{"id":470491,"date":"2024-01-04T19:00:00","date_gmt":"2024-01-05T00:00:00","guid":{"rendered":"https:\/\/platohealth.ai\/decoding-the-spatiotemporal-regulation-of-transcription-factors-during-human-spinal-cord-development-cell-research\/"},"modified":"2024-01-04T23:56:22","modified_gmt":"2024-01-05T04:56:22","slug":"decoding-the-spatiotemporal-regulation-of-transcription-factors-during-human-spinal-cord-development-cell-research","status":"publish","type":"post","link":"https:\/\/platohealth.ai\/decoding-the-spatiotemporal-regulation-of-transcription-factors-during-human-spinal-cord-development-cell-research\/","title":{"rendered":"Decoding the spatiotemporal regulation of transcription factors during human spinal cord development – Cell Research","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
We present here a comprehensive spatiotemporal cellular atlas of the developing human spinal cord, based on two sources of transcriptomic data at single-cell resolution. The first dataset was obtained by performing scRNA-seq on individual cells in the developing human spinal cord, covering the early first to early third trimester (GW7\u201325), resulting in a transcriptional profile of 217,636 single cells. The second dataset consists of two sets of spatially resolved transcriptomic data, obtained using the 10\u00d7\u2009Visium and a single-cell transcriptomics method that quantified the expression levels of 1085 TFs (TF-seqFISH, see Materials and Methods), respectively (Fig. 1a<\/a>). To create a more comprehensive cellular landscape, we integrated our scRNA-seq data with data from a previous study,14<\/a><\/sup> resulting in a transcriptional census of 912,514 single cells across GW7 to GW25 (Fig. 1b<\/a>; Supplementary information, Fig. S1a<\/a>). Twenty major cell clusters were identified through unsupervised clustering and visualized via Uniform Manifold Approximation and Projection (UMAP), which we annotated using well-known cell type-specific markers (Fig. 1b<\/a>; Supplementary information, Fig. S1b<\/a> and Tables S1<\/a>, S2<\/a>). Our analysis revealed the existence of three cell types in the cell division cycle, labeled as cell cycle-1, -2, and -3, as well as neural progenitor cells (NPCs) and the neuronal types of In-INs, Ex-INs, and MNs (Fig. 1b<\/a>). In addition, we detected distinct cell types in the oligodendrocyte lineage, including oligodendrocyte progenitor cells (OPCs), committed oligodendrocyte progenitors (COPs), myelin-forming oligodendrocytes (MFOLs), and mature oligodendrocytes (MOLs) (Fig. 1b<\/a>). We also identified astrocytes and astrocyte progenitor cells (APCs), as well as one group of glial progenitor cells (GPCs) situated between the astrocyte and oligodendrocyte lineages in this UMAP embedding (Fig. 1b<\/a>). Gene expression analysis revealed that, in addition to markers like HOPX<\/i> and EGFR<\/i>, GPCs displayed expression profiles of both astrocytic (GFAP<\/i>, AQP4<\/i>, and SLC1A3<\/i>) and oligodendrocytic lineages (OLIG2<\/i> and OLIG2<\/i>) (Supplementary information, Fig. S1c<\/a>). Trajectory analysis further supported the potential bipotent nature of GPCs (Supplementary information, Fig. S1c<\/a>). Additionally, we discerned microglia, endothelial cells, pericytes, neural crest cells, mesoderm, and ependymal cells (Fig. 1b<\/a>).<\/p>\n a<\/b> A comprehensive spatiotemporally resolved transcriptomic landscape of the developing human spinal cord across GW7\u201327 was obtained through the use of scRNA-seq, TF-seqFISH, and 10\u00d7\u2009Visium technologies. This schematic provides an overview of the experimental design and analysis. b<\/b> The developing human spinal cord from GW7\u201325 was segmented into cervical, thoracic, and lumbar regions and then analyzed using scRNA-seq. Through unsupervised clustering and marker gene expression, 20 distinct cell types were identified and annotated. The expression profiles of marker genes are depicted through a visualization where each dot represents an individual cell colored according to the expression level (dark red, high expression; light blue, low). c<\/b> We present an analysis of the neurogenic, astrocytic, and oligodendrocytic lineages in the developing human spinal cord using RNA velocity (left). To aid interpretation, we have highlighted the three lineage trajectories separately (right). The changes in cell ratio over GWs are depicted by fitted curves, while a density map shows cell density at different gestational stages, with high density denoted by dark red. Unrelated cells were excluded from this UMAP analysis. d<\/b> At GW6, the MNX1+<\/sup> MNs, VGLUT1+<\/sup> Ex-INs, and GABA+<\/sup> In-INs could be easily identified in the ventral region of the spinal cord. The regions of interest are magnified in the boxes. Scale bars,100\u2009\u03bcm (left) and 20\u2009\u03bcm (right). e<\/b> The spatial separation of MNX1+<\/sup> MNs and VGLUT1+<\/sup> Ex-INs in the human spinal cord at GW9 is demonstrated in this image. The areas within the boxes are shown in higher magnification to provide a better visualization. Scale bars, 100\u2009\u03bcm (left), 20\u2009\u03bcm (right). f<\/b> The expression patterns of GFAP, OLIG2, and PDGFRA in the human spinal cord at GW11 are presented. The boxed regions are displayed at high magnification. Scale bar, 100\u2009\u03bcm. g<\/b> Expression of MBP in human spinal cord samples at GW22 is shown. Scale bars, 100\u2009\u03bcm.<\/p>\n<\/div>\n<\/div>\n Next, we employed RNA velocity analysis to infer the differentiation lineages leading to neuronal, astrocytic, and oligodendrocytic fates (Fig. 1c<\/a>). Intriguingly, the differentiation trajectories revealed that the three cell groups in cell division (cell cycle-1\/2\/3) are involved in astrocytic, neuronal, and oligodendrocytic lineages, respectively, with distinct differentially expressed genes (DEGs) (Fig. 1c<\/a>; Supplementary information, Fig. S1d<\/a> and Table S2<\/a>). These findings suggested that the genetic cues instructing the neurogenic, astrocytic, or oligodendrocytic fates emerge early, even in dividing cells. To elucidate the timing of neurogenic and gliogenic events in the developing human spinal cord, we analyzed the cell ratio dynamics for cells implicated in the neuronal, astrocytic, and oligodendrocytic lineages across GWs (Fig. 1c<\/a>; Supplementary information, Fig. S1e, f<\/a>). Our results demonstrate that neurogenesis in the developing human spinal cord is initiated very early, as neurogenic dividing cells (cell cycle-2) and NPCs are present in high proportion at GW7, the earliest stage included in our dataset (Fig. 1c<\/a>; Supplementary information, Fig. S1e, f<\/a>). Accordingly, abundant neuronal progenies, including MNs, Ex-INs, and In-INs classified according to their neurotransmitter properties, are detectable as early as GW7 (Fig. 1c<\/a>; Supplementary information, Fig. S1e, f<\/a>).<\/p>\n In contrast, gliogenic events are more hysteretic. In the astrocytic lineage, astrocytic dividing cells and APCs are detectable around GW10 and peak at GW15, while astrocytes increase gradually around GW10 and peak at GW20 (Fig. 1c<\/a>; Supplementary information, Fig. S1e, f<\/a>). In the oligodendrocyte lineage, the cell types at low differentiation states, such as the oligodendrocytic dividing cells (cell cycle-3), OPCs, and COPs, emerge around GW10 and gradually increase thereafter (Fig. 1c<\/a>; Supplementary information, Fig. S1e, f<\/a>). Well-differentiated MFOLs and MOLs arise from GW18 and increase substantially after GW20 (Fig. 1c<\/a>; Supplementary information, Fig. S1e, f<\/a>). These temporal rules of neurogenic and gliogenic events in the developing human spinal cord are further supported by immunofluorescence staining, where the MNX1+<\/sup> MNs, VGLUT1+<\/sup> Ex-INs, and GABA+<\/sup> In-INs can be detected as early as GW6, particularly in the ventral horn, indicating that neurogenesis in the human spinal cord may occur even earlier than GW6 (Fig. 1d<\/a>). Besides, we also probed the spatially characterized organization of MNX1+<\/sup> cells and VGLUT1+<\/sup> cells at GW9 (Fig. 1e<\/a>). Furthermore, we conducted immunostaining for genes related to glia differentiation in embryonic human spinal cord. Although absent at GW8, abundant GFAP+ astrocytes and PDGFRA+<\/sup>OLIG2+<\/sup> OPCs could be detected at GW11 (Fig. 1f<\/a>; Supplementary information, Fig. S1g<\/a>). Additionally, at GW22, MBP+<\/sup> oligodendrocytes were detected in the white matter of the spinal cord (Fig. 1g<\/a>; Supplementary information, Fig. S1g<\/a>). Taken together, our results demonstrate the early initiation of neurogenesis in the human spinal cord around GW6 and the hysteretic gliogenesis occurring around GW10. Compared with the counterpart events in the developing human cerebral cortex,15<\/a>,16<\/a>,17<\/a>,18<\/a><\/sup> the early initiation of neurogenesis and gliogenesis, especially the gliogenic events, is evident in the developing human spinal cord (Supplementary information, Fig. S1h<\/a>).<\/p>\n To investigate the molecular characteristics and spatial organization principles of neurogenesis in the embryonic human spinal cord, we conducted scRNA-seq analysis on extracted NPCs and neurons. Through the implementation of unsupervised clustering, RNA velocity analysis and subsequent visualization using UMAP, we identified distinct neuronal lineages that exhibited unique molecular signatures (Fig. 2a<\/a>; Supplementary information, Fig. S2a<\/a>). The NPCs within each lineage were found to have low pseudotime values and served as initiating cells (Fig. 2a<\/a>). We annotated these lineages based on characteristic expression of accredited TF markers, defining distinct NPC pools of dp1, dp2, dp3\u20136, p0, p0\/1, pMN, and p2\/3, and corresponding neuronal types of dI1, dI2, dI3, dI4, dI5, dI6, V0, V1, MN, V2, and V3 (Fig. 2a\u2013c<\/a>; Supplementary information, Fig. S2a<\/a>). For instance, mouse TF markers for dp1 (e.g. MSX1<\/i>, MSX2<\/i>) and dI1 (e.g. LHX2<\/i>, LHX9<\/i>, and BARHL1<\/i>) were found to be specifically enriched in human NPCs and neurons in lineage 1, which we therefore defined as dp1 and dI1, respectively (Fig. 2a\u2013c<\/a>; Supplementary information, Fig. S2a<\/a>). We identified six neuronal lineages in the developing human spinal cord, each with a characteristic cell composition. Lineage 1 consisted of dp1 and dI1; lineage 2 depicted the differentiation from dp2 towards dI2; lineage 3 comprised heterogeneous cell types of dp3\u20136 and dI3, dI4, dI5, dI6; lineage 4 illustrated the bifurcated differentiation of p0\u20131 towards V1 and V0; lineage 5 demonstrated the specification of pMN to MN; and lineage 6 illustrated the differentiation of p2\u20133 towards V2 and V3 (Fig. 2a<\/a>; Supplementary information, Fig. S2a<\/a>). Notably, certain cell types that are well-distinguished in mice were difficult to discern in the human fetal spinal cord due to the similarity in molecular features, such as the dp3\u20136, p0\u20131, and p2\u20133 (Fig. 2a\u2013c<\/a>; Supplementary information, Fig. S2a<\/a>). Besides, we observed that neurons with the same neurotransmitter properties were consistently clustered together, with glutamatergic dI3 and dI5 neurons close together in the UMAP, as were the GABAergic dI4 and dI6 (Fig. 2a<\/a>; Supplementary information, Fig. S2a<\/a>). Additionally, we identified novel gene expression patterns beyond well-known genetic markers. For instance, we found specific expression of CRABP1<\/i>, LHX2<\/i> and BARHL2<\/i> in dI1, EBF1<\/i>, EBF3<\/i>, and AJAP1<\/i> in dI5, THSZ2<\/i>, SST<\/i>, and CHL1<\/i> in dI6, as well as SLIT3<\/i>, ISL1<\/i> and ISL2<\/i> in MNs (Fig. 2c<\/a>). Immunostaining was performed for CRABP1 along with the classical dI1 marker LHX2, as well as for SLIT3 along with the MN-specific CHAT. The co-expression of the newly identified genes alongside well-established cell type markers validates our identification of novel genes characteristic of certain cell types (Supplementary information, Fig. S2b<\/a>).<\/p>\n a<\/b> The distinct neuronal lineages comprising NPCs and neurons are visualized using a UMAP plot that integrates the outcomes of RNA velocity analysis. To enhance clarity, each neuronal lineage is also depicted individually, and illustrated with gray dotted lines. The cells are colored based on their identities as NPCs, Ex-INs, In-INs, and MNs, their pseudotime scores, or based on their identities in different lineages. b<\/b> The expression profiles of DEGs across diverse NPC subtypes are shown in a dotplot. c<\/b> A heatmap depicting the expression patterns of DEGs enriched in various neuronal types in the developing human spinal cord is presented. d<\/b> RNA signals for 31 highly variable genes on a coronal section of the human spinal cord at GW8 are spatially presented, with each dot representing a single molecule of RNA. The spatial expression profiles of some example genes are shown individually for better visualization. e<\/b> The spatiotemporal arrangement of NPCs along the dorsoventral axis in the VZ of human spinal cord is inferred through an integrated analysis of scRNA-seq and TF-seqFISH data. The spatial distribution of NPC cell types is represented as a simplified density map based on the dorsoventral score. The NPCs are arranged along the DV axis from dorsal to ventral as: dp1, dp2, dp3\u20136, p0\u20132, pMN, and p3. f<\/b> The spatial expression patterns of TFs in the VZ along the dorsoventral axis of the human spinal cord at GW8 are visualized. Each dot represents an individual cell and is colored according to the expression level of the TF (red, high; gray, low). The dotted lines indicate the boundaries of the VZ in the human spinal cord at GW8. g<\/b> The immunostaining results visualize the spatial expression patterns of TFs, including OLIG3, PAX3, NKX6-1, ASCL1, FOXA2, OLIG2, and NKX2-2, in the VZ of the human spinal cord at GW8. Scale bar, 50 \u03bcm.<\/p>\n<\/div>\n<\/div>\n TFs are essential in regulating neural progenitor domains, cell specification, and the patterning of brain regions during development.19<\/a>,20<\/a><\/sup> In this study, we introduced an image-based single-cell TF-seqFISH technique to examine the expression profiles of 1085 TFs in the developing human spinal cord, in a spatial context. Our emphasis was on deciphering the spatial distribution of TFs within the ventricular zone (VZ)-resident NPCs. So, we performed a comprehensive analysis to determine the suitable spinal tissues for performing TF-seqFISH. We initially performed immunostaining for SOX2 and NEUN to label NPCs in the developing human spinal cord at various stages. The results demonstrated that SOX2+<\/sup> NPCs were notably abundant at GW8, congregating within a thickened VZ region. However, as development progresses, the SOX2+<\/sup>\u2009NPC-enriched VZ layer gradually diminishes (Supplementary information, Fig. S2c<\/a>). Next, we analyzed RNA levels in the NPCs of spinal cord tissues at different time points, revealing that a higher number of RNA molecules were detected at an early stage, especially at GW7\u20139. (Supplementary information, Fig. S2d<\/a>). Furthermore, we also took into account the availability of spinal cord tissues and the temporal alignment with the scRNA-seq dataset. Thus, we conducted TF-seqFISH on human spinal tissues collected at GW8, yielding a cellular-resolution spatial mapping of TF mRNA within the developing spinal cord slices containing 12,274 cells (Supplementary information, Fig. S2e<\/a>). Our visualization of 31 highly variable genes with spatial expression profiles revealed the heterogeneity of TF mRNA in VZ-restricted NPCs along the dorsoventral axis (Fig. 2d<\/a>; Supplementary information, Fig. S2e<\/a>). By integrating scRNA-seq and TF-seqFISH datasets, we were able to assign unique spatial identities to each VZ cell, including dp1, dp2, dp3\u20136, p0\/1, p2\/3, and pMN, respectively, along the dorsoventral axis (Supplementary information, Fig. S2f<\/a>). The spatial assignments of NPCs along the dorsoventral axis (dp1\u2192dp2\u2192dp3\u20136\u2192p0\/1\u2192pMN\u2192p2\/3) in our study are similar to those reported in developing mouse and chick spinal cords7<\/a>,13<\/a><\/sup> (Fig. 2e<\/a>).<\/p>\n Furthermore, we investigated the spatial organization pattern of TF configuration that contributes to progenitor subdivision in humans. Our results showed the spatially restricted expression of TFs that are crucial for defining progenitor identities in the VZ region (Fig. 2f<\/a>; Supplementary information, Fig. S2g<\/a>). Specifically, ATOH1<\/i>, which is dp1-specific, is mainly enriched in the dorsal-most cells, while dp3\u20136 enriched PAX3<\/i>, PAX7<\/i>, GSX1<\/i>, and GSX2<\/i> are widely distributed in the middle-upper part (Fig. 2f<\/a>; Supplementary information, Fig. S2g<\/a>). Additionally, we observed the ventrally restricted expression of TFs such as PRDM8<\/i>, NKX6-1<\/i>, NKX2-2<\/i>, NKX2-8<\/i>, and FOXA2<\/i> (Fig. 2f<\/a>; Supplementary information, Fig. S2g<\/a>). These findings were validated by immunostaining (Fig. 2g<\/a>). While our study unveiled the remarkable conservation of spatial gene expression in spinal NPCs across species evolution, we also identified distinct features unique to human spinal cord development. For instance, PTF1A<\/i> is widely distributed in the human spinal cord, with a pattern comparable to that of dp3\u20136 specific GSX1<\/i>, GSX2<\/i>, PAX3<\/i>, and PAX7<\/i> (Fig. 2f<\/a>; Supplementary information, Fig. S2g<\/a>), unlike its restriction to dp4 neurons in the mouse spinal cord.21<\/a><\/sup> This difference may explain the indiscernibility of dp3\u20136 progenitor cells in the human spinal cord.<\/p>\n In summary, our study not only identified transcriptionally diverse NPCs based on scRNA-seq analysis but also elucidated the spatial organization pattern of NPC and TF expression in the developing human spinal cord. Our findings suggested that, although the spatial transcriptome of NPCs was largely conserved in species evolution, human fetal spinal cord development followed subtly different genetic regulatory rules.<\/p>\n Following the delineation of the spatial transcriptome of NPCs, we sought to elucidate the rules governing neuronal organization in the developing human spinal cord. To this end, we employed unsupervised classification based on the expression of 1085 TFs in the TF-seqFISH datasets, which resulted in the identification of 18 distinct molecular clusters (Fig. 3a, b<\/a>; Supplementary information, Fig. S3a, b<\/a>). These clusters were further demarcated into 5 spatially discrete domains (Region 1\u20135) through a correlation-based analysis (Fig. 3a<\/a>; Supplementary information, Fig. S3b\u2013c<\/a>). Specifically, Clusters 2, 7, 9, 11, 12, 16, and 18 were associated with Region 1; Clusters 4 and 13 were assigned to Region 2; Cluster 10 was related to Region 3; Clusters 14 and 5 were located in Region 4; Clusters 1, 3, 6, 8, and 17 were grouped in Region 5 (Fig. 3a<\/a>; Supplementary information, Fig. S3b, c<\/a>). According to the developing human spinal cord atlas,22<\/a><\/sup> Region 1, located medially, primarily consists of dividing progenitor cells. Region 2, an intermediate zone (IZ) adjacent to the proliferative areas, is designated as a Sojourn zone, and contains premigratory neurons. The dorsally localized mantle zone, Region 3, is comprised of migrating and settling dorsal horn neurons. Region 4, an in-between mantle zone (MZ), contains intermediate interneurons and migrating neurons towards the ventral horn. Finally, the ventral horn-specific Region 5 is mainly composed of settled ventral horn neurons (Fig. 3a<\/a>). These TF-classified spatial territories provide a valuable molecular basis for delineating neuroanatomical structures.<\/p>\n a<\/b> Unsupervised clustering in TF-seqFISH dataset identified 18 molecularly defined clusters and their corresponding spatial localization in a slice of developing human spinal cord at GW8. Individual cells are represented by dots and colored based on the clusters, revealing five distinct anatomical regions based on spatially restricted cell distribution. To aid clarity, a schematic illustration is also provided, where distinct regions are denoted by digits and blue arrows indicate migratory directions of neurons in the spinal cord. b<\/b> Visualization of the imputed gene expression profiles in the developing human spinal cord, with each dot representing an individual cell and color coded according to the expression level (red, high; gray, low). c<\/b> A heatmap depicting the expression profiles of DEGs among cells located in Regions 1, 2, and 3 of the developing human spinal cord is presented, with enriched GO terms for the DEGs listed. d<\/b> Visualization of laminar cell organization and gene expression, including TFs and non-TF genes, along the mediolateral axis in the dorsal horn of the developing human spinal cord facilitated by utilizing the imputed data. Each dot represents an individual cell and is colored based on expression level (red, high; gray, low). The leftmost column is a zoomed-in image of the boxed area in a<\/b>. e<\/b> A dot plot illustrating the expression profiles of DEGs among Ex-INs, In-INs, and MNs in the developing human spinal cord. f<\/b> The spatial organization of Ex-INs, In-INs, and MNs in the developing human spinal cord is displayed, with the spatially-restricted expression of genes specific for Ex-INs and In-INs in the dorsal horn also depicted in the boxed area based on the imputed data. A schematic representation of the sandwich-like arrangement of Ex-INs and In-INs in the dorsal horn is provided for clarity.<\/p>\n<\/div>\n<\/div>\n Then, we have imputed expression profiles of transcriptome-wide genes by combining the TF-seqFISH and scRNA-seq data (Supplementary information, Fig. S2f<\/a>). Leveraging the imputed data, we deduced the spatial expression patterns of non-TF genes as well, providing a comprehensive resource for decoding the cellular and molecular organization principles of the developing human fetal spinal cord. With this dataset, we aimed to unravel the enigma of neuron migration and differentiation in the human spinal cord. In the dorsal horn, Regions 1, 2, and 3 are highly laminated (Fig. 3a<\/a>). Analysis of DEGs and gene ontology (GO) reveals that VZ cells in Region 1 are involved in cell division, maintenance of the neuronal stem cell population, and regulation of neurogenesis (Fig. 3c<\/a>; Supplementary information, Table S3<\/a>). Conversely, cells in the laterally localized Regions 2 (IZ) and 3 (MZ) are implicated in neuron differentiation, migration, projection development, and dorsal spinal cord development (Fig. 3c<\/a>; Supplementary information, Table S3<\/a>).<\/p>\n The transition from a proliferative progenitor cell to a post-mitotic neuron is tightly regulated by different molecules. For instance, we found that SOX2<\/i>, VIM<\/i>, and HES5<\/i> are enriched in VZ region (Region 1), whereas ST18<\/i>, NHLH1<\/i>, ROBO3<\/i>, and INSM1<\/i> are spatially restricted to IZ (Region 2). In comparison, DCX<\/i>, CRABP1<\/i>, and ELAVL4<\/i> expressions are confined in the more laterally MZ (Region 3) (Fig. 3c, d<\/a>; Supplementary information, Table S3<\/a>). Among the genes enriched in Region 2 cells, NHLH1<\/i> and INSM1<\/i> are transiently expressed in late neuronal progenitors and nascent neurons,23<\/a>,24<\/a><\/sup> further indicating the nascent identity of Region 2 cells that have not yet fully differentiated and reached their final locations. Hence, by probing the specific gene expression in Region2, we identified a group of early markers for post-mitotic neurons in the IZ of the developing human spinal cord, including NHLH1<\/i>, INSM1<\/i>, ST18<\/i>, ROBO3<\/i>, KLHL35<\/i>, and more. Additionally, we identified genes expressed in later, more mature, post-mitotic cells in the lateral MZ (Region 3), such as CRABP1<\/i>, STMN2<\/i>, DCX<\/i>, and ZFHX3<\/i> (Fig. 3c, d<\/a>). Similar results were achieved in the ventral part of the developing spinal cord (Supplementary information, Fig. S3d, e<\/a> and Table S4<\/a>). However, the ventral proliferative VZ region is much thinner compared to its dorsal counterpart, potentially due to earlier neurogenesis in the ventral region (Fig. 3a<\/a>; Supplementary information, Fig. S3e<\/a>).<\/p>\n Overall, our TF-seqFISH dataset provided an invaluable resource for understanding the cellular and molecular basis underlying the neuroanatomical demarcation in the developing human spinal cord. Furthermore, our dataset provided a detailed understanding of the spatial and molecular programs underlying neuron generation, migration, and differentiation.<\/p>\n The spinal cord neurons can be further categorized based on their neurotransmitter properties, which play crucial roles in neuronal function and circuit connectivity. There are three main types of neurotransmitter-defined neurons in the spinal cord: glutamatergic Ex-INs, GABAergic\/glycinergic In-INs, and acetylcholinergic MNs. Through DEG analysis of the scRNA-seq dataset, we identified characteristic molecular profiles that distinguish these neuronal types, such as NEFM<\/i>, SLIT2<\/i>, SNCG<\/i>, ECEL1<\/i>, and NRP2<\/i> in MNs, LAMP5<\/i>, PAX2<\/i>, PAX8<\/i>, and ID2<\/i> in In-IN, and TLX3<\/i>, NRN1<\/i>, EBF3<\/i>, CACNA2D1<\/i>, and CRABP1<\/i> in Ex-IN (Fig. 3e<\/a>; Supplementary information, Table S5<\/a>). To further investigate the spatial distribution of these three neuronal types, we mapped the TF-seqFISH cells to the transcriptomic taxonomy and assigned the best-matched neuronal identities. The results showed that MNs exclusively reside in the ventral horn, while Ex-IN and In-IN are mainly distributed in the dorsal horn and the intermediate part (Fig. 3f<\/a>). Interestingly, at GW8, the Ex-INs are sandwiched between two spatially bifurcated groups of In-INs in the dorsal horn (Fig. 3f<\/a>). Notably, the In-INs located in close proximity to VZ show an intermingling pattern with some Ex-INs (Fig. 3f<\/a>). The spatial expression profiles of Ex-INs and In-INs enriched genes also reveal this distinctive organization pattern, which remains consistent across the entire spinal cord (Fig. 3f<\/a>; Supplementary information, Fig. S3f<\/a>). In the adult spinal cord, dorsal horn neurons are organized into distinct laminae that receive specific sensory inputs.25<\/a><\/sup> The sandwich-like organization of Ex-INs and In-INs in the dorsal horn of the human fetal spinal cord may be an intermediate state for the formation of appropriate circuit connections for perceiving environmental and internal signals.<\/p>\n Next, we aimed to uncover the molecular and cellular events underlying the development of the dorsal horn over a broad time range. To this end, we enriched our resources of spatial transcriptome data by conducting 10\u00d7\u2009Visium experiments on developing human spinal cord tissues at various gestational ages (GW8, GW9, GW13, and GW27). Using unsupervised clustering, we categorized 13 molecularly distinctive spot clusters based on molecular features and spatial location, including ventral gray cells, dorsal gray cells, intermediate gray cells, lateral corticospinal tract, dividing cells, ependyma, mesoderm, dorsal root ganglion (DRG) cells (merely captured in the tissues from GW8 and GW9), dorsal and ventral glial cells, and two groups of MNs (Fig. 4a, b<\/a>; Supplementary information, Fig. S4a\u2013d<\/a> and Table S6<\/a>). We found that a group of spots, mainly comprising Clusters 12 and 13 (outlined by red circles), was absent in the GW8 samples (Fig. 4c<\/a>), and this group of spots was located superficially in the dorsal horn when spatially visualized (Fig. 4d<\/a>). At GW9, GW13, and GW27, in addition to some yellow spots with identities of Cluster 8, we could also detect the spots of Clusters 12 and 13 in the dorsal horn (Fig. 4d<\/a>). However, at GW8, only spots of Cluster 8 could be detected in the dorsal horn while those of Clusters 12 and 13 were entirely missing (Fig. 4c, d<\/a>). Our results indicated that there were early-born neurons that emerged at or before GW8 (Cluster 8) as well as late-born neurons that arose after GW8 (Clusters 12 and 13). The emergence of late-born neurons at GW9 coincides with the expansion of the dorsal horn in the developing human spinal cord, indicating their importance in dorsal horn construction (Fig. 4d<\/a>).<\/p>\n a<\/b> A molecular-based classification of cells in the developing human spinal cord at various stages was performed using the dataset of 10\u00d7\u2009Visium. Unsupervised clustering was used to identify spots with a shared molecular profile, which were then annotated according to gene expression and spatial position (left). The spatial localization of these molecularly-defined cell groups was visualized in tissue slices at distinct timepoints (right). b<\/b> The UMAP visualization displays the expression profiles of marker genes utilized to annotate spot clusters. Each dot represents an individual spot, colored based on the expression level (red, high; gray, low). c<\/b> Distinct groups of spots derived from different embryonic stages were displayed separately. Notably, a group of spots primarily composed of Clusters 12 and 13, which were absent at GW8, were outlined with red circles. Furthermore, a group of spots mainly composed of Clusters 7 and 11, which were not detected at GW8 and GW9, were also identified and marked with blue circles. d<\/b> The H&E staining images, with red insets highlighting the dorsal horns of the spinal cord at GW8 and GW9, reveal a marked expansion of the dorsal horn from GW8 to GW9. Notably, only spots in cluster 8 are detectable in the dorsal horn at GW8, whereas at GW9, spots from Clusters 12 and 13 emerge and are situated in a superficial layer of the dorsal horn. e<\/b> A dotplot was generated to display the expression profiles of DEGs among Clusters 8, 12, and 13. The size of each dot in the plot is related to the detection rate, while the color bar is scaled with the average gene expression levels. f<\/b> Spatially-resolved transcriptomic expression features of Cluster 8-specific CRABP1<\/i>, as well as Clusters 12 and 13-specific DRGX<\/i> at different developmental stages, are visualized on the 10\u00d7\u2009Visium platform. These features are overlaid with H&E staining to illustrate the anatomical structures. The DRG regions are outlined. g<\/b> Immunostaining was performed on human spinal cord slices at GW8, 9, and 11 to examine the spatial expression features of DRGX and CRABP1. Scale bars, 100 \u03bcm. h<\/b> The neuronal identities of Clusters 8, 12, and 13 have been determined through an integrated analysis of the datasets from 10\u00d7\u2009Visium and scRNA-seq, and are depicted in a Sankey diagram. i<\/b> Schematic depiction of the cellular rules underlying the expansion and lamina formation of the dorsal horn during human spinal cord development.<\/p>\n<\/div>\n<\/div>\n Moreover, we performed a DEG analysis to probe the molecular characteristics of early- and late-born neurons in the dorsal horn. We found that the early-born neurons in Cluster 8 were specific for CRABP1<\/i>, RBP1<\/i>, ROBO3<\/i>, MEIS2<\/i>, and OTP<\/i>, whereas the late-born neurons in clusters 12 and 13 exhibited a distinctive molecular specificity (Fig. 4e<\/a>). Cluster 12 was distinctive for ENC1<\/i>, SST<\/i>, EBF3<\/i>, GRIA2<\/i>, and PCDH8<\/i>, while Cluster 13 was characteristic for MAF<\/i>, MAFA<\/i>, CCK<\/i>, and RORB<\/i> (Fig. 4e<\/a>). The specific enrichment of CRABP1<\/i> in Cluster 8 was also established by analyzing its expression across all distinct clusters (Supplementary information, Fig. S4e<\/a>). The modular architecture of the dorsal horn was further illustrated by the spatial expression profiles of CRABP1<\/i> and DRGX<\/i>, which symbolize early- and late-born neurons, respectively (Fig. 4f, g<\/a>). Notably, DRGX exhibited enriched expression specifically in the DRG cells but not in the spinal cord at GW8 (Fig. 4f<\/a>). We also identified the neuronal identities of Clusters 8, 12, and 13 by integrating scRNA-seq and 10\u00d7\u2009Visium datasets. The early-born neurons in Cluster 8 were mapped to the neuronal types of dI1, dI2, dI3, and dI4, while those of late-born Clusters 12 and 13 were mainly assigned to dI5 and dI6 (Fig. 4h<\/a>). Furthermore, our analysis of the scRNA-seq dataset for GW8 and GW9 revealed that the majority of dI5 and dI6 neurons originate from GW9, suggesting that these neurons possess a late-born identity (Supplementary information, Fig. S4f<\/a>). Intriguingly, aligning with the observations delineated in Fig. 1c\u2013g<\/a>, our investigation also revealed the tardy emergence of dorsal and ventral glial cells (Fig. 4c<\/a>; Supplementary information, Fig. S4g<\/a>).<\/p>\n Thus, by adopting a spatial perspective, we have conducted a comprehensive inquiry into the cellular and molecular mechanisms that underpin the development of the dorsal horn in the human spinal cord (Fig. 4i<\/a>). During the initial embryonic stages, we have unearthed the prototypical dorsoventral laminar configuration of neurons in the dorsal horn, which lays the groundwork for the constitution of somatosensory circuits. Employing a spatial transcriptomic dataset, we have not only identified early neurogenic events, but also the delayed ones in the dorsal horn of the developing human spinal cord, with the epoch of GW8 ascribing a pivotal temporal interval for the ontogeny of the dorsal horn.<\/p>\n Spinal motor neurons exhibit diversity in their morphology, connectivity, and functional characteristics.26<\/a>,27<\/a><\/sup> A notable aspect of the MN diversity is their categorization into distinct columnar groups, each of which occupies a specific rostrocaudal position within the spinal cord and innervates a unique array of peripheral target tissues. There are four key categories of MN columnar groups: LMC (Lateral Motor Column), which can be further divided into lateral (LMCl) and medial (LMCm) subcolumns; PGC (Visceral Preganglionic Column); HMC (Hypaxial Motor Column) and MMC (Median Motor Column) (Fig. 5a<\/a>).<\/p>\n<\/a><\/div>\n
Unraveling the spatial organization of neural progenitor cells with TF-seqFISH<\/h3>\n
<\/a><\/div>\n
Revealing spatial dynamics of neuronal differentiation and migration along the medial-lateral axis using TF-seqFISH<\/h3>\n
<\/a><\/div>\n
Spinal dorsal horn development in the developing human spinal cord<\/h3>\n
<\/a><\/div>\n
Molecular and spatial diversification of human MNs throughout spinal cord development<\/h3>\n