Wildtype C57BL/6 male mice were purchased from Jackson Laboratories (Bar Harbor, ME) and were 3–4 months or 18–19 months of age at the time of injury. Mice were housed 5 per cage under 12:12 h light:dark cycles with food and water available ad libitum, and maintained at a temperature of 22°C–24°C and 40%–60% humidity. All procedures were conducted in accordance with protocols approved by the Georgetown University Animal Care and Use Committee. Experiments adhered to guidelines from the Guide for the Care and Use of Laboratory Animals, U.S. Department of Health and Human Services.

Controlled cortical impact (CCI) was conducted as previously described (Main et al., 2018; Korthas et al., 2022). Briefly, surgical anesthesia was induced using 4% isoflurane with maintenance at 2%, at flow rate of 1–1.5 L/min in freely breathing oxygen. The anesthetized animal was mounted in a stereotaxic frame with built-in heating bed to maintain body temperature at 37°C for the duration of the surgery. The surgical site over the top of the head was clipped and aseptically sterilized with alcohol swabs and topical iodine application. A 10 mm midline incision was made over the skull, the skin and fascia reflected and craniotomy performed (4 mm) on the central aspect of the left parietal bone. The impounder tip of a Leica StereoOne Impactor was sterilized, positioned to the surface of the exposed dura, and set to impact the cortical surface at 5.25-m/s velocity, 1.5 mm tissue deformation. Sham animals received isoflurane anesthesia, skin incision, and reflection, but no craniotomy or impact. After injury, the incision was closed with wound clips, anesthesia discontinued, 1 mL saline administered by intraperitoneal (i.p) injection and mouse placed in a heated cage to maintain normothermia for a 45 min recovery period. All animals were monitored carefully for 4 h after surgery and then daily.

In this study design, four experimental groups; Young Sham, Aged Sham, Young TBI and Aged TBI were assessed across two different time points of 7 days and 1 month post TBI. Experimentally, TBI or sham surgery was performed on n = 6 mice for each time points. Due to RNA yield from each tissue, dural meninges from 2 mice were combined to create 1 biological replicate at each time point (Figure 1A).

Mice were euthanized with CO2, followed by bilateral thoracotomy and transcardiac perfusion with ice-cold 1× PBS, using a peristaltic pump set to a flow rate of 5 mL/min. Following perfusion, skin and muscle were stripped from the outer skull. Skullcaps were removed via the ventral side of the jaw, with the lower jaw removed from the mandibular junction, before incisions dorsal to the ear canal, removal of the brain and collection of the skullcap containing the adherent meninges, as previously described (Louveau et al., 2018a). Skullcaps were immediately placed in a 10-mm petri dish with 1× PBS, and using a stereomicroscope (M80, Leica), meninges were scraped and removed intact (dura mater and arachnoid) using Dumont #5 forceps (Fine Science Tools). After meningeal collection from the skullcap, meninges were placed in 500 μL TRIzol Reagent (15596018, Life Technologies), immediately snap-frozen and stored at −80°C until further use.

For each of the four experimental groups (Young Sham, Aged Sham, Young TBI and Aged TBI), two meningeal samples were pooled to create one biological replicate. RNA was then extracted from each meningeal biological replicate using TRIzol® reagent (15596026, Invitrogen), as described in detail previously (Buenaventura et al., 2022). Concentration and purity of the RNA was assessed using a NanoDrop 1,000 spectrophotometer (ThermoScientific). RNA was aliquoted stored immediately at −80°C before use in RT-QPCR applications and high throughput RNA sequencing pipelines (Sloley et al., 2021).

The raw RNA-Seq reads (FASTQ files) were aligned to the mouse reference genome (mm10) using STAR (Dobin et al., 2013) and expression counts per transcript quantified using eXpress (Roberts and Pachter, 2013). From the raw counts per transcript, we used DESeq2 (Love et al., 2014) to normalize data and estimate differential expression between groups, and generate datasets containing contrasts. Briefly, from raw counts our DESeq2 pipeline estimated size factors (to account for differences in library depth) and gene-wise dispersions before shrinking these estimates to generate an accurate dispersion to model the counts. Then we fit the negative binomial model for differential expression testing to generate contrasts per region and log2 fold changes (LFC) for each sample group, before performing hypothesis testing using the Wald test. To improve power detection, outliers of genes whose mean of normalized counts was below the default threshold were removed using Cook’s distance cutoffs. To determine significance, the Wald test takes the LFC and divides it by its standard error, resulting in a z-statistic. This z-statistic is then compared to a standard normal distribution, in order to compute a p-value for each specific gene. To correct for multiple testing and control the false discovery rate (FDR), the Benjamini-Hochberg (BH) algorithm was used set to a threshold of 0.1. Results tables were generated for pairwise comparisons of interest.

Following DESeq2 analysis and generation of variance stabilized contrasts with statistically significant log2 fold changes, principal component analysis (PCA) and volcano plots and were generated from datasets using ggplot2, lattice and enhanced volcano R packages. Heatmaps were generated using the pheatmap R package, and Morpheus software.¹

The Gene Ontology (GO) Resource² was used for analysis of ontologies, biological functions and relationships (Ashburner et al., 2000; Gene Ontology, 2021). Enrichment analysis of genes was performed using the PANTHER (Protein ANalysis THrough Evolutionary Relationships) classification system (Mi et al., 2019). Using PANTHER (version 17.0), gene sets were uploaded for biological process (BP) analysis to identify hierarchical relations between over-represented or enriched functional classes. Fisher’s exact test followed by FDR correction determined statistically enriched categories. Dot plots were used to visualized this analysis, using ggplot2 R package.

CIBERSORTx was used to deconvolve meningeal bulk RNA-Seq data in silico against validated mouse (Mostafavi et al., 2016, Immgen GeoAccession: GSE75202) and human (Newman et al., 2015). reference profiles, which distinguish distinct immune cell subpopulations (Chen et al., 2018). Using CIBERSORTx impute cell fractions tool, bulk RNA-Seq data were inputted in CPM (counts per million) format and batch corrected under B-mode against the validated microarray matrix to reduce the impact of cross-platform variation (Newman et al., 2015; Steen et al., 2020). Bulk RNA-Seq data were averaged per sample phenotype, deconvolved separately against mouse and human immune cell populations, and presented in relative percentage format for each existing cell expression fraction.

Ingenuity Pathways Analysis (IPA, Qiagen), was performed to identify significant upstream regulators and canonical pathways (Kramer et al., 2014). Across our comparison groups of TBI and age, significantly expressed gene names and fold changes were submitted to compare expression patterns in our dataset to the IPA database. IPA results for canonical pathways and master regulators with p < 0.05 were considered significant.

The GWAS Catalog³ procured by the European Bioinformatic Institute and the National Human Genome Research Institute (Sollis et al., 2023), was used to identify genes associated with diseases (Neurodegenerative, Alzheimer’s, Parkinson’s) or molecular traits (inflammatory biomarker, memory, cognition). We then group wise compared significantly DEGs in our datasets with GWAS identified genes to assess translation of our results to known neurodegenerative processes. Visualization of these overlaps was performed using circus plots made with the R packages circlize and dichromat.

Statistical testing for RNAseq experiments is discussed in the relevant methods section. All sequencing analysis was performed using RStudio (v4.1.3) and GraphPad Prism (v9.5.0). The criteria for statistical significance were preset at p < 0.05 for analysis including IPA and gene ontology enrichment.

It is well established that age and aging related processes remain key factors that influence outcomes after TBI. Along with age, neuro-immune cascades from both central and peripheral mediators influence functional recovery timeframes. Importantly, these responses are time and context dependent, therefore we sought to investigate the temporal series of meningeal transcriptomic responses, and how age and post-injury interval effect these responses. Young and aged C57BL/6 wildtype male mice were subject to TBI, before assessment of the meningeal transcriptional response at 7 days post-injury. Hierarchical clustering of the aligned and normalized read counts indicated that age was the main factor driving separation of differential gene expression (Figure 1B). This was confirmed with principal component analysis (PCA) showing young and aged groups clustered furthest apart. While age was a clear separator, mice that received TBI were also spread, with Young TBI and Aged TBI mice having varied clusters at locations separated clear of each other (Figure 1C). Indeed, when we looked at the number of differentially expressed genes (DEGs) between experimental comparison groups, Young TBI mice have only 413 DEGs (211 upregulated, 202 downregulated), while Aged TBI mice display 695 DEGs (381 upregulated, 314 downregulated) at 7 days post-injury (Figure 1D; Supplementary Table S1). Interestingly, 2,392 differentially expressed genes (1,392 upregulated, 1,000 downregulated) were observed when comparing Aged Sham mice to Young Sham mice (Figure 1D; Supplementary Table S1). This effect was further confirmed via volcano plot visualization, showing that not only is there more significantly altered genes (red), but the overall level of transcriptomic variation is higher as a result of aging (Figure 1E). To gain an understanding of the type of immune cells affected by age and TBI, we deconvolved bulk RNAseq data using in silico CIBERSORTx software. At 7 days post-injury, meningeal macrophage responses are enhanced by TBI in young mice, a response which is dampened in aged TBI mice (Supplementary Figure 1; Supplementary Table S1). Comparison of DEGs in Young TBI and Aged TBI groups revealed 76 overlapping genes (Figure 1F). PANTHER ontology analysis showed these genes are involved in extracellular matrix (GO:0030198) and collagen organization (GO:0030199) biological processes, with significant fold enrichment (F.E 13.8–17.8) in these categories (Figure 1F; Supplementary Table S7). The extracellular matrix is further characterized by a trend of decreased subacute laminin expression, including lama, lamb, and lamc genes independent of age (Supplementary Table S7). Of the 76 overlapping genes, the top upregulated genes based on Log2 fold change are collagen related genes in Young TBI and Aged TBI (Figure 1G), further evidencing the potential extracellular matrix repair and resolution in the meninges early after injury.

To gain insights into mechanisms and pathways associated with the DEGs in our datasets, we performed Ingenuity Pathway Analysis (IPA) on young and aged comparison groups at 7 days post TBI. All raw outputs from IPA are included in Supplementary Table S2. IPA showed 71 unique genetic regulators modulated in Young TBI mice. Assessment of significantly activated upstream regulators (p < 0.05; absolute Z-Score > 2) showed lymphocyte and acute immune signatures in Young TBI mice, evidenced by IL2, IL5, CEBPB and ANGPT2 activation (Figure 2A). Enhancing the notion of upstream immune responses, cytokines (TGFB) and oxidative stress mediators (NOS2, AT4) were also activated in Young TBI mice. Alongside activation, a host of upstream mediators were inhibited (p < 0.05; absolute Z-Score > −2), including complement receptor Cr11 and miR-338-3p, miR-29b-3p, miR-124-3p which inhibit MAPK, JAK/STAT and NF-kB signaling (Figure 2A; Supplementary Table S8). At 7 days post injury, Aged TBI mice have 66 unique upstream regulators compared to Aged Sham mice. Aged TBI also have significant activation of upstream cytokines including TGFB, as well as the endosomal trafficking and risk factor for Alzheimer’s disease gene SORL1 (Figure 2B). The glucose metabolism gene PPARGC1A had the lowest Z-score, demonstrating inhibition upstream, followed by the inhibition of alpha catenin family which plays a key role in stabilizing adherens junctions and sealing membranes in the maintenance of Blood Brain Barrier (BBB) integrity (Figure 2B). Comparison upstream regulators in both Young and Aged TBI groups showed 38 (21.7%) shared regulators, with these mediators having ontology profiles that evidence their role in apoptotic processes (GO:0006925) and T lymphocyte driven immune processes (GO:0072540; Figure 2C; Supplementary Table S2). As aging profoundly affected meningeal gene expression (Figure 1), we also assessed the influence of age on upstream regulators and canonical pathways. Notably, age alone had 181 unique upstream regulators, which accounts for over 50% of the total regulators when all three groups are compared (Figure 2D). Key upstream master regulators that are activated due to aging include type-1 and type-2 interferons (IFNG; Supplementary Table S8), regulators of NF-kB signaling (HGF) and T cells (ESR1, CSF2), as well as transcriptional regulators of cell proliferation cascades (ERG, ERBB2, RICTOR, CD23, TCF7L2; Figure 2E). Similarly, when assessing significant canonical pathways modulated by aging, amplified inflammatory cascades are observed, including NF-kB signaling, PKR interferon induction, B cell and iNOS signaling (Figure 2F). In addition, oxidative phosphorylation, PTEN signaling, and PPARα/RXRα pathways were inhibited canonical pathway, indicating depleted metabolic regulation in aged mice (Figure 2F).

Young and aged CB57BL/6 wildtype male mice were subject to TBI, before assessment of the meningeal transcriptional response at 1 month post injury (Supplementary Table S3). Heatmap clustering of significantly DEGs at showed similar levels of genetic expression in both Aged TBI (112 upregulated, 159 downregulated) and Young TBI (220 upregulated, 169 downregulated) groups. Similar to 7dpi, the most amount of DEGs was observed as a factor of aging alone (264 upregulated, 239 downregulated; Figure 3A; Supplementary Table S3). This was further evidenced by volcano plot analysis, which also showed that the highest Log2 FC genes in all groups belonged to antibody production, with Ighv6-6, Igkv4-86, and Igkv6-20 the highest significantly upregulated gene in Young TBI, Aged TBI, and Aged Sham groups, respectively (Figure 3B). In silico deconvolution using CIBERSORTx, showed that at this chronic timepoint, young TBI mice still contain elevated macrophage contributions compared to Aged TBI mice. It also revealed a unique contribution of monocytes and plasma cells in Aged TBI compared to Young TBI (Supplementary Figure 1; Supplementary Table S3). Assessment of gene ontology (GO) biological processes (BP) showed that Young TBI mice display elevated activation of extracellular matrix processes, with matrix structure (GO:0030198) and collagen fibril organization (GO:0030199) significantly enriched (Figure 3C; Supplementary Table S7). Aged TBI mice display enhanced inflammatory based responses with immune response, immunoglobulin production (GO:0002377) and production of molecular mediators in immune responses (GO:0002440) significantly enriched (Figure 3D). This immune phenotype after TBI may be specific to Aged TBI mice, with assessment of Aged TBI v Young TBI revealing that the Aged TBI differentiates through enriched immune responses in cytokine production (GO:0006955), T cell/lymphocyte proliferation (GO:0042102), interferon-gamma (GO:0032729) and immunoglobulin production (GO:0002377; Figure 3E; Supplementary Tables S5, S8). Ingenuity Pathway Analysis (IPA) of upstream regulators (Supplementary Table S4) in Young TBI mice at 1 month post injury showed activation of mediators involved in immune suppression (FOXC1), cytokine expression [ZEB1, AKT, SOX9, SORL1, and monocyte recruitment (CCR2)], while inhibition of mediators (IL1B, IKBKG) involved in pro-inflammatory responses (Figure 3F). Meanwhile, Aged TBI mice display activated upstream regulators including immunoglobulins, cytokines (IL33) and T cell receptors (CD3), with inhibited regulators including IRF2BP2, SOX2, and WNT pathway transcription factor (TCF7L2) at 1 month post injury (Figure 3G). This enhanced inflammatory response is confirmed when assessing upstream regulators that are unique to Aged TBI vs. Young TBI. Specifically, a robust interferon master regulator signature is observed upstream, with activation in the interferon alpha family, IFNB1 (interferon beta), IFNG (interferon gamma) and IRF7 (interferon regulatory factor 7; Figure 3H; Supplementary Table S8). In contrast ani-inflammatory mediators including (TFGB, EGF) are uniquely inhibited in Aged TBI mice 1 month post injury (Figure 3H).

Given aging itself resulted in substantially different gene expression patterns (Figure 4A), we decided to look more closely at these differences within the bulk RNA-seq dataset. To gain insights into the molecular pathways associated with age related changes in gene expression, we performed gene ontology (GO) enrichment. Analysis of biological process (BP) categories revealed that age alone resulted in elevated inflammatory signatures, including immunoglobulin production (GO:0002377), immune system process (GO:0002376) and regulation of B cell activation (GO:0050864; Figure 4B; Supplementary Table S5), compared to Young Sham mice. The genetic components related to each of these processes were further profiled and show a significant pattern of separation in overall adaptive immune upregulation in aged mice (Figures 4C–H). Key immunoglobulin clusters are highly active (Figure 4C), with regulators like Igkv8-18 and Igkv3-10 presenting 8-fold upregulation (Figure 4D). The constant secretion of antigen processes indicates that the meninges in aged mice have a baseline adaptive immunity that may be constantly activated for defense. Adaptive immune responses are also populated by T cell mediators (Cd8b1, Cd247, Cd3g), leukocytes (Lax1), and immunoglobulin production (Ighg2c, Igha, Ighv6-6; Figure 4E). It is evident that the genetic expression of prominent adaptive immune response activity by Ighg2c, Jchain, Igha, and Ighg1 in aged mice is significantly upregulated compared to young mice (Figure 4F). Aged lymphocyte activation is characterized by increased expression of Lax1, Lck, Igkc, Card11, Slamf6, Cd27, Ighg1, Tbx21, Mpzl2, Tnfsf8, Hdac9, Ikzf3, Cd3g (Figure 4G) and aged B cell activation is regulated by an increase of Ighg2c, Ughv6-6, Igha, Cdkn2a, Ugkc, Card11, Ighv1-39, Cd27, Tbx21, Tnsfs13b, and Ikzf3 (Figure 4H). These transcriptomic profiles validate the chronic upregulation of the adaptive immune response and provide context for the immunological status of the meninges in aging.

Transcriptomic analyses between age and TBI at sub-acute and chronic time points reveal upregulated inflammatory activity in the meninges. To further understand how the inflammatory signature in aging and TBI is related to functional consequences, we evaluated meningeal genomic data through a human pathophysiology context. Genome-wide association study (GWAS) analysis of common genes with high activity in the aging brain show significant overlap in a number of functional traits including behavior, neurodegeneration, inflammation, cognition, memory, anxiety, and lymphatic system disease (Figures 5A,B; Supplementary Table S6). To further investigate the inflammatory biomarkers from TBI, we compared the transcriptomic profiles of each time point and assessed for similarities. The aged brain has a unique inflammatory profile at both time points (67% at 7dpi and 64.7% at 1mpi), which provides evidence of baseline inflammation with aging independently (Figures 5C,D; Supplementary Table S6). The differential expression of inflammatory biomarkers in TBI with age also show variant profiles at the subacute (10.2% in Young TBI and 10.2% in Aged TBI) and chronic (13.8% in Young TBI and 15.4% in Aged TBI) time points (Figures 5C,D). Acutely, we found 13 genes in common between Aged Sham and Aged TBI mice (RUNX1, CUL4A, KMT2A, ANGPTL4, RCN3, FAM114A2, PPP5C, CDK6, PDZRN4, RASA2, ZC3H7B, SP140, TRIOBP), 9 genes in common between Aged Sham and Young TBI mice (DDAH1, IQGAP1, TRIB1, ADCY5, TMEM106B, NUCKS1, RCN1, PLA2G5, ETS2), and 4 genes in common between both TBI groups (ATP2A3, COL18A1, SPTBN1, VMP1; Figure 5C). Chronically, we found 1 common gene between Aged Sham and Aged TBI mice (PIM3) and 3 similar genes between Aged Sham and Young TBI mice (YWHAZ, BANP, TMEM18; Figure 5D). These transcriptomic differences reveal unique inflammatory responses with age and TBI evidenced by specific genomic trait profiles per condition. We then assessed the differential trait interaction patterns with age as a factor of TBI (Figures 5E,F; Supplementary Table S6). Compared to Young TBI (Figure 5E), Aged TBI (Figure 5F) appears to have increased dynamic connectivity in relation to behavioral genomics. Acutely, Young TBI reveal behavior-related genes SLC4A10, NAV1, TRIB1, CUL9, RERE, EXT1, PDGFD, DDX18, KIF5A are linked to inflammation, neurodegeneration, sleep duration, sleep performance, and anxiety (Figure 5E). Cognitive disability is also related to behavior and inflammation through SPRED2 and MYOC. In comparison, Aged TBI have more prominent genomic associated overlaps, where behavior-related genes CD36, CDK6, RASA2, ANGPTL4, PCSK5, SNX13, LAMA2, ARNTL, ACE, PCDH7, NTRK3, CDKAL1, FOXN3, EFNA, and PDE4B are linked to inflammation and neurodegeneration. Memory deficits are related to altered behavior, inflammation, and neurodegeneration through ARNTL, SORCS2, ETV1, MAP2K5, BCAM, PSD5B, IFITIM10, and NCAPD3, while cognitive disability is linked to neurodegeneration, behavior, and lymphatics through ID2, G2E3, GBA, IRF2, MYOC, and PSD5B (Figure 5F). While TBI influences meningeal transcriptomic alterations, Aged TBI presents a progressively active interaction network between genes associated with inflammation and potential neurodegenerative deficits.

As the effects of aging and TBI are known to increase susceptibility to other neurodegenerative disorders, we compared meningeal signatures to the human pathophysiology of known chronic neurobiological diseases at 7 days and 1 month post injury (Figures 6A,B; Supplementary Table S6). With Alzheimer’s disease as a common risk factor in TBI, we further extrapolated genomic profiles at subacute and chronic time points with age. We found that aged mice present unique profiles (66.3% at 7dpi and 54.3% at 1 month) which indicate high activity of Alzheimer’s related genes (Figures 6C,D). Transcriptomic expression related to Alzheimer’s with TBI also show differences with age at the subacute (4.1% in Young TBI and 14.3% in Aged TBI) and chronic (21.7% in Young TBI and 15.2% in Aged TBI) time points (Figures 6C,D; Supplementary Table S6). At 7 days post injury, we found 8 common genes between Aged Sham and Aged TBI, 4 common genes between Aged Sham and Young TBI, and 3 common genes between Aged TBI and Young TBI (Figure 6C). At 1 month post injury, we found 3 common genes between Aged Sham and Aged TBI, 1 common gene between Aged Sham and Young TBI, and no similar genes between Aged TBI and Young TBI (Figure 6D). Overall, Aged Sham, Aged TBI, and Young TBI present unique transcriptomic expression of disease relative to each condition. The TBI signature is consistent with other findings of upregulated CACNA1G with brain swelling in response to head injury. COL25A appears to indicate membrane-associated brain-specific collagen repair, which is unsurprising given the active extracellular matrix affected by injury. Extracellular matrix repair patterns appear to present an increase of collagens and laminins at 7 days post-injury and a dampened response 1 month of injury independent of age. Further, ELL appears to have a role in facilitating RNA Polymerase II elongation within epidermal proliferation, likely a result of the transcriptional regulation necessary within the extracellular matrix after injury. To determine how neurodegenerative disease is affected by TBI in the aged brain, we visualized the interactivity of disease genomics at 1 month post injury (Figures 6E,F; Supplementary Table S6). We identified that aging has significant overlap in a number of diseases including Alzheimer’s, ALS, stroke, Parkinson’s, dementia, epilepsy, and depression (Figure 6E). At the chronic time point in Young TBI 1 month post injury, several prominent genes like CACNA1G, SENP7, ETF1, EGR1, KDM3B, CCR2, GRAMD1A, ARAP2, GOLIM4, THSD4, ADAMTS1, CELF1, SUCLG2, TOP1, NR3C2, DDX18, CDC5L, and AP2A2 that are related to Alzheimer’s are also linked to Parkinson’s, epilepsy, and behavior (Figure 6E). Interestingly, GWAS associations in Aged TBI 1 month post injury reveals a significantly altered genomic network pattern, where Alzheimer’s related genes FOS8, MPZL1, LIMS2, ELL, EXOC3L2, STK11, TOP1, CDC5L, ABCA1, and KANSL1 are linked to neurodegeneration, inflammation, Parkinson’s, stroke and dementia. Dementia-related genes ARHGAP27, CYP1B1, KANSL1, KRTCAP2, LSM7, and GBA are also linked to altered behavior, neurodegeneration, inflammation, and Parkinson’s (Figure 6F). Notably, TBI appears to increase neurodegenerative disease interactions chronically (Figures 6E,F), which may indicate an amplified disease-state environment with aging. The neurodegenerative interaction patterns in Aged TBI (Figure 6F) are more sporadic and robust compared to Aged Sham (Figure 6E). While aging alone presents genomic associations with neurodegeneration, TBI-induced chronic inflammation in the aged brain presents further exacerbation of neurodegenerative consequences.