A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain

The 10xv3 and 10xv2 cells were first clustered separately and then integrated, resulting in an initial joint transcriptomic cell-type taxonomy with 5,283 clusters.

“The 10xv3 and 10xv2 cells were first clustered separately and then integrated with methods we developed previously23, resulting in an initial joint transcriptomic cell-type taxonomy with 5,283 clusters (Extended Data Fig. 1a).”

• cites [23] Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241.e3226 (2021). sourceSupported

The Subpallium-GABA neighbourhood (subclasses 39-90, total 1,051 clusters) was divided into 7 classes that are likely related to their distinct developmental origins.

“On the basis of the molecular signature and regional specificity of each subclass, the Subpallium-GABA neighbourhood (subclasses 39–90, total 1,051 clusters) was divided into 7 classes that are likely related to their distinct developmental origins29,30”

• cites [29] Flames, N. et al. Delineation of multiple subpallial progenitor domains by the combinatorial expression of transcriptional codes. J. Neurosci. 27, 9682–9695 (2007). sourceSupported
• cites [30] Turrero Garcia, M. & Harwell, C. C. Radial glia in the ventral telencephalon. FEBS Lett. 591, 3942–3959 (2017). sourcePartly supported

  The Introduction says subpallial cell diversity varies with germinal zone and developmental timing, supporting the general link between cell types and developmental origins. The source does not report a Subpallium-GABA neighbourhood, subclasses 39–90, 1,051 clusters, or a division into seven classes. The indexed passages have section labels but no page numbers. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

The NP-CT-L6b-like subclasses emerge as a group highly distinct from the IT-ET-like subclasses, forming two distinct classes, IT-ET Glut and NP-CT-L6b Glut.

“We also observed that the NP–CT–L6b-like (NP, near-projecting; CT, corticothalamic; L6b, layer 6b) subclasses emerge as a group highly distinct from the IT–ET-like (IT, intratelencephalic; ET, extratelencephalic) subclasses13,23,26,27, forming two distinct classes, IT–ET Glut and NP–CT–L6b Glut.”

• cites [13] Brain Initiative Cell Census Network. A multimodal cell census and atlas of the mammalian primary motor cortex. Nature 598, 86–102 (2021). sourcePartly supported

  The Methods section on L5 ET nomenclature describes a hierarchical taxonomy with distinct IT, L5 ET, L6 CT, NP and L6b glutamatergic subclasses. It does not report that these subclasses form the two named classes, IT–ET Glut and NP–CT–L6b Glut; that grouping is presented as the citing authors’ observation. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [23] Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241.e3226 (2021). sourceSupported
• cites [26] Tasic, B. et al. Shared and distinct transcriptomic cell types across neocortical areas. Nature 563, 72–78 (2018). sourcePartly supported

  In “Glutamatergic taxonomy by scRNA-seq and projections” (page unavailable), the paper identifies IT, near-projecting and CT as glutamatergic subclasses and describes L6b types. It does not report that NP–CT–L6b and IT–ET form two distinct classes; “Overall cell type taxonomy” instead uses “glutamatergic” as the class and “L6b” as a subclass. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [27] Yao, Z. et al. A transcriptomic and epigenomic cell atlas of the mouse primary motor cortex. Nature 598, 103–110 (2021). sourceSupported

The parallel relationships of the different subclasses of glutamatergic neurons between isocortex and HPF that the authors had reported previously extend to other pallium structures-that is, OLF and CTXsp.

“We found that the parallel relationships of the different subclasses of glutamatergic neurons between isocortex and HPF that we had reported previously23 extend to other pallium structures—that is, OLF and CTXsp.”

• cites [23] Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241.e3226 (2021). sourceSupported

The Glut-GABA clusters include four clusters in the isocortex and hippocampus and three clusters in globus pallidus, internal segment, which probably correspond to previously well-characterized glutamate-GABA co-releasing neuronal types in these regions.

“They include four clusters in the isocortex and hippocampus and three clusters in globus pallidus, internal segment (GPi), which probably correspond to previously well-characterized glutamate–GABA co-releasing neuronal types in these regions41,42.”

• cites [41] Kim, S., Wallace, M. L., El-Rifai, M., Knudsen, A. R. & Sabatini, B. L. Co-packaging of opposing neurotransmitters in individual synaptic vesicles in the central nervous system. Neuron 110, 1371–1384.e1377 (2022). sourcePartly supported

  Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured
• cites [42] Pelkey, K. A. et al. Paradoxical network excitation by glutamate release from VGluT3+ GABAergic interneurons. eLife 9, e51996 (2020). sourcePartly supported

  Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured

Both RT and ZI neurons may have originated from the prethalamus or the zona limitans intrathalamica.

“Both RT and ZI neurons may have originated from the prethalamus or the zona limitans intrathalamica (ZLI)34–38.”

• cites [34] Delaunay, D. et al. Genetic tracing of subpopulation neurons in the prethalamus of mice (Mus musculus). J. Comp. Neurol. 512, 74–83 (2009). sourceNot checked — no full text of the source
• cites [35] Govek, K. W. et al. Developmental trajectories of thalamic progenitors revealed by single-cell transcriptome profiling and Shh perturbation. Cell Rep. 41, 111768 (2022). sourceCould not verify

  The Results section on molecular identities at E18.5 identifies zona incerta (ZI) and reticular complex (RT) cell populations, while the Introduction describes the ZLI as separating the thalamus from the prethalamus. Neither passage, nor searches for RT and ZI origins, says that both populations may originate in the prethalamus or ZLI. Page numbers were not available in our copy. (LLM, conf=0.72)

  Experimental — accuracy not yet measured
• cites [36] Inamura, N., Ono, K., Takebayashi, H., Zalc, B. & Ikenaka, K. Olig2 lineage cells generate GABAergic neurons in the prethalamic nuclei, including the zona incerta, ventral lateral geniculate nucleus and reticular thalamic nucleus. Dev. Neurosci. 33, 118–129 (2011). sourceNot checked — no full text of the source
• cites [37] Puelles, L. et al. LacZ-reporter mapping of Dlx5/6 expression and genoarchitectural analysis of the postnatal mouse prethalamus. J. Comp. Neurol. 529, 367–420 (2021). sourcePartly supported

  The Results section, “Regional and areal constitution of the prethalamus” (page unavailable), identifies RT and the rostral and caudal ZI as prethalamic derivatives. The Results section, “Adult prethalamic cell populations at P40 and P140” (page unavailable), instead identifies the zona-limitans-derived retroincertal nucleus as distinct from the caudal ZI; it does not establish a zona limitans origin for both RT and ZI neurons. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [38] Shimogori, T. et al. A genomic atlas of mouse hypothalamic development. Nat. Neurosci. 13, 767–775 (2010). sourceCould not verify

  The Results section on the anterior diencephalon describes markers of prethalamic neuroepithelium, and the section on basal hypothalamic Shh discusses the ZLI. Neither passage addresses whether RT or ZI neurons originate from either region. Searches for reticular nucleus and zona incerta did not locate a passage establishing that origin. (LLM, conf=0.72)

  Experimental — accuracy not yet measured

The HY-EA-Glut-GABA neighbourhood (including subclasses 66 and 73-144, total 1,404 clusters) contains a set of closely related neuronal subclasses from the entire hypothalamus, as well as the striatum-like amygdalar nuclei and caudal PAL regions of CNU that are also known as the extended amygdala.

“The HY–EA-Glut–GABA neighbourhood (including subclasses 66 and 73–144, total 1,404 clusters) contains a set of closely related neuronal subclasses from the entire hypothalamus32,33, as well as the striatum-like amygdalar nuclei (sAMY) and caudal PAL regions of CNU that are also known as the extended amygdala”

• cites [32] Saper, C. B. & Lowell, B. B. The hypothalamus. Curr. Biol. 24, R1111–R1116 (2014). sourceNot checked — no full text of the source
• cites [33] Steuernagel, L. et al. HypoMap-a unified single-cell gene expression atlas of the murine hypothalamus. Nat. Metab. 4, 1402–1419 (2022). sourcePartly supported

  Split into 2 atoms — 1 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 1 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured

Cell-type specificity of neuropeptide gene expression varies widely, as measured by Tau score.

“We measured the cell-type specificity of expression of these genes using the Tau score60 and found a wide range of variation”

• cites [60] Yanai, I. et al. Genome-wide midrange transcription profiles reveal expression level relationships in human tissue specification. Bioinformatics 21, 650–659 (2005). sourceSupported

The HY Gnrh1 Glut class is the hypothalamic Gnrh1 neuronal type developmentally originated from the embryonic olfactory epithelium.

“The HY Gnrh1 Glut class is the hypothalamic Gnrh1 neuronal type developmentally originated from the embryonic olfactory epithelium39.”

• cites [39] Duittoz, A. H. et al. Development of the gonadotropin-releasing hormone system. J. Neuroendocrinol. 34, e13087 (2022). sourcePartly supported

  The Introduction and “Olfactory placode and GnRH ontogenesis” section report that GnRH neurons arise in the developing olfactory region and migrate into the forebrain. The paper places their origin in the olfactory placode, likely near the vomeronasal organ primordium, rather than specifically in the olfactory epithelium. It does not identify an “HY Gnrh1 Glut” class or establish that class’s developmental origin; page numbers were not available in our copy. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Of all non-neuronal cell types, the Astro-Epen class exhibits the most diverse spatial patterns.

“Of all the non-neuronal cell types, the Astro–Epen class exhibits the most diverse spatial patterns68,69.”

• cites [68] Endo, F. et al. Molecular basis of astrocyte diversity and morphology across the CNS in health and disease. Science 378, eadc9020 (2022). sourcePartly supported

  The Results section, “Region-specific astrocyte genes and pathways” (page unavailable), documents differences among astrocytes across CNS regions. It does not identify an Astro–Epen class or compare its spatial patterns with those of all other non-neuronal cell types, so the claim that it is the most diverse is unsupported. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [69] John Lin, C. C. et al. Identification of diverse astrocyte populations and their malignant analogs. Nat. Neurosci. 20, 396–405 (2017). sourcePartly supported

  The Results (page unavailable) describe astrocyte subpopulations across several brain regions and differing locations within the cortex, supporting spatial diversity among astrocytes. The paper does not identify an “Astro–Epen” class or compare its spatial patterns with those of all other non-neuronal cell types, so it does not support the superlative claim. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Astro-TE clusters 5229 and 5230 and clusters in the Astro-OLF subclass match the path of the rostral migratory stream (RMS).

“Astro-TE clusters 5229 and 5230 and clusters in the Astro-OLF subclass match the path of the rostral migratory stream70–72 (RMS; see below).”

• cites [70] Garcia-Marques, J., De Carlos, J. A., Greer, C. A. & Lopez-Mascaraque, L. Different astroglia permissivity controls the migration of olfactory bulb interneuron precursors. Glia 58, 218–230 (2010). sourceSupported
• cites [71] Kaneko, N. et al. New neurons clear the path of astrocytic processes for their rapid migration in the adult brain. Neuron 67, 213–223 (2010). sourcePartly supported

  The Introduction (page unavailable) describes the RMS as the route from the subventricular zone to the olfactory bulb, and the Results describe astrocytes along that pathway. The paper does not identify Astro-TE clusters 5229 or 5230, Astro-OLF clusters, or show that those clusters match the RMS path. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [72] Lim, D. A. & Alvarez-Buylla, A. The adult ventricular–subventricular zone (V-SVZ) and olfactory bulb (OB) neurogenesis. Cold Spring Harb. Perspect. Biol. 8, a018820 (2016). sourceSupported

This corroborates findings indicating the existence of neural stem cells in the OVLT, SFO, ME and AP.

“which corroborates findings indicating the existence of neural stem cells in the OVLT, SFO, ME and AP78–81.”

• cites [78] Bennett, L., Yang, M., Enikolopov, G. & Iacovitti, L. Circumventricular organs: a novel site of neural stem cells in the adult brain. Mol. Cell. Neurosci. 41, 337–347 (2009). sourceSupported
• cites [79] Furube, E., Morita, M. & Miyata, S. Characterization of neural stem cells and their progeny in the sensory circumventricular organs of adult mouse. Cell Tissue Res. 362, 347–365 (2015). sourceNot checked — no full text of the source
• cites [80] Lee, D. A. et al. Tanycytes of the hypothalamic median eminence form a diet-responsive neurogenic niche. Nat. Neurosci. 15, 700–702 (2012). sourcePartly supported

  The paper’s main text (indexed under Abstract; page unavailable) identifies a neural progenitor pool in the median eminence (ME) and reports that its β2-tanycytes generate newborn neurons. It does not provide findings for neural stem cells in the OVLT, SFO, or AP, so it supports only the ME portion of the cited list. (LLM, conf=0.88)

  Experimental — accuracy not yet measured
• cites [81] Robins, S. C. et al. α-Tanycytes of the adult hypothalamic third ventricle include distinct populations of FGF-responsive neural progenitors. Nat. Commun. 4, 2049 (2013). sourcePartly supported

  The Discussion (page unavailable) reports evidence of neural stem-cell potential in median eminence (ME) β2-tanycytes in postnatal or juvenile animals, while distinguishing that finding from this paper’s results in adults. The cited paper does not address neural stem cells in the OVLT, SFO or AP, so it supports only the ME portion of the attributed claim. (LLM, conf=0.87)

  Experimental — accuracy not yet measured

Interactions between various VLMC and ependymal cell clusters, together with ABCs, likely regulate the movement of molecules and cells across the barriers between brain and blood or CSF.

“Interactions between various VLMC and ependymal cell clusters, together with ABCs, likely regulate the movement of molecules and cells across the barriers between brain and blood or CSF82.”

• cites [82] Derk, J., Jones, H. E., Como, C., Pawlikowski, B. & Siegenthaler, J. A. Living on the edge of the CNS: meninges cell diversity in health and disease. Front. Cellular Neurosci. 15, 703944 (2021). sourcePartly supported

  The Meningeal Cell Types section and Figure 2 describe arachnoid barrier cells regulating the movement of molecules and cells between the dura and CSF-filled subarachnoid space. The Conclusion mentions vascular leptomeningeal cell clusters, but the cited passages do not establish that interactions among VLMC and ependymal cell clusters, together with arachnoid barrier cells, regulate that movement. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

VLMC types show highly specific spatial and colocalization patterns.

“VLMC types66,82 also show highly specific spatial and colocalization patterns.”

• cites [66] Zeisel, A. et al. Molecular architecture of the mouse nervous system. Cell 174, 999–1014.e1022 (2018). sourcePartly supported

  The Results section, “Vascular Cells and a Family of Broadly Distributed Mesothelial Fibroblasts” (page unavailable), identifies distinct VLMC-like types and reports specific locations, including enteric mesothelial fibroblasts found exclusively in the enteric nervous system. It does not establish highly specific colocalization patterns for VLMC types, so that part of the claim is unsupported. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites [82] Derk, J., Jones, H. E., Como, C., Pawlikowski, B. & Siegenthaler, J. A. Living on the edge of the CNS: meninges cell diversity in health and disease. Front. Cellular Neurosci. 15, 703944 (2021). sourcePartly supported

  The Fibroblasts section (page unavailable) describes meningeal fibroblast populations in distinct layers and perivascular fibroblasts beside particular blood vessels, supporting spatial specificity. It does not establish highly specific colocalization patterns for VLMC types; the Conclusion (page unavailable) says there is a lack of consensus on these populations’ spatial heterogeneity. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

The migrating neurons in the RMS are separated from the parenchyma by astrocytes that form tunnels through which the cells migrate.

“The migrating neurons in the RMS are separated from the parenchyma by astrocytes that form tunnels through which the cells migrate71,86.”

• cites [71] Kaneko, N. et al. New neurons clear the path of astrocytic processes for their rapid migration in the adult brain. Neuron 67, 213–223 (2010). sourceSupported
• cites [86] Lois, C., Garcia-Verdugo, J. M. & Alvarez-Buylla, A. Chain migration of neuronal precursors. Science 271, 978–981 (1996). sourceNot checked — no full text of the source

Prediction of GABAergic subclasses and their spatial location in both directions appears to have more confusions, especially in the Subpallium-GABA neighbourhood, consistent with the more widespread distribution across multiple cortical areas of many GABAergic cell types.

“Notably, the prediction of GABAergic subclasses and their spatial location in both directions appears to have more confusions, especially in the Subpallium-GABA neighbourhood, consistent with the more widespread distribution across multiple cortical areas of many GABAergic cell types23,26.”

• cites [23] Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241.e3226 (2021). sourceSupported
• cites [26] Tasic, B. et al. Shared and distinct transcriptomic cell types across neocortical areas. Nature 563, 72–78 (2018). sourceSupported

Cortical GABAergic cell types are shared across all cortical areas.

“One exception is cortical GABAergic cell types, which are shared across all cortical areas as we previously reported23,26”

• cites [23] Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241.e3226 (2021). sourceSupported
• cites [26] Tasic, B. et al. Shared and distinct transcriptomic cell types across neocortical areas. Nature 563, 72–78 (2018). sourcePartly supported

  The Abstract reports that nearly all GABAergic neuron types were shared between the two mouse neocortical areas studied, ALM and VISp. The Overall cell type taxonomy section identifies two VISp-specific exceptions. The claim overstates both the extent of sharing and the number of cortical areas examined. (LLM, conf=0.91)

  Experimental — accuracy not yet measured

By performing all pairwise cluster comparisons in the initial transcriptomic taxonomy, the authors derived 8,460 differentially expressed genes differentiating all pairs of clusters.

“By performing all pairwise cluster comparisons in this initial transcriptomic taxonomy, we derived 8,460 differentially expressed genes (DEGs) (Supplementary Table 5) differentiating all pairs of clusters.”

The anatomical annotations at subclass level are largely consistent between the X.Z. laboratory and the AIBS MERFISH datasets.

“As the anatomical annotations at subclass level are largely consistent between the X.Z. laboratory and the AIBS MERFISH datasets, the AIBS MERFISH dataset is used to illustrate our results and findings in the subsequent sections of this manuscript.”

The paper identified a high degree of correspondence between transcriptomic identity and spatial specificity for each cell type.

“We systematically analysed the neuronal and non-neuronal cell types across the brain and identified a high degree of correspondence between transcriptomic identity and spatial specificity for each cell type.”

• Also stated in: § Neuronal cell types across the mouse brain · boosted

  “The results shown in Fig. 2 reveal a marked correspondence between transcriptomic specificity and relatedness and spatial specificity and relatedness among the different neuronal subclasses.”

Transcription factors are major determinants of cell-type classification.

“we found that transcription factors are major determinants of cell-type classification”

Glutamatergic neurons from all pallium structures, including isocortex, HPF, OLF and CTXsp, form a distinct Pallium-Glut neighbourhood that includes subclasses 1-38 and a total of 517 clusters.

“Glutamatergic neurons from all pallium structures, including isocortex, HPF, OLF and CTXsp, form a distinct Pallium-Glut neighbourhood that includes subclasses 1–38 and a total of 517 clusters (Figs. 1a and 2a,b, Extended Data Table 1, Extended Data Fig. 6 and Supplementary Table 7).”

The atlas is hierarchically organized into four nested levels of classification: 34 classes, 338 subclasses, 1,201 supertypes and 5,322 clusters.

“The atlas is hierarchically organized into 4 nested levels of classification: 34 classes, 338 subclasses, 1,201 supertypes and 5,322 clusters.”

• Also stated in: § Creation of the mouse brain cell-type atlas · plain

  “we present a high-resolution transcriptomic and spatial cell-type atlas for the whole mouse brain with four nested levels of classification: 34 classes, 338 subclasses, 1,201 supertypes and 5,322 clusters or types”

The paper reports a comprehensive and high-resolution transcriptomic and spatial cell-type atlas for the whole adult mouse brain.

“Here we report a comprehensive and high-resolution transcriptomic and spatial cell-type atlas for the whole adult mouse brain.”

• AssumptionRests on an assumption: The mouse is a suitable model for a comprehensive definition of mammalian brain composition and architecture.

The paper identified a combinatorial transcription factor code that defines cell types across all parts of the brain.

“identified a combinatorial transcription factor code that defines cell types across all parts of the brain”

A set of developmental subpallium-derived GABAergic neuronal subclasses, including all GABAergic neurons found in pallium structures and those in the subpallial CNU, form a Subpallium-GABA neighbourhood.

“A set of developmental subpallium-derived GABAergic (γ-aminobutyric acid-producing) neuronal subclasses, including all GABAergic neurons found in pallium structures and those in the subpallial CNU, including dorsal STR (STRd) and ventral STR (STRv), lateral septal complex (LSX), and dorsal PAL (PALd), ventral PAL (PALv) and medial PAL (PALm), form a Subpallium-GABA neighbourhood (Figs. 1a and 2c,d”

Each neuronal subclass exhibits layer and/or region specificity.

“Here, each neuronal subclass exhibits layer and/or region specificity (Fig. 2a,b).”

The paper presents an online platform, Allen Brain Cell Atlas, to visualize the mouse whole-brain cell-type atlas along with the scRNA-seq and MERFISH datasets.

“We present an online platform, Allen Brain Cell Atlas, to visualize the mouse whole-brain cell-type atlas along with the single-cell RNA-sequencing and MERFISH datasets.”

The results reveal unique features of cell-type organization in different brain regions, in particular a dichotomy between the dorsal and ventral parts of the brain.

“The results reveal unique features of cell-type organization in different brain regions—in particular, a dichotomy between the dorsal and ventral parts of the brain.”

The study uncovered extraordinary diversity and heterogeneity in neurotransmitter and neuropeptide expression and co-expression patterns in different cell types.

“Our study also uncovered extraordinary diversity and heterogeneity in neurotransmitter and neuropeptide expression and co-expression patterns in different cell types.”

The DG-IMN Glut class contains both the dentate gyrus granule cells and the immature neurons found in DG and the piriform cortex that are involved in adult neurogenesis.

“Finally, this neighbourhood includes the DG–IMN Glut class which contains both the dentate gyrus (DG) granule cells and the immature neurons found in DG and the piriform cortex (PIR) that are involved in adult neurogenesis (see below).”

Cell types in midbrain and hindbrain that were depleted in the scRNA-seq dataset were supplemented with 10x Multiome snRNA-seq data, with a total of 1,687 10xMulti nuclei across 33 clusters added to the taxonomy.

“We further refined the clustering (Methods) and identified cell types in midbrain and hindbrain that were depleted in our scRNA-seq dataset. These cell types were supplemented with 10x Multiome snRNA-seq data (10xMulti), with a total of 1,687 10xMulti nuclei across 33 clusters added to the taxonomy (Extended Data Fig. 1a and Methods).”

Some neuropeptides are widely expressed in many cell types or clusters and at high levels (e.g., Cck, Pnoc, Adcyap1, Penk, Sst and Tac1).

“Some neuropeptides are widely expressed in many cell types or clusters and at high levels (for example, Cck, Pnoc, Adcyap1, Penk, Sst and Tac1)”

The MB-HB-CB-GABA neighbourhood includes 4 classes (MB GABA, P GABA, MY GABA and CB GABA), 75 subclasses and 1,040 clusters.

“This neighbourhood includes 4 classes (MB GABA, P GABA, MY GABA and CB GABA), 75 subclasses and 1,040 clusters”

Dopaminergic neuronal types were found in other brain regions, including 22 clusters in RAmb and PAG.

“we also found dopaminergic neuronal types in other brain regions, including many clusters (22) in RAmb and periaqueductal gray (PAG)”

Some neuropeptides are highly expressed specifically in only one or few clusters (e.g., Avp, Agrp, Pomc, Pmch, Oxt, Rln3, Npw, Nps, Ucn, Hcrt, Gnrh1, Gcg and Pyy).

“others are highly expressed specifically in only one or few clusters (for example, Avp, Agrp, Pomc, Pmch, Oxt, Rln3, Npw, Nps, Ucn, Hcrt, Gnrh1, Gcg and Pyy)”

All histaminergic neurons co-release GABA.

“and all co-release GABA54.”

Neuronal types in the most anterior part of the hypothalamus, the POA, are highly similar to neuronal types in sAMY and PAL.

“Neuronal types in the most anterior part of hypothalamus, the POA, are highly similar to neuronal types in sAMY and PAL.”

Many GABAergic neuronal clusters in midbrain and hindbrain co-release glycine.

“Many GABAergic neuronal clusters in midbrain and hindbrain co-release glycine.”

No clusters co-express the key synthesis genes Th and Tph2.

“we found no clusters that could co-release serotonin and dopamine, as no clusters co-express the key synthesis genes Th and Tph2.”

Histaminergic neurons are found exclusively in the tuberomammillary nucleus, dorsal part (TMd) and ventral part (TMv), of hypothalamus (5 clusters in subclass 92).

“Histaminergic neurons are found exclusively in the tuberomammillary nucleus, dorsal part (TMd) and ventral part (TMv), of hypothalamus (5 clusters in subclass 92)”

Except for the three Glut-GABA clusters that form an exclusive subclass in GPi (subclass 112), the other Glut-GABA clusters are present in subclasses that also contain closely related single-neurotransmitter (glutamate or GABA) clusters.

“Notably, except for the three Glut–GABA clusters that form an exclusive subclass in GPi (subclass 112), the other Glut–GABA clusters are present in subclasses that also contain closely related single-neurotransmitter (glutamate or GABA) clusters (Extended Data Fig. 9 and Supplementary Table 7).”

Dopaminergic neuronal types were found in DMH (3 clusters).

“as well as 3 clusters in dorsomedial nucleus of the hypothalamus (DMH).”

These non-neuronal cell types can be distinguished by highly specific marker genes at all levels of hierarchy.

“which can be distinguished by highly specific marker genes at all levels of hierarchy”

There are numerous co-expression combinations of different neuropeptides in many clusters, with high degrees of variations within subclasses.

“there are numerous co-expression combinations of different neuropeptides in many clusters, with high degrees of variations within subclasses”

Subclass 215 displays the most heterogeneous neurotransmitter content.

“displays the most heterogeneous neurotransmitter content.”

The fourth neuronal neighbourhood, TH-EPI-Glut (subclasses 145-154, total 148 clusters), contains all glutamatergic neuronal subclasses located in the thalamus, as well as the medial habenula and lateral habenula, which collectively compose the epithalamus.

“The fourth neuronal neighbourhood, TH–EPI-Glut (subclasses 145–154, total 148 clusters), contains all glutamatergic neuronal subclasses located in the thalamus, as well as the medial habenula (MH) and lateral habenula (LH), which collectively compose the epithalamus (EPI)”

The sixth neuronal neighbourhood, MB-HB-CB-GABA, contains all GABAergic subclasses located in midbrain, hindbrain and cerebellum.

“The sixth and final neuronal neighbourhood, MB–HB–CB-GABA, contains all GABAergic subclasses located in midbrain, hindbrain and cerebellum”

The study defined a comprehensive set of cell types with unique and differing neurotransmitter content that can be identified through combinations of marker genes.

“our study defined a comprehensive set of cell types with unique and differing neurotransmitter content that can be identified through combinations of marker genes.”

About 79% of all clusters express at least one neuropeptide gene.

“About 79% of all clusters express at least one neuropeptide gene”

The seventh and final neighbourhood, NN-IMN-GC, contains a mixed collection of highly distinct non-neuronal cell types, immature neuronal types and granule cell types.

“the seventh and final neighbourhood, NN–IMN–GC, contains a mixed collection of highly distinct non-neuronal cell types, immature neuronal types and granule cell types”

All serotonergic clusters reside in various raphe nuclei within midbrain or medulla.

“All of these clusters reside in the various raphe nuclei within midbrain or medulla.”

The Oligo subclass is divided into four supertypes corresponding to different stages of oligodendrocyte maturation: COP, NFOL, MFOL, and MOL.

“The Oligo subclass is further divided into four supertypes corresponding to different stages of oligodendrocyte maturation: committed oligodendrocyte precursors (COP), newly formed oligodendrocytes (NFOL), myelin-forming oligodendrocytes (MFOL), and mature oligodendrocytes (MOL)”

All non-neuronal cell types across the mouse brain are classified into 5 classes, 23 subclasses, 45 supertypes and 117 clusters.

“All non-neuronal cell types across the mouse brain are classified into 5 classes, 23 subclasses, 45 supertypes and 117 clusters”

The Glut-GABA clusters are widely distributed in different parts of the brain.

“These clusters are widely distributed in different parts of the brain.”

The fifth neuronal neighbourhood, MB-HB-Glut-Sero-Dopa, contains all glutamatergic, serotonergic and dopaminergic neuronal types in midbrain and hindbrain.

“The fifth neuronal neighbourhood, MB–HB-Glut–Sero–Dopa, contains all glutamatergic, serotonergic and dopaminergic neuronal types in midbrain (MB) and hindbrain (HB)”

Additional 28 dopaminergic clusters are spread across 14 subclasses.

“as well as an additional 28 clusters spread across 14 subclasses.”

Some neuropeptides are expressed at high levels in a moderate number of clusters (e.g., Cartpt, Nts, Pdyn, Gal, Tac2, Grp, Vip, Crh, Trh and Cort).

“some are expressed at high levels in a moderate number of clusters (for example, Cartpt, Nts, Pdyn, Gal, Tac2, Grp, Vip, Crh, Trh and Cort)”

The OPC-Oligo class contains two subclasses, oligodendrocyte precursor cells (OPC) and oligodendrocytes (Oligo).

“The OPC–Oligo class contains two subclasses, oligodendrocyte precursor cells (OPC) and oligodendrocytes (Oligo).”

Noradrenergic neurons are found exclusively in subclass 251.

“Noradrenergic neurons52,53 are found exclusively in subclass 251.”

The study identified clusters corresponding anatomically to all classically defined dopaminergic neuron groups A8-A16 across the brain.

“We identified clusters that correspond anatomically to all classically defined dopaminergic neuron groups—A8–A16—across the brain”

Most (35) of the 39 dopaminergic clusters also co-release glutamate (11 clusters) or GABA (10 clusters), or both glutamate and GABA (14 clusters).

“Most (35) of the 39 dopaminergic clusters also co-release glutamate (11 clusters) or GABA (10 clusters), or both glutamate and GABA (14 clusters).”

The majority of neuronal clusters release a single neurotransmitter-either glutamate or GABA.

“These marker genes indicate that the majority of neuronal clusters release a single neurotransmitter—either glutamate or GABA.”

Subclass 215 is located in substantia nigra, compact part (SNc), VTA and midbrain raphe nuclei (RAmb) areas.

“Subclass 215, located in substantia nigra, compact part (SNc), VTA and midbrain raphe nuclei (RAmb) areas”

The Vascular class consists of 5 subclasses: ABC, VLMC, Peri, SMC and Endo.

“The Vascular class consists of 5 subclasses: arachnoid barrier cells (ABC), vascular leptomeningeal cells (VLMC), pericytes (Peri), smooth muscle cells (SMC) and endothelial cells (Endo).”

Region-specific astrocytes Astro-OLF, Astro-TE, Astro-NT and Astro-CB are arranged in the UMAP in an anterior-to-posterior order, consistent with their spatial patterning.

“Region-specific astrocytes Astro-OLF, Astro-TE, Astro-NT and Astro-CB are arranged in the UMAP in an anterior-to-posterior order (Fig. 4b), consistent with their spatial patterning.”

Immature neurons in the SGZ, SVZ and RMS have a shared gene expression pattern that includes expression of immature neuron markers such as Draxin, Prox1, Mex3a and Dcx.

“Immature neurons in the SGZ, SVZ and RMS have a shared gene expression pattern that includes the expression of immature neuron markers such as Draxin, Prox1, Mex3a and Dcx”

Distinct gene expression patterns include more lineage-specific genes, such as Rbfox3 and Frmd7 for more mature OB neurons, and C1ql2 and Smad3 for mature DG neurons.

“distinct gene expression patterns include more lineage-specific genes, such as Rbfox3 and Frmd7 for more mature OB neurons, and C1ql2 and Smad3 for mature DG neurons”

An inside-out spatial gradient was observed in MOB among the four OEC clusters.

“We observed an inside-out spatial gradient in MOB among the four OEC clusters”

The study identified transcription factors that potentially serve as master regulators for many non-neuronal cell types.

“We identified transcription factors that potentially serve as master regulators for many of these non-neuronal cell types”

The 1180 OPC NN_2 supertype is found mostly in grey matter areas.

“by contrast, the 1180 OPC NN_2 supertype is found mostly in grey matter areas”

Marker genes for VLMC clusters are enriched in extracellular matrix components and transmembrane transporters, including collagens and solute carriers with distinct cell-type specificity.

“Marker genes for VLMC clusters are enriched in extracellular matrix components and transmembrane transporters, including collagens and solute carriers with distinct cell-type specificity”

The scRNA-seq data show a trajectory from immature neurons to mature neurons in DG.

“The scRNA-seq data show a trajectory from immature neurons to mature neurons in DG”

Immature neurons in the SVZ and RMS are marked by expression of cell cycle-associated genes like Top2a and Mki67.

“Immature neurons in the SVZ and RMS are marked by the expression of cell cycle-associated genes like Top2a and Mki67”

Astrocytes that follow the same trajectory as immature neurons in the RMS also show changes in gene expression along the trajectory that are similar to the gene expression changes in the IMN population.

“Astrocytes that follow the same trajectory as the immature neurons in the RMS also show changes in gene expression along the trajectory that are similar to the gene expression changes in the IMN population.”

The numbers of clusters from different regions do not correlate with the numbers of cells profiled by scRNA-seq even when corrected for brain region volumes; rather, region-specific characteristics dominate.

“We found that the numbers of clusters from different regions do not correlate with the numbers of cells profiled by scRNA-seq even when corrected for brain region volumes (Fig. 6b,c); rather, region-specific characteristics dominate.”

Many co-expressed homologues show subtle but interesting distinctions.

“Many co-expressed homologues show subtle but interesting distinctions.”

Both Gini coefficient and Shannon diversity index reveal very high inequality (highly localized patterns) in spatial distribution of each neuronal subclass.

“We further used the Gini coefficient and Shannon diversity index to measure the extent of variation in spatial distribution among subclasses (Fig. 6a; also see Extended Data Fig. 14 for Gini coefficient), and both reveal very high inequality (that is, highly localized patterns) in spatial distribution of each neuronal”

Each major brain region has its own specific sets of both glutamatergic and GABAergic neuronal subclasses that are mostly colocalized.

“Each major brain region has its own specific sets of both glutamatergic and GABAergic neuronal subclasses that are mostly colocalized.”

CNU exhibits an intermediate level of diversity.

“CNU exhibits an intermediate level of diversity.”

Lhx3 and Lhx4 are co-expressed in very specific glutamatergic types in hindbrain and pineal gland.

“Lhx3 and Lhx4 are co-expressed in very specific glutamatergic types in hindbrain and pineal gland.”

The results presented above showed that each cell type has a specific spatial localization.

“The results presented above showed that each cell type has a specific spatial localization.”

For most other modules, each module consisted of a few transcription factors that are homologues-for example, Nfia, Nfib and Nfix, the Zic family, the Irx family, the Ebf family, En1 and En2, Lhx6 and Lhx8, Six3 and Six6, and Pou4f1, Pou4f2 and Pou4f3.

“For most other modules, each module consisted of a few transcription factors that are homologues—for example, Nfia, Nfib and Nfix, the Zic family, the Irx family, the Ebf family, En1 and En2, Lhx6 and Lhx8, Six3 and Six6, and Pou4f1, Pou4f2 and Pou4f3.”

After volume normalization, pallium, CNU, thalamus and cerebellum contain smaller numbers of clusters, suggesting lower complexity than hypothalamus, midbrain and hindbrain.

“Overall, after volume normalization, pallium, CNU, thalamus and cerebellum contain smaller numbers of clusters, suggesting lower complexity than hypothalamus, midbrain and hindbrain.”

The median cluster recall accuracy of cross-validation with transcription factors is between that of all DEGs and the random subset of DEGs.

“The median cluster recall accuracy of cross-validation with transcription factors is between that of all DEGs and the random subset of DEGs.”

Transcription factor marker genes show the lowest correlations in gene expression among all clusters compared with functional genes, adhesion molecules and all DEGs.

“We found that transcription factor marker genes show the lowest correlations in gene expression among all clusters compared with functional genes, adhesion molecules and all DEGs”

Transcriptomically more similar cell types are located closer to each other spatially-for example, neuronal subclasses within the same class are mostly colocalized within the same major brain region.

“Transcriptomically more similar cell types are located closer to each other spatially—for example, neuronal subclasses within the same class are mostly colocalized within the same major brain region.”

The LIM homeobox family can be split into multiple groups with complementary expression patterns that together cover most neuronal types in the brain.

“Particularly intriguing is the LIM homeobox family, which can be split into multiple groups with complementary expression patterns that together cover most neuronal types in the brain.”

Cluster sizes vary among major brain regions, with hypothalamus, midbrain and hindbrain containing more smaller clusters.

“Furthermore, cluster sizes (tha is, the number of cells in each cluster) also vary among major brain regions (Extended Data Fig. 17b), with hypothalamus, midbrain and hindbrain containing more smaller clusters.”

Conversely, transcriptomically more distant cell types are spatially further apart from each other.

“Conversely, transcriptomically more distant cell types are spatially further apart from each other.”

Owing to the sheer scale of the atlas, the study has not extensively incorporated the vast amount of existing data and knowledge about cell types in many parts of the brain to help better annotate the cell-type atlas.

“Lastly, owing to the sheer scale of the atlas, we have not extensively incorporated the vast amount of existing data and knowledge about cell types in many parts of the brain to help better annotate the cell-type atlas.”

There are distinct features of cell-type organization in different major brain structures.

“Another notable finding is the distinct features of cell-type organization in different major brain structures (Fig. 6).”

In the adult brain, transcription factors are major determinants in defining cell types across all regions of the brain.

“Here we found that in the adult brain, transcription factors also are major determinants in defining cell types across all regions of the brain.”

There is a wide spectrum of expression patterns among different neuropeptide genes-some are widely expressed in many cell types, whereas others are highly specific to one or few cell types.

“Similarly, there is a wide spectrum of expression patterns among the different neuropeptide genes—some are widely expressed in many cell types, whereas others are highly specific to one or few cell types (Extended Data Fig. 10).”

The anterior and dorsal brain regions, including OLF, isocortex, HPF, STR, thalamus and cerebellum, contain cell classes and types that are highly distinct from the other parts of the brain.

“The anterior and dorsal brain regions, including OLF, isocortex, HPF, STR, thalamus and cerebellum, contain cell classes and types that are highly distinct from the other parts of the brain.”

Comparison of gene expression correlation matrices among all pairs of clusters showed that transcription factors have the greatest overall power to distinguish cell types.

“Comparison of gene expression correlation matrices among all pairs of clusters showed that transcription factors have the greatest overall power to distinguish cell types (Fig. 5 and Extended Data Fig. 3).”

The re-implemented DEG method computed DEGs between ~13.5 million pairs of clusters within a day on a single Linux server.

“In this way, we were able to compute DEGs between ~13.5 million pairs of clusters within a day on a single Linux server.”

The third mode is intersection between different sets of transcription factors that define molecular identity or spatial specificity, respectively, within a cell type.

“The third represents the intersection between different sets of transcription factors that define molecular identity or spatial specificity, respectively, within a cell type.”

The scRNA-seq dataset contained many types of low-quality cells, including damaged cells and mixed debris of various cell-type combinations.

“First, owing to the difficulty in dissociating and isolating intact cells from the adult brain tissue—especially in highly myelinated areas—our scRNA-seq dataset contained many types of low-quality cells, including damaged cells and mixed debris of various cell-type combinations.”

The vascular 329 ABC NN subclass also displays a similar phase shift.

“Of note, the vascular 329 ABC NN subclass also displays a similar phase shift.”

There is a high degree of correspondence between transcriptomic identity and spatial specificity.

“One of the most notable findings from our study is the high degree of correspondence between transcriptomic identity and spatial specificity (Figs. 2–4 and 6).”

The study created a comprehensive, high-resolution transcriptomic cell-type atlas for the whole adult mouse brain based on combination of whole-brain-scale scRNA-seq and MERFISH datasets.

“In this study, we created a comprehensive, high-resolution transcriptomic cell-type atlas for the whole adult mouse brain based on the combination of whole-brain-scale scRNA-seq and MERFISH datasets.”

There is a diverse set of neuronal clusters with glutamate-GABA co-transmission from many brain regions.

“We found a diverse set of neuronal clusters with glutamate–GABA co-transmission from many brain regions (Extended Data Fig. 9).”

Detailed gene expression analysis at class and subclass levels revealed widespread expression differences of canonical circadian clock genes between the light and dark phases.

“Detailed gene expression analysis at class and subclass levels revealed widespread expression differences of canonical circadian clock genes between the light and dark phases (Extended Data Fig. 18).”

On the basis of the MERFISH data, these sex-skewed clusters mostly reside in specific regions of PAL, sAMY, hypothalamus and hindbrain.

“On the basis of the MERFISH data, these clusters mostly reside in specific regions of PAL, sAMY, hypothalamus and hindbrain.”

The cell-type atlas was hierarchically organized into four nested levels: 34 classes, 338 subclasses, 1,201 supertypes and 5,322 clusters.

“The cell-type atlas was hierarchically organized into four nested levels: 34 classes, 338 subclasses, 1,201 supertypes and 5,322 clusters (Fig. 1).”

Cells from the ventral part of the brain-including ventral PAL, extended amygdala, hypothalamus, midbrain, pons and medulla-form many small clusters that are closely related to each other.

“By contrast, cells from the ventral part of the brain—including ventral PAL, extended amygdala, hypothalamus, midbrain, pons and medulla—form many small clusters that are closely related to each other.”

Every subclass (and all supertypes and many clusters within each) has a unique and specific spatial localization pattern within the brain.

“Every subclass (and all supertypes and many clusters within each) has a unique and specific spatial localization pattern within the brain.”

The first mode of coordination among transcription factors is coordinated expression of different transcription factors (often pairs) within the same transcription factor gene family in specific cell types.

“The first mode is the coordinated expression of different transcription factors (often pairs of transcription factors) within the same transcription factor gene family in specific cell types.”

The 262 Pineal Crx Glut subclass has particularly strong circadian gene expression fluctuations.

“The 262 Pineal Crx Glut subclass, located in the dorsal part of the third ventricle and on top of superior colliculus (SC) in the MERFISH data, which probably represents the pinealocytes that evolved from photoreceptor cells and secret melatonin116, has particularly strong circadian gene expression fluctuations (Extended Data Fig. 18b,c).”

• cites [116] Maronde, E. & Stehle, J. H. The mammalian pineal gland: known facts, unknown facets. Trends Endocrinol. Metab. 18, 142–149 (2007). sourceNot checked — no full text of the source

Transitional cell types in the transcriptomic space are also found crossing regional boundaries.

“Transitional cell types in the transcriptomic space are also found crossing regional boundaries.”

The study identified transcription factor genes and co-expression modules that are specific at different hierarchical levels.

“We identified transcription factor genes and co-expression modules that are specific at different hierarchical levels (Fig. 5d, Extended Data Fig. 13 and Supplementary Table 8).”

In the 094 SCH Six6 Cdc14a Gaba subclass, most clock genes (for example, Per1, Per3, Dbp, Nr1d1 and Nr1d2) have higher levels of expression in the light phase than the dark phase.

“Furthermore, in the 094 SCH Six6 Cdc14a Gaba subclass, which is specific to the suprachiasmatic nucleus (SCH), the circadian pacemaker of the brain, most clock genes (for example, Per1, Per3, Dbp, Nr1d1 and Nr1d2) have higher levels of expression in the light phase than the dark phase”

• AssumptionRests on an assumption: The 094 SCH Six6 Cdc14a Gaba subclass is specific to the suprachiasmatic nucleus, which is the circadian pacemaker of the brain.

The relative relatedness between transcriptomic types is strongly correlated with the spatial relationship between them.

“The relative relatedness between transcriptomic types is strongly correlated with the spatial relationship between them (Fig. 6a and Extended Data Figs. 3, 15 and 16).”

The study identified many sets of neuronal types with varying degrees of similarity, including highly distinct neuronal types and transitional neuronal types across regions.

“We identified many sets of neuronal types with varying degrees of similarity with each other, including highly distinct neuronal types as well as transitional neuronal types across regions.”

Neuromodulatory cell types often have closely related glutamatergic and/or GABAergic clusters within the same subclass, showing a high degree of heterogeneity in neurotransmitter content in these cell populations.

“The neuromodulatory cell types often have closely related glutamatergic and/or GABAergic clusters within the same subclass, showing a high degree of heterogeneity in neurotransmitter content in these cell populations (Fig. 3).”

There are numerous co-expression combinations of two or more neuropeptide genes in many neuronal clusters.

“Furthermore, there are numerous co-expression combinations of two or more neuropeptide genes in many neuronal clusters (Supplementary Table 7).”

The second mode is combination of transcription factors at different hierarchical branch levels to collectively define the identity of leaf-node cell types.

“The second is the combination of transcription factors at different hierarchical branch levels to collectively define the identity of the leaf-node cell types.”

At the cluster level, non-neuronal cell types show great spatial specificity, especially astrocytes, ependymal cells, tanycytes and VLMCs.

“However, at the cluster level, we also observed a great degree of spatial specificity in non-neuronal cell types, especially for astrocytes, ependymal cells, tanycytes and VLMCs, indicating specific neuron–glia and glia–vasculature interactions (Fig. 4).”

Transcriptomically similar cell types are often found in the same region, or in some cases in related regions that have a common developmental origin.

“Transcriptomically similar cell types are often found in the same region, or in some cases in related regions that have a common developmental origin.”

Within the whole mouse brain scRNA-seq dataset, a complete subset of data covering all brain regions was collected from the dark phase of the circadian cycle, totaling 1,121,542 10xv3 cells.

“Within the whole mouse brain scRNA-seq dataset, we also collected a complete subset of data covering all brain regions from the dark phase of the circadian cycle (Supplementary Table 2, total 1,121,542 10xv3 cells).”

Some hindbrain neuronal clusters had high within-cluster heterogeneity.

“We also identified some hindbrain neuronal clusters that had high within-cluster heterogeneity.”

The Purkinje neuron population was very small in the dataset.

“This is what we observed in Purkinje neuron population, which was very small in our dataset.”

The imputation strategy for MERFISH helped reduce the effect of contamination from neighbouring cells due to imperfect segmentation.

“This strategy still helped to reduce the effect of contamination from neighbouring cells due to imperfect segmentation.”

Confusion in transcriptomic signatures usually only exists among neighbouring CCFv3 regions.

“confusion usually only exists among neighbouring regions.”

Partial separation of upper and lower cortical layers is observable, due to enrichment of CGE and MGE neurons respectively.

“We could still observe partial separation of upper and lower cortical layers due to enrichment of CGE and MGE neurons respectively.”

Mapping 10xMulti snRNA-seq data to the scRNA-seq taxonomy identified several 10xMulti clusters with very few mapped scRNA-seq cells and clear transcriptional signature.

“Through mapping our newly generated 10xMulti snRNA-seq data with the scRNA-seq taxonomy (see ‘Assigning cell-type identities’), we identified several 10xMulti clusters that have very few mapped scRNA-seq cells with clear transcriptional signature.”

For most CCFv3 regions, each region contains cells with distinct transcriptomic signatures from other regions.

“This analysis showed that for most of the CCFv3 regions, each region contains cells with distinct transcriptomic signatures from other regions”

A simpler imputation strategy computing KNNs within the same cluster and/or subclass could not preserve transitions between clusters or subclasses, exaggerating separations between cell types, especially at finer level.

“However, the imputation values using this strategy could not preserve the transitions between clusters or subclasses, exaggerating the separations between cell types, especially at the finer level.”

Clustering using 534 transcription factor marker genes recapitulated existing knowledge of cell types including spatial distribution and lineage relationships better than using all 8,460 marker genes.

“We tried clustering using either all 8,460 marker genes or 534 transcription factor marker genes only, and found the result based on transcription factors recapitulate existing knowledge of cell types including spatial distribution and lineage relationships better.”

Examination of MERFISH spatial distribution revealed that some existing clusters in the Astro-Epen class did not fully capture the rich spatial gradients present in the dataset.

“After careful examination of spatial distribution of each cell types based on the MERFISH dataset, we realized that the some of the existing clusters in Astro–Epen class did not fully capture the rich spatial gradients present in the dataset.”

After removing additional noise clusters, the dataset had 5,200 clusters with 4,041,289 cells.

“After removing these additional noise clusters, we had 5,200 clusters with 4,041,289 cells.”

Only 51 of 59 sections fell within the boundaries of the CCFv3.

“Only 51/59 sections fell within the boundaries of the CCFv3.”

The MERFISH dataset can resolve the vast majority of clusters owing to strong correlation of DEG expression between 10xv3 and MERFISH clusters.

“The MERFISH dataset can resolve the vast majority of clusters owing to strong correlation of DEG expression between 10xv3 and MERFISH clusters (Extended Data Fig. 7e–g).”

The authors developed a set of stringent quality control metrics guided by pilot clustering results that informed them on characteristics of low-quality single-cell transcriptomes.

“We developed a set of stringent quality control (QC) metrics guided by pilot clustering results that informed us on characteristics of low-quality single-cell transcriptomes (Extended Data Fig. 1a–c, Supplementary Table 4 and Methods).”

The high quality of the scRNA-seq and snRNA-seq data included in the final taxonomy is indicated by the high gene counts and unique molecular identifier counts across cell types.

“The high quality of the scRNA-seq and snRNA-seq data included in the final taxonomy is indicated by the high gene counts and unique molecular identifier (UMI) counts across cell types (Extended Data Fig. 5a–d).”

The imputed spatial expression patterns were consistent with the actual expression patterns by both MERFISH and in situ hybridization from the Allen Brain Atlas for genes that are on the 500 MERFISH gene panel (Calb2, Baiap3 and Lypd1) and those that are not (Foxp2).

“The imputed spatial expression patterns were consistent with the actual expression patterns by both MERFISH and in situ hybridization from the Allen Brain Atlas25 for genes that are on the 500 MERFISH gene panel (Calb2, Baiap3 and Lypd1) and those that are not (Foxp2) (Extended Data Fig. 8b–f).”

• cites [25] Allen Mouse Brain Atlas. Allen Institute for Brain Science https://mouse.brain-map.org/ (2004).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

After removing additional noise clusters, the final set contained around 4.0 million high-quality single-cell transcriptomes.

“During this process, we identified and removed an additional set of ‘noise’ clusters (usually doublets or mixed debris; Methods) that had escaped the initial QC process, resulting in a final set of around 4.0 million high-quality single-cell transcriptomes (Extended Data Fig. 1a,d,e).”

High correlations between imputed expression and the original MERFISH expression, as well as the reference 10xv3 expression for each gene, were observed.

“High correlations between imputed expression and the original MERFISH expression, as well as the reference 10xv3 expression for each gene were observed (Extended Data Fig. 8a).”

Nearly all subclasses with a dominant modulatory neurotransmitter contain clusters transmitting glutamate and/or GABA only, as well as various patterns of co-transmission.

“Overall, a pattern emerged where nearly all subclasses with a dominant modulatory neurotransmitter contain clusters transmitting glutamate and/or GABA only, as well as various patterns of co-transmission”

No clusters could co-release serotonin and dopamine.

“we found no clusters that could co-release serotonin and dopamine”

This indicates a high degree of heterogeneity in neurotransmitter release and co-release among closely related neuronal types that may have common developmental origins.

“indicating a high degree of heterogeneity in neurotransmitter release and co-release among closely related neuronal types that may have common developmental origins.”

The QC process excluded the possibility of doublet or low-quality cell contamination accounting for the heterogeneity.

“Our QC process excluded the possibility of doublet or low-quality cell contamination accounting for the heterogeneity.”

The datasets provide a rich resource for exploration of neuropeptide ligand-receptor interactions across the entire brain.

“Our datasets provide a rich resource for the exploration of neuropeptide ligand–receptor interactions across the entire brain.”

The spatial distribution of all non-neuronal cell types in the mouse brain was confirmed and further refined by the MERFISH data.

“The spatial distribution of all non-neuronal cell types in the mouse brain was confirmed and further refined by the MERFISH data.”

This suggests that transcription factors have the greatest capability to differentiate cell types.

“suggesting that transcription factors have the greatest capability to differentiate cell types.”

This analysis indicates that most CCFv3 structures contain distinct neuronal cell types.

“This analysis indicates that most CCFv3 structures contain distinct neuronal cell types.”

These results quantify the major roles transcription factors can have in defining cell-type identities.

“These results quantify the major roles transcription factors can have in defining cell-type identities.”

These ventral cell types often have restricted spatial localization that likely underlies the small nuclei characteristic of these regions.

“These cell types often have restricted spatial localization that likely underlies the small nuclei characteristic of these regions.”

The atlas enables creation of cell-type-targeting tools for labelling and manipulating specific cell types to probe and modify their functions in vivo.

“It enables the creation of cell-type-targeting tools for labelling and manipulating specific cell types to probe and modify their functions in vivo.”

These results support the extraordinary diversity in intercellular communications in the brain.

“These results support the extraordinary diversity in intercellular communications in the brain.”

Comparison with subsequently generated MERFISH data showed that CCFv3-based microdissections were largely accurate at cell subclass and major brain region levels.

“Comparison with subsequently generated MERFISH data showed that our CCFv3-based microdissections were largely accurate at cell subclass and major brain region levels (Extended Data Fig. 2h).”

This spatial specificity indicates specific neuron-glia and glia-vasculature interactions.

“indicating specific neuron–glia and glia–vasculature interactions”

These findings reveal how transcription factors form a combinatorial code that lays out the highly complex cell-type landscape.

“These findings reveal how transcription factors form a combinatorial code that lays out the highly complex cell-type landscape.”

The whole mouse brain transcriptomic cell-type atlas captured circadian state-dependent gene expression changes.

“These results suggest that our whole mouse brain transcriptomic cell-type atlas also captured circadian state-dependent gene expression changes.”

The cell-type classification is not significantly affected by different circadian states.

“Although supervised analysis can reveal these changes, our cell-type classification is not significantly affected by the different circadian states.”

The 094 SCH Six6 Cdc14a Gaba subclass pacemaker cells are at a different phase of the circadian cycle of gene expression from the rest of the brain.

“suggesting that the pacemaker cells are at a different phase of the circadian cycle of gene expression from the rest of the brain, consistent with previous findings117”

• cites [117] Wen, S. et al. Spatiotemporal single-cell analysis of gene expression in the mouse suprachiasmatic nucleus. Nat. Neurosci. 23, 456–467 (2020). sourceNot checked — no full text of the source

• AssumptionRests on an assumption: The 094 SCH Six6 Cdc14a Gaba subclass is specific to the suprachiasmatic nucleus, which is the circadian pacemaker of the brain.

Cell types in the midbrain and hindbrain may not be fully represented or fully resolved in the atlas.

“Thus, cell types in the midbrain and hindbrain may not be fully represented or fully resolved in our atlas.”

The atlas provides baseline gene expression patterns that enable investigation of dynamic changes in gene expression and cellular function in different physiological and diseased conditions.

“The atlas provides baseline gene expression patterns that enable investigation of the dynamic changes in gene expression and cellular function in different physiological and diseased conditions.”

During QC, it is likely that some cell types were more selectively depleted than others, especially large neurons vulnerable to damage during tissue dissociation, such as Purkinje cells and large motor neurons in the midbrain and hindbrain.

“During this process, it is likely that some cell types were more selectively depleted than others, especially large neurons that are more vulnerable to damage during tissue dissociation—such as Purkinje cells and large motor neurons in the midbrain and hindbrain.”

The automatic iterative clustering method ensured that all pairs of clusters, even at the finest level, were separable by stringent differential gene expression criteria.

“We used the automatic iterative clustering method, iter_clust_big, which performed clustering in top down manner into cell types of increasingly finer resolution without any human intervention, while ensuring that all pairs of clusters, even at the finest level, were separable by stringent differential gene expression criteria as follows:”

Individual mouse variability should not affect cell-type identities.

“Thus, individual mouse variability should not affect cell-type identities (Extended Data Fig. 5).”

The missing of these 10xMulti clusters in the scRNA-seq taxonomy led to poor mapping and 'holes' in MERFISH cell-type annotation.

“The missing of these clusters in our scRNA-seq taxonomy led to poor mapping and ‘holes’ in our MERFISH cell-type annotation.”

Full-scale integration of the entire 10xMulti snRNA-seq dataset could compromise the many high-resolution clusters already present in the scRNA-seq taxonomy.

“we did not proceed with full-scale integration of the entire 10xMulti snRNA-seq dataset which could compromise the many high-resolution clusters already present in our scRNA-seq taxonomy.”

• AssumptionRests on an assumption: Full-scale integration of the entire 10xMulti snRNA-seq dataset could compromise the many high-resolution clusters already present in the scRNA-seq taxonomy.

Most additional noise clusters were very small, likely doublets of damaged cells.

“Most of these clusters were very small, likely doublets of damaged cells.”

Cosine metric was chosen for KNN imputation because of overall conservation of gene expression at the cluster level.

“We chose cosine metric due to overall conservation of gene expression at the cluster level (Extended Data Fig. 7d).”

The study did not include Slc6a1 in neurotransmitter criteria to minimize false positives, even at risk of false negatives.

“Therefore, we did not include Slc6a1 in our criteria in order to minimize false positives, even at the risk of having some false negatives.”

• AssumptionRests on an assumption: Slc6a1 is widely expressed in all GABAergic neurons, as well as astrocytes, and even many subcortical glutamatergic neurons.

Using too many genes to infer KNNs could make the inference more susceptible to platform differences because conservation is not perfect.

“On the other hand, the conservation is not perfect, and using too many genes to infer KNNs could make the inference more susceptible to platform differences.”

The precision of the mutual prediction analysis is highly dependent on the accuracy of CCFv3 registration.

“Note that the precision of this analysis is highly dependent on the accuracy of CCFv3 registration.”

Because of more gradient transition of cell-type composition between cortical regions, the specificity of cortical plate regions is limited to isocortex, OLF and HPF despite more granular dissection regions.

“Because of the more gradient transition of cell-type composition between cortical regions, the specificity of cortical plate regions is limited to isocortex, OLF and HPF despite more granular dissection regions.”