Neuron

Escape from threat is one of the most ancient and conserved sensorimotor transformations and must satisfy two competing demands. It must be fast and reliable, because failing to escape genuine threats risks death, yet also selective, because escaping indiscriminately costs energy and missed resources. Speed and reliability favour a system with a low threshold that responds to the slightest indication of danger, whereas selectivity demands a high threshold that filters out innocuous stimuli - yet both must be achieved simultaneously. While previous work has identified mechanisms for implementing selectivity, it is not known how the mammalian brain achieves speed and reliability once genuine threats have been identified. Here we use whole-cell recordings from dorsal periaqueductal gray (dPAG) neurons in behaving mice to show that the escape circuit resolves this tension via high intrinsic excitability at the single cell level. We find that while the membrane potential of dPAG neurons does not reach action potential threshold during exploratory behaviour, they require only a small amount of current to fire. Threat stimuli cause a sparse increase in synaptic input rate that, because of the high excitability, produces a sustained depolarizing voltage step that drives spiking and escape behaviour. Individual dPAG neurons respond similarly on escape and non-escape trials - what determines escape is the fraction of the dPAG population that is recruited, a finding we confirm with single-unit recordings in freely moving mice. The membrane voltage step response is also invariant to threat type, in contrast to superior colliculus neurons where membrane potential dynamics reflect stimulus identity. These findings reveal a single neuron mechanism for fast and reliable escape, in which the high intrinsic excitability of dPAG neurons transforms sparse, threat-evoked synaptic input into a rapid and stereotyped voltage signal, converting diverse threat representations into a uniform escape command.

Flexible behavior requires using past experiences to configure cortical computations to suit current task demands. A central question in neuroscience is how memory representations control such reconfigurations. Although hippocampal (HPC) engrams can drive learned behaviors, prior studies have been largely limited to a fixed stimulus-response computation. Thus, whether engrams can drive retrieval of a set of stimulus-response mappings, rather than a specific response itself, remains unresolved. Moreover, how engrams affect cortical task representations and dynamics so as to produce engram-consistent behavior remains unstudied. Here, we address both questions by combining tagging and reactivation of HPC engrams with simultaneous large-scale recordings in medial prefrontal cortex (mPFC) during a context-dependent task-switching paradigm in mice. We report that HPC engram reactivation caused mice to apply engram-consistent decision-rules rather than a specific motor output. Simultaneous mPFC recordings revealed that reactivating the HPC engram for a given context reinstated the representation of that context in mPFC within hundreds of milliseconds, indicating that it was mediated by rapid network effects, leading to choice behavior consistent with the reactivated HPC engram. Other aspects of endogenous dynamics in mPFC were left remarkably intact. Together, our findings provide direct causal evidence that HPC engrams can configure task-relevant population states of downstream cortical circuits in real time, establishing a neural mechanism by which memory traces control flexible behavior.

Human genetics implicates innate immunity as a key modifier of tau toxicity, yet human-specific neuroimmune mechanisms remain difficult to test in vivo. Here, we developed HuMiNAX, the first humanized iPSC-based neuroimmune xenograft model of tau-associated neurodegeneration, enabling human microglia to interact with human neurons and astrocytes in the adult mouse brain. In HuMiNAX, tau seeding induced aggregation only in mutation-carrying human neural grafts, causing neuron loss and inflammatory activation of human microglia. Progranulin-overexpressing human microglia dampened tau-associated inflammation, preserved neurons, and restored neuronal gene-expression and RNA-splicing programs, supporting microglial control of neuronal resilience. CRISPRi knockdown of the human-specific lncRNA HNRNPK-AS1 also protected neurons in HuMiNAX. These findings establish HuMiNAX as a human neuroimmune model of tauopathy and identify microglial and RNA-mediated strategies of neuronal resilience.

Animals must execute learned behaviors and update them when outcomes change, yet the neural substrates controlling this phenomenon are not fully understood. Here, we show that N-Methyl-D-Aspartate Receptors (NMDARs) in the dorsomedial striatum are necessary for learning from previously rewarded actions. Moreover, blocking of striatal NMDARs almost fully abolished striatal calcium dynamics, but not action potential activity, suggesting a unique function of NMDAR-driven striatal calcium activity in updating action policy.

Flexible decisions require the brain to transform sensory evidence into abstract, task-relevant variables and then into actions. Understanding this process requires identifying how distributed neural populations represent sensory, cognitive, and motor variables, and how interareal interactions mediate transformations between them. We simultaneously recorded population activity in frontal eye field (FEF), lateral intraparietal area (LIP) and superior colliculus (SC) while monkeys performed a flexible yet urgent visual motion-categorization task. Within this network, FEF first encoded abstract categories, followed by SC and then LIP. LIP showed the earliest encoding of visual stimulus features, but a later encoding of upcoming saccades. Single-trial analyses revealed directed information flow from FEF to LIP populations for category- and choice-related signals. Reversible FEF inactivation impaired categorization and saccadic choice, causally implicating FEF in category-guided action. These findings reveal a differentiated FEF-LIP-SC circuit for transforming sensory evidence into abstract categorical decisions and the actions used to report them.

Striatal cholinergic interneurons (CINs) exhibit a transient pause in tonic firing in response to salient stimuli, a hallmark of reinforcement learning that becomes synchronized with learning. Although thalamostriatal and dopaminergic inputs have been implicated in this pause, the underlying circuit mechanisms remain unclear. Here, we combine optogenetics, electrophysiology, and genetic approaches to examine inhibitory interactions within the CIN network. Synchronized activation of CINs in striatal slices elicited robust feedback inhibition in CINs, suppressing firing and generating pause-like responses. This inhibition was mediated by GABAA receptors and required beta2-containing nicotinic acetylcholine receptors (beta2-nAChR), and could be recruited by thalamostriatal activation. Dopamine is not required for this circuit but modulates it via D2 receptors. Surprisingly, cell-type-specific silencing and striatal beta2-nAChR deletion excluded local GABAergic sources, whereas retrograde beta2-nAChR deletion abolished inhibition, revealing an extrastriatal pathway. These findings identify a long-range inhibitory mechanism linking synchronized cholinergic activity to pause generation in striatal circuits.

Adaptive behavior requires sensory systems to prioritize cues that predict meaningful outcomes while suppressing irrelevant stimuli, but how relevance-based filtering is implemented along early sensory pathways remains unclear. Using deep brain two-photon imaging and causal circuit manipulations in mice performing an audiovisual detection task, we show that inhibition from thalamic reticular nucleus dynamically tunes sensory thalamus according to learned value. As animals learned stimulus-outcome associations, neurons in medial geniculate body developed biased responses favoring reward-predicting cues and suppressing non-rewarded stimuli. Silencing inhibitory input from thalamic reticular nucleus broadly disinhibited thalamic responses and abolished this value bias. Notably, stimulus identity decoding was unaffected by loss of inhibition. However, the geometry of MGB population activity was reorganized: coding axes rotated, action-related coding was strongly impaired, and thalamic representations became misaligned with learned behavioral readout. This altered population code impaired behavioral performance despite preserved sensory separability. Thus, inhibition of sensory thalamus by the reticular nucleus does not simply gate sensory throughput; it acts as a subcortical coding mechanism that aligns neuronal representations with learned value and behavioral goals, with implications for disorders of perception and cognition.

AbstractThe basal ganglia (BG) are a set of topographically organized, interconnected structures that are pivotal for regulating volitional movement and other aspects of cognitive, motivational, and affective behavior. Recently generated taxonomies of transcriptomically-defined cell types (T-types) have revealed both fine-grained distinctions in gene expression between neurons in these structures as well as continuous transcriptomic variation across similar T-types1-9, which are both related to location within a structure. However, it remains unclear to what extent these and other cellular properties co-vary with each other. Therefore, we performed Patch-seq experiments10,11 on over 900 neurons in mouse brain slices from BG to provide an integrated view of the co-variation between gene expression, location, physiology, and morphology measured from the same neurons. Medium spiny neurons (MSNs) from both the direct and indirect pathways across the dorsal and ventral striatum follow a gene expression gradient that varies in a dorsolateral to ventromedial direction; we find that this gradient also corresponds with systematic differences in action potential kinetics and dendritic arborization. Our analysis also characterizes additional multimodal dimensions of MSN variation, such as those between direct and indirect pathway neurons and between matrix and striosome neurons. Furthermore, through comparison with Patch-seq data from macaque, we demonstrate that the relationship between the transcriptomic/spatial gradient and electrophysiological and morphological properties is conserved across these two species. We also find that properties of striatal interneurons, such as action potential kinetics, vary across the striatum in a manner consistent with the MSN gradient. Outside the striatum, our multimodal Patch-seq dataset from the globus pallidus, subthalamic nucleus, and substantia nigra enabled us to characterize transcriptomically-defined types and link them to prior descriptions of cell types in these structures. Finally, we examined to what extent the MSN transcriptomic/spatial gradient persisted across different stages of the BG circuit by comparing Patch-seq neurons to reconstructed whole-neuron morphologies and the topography of their interareal projections, finding that the gradient is better preserved in GPe and GPi compared to SNr. Our study links transcriptomic variation across T-types in mouse BG to spatial localization and phenotypic differences at the level of individual cells, improving our understanding of cell type architecture in topographically organized circuits of the brain.

Astrocytes influence synaptic plasticity and neuronal function through astrocytic calcium dynamics (ACD). However, astrocyte contribution to cognitive operations like reinforcement learning (RL) remains unclear. To examine this, we trained mice on a RL dependent probabilistic decision-making task. We attenuated ACD across distinct striatal regions, and found ACD attenuation specifically in ventral striatum (VS) increased decision noisiness and impaired reward-guided choice performance. This effect was largely due to a reduction in win-stay behavior. Using in-vivo calcium imaging, we found that VS ACD correlated with reward prediction errors (RPEs). Furthermore, these trial-by-trial ACD fluctuations predicted trial-by-trial choice variability. In-silico lesions of a biologically constrained circuit model suggest that astrocytes could regulate behavioral variability by sharing RPE signals across populations of striatal neurons. Together, these results suggest that VS astrocytes contribute to cortico-striatal functions to mediate decision noisiness.

We utilize internal representations of meaning for two purposes: to understand the words we hear and to generate our own speech. This dual requirement necessitates abstract, modality-agnostic representations. Building on work identifying it as a hub for relational mapping, we hypothesized that the hippocampus supports abstract, cross-person representations, and uses shared semantic geometries to do so. We tested this hypothesis by examining hippocampal activity in a remarkable single-neuron dataset derived from conversational speech. Neurons robustly encoded meanings of both spoken and heard words, and used common geometric embeddings for both, leading to abstract meaning performance. Speaker identity was aligned with meaning via partial subspace alignment, which affords speaker-meaning binding by partitioning meaning by speaker while maintaining cross-speaker generalization. Degrees of subspace rotation varied on a single word level and depended systematically on semantic category. Together, these findings indicate how geometric principles allow for abstract cross-personal meanings while preserving binding to speaker identity.

Across mammals, brain regions can duplicate, expand, and diversify, requiring long-range connectivity to accommodate species-specific specializations while preserving globally ordered wiring. This challenge is especially pronounced in primate thalamocortical circuits, where select cortical fields and their thalamic partners have expanded disproportionately. Although gene expression in the thalamus follows broad and conserved gradients, how thalamocortical projections are organized at single-neuron resolution, and how this organization is reshaped by expansion, remain unknown. Here, we investigate thalamocortical projection organization by in situ sequencing and BARseq projection mapping in marmosets and mice. We profiled the gene expression of 1.5 million marmoset neurons and jointly measured gene expression and cortical projections in 708 marmoset and 1,518 mouse neurons that spanned multiple thalamic nuclei. In both species, projections of individual neurons targeted diverse areas that together spanned a large fraction of the cortex. Comparing projections at the single-neuron level and local neighborhood level revealed that marmoset thalamocortical projections were more spatially segregated, producing a point-to-point architecture. Strikingly, this local specialization coexisted with a conserved gradient that predominated over discrete anatomical borders: In the higher-order sensory thalamus of both species, gene expression and projections varied continuously across nucleus borders, and borders had only a small effect on projections. Furthermore, in both marmoset and mouse, gene expression gradients were associated with the anteroposterior locations of cortical targets. These results reconcile discrete nucleus and gradient-based models of thalamic organization and suggest that primate circuit specialization is superimposed on a conserved molecular-projection gradient.

Episodic memory requires reconstructing the position of the self within a remembered environment, yet whether and how memory engrams incorporate self-referenced spatial information remains unknown. Using activity-dependent engram tagging, longitudinal calcium imaging and optogenetic perturbation in the retrosplenial cortex, we found that cortical memory engrams are enriched for egocentric and boundary-coding neurons that encode self-position relative to environmental boundaries. Longitudinal imaging revealed that future engram neurons are preferentially recruited from a pre-existing spatial scaffold rather than generated de novo during learning. During memory retrieval, scaffold populations underwent coordinated refinement and became transiently reorganized into a scaffold-engram network state whose dynamics tracked memory expression. Silencing scaffold neurons reduced memory expression while preserving recall dynamics, whereas engram silencing abolished recall dynamics while maintaining a stable low-memory state. Together, these findings identify a scaffold-engram architecture through which episodic memories are reconstructed by reinstating the self-referenced spatial representations within remembered space.

Slowly varying internal states govern our ability to sustain goal-directed behavior over minutes to hours, imposing fundamental constraints on cognitive performance, yet the neural signals that track these states remain poorly defined. Dopamine is a powerful modulator of motivated behavior, but its best-characterized signals are in the form of reward prediction errors (RPEs) that operate at seconds timescales. Whether these fast signals also carry information about slower, ongoing states of behavioral performance has not been directly tested. Here, we recorded subsecond dopamine concentration changes across multiple sites in the caudate nucleus (CN) and putamen of rhesus monkeys performing a reward-guided saccade task. Task performance oscillated over tens to hundreds of trials, demonstrating fluctuating internal states of sustained engagement. We found that single-trial dopamine signals were strongly modulated by these slowly evolving performance states, in a manner dissociable from both trial-level RPE and cumulative reward rate. This relationship was biased toward future rather than past performance windows, indicating that dopamine tracks motivational state prospectively rather than simply reflecting the reward history on which prediction errors are computed. Furthermore, performance-state modulation was concentrated in the CN rather than the putamen, consistent with the distinct roles of these regions in oculomotor and skeletomotor control and suggesting that state-dependent dopamine signals are organized according to the behavioral demands of the task. These findings demonstrate that phasic dopamine signals in the dorsal striatum simultaneously reflect fast learning signals and slowly evolving internal states, linking dopamines canonical RPE function to its broader role in sustaining motivated performance.

Hippocampal place cell activity represents an animals location in space; yet, how hippocampal neuronal population dynamics change with spatial learning and the mechanisms underlying these activity changes, which drive allocentric navigation to a learned goal, are poorly understood. To address these questions, we performed calcium imaging with a novel wire-free waterproof miniaturized microscope to image the activity of large populations of hippocampal CA1 neurons during spatial learning of a two-dimensional navigational task, the Morris water maze. We followed the same cells during learning and were able to directly examine how each neuron in the ensemble, and the ensemble as a whole, changes its response properties. We found that neuronal spatial selectivity increased and population decoding of spatial location improved as mice learned to navigate to the goal. Viral CRISPR knock out of Grin1 (encoding the essential GluN1 NMDA receptor subunit) in dorsal hippocampal neurons, dramatically reduced long-term potentiation in CA1. This manipulation also prevented the increase in spatial selectivity and improvement of population decoding with spatial learning and resulted in learning deficits in the Morris water maze. Together, our results show that dorsal hippocampus NMDAR-dependent synaptic plasticity is essential for the learning-dependent refinement of CA1 place selectivity and improvement in population decoding of space.

System consolidation transforms temporary hippocampal representation of memory into long-term storage in cortex. The underlying neural substrate, however, remains enigmatic. Here, we tracked the spatiotemporal evolution of hippocampus (HPC)-cortex local field potentials and single-neuron spikes in behaving animals during fear memory formation. During learning, HPC fast gamma exhibited a progressive phase shift relative to PFC theta oscillations, with gamma power aligning to progressively later phases of the PFC theta cycle. Strikingly, a related phase-shifted coupling pattern re-emerged during subsequent consolidation in association with hippocampal sharp-wave ripples and transient PFC spindle events during NREM sleep. Across this process, interregional interactions evolved from HPC-driven cortical gamma coherence at recent stages to PFC-mediated cortical low-frequency coherence at remote stages. Using closed-loop optogenetic perturbations, we demonstrated a stepwise causal chain of coupling events underlying remote memory formation. Our study revealed HPC-PFC coupling phase shift as a feasible substrate mediating recent-to-remote transformation of memory.

Throughout the mammalian cortex, populations of neurons must work together to enact behaviors. While population recordings in motor cortex have revealed many aspects of when neurons fire during behaviors, limitations in causal experiments have made it difficult to identify which features of neural activity directly drive movements and which do not. Here, we explicitly test the principles of neural coding using high temporal precision multiphoton holographic optogenetics in the motor cortex. We show that activation of a small number (50-75) of Layer 2/3 excitatory neurons in the motor cortex is sufficient to drive movements. The efficacy of stimulation-driven movement depends on the state of the local circuit, and, to a lesser extent, the identity of which neurons are stimulated. We test whether evoked activity is acting through a rate code or a timing code by holding the firing rate and cell identities constant while varying the millisecond precise timing of activation. We find an unexpected and strong dependence on inter-cell synchrony when evoking movements. This neural synchrony recruits distinct patterns of recurrent excitation and inhibition. These findings provide evidence that the timing code, more so than the rate code, drives motor output.

Prediction errors (PEs) drive perceptual learning by updating internal models of the sensory environment, yet it remains unclear how attention reshapes their representation across distributed thalamocortical circuits. Using intracranial stereoelectroencephalography (sEEG) from 17 patients performing a roving auditory oddball task under attended and unattended conditions, we quantified PE encoding using mutual information and co-information to capture redundant and synergistic PE representations. Attention modulated PE encoding in both the thalamus and the temporal cortex, but with distinct informational dynamics. Thalamic encoding showed a stable reduction of PE information during distraction, consistent with state-dependent thalamocortical gating. In contrast, the temporal cortex expressed two opposing learning trajectories during attended listening that converged once attention was diverted, revealing distinct cortical learning regimes rather than a uniform attentional effect. Attention further reorganized the informational content of cortical PE representations by altering the balance between redundant and synergistic information. A biologically constrained neural network showed that attention-dependent changes in inhibition and long-range connectivity reproduced these dynamics through Hebbian learning. Together, these findings suggest that attention regulates predictive learning not simply by changing the strength of PE responses, but by reshaping how distributed thalamocortical circuits represent and integrate sensory evidence over time.

Animals must constantly calibrate the costs and benefits of exploration of an environment based on expectations of danger. These decisions are strongly shaped by past experience of perceived threats within that environment and by internal state, which is strongly modulated by circulating gonadal hormones. Although the circuits underlying threat detection are relatively well characterized, how sex hormones shape the long-term behavioral consequences of prior threat experience, and whether this differs across sexes, remains unknown. Here, we show that female mice, like males, exhibit robust long-term threat avoidance (LTTA), avoiding a location where they previously experienced a single visual threat. However, we find that in females this behavior shows strong modulation by the estrous cycle. Surprisingly, we find that though male and female LTTA is driven through glutamate release by the melanopsin-projecting intrinsically photosensitive retinal ganglion cells (ipRGCs) in the thalamic perihabenular nucleus, disruption of this circuit drives completely opposing effects on male versus female LTTA. Moreover, hormonal modulation of LTTA in females requires functional ipRGC input. Thus, despite similar circuit architecture and behavioral outcomes, the individual components of the LTTA circuit play opposing roles in shaping this behavior in males and females, and female LTTA is further tuned by hormonal status.

The medial entorhinal cortex (mEC) contains functionally specific neurons crucial for spatial memory and navigation, including grid, border, spatial, head direction, and cue cells. However, how these neurons interact to build spatial maps remains unclear. Here, using simultaneous recordings from hundreds of functionally defined mEC neurons in mice navigating virtual tracks with varying numbers of visual landmarks, we uncovered connectivity motifs that define the mEC functional network architecture and link it to the principles governing allocentric map formation. We found that connectivity between cells was cell-type-specific and grouped into subnetworks based on their function and theta modulation, with theta modulated connections dominating over non-theta. Functionally distinct neurons preferentially connected to their own type, and their interactions were coordinated by a shared inhibitory pool of interneurons, with a higher proportion of excitatory-to-inhibitory connections than between excitatory cells. This was accompanied by the number of fields formed across all mEC cell types increasing sublinearly with the number of available cues. Grid cells showed the strongest relative connectivity to interneurons regardless of their theta modulation, linking otherwise largely isolated theta and non-theta streams. Grid cells were also least likely to form a field in response to cues. Together, these findings reveal an mEC network architecture organised by cell type and theta dependency, in which inhibitory interactions, with grid cells as the strongest hub, play a central role in building the mEC allocentric map.

One SentenceMaternal experience fundamentally reorganizes infant cue processing across the entire brain into a more efficient and specialized caregiving state. Pup calls signal the presence of a mouse infant in need of care (Bernd Haack et al., 1983; Ehret, 2005). Previous work demonstrated that experience-dependent plasticity within the left primary auditory cortex is oxytocin-facilitated and enhances neural and behavioral responses to these distress vocalizations (Marlin et al., 2015). However, whether maternal experience reshapes pup call processing beyond auditory regions remains unknown. Here, we show that both innate and learned maternal experience shape pup call-evoked behavior and brain-wide neural responses. Combining behavioral assays, whole-brain activity mapping, and oxytocinergic projection analyses, we find that pup calls recruit distinct large-scale neural networks in naive virgins, experienced virgins, and mothers. With more maternal experience we observe a distributed pup call response network that is sparser yet more strongly coactivated, consistent with a more efficient and specialized neural representation of infant cues. We identified an oxytocin-dense circuit that shows an experience-dependent pup call response pattern. Therefore, we propose that experience-dependent refinement of the brain-wide response to pup calls is facilitated by oxytocin. Taken together, our findings reveal that maternal experience refines brain-wide sensory processing to support adaptive caregiving behavior.