Characterizing progenitor cells in developing and injured spinal cord: Insights from single-nucleus transcriptomics and lineage tracing

Abstract

Various mature tissue-resident cells exhibit progenitor characteristics following injury. However, the existence of endogenous stem cells with multiple lineage potentials in the adult spinal cord remains a compelling area of research. In this study, we present a cross-species investigation that extends from development to injury. We used single-nucleus transcriptomic sequencing and genetic lineage tracing to characterize neural cells in the spinal cord. Our findings show that ciliated ependymal cells lose neural progenitor gene signatures and proliferation ability following the differentiation of NPCs within the ventricular zone. By combining single-nucleus transcriptome datasets from the rhesus macaque spinal cord injury (SCI) model with developmental human spinal cord datasets, we revealed that ciliated ependymal cells respond minimally to injury and cannot revert to a developmental progenitor state. Intriguingly, we observed astrocytes transdifferentiating into mature oligodendrocytes postinjury through lineage tracing experiments. Further analysis identifies an intermediate-state glial cell population expressing both astrocyte and oligodendrocyte feature genes in adult spinal cords. The transition ratio from astrocytes into oligodendrocytes increased after remodeling injury microenvironment by functional scaffolds. Overall, our results highlight the remarkable multilineage potential of astrocytes in the adult spinal cord.

Keywords: astrocyte; lineage tracing; single-nucleus RNA sequencing; spinal cord injury; transdifferentiation.

Fig. 1.

Dynamic alterations of the human spinal cord NPCs and ciliated ependymal cells within the ventricular zone during development. (A) UMAP visualization of SOX2 expression glial cells integrated from datasets of the human spinal cord samples at different stages showing neural progenitor cells differentiation tendency. The population is categorized into SOX2 progenitor cells, ciliated ependymal cells, ALDH1L1-astrocytes, and oligodendrocytes, distinguished by their gene profiles. (B) Violin plot depicting the expression level of key marker genes for SOX2 progenitor cells (SOX2, MKI67, SOX9, NES, and EGFR), ciliated ependymal cells (DNAAF1, SHH, and GDF7), astrocytes (FAM189A2, ALDH1L1, and PAX3), oligodendrocyte lineage (OLIG2 and SOX10), and neuronal progenitor cells (DCX) in each cell clusters. (C) Split UMAP plots illustrating changes in cell composition across various development stages for SOX2-progenitor, ciliated ependymal, ALDH1L1-astrocyte, and oligodendrocyte populations identified in (A). Ciliated ependymal cells were not observed in the adult human spinal cord dataset due to their low proportion. (D) Immunofluorescence images of the spinal cord ventricular zone at different development stages for SOX2 (green), SOX9 (red), SOX10 (gray), and nuclei (DAPI, blue). (Scale bar, 50 μm.) (E) Bar graph showing the proportional changes of proliferation cells in the human spinal cord ventricular zone across development stages. A significant decrease in proliferation rate within the ventricular zone was observed between GW10 and GW14. N = 3 per group. Error bars represent mean ± SEM; ***P value < 0.001. (F) Immunofluorescence images of the human spinal cord at GW10 and GW26 for ciliated ependymal cell marker DNAAF1 (red), and nuclei (DAPI, blue), focused on the ventricular zone. (Scale bar, 50 μm.) (G) Spatial distribution diagram of the glial population (SOX2 progenitor cells, ciliated ependymal cells, ALDH1L1-astrocytes, and oligodendrocytes) in the ventricular zone of the adult spinal cord.

Fig. 2.

Absence of ciliated ependymal cells in ex vivo–cultured glial cells from the human fetal spinal cord. (A) UMAP plot illustrating the clustering of in vitro expanded glial cells from GW13 human spinal cord based on their gene expression profiles: proliferation cells, astrocyte lineage, oligodendrocyte lineage, and fibroblast. (B) Clusters are shown with proliferative (MKI67 and PIF1) potential related to astrocyte (SOX9 and GFAP) and oligodendrocyte (SOX10 and OLIG2) lineages, but no ciliated ependymal cells (DNAAF1 and LRRIQ1) were observed. The color intensity indicates the average expression level, and the size of the dot indicates the percentage of cells expressing the gene in each cluster. (C) Bar graph illustrating that the proportion of GFAP, OLIG2, and DNAAF1-expressing cells in the expanded culture was 43.92 ± 6.39%, 5.20 ± 1.08%, and 0%, respectively. N = 3 per group. (D) Immunofluorescence images displaying the expression of GFAP (green), OLIG2 (red), and DNAAF1 (gray) in the progeny from in vitro expanded GW13 spinal cord tissue. (Scale bar, 25 μm.) (E) Immunofluorescence staining of GFAP (green), OLIG2 (red), and FOXJ1 (gray) from in vitro expanded GW13 human spinal cord tissue in 3D method. (Scale bar, 25 μm.) (F) Immunofluorescence staining of GFAP (green), OLIG2 (red), and FOXJ1 (gray) from in vitro expanded e15.5 mice spinal cord tissue in 3D method. (Scale bar, 25 μm.)

Fig. 3.

Integrated single-cell RNA-seq analysis of developmental and injured datasets reveals the limited activation of ciliated ependymal cells following SCI in primates and rodents. (A) Schematic diagram illustrating the UMAP plot and spatial distribution of ciliated ependymal cells. (B) UMAP visualization depicting the diverse cell populations from selected human and rhesus macaque datasets. The red arrow indicates ciliated ependymal cell cluster. (C) Feature plots showing that the ciliated ependymal cell population (DNAAF1) does not express the proliferation gene MKI67. (D) Gene set mRNA analysis revealing interest signature genes expression in ciliated ependymal cells cluster between development and injury, including immune signatures, M phase, S phase, G2 phase, neural progenitor, and ciliated cell feature genes. (E) Heatmap and dot plot summarizing the dynamic changes in gene expression relevant to cell type signatures over developmental stages and post-SCI conditions, highlighting the intensity and prevalence of specific marker genes. In the dot plot, color represents average expression level, and percentage represents cell proportion. (F) Integration of snRNA-seq datasets from rat and rhesus macaque. The ciliated ependymal population is indicated by the red arrow. (G) Heatmap showing injury-induced transcriptomic changes within the ciliated ependymal population of rat and rhesus macaque. Changes in NES, VIM, and VWA3A expression (red arrow) induced by spinal cord injury (SCI) differ between species. No observable changes in expression of the feature genes related to neuronal, astrocyte, or oligodendrocyte lineages are observed after injury in both species.

Fig. 4.

SOX9-traced glial cells transdifferentiate into oligodendrocytes after SCI. (A) Schematic diagram illustrating the UMAP plot and spatial distribution of SOX9-expressing glial cells. (B) Immunofluorescence images depicting the distribution of SOX9-traced cells (green) in the dorsal, central, ventral, and T7/8 segments of the spinal cord. (Scale bar, 100 μm.) (C) Immunofluorescence staining demonstrates the expression of tdTomato (red), BrdU (green), and GFAP (gray) in transgenic mice at uninjured and 14 dpi. (Scale bar, 100 μm.) (D) Bar graph showing a significant increase of the proportion of SOX9-traced cells expressed APC (green) at uninjured, 14 dpi, and 56 dpi (0%, 1.63% ± 0.05%, and 6.17% ± 0.22%, respectively). N = 6 per group. *P value < 0.05, **P value < 0.01, and ***P value < 0.001. (E) Immunofluorescence images of the spinal cord revealing APC-expressing cells (green) derived from SOX9-traced cells (red) at 7 dpi and 14 dpi. Panels a and b show higher magnification views from the boxed areas in the merged images, with white arrows indicating cells coexpressing APC and tdTomato. (Scale bar, 50 μm.)

Fig. 5.

Lack of transdifferentiation of NESTIN-expressing glial cells into oligodendrocytes after SCI in NES/CreERT2-tdTomato transgenic mice. (A) Schematic diagram depicting the UMAP plot and spatial distribution of NESTIN-expressing glial cells. (B) Immunofluorescence staining showing single-channel and merged images of NES/CreERT2-tdTomato specifically labeled a subset of SOX9-expressing glial cells but not oligodendrocytes (APC) in uninjured transgenic mice. (Scale bar, 50 μm.) (C) Immunofluorescence staining images demonstrating minimal contribution of NES/CreERT2-tdTomato traced glial cells to scar formation (SOX9), the oligodendrocyte lineage (APC), or neuronal lineage (TUJ1) cells. (Scale bar, 50 μm.) (D) Overview and high-resolution images of the spinal cord at 14 dpi stained for SOX9, tdTomato, and GFAP. Panels a1-a4 and b1-b4 display magnified views from boxed areas in the larger context image, showing cellular details of marker expression and colocalization. (Scale bar, 50 μm.)

Fig. 6.

ALDH1L1-traced astrocytes transdifferentiated into oligodendrocytes after SCI. (A) Schematic diagram illustrating the UMAP plot and spatial distribution of ALDH1L1-astrocytes. (B) Bar graph illustrating the proportion of ALDH1L1 traced cells expressing GFAP, SOX9, and S100β in the spinal cord of uninjured transgenic mice. N = 3 per group. (C) Bar graph showing the proportion of ALDH1L1traced cells expressing APC in the spinal cord before and after SCI. A significant increase in the APC-positive cell proportion was observed at 14 dpi and 56 dpi. N = 3 per group. ***P value < 0.001. (D) Immunofluorescence images displaying colocalization of APC (green) and tdTomato (red) at various stages. White arrows indicate cells coexpressing both markers. These images demonstrate increasing transdifferentiation of astrocytes into oligodendrocytes. (Scale bar, 50 μm.) (E) Detailed images of the spinal cord sections labeled for APC (green), tdTomato (red), and MAG (gray) at different injury stages. Arrows indicate APC (green) and MAG (gray) coexpressing cells derived from ALDH1L1-traced astrocytes (red). (Scale bar, 25 μm.) (F) Immunofluorescence images showed colocalization of ALDH1L1-tdT (red) and MBP (gray) (arrow), axons stained against neurofilament (green). (Scale bar, 25 μm.) (G) Bar graph showing the proportion of ALDH1L1-traced cells expressing MAG in the spinal cord before and after SCI. A significant increase in the APC-positive cell proportion was observed at 56 dpi. N = 5 per group. *P value < 0.05, ***P value < 0.001.

Fig. 7.

Integrated single-cell RNA-seq data from development and injured primate spinal cord datasets revealed an intermediate-state subpopulation of spinal cord astrocytes in primates. (A) Violin plots depicting the expression of feature genes of the astrocyte lineage (FAM189A2, SOX9, and EGFR) and oligodendrocyte lineage (PDGFRA, SOX10, and TCF7L2) in the intermediate-state astrocyte cluster (cluster 22). (B) Immunofluorescence images show coexpression of SOX9 (red), SOX10 (green), and GFAP (gray) coexpressed cell in the rhesus macaque spinal cord at 7 dpi. SOX9-SOX10 coexpressing cells were observed in the central glial substance. Subpanels a1-a3 provide higher magnification views of the boxed area in panel a. Arrows indicate SOX9-SOX10 coexpressing cells. (Scale bar, 100 μm.) (C) Bar graph displaying the percentage of SOX9-SOX10 coexpressing cells among SOX9-positive cells in the rhesus macaque SCI model at different timepoints. SOX9-SOX10 coexpressing cells were observed in the adult spinal cord under both normal and injured conditions without significant changes. N = 3 per group. (D) Single-cell trajectory analysis of astrocytes (clusters 0, 14, 18, and 22), OPCs (clusters 8 and 21), maturing oligodendrocytes (cluster 20), and myelination oligodendrocyte (cluster 3) population using Monocle. (E) Velocities derived from astrocytes and oligodendrocytes are visualized as streamlines in a UMAP-based embedding. The arrow indicates the intermediate-state astrocyte. The velocity arrow is attached to the position for which the velocity has been calculated. The arrowhead points to the future state. (F) Immunofluorescence staining images showing coexpressing of SOX10 (green) and SOX9 (red) in tdTomato (red) traced cells in ALDH1L1-tdTomato transgenic mice at 14 dpi within the injury area. Subpanels b1-b3 provide higher magnification views of the boxed area in panel b. Arrows indicate SOX9-SOX10 coexpressing cells. (Scale bar, 50 μm.) (G) Expression of Tnfrsf19 could be found in the intermediate-state astrocyte (cluster 22) cluster in primate datasets.

Fig. 8.

Transplantation of “CS-CTBNG” scaffolds promotes the transdifferentiation of ALDH1L1-expressing astrocytes into oligodendrocytes in mice after SCI. (A) Schematic of the experimental timeline showing the administration of tamoxifen in ALDH1L1 -tdTomato mice to trace astrocytes, followed by SCI. (B) Bar chart depicting the proportion of ALDH1L1-traced cells expressing APC in uninjured, 14 dpi, and 56 dpi, with and without the addition treatment of a collagen scaffold. The results indicate that scaffolds containing bioactive molecules accelerated the transdifferentiation efficiency. N = 3 per group. *P < 0.05, **P < 0.01, and ***P < 0.001, ns: not significant. (C and D) Single channel and merged immunofluorescence staining showing expression of APC in transgenic mice at 14 dpi and 56 dpi, under SCI and SCI+CS-CTBNG treatment conditions. Arrows indicated APC-expressing cells (green) derived from ALDH1L1-traced astrocytes (red). (Scale bar 50 μm.)