Spatial Transcriptomics: Molecular Maps of the Mammalian Brain.

Maps of the nervous system inspire experiments and theories in neuroscience. Advances in molecular biology over the past decades have revolutionized the definition of cell and tissue identity. Spatial transcriptomics has opened up a new era in neuroanatomy, where the unsupervised and unbiased exploration of the molecular signatures of tissue organization will give rise to a new generation of brain maps. We propose that the molecular classification of brain regions on the basis of their gene expression profile can circumvent subjective neuroanatomical definitions and produce common reference frameworks that can incorporate cell types, connectivity, activity, and other modalities. Here we review the technological and conceptual advances made possible by spatial transcriptomics in the context of advancing neuroanatomy and discuss how molecular neuroanatomy can redefine mapping of the nervous system.

The spatial landscape of gene expression isoforms in tissue sections.

In situ capturing technologies add tissue context to gene expression data, with the potential of providing a greater understanding of complex biological systems. However, splicing variants and full-length sequence heterogeneity cannot be characterized at spatial resolution with current transcriptome profiling methods. To that end, we introduce spatial isoform transcriptomics (SiT), an explorative method for characterizing spatial isoform variation and sequence heterogeneity using long-read sequencing. We show in mouse brain how SiT can be used to profile isoform expression and sequence heterogeneity in different areas of the tissue. SiT reveals regional isoform switching of Plp1 gene between different layers of the olfactory bulb, and the use of external single-cell data allows the nomination of cell types expressing each isoform. Furthermore, SiT identifies differential isoform usage for several major genes implicated in brain function (Snap25, Bin1, Gnas) that are independently validated by in situ sequencing. SiT also provides for the first time an in-depth A-to-I RNA editing map of the adult mouse brain. Data exploration can be performed through an online resource (https://www.isomics.eu), where isoform expression and RNA editing can be visualized in a spatial context.

Assessing the replicability of spatial gene expression using atlas data from the adult mouse brain.

High-throughput, spatially resolved gene expression techniques are poised to be transformative across biology by overcoming a central limitation in single-cell biology: the lack of information on relationships that organize the cells into the functional groupings characteristic of tissues in complex multicellular organisms. Spatial expression is particularly interesting in the mammalian brain, which has a highly defined structure, strong spatial constraint in its organization, and detailed multimodal phenotypes for cells and ensembles of cells that can be linked to mesoscale properties such as projection patterns, and from there, to circuits generating behavior. However, as with any type of expression data, cross-dataset benchmarking of spatial data is a crucial first step. Here, we assess the replicability, with reference to canonical brain subdivisions, between the Allen Institute's in situ hybridization data from the adult mouse brain (Allen Brain Atlas (ABA)) and a similar dataset collected using spatial transcriptomics (ST). With the advent of tractable spatial techniques, for the first time, we are able to benchmark the Allen Institute's whole-brain, whole-transcriptome spatial expression dataset with a second independent dataset that similarly spans the whole brain and transcriptome. We use regularized linear regression (LASSO), linear regression, and correlation-based feature selection in a supervised learning framework to classify expression samples relative to their assayed location. We show that Allen Reference Atlas labels are classifiable using transcription in both data sets, but that performance is higher in the ABA than in ST. Furthermore, models trained in one dataset and tested in the opposite dataset do not reproduce classification performance bidirectionally. While an identifying expression profile can be found for a given brain area, it does not generalize to the opposite dataset. In general, we found that canonical brain area labels are classifiable in gene expression space within dataset and that our observed performance is not merely reflecting physical distance in the brain. However, we also show that cross-platform classification is not robust. Emerging spatial datasets from the mouse brain will allow further characterization of cross-dataset replicability ultimately providing a valuable reference set for understanding the cell biology of the brain.

Adult trkB Signaling in Parvalbumin Interneurons is Essential to Prefrontal Network Dynamics.

Inhibitory interneurons expressing parvalbumin (PV) are central to cortical network dynamics, generation of γ oscillations, and cognition. Dysfunction of PV interneurons disrupts cortical information processing and cognitive behavior. Brain-derived neurotrophic factor (BDNF)/tyrosine receptor kinase B (trkB) signaling regulates the maturation of cortical PV interneurons but is also implicated in their adult multidimensional functions. Using a novel viral strategy for cell-type-specific and spatially restricted expression of a dominant-negative trkB (trkB.DN), we show that BDNF/trkB signaling is essential to the integrity and maintenance of prefrontal PV interneurons in adult male and female mice. Reduced BDNF/trkB signaling in PV interneurons in the medial prefrontal cortex (mPFC) resulted in deficient PV inhibition and increased baseline local field potential (LFP) activity in a broad frequency band. The altered network activity was particularly pronounced during increased activation of the prefrontal network and was associated with changed dynamics of local excitatory neurons, as well as decreased modulation of the LFP, abnormalities that appeared to generalize across stimuli and brain states. In addition, our findings link reduced BDNF/trkB signaling in prefrontal PV interneurons to increased aggression. Together our investigations demonstrate that BDNF/trkB signaling in PV interneurons in the adult mPFC is essential to local network dynamics and cognitive behavior. Our data provide direct support for the suggested association between decreased trkB signaling, deficient PV inhibition, and altered prefrontal circuitry.SIGNIFICANCE STATEMENT Brain-derived neurotrophic factor (BDNF)/tyrosine receptor kinase B (trkB) signaling promotes the maturation of inhibitory parvalbumin (PV) interneurons, neurons central to local cortical dynamics, γ rhythms, and cognition. Here, we used a novel viral approach for reduced BDNF/trkB signaling in PV interneurons in the medial prefrontal cortex (mPFC) to establish the role of BDNF/trkB signaling in adult prefrontal network activities. Reduced BDNF/trkB signaling caused pronounced morphologic alterations, reduced PV inhibition, and deficient prefrontal network dynamics. The altered network activity appeared to manifest across stimuli and brain states and was associated with aberrant local field potential (LFP) activities and increased aggression. The results demonstrate that adult BDNF/trkB signaling is essential to PV inhibition and prefrontal circuit function and directly links BDNF/trkB signaling to network integrity in the adult brain.

Brain-wide genetic mapping identifies the indusium griseum as a prenatal target of pharmacologically unrelated psychostimulants.

Psychostimulant use is an ever-increasing socioeconomic burden, including a dramatic rise during pregnancy. Nevertheless, brain-wide effects of psychostimulant exposure are incompletely understood. Here, we performed Fos-CreERT2-based activity mapping, correlated for pregnant mouse dams and their fetuses with amphetamine, nicotine, and caffeine applied acutely during midgestation. While light-sheet microscopy-assisted intact tissue imaging revealed drug- and age-specific neuronal activation, the indusium griseum (IG) appeared indiscriminately affected. By using GAD67gfp/+ mice we subdivided the IG into a dorsolateral domain populated by γ-aminobutyric acidergic interneurons and a ventromedial segment containing glutamatergic neurons, many showing drug-induced activation and sequentially expressing Pou3f3/Brn1 and secretagogin (Scgn) during differentiation. We then combined Patch-seq and circuit mapping to show that the ventromedial IG is a quasi-continuum of glutamatergic neurons (IG-Vglut1+) reminiscent of dentate granule cells in both rodents and humans, whose dendrites emanate perpendicularly toward while their axons course parallel with the superior longitudinal fissure. IG-Vglut1+ neurons receive VGLUT1+ and VGLUT2+ excitatory afferents that topologically segregate along their somatodendritic axis. In turn, their efferents terminate in the olfactory bulb, thus being integral to a multisynaptic circuit that could feed information antiparallel to the olfactory-cortical pathway. In IG-Vglut1+ neurons, prenatal psychostimulant exposure delayed the onset of Scgn expression. Genetic ablation of Scgn was then found to sensitize adult mice toward methamphetamine-induced epilepsy. Overall, our study identifies brain-wide targets of the most common psychostimulants, among which Scgn+/Vglut1+ neurons of the IG link limbic and olfactory circuits.

A hypothalamus-habenula circuit controls aversion.

Encoding and predicting aversive events are critical functions of circuits that support survival and emotional well-being. Maladaptive circuit changes in emotional valence processing can underlie the pathophysiology of affective disorders. The lateral habenula (LHb) has been linked to aversion and mood regulation through modulation of the dopamine and serotonin systems. We have defined the identity and function of glutamatergic (Vglut2) control of the LHb, comparing the role of inputs originating in the globus pallidus internal segment (GPi), and lateral hypothalamic area (LHA), respectively. We found that LHb-projecting LHA neurons, and not the proposed GABA/glutamate co-releasing GPi neurons, are responsible for encoding negative value. Monosynaptic rabies tracing of the presynaptic organization revealed a predominantly limbic input onto LHA Vglut2 neurons, while sensorimotor inputs were more prominent onto GABA/glutamate co-releasing GPi neurons. We further recorded the activity of LHA Vglut2 neurons, by imaging calcium dynamics in response to appetitive versus aversive events in conditioning paradigms. LHA Vglut2 neurons formed activity clusters representing distinct reward or aversion signals, including a population that responded to mild foot shocks and predicted aversive events. We found that the LHb-projecting LHA Vglut2 neurons encode negative valence and rapidly develop a prediction signal for negative events. These findings establish the glutamatergic LHA-LHb circuit as a critical node in value processing.

Neuronal circuitry mechanism regulating adult quiescent neural stem-cell fate decision.

Adult neurogenesis arises from neural stem cells within specialized niches. Neuronal activity and experience, presumably acting on this local niche, regulate multiple stages of adult neurogenesis, from neural progenitor proliferation to new neuron maturation, synaptic integration and survival. It is unknown whether local neuronal circuitry has a direct impact on adult neural stem cells. Here we show that, in the adult mouse hippocampus, nestin-expressing radial glia-like quiescent neural stem cells (RGLs) respond tonically to the neurotransmitter γ-aminobutyric acid (GABA) by means of γ2-subunit-containing GABAA receptors. Clonal analysis of individual RGLs revealed a rapid exit from quiescence and enhanced symmetrical self-renewal after conditional deletion of γ2. RGLs are in close proximity to terminals expressing 67-kDa glutamic acid decarboxylase (GAD67) of parvalbumin-expressing (PV+) interneurons and respond tonically to GABA released from these neurons. Functionally, optogenetic control of the activity of dentate PV+ interneurons, but not that of somatostatin-expressing or vasoactive intestinal polypeptide (VIP)-expressing interneurons, can dictate the RGL choice between quiescence and activation. Furthermore, PV+ interneuron activation restores RGL quiescence after social isolation, an experience that induces RGL activation and symmetrical division. Our study identifies a niche cell­signal­receptor trio and a local circuitry mechanism that control the activation and self-renewal mode of quiescent adult neural stem cells in response to neuronal activity and experience.

Driving fast-spiking cells induces gamma rhythm and controls sensory responses.

Cortical gamma oscillations (20-80 Hz) predict increases in focused attention, and failure in gamma regulation is a hallmark of neurological and psychiatric disease. Current theory predicts that gamma oscillations are generated by synchronous activity of fast-spiking inhibitory interneurons, with the resulting rhythmic inhibition producing neural ensemble synchrony by generating a narrow window for effective excitation. We causally tested these hypotheses in barrel cortex in vivo by targeting optogenetic manipulation selectively to fast-spiking interneurons. Here we show that light-driven activation of fast-spiking interneurons at varied frequencies (8-200 Hz) selectively amplifies gamma oscillations. In contrast, pyramidal neuron activation amplifies only lower frequency oscillations, a cell-type-specific double dissociation. We found that the timing of a sensory input relative to a gamma cycle determined the amplitude and precision of evoked responses. Our data directly support the fast-spiking-gamma hypothesis and provide the first causal evidence that distinct network activity states can be induced in vivo by cell-type-specific activation.

Target selectivity of feedforward inhibition by striatal fast-spiking interneurons.

The striatal microcircuitry consists of a vast majority of projection neurons, the medium spiny neurons (MSNs), and a small yet diverse population of interneurons. To understand how activity is orchestrated within the striatum, it is essential to unravel the functional connectivity between the different neuronal types. Fast-spiking (FS) interneurons provide feedforward inhibition to both direct and indirect pathway MSNs and are important in sculpting their output to downstream basal ganglia nuclei. FS interneurons are also interconnected with each other via electrical and chemical synapses; however, whether and how they inhibit other striatal interneuron types remains unknown. In this study we combined multineuron whole-cell recordings with optogenetics to determine the target selectivity of feedforward inhibition by striatal FS interneurons. Using transgenic and viral approaches we directed expression of channelrhodopsin 2 (ChR2) to FS interneurons to study their connectivity within the mouse striatal microcircuit. Optogenetic stimulation of ChR2-expressing FS interneurons generated strong and reliable GABA(A)-dependent synaptic inputs in MSNs. In sharp contrast, simultaneously recorded neighboring cholinergic interneurons did not receive any synaptic inputs from photostimulated FS cells, and a minority of low-threshold spiking (LTS) interneurons responded weakly. We further tested the synaptic connectivity between FS and LTS interneurons using paired recordings, which showed only sparse connectivity. Our results show that striatal FS interneurons form a feedforward inhibitory circuit that is target selective, inhibiting projection neurons while avoiding cholinergic interneurons and sparsely contacting LTS interneurons, thus supporting independent modulation of MSN activity by the different types of striatal interneurons.

Cell-type-specific representation of spatial context in the rat prefrontal cortex.

The ability to represent one's own position in relation to cues, goals, or threats is crucial to successful goal-directed behavior. Using optotagging in knock-in rats expressing Cre recombinase in parvalbumin (PV) neurons (PV-Cre rats), we demonstrate cell-type-specific encoding of spatial and movement variables in the medial prefrontal cortex (mPFC) during goal-directed reward seeking. Single neurons encoded the conjunction of the animal's spatial position and the run direction, referred to as the spatial context. The spatial context was most prominently represented by the inhibitory PV interneurons. Movement toward the reward was signified by increased local field potential (LFP) oscillations in the gamma band but this LFP signature was not related to the spatial information in the neuronal firing. The results highlight how spatial information is incorporated into cognitive operations in the mPFC. The presented PV-Cre line opens the door for expanded research approaches in rats.

Striosomes Target Nigral Dopamine-Containing Neurons via Direct-D1 and Indirect-D2 Pathways Paralleling Classic Direct-Indirect Basal Ganglia Systems.

The classic output pathways of the basal ganglia are known as the direct-D1 and indirect-D2, or "Go/No-Go", pathways. Balance of the activity in these canonical direct-indirect pathways is considered a core requirement for normal movement control, and their imbalance is a major etiologic factor in movement disorders including Parkinson's disease. We present evidence for a conceptually equivalent parallel system of direct-D1 and indirect-D2 pathways that arise from striatal projection neurons (SPNs) of the striosome compartment rather than from the matrix. These striosomal direct (S-D1) and indirect (S-D2) pathways, as a pair, target dopamine-containing neurons of the substantia nigra (SNpc) instead of the motor output nuclei of the basal ganglia. The novel anatomically and functionally distinct indirect-D2 striosomal pathway targets dopaminergic SNpc cells indirectly via a core region of the external pallidum (GPe). We demonstrate that these S-D1 and S-D2 pathways oppositely modulate striatal dopamine release in freely behaving mice under open-field conditions and oppositely modulate locomotor and other movements. These S-D1 and S-D2 pathways further exhibit different, time-dependent responses during performance of a probabilistic decision-making maze task and respond differently to rewarding and aversive stimuli. These contrasts depend on mediolateral and anteroposterior striatal locations of the SPNs as are the classic direct and indirect pathways. The effects of S-D1 and S-D2 stimulation on striatal dopamine release and voluntary locomotion are nearly opposite. The parallelism of the direct-indirect circuit design motifs of the striosomal S-D and S-D2 circuits and canonical matrix M-D1 and M-D2, and their contrasting behavioral effects, call for a major reformulation of the classic direct-indirect pathway model of basal ganglia function. Given that some striosomes receive limbic and association cortical inputs, the S-D1 and S-D2 circuits likely influence motivation for action and behavioral learning, complementing and possibly reorienting the motoric activities of the canonical matrix pathways. At a fundamental level, these findings suggest a unifying framework for aligning two sets of circuits that share the organizational motif of opponent D1 and D2 regulation, but that have different outputs and can even have opposite polarities in their targets and effects, albeit conditioned by striatal topography. Our findings further delineate a potentially therapeutically important set of pathways influencing dopamine, including a D2 receptor-linked S-D2 pathway likely unknowingly targeted by administration of many therapeutic drugs including those for Parkinson's disease. The novel parallel pathway model that we propose here could help to account for the normally integrated modulatory influence of the basal ganglia on motivation for actions as well as the actions themselves.

Spatially resolved multiomics on the neuronal effects induced by spaceflight in mice.

Impairment of the central nervous system (CNS) poses a significant health risk for astronauts during long-duration space missions. In this study, we employed an innovative approach by integrating single-cell multiomics (transcriptomics and chromatin accessibility) with spatial transcriptomics to elucidate the impact of spaceflight on the mouse brain in female mice. Our comparative analysis between ground control and spaceflight-exposed animals revealed significant alterations in essential brain processes including neurogenesis, synaptogenesis and synaptic transmission, particularly affecting the cortex, hippocampus, striatum and neuroendocrine structures. Additionally, we observed astrocyte activation and signs of immune dysfunction. At the pathway level, some spaceflight-induced changes in the brain exhibit similarities with neurodegenerative disorders, marked by oxidative stress and protein misfolding. Our integrated spatial multiomics approach serves as a stepping stone towards understanding spaceflight-induced CNS impairments at the level of individual brain regions and cell types, and provides a basis for comparison in future spaceflight studies. For broader scientific impact, all datasets from this study are available through an interactive data portal, as well as the National Aeronautics and Space Administration (NASA) Open Science Data Repository (OSDR).

Efficient reprogramming of adult neural stem cells to monocytes by ectopic expression of a single gene.

Neural stem cells have a broad differentiation repertoire during embryonic development and can be reprogrammed to pluripotency comparatively easily. We report that adult neural stem cells can be reprogrammed at very high efficiency to monocytes, a differentiated fate of an unrelated somatic lineage, by ectopic expression of the Ets transcription factor PU.1. The reprogrammed cells display a marker profile and functional characteristics of monocytes and integrate into tissues after transplantation. The failure to reprogram lineage-committed neural cells to monocytes with PU.1 suggests that neural stem cells are uniquely amenable to reprogramming.

A Selective Projection from the Subthalamic Nucleus to Parvalbumin-Expressing Interneurons of the Striatum.

The striatum and subthalamic nucleus (STN) are considered to be the primary input nuclei of the basal ganglia. Projection neurons of both striatum and STN can extensively interact with other basal ganglia nuclei, and there is growing anatomic evidence of direct axonal connections from the STN to striatum. There remains, however, a pressing need to elucidate the organization and impact of these subthalamostriatal projections in the context of the diverse cell types constituting the striatum. To address this, we conducted monosynaptic retrograde tracing from genetically-defined populations of dorsal striatal neurons in adult male and female mice, quantifying the connectivity from STN neurons to spiny projection neurons, GABAergic interneurons, and cholinergic interneurons. In parallel, we used a combination of ex vivo electrophysiology and optogenetics to characterize the responses of a complementary range of dorsal striatal neuron types to activation of STN axons. Our tracing studies showed that the connectivity from STN neurons to striatal parvalbumin-expressing interneurons is significantly higher (∼4- to 8-fold) than that from STN to any of the four other striatal cell types examined. In agreement, our recording experiments showed that parvalbumin-expressing interneurons, but not the other cell types tested, commonly exhibited robust monosynaptic excitatory responses to subthalamostriatal inputs. Taken together, our data collectively demonstrate that the subthalamostriatal projection is highly selective for target cell type. We conclude that glutamatergic STN neurons are positioned to directly and powerfully influence striatal activity dynamics by virtue of their enriched innervation of GABAergic parvalbumin-expressing interneurons.

Esr1+ hypothalamic-habenula neurons shape aversive states.

Excitatory projections from the lateral hypothalamic area (LHA) to the lateral habenula (LHb) drive aversive responses. We used patch-sequencing (Patch-seq) guided multimodal classification to define the structural and functional heterogeneity of the LHA-LHb pathway. Our classification identified six glutamatergic neuron types with unique electrophysiological properties, molecular profiles and projection patterns. We found that genetically defined LHA-LHb neurons signal distinct aspects of emotional or naturalistic behaviors, such as estrogen receptor 1-expressing (Esr1+) LHA-LHb neurons induce aversion, whereas neuropeptide Y-expressing (Npy+) LHA-LHb neurons control rearing behavior. Repeated optogenetic drive of Esr1+ LHA-LHb neurons induces a behaviorally persistent aversive state, and large-scale recordings showed a region-specific neural representation of the aversive signals in the prelimbic region of the prefrontal cortex. We further found that exposure to unpredictable mild shocks induced a sex-specific sensitivity to develop a stress state in female mice, which was associated with a specific shift in the intrinsic properties of bursting-type Esr1+ LHA-LHb neurons. In summary, we describe the diversity of LHA-LHb neuron types and provide evidence for the role of Esr1+ neurons in aversion and sexually dimorphic stress sensitivity.

Sensorimotor Integration Supporting Perception Requires Syngap1 Expression in Cortex.

Perception, a cognitive construct, emerges through sensorimotor integration (SMI). The molecular and cellular mechanisms that shape SMI within circuits that promote cognition are poorly understood. Here, we demonstrate that expression of the autism/intellectual disability gene, Syngap1, in mouse cortical excitatory neurons promotes touch sensitivity required to elicit perceptual behaviors. Cortical Syngap1 expression enabled touch-induced feedback signals within sensorimotor loops by assembling circuits that support tactile sensitivity. These circuits also encoded correlates of attention that promoted self-generated whisker movements underlying purposeful and sustained object exploration. As Syngap1 deficient animals explored objects with whiskers, relatively weak touch signals were integrated with relatively strong motor signals. This produced a signal-to-noise deficit consistent with impaired tactile sensitivity, reduced tactile exploration, and weak tactile learning. Thus, Syngap1 expression in cortex promotes tactile perception by assembling circuits that integrate touch and whisker motor signals. Deficient Syngap1 expression likely contributes to cognitive impairment through abnormal top-down SMI.

Centrosome motility is essential for initial axon formation in the neocortex.

The mechanisms underlying the normal development of neuronal morphology remain a fundamental question in neurobiology. Studies in cultured neurons have suggested that the position of the centrosome and the Golgi may predict the site of axon outgrowth. During neuronal migration in the developing cortex, however, the centrosome and Golgi are oriented toward the cortical plate at a time when axons grow toward the ventricular zone. In the current work, we use in situ live imaging to demonstrate that the centrosome and the accompanying polarized cytoplasm exhibit apical translocation in newborn cortical neurons preceding initial axon outgrowth. Disruption of centrosomal activity or downregulation of the centriolar satellite protein PCM-1 affects axon formation. We further show that downregulation of the centrosomal protein Cep120 impairs microtubule organization, resulting in increased centrosome motility. Decreased centrosome motility resulting from microtubule stabilization causes an aberrant centrosomal localization, leading to misplaced axonal outgrowth. Our results reveal the dynamic nature of the centrosome in developing cortical neurons, and implicate centrosome translocation and microtubule organization during the multipolar stage as important determinants of axon formation.

Genetic manipulation of adult mouse neurogenic niches by in vivo electroporation.

Targeted ectopic expression of genes in the adult brain is an invaluable approach for studying many biological processes. This can be accomplished by generating transgenic mice or by virally mediated gene transfer, but these methods are costly and labor intensive. We devised a rapid strategy that allows localized in vivo transfection of plasmid DNA within the adult neurogenic niches without detectable brain damage. Injection of plasmid DNA into the ventricular system or directly into the hippocampus of adult mice, followed by application of electrical current via external electrodes, resulted in transfection of neural stem or progenitor cells and mature neurons. We showed that this strategy can be used for both fate mapping and gain- or loss-of-function experiments. Using this approach, we identified an essential role for cadherins in maintaining the integrity of the lateral ventricle wall. Thus, in vivo electroporation provides a new approach to study the adult brain.

Spinal cord injury reveals multilineage differentiation of ependymal cells.

Spinal cord injury often results in permanent functional impairment. Neural stem cells present in the adult spinal cord can be expanded in vitro and improve recovery when transplanted to the injured spinal cord, demonstrating the presence of cells that can promote regeneration but that normally fail to do so efficiently. Using genetic fate mapping, we show that close to all in vitro neural stem cell potential in the adult spinal cord resides within the population of ependymal cells lining the central canal. These cells are recruited by spinal cord injury and produce not only scar-forming glial cells, but also, to a lesser degree, oligodendrocytes. Modulating the fate of ependymal progeny after spinal cord injury may offer an alternative to cell transplantation for cell replacement therapies in spinal cord injury.

A hypoxic niche regulates glioblastoma stem cells through hypoxia inducible factor 2 alpha.

Glioma growth and progression depend on a specialized subpopulation of tumour cells, termed tumour stem cells. Thus, tumour stem cells represent a critical therapeutic target, but the molecular mechanisms that regulate them are poorly understood. Hypoxia plays a key role in tumour progression and in this study we provide evidence that the hypoxic tumour microenvironment also controls tumour stem cells. We define a detailed molecular signature of tumour stem cell genes, which are overexpressed by tumour cells in vascular and perinecrotic/hypoxic niches. Mechanistically, we show that hypoxia plays a key role in the regulation of the tumour stem cell phenotype through hypoxia-inducible factor 2alpha and subsequent induction of specific tumour stem cell signature genes, including mastermind-like protein 3 (Notch pathway), nuclear factor of activated T cells 2 (calcineurin pathway) and aspartate beta-hydroxylase domain-containing protein 2. Notably, a number of these genes belong to pathways regulating the stem cell phenotype. Consistently, tumour stem cell signature genes are overexpressed in newly formed gliomas and are associated with worse clinical prognosis. We propose that tumour stem cells are maintained within a hypoxic niche, providing a functional link between the well-established role of hypoxia in stem cell and tumour biology. The identification of molecular regulators of tumour stem cells in the hypoxic niche points to specific signalling mechanisms that may be used to target the glioblastoma stem cell population.