Mild Subcortical Stroke Induces Widespread Chronic Reactive Astrogliosis That Is Largely Unaffected by Microglia and Age

Abstract

Ischemic stroke induces a plethora of pathophysiological changes, including neuroinflammation and chronic cerebrovascular dysfunction. In humans, even small, silent strokes can trigger these pathologies, which can spread to brain regions far beyond the infarct and persist chronically, ultimately worsening prognosis and increasing the risk for vascular dementia and Alzheimer’s disease. The cause of this pathology is unknown, but reactive astrocytes and microglia are likely contributors. Here, we describe an optimized short-duration middle cerebral artery occlusion model that produces a clinically relevant small stroke mostly confined to subcortical regions, similar to many silent strokes in humans. We termed this model the mild subcortical infarct (MSCI). We then mapped the spatiotemporal extent of reactive astrocytes and microglia during the sub-acute period (1, 3, and 7 days) following MSCI. We observed that reactive astrogliosis develops more rapidly and spreads more extensively, compared to the reactive microglia response following this small infarct. Microglial depletion resulted in larger infarct sizes but did not prevent reactive astrocytes. Aging mice exposed to MSCI exhibited a comparably strong response of reactive astrocytes and microglia as young mice. Lastly, reactive astrocytes persisted for at least nine months after MSCI. We propose that this mild ischemia model is valuable for examining the chronic effects of subcortical stroke. It may be especially useful for investigating the functional impact of reactive astrogliosis in regions distal from the primary injury site.

1. Introduction

Stroke is a leading contributor to major disability in the aging population (Benjamin et al., 2018) and one of the strongest risk factors for dementia (Hachinski, 2018), a growing concern for the aging population worldwide. Understanding the complexities of stroke-induced neurological dysfunction is important for discovering new interventions to improve clinical outcomes and quality of life (Feske, 2021).

Ischemia and the resulting cell death (infarct) can trigger gliosis, a process wherein astrocytes and microglia undergo changes in their morphological and gene expression profiles. These changes are mainly geared toward the protection of tissue, likely to repair and maintain the infarct border, preventing further damage (Amantea et al., 2015; Burda & Sofroniew, 2014). For example, microglia clear dead cells and debris via phagocytosis (Jia et al., 2021), while astrocytes form glial scars that limit the spread of injury (Amantea et al., 2015; Anderson et al., 2016). However, reactive microglia and astrocytes may also release detrimental signals, such as pro-inflammatory cytokines that can exacerbate neuronal damage in the long term (Amantea et al., 2015; Gong et al., 2023). After ischemic infarction, gliosis spreads beyond the immediate infarct and its border regions in an outward gradient, affecting otherwise intact tissue (Barreto et al., 2011; Gong et al., 2023). However, the precise spatiotemporal extent of gliosis after stroke remains unclear. In certain pathological conditions, microglial activation induces specific subtypes of reactive astrocytes, as seen in response to lipopolysaccharide (LPS), a model of bacterial infection (Liddelow et al., 2017), and traumatic brain injury (Witcher et al., 2018). Whether, and to which extent, reactive astrogliosis in other pathological contexts, such as after stroke, depend on signals from microglia is also unknown.

Many ischemic stroke studies have focused on the mechanisms underlying infarct development during the acute period (<1 day after stroke). The sub-acute (1–2 weeks after stroke) and chronic (months and years later) effects of stroke on areas distant from the initial infarct remain poorly understood, despite long-term deficits extending beyond infarcted regions. Areas distant from the infarct site exhibit abnormalities such as decreased neurovascular coupling (Krainik et al., 2005; Li et al., 2021; Lin et al., 2011; Salinet et al., 2015), vascular dysfunction, and cognitive deficiencies (Levine et al., 2015; Mijajlovic et al., 2017). Individuals who experience small, silent, and recurring infarcts can develop vascular dementia and Alzheimer’s disease in later life (Miklossy, 2003). Notably, subcortical damage without acute cortical injury is especially clinically relevant, as insults in these regions are more closely associated with dementia (van Rooij et al., 2016; Vermeer et al., 2007). Given that both microglia and astrocytes play a role in modulating cerebral blood flow (Bisht et al., 2021; Mishra et al., 2024), long-term gliosis may alter their role in this context and contribute to neurovascular dysfunction in regions beyond the infarct. This could lead to metabolic stress and drive dementia in later life (Ghosh et al., 2013; Hickman et al., 2018).

The 60-minute transient middle cerebral artery (MCA) occlusion (MCAo) model is widely used to study ischemic stroke (Longa et al., 1989; Sozmen et al., 2012). This model results in an infarct that spans the subcortical regions, including the striatum, and much of the MCA territory of the cortex (Figure 1(B), top). Large strokes of this magnitude are considered malignant strokes in humans and are frequently lethal (Barthels & Das, 2020; Carmichael, 2005). Although rodents can survive a 60-minute MCAo for several days, this model is impractical for studying the more chronic effects of stroke. A model that produces a reliable, clinically relevant, subcortical infarct would, thus, be valuable for studying how strokes contribute to complex pathologies that develop over long periods, such as vascular dementia and Alzheimer’s disease.

Figure 1.

A 30 min middle cerebral artery (MCA) occlusion (MCAo) produces a mild subcortical infarct. (A) Example brain sections (2 mm thick) stained with 2,3,5-triphenyltetrazolium chloride (TTC) 24 hours after a 60 min (left) or 30 min (right) MCAo. Infarcted tissue does not take up TTC and remains white (infarct denoted by arrow in top panel; 30 min MCAo resulted in a diffuse reduction in striatal TTC stain). (B) Change in cerebral blood flow (measured by laser Doppler flowmetry) in the MCA territory before, during, and after occlusion. (C) 30 min MCAo produced a similar infarcted area (expressed as percent of contralateral region) within the striatum as 60 min MCAo, while the percentage of area infarcted in the cortex and the whole hemisphere were significantly smaller after 30 min MCAO. Data displayed as mean ± SEM, unpaired t-test. (D) Schematic depicting regions analyzed in this study (Created in BioRender. Stackhouse, T. (2026). https://BioRender.com/g6wxs3a). ACA, anterior cerebral artery territory of the cortex; MCA, middle cerebral artery territory of the cortex. Each dot represents an animal.

In this study, we describe a new optimized model of stroke for studying the effects of small, mild subcortical infarct (MSCI) using a 30-min MCAo and examine the extent of reactive astrocytes and microglia during the sub-acute period (first week after stroke; Birenbaum et al., 2011). We observed that MSCI leads to an increase in both reactive astrogliosis and microglial reactivity during the subsequent week, though their spatiotemporal progression patterns differ. Severe depletion of microglia increased the infarct size but did not abolish reactive astrocytes. Additionally, aged mice displayed higher baseline levels of astrocyte and microglia markers, indicating aging-related glial changes, but their response to stroke remained robust and comparable to that of mid-life adults. When MSCI was induced in mid-life adults, we found that reactive astrocytes persist for at least 9 months after ischemic injury. Together, these findings suggest that reactive astrocytes do not completely depend on reactive microglia following ischemia, and that reactive astrogliosis induced by a single ischemic event may not resolve fully for long periods.

2. Materials and Methods

2.1. Animals

All experiments were done in accordance with the policies of Oregon Health & Science University’s Institutional Animal Care and Use Committee. Male and female 4–6 months old C57/B6 mice were used as mid-life adult animals for all experiments, except for the aging experiment (Figures 7–9), where only 20-month-old female mice were available.

Figure 7.

GFAP-labeled reactive astrocytes spread throughout the MCA territory after MSCI in aging animals. (A,B) Example GFAP-labeled coronal mouse brain section 7-days after a sham (A) or MSCI (B) in 20-month-old animals. Top image shows the infarct border depicted by dotted line and numbered regions indicate higher magnification images shown below. (C) Global GFAP intensity ratio (ipsilesional:contralateral) from each region from sham (white bars) or MSCI (red bars) animals. Red asterisks denote significance compared to 1.0 (no difference between hemispheres). Black asterisks denote significant difference between conditions. (D) Density of GFAP-positive astrocyte cell bodies in sham (white), contralateral (grey), and ipsilesional (red) regions. Black asterisks denote significant difference between groups. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05, **p < 0.01. Scale bars = 50 µm.

Figure 8.

Vimentin-labeled reactive astrocytes spread to the MCA territory after MSCI in aging animals. (A,B) Example vimentin-labeled coronal mouse brain section 7-days after a sham (A) or MSCI (B) surgery in 20-month-old animals. Top image shows the infarct border depicted by dotted line and numbered regions indicate higher magnification images shown below. (C) Global vimentin intensity ratio (ipsilesional:contralateral) for each region from sham (white) or MSCI (red) animals. Red asterisks denote significance compared to 1.0 (no difference between hemispheres). Black asterisks denote significant difference between conditions. (D) Density of vimentin-positive astrocyte cell bodies in sham (white), contralateral (grey), and ipsilesional (red) hemispheres. Black asterisks denote significant difference between groups. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bars = 50 µm.

Figure 9.

Iba1 reactive microglia increase near the infarct after MSCI in aging animals. (A,B) Example Iba1-labeled coronal mouse brain section 7-days after a sham (A) or MSCI (B) surgery in 20-month-old animals. Top image shown the infarct border depicted by dotted line and numbered regions indicate higher magnification images shown below. (C) Global Iba1 intensity ratio (ipsilesional:contralateral) for each region from sham (white) or MSCI (red) animals. Red asterisks denote significance compared to 1.0 (no difference between hemispheres). Black asterisks denote significance between conditions. (D) Density of Iba1-positive microglia cell bodies in sham (white), contralateral (grey), and ipsilesional (red) hemispheres. Black asterisks denote change between groups. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05, **p < 0.01. Scale bars = 50 µm.

2.2. Transient MCA Occlusion (MCAo)

MCA occlusion was performed following previously reported methods (Li et al., 2021; Longa et al., 1989). For Figure 1, both 60-min and 30-min occlusions were performed, while the remaining study used only 30-min occlusions or sham surgeries. Prior to the occlusion, a laser Doppler flowmeter (Moor Instruments) probe was affixed over the right parietal bone overlying the MCA territory to monitor changes in cerebral blood flow. A ventral midline incision was made over the neck, the right common carotid artery (CCA) bifurcation was exposed by gentle dissection and tissue retraction, and the external carotid artery (ECA) was permanently ligated distal to the occipital artery using electrocautery. The right ECA and internal carotid artery (ICA) were temporarily closed with reversible slip knots before an arteriotomy was made in the ECA stump. A 5.0 nylon (silicone-coated) monofilament (Doccol Corporation) appropriate for the weight of the mouse was inserted into the ICA via the arteriotomy and gently advanced to the ICA/MCA bifurcation to occlude cerebral blood flow to the MCA territory, confirmed by a drop in flow measured by laser Doppler flowmetry. After a 30- or 60-min occlusion, the filament was gently retracted to allow reperfusion, the ECA permanently ligated, the slip knot on the CCA removed, and the incision sites sutured closed. The sham surgery involved exposure of the CCA and ECA without any arteriotomy or ligations. Mice remained under anesthesia for the entire 30 min and then the incision site was closed. Sham and MCAo mice were allowed to recover for up to 9 months as indicated by the experimental design.

Infarct size was measured 24 h after MCAo in 2 mm thick coronal brain sections (four total) using 2,3,5-triphenyltetrazolium chloride (TTC) staining and digital image analysis. Sections were incubated in 1.2% TTC in saline for 15 min at 37 °C and then fixed in formalin for 24 h.

2.3. PLX3397 Administration

All specialty feed was produced by ResearchDiets. Animals were fed ad libitum and monitored daily for consumption. Open Standard Diet from ResearchDiets (Cat. No. D11112201) was impregnated with Pexidartinib (also called PLX3397, Cat. No. C-1271, Chemgood) at a concentration of 290 mg/kg (Elmore et al., 2014). The control diet was a nutrition-matched Open Standard Diet. Separate cohorts of mice were fed the PLX3397 impregnated or matching control diet for 3 weeks prior to stroke induction and remained on the same diet until euthanasia and tissue collection.

2.4. Perfusion

Mice were transcardially perfused with 1% heparinized saline via a needle placed in the left ventricle, followed by fresh-made room temperature 4% paraformaldehyde (PFA) prepared in phosphate-buffered saline (PBS). Following complete fixation (confirmed by drainage of blood from the cut right atrium, visualization of liver paling, and stiffness of the body), brains were extracted and placed in 4% PFA overnight at 4 °C before storing in PBS for subsequent immunohistochemistry analysis.

2.5. Immunohistochemistry

Perfusion-fixed brains were paraffin-embedded, 6-µm sections were cut, antigen retrieval was performed using standard methods (30 min incubation in citrate buffer, pH 6.0, at 80 °C), tissue sections were blocked with 3% nonfat dry milk in PBS and immunolabeled with primary antibodies: rabbit anti-GFAP, 1:500 (Cat. No. 16825-1-AP, Proteintech); rabbit anti-vimentin 1:500 (Cat No. 10366-1-AP, Proteintech); rabbit anti-Iba1, 1:5000 (Cat. No. 10904-1-AP, Proteintech). Appropriate secondary antibodies conjugated to horseradish peroxidase were then used and reacted with 3,3′-diaminobenzidine (DAB) for visualization. Cell nuclei were lightly counterstained with hematoxylin.

2.6. Axioscan Imaging

Immunolabeled slides containing whole brain sections were imaged on a Zeiss Axioscan.Z1 with either a 10×/0.45 NA objective and a Hitachi HV-F202 camera (resulting in a pixel size of 0.441 µm × 0.441 µm) or a 20×/0.8 NA objective under the same system (resulting in a pixel size of 0.220 µm × 0.220 µm).

2.7. Region of Interest and Image Selection

2.7.1. GFAP and Vimentin

All analysis was conducted in three regions: (1) the anterior cerebral artery territory of the cortex (termed ACA; a region that does not receive any ischemia, although it may see some alterations in blood flow due to system-wide effects of occlusion), (2) the middle cerebral artery territory of the cortex (termed MCA; a region that experiences transient ischemia), and (3) the intact, non-infarcted portions of the striatum (a region that experiences transient ischemia and borders the infarct). ROIs were selected while avoiding areas with visible MCA infarction, which was identified by loss of H&E labeling and visible cell death. GFAP-positive astrocytes also die within the infarct, providing additional confirmation.

2.7.2. Iba1

Four regions in the ipsilesional hemisphere were selected: (1) the ACA cortex, (2) the MCA cortex, (3) the intact, non-infarcted portion of the striatum (termed border striatum), and (4) the infarcted region of the striatum (termed infarct). In the contralateral hemisphere, analogous regions of the ACA cortex, MCA cortex, and striatum were selected. The same contralateral striatum regions were used for comparison with both the border (3) and the infarcted (4) ipsilesional striatum. The infarct could be identified by high Iba1 signal intensity but was defined as the region without GFAP-labeled cells in the adjacent section (as astrocytes also die within the infarct) with a border of GFAP positive cells to avoid subjective bias.

Regions were identified based on structural morphology landmarks in H&E staining using the Allen Brain Reference Atlas. All reasonable effort was made to match coronal sections between animals and across labels. For each brain region, three representative regions of interest (ROIs; 221² µm) were selected per animal, except in cases where tissue integrity was compromised, in which case only two ROIs were used (which occurred in a few sections). The values from these three ROIs were averaged to obtain one representative value per brain region, per animal. The same ROIs were used for both intensity and cell count analyses.

2.8. Intensity Analysis

ImageJ software (Fiji) was used for signal intensity analysis. High-resolution images were converted to grayscale and inverted such that positive pixels were displayed in white. The average raw intensity for each ROI was measured using the intensity measure tool. The global intensity values from all ROIs from each region were averaged to determine the overall signal intensity of that region for each animal. To overcome batch differences in labeling, the ipsilesional intensity values were normalized to the contralateral intensity values within each section. Thus, the ratio reflects the change in the ipsilesional hemisphere signal compared to the contralateral signal, which serves as a proxy for baseline intensity. A ratio of 1.0 indicates no change between hemispheres, values above 1.0 indicate an increase in intensity within the ipsilesional hemisphere.

2.9. Cell Counting

A cloud-based deep learning segmentation platform (Biodock, AI Software Platform, Biodock 2024. Available from biodock.ai) was used to count the number of cells positive for each marker (GFAP, vimentin, and Iba1) to quantify density of cell bodies. We trained this artificial neural network (ANN) on regions that were not selected for final analysis but were analogous and covered the range of label intensity and cell morphology. An iterative approach was used to accelerate the training process. Initially, a relatively small number of cells were traced by a human observer, these inputs were then used to train a preliminary model, and the predictions of the preliminary model were then further refined by a human expert, which then fed into the final training sets.

We trained three separate models for GFAP, vimentin, and Iba1-labeled cells. Biodock was trained on 1717 (GFAP; across 14 iterations) or 1603 (vimentin; across 11 iterations) labeled astrocytes, or 3258 labeled microglia (Iba1; across 8 iterations). Our models are publicly available at biodock.ai under the following names: (1) Stackhouse_Mishra GFAP+ astrocyte cell counter, (2) Stackhouse_Mishra vimentin + astrocyte cell counter, (3) Stackhouse _Mishra Iba1+ microglia cell counter.

To avoid including small spurious false positives, the output cell counts were further refined based on an area cutoff, which was determined by the typical segmentation area per cell reported by a human expert observer. Cells below the following area were excluded: 39 μm² (GFAP and Iba1 stains) or 78 μm² (vimentin). We validated each model by comparing Biodock-generated cell counts to manual counts conducted by three independent blind experts (co-authors TLS, BDM, and AM) on a randomly selected and de-identified subset of test images (Supplementary Figure 1). We concluded training when ≥90% of the difference between the ANN and human experts fell within the 95% confidence interval (SF1, 1.96 ± SD, dotted lines). We reported the density of positive cells as the count for each ROI normalized to the area (cells/mm²), averaged for each animal.

2.10. Statistical Analysis

Since all groups had an n < 20, we tested for normality and difference of variance to determine appropriate statistical tests. For intensity ratios, we tested each ratio using a one-sample t-test against the null hypothesis 1.0, which would indicate no change in label intensity in the ipsilesional hemisphere compared to the contralateral side. In figures, a red * indicates significance with a one-sample t test. We followed a decision tree for testing statistical significance between groups. For paired groups (ipsilesional vs. contralateral), a paired t test was performed when normality and variance assumptions were met. When normality and variance assumptions failed, a Wilcoxon ranked test was run. For unpaired group comparisons (stroke vs. sham), a Student’s t-test was performed when normality and variance assumptions were met, and a Mann–Whitney test was run when they were not met. For comparison of more than two groups, an ANOVA was used. In the case of normality and equal variance, a one-way ANOVA was performed. In the case of normality with unequal variance, a Welch’s ANOVA was performed. When normality failed, a Kruskal–Wallis test was performed. ANN method bias (Supplementary Figure 1) was evaluated using a Bland–Altman plot followed by one-sample t-test against the null hypothesis of calculated bias being 0.0. Significance was defined as p-values below 0.05 unless otherwise indicated. Infarct size and CBF data in Figures 1 and 5 are displayed as mean ± SEM. All other data are displayed as median ± IQR. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 5.

Microglia depletion increases infarct size after 30-min MCAo. (A) Schematic depicting experimental timeline for microglia depletion (Created in BioRender. Stackhouse, T. (2026). https://BioRender.com/9g968o7). Mice were fed chow containing 290 mg/kg Pexidartinib (PLX3397), a CSF1R inhibitor, or nutrition matched control chow, for 3–4 weeks. Two cohorts of animals were used, one for infarct assessment 24-hrs after 30 min MCAo and the second for histology 7 days after 30 min MCAo. (B) Normalized blood flow in the MCA territory (measured by laser Doppler flowmetry) before, during, and after occlusion in control and PLX3397-treated mice. (C) Example TTC stained brain sections from control or PLX3397-treated animals 24 hrs after they were exposed to 30 min MCAO. TTC stains metabolically active tissue but not infarcted regions (white arrowheads). (D) Quantification performed 1-day after MCAo shows a significant increase in infarct area in overall hemisphere, striatum, and cortex in PLX3397-treated mice compared to controls (black bars). Data displayed as mean ± SEM. Each dot represents an animal. *p < 0.05.

3. Results

3.1. Optimizing a Mild Stroke Model for Chronic Studies

We tested MCA occlusion (Figure 1(A)) durations shorter than 60 min and found that a 30 min MCAo results in a subcortical infarct confined mainly to the striatum while sparing the cortex, even though the decrease in MCA flow is similar during occlusion (Figure 1(B–D)). The striatal infarct area induced by 30 min MCAo (57.4 ± 2.6%) was similar to 60 min MCAo (59.0 ± 7.2%; p = 0.99), but the cortical infarct area was much smaller (30 min: 6.8 ± 7% vs. 60 min: 34 ± 8.4%; p = 0.003), resulting in an overall smaller hemispheric infarction (30 min: 11.18 ± 2% vs. 60 min: 28 ± 5.8%; p = 0.009). The variability in infarct size observed in both 60 and 30 min MCAo is likely due to variability in the circle of Willis, which is reported to occur in C57/B6 mice (Doyle et al., 2012). We refer to the 30-min MCAo as the mild sub-cortical infarct (MSCI) model.

3.2. Reactive Astrogliosis Spreads Throughout the MCA Territory in the First Week after MSCI

We next examined the extent and progression of reactive astrogliosis induced by MSCI during the first week by immunohistochemical labeling for GFAP and vimentin. GFAP is an intermediate filament protein that is expressed largely in astrocytes in the brain and is broadly used as an indicator of reactive astrocytes due to its increased detection following injury. In contrast, vimentin is minimally expressed in healthy mature astrocytes but is upregulated in reactive astrocytes (Escartin et al., 2021). We analyzed GFAP and vimentin in three regions: the ACA cortex (which experiences no blood flow reduction during MSCI), the MCA cortex (which experiences blood flow reduction but minimal infarction during MSCI), and the striatum (which experiences both blood flow reduction and infarction during MSCI) at 1, 3, and 7 days after MSCI (Figures 2 and 3).

Figure 2.

GFAP-labeled reactive astrocytes spread throughout the MCA territory in the sub-acute phase after mild subcortical infarct (MSCI). (A–C) Example coronal mouse brain sections immunolabeled for GFAP 1-day (A), 3-days (B), or 7-days (C) after MSCI. Top image shows the infarct border depicted by dotted line and numbered regions indicate the higher magnification regions shown below. (D) Global GFAP signal intensity ratio for the ACA cortex, MCA cortex and striatum at each timepoint (average intensity of 3 ipsilesional ROIs normalized to the average of the 3 contralateral ROIs for each region). Red asterisks denote significance compared to 1.0 (no difference between hemispheres). (E) Density of GFAP-positive astrocyte cell bodies in the ACA cortex, MCA cortex and striatum at each timepoint. Black asterisks denote significant difference between hemispheres. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05. Scale bars = 50 µm.

Figure 3.

Vimentin-labeled reactive astrocytes increase in the sub-acute period after MSCI. (A–C) Example coronal mouse brain sections immunolabeled for vimentin 1-day (A), 3-days (B), or 7-days (C) after MSCI. Top image shows the infarct border depicted by dotted line and numbered regions indicate higher magnification images shown below. (D) Global vimentin signal intensity ratio for the ACA cortex, MCA cortex and striatum at each timepoint (average intensity of 3 ipsilesional ROIs normalized to the average of the 3 contralateral ROIs for each region). Red asterisks denote significance compared to 1.0 (no difference between hemispheres). (E) Density of vimentin-positive astrocyte cell bodies in each region at each timepoint. Black asterisks denote significant difference between hemispheres. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05, ****p < 0.0001. Scale bars = 50 µm.

To assess changes in GFAP in the ischemic hemisphere, we calculated the ratio of ipsilesional to contralateral GFAP intensity signal (a ratio of 1.0 indicates no change) as well as the number of GFAP positive cells present in each region. GFAP intensity increased in the MCA cortex and striatum within 1 day and became more pronounced by 7 days after MSCI (Figure 2(A–D)). The number of GFAP positive cells also increased in both the MCA cortex and striatum by 3 days and remained until 7 days after MSCI (Figure 2(E)). The ipsilesional ACA cortex remained unchanged during the first 7 days after MSCI.

Vimentin followed a similar pattern. The vimentin labeling intensity ratio (ipsilesional:contralateral) increased in the MCA cortex and striatum, but not the ACA cortex, 1 day after MSCI and remained elevated at 7 days, but with evident variability at 3 days (Figure 3(A–D)). The number of vimentin-positive cells increased within MCA cortex by day 3 and remained high until day 7 (Figure 3(E)). Increased vimentin-positive cells in the striatum was pronounced, but not statistically significant (p = 0.058) at 3 days. At 7 days, the striatum showed the strongest increase in vimentin positive cells (Figure 3(E)). As with GFAP, no changes were observed in the ACA region of the cortex. Together, these data suggest that the reactive astrogliosis response begins within the first day after a subcortical stroke and progressively increases throughout the first week, or sub-acute phase.

3.3. Microglia Reactivity Is Restricted near the Infarcted Region after MSCI

Microglia play a crucial role in injury response and recovery, including after ischemic stroke (Gao et al., 2023; Paolicelli et al., 2022). Microglia migrate to and proliferate at the site of injury; therefore, we considered four regions for analysis: the ACA cortex, the MCA cortex, the striatum border, and the infarcted striatum. The infarct border was identified using H&E staining and confirmed by the absence of GFAP labeling in adjacent sections (dotted lines, Figure 4(A–C), compare to Figure 2(A–C)).

Figure 4.

Microglia reactivity increases near the infarct in the sub-acute phase after MSCI. (A–C) Example coronal mouse brain sections immunolabeled for Iba1 1-day (A), 3-days (B), or 7-days (C) after MSCI. Top image shows the infarct border depicted by dotted line and numbered regions indicate higher magnification images shown below. (D) Global Iba1 intensity ratio for the ACA region, MCA region, and striatum at each timepoint (average intensity of 3 ipsilesional ROIs normalized to the average of the 3 contralateral ROIs for each region). Red asterisks denote significance compared to 1.0 (no difference between hemispheres). (E) Density of Iba1-positive microglial cell bodies in each region at each timepoint. Both the border and infarct regions of the ipsilesional striatum were compared to the contralateral striatum. Black asterisks denote significant difference between hemispheres. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05, **p < 0.01, ***p < 0.001. Scale bars = 50 µm.

The Iba1 intensity ratio did not increase within the infarct until 7 days after MSCI (Figure 4(D)). The striatum shows a slight decrease in intensity within the infarct border region of the striatum at 1 day MSCI but trended to an increase in intensity throughout the first week. Although the ACA cortex showed a statistically significant change at 1 and 3 days, the change was minimal and significance attributable to the low variability. Accordingly, the number of Iba1-positive microglia remained unchanged in the ACA and MCA cortices but increased within the striatum, with the highest numbers of Iba1-positive cells observed in the infarct at 7 days post-MSCI (Figure 4(E) and Supplementary Table 1). These data suggest that the microglial reactivity response is largely localized to the infarcted region. Overall, the widespread changes, both regionally and temporally, in astrocyte reactivity was not reflected by microglia in our model.

3.4. Microglia Limit Infarct Size but Are Not Necessary for Reactive Astrogliosis after Stroke

The delayed and spatially restricted microglia reactivity we observed, in contrast to the widespread reactive astrogliosis, indicates that post-ischemic astrogliosis may not be initiated by microglial signals, as has been shown in other pathologies (Liddelow et al., 2017; Shinozaki et al., 2017). To investigate the relationship between reactive microglia and astrocytes after stroke, we depleted microglia in our MSCI model using Pexidartinib (PLX3397), a colony-stimulating factor 1 receptor (CSF1R) antagonist. CSF1 regulates the survival, proliferation, and differentiation of monocyte and macrophage lineage cells (Fujiwara et al., 2021; Hagan et al., 2020), and CSF1R inhibition depletes microglia (Elmore et al., 2014). To examine the microglial contribution to astrogliosis after MSCI, we exclusively fed mice chow containing PLX3397 (290 mg/kg) continuously for 3 weeks to deplete microglia, performed either 30 min MCAo or sham surgeries, and continued them on the PLX3397 diet for another week. Since the 7-day timepoint revealed the most robust microglia response, we harvested tissue and examined reactive microglia and astrocytes at 7 days post injury (Figure 5(A)).

We observed that infarct size following 30-minute MCAo was significantly larger in mice that were fed PLX3397-impregnated chow, with infarcts extending into the cortex compared to those fed the control diet (Figure 5(C,D)), despite the MCAo producing a similar decrease in blood flow during the occlusion (Figure 5(B)). As expected, PLX3397 treatment drastically depleted microglia (Figure 6(C)) and eliminated hemispheric differences in Iba1 labeling intensity after MSCI across all regions (Figure 6(D), right panel and Supplementary Table 2). A small increase in Iba1-positive cell count was observed within the infarcted striatum (Figure 6(E), right panel and Supplementary Table 2), but these cells were sparse (only 12.7% of the number of cells counted in mice with intact microglia; Figure 4(E)). Although a robust decrease in microglia was achieved, the presence of Iba1-positive cells in the infarct indicates that the remaining microglia are still able to respond to ischemic injury.

Figure 6.

Microglia depletion does not abolish astrogliosis. (A–C) Example GFAP (A), vimentin (B) or Iba1 (C) immunolabeled coronal brain sections 7-days after MSCI in PLX3397-treated mice. Top image shows the infarct border depicted by dotted line and numbered regions indicate higher magnification images shown below. (D) Global intensity ratio (ipsilesional:contralateral) shown by region (ACA cortex, MCA cortex, and striatum) for astrocyte (GFAP, vimentin) or microglia (Iba1) immunolabels in PLX3397-treated sham (white) and MSCI (red) animals. Red asterisks denote significance compared to 1.0 (no difference between hemispheres). Black asterisks denote significant difference between groups. (E) Density of cells positive for astrocyte (GFAP, vimentin) or microglia (Iba1) immunolabels. Black asterisks denote significant difference between groups. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bars = 50 µm.

We next examined reactive astrogliosis at 7 days after MSCI, when astrogliosis was most pronounced in control mice (Figures 2 and 3), in mice lacking microglia. We observed an increase in both labeling intensity and number of positive cells for both GFAP and vimentin in the ipsilesional MCA cortex and striatum compared to analogous regions in the contralateral hemisphere and animals fed the control diet (Figure 6(D,E), left and middle panels). As in animals with intact microglia, these changes were significant by 3 days and remained high until 7 days post-MSCI. These data indicate that microglia are important in limiting the infarct size (Figure 5), but that stroke-induced reactive astrogliosis remains largely intact even when microglia are severely depleted.

3.5. MSCI-Induced Reactive Astrogliosis and Microglial Response Remain Intact in Aged Animals

Age is a key factor in stroke recovery, with older patients generally experiencing worse outcomes after an ischemic stroke (Benjamin et al., 2018). We examined whether age influences the progression of reactive astrogliosis and microglia responses by inducing MSCI in ∼20-month-old mice. At baseline, we observed a higher number of GFAP-positive cells in both the ACA and MCA cortices of the contralateral hemisphere of aged MSCI mice, as well as in those that underwent a sham procedure, compared to young mice (Figure 7 and Supplementary Table 3). At 7 days after MSCI, the GFAP intensity ratio (ipsilesional:contralesional) increased in the ACA and MCA, with the striatum showing a similar trend (Figure 7(C)). Additionally, GFAP-positive cell counts increased, particularly in the striatum (Figure 7(D) and Supplementary Table 3). We observed a similar increase in vimentin signal, with intensity increasing in the striatum and cell counts increasing in both the MCA cortex and the striatum (Figure 8(A–D)). Iba1 labeling also appeared stronger in aging mice at baseline (Figure 9 and Supplementary Table 3). At 7 days post-MSCI, Iba1 labeling intensity and the number of Iba1-positive cells increased within both the striatal infarct and infarct border (Figure 9(C,D)). Overall, these data suggest that, despite a higher baseline level of GFAP and Iba1, reactive astrogliosis and microglia response to stroke remain intact in older animals.

3.6. Reactive Astrocytes Persist for at Least 9 Months after MSCI

In a subset of animals, we induced MSCI at 6 months of age and examined reactive astrogliosis at a chronic time point using immunohistochemistry against GFAP 9 months after injury (Figure 10), when the mice were 15 months of age. Even at this chronic time point, we observed higher GFAP intensity in the ipsilesional striatum (Figure 10(C)), and an increase in the number of cells positive for GFAP in both the striatum and the MCA regions (Figure 10(D)). This suggests that reactive astrocytes may persist for many months after stroke, with implications for their functional roles within the neuroglial and neurovascular interfaces.

Figure 10.

Reactive astrocytes remain chronically elevated in the MCA territory up to 9 months later. (A,B) Example GFAP-labeled coronal mouse brain sections (top) 9 months after a sham (A) or MSCI (B) surgery, with numbered regions indicating higher magnification images shown below. (C) Global GFAP intensity ratio (ipsilesional:contralateral) from each region from sham (white bars) or MSCI (red bars) animals 9 months after surgery. Red asterisks denote significance compared to 1.0 (no difference between hemispheres). Black asterisks denote significance between groups. (D) Density of GFAP-positive astrocyte cell bodies in each region of sham (white) mice and the contralateral (grey) and ipsilesional (red) hemispheres of MSCI mice, 9 months after surgery. Black asterisks denote significant difference between groups. Data displayed as median (IQR). Each dot represents an animal. *p < 0.05, ***p < 0.001, ****p < 0.0001. Scale bars = 100 µm.

4. Discussion

We report an optimized model of MCAo with a reduced occlusion period of 30 minutes. This model provides a clinically relevant small subcortical infarct (Figure 1) for studying the effects of stroke in regions distant from the infarct, including an intact cortex. The MCA cortex is particularly relevant in this model, as it experiences decreased blood flow without infarction and hence may represent a model of transient ischemic attack. As mice recover well from this shorter occlusion and survive for at least 9 months after ischemia (Figure 10), this model shows promise for exploring chronic post-stroke outcomes. Additionally, provided there is careful control of age and other variables that influence infarct size, it may be suitable for studying striatal-specific strokes while minimizing extensive extra-striatal infarction.

Our analysis of gliosis using this small stroke model demonstrates that reactive astrogliosis is more pronounced at earlier timepoints after minor ischemic injuries compared to microglia reactivity (Figures 2–4). Importantly, reactive astrogliosis is more widespread, extending into the MCA region of the cortex, whereas microglia reactivity remains largely restricted to the area of infarct and its surrounding border regions. Additionally, despite the physical proximity of MCA and ACA regions, reactive astrogliosis in the cortex remains largely restricted to the MCA territory during the sub-acute phase, suggesting that it may primarily be a response to the transient ischemia or hypoperfusion experienced by the tissue itself rather than by diffusible signals released from the injured tissue. Future long-term studies may shed light on whether diffusible signals contribute to the spread of gliosis at more chronic timepoints. The widespread nature of reactive astrogliosis, especially at sub-acute and chronic timepoints, raises new questions. How are the broader astrocyte functions, including neuronal circuit modulation, metabolic and ionic homeostasis, and cerebrovascular regulation altered in these distal regions after a small localized ischemic injury, and how do they evolve over time? These are important avenues for future research.

Our study focuses primarily on the markers that report structural manifestations of reactive astrogliosis and does not evaluate potential changes in astrocyte functional states or phenotypic heterogeneity. These measures may also not encompass all reactive astrocyte populations, including any population not immunoreactive for vimentin or GFAP (Escartin et al., 2021). Reactive astrogliosis is a complex process involving morphological, transcriptional, translational, and functional changes in astrocytes (Escartin et al., 2021). Thus, relying on these markers provides only a limited understanding of astrogliosis. Nevertheless, immunolabeling for specific markers represents a first step in identifying distal brain regions potentially affected by subcortical stroke for further functional investigations. Although GFAP is widely used as a marker for reactive astrocytes, its baseline expression is naturally higher in some regions (e.g., white matter, hippocampus) compared to others (e.g., cortex) (Kimelberg, 2004) and can also be increased by physiological processes (Escartin et al., 2021). We suggest that vimentin, which is low in mature grey matter astrocytes but highly induced following injury (Figure 3), may serve as a more reliable reporter than GFAP for assessing astrogliosis particularly in the sub-acute phase. Although the level of vimentin expression observed varied considerably between animals, the induction of its expression itself served as a strong indicator of injury.

In our study, Iba1 labeling did not change immediately within the infarcted region 1 day after stroke but was robustly increased by 7 days. This was unexpected, as previous studies have reported rapid recruitment and proliferation of microglia at the site of infarct or ischemic injury (Hu et al., 2012; Huang et al., 2023). Several factors could explain this discrepancy, including the smaller stroke size in our model or the possibility that microglia responding to the ischemic injury do not immediately increase Iba1 expression. Indeed, Iba1 labeling showed less ramified, more amoeboid cells in the striatal infarct and border regions even 1 day after stroke (Figure 4(A)). Given that Iba1 is not the only marker of reactive microglia, other populations might be overlooked in this study. Further, blood-brain barrier impairment in the infarct leads to blood monocytes infiltrating the brain, where they become macrophages that adopt microglial identity and properties (Han et al., 2020). The large increase in Iba1 positive cell population likely reflects both locally migrating microglia and these infiltrating macrophages, which are also not discerned in our study.

The striking increase in infarct size in mice depleted of microglia (Figure 5), similar to previous findings (Huang et al., 2023), indicates that microglia are crucial for maintaining the infarct border and limiting injury. Microglia are thought to be necessary for the induction of border (scar) forming reactive astrocyte subtypes, which protect against neurotoxic inflammation (Sofroniew, 2015). Although we observed robust reactive astrogliosis after microglia depletion (Figure 6), their ability to form the injury border may be impaired in the absence of microglia. Other key microglial functions such as debris clearance, reactive oxygen species sequestration, or inflammatory signaling regulation (Wang et al., 2022; Xiong et al., 2016), might also contribute to the increased infarct size in the microglia depleted condition. Importantly, our observation that reactive astrogliosis persists in microglia-depleted mice suggests that it is not solely dependent on signals from microglia in ischemic stroke, unlike in infection-like contexts (Liddelow et al., 2017). Instead, astrogliosis following mild ischemia may be influenced directly by factors such as reduced blood flow during stroke, hypoxia, metabolic changes, oxidative stress, or signals released from injured tissue. In the microglia-depleted condition, the increased infarct area may in fact result in alternate signals that induce reactive astrogliosis. It is possible that the small number of residual Iba1+ cells remaining in the PLX3397 treated mice, as also reported by others (Okojie et al., 2023), were enough to trigger reactive astrogliosis, although it is unlikely given the sparsity of these remaining microglia (Figure 6(C–E)). Single-cell analyses have identified multiple subclusters of reactive astrocytes after LPS exposure (Hasel et al., 2021) and in the days and weeks following a stroke (Guo et al., 2021; Zhang et al., 2022). It is possible that only certain subsets of reactive astrocytes are dependent on signals from microglia (potentially border/scar forming subtypes), while others, as observed here, are not. Future single-cell transcriptomic and proteomic analysis of mild stroke tissues will help define the heterogeneity of reactive astrocytes and microglia induced after stroke, while analysis of the stroke-induced reactive astrocytes under microglia intact and depleted conditions could elucidate the exact nature of reactive astrocyte populations that do or do not depend on microglial signals. These approaches will also help prioritize potential mechanisms that drive widespread and persistent astrogliosis following mild ischemia (e.g., cytokine signaling, blood-brain barrier disruption) as well as their functional consequences for further investigation. These are intriguing avenues for future research.

Reactive astrogliosis and reactive microglia may be “primed” for increased reactivity based on many environmental factors, including age. Indeed, we observed an increase in GFAP and Iba1 in aged mice even at baseline (although direct statistical comparisons were not performed as the tissues were processed in separate batches). This finding generally aligns with previous characterizations of astrocytes (Clarke et al., 2018; Suda et al., 2021) and microglia (Suda et al., 2021) in aging. Despite these baseline differences, the MSCI-induced increase in both reactive astrocytes and microglia, as well as their spatiotemporal patterns, remained similar between aged mice and middle-aged adult mice (Figures 7–9). However, as our aging cohort was limited to a small sample of female mice only, these data should be interpreted with caution. Age could also impact the size of the infarct, which in turn could affect the extent of reactive astrogliosis or microglia reactivity. Although infarct size in aged mice was not quantified in this study, we observed that one of the aged mice exhibited an infarct that spread even to the hippocampus (example shown in panel B of Figures 7–9), an area that is normally supplied by the posterior cerebral artery rather than the MCA. Whether this is due to aging itself or whether this mouse had a misformed arterial structure in which the MCA supplied the hippocampus is unknown. A related limitation of our study is that all experiments were conducted in C57/B6 mice, wherein the circle of Willis can be variable (Doyle et al., 2012), which may have contributed to the larger infarct in this particular mouse. The effect of age on the impact of stroke presents an intriguing avenue for future research, as understanding the relationship between age, sex, infarct size, and reactive glial responses could yield significant insights into stroke pathology and recovery, as well as the underlying causes of stroke-induced dementia.

Importantly, we found that reactive astrogliosis persisted chronically in the ipsilesional hemisphere up to 9 months after mild stroke even though the infarct itself had been resolved. Although our present study is limited to an anatomical, spatiotemporal analysis based on astrocyte cytoskeletal proteins, it is important to examine the full extent of changes occurring in these persistent reactive astrocytes and their functional impact on neurovascular, neurophysiological, cognitive, and other post-stroke outcomes. Stroke patients continue to exhibit neurovascular deficits many years after the initial stroke (Krainik et al., 2005) and have a higher risk of dementia (Miklossy, 2003). It is tempting to hypothesize that mild reactive astrogliosis is a semi-permanent state following ischemic injury driving these deficits. Further investigations, including functional consequences and analysis of heterogeneous reactive astrocyte subtypes, are necessary to better understand the pathophysiological relevance of elevated GFAP+ cells in the ipsilesional hemisphere at chronic timepoints.

Conclusion

Here we describe the MSCI model suitable for chronic studies on the effects of ischemic stroke. We show that, in the sub-acute phase (up to 7 days after MSCI), reactive astrogliosis precedes reactive microglia, and this response is restricted primarily to the arterial territory that experienced ischemia. We determined that stroke-induced reactive astrogliosis is, to a great extent, not dependent on microglia. We show that aging itself is associated with some increase in astrocyte and microglia markers, but the response of both cell types to stroke is conserved, at least during the sub-acute period. Finally, we show that reactive astrogliosis persists at chronic post-stroke timepoints, for at least 9 months. Our findings raise several important questions: what signals induce astrocyte and microglia reactivity after ischemic stroke, and how do they diverge? What is the heterogeneity of reactive microglia and astrocyte populations induced by mild ischemia? How and when do reactive astrogliosis and microglia resolve, and might there be permanent changes in astrocyte or microglia state after injury? And perhaps most importantly, what are the functional consequences of widespread reactive gliosis during the sub-acute and chronic phases, and how do they contribute to neurodegeneration, such as in post-stroke dementia? These should be the foci of future inquiries.

Funding Statement

This work was funded by a Lacroute fellowship (TLS), JTMF Foundation award (AM) and the following NIH grants: NIA NRSA T32AG055378 (TLS), NHLBI NRSA T32HL094294 (HM), NINDS R01NS110690 and R01NS134592 (AM), and NIA P30AG066518 (AM, RW).

Ethics statement

The animal study was reviewed and approved by the Oregon Health & Science University Institutional Animal Care and Use Committee.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data presented in this study are available upon reasonable request.