Shared transcriptomic signatures in perilesional and contralesional cortex after ischemic stroke

Abstract

Stroke induces a transient period of heightened plasticity during which functional recovery is most pronounced. Experimental models have identified repair-associated processes in both the ipsilesional and contralesional cortex, indicating that stroke recovery involves regions both remote and near the lesion. However, most transcriptional studies have focused on the infarct core and peri-lesional cortex (PLC), leaving it unclear whether comparable molecular responses occur in the contralesional cortex (CLC), a region that undergoes substantial remodeling in the absence of direct tissue injury, necrosis, or widespread cellular infiltration. In addition, potential sex-dependent differences in these responses remain incompletely defined, despite known influences of biological sex on post-stroke inflammation and vascular remodeling. To address these gaps, we performed bulk RNA-sequencing of the PLC and CLC at 7 days after photothrombotic stroke, a subacute time point associated with the initiation of repair, in male and female mice. Despite distinct positions relative to the lesion, both regions exhibited robust upregulation of inflammatory signaling, including cytokine-, astrocyte-, and myeloid-lineage-associated pathways. The CLC did not demonstrate a distinct region-specific transcriptional profile; instead, shared signatures between PLC and CLC included genes strongly associated with reactive microglial phenotypes. This shared neuroinflammatory response was largely conserved across sexes. Consistent with these findings, male and female mice exhibited comparable corticospinal tract axonal sprouting originating from the CLC at 6 weeks post-stroke. Together, these findings support a shared neuroinflammatory transcriptional response as a prominent early feature of cortical regions associated with post-stroke plasticity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12974-026-03885-1.

Background

Stroke patients present with a wide range of deficits, often followed by a remarkable period of improvement that spans the first 3–6 months after stroke, known as the critical window [8, 37]. During this time, the brain exhibits a heightened capacity for neuronal plasticity that supports functional reorganization. However, most individuals experience a plateau in recovery and are left with chronic deficits that reduce independence and quality of life [29, 32]. Investigations into the mechanisms underlying this critical window may help drive the discovery of new therapies for stroke recovery. Preclinical models of stroke have been instrumental in advancing our understanding of these mechanisms. Rodent models replicate key aspects of human stroke recovery and are highly amenable to molecular and genetic manipulation, making them ideal for dissecting the biological processes that underlie post-stroke plasticity. A consistent finding across models is that recovery depends heavily on the capacity of spared neural circuits to rewire and compensate for damaged networks, which occurs through numerous forms of plasticity, including dendritic arborization, synaptic tuning, and axonal sprouting [16, 25, 69, 99, 112]. This is especially evident in the context of focal motor cortical strokes, which elicit robust plasticity in both local and remote regions [21, 45, 81].

Among regions undergoing post-stroke plasticity, the perilesional cortex (PLC) and contralesional cortex (CLC) are the most well characterized. In the PLC, new connections formed within spared regions, including somatosensory and premotor cortex, directly contribute to recovery [16, 60, 61, 74, 79]. These structural changes are accompanied by well-defined transcriptional responses, which have provided key insight into both cell-intrinsic and environmental drivers of stroke repair. The CLC likewise undergoes robust plasticity and contributes to functional reorganization, although its effects on recovery are more variable and depend on lesion characteristics, timing, and behavioral context [2, 4, 11, 17, 40, 86]. In some settings, particularly after larger infarcts or complete CST injury, the CLC appears to support recovery [12, 99], whereas in others it is neutral or may even interfere with functional improvement [10, 69]. While gene expression changes in the CLC have also been linked to plasticity and recovery [36, 100], its molecular characterization remains comparatively limited.

Although the PLC and CLC exhibit broadly similar responses to stroke, relatively few studies have directly compared their transcriptomes, highlighting an opportunity to better understand regional plasticity programs [18, 33, 42, 49]. The PLC lies adjacent to the infarct core and glial scar and is exposed to pronounced inflammatory and injury-related signaling, conditions thought to create a permissive niche for structural remodeling [20], [68]. In contrast, CLC is anatomically distant from the lesion and does not experience direct ischemic injury or the same degree of damage-associated cues. How a remote region engages a similarly robust plasticity program under these conditions remains unclear. Direct comparison of the molecular landscape across these regions therefore provides a unique opportunity to distinguish transcriptional responses that are common to post-stroke plasticity from those shaped by local environment, offering new insight into the mechanisms that govern region-specific remodeling after injury.

Sex has been extensively investigated in stroke epidemiology and early injury responses [72, 84], yet its role in recovery and the molecular programs that govern repair remain incompletely understood. Studies in both humans and mice demonstrate that key processes underlying repair exhibit sex-dependent regulation, particularly inflammatory signaling, immune cell activation, and blood–brain barrier dynamics [71, 105]. These differences include sex-biased cytokine production, divergent microglial activation states, and alterations in leukocyte recruitment and neurovascular integrity, all of which influence the local environment that supports or constrains remodeling [5, 23, 30, 63, 67, 101]. Because these processes are also differentially engaged across the PLC and CLC, stroke associated transcriptional responses may be not only region-specific, but also sex-dependent, with males and females potentially diverging in how they respond to injury or initiate remodeling. Despite this, how sex shapes the transcriptional response to stroke within the PLC and CLC remains largely unexplored.

In this study, we aimed to define and compare the transcriptomic landscape of the PLC and CLC at 7 days post-stroke, a time point associated with the initiation of post-stroke plasticity. We further compared these molecular programs across sexes and whether these molecular changes were accompanied by changes in post-stroke axonal sprouting in both male and female mice.

Materials and methods

Animals: C57BL/6J wild-type (WT) male and female mice (8–16 weeks old, stock no. 000664, Jackson Labs) were housed under a 12 h light/12 h dark cycle with standard laboratory chow and water available ad libitum. All animal procedures were approved by the Institutional Animal Care and Use Committee of UT San Antonio and UT Southwestern Medical Center and were performed in accordance with institutional guidelines. All bulk RNA-seq and histological studies were performed in mice 8–12 weeks of age. The tracing studies included a broader age range, but ages were balanced between males and females. A total of 30 mice were included in the original bulk RNA-seq study design (16 female, 14 male), however 6 males had to be excluded; exclusion criteria are discussed in the RNA-seq analysis section below. For histological analyses, 16 mice were used (10 stroke (5 male, 5 female) and 6 sham (3 male, 3 female)). For axonal tracing studies, 12 mice were used (6 stroke (3 male, 3 female) and 6 sham (3 male, 3 female)). Animals were randomly assigned to experimental groups prior to tissue collection. Because stroke was readily apparent at the time of dissection and histological evaluation, investigators were not blinded to injury status.

Photothrombotic stroke: Focal ischemic lesions were induced as previously described [53] via photothrombosis of vessels supplying primary motor cortex (M1). Mice were anesthetized (2–4% isoflurane/70% NO₂/30% O₂) and positioned in a stereotaxic frame. After shaving and disinfecting the scalp, a midline incision was made, and the skull overlying M1 was exposed and cleared of periosteum. Rose Bengal (40 mg/kg, intraperitoneal, Millipore Sigma, 330,000-5G) was administered, and a 45 mW laser (Sapphire 561-FP OEM Laser System, 120 mW,Coherent, Santa Clara, CA, USA) was directed to stereotaxic coordinates (−1.7 ML, 0.0 AP) corresponding to M1 for 15 min to induce endothelial injury and local thrombus formation. At 24 h post-stroke, mice were screened for incomplete Rose Bengal delivery, and one mouse (female) was excluded due to a subcutaneous injection, as indicated by pink discoloration of the ventral skin. Sham surgeries were performed identically to stroke procedures without the use of the laser.

Infarct volume quantification: All MRIs were acquired by the Research Imaging Institute at UT Health San Antonio. Infarct volumes were quantified from rapid 11.7T MRI acquired 24 h after focal cortical stroke induction in one cohort from the bulk RNA-seq study (n = 7; 4 males, 3 females). The 24 h time point was selected because it is a commonly used time point for standardized assessment of infarct size across animals and because it maximized the interval between MRI anesthesia exposure and tissue collection, thereby reducing the potential influence of anesthesia on gene expression [90, 92]. Mice were anesthetized, and contiguous 1 mm coronal slices spanning the whole brain were obtained. Hyperintense lesions were quantified on T2-weighted images by direct manual tracing with automated volume measurement in ImageJ by an investigator blinded to sex. Stroke volume was calculated as the sum of lesion area across slices multiplied by slice thickness, and final volumes are reported in cubic millimeters. Because this approach can overestimate lesion size, these measurements were used for relative comparisons across animals rather than as a direct measure of final tissue loss.

Tissue collection and RNA extraction: Tissue for bulk RNA sequencing was collected 7 days after stroke. Mice were deeply anesthetized and transcardially perfused with ice-cold, RNase-free 1 × PBS. Brains were rapidly removed and placed on a chilled RNase-free Petri dish, and 2-mm biopsy punches were obtained from both PLC and CLC. Careful visualization, comparison of tissue, and ruler measurements were performed to ensure consistent sampling across animals in both stroke and sham groups. The PLC punch was positioned approximately 3 mm lateral and 1 mm rostral to Bregma. In stroke animals, the punch was visually adjusted to avoid inclusion of infarcted tissue. The PLC included regions classically called peri-infarct tissue and thus included spared cortical regions known to undergo post-stroke plasticity and contribute to motor recovery, including motor, premotor, and primary and secondary somatosensory cortex. The CLC punch was centered approximately 2 mm lateral to Bregma.

Microdissected tissue was immediately transferred into TRIzol reagent (Thermo Fisher Scientific) and homogenized by sequential passage through 18- and 27-gauge needles until no visible fragments remained, followed by RNA extraction as previously described [31]. Briefly, total RNA was freeze/thawed/vortexed three times, followed by precipitation with chloroform and isopropanol precipitation. RNA pellets were washed twice with 75% ethanol, air-dried, and resuspended in 50 uL molecular biology grade water. A nanodrop was used to measure RNA concentration, 260/280 ratios, and 260/230 ratios. Quality control and paired-end 150bp sequencing was performed by Novogene. Only samples with RNA integrity number (RIN) ≥ 7, OD260/280 ≥ 2, OD260/230 ≥ 2, and no degradation or contamination were used for library preparation.

RNA-seq analysis: Libraries were sequenced as paired-end 150 bp reads on an Illumina platform by Novogene (Sacramento, CA). Sequencing quality was high across included samples, with Q30 values ranging from 92.0% to 95.3%, mean error rates of 0.01%−0.03%, GC content of 49.8%−51.0%, and substantial clean-read depth per library (approximately 39.3–107.6 million reads, depending on sequencing batch). Reads were processed with Novogene’s standard RNA-seq pipeline, aligned with HISAT2 to the mouse reference genome mm10 (GRCm38), and quantified with featureCounts. Sample metadata included condition, biological sex, anatomical location, and sequencing batch. Samples were processed in multiple experimental cohorts. Counts from separate cohorts were batch corrected using ComBat-seq [111]. One cohort processed at a different institution (n = 6) remained segregated by batch rather than injury condition and was therefore excluded from downstream analysis. This prevented a sex-balanced design and limited statistical power for sex-stratified comparisons. The final dataset included 12 biological replicates per condition and brain region, comprising 8 females and 4 males per group.

Low-abundance genes were filtered before differential expression analysis by retaining genes with ≥ 10 counts in the minimum group size for each comparison. Stroke versus sham differential expression analyses were performed within each region using both pooled-sex and sex-stratified comparisons. For pooled-sex analyses, sex-chromosome-linked transcripts (Xist, Ddx3y, Eif2s3y, Kdm5d, Uty, Gm29650, Sry, Usp9y, and Eif2s3x) that separated samples by biological sex were excluded but retained for analyses of sex-dependent transcriptional differences. Gene-level counts were analyzed using DESeq2 with size-factor normalization. Genes with log₂FC ≥|0.3| and FDR < 0.1 were considered differentially expressed. Given the discovery-focused nature of this study and the heterogeneity of bulk cortical tissue, these criteria were chosen to retain modest but potentially coordinated expression changes. Interpretation was based primarily on shared patterns across genes, pathways, and network analyses rather than individual transcripts alone. Gene ontology enrichment analyses were performed in Metascape [115] using Gene Ontology Biological Process (GOBP) terms only. For each analysis, the background gene set was defined as all genes detected in the bulk RNA-seq count matrix, and Metascape default parameters and significance thresholds were applied. Sequencing data are available in NCBI GEO under accession number (pending).

Weighted gene coexpression network analysis (WGCNA): WGCNA was performed using the WGCNA package [54, 55]. Batch-corrected, variance-stabilized expression values were used as input after removing lowly expressed (genes with less than 15 counts in more than 75% of samples) and low-variance genes (variance threshold < 0.01) to reduce noise. A signed coexpression network was constructed using biweight midcorrelation, soft-thresholding power β = 12, minimum module size 50, merge cut height 0.25, maximum block size 15,000, and a fixed random seed, yielding color-coded gene modules and corresponding module eigengenes. The resulting adjacency matrix was transformed into a topological overlap matrix (TOM), and genes were clustered by hierarchical clustering of TOM-based dissimilarity. Gene modules were defined by dynamic tree cutting, and closely related modules were merged based on module eigengene similarity. Module eigengenes were then correlated with experimental traits (e.g., injury status, sex, and cortical region) to identify stroke- and sex-associated gene modules. Gene ontology and pathway enrichment analyses were performed on module gene lists to infer biological processes associated with each module. Top 10 hub genes per module were identified using Cytoscape [91] plugin CytoHubba [22], where Hubba nodes were ranked by highest degree score.

Protein-protein interaction analysis: Network analysis was performed using the STRING database (v12.0; [94]) via the web interface (string-db.org). Common upregulated DEGs shared between the PLC and CLC were submitted as a gene list using Mus musculus as the organism. Networks were constructed using a minimum combined interaction score (MCC) of 0.400 and the "Pathways" network source. The resulting network was exported and visualized in STRING website.

Microglia histology: At 7 days post-stroke, mice from the sham (n = 6) and stroke (n = 10) groups were euthanized by isoflurane overdose and transcardially perfused with phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were dissected and post-fixed in 4% PFA for 24 h at 4 °C, then cryoprotected in 30% sucrose in PBS. Frozen tissue was sectioned at 20 µm and mounted onto slides. Sections were washed in PBST (PBS + 0.1% Tween-20), blocked in 5% BSA in PBST, and incubated with an anti-Iba1 primary (Fujifilm Wako, 019–19741; 1:500) overnight at 4 °C. Following PBST washes, sections were incubated with secondary antibody (A-11008, Invitrogen/Thermo Fisher; 1:1000). Sections were washed in PBST, coverslipped, and imaged with the Axioscan (Carl Zeiss) using consistent acquisition settings across groups. Quantification was performed using FIJI (ImageJ). Images were processed using a consistent workflow that included background correction and application of a uniform intensity threshold to isolate Iba1-positive signal. Investigators were blinded to sex but not injury status as the injury was evident in stroke sections. Regions of interest (ROI) were manually delineated using the freehand tool to encompass the ipsilesional secondary somatosensory cortex (S2) and the contralesional motor cortex (M1). Sections were excluded if ROIs were significantly damaged. Iba1 immunoreactivity was summarized as the area fraction within each ROI. Values were averaged across sections to generate a single value per animal. Statistical analyses were performed in GraphPad Prism, and sham versus stroke groups were compared using an unpaired two-tailed t-test.

Corticospinal tract sprouting: Four weeks after stroke induction, mice received unilateral intracortical AAV injections targeting the right primary motor cortex. Anesthetized mice (1–2% isoflurane in 70% nitrous oxide/30% oxygen) were positioned in a stereotaxic frame and continuously monitored for temperature and respiratory rate. A small burr hole was drilled over M1, + 0.5 mm anterior and 1.5 mm lateral to Bregma. A pulled glass micropipette connected to a microinjector (R-480 Nanoliter Microinjection Pump, World Precision Instruments) was lowered to a depth of 0.5 mm from the skull surface, and 1 µL of Synaptotag4 AAV (AAVD/J-HSyn-tdTomatoGap43-EGFPSyb2, titer ~ 10¹³ gc/mL) membrane-targeted tdTomato and synaptobrevin2 fused eGFP was delivered at 100 nl/min. After the virus was delivered, the pipette was left in place for 1 min, raised 0.2 mm, held again for 1 min, and removed after injection. Buprenorphine was administered postoperatively for pain management. Moist food was provided for the first 72 h after stroke.

At 6 weeks post-stroke, mice were transcardially perfused using room temperature PBS followed by ice-cold 4% PFA. Whole brain and spinal cords were removed, submerged in 4% PFA overnight. Whole brains were examined under a fluorescent microscope to verify injection site and expression. Animals with a visibly dim injection core were excluded from further analysis (n = 6). Cervical cords were processed using the SpinalTRAQ imaging, atlasing, and analysis pipeline outlined in detail previously [81]. Briefly, cervical cords (pyramidal decussation to T1) were cut from the whole CNS prep and volumetrically imaged using serial two-photon tomography (STPT) on a Tissuecyte 1000 (TissueVision), registered into the ST-CRA template, subjected to semi-supervised pixel classification of axons and synapses using ilastik [7], and quantified via custom R scripts for level and region-based analyses (10.5281/zenodo.14750099). All samples were processed in a standardized manner and investigators were blinded to injury status and sex.

Statistical Analysis: All non-transcriptomic statistical analyses were performed in GraphPad Prism (v10), and transcriptomic analyses were performed in R (v4.5.2). RNA-sequencing differential expression analyses were performed using DESeq2 (v 1.50.2) with size-factor normalization and Wald testing. Batch correction was performed using ComBat-seq via the sva package (v 3.58.0). Gene identifier mapping was performed with biomaRt (v2.66.1). P-values adjusted for multiple comparisons using the Benjamini–Hochberg correction and genes were considered differentially expressed at an FDR-adjusted p-value < 0.1 and a Log₂FC ≥|0.3|. No additional correction for multiple comparisons across the six independent DGE analyses was applied as each analysis addressed a distinct biological comparison. Infarct volumes of male and female mice (Fig. 4A) were compared using an unpaired Student's t-test. Iba1 area-fraction quantification (Fig. 2D) compared sham versus stroke groups within each region using unpaired Student's t-tests. Each animal contributed a single value, computed as the mean across 4 sections within an ROI. Cervical spinal cord synaptic plasticity (SpinalTRAQ) measured normalized synapse density per animal per cervical level (C1–C8) per lamina. Sham versus stroke comparisons were performed within each level using a Kruskal–Wallis test, with Benjamini–Hochberg correction for multiple comparisons.

Fig. 4.

Sex-specific transcriptional responses in the PLC 7 days post-stroke. A Quantification of infarct volume 24 h post-stroke separated by sex (p = 0.951, unpaired t-test). B PLC DEGs separated by sex, showing upregulated (red) and downregulated (blue) genes (expression relative to the PLC region in sham brains). C Venn diagrams comparing PLC DEGs across sexes, GOBP enrichment for PLC DEGs that are (D) common among sexes, (E) unique to female, and F unique to male

Fig. 2.

Common inflammatory signaling pathways suggest shared myeloid-lineage-cell response. A Venn diagrams comparing PLC (purple) and CLC (green) DEGs B Top shared upregulated DEGs in both PLC (purple) and CLC (green). Asterisks indicate genes that meet DEG criteria of Log₂FC ≥|0.3| and FDR < 0.1. C STRING network analysis of common upregulated DEGs and pathway enrichment of largest clusters. D Representative coronal brain sections stained for IBA1 at 7 days post-stroke, shown for sham and stroke conditions in female and male mice. Quantification shows percent IBA1 + area within PLC and CLC regions of interest (ROI). Males are circles and females are triangles. Whole brain scale bar = 200 µm and zoom scale bar = 20 µm. Statistical comparisons were performed using unpaired t-tests (PLC: p = 0.1783; CLC: p = 0.0853)

Results

PLC and CLC transcriptomes after stroke

We performed photothrombotic stroke or sham surgery in male (n = 8, 4 stroke, 4 sham) and female (n = 16, 8 stroke, 8 sham) adult C57Bl6/J mice and measured T2-weighted MRI lesion volumes from a subset of mice (n = 7) at 24 h after stroke (Fig. 1A). Stroke volumes ranged from 24.5 mm³ to 39.7 mm³ with no mortalities. Strokes were confined to the ipsilesional hemisphere and affected the entire primary motor cortex, part of the primary sensory cortex, and extended just below the corpus callosum (Fig. 1B).

Fig. 1.

RNA-seq reveals heightened inflammatory signaling in the perilesional than the contralesional cortex at 7 days post-stroke. (A) Experimental timeline of study and biopsy punch locations of PLC and CLC. Filled circle indicates approximate infarct location. (B) T2 MRI’s of representative stroke mouse with sections spanning the length of the brain rostral (top left) to caudal (bottom right). (C) Volcano plots for PLC and CLC depicting upregulated (red), downregulated (blue), and nonsignificant DEGs (grey). (D) GO enrichment analysis of the PLC upregulated (left) and downregulated DEGs (right). (E) Bar chart of Top 5 upregulated and downregulated PLC DEGs. Asterisk signifies gene that meets DEG criteria. (F) GO enrichment analysis of the CLC upregulated (left) and downregulated DEGs (right). (G) Bar chart of Top 5 upregulated and downregulated CLC DEGs. Asterisk signifies gene that meets DEG criteria of Log2FC ≥ |0.3| and FDR < 0.1.

Region-specific transcriptomic profiles of PLC and CLC were obtained one week following primary motor cortical stroke (Fig. 1). PLC comprised ipsilateral cortical tissue immediately adjacent to the infarct and CLC was defined as the cortical region homotopic to the lesion, with neither region demonstrating positive TUNEL staining, indicating a lack of ongoing cell death at this timepoint (Supplemental Fig. 1). Differential expression was determined by comparing the gene expression of each site with the same region from sham-operated brains (Log₂FC ≥|0.3|, FDR < 0.1). Bulk RNA-sequencing of 48 samples identified 1,891 differentially expressed genes (DEGs) in the PLC (1,365 upregulated, 526 downregulated) and 274 DEGs in the CLC (245 upregulated, 29 downregulated) (Fig. 1C). The PLC highly upregulated genes that align with an acute, localized response to injury, including chemokines (Ccl2, Ccl3, Cxcl9, Cxcl10), innate and adaptive immune system signals (Clec7a, Fcgr4, Cd300ld, Cd300lb, Siglec1, Lilrb4a, Cd22, Ly9, Itgax, Hcar2, Fgr), and interferon stimulated genes (Ifi204, Ifi206, Ifi207, Ifi209, Ifi27l2a). These genes were enriched for GO Biological Processes (GOBP) such as response to IFN-β, positive regulation of IL-1 production, cytokine production, and inflammatory response (Fig. 1D, left). In contrast, downregulated genes in the PLC were linked to processes for neuronal activity and signaling, including neurofilament cytoskeletal organization, chemical synaptic transmission, and neurotransmitter transport (Fig. 1D, right). Despite the remote position of the CLC relative to the lesion, highly upregulated genes also included chemokines (Ccl3, Ccl6, Ccl9, Cxcl16), antiviral response proteins (Irf7, Mx1, Usp18), and genes associated with glial reactivity and neuroinflammation (Clec7a, Tlr2, Lag3, Itgax, Gpr84, Cd44). These genes were enriched for neuroinflammatory pathways such as IFN-α production, leukocyte-mediated immunity, and innate immune responses. The small set of downregulated genes in CLC (29) returned two significant terms (peptidase and endopeptidase activity, FDR < 0.05; Fig. 1F). Given the limited small number of genes, we interpret this result as exploratory and do not draw conclusions from it.

We next examined the overlapping and unique DEGs for the PLC and CLC. The PLC uniquely expressed 1701 genes (1184 upregulated, 517 downregulated) that were enriched for inflammatory processes such as regulation of the innate immune response, positive regulation of cytokine production, and regulation of leukocyte activation (Fig. 2A, Supplemental Fig. 2). Top differentially expressed genes unique to the PLC included Mmp12, Cd5l, Runx3, Cxcl9, and Cxcl10, and Hcar2. Surprisingly, the CLC uniquely expressed only 84 genes (64 upregulated, 20 downregulated). Top unique upregulated genes have been linked to chromatin remodeling (Xlr4b) [26], neuronal survival (Sh3rf2) [107], and synaptic integrity (Fat2) [103, 104]. Unique downregulated genes included Entppl, Hif3a, and Cbln4, which are involved in reactive astrocyte signaling [98], hypoxia responses [66], and synaptic function [82]. However, at the pathway level, GOBP enrichment of CLC-unique genes yielded broad categories, limiting interpretability (Fig. 2A, Supplemental Fig. 2).

We identified a substantial set of common DEGs across regions, indicating that CLC converges on a shared transcriptomic response at one-week post-stroke despite its remote anatomical location. Among 181 shared upregulated DEGs, top transcripts, including Clec7a, Itgax, Lag3, Lgals3bp, Gfap, and C4b, were similarly increased in both regions (Fig. 2B) and enriched for inflammatory pathways, including IFN-α signaling, adaptive immune response, and innate immune response. Many of these genes are markers for activated microglial phenotypes, including disease-associated microglia (DAMs), which have been increasingly implicated in stroke. We therefore next tested whether the shared signature aligned with published DAM signatures [6], which revealed a significant overlap with DAM signatures from multiple settings (Supplemental Fig. 3). STRING protein–protein interaction (PPI) analysis further [94] identified a large, densely connected network of genes that associated with immune effector signaling, microglial phagocytosis, and Tyrosine Protein Tyrosine Kinase Binding Protein (TYROBP)-causal network pathways (Fig. 2C). This interconnected gene network highlighted interactions that are known to activate both stroke associated microglia and infiltrating peripheral macrophages after stroke [9, 48, 102, 110]. A second prominent cluster was comprised of interferon response genes (Irf7, Ifih1, Oas2, Usp18) that linked to anti-viral response pathways, which may arise downstream of microglial activation [15, 28, 78]. Additional peripheral networks were observed, including an astrocyte-associated group containing Gfap, Aqp4, and Vim.

Given that the shared PLC–CLC upregulated signature was largely driven by myeloid lineage cells, we performed histological assessment of Iba1 signal in the PLC and CLC at 7 days post-stroke. We assessed myeloid morphology by measuring the percent surface area occupied by thresholded Iba1 signal within each region. Iba1⁺ area fraction did not increase and, instead, trended lower (Fig. 2D). While area-fraction measurements have limited interpretability, one possible interpretation is that transcriptional profiling is more sensitive to cellular state transitions than surface area-based measurements [75, 90]. This may be particularly relevant in remote regions, where prior studies have shown persistent or delayed myeloid responses linked to secondary degeneration and circuit remodeling after stroke [19, 35, 52, 76]. Together, these findings raise the possibility that reactive myeloid phenotypes emerge not only locally after ischemic injury, but also in the contralesional cortex, a remote region implicated in post-stroke plasticity.

Transcriptional correlates of infarct volume

We performed gene-wise correlations between cortical gene expression and infarct volume to assess whether the shared reactive myeloid signal identified by differential expression tracked with lesion severity. The myeloid activation genes identified by differential expression analyses did not correlate with infarct volume (Supplemental Fig. 4I-L). Classic injury-associated inflammatory markers, including Gfap, Aif1, and Il6 [1, 57] also did not correlate with infarct volume. However, both the PLC and CLC contained numerous transcripts whose expression was strongly associated with stroke volume (R² ≥ 0.9) (Supplemental Fig. 4A-C). Many of the most highly correlated genes showed negative correlations. In the PLC, these included Nek6 (R² = 0.98, p < 0.001), a serine/threonine kinase reported to exacerbate injury by promoting astrogliosis [109], and Fstl4 (R² = 0.92), which has been implicated as a negative regulator of BDNF signaling [93] (Supplemental Fig. 4 D,E). In CLC, strongly negatively correlated genes included the deSUMOylating enzyme SENP6 (R² = 0.99), which has been linked to ischemia-induced neuronal apoptosis [108] and Tspoap1 (R² = 0.98), which has not, to our knowledge, been directly associated with stroke but participates in the regulation of neurotransmitter release [65] (Supplemental Fig. 4 F, G). Surprisingly, only one gene, Acly, encoding ATP-citrate lyase, a key enzyme for acetyl-CoA synthesis, was shared between regions. Acly was negatively correlated with infarct volume, consistent with prior work indicating a neuroprotective role [103, 104] (Supplemental Fig. 4H). Together, these findings suggest that the shared myeloid activation gene program in the PLC and CLC is not restricted to larger lesions. Rather, within the relatively narrow range of infarct sizes in our cohort, a distinct set of transcripts may provide greater sensitivity to lesion-associated transcriptional variation than canonical inflammatory markers.

Hierarchical clustering of PLC and CLC transcriptomes

To determine whether gene expression signatures were consistent across animals, we performed hierarchical clustering of our 48 stroke and sham samples, which provided sufficient depth to resolve inter-animal variability and identify potential transcriptional subgroups. Hierarchical clustering of the top 50 up-regulated DEGs in both regions revealed clear separation of stroke and sham samples (Fig. 3). In the PLC, stroke samples segregated into two groups. Group 1 was defined by the upregulation of genes associated with several myeloid lineage cell responses, including phagocytosis (Mmp12, Lgals3, Gpnmb, Clec12a), and pro-inflammatory signaling (Itgax, Cst7, Ifi27l2a, Clec7a, and C3), and chemokines and cytokine secretion (Ccl3, Ccl4, Ccl5, Ccl12, Cxcl9, Cxcl10, Il1rn, and Tnfrsf26) (Fig. 3A) [50, 106]. Group 2 was characterized by uniformly lower expression of all top 50 upregulated DEGs, indicating differences in the strength of the transcriptional response to stroke. This variation was not driven by sex or lesion severity, as Groups 1 and 2 contained both sexes and a range of infarct volumes. Hierarchical clustering of the top 50 upregulated DEGs in the CLC also identified stroke subgroups that corresponded to the PLC transcriptional subgroups (Fig. 3B). Specifically, the same animals that were classified as transcriptionally distinct animals within CLC subgroup were similarly segregated in PLC subgroups. Like the PLC, CLC subgroups were distinguished by a higher expression of genes associated with activated myeloid phenotypes (Treml2, Gpr84, Clec7a, Lgals3bp, Cst7, Lag3, Lyz3, Ifi27l2a) [14, 64, 113] (Fig. 3B). Thus, animals distinguished by stronger neuroinflammatory gene expression in the PLC showed a corresponding elevation of this signature in the CLC.

Fig. 3.

Sample-level transcriptomic analyses of PLC and contralesional cortical stroke responses. Heatmap showing z-scored expression of top 50 upregulated DEGs across condition for (A) PLC and B CLC across samples

Cortical transcriptional responses to stroke in females and males

Our analyses thus far did not reveal overt sex-dependent trends. However, sex effects may be difficult to detect in pooled analyses. We therefore repeated differential expression and pathway analyses separately in male and female mice to more thoroughly evaluate sex-specific transcriptional responses. There were no differences in the lesion volumes of the subgroups measured in male vs female mice (Fig. 4A). The PLC exhibited a strong transcriptional response in both females (1597 DEGs) and males (1889 DEGs) (Fig. 4B). These genes were mostly upregulated and demonstrated substantial overlap (Fig. 4C). GO terms for shared upregulated genes suggested a robust inflammatory response, including regulation of the innate and adaptive immune responses as well as cell proliferation and migration (Fig. 4D). Highly expressed shared DEGs included Toll-like receptors (Tlr1, Tlr2, Tlr7, Tlr9, Tlr12), complement proteins (C1qa, C1qb, C1qc, C1qtnf6, C3, C4b), cytokine and chemokines (Ccl2, Cxcl10, Ccr5, Ccl12, Il1a), and phagocytic effectors (Aif1, Lgals3, Ctss, Itgam, Itgax). GO terms for shared downregulated genes (202), which encoded synaptic proteins (Snap25, Vamp1, Homer1, Cbln2) and voltage-gated ion channels (Cacna1i, Cacng2, Cacnb4, Hcn1, Scn1a, Kcna2), included trans-synaptic signaling, GABA signaling pathway, neuronal action potential propagation, and neurofilament cytoskeleton organization (Fig. 4D).

In the PLC, females uniquely upregulated genes encoding pattern recognition receptors (TLR3, TLR4, TLR8, Nlrp3, Nlrp1, Rig-I, Lgp2, Aim2) [3] and interferon-stimulated genes (Ifit1, Ifitm3, Rsad2, Usp18), suggesting increased inflammatory signaling. GO analysis of these genes indicated a positive regulation of IFN-β, IL-1, and IL-6 production (Fig. 4E). Conversely, male-specific upregulated DEGs encompassed more diverse cellular processes such as synaptic signaling, astrocyte activation, and angiogenesis (Fig. 4F). Downregulated genes unique to females (46) and males (227) were enriched for both neuronal signaling and plasticity. However, female-specific GO terms, including regulation of membrane potential, GABA signaling, long-term memory, and post-synapse organization, were distinct from those in males, including action potential, trans-synaptic signaling, and axon development (Fig. 4F). These subtle differences may suggest a sex-dependent suppression of neuronal repair processes in the PLC.

The CLC exhibited a comparable transcriptional response in females and males, with 114 DEGs in females and 171 in males (Fig. 5A). Across sexes, 43 genes were commonly upregulated, and these converged on immune-signaling pathways, including IL-6 production, leukocyte activation, and inflammatory response (Fig. 5B, C). Interestingly, many of these shared transcripts have been linked to microglial activation, including Ctss, Lag3, Ccl6, Irf7, Abca1, Cd33, Cd44, and Lgals3bp [38, 47, 58, 59, 89]. In contrast, only two genes were commonly downregulated (Cbln4 and Serpinb8), neither of which has a well-established connection to stroke.

Fig. 5.

Sex-specific transcriptional responses in the CLC 7 days post-stroke.A CLC DEGs separated by sex. B Venn diagrams comparing CLC DEGs in both sexes. C-E GOBP enrichment of genes (C) common among sexes, (D) unique to female, and E unique to male

Female-specific upregulated genes (56) were enriched for pathways related to IL-1 production, response to type II interferon, regulation of the innate immune system, and vascular development (Fig. 5D). Male-specific upregulated genes (105) included a broader set of enriched terms, including synapse pruning, microglial activation, and inflammatory signaling (Fig. 5E). Notably, gliogenesis pathways were detected in both sexes, but the contributing genes differed. In females, this signal was driven by genes such as Gfap, Adgrg1, Sox13, Tgfb1, Vim, and Apcdd1, whereas males expressed C1qa, Col3a1, Csf1r, Hes5, Hexb, Kcnj10, Lyn, Nab2, Sox11, Plpp3, and Trem2. Very few genes were downregulated in the CLC across both sexes and exhibited essentially no overlap.

Coexpression network analysis of stroke- and sex-associated gene modules

To complement differential expression analysis, we examined coordinated gene expression changes via weighted gene coexpression network analysis (WGCNA). This approach identified 12 coexpression modules in the PLC and 3 in CLC (Fig. 6A). In the PLC, four modules were sex-independent (Fig. 6B and C). Two modules (Fig. 6B) were positively correlated with stroke (“stroke-activated”) and enriched for reactive astrocyte signaling, microgliosis, and inflammatory pathways. The two negatively correlated PLC modules (Fig. 6C) were enriched for neuronal pathways, including action potentials, modulation of chemical synaptic transmission, and regulation of neuronal differentiation.

Fig. 6.

Weighted gene coexpression network analysis reveals sex-specific and regionally localized transcriptional networks following stroke. A Module-trait relationship heatmaps showing Pearson correlations between WGCNA module eigengenes and cortical region (PLC vs CLC) within stroke samples, shown separately for males and females. Color intensity indicates the strength and direction of the correlation. Asterisks indicate significance of the correlation between each module eigengene and sex/group. (*P < 0.05, **P < 0.01, ***P < 0.001). GOBP enrichment analysis for sex-independent (B-D) and sex-correlated E, F modules

Sex-dependent modules further supported suppression of neuronal signaling and plasticity in the PLC (Fig. 6E and F), but through distinct gene expression. In females, the red module was negatively correlated with stroke and was enriched for pathways related to membrane potential regulation and membrane organization, and included genes involved in synaptic function (Snap25, Gabra1, Gabrg2, Grin2c, Cnih2, Dlg1) and protein homeostasis (Hsph1, Rb1cc1, Vmp1, Hspa4, Ube2g1) (Fig. 6E). In males, the yellow module was negatively correlated and included transcripts linked to axon guidance cues (Epha5, Ephb3, Sema3f, Nrp2, Unc5a, NogoR), ion channels (Kcnq4, Kcnb1, Trpm4), and cellular stress response proteins (Xrcc4, Prdx1, Nqo1, Sesn3) (Fig. 6F). Together, these sex-specific modules suggest that suppression of neuronal plasticity in the PLC may affect synaptic signaling in females and structural remodeling in males.

While PLC gene expression has consistently suggested a suppression of neuronal signaling and plasticity under stroke conditions, one male-specific positively correlated module (Fig. 6F) was upregulated in the PLC and was notable for genes linked to neuronal growth and repair; this module contained 182 upregulated genes, including Esr1, which was upregulated in males but not females, and featured hub genes such as GAP43, Esr1, and Calb2. Females exhibited one positively correlated module (tan); however, its members were enriched for leptomeningeal, choroid-plexus, and extracellular matrix markers (Cdh1, Fbln1, Foxc1), which can sometimes indicate variation in tissue composition or sample handling [70].

In CLC, WGCNA yielded three modules, of which only one (cyan) was significantly associated with stroke (Fig. 6A and D). This module was enriched for microglial activation, synaptic pruning, and gliogenesis. The hub genes for the cyan module (Cd68, Gfap, Apoe, Csf1r, Fcgr3, B2m, and Cd44) suggest microglial- and astrocyte-mediated inflammatory signaling in the CLC across both sexes. Overall, WGCNA supports findings of sex-independent inflammatory signaling and suppression of neuronal plasticity in the PLC and a glial-driven neuroinflammatory response in the CLC.

Post-stroke CST plasticity across sexes

We next asked whether sex-dependent transcriptional programs correspond to differences in post-stroke plasticity. To examine this question, we quantified axonal sprouting of the contralesional corticospinal tract (CST), a robust, functionally relevant form of post-stroke plasticity, in both female and male mice. Using our previously developed spinal cord analysis pipeline, SpinalTRAQ [81], we mapped and quantified contralesional CST synapses throughout the entire cervical spinal cord (Fig. 7A). In sham animals, both sexes showed strikingly similar CST innervation patterns, with higher normalized synapse density in dorsal and intermediate laminae (laminae 5, 7–10) across C1–C8 and minimal innervation of dorsal laminae 1–2 (Fig. 7B,C). Six weeks after stroke, CST sprouting and synapse formation into the denervated hemicord significantly increased in both females and males. In line with our previous findings [81], both sexes showed the largest gain in synapses in lamina 5 and 7 across all cervical levels after stroke. Additionally, we found increases in lamina 4 and 9 synapses that localized to caudal cervical sections. This indicates that, despite robust differences in cortical CLC gene expression between males and females, we found no evidence for sex differences in contralesional CST synaptic plasticity at the level of the cervical spinal cord.

Fig. 7.

Comparison of chronic post-stroke spinal plasticity by sex. A Experimental schematic: adult C57BL/6 mice (n = 3 per group) underwent PT stroke or sham surgery followed by anterograde synaptic labelling (Synaptotag4) at 4 weeks post-stroke and SpinalTRAQ analysis at 6 weeks post-stroke. B Maximum intensity projections (MIPs) of synapse-classified pixels from the volumetric cervical spinal cord dataset of a representative animal. White outlines denote spinal cord boundaries, and the red line indicates the midline. C Heatmap of classified synaptic signal showing group-mean normalized intensity (log₁₀ scale) across cervical spinal levels and lamina. Laminae are hierarchically clustered based on presynaptic terminal distributions in the contralateral hemicord

Discussion

Stroke induces plasticity in both lesion-adjacent and lesion-remote regions. In rodent models, the contralesional cortex (CLC) is one of the most studied remote sites of post-stroke remodeling. Our lab and others have shown that both the perilesional cortex (PLC) and CLC undergo significant, functionally relevant plasticity after stroke despite markedly different local environments [21, 25, 56, 81, 112]. This raises a central question in stroke recovery: what are the drivers for regionally localized post-stroke plasticity? To investigate this, we used bulk RNA sequencing at 7 days post-stroke to compare the transcriptomes of the PLC and CLC.

A defining feature of the CLC is that it lies outside the infarct and does not experience direct ischemic injury, yet it exhibits robust, functionally relevant structural plasticity that is otherwise uncharacteristic of the mature nervous system. Therefore, we hypothesized that the CLC would exhibit a distinct transcriptomic signature from the PLC, where neurons are exposed to lesion-adjacent injury signals and exposed to a highly inflammatory milieu. Instead, we observed substantial overlap between the two regions. Consistent with prior work, the PLC exhibited a broad induction of innate and adaptive immune pathways and suppression of neuronal genes [24, 62, 83, 85]. While the CLC displayed a significantly more modest response, its gene expression signature was also enriched for neuroinflammatory pathways and, unlike the PLC, preserved neuronal gene expression. The CLC demonstrated a very limited set of uniquely expressed genes that could not be clearly characterized. Instead, DEGs in the CLC were also often found in the PLC and converged on inflammatory and myeloid lineage cell activation pathways. More specifically, these genes were classically involved in cytokine and pro-inflammatory signaling pathways frequently associated with activated myeloid populations [15, 96].

The shared signature was strongly enriched for markers of reactive myeloid-lineage cells and overlapped with marker sets linked to interferon-responsive (IRM) and disease-associated (DAM) microglial states previously described after stroke [6, 28]. DAM-like microglial populations (Clec7a, Cst7, and Itgax) have been reported in remote post-stroke regions on the ipsilesional side [19], and while their role in stroke recovery remains unknown, they have been proposed to reflect a phagocytic state that may support structural reorganization [28, 43, 102]. We also observed marked upregulation of Ifi27l2a, a marker for a prominent pro-inflammatory microglial population found in the PLC [51]. Its upregulation in CLC suggests that future studies should determine whether similar inflammatory myeloid states emerge in lesion-remote regions and whether they contribute to post-stroke remodeling. Shared upregulation of Ccl3, Tlr2, and Cd22, genes implicated in both peripheral immune and reactive myeloid signaling, supports prior work establishing immune cell responses in remote regions after stroke [46, 80, 88, 95, 97]. Because several of these transcripts are not exclusive to microglia, we considered whether infiltrating immune populations could also contribute, particularly neutrophils, which release chemokines and interferon-responsive mediators that overlap with our observed signal. We found no transcriptional evidence for a substantial neutrophil contribution in the CLC, since canonical neutrophil markers (Ly6g, Mpo, Elane, Ngp, Cxcr2) were not differentially expressed. This is consistent with prior work showing that neutrophil infiltration after focal ischemia is largely confined to the ipsilesional leptomeninges, perivascular spaces, and infarct core, with neutrophils essentially absent from contralateral brain tissue [27, 73, 77]. We therefore interpret the shared inflammatory transcriptional signature as most consistent with myeloid-lineage-cell activity, while acknowledging that bulk profiling cannot resolve whether this reflects resident microglia, recruited peripheral myeloid cells, or both.

Although the molecular signals thought to support later remodeling, including axonal sprouting, are believed to be strongly engaged by 7 days after stroke [60], we did not observe a robust induction of neuronal plasticity programs. In our data, most DEGs were classically associated with non-neuronal cell types, and in the PLC, genes linked to neuronal plasticity were suppressed despite the absence of cell death in this region. We interpret this cautiously. Bulk RNA-sequencing enables a useful comparison of two key regions involved in stroke recovery but has limited sensitivity for neuronal transcriptional responses, particularly when large shifts in cellular composition can dilute neuronal signal. It may be further compounded by the fact that neurons with direct axonal injury may engage distinct transcriptional responses from those with circuit disruption, differences that can be masked by bulk profiling, which averages signals across diverse neuronal states and environmental cues.

Because sex influences processes that influence the post-stroke molecular environment, including inflammatory signaling, microglial activation states, and leukocyte recruitment, we asked whether the shared PLC/CLC inflammatory landscape is sex-dependent. Females and males engaged in partially distinct molecular responses to stroke. In the PLC, both sexes showed robust induction of innate and adaptive immune pathways and downregulation of synaptic and excitability genes. However, females exhibited stronger upregulation of pattern recognition receptors and interferon-stimulated genes, with GO terms pointing to enhanced IFN-β, IL-1, and IL-6 production. Males, in contrast, showed additional enrichment for synaptic signaling, astrocyte activation, and angiogenesis. WGCNA provided additional support by identifying female-specific modules with suppressed pathways for synaptic signaling and excitability, and a male-specific module enriched for neuronal growth and repair, with hub genes such as GAP43, Calb2, and Esr1. These differences suggest that sex differences in neuroinflammatory signaling should be further explored, particularly since transcriptional studies in female mice are extremely limited [85]. Our unbalanced groups reduced power for sex-stratified comparisons, and larger balanced cohorts will be needed to determine whether these differences are robust and therapeutically informative.

Sex-dependent transcriptional differences were detected at 7 days post-stroke, a time point considered an early, initiating phase of repair, when large-scale structural plasticity is not evident. Therefore, we examined corticospinal plasticity at 6 weeks post-stroke, a chronic time point when sprouting and circuit remodeling are more commonly assessed in the stroke literature and are thought to be relatively stabilized. We chose to compare structural plasticity across sexes using SpinalTRAQ, a pipeline previously developed in our laboratory for automated quantification and anatomical mapping of fluorescent signal throughout the bilateral cervical spinal cord [81]. This approach is particularly informative for contralesional CST projections arising from the CLC, as these axons show a predominantly unilateral innervation pattern in the uninjured spinal cord, allowing for highly sensitive detection of post-stroke changes. Females and males showed broadly similar CST innervation patterns at baseline, and the patterns remained comparable at 6 weeks after stroke. The stronger inflammatory activation observed in females during the early subacute phase, therefore, did not translate into detectable sex differences in contralesional CST remodeling at this later stage. It remains possible that sex influences the timing or mechanisms of plasticity, or other dimensions of recovery not captured by contralesional CST sprouting alone.

This study has several important limitations. Because transcriptomic profiling was restricted to a single subacute time point, we cannot determine whether the observed sex- or region-specific differences are transient or persistent. In addition, the use of bulk RNA-seq from heterogeneous cortical tissue limits cell-type resolution and does not allow us to assign these changes definitively to specific microglial, astrocytic, neuronal, or peripheral immune populations, nor does it address protein-level or functional consequences. Our measure of post-stroke plasticity was also intentionally focused on one major form of plasticity, axonal sprouting, within a single pathway that contributes to stroke recovery, the uninjured CST. While this enabled a highly sensitive and comprehensive comparison of contralesional CST remodeling, it did not capture other circuits or forms of plasticity that may differ by sex. Finally, we did not directly link transcriptional responses to behavioral outcomes. Future studies should address these gaps by combining single-cell or spatial transcriptomics with parallel measurements of plasticity and behavior. Although these approaches have identified important cell-type-specific contributions after stroke, they have not been integrated across modalities or compared across multiple regions within the same study [13, 34, 39, 41, 44, 51, 87, 114]. Future studies examining molecular, cellular, and behavioral features in parallel may provide a clearer view of which mechanisms are most actively driving recovery.

In summary, this work identifies a shared cortical transcriptional state at day 7 after focal motor cortical stroke, enriched for inflammatory and reactive myeloid-lineage-associated gene expression in the PLC and CLC. These findings show that a region remote from the stroke can engage a robust transcriptional response to stroke, and that post-stroke inflammatory responses are not confined to lesion-adjacent cortex but extend across multiple regions associated with post-stroke plasticity. This overall response was broadly similar in females and males, with both sexes showing shared neuroinflammatory-associated gene expression across the PLC and CLC. At the same time, we identified sex-specific transcriptional differences that did not associate with measures of structural post-stroke plasticity. Although the overall remodeling pattern was similar, these findings suggest that sex may be an important modifier of the post-stroke molecular environment and should be considered in future efforts to understand or therapeutically target recovery. By linking unbiased cortical transcriptomics with cervical spinal cord mapping of CST plasticity, our study provides a framework for identifying molecular candidates and prioritizing future mechanistic studies across regions, cell types, and sex. More broadly, the description of convergent molecular signatures across spatially remote brain regions has potential relevance for the discovery of novel plasticity biomarkers and strategies to enhance repair after stroke and other CNS injuries.

Abbreviations

PLC

Perilesional Cortex

CLC

Contralesional Cortex

PT

Photothrombotic

RNA-seq

RNA sequencing

PPI

Protein–protein interaction

DAM

Disease-associated Microglia

EAE

Experimental autoimmune encephalomyelitis

5xFAD

5 Familial Alzheimer’s Disease

GOBP

Gene Ontology Biological Processes

GEO

Gene Expression Omnibus

Authors’ contributions

DB, VAA, MK, AMS, PMD, and MPG designed research; DB, VAA, MK, KRZ, RC, PR, EJP, and DMOR performed research; DB, VAA, MK, KRZ, RC, MPG analyzed data; DB, VAA, AMS, and MPG wrote the paper.

Funding

This work was supported by National Institutes of Health (NIH) awards T32AG082661 (DB), T32GM -113896 and -145432 (DB), T32NS082145-08 (DB), AHA Fellowship 23PRE1018993 (DB), GSBS Neuroscience Training Fellowship (MK), NS088555 (AMS), OT2OD032581 (MPG), and by gifts from the Haggerty Foundation (MPG) and Meier Family Foundation (MPG). P.M.D and the members of the Douglas lab were supported by the Clayton Foundation for Research, the Welch Foundation (I -2061 -20210327), the American Federation of Aging Research (AFAR 2023), the National Institutes of Health (R01AG076529, R01GM15385).

Data availability

The datasets generated and/or analyzed during the current study are available in the Gene Expression Omnibus repository.

Declarations

Ethics approval and consent to participate

All animal procedures were approved by the Institutional Animal Care and Use Committee of UT San Antonio and UT Southwestern Medical Center and were performed in accordance with institutional guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.