cli\u003eHikida, T.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Pathway-specific modulation of nucleus accumbens in reward and aversive behavior via selective transmitter receptors. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e \u003cstrong\u003e110\u003c/strong\u003e, 342-347, doi:10.1073/pnas.1220358110 (2013).\u003c/li\u003e\n \u003cli\u003eYadav, N.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Prefrontal feature representations drive memory recall. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e608\u003c/strong\u003e, 153-160, doi:10.1038/s41586-022-04936-2 (2022).\u003c/li\u003e\n \u003cli\u003eO\u0026apos;Reilly, R. C. \u0026amp; Rudy, J. W. Conjunctive representations in learning and memory: principles of cortical and hippocampal function. \u003cem\u003ePsychol Rev\u003c/em\u003e \u003cstrong\u003e108\u003c/strong\u003e, 311-345, doi:10.1037/0033-295x.108.2.311 (2001).\u003c/li\u003e\n \u003cli\u003eHuckleberry, K. A., Ferguson, L. B. \u0026amp; Drew, M. R. Behavioral mechanisms of context fear generalization in mice. \u003cem\u003eLearn Mem\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 703-709, doi:10.1101/lm.042374.116 (2016).\u003c/li\u003e\n \u003cli\u003eHoward, M. W., Fotedar, M. S., Datey, A. V. \u0026amp; Hasselmo, M. E. The temporal context model in spatial navigation and relational learning: toward a common explanation of medial temporal lobe function across domains. \u003cem\u003ePsychol Rev\u003c/em\u003e \u003cstrong\u003e112\u003c/strong\u003e, 75-116, doi:10.1037/0033-295X.112.1.75 (2005).\u003c/li\u003e\n \u003cli\u003eMcTighe, S. M., Cowell, R. A., Winters, B. D., Bussey, T. J. \u0026amp; Saksida, L. M. Paradoxical false memory for objects after brain damage. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e330\u003c/strong\u003e, 1408-1410, doi:10.1126/science.1194780 (2010).\u003c/li\u003e\n \u003cli\u003eBesnard, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Dorsolateral septum somatostatin interneurons gate mobility to calibrate context-specific behavioral fear responses. \u003cem\u003eNat Neurosci\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 436-446, doi:10.1038/s41593-018-0330-y (2019).\u003c/li\u003e\n \u003cli\u003eMcClelland, J. L., McNaughton, B. L. \u0026amp; O\u0026apos;Reilly, R. C. Why There Are Complementary Learning Systems in the hippocampus and Neocortex: Insights From the Successes and Failures of Connectionist Models of Learning and Memory. \u003cem\u003ePsychological Review\u003c/em\u003e, doi:https://doi.org/10.1037/0033-295X.102.3.419 (1995).\u003c/li\u003e\n \u003cli\u003eWillshaw, D. J., Dayan, P. \u0026amp; Morris, R. G. Memory, modelling and Marr: a commentary on Marr (1971) \u0026apos;Simple memory: a theory of archicortex\u0026apos;. \u003cem\u003ePhilos Trans R Soc Lond B Biol Sci\u003c/em\u003e \u003cstrong\u003e370\u003c/strong\u003e, doi:10.1098/rstb.2014.0383 (2015).\u003c/li\u003e\n \u003cli\u003eSun, X.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Functionally Distinct Neuronal Ensembles within the Memory Engram. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e181\u003c/strong\u003e, 410-423 e417, doi:10.1016/j.cell.2020.02.055 (2020).\u003c/li\u003e\n \u003cli\u003eVincent, C. J., Aguilar-Alvarez, R., Vanderhoof, S. O., Mott, D. D. \u0026amp; Jasnow, A. M. An amygdala-cortical circuit for encoding generalized fear memories. \u003cem\u003eMol Psychiatry\u003c/em\u003e, doi:10.1038/s41380-025-03140-8 (2025).\u003c/li\u003e\n \u003cli\u003eBai, T.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Learning-prolonged maintenance of stimulus information in CA1 and subiculum during trace fear conditioning. \u003cem\u003eCell Rep\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 112853, doi:10.1016/j.celrep.2023.112853 (2023).\u003c/li\u003e\n \u003cli\u003eZhong, Q.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e High-definition imaging using line-illumination modulation microscopy. \u003cem\u003eNature Methods\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 309-315, doi:10.1038/s41592-021-01074-x (2021).\u003c/li\u003e\n \u003cli\u003eRohlfing, T. \u0026amp; Maurer, C. R., Jr. Nonrigid image registration in shared-memory multiprocessor environments with application to brains, breasts, and bees. \u003cem\u003eIEEE Trans Inf Technol Biomed\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 16-25, doi:10.1109/titb.2003.808506 (2003).\u003c/li\u003e\n \u003cli\u003eHinton, G. E. \u0026amp; Salakhutdinov, R. R. Reducing the dimensionality of data with neural networks. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e313\u003c/strong\u003e, 504-507, doi:10.1126/science.1127647 (2006).\u003c/li\u003e\n \u003cli\u003eZheng, G. X. Y.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Massively parallel digital transcriptional profiling of single cells. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, doi:10.1038/ncomms14049 (2017).\u003c/li\u003e\n \u003cli\u003eStuart, T.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Comprehensive Integration of Single-Cell Data. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e177\u003c/strong\u003e, 1888-1902 e1821, doi:10.1016/j.cell.2019.05.031 (2019).\u003c/li\u003e\n \u003cli\u003eHao, Y.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Integrated analysis of multimodal single-cell data. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e, 3573-3587 e3529, doi:10.1016/j.cell.2021.04.048 (2021).\u003c/li\u003e\n \u003cli\u003eRozeske, R. R.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Prefrontal-Periaqueductal Gray-Projecting Neurons Mediate Context Fear Discrimination. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e97\u003c/strong\u003e, 898-910 e896, doi:10.1016/j.neuron.2017.12.044 (2018).\u003c/li\u003e\n \u003cli\u003eZhou, P.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Efficient and accurate extraction of in vivo calcium signals from microendoscopic video data. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, doi:10.7554/eLife.28728 (2018).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7828569/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7828569/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe balance between memory specificity and generalization is essential for animals\u0026rsquo; adaptive behaviours in diverse contexts, and is controlled by thalamus and hippocampus. Yet, whether and how thalamo-hippocampal circuitry regulates this balance remains elusive. Using single-neuron projectome analysis, we found that dorsal and ventral subregions of ventral CA1 (vCA1d and vCA1v) received inputs from two distinct neuronal populations in the nucleus of reuniens (NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons), respectively. Cell type-specific trans-synaptic retrograde tracing and circuit-barcoded single-cell RNA sequencing of NRe\u003csup\u003evCA1d\u003c/sup\u003e and NRe\u003csup\u003evCA1v\u003c/sup\u003e neurons uncovered distinct presynaptic inputs as well as gene expression patterns underlying their different intrinsic excitabilities and spike waveforms. Using single-unit recording, circuit specific chemogenetic inhibitions, and miniscope Ca\u003csup\u003e2+\u003c/sup\u003e imaging, we found that NRe\u003csup\u003evCA1d\u003c/sup\u003e (but not NRe\u003csup\u003evCA1v\u003c/sup\u003e) neurons and downstream vCA1d neurons exhibited reduced contextual discriminability, promoting memory generalization after threat learning. These results demonstrate that the NRe-vCA1d circuit regulates memory precision by promoting memory generalization over specificity.\u003c/p\u003e","manuscriptTitle":"A thalamo-hippocampal circuit regulating memory precision during contextual learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-08 12:29:21","doi":"10.21203/rs.3.rs-7828569/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"c71875c7-04f7-4010-928a-be751a97dbf3","owner":[],"postedDate":"December 8th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":56358708,"name":"Biological sciences/Neuroscience/Learning and memory/Hippocampus"},{"id":56358709,"name":"Biological sciences/Molecular biology/Transcriptomics"}],"tags":[],"updatedAt":"2025-12-08T12:29:21+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-08 12:29:21","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7828569","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7828569","identity":"rs-7828569","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}