Disease-associated microglial states are thought to contribute to Alzheimer's disease (AD) progression, but characterizing them and their relationships to pathology remains challenging. Here we introduce CODEX-CNS-a multiplexed protein imaging technology with a custom data analysis pipeline for use in human brain samples. We profiled 704,706 cells in samples from the frontal cortex of 8 people with AD and 8 healthy controls and mapped features including blood-brain barrier, meningeal components and cell-cell interactions within the same tissue sections. Amongst the myeloid cell populations we identified, we found a border-associated macrophage-like microglial subset associated with aging. Further classifying myeloid cell subsets based on their spatial neighborhood, we identified a border-associated macrophage-like microglial subpopulation that was associated significantly with dense amyloid-β plaques, which we termed human plaque-associated microglia. This work offers insights into myeloid cell heterogeneity in AD and provides a new spatial approach to characterizing brain cells at the single-cell protein level.

It is often assumed that phenotypic heterogeneity in autism reflects underlying pathobiological variation. However, direct evidence supporting this link is lacking. Leveraging cross-species functional neuroimaging, we show that brain dysconnectivity patterns in autism can be parsed into biologically dissociable subtypes. Specifically, we found that functional magnetic resonance imaging (fMRI) connectivity alterations in 20 distinct genetic mouse models of autism cluster into hypoconnectivity-dominant and hyperconnectivity-dominant subtypes. These subtypes are linked to distinct biological pathways, with hypoconnectivity being associated with synaptic dysfunction and hyperconnectivity reflecting transcriptional and immune-related alterations. Here we identified analogous hypoconnectivity and hyperconnectivity subtypes in a multicenter human fMRI dataset of n = 940 individuals with idiopathic autism and n = 1,036 neurotypical individuals. The human autism subtypes are highly replicable, are associated with distinct functional network architectures and behavioral profiles and recapitulate the synaptic and immune-related pathways identified in the rodent dataset. Our work provides a new empirical framework for targeted subtyping of the autism spectrum.

Synaptic inputs onto the dendrites and cell body integrate and trigger action potentials (APs) at the axon initial segment (AIS). The AIS receives GABAergic synaptic inputs; however, the structure and function of glutamatergic synapses at the AIS remain poorly understood. Here, we show that, in adult mice, the AIS exhibits axonic spines in about half of the neurons examined in three brain regions: the dorsal lateral septum (dLS), bed nucleus of the stria terminalis and striatum. In the dLS, axonic spines express ionotropic glutamate receptors and undergo structural plasticity. Voltage-gated Na+ channels at the AIS boost the synaptic responses of axonic spines and thus AP generation. Although hippocampal dorsal CA3 neurons synapse onto both axonic spine neurons (ASNs) and non-ASNs, they preferentially activate ASNs and subsequently inhibit non-ASNs through feedforward inhibition. Together, these results indicate that axonic spines jump-start APs in dLS neurons and route information flow from the hippocampus to downstream brain regions.

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder marked by progressive motor neuron (MN) degeneration in the brain and spinal cord. Although neuroinflammation is increasingly recognized as a hallmark of ALS, the precise molecular programs linking immune responses to MN pathology remain poorly defined. Using an integrated approach that combines single-cell and bulk RNA sequencing with spatial proteogenomics, we characterized both shared and distinct immune dynamics in peripheral blood and spinal cord tissues from patients with sporadic ALS and those carrying C9orf72 repeat expansions. Our analysis revealed broad immune remodeling in C9orf72 ALS, ALS subtype-specific and progression-associated differences in monocyte activation and antigen-experienced CD8 effector memory T cells with clonal features consistent with antigen-driven responses. Spatial mapping revealed complement activation and lipid-programmed myeloid states converging at sites of MN loss and TDP-43 pathology. Together, these findings connect peripheral and central immune alterations to ALS heterogeneity and highlight stratified immunomodulation as a potential therapeutic strategy.

Motor learning relies on signals that instruct adaptive plasticity following errors. In the cerebellum, climbing fibers (CFs) provide these instructions to Purkinje cells (PCs). Yet CFs fire continuously, even without errors, requiring molecular layer interneuron (MLI) inhibition of PCs to counteract CF excitation and prevent maladaptive plasticity. Here, to identify how this regulatory inhibition is contextually suppressed to selectively permit error-driven learning in mice, we combined connectomics, functional recordings, computational modeling and behavioral manipulations. We discovered that CFs target not only PCs but also a specific MLI subtype that inhibits PC-targeting MLIs, creating serial disinhibition. These disinhibitory MLIs integrate multiple CFs, causing increased activation with CF synchrony. This stronger disinhibitory drive allows larger CF-evoked calcium responses in PCs. Disruption of MLI-to-MLI inhibition prevents CF-instructed motor learning, confirming the necessity of this disinhibitory pathway. Therefore, population synchrony selectively enables CF-driven plasticity through disinhibitory network interactions, demonstrating that instructive signaling is a product of circuit-level processing.

Spontaneous fluctuations in attention can impede adaptation to changing goals and environments. Endogenous control over attentional shifts, referred to as attentional flexibility, is prone to disruption in children with attention deficit disorders. Here we studied in vivo intracranial recordings in children with epilepsy to identify a reproducible neural signature of attentional control that could predict and prevent impending lapses in real time. Machine learning classifiers were trained on intracranial signals while each child performed an attentional set-shifting task and predicted delays in attention shifting over multiple days and across several pediatric populations. Intracranial electrical stimulation in response to impending delays rescued attention shifts indexed by eye tracking, reaction time and accuracy. Simultaneous electroencephalography identified corresponding scalp signatures that enabled noninvasive modulation of attention shifting in healthy participants. These findings provide insight into the neural basis of attentional shifts with implications for targeted neuromodulation and exogenous attentional control.

Our perception of the world depends on a complex interplay between external sensory inputs and our internal states. How and where in the brain these interactions are implemented remain poorly understood. To address these questions, we measured membrane potential (Vm) of single V1 neurons in macaque monkeys performing a reaction-time visual detection task. Here we show that most V1 neurons gradually depolarize in preparation for target onset, and that variations in this buildup are correlated with the monkey's reaction times. Further, we show that fluctuations in Vm after target onset are correlated with choice, and that these covariations strongly depend on the location and contrast of the target. Finally, we show that a simple computational model with fluctuating multiplicative gain can account for our results. Thus, the surprising covariations between Vm of single V1 neurons and behavior are implemented by internal-state-related nonlinear modulations operating at, or before, V1.

Spatial transcriptomics has emerged as a transformative approach for in situ mapping of cellular heterogeneity and interactions, yet existing methods often compromise throughput, cost and tissue coverage. Here we introduce Imaging Reconstruction using Indexed Sequencing (IRISeq): an optics-free, cost-effective platform that leverages spatial interaction mapping by indexed sequencing to profile tissues at adjustable sizes and resolutions (5-50 µm). We applied IRISeq to map gene expression across more than 70 coronal sections from both adult and aged mouse brains, including wild-type and two lymphocyte-deficient models (Rag1 and Prkdc mutants) and generated more than 460,000 spatial transcriptome profiles. Our integrated analysis with 783,264 single-cell transcriptomes revealed region-specific aging signatures that are lymphocyte dependent, notably a downregulation of interferon signaling and inflammation in ventricular regions upon lymphocyte depletion, alongside mutant-specific upregulation of senescence pathways. Furthermore, lymphocyte deficiency was linked to preserved abundance of ependymal cells that line the brain's ventricles and to distinct microglial state dynamics, highlighting a key role for lymphocytes in driving inflammatory processes during brain aging. Overall, IRISeq provides a high-throughput and cost-effective solution for spatially resolved transcriptomic profiling, opening new avenues for elucidating region-specific cellular mechanisms underlying aging and identifying potential therapeutic targets to preserve brain homeostasis.

Animals can learn about danger by observing conspecifics, but whether and how they acquire behaviors through positive affective states of others is not understood. Here we show that by observing a demonstrator, mice learn to take actions that benefit others and that are goal-directed and flexible. Chemogenetic silencing experiments showed that activity in hippocampal dorsal CA1 (dCA1) was required for observers to learn action-outcome associations in a social context. Fiber photometry recordings revealed inter-individual differences in dCA1 activity patterns during observation that tracked the observer's subsequent prosocial or selfish behavioral propensities. Optogenetic manipulations demonstrated that the dCA1 is key during observation and can orient a mouse's actions toward prosocial or selfish choices in future interactions. Our study provides a mouse model of social transmission of knowledge that guides prosocial behavior and that may be relevant for studying disorders in which the ability to learn from others' actions is compromised.