Molecular pathology of acute spinal cord injury in middle-aged mice

Abstract

The median age at which spinal cord injuries occur has steadily increased from 29 to 43 over the last several decades. Although more pre-clinical studies in aged rodents are being done to address this shift in demographics, comprehensive transcriptomic studies investigating SCI pathobiology in middle-aged mice are lacking. To address this gap in knowledge, we compared behavioral, histopathological, and transcriptional outcomes in young (2-4 months old) and middle-aged (10-12 months old) mice. In contrast to most previous studies, open field tests showed no differences in locomotor recovery between the young and middle-aged mice over a one-month period. The injury site also demonstrated similar histopathology in terms of lesion size, and numbers of macrophages and fibroblasts. Acutely after injury, proliferation of macrophages, fibroblasts, and astrocytes were also similar between the two age groups. In addition, spatial transcriptomics showed similar, transcriptionally defined regions around the injury site at 3 days post-injury. However, single cell RNA-sequencing of the cells at the injury site and surrounding spared tissue showed differences in select cell subpopulations. Taken together, our results indicate that although young and middle-aged mice display similar locomotor recovery and histopathology after SCI, changes in cell subpopulations may underlie a decline in repair mechanisms that manifest after middle age.

Keywords: Age as a biological variable; Aging; Immune response; Inflammation; Middle-age; Single-cell RNA sequencing; Spatial transcriptomics; Spinal cord injury.

Fig. 1

Young and middle-aged mice show similar locomotor recovery and lesion pathology at 4 weeks after SCI. (A) Basso Mouse Scale (BMS) scores and (B) subscores of young (n=10) and middle-aged (n=9) mice over a 4-week period post-spinal cord injury (SCI). Two-way repeated measures ANOVA with Bonferroni post-test (n=10 young and n=9 middle aged mice). Representative images of GFAP+ (gray C, D), CD11b (red F, G), and PDGFRβ (green, I, J) immunostaining in spinal cord lesion areas of young and middle-aged mice. (E) Quantification of the GFAP-negative lesion area did not show differences between the two groups. (n=10 young and n=6 middle aged mice) (H) Quantification of the density of CD11b+ myeloid cells in the lesion area did not show differences between the two groups. (n=10 young and n=5 middle aged mice) (K) Quantification of the density of PDGFRβ+ fibroblasts in the lesion area did not show differences between the two groups. (n=10 young and n=5 middle aged mice) Data are presented as mean ± SEM. Unpaired Student's t-test, each data point is a biological replicate. Scale bar = 200 µm.

Fig. 2

Young and middle-aged mice show similar cell proliferation and lesion pathology acutely after SCI. (A, B) Representative images of CD11b + myeloid cells (green) co-stained with EdU (magenta) in young (A) and middle-aged (B) mice at 4dpi. (C) Quantification of CD11b+/EdU + cells as a percentage of all CD11b + cells within the GFAP-negative lesion area. (n = 7 young and n = 8 middle-aged mice) (D, E) Representative images of PDGFRβ + fibroblasts (green) co-stained with EdU (magenta) in young (D) and middle-aged (E) mice at 4 dpi. (F) Quantification of PDGFRβ+/EdU + cells as a percentage of all PDGFRβ + cells within the GFAP-negative lesion area (n = 6 young and n = 7 middle-aged mice). (G, H) Representative images of GFAP + astrocytes (green) co-stained with Edu (magenta) in young (G) and middle-aged (H) mice at 4 dpi. (I) Quantification of GFAP+/Edu + cells as a percentage of all GFAP + cells within the perilesional region (250 µM region bordering the lesion) (n = 6 young and n = 7 middle-aged mice). (J, K) Density of CD11b + myeloid cells (J) or PDGFRβ + fibroblasts (K) within the lesion area in young and middle-aged mice (n = 7 young and n = 8 middle-aged mice). (L) Area of the GFAP-negative lesion in young and middle-aged mice (n = 6 young and n = 8 middle-aged mice). All counted cells were DAPI+ (nucleus in blue). Error bars = SEM. Each data point is a biological replicate. Unpaired Student’s t-test. Boxed region is enlarged in the adjacent image on the right. Scale bar = 400 μm for low magnification images, and 100 μm for high magnification images

Fig. 3

Spatial transcriptomics identifies distinct domains in and around the injury site that are similar between young and middle-aged mice. (A) 10X Visium spatial transcriptomic spots were clustered based on gene expression and mapped on the injury site to identify eight distinct spatial domains that were named based on a combination of marker genes, Gene Ontology terms, and spatial location. (B) Heatmap of top differentially expressed genes for each spatial domain. The color corresponds to the level of gene expression by z-score. (C) Top Gene Ontology Biological Process terms for each spatial layer ranked according to log-adjusted p-value. Dashed line represents the threshold for logP = 0.05. Bars are colored by the total number of differentially expressed genes in each term. (D) Volcano plots of differentially expressed genes between young (left side) and middle-aged (right side) mice for each spatial domain. Genes were filtered by avg_log2FC < 0.5 and p_val_adj > 1e-3. Genes that passed both thresholds are colored according to their spatial domains. n = 2 for young and n = 2 for middle-aged mice, with one n being a single section per-animal

Fig. 4

Single cell RNAseq analysis of the injury site from young and middle-aged mice show similar clustering and marker genes for general cell populations. (A) UMAP representing all cells from uninjured and 3 dpi spinal cords of young and middle-aged mice, colored by cell type. A total of 4 young uninjured, 3 middle-aged uninjured, 3 young SCI, and 3 middle-aged SCI biological replicates represent the combined dataset for a total of 71,794 cells, including 6,945 macrophages, 3,830 monocytes, 19,838 microglia, 12,940 endothelial cells, 1,304 tip cells, and 2,028 astrocytes (see Supp. Fig. 6 for full list). Cells are annotated using canonical markers and alignment with other single cell references using SingleR. UMAP of all cells colored by age (B) or injury condition (C). Cells are shuffled in depth to show the distribution and density in localization in the UMAP. (D) Dotplot of marker genes used to annotate the general cell types. Circle size corresponds to the percentage of cells in the group which express at least one count of the gene. The color of the circle corresponds to the level of gene expression by z-score. Each of the 13 biological replicates represents one unenriched and one astrocyte-enriched 8mm section of spinal cord centered on T8.

Fig. 5

Single cell RNA-seq analysis of myeloid subpopulations in young and middle-aged mice after acute SCI. (A) UMAP of leukocyte subpopulations present in uninjured and 3 dpi spinal cords of young and middle-aged mice. Subpopulations are annotated using canonical markers and alignment with other single cell references using SingleR. (B) UMAP of leukocyte subpopulations separated by age (B) or injury condition (C). Cells are shuffled in depth to visualize the distribution and density in localization in the UMAP. Boxplots of microglia (D), monocyte/macrophages (F), neutrophil (H) subpopulation proportions that are shown as a percentage of all myeloid cells per-sample. Percentages are individually calculated for each sample and each spot represents one biological replicate. Boxplots are split by cell subtype with independent Y axes. Cell type proportions were statistically compared with ANOVA and Tukey’s post-test using a threshold of p<0.05. Gene Ontology Biological Processes terms based on differentially expressed genes between the noted subpopulations of microglia (E), monocytes/macrophages (G), and neutrophils (I) within their respective age groups. Micro-B was compared to Micro-Homeo, Macro-A and Macro-B were compared to monocytes, and Neutro-Activated was compared to Neutro-Homeo all within their respective age groups. Terms which are unique to each age group within a cellular response are bolded. Circle size represents the odds ratio, and color represents the number of genes which map to each term. Volcano plots of DEG results are in supplemental figure 3. Abbreviations - Micro-Homeo, Homeostatic Microglia; Micro-A, Microglia-A; Micro-B, Microglia-B; Micro-Ifn, Interferon-associated microglia; Micro-Div, Dividing Microglia; BAM, Border-associated Macrophages; Macro-A, Macrophage-A; Macro-B, Macrophage-B; Neutro-Activated, Activated Neutrophils; Neutro-Homeo, Homeostatic Neutrophils

Fig. 6

Single cell RNA-seq analysis of glial subpopulations in young and middle-aged mice after acute SCI. (A) UMAP of glial subpopulations present in uninjured and 3 dpi spinal cords of young and middle-aged mice. Subpopulations are annotated using canonical markers and alignment with other single cell references using SingleR. (B) UMAP of glial subpopulations separated by age (B) or injury condition (C). Cells are shuffled in depth to visualize the distribution and density in localization in the UMAP. Boxplots of ependymal cell subpopulations (D) and their Gene Ontology Biological Processes based on DEG comparing combined ependymal cells A and B to the astroependymal A subpopulation (E) or the astroependymal B subpopulation (F). Boxplots of astrocyte subpopulations (G) and their Gene Ontology Biological Processes based on DEG comparing homeostatic and reactive astrocytes (H). Boxplots of oligodendrocyte progenitor cell (OPC) subpopulations (I) and their Gene Ontology Biological Processes based on DEG comparing OPC-a2 and OPC-b (J). Boxplot percentages are individually calculated for each sample and each spot represents one biological replicate. Boxplots are split by cell subtype with independent Y axes. Cell type proportions were statistically compared with ANOVA and Tukey’s post-test with p<0.05. GO terms are from comparisons of cell types within their respective age group. Terms which are unique to each age group within a cellular response are bolded. Circle size represents the odds ratio, and color represents the number of genes which map to each term. Volcano plots of DEG results are in supplemental figure 4. Abbreviations– Astroepen. Astroependymal; Astroepen-Div, Astroependymal-Dividing; Astrocyte-Homeo, Astrocyte-Homeostatic; Neuron-CSF, Cerebrospinalfluid-contacting Neurons; Neuron-SstExc, Sst positive and Excitatory Neurons; Neuron-Inhib, Inhibitory Neurons; OPC, Oligodendrocyte Precursor Cell; OPC-Div, Oligodendrocyte Precursor Cell- Dividing

Fig. 7

Single cell RNA-seq analysis of vascular subpopulations in young and middle-aged mice after acute SCI. (A) UMAP of vascular subpopulations present in uninjured and 3 dpi spinal cords of young and middle-aged mice. Subpopulations are annotated using canonical markers and alignment with other single cell references using SingleR. UMAP of vascular subpopulations separated by age (B) or injury condition (C). Cells are shuffled in depth to visualize the distribution and density in localization in the UMAP. Boxplots of endothelial subpopulations (D) and their Gene Ontology Biological Processes based on DEG comparing arterial uninjured and injured or capillary to tip cells (E). Boxplots of fibroblast subpopulations (F) and their Gene Ontology Biological Processes based on DEG comparing fibroblast-3 to fibroblast-2 or uninjured fibroblast-1 to injured fibroblast-1 (G). Boxplot percentages are individually calculated for each sample and each spot represents one biological replicate. Boxplots are split by cell subtype with independent Y axes. Cell type proportions were statistically compared with ANOVA and Tukey’s post-test using a threshold of p < 0.05. GO terms are from comparisons of cell types within their respective age group. Terms which are unique to each age group within a cellular response are bolded. Circle size represents the odds ratio, and color represents the number of genes which map to each term. Volcano plots of DEG results are in supplemental Figs. 5. Abbreviations– Endoth, Endothelial; Cap, Capillary; VSMC, Vascular Smooth Muscle Cell

Fig. 8

A direct comparison between young and middle-aged cell subpopulations. (A) heatmap of how many genes were differentially expressed in young (top left split) and middle-aged mice (bottom right split) for each timepoint (x axis) and cell subtype (y axis). Genes were obtained using DESeq2 and regressing out batch metadata. All genes passing padj < 0.05 are counted. Groups are not plotted or are greyed out when there were fewer than three samples with > 5 cells per group available for testing, or no DEGs passing in any group. (B) Volcano plots of selected groups from figure A. Horizontal line indicates padj < 0.05, vertical lines represent a log2FC +/- 0.5. All genes passing padj < 0.5 are plotted. The top 20 DEGs are labeled in each group. (C) Gene ontology terms which mapped to differentially expressed genes between ages for cell subtypes at each timepoint. All genes passing padj < 0.05 were processed per-group. Circle size represents the odds ratio, and color represents the number of genes which map to each term. Vertical line at X intercept represents padj = 0.05