Environmental enrichment modulates chronic poststroke inflammation and links white matter TREM2‐positive microglia in recovery in mice

Abstract

Background

Environmental enrichment (EE) has been widely reported to improve functional recovery after stroke, yet its effects on chronic inflammation and white matter pathology remain less well defined. The aim of this study was to investigate whether poststroke EE modulates chronic neuroinflammation and white matter pathology and how these changes relate to functional recovery.

Methods

Here, we investigated how EE influences the microglial response and myelin integrity during the recovery phase after photothrombotic stroke in male C57Bl/6J mice. Mice were randomized by using a computer‐generated sequence into standard environment (SE) or EE housing conditions 2 days poststroke (SE: n = 6; EE: n = 7 for histological analyses; behavioral data unavailable for 2 SE mice). Sensorimotor performance was assessed at 7, 14, and 21 days poststroke using paw placement, foot fault, and limb symmetry tests, integrated into a composite Neuroscore. Housing in EE improved and sustained behavioral recovery up to 21 days. To explore whether specific microglial subpopulations are involved in tissue reorganization and functional outcome, immunofluorescence was performed for ionized calcium binding adaptor molecule 1 (Iba1), galectin‐3 (Gal3), purinergic receptor P2Y12 (P2RY12), cluster of differentiation 68 (CD68), and triggering receptor expressed on myeloid cells 2 (TREM2), together with Black Gold myelin staining, in peri‐infarct and white matter regions. Statistical analyses included correlation analyses and analysis of covariance (ANCOVA).

Results

In SE mice, the inflammatory microglial marker Gal3 and the phagocytic marker CD68 correlated positively with infarct size (r = 0.937, p = 0.006; r = 0.845, p = 0.034, respectively), and the accumulation of myelin debris (r = 0.865, p = 0.026) together with loss of myelin coverage (r = −0.907, p = 0.013) followed lesion‐driven trajectories. By contrast, these correlations were absent in EE mice (r = −0.028, p = 0.953; r = 0.671, p= 0.099; r = 0.053, p = 0.910; r = 0.292, p = 0.525) indicating that enrichment attenuated lesion‐driven inflammatory responses and myelin damage. Importantly, Gal3 expression correlated specifically with peri‐infarct myelin debris (r = 0.82, p = 0.048), reinforcing its role as a pro‐inflammatory mediator. Finally, correlation analysis between Neuroscore and both microglial and myelin markers revealed a unique positive association between TREM2 coverage in white matter and improved Neuroscore in EE mice (r = 0.890, p = 0.007; ANCOVA p = 0.039).

Conclusion

These results support the idea that EE modulates processes of chronic post‐stroke inflammation, peri‐infarct myelin debris clearance and demyelination, and highlight white matter TREM2‐positive microglia as a potential cellular link to functional recovery.

Schematic overview illustrating the experimental design and main findings of the study. Following photothrombotic stroke, mice were housed in either standard environment or enriched environment. Environmental enrichment attenuated lesion‐driven microglial inflammation and myelin damage while promoting an association between white matter triggering receptor expressed on myeloid cells 2 (TREM2)‐positive microglia and improved functional recovery. CD68, cluster of differentiation 68; Gal3, galectin‐3; Iba1, ionized calcium binding adaptor molecule 1.

Highlights

• Environmental enrichment (EE) promotes sustained sensorimotor recovery after photothrombotic (PT) stroke in mice, independent of lesion size.

• EE dissociates the positive correlation between infarct size and pro‐inflammatory (galectin‐3) and phagocytic (cluster of differentiation 68) microglial markers observed in standard‐housed mice.

• The accumulation of peri‐infarct myelin debris and white matter demyelination is lesion‐driven in standard‐housed mice but attenuated by EE.

• Triggering receptor expressed on myeloid cells 2‐positive microglia in white matter are uniquely associated with improved functional recovery in environmentally enriched mice.

1. INTRODUCTION

Stroke remains the second leading cause of death worldwide after coronary disease, and its global burden in terms of mortality, disability, and cost is projected to double by 2050. ¹ Although some neurological functions recover, up to 50% of patients who have had a stroke remain chronically disabled, with motor deficits being particularly common. The early recovery period during the first weeks after the insult is a critical window that largely determines long‐term outcome. ² , ³

Physical activity has been associated with improved post‐stroke outcome in both clinical and experimental settings. ⁴ , ⁵ , ⁶ In animal models, environmental enrichment (EE), a paradigm combining voluntary exercise, novel objects, and social interaction; significantly improves recovery through, enhanced plasticity, but also reduced inflammation. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ In particular, enriched environment (EE) attenuates the number of toll‐like receptor 2 (TLR2) positive microglia ¹² and reduced the levels of pro‐inflammatory mediators in the ischemic territory. ¹¹ Results have been confirmed in adapted stroke and EE models. ¹⁴ The impact of cell genesis on EE‐induced recovery remains open and might depend on the experimental models used to assess neuronal progenitor cells. ¹⁵ , ¹⁶

Microglia, as the resident immune cells of the brain, play a central role in the inflammatory response in the ischemic territory. They contribute both to repair through debris clearance and to secondary damage through sustained proinflammatory activation. ¹⁷ The expression of specific markers such as galectin‐3 (Gal3), cluster of differentiation 68 (CD68), purinergic receptor P2Y12 (P2RY12), and triggering receptor expressed on myeloid cells 2 (TREM2) on microglia can provide insights into the balance between pro‐inflammatory, phagocytic, and homeostatic states. ¹⁸ , ¹⁹ , ²⁰ , ²¹ Although EE has been shown to improve behavioral outcome after stroke, ¹⁵ , ²² , ²³ , ²⁴ , ²⁵ specific cell populations underlying these effects, as well as the influence of EE on post‐ischemic inflammation have not been studied in detail. Moreover, the abundance of various microglial cell populations at later stages poststroke after the acute response has subsided and tissue reorganization processes dominate, has not been properly characterized.

Using established marker proteins of microglial cells, here we have investigated how EE affects the distribution of different microglia populations in the postischemic brain and analyzed their impact on lesion size, functional outcome, and tissue reorganization following experimental stroke induced by photothrombosis.

2. METHODS

2.1. Animals and experimental design

Male C57Bl/6J mice (Charles River Laboratories), age 10–12 weeks, body weight 23–27 g, were included and housed under a 12 h light/dark cycle with ad libitum access to food and water. On Day 2 following stroke, mice were randomized into standard environment (SE) or EE housing and kept in the assigned condition for 19 days. Randomization was performed using a computer‐generated sequence. On Days 7, 14, and 21 after the insult, neurological functions were tested as described below. ²⁶ Body weight was monitored throughout the study and did not show differences between animals in different housing conditions (Supporting Information S1: Figure 1A). In vivo studies were approved by the Malmö‐Lund review committee (ethical permit numbers: M 50‐15 and M 25‐12).

2.2. Photothrombotic stroke

C57Bl/6J male mice (Charles River) were used for this study. Mice were subjected to unilateral photothrombotic (PT) cortical infarct in the left sensori‐motor cortex as previously described in Quattromani et al. ¹² In brief, one intraperitoneal (i.p.) injection of 0.25 mL Rose Bengal (10 mg/mL in 0.9% saline; Cat# 330000; Sigma‐Aldrich) was administered 5 min prior to light illumination. An optic fiber (diameter 2.5 mm) was placed above the sensori‐motor cortex, 0.5 mm anterior to bregma and 1.5 mm lateral to midline. Thereafter, the brain was illuminated with cold light for 20 min (Schott KL1500 LCD). During surgery, the body temperature of the animals was kept at 37°C using a self‐regulating heating pad.

2.3. Housing environment

Two days after stroke induction, mice were allocated either to standard housing conditions (SE, n = 6; cage area 17.0 cm × 16.0 cm × 34.5 cm) or EE (EE, n = 7; cage area 30.0 cm × 27.0 cm × 43.0 cm). EE housing included a higher number of mice in the cage and contained various colored objects such as plastic tunnels, small houses, slides and rodent running wheels, with frequent changes in the positioning of housing contents every other day. ¹⁶

2.4. Assessment of neurological function

Paw placement test

Spontaneous placement and mobility of the impaired forepaw were scored on a scale from 0 to 5: 0, no movement, paw hanging down; 1, minimal horizontal movement; 2, horizontal and vertical movement below the table surface; 3, movement up to table surface level; 4, full mobility in all directions, occasional contact with surface; 5, normal, immediate placement on surface or no hanging observed.

Grid‐walking test (foot fault on grid)

Mice traversed a 46.5 cm grid (step size 2 mm, interspace 1.5 cm) after 2–3 training sessions. Left forelimb foot faults were counted over 27 steps and scored on a 0‐6 scale: 0, ≥6 faults; 1, 5 faults; 2, 4 faults; 3, 3 faults; 4, 2 faults; 5, 1 fault; 6, no faults.

Tail suspension test (asymmetry of limb movement)

Mice were suspended by the tail and forelimb posture and mobility were scored on a 0‐4 scale: 0, impaired paw immobile and tightly held to the trunk; 1, impaired paw mostly held to trunk, slight mobility; 2, impaired paw mobile but with marked asymmetry compared to contralateral paw; 3, slight asymmetry with both paws mobile; 4, normal mobility.

Scores from the three tests were combined into a single score (referred as Neuroscore), by summation of score from each test in each mouse. The experimenters were blinded against the experimental group during data recording and the primary data analysis. Training has been performed three consecutive days before induction of stroke on Day 0. On Day −1 baseline neurological scores were obtained from all mice. Mice not performing the tests with full scores were excluded at this time point and kept as companion animal in the cage. To ensure that mice assigned to standard or enriched cages started with comparable levels of functional deficit, selective sorting was performed on Day 2 poststroke, a time point at which detrimental cell death processes contributing to lesion size have been subsided. Only animals that exhibited severe deficits in the paw placement test (score = 0) were included in the study (one mouse in SE group was not included due to this). In total, 6 mice for SE and 7 mice for EE were analyzed for immunohistochemistry, but behavioral measurements were not obtained in two mice in SE group.

2.5. Tissue collection

At 21 days after surgery, mice were anesthetized with 2% isoflurane, and lethally injected with 10 mg/mL of chloral hydrate (i.p., 150 mg/kg). They were then perfused transcardially with 0.9% saline followed by ice‐cold phosphate‐buffered 4% paraformaldehyde solution (pH 7.4, paraformaldehyde [PFA]; Sigma Co.). Tissue samples were post‐fixed overnight in 4% PFA solution and equilibrated in phosphate‐buffered 30% sucrose solution for additional 48 h at 4°C. The brains were then sectioned on a microtome into 30‐μm‐thick coronal sections. These were collected in 8 subsequent series and stored at −20°C in an antifreeze solution (phosphate‐buffered 30% glycerol/30% ethylene glycol solution.

2.6. Immunofluorescence

Free‐floating brain sections (+0.96 mm to −0.04 mm relative to bregma) were initially rinsed three times in phosphate‐buffered saline (PBS) for 5 min, followed by another rinse round in PBS with 0.25% Tween‐20 (T‐20; Cat# P1379; Sigma‐Aldrich). After washing, the samples were blocked with a 10% normal donkey serum (NDS, Cat# S30; Millipore) in PBS 0.25% T‐20 blocking solution for 1 h at room temperature. They were then incubated at 4°C overnight with one of the following antibodies: Iba1 (1:500, rabbit, Cat# 016‐26721; Wako), Gal3 (1:750, goat, Cat# AF1197; R&D Systems) P2RY12 (1:200, rat, Cat# S16007D; Biolegend). Following overnight incubation, the samples were taken for an initial rinse with PBS 0.25% T‐20, followed by a 1‐h incubation in dark conditions with the respective secondary antibodies (1:500; donkey anti‐goat Alexa Fluor 488, donkey anti‐rat Alexa Fluor 555, donkey anti‐rabbit Alexa Fluor 647; all from Invitrogen; Cat# A11055, A48270, A31573, respectively) in 10% NDS blocking solution. Subsequently, they were incubated with 4′,6‐diamidino‐2‐phenylindole (DAPI, 1:1000, Cat# 62248; Invitrogen) in PBS 0.25% T‐20 for 5 min, followed by another round of rinse with PBS. Thereafter, sections were mounted onto slides (Invitrogen ProLong Glass Antifade Mountant, Epredia Superfrost Plus Adhesion microscope slides). The second staining protocol included a 48‐h incubation with the antibodies Iba1 (1:500, rabbit; Wako), TREM2 (1:200, sheep, Cat# AF1729; R&D systems), and CD68 (1:1000, rat, Cat# MCA1957; Biorad). Following two rounds of rinsing with PBS, and a 1‐h incubation with 10% NDS blocking solution in PBS, the samples were similarly incubated with the respective primary antibodies. Upon the completion of the 48‐h incubation, the sections were rinsed with PBS and incubated with the appropriate secondary antibodies (all 1:500; donkey anti‐rabbit Alexa Fluor 405, donkey anti‐sheep Alexa Fluor 555, donkey anti‐rat Alexa Fluor 647, all from Invitrogen, Cat# A48258, A21436, A48272TR, respectively). Upon 2‐h incubation with the secondary antibody solution and an additional rinse with PBS, the sections were incubated with Bodipy (1:1000, Invitrogen Cat# D3922) for 20 min and washed with PBS before mounting and coverslipping.

2.7. BlackGold II myelin staining

Myelin staining was performed using BlackGold II (Black‐Gold® II RTDTM kit; Biosensis), a fluorescent dye that selectively labels myelinated fibers. Free‐floating sections (−0.04 mm to −1.04 mm relative to bregma) were first rinsed in distilled water, placed onto slides and left to air‐dry at 50–60°C for 35 min until they were visibly dry. Subsequently, the sections were rehydrated with distilled water for 10 to 15 s. In a Coplin jar, the first incubation solution was prepared with 9 parts distilled water and 1 part Black‐Gold II and heated to 65°C in the incubator. Upon reaching the temperature, the slides were then placed in the Coplin jar for incubation in dark conditions, with microscopic monitoring of the labeling checked every 3 min. Upon a total incubation time of 24 min, the slides were then rinsed with water in a Coplin jar for 2 min. In a similar fashion to the Black‐Gold II labeling solution, the fixative solution was prepared with 1 part sodium thiosulfate and 9 parts distilled water in a clean Coplin jar. The slides were placed inside for a 3‐min incubation, followed by 3 rinses with distilled water in separate Coplin jars. Sections were dehydrated through series of ethylalcohol solutions starting with a 70% solution for 5 min, 90% for 3 min, 96% for 3 min, and 99.5% for 3 min. After immersion in xylene for 2 min, the slides were coverslipped with mounting medium (Permount; Electron Microscopy Sciences).

2.8. Image quantification

Immunofluorescence imaging was performed using a Nikon A1RHD confocal laser scanning microscope (Nikon Instruments) with ×20 magnification objective. Tile scans were taken for the peri‐infarct area and the surrounding tissues of interest, and three areas were analyzed: (1) the peri‐infarct area, delineated as the tissue spanning 400 μm outward from the infarct border across all directions using DAPI as reference; (2) the white matter region, which included the ipsilateral portion of corpus callosum and external capsule subjacent to the infarct zone; and (3) the ipsilateral striatum. A summary of markers analyzed in the different regions can be found in Table 1.

Table 1.

Overview of markers analyzed across different brain regions after photothrombotic stroke in male C57Bl/6J mice.

Region of interest	Iba1	Gal3	P2RY12	CD68	TREM2	Bodipy	BlackGold myelin
Peri‐infarct	x	x	x	x	x	x	x
White matter	x	x	x	x	x	‐	x
Striatum	x	x	x	‐	‐	‐	‐

Note: Summary of immunohistochemical and histochemical markers quantified in each region of interest. The peri‐infarct area was analyzed for all markers, including indicators of microglial activation (ionized calcium binding adaptor molecule 1 [Iba1], galectin‐3 [Gal3], cluster of differentiation 68 [CD68]), homeostatic microglia (purinergic receptor P2Y12 [P2RY12]), and reparative signaling (triggering receptor expressed on myeloid cells 2 [TREM2]), as well as lipid accumulation (4,4‐difluoro‐4‐bora‐3a,4a‐diaza‐s‐indacene [Bodipy]) and myelin integrity (BlackGold). In white matter, all markers except Bodipy were assessed, while the striatum was analyzed for selected microglial markers (Iba1, Gal3, P2RY12). x, analyzed; −, not applicable.

Image analysis was performed using Fiji by ImageJ software (version 2.13.1, National Institutes of Health, USA; https://imagej.net/software/fiji/). Positive signal cell coverage: Images were subjected to consistent binary thresholding and area quantification % for Gal3 (Iba1+), Iba1, and P2RY12 (Iba1+) signals across all regions of interest. For the staining of CD68, Bodipy, and TREM2, two neighbouring images subjacent to the border of the infarct core were quantified for peri‐infarct, and two additional images of white matter were quantified.

Infarct size measurements: To measure the area of the infarcted tissue, indirect lesion size was manually calculated delineating the affected ischemic area, the noninfarcted area of the ischemic hemisphere, and the contralateral hemisphere based on the morphology of DAPI‐positive nuclei and the resulting area was reported in mm².

Imaging of myelin‐stained slices were performed in brightfield technique using Zeiss Axio Imager M2 (×10, Zeiss Axiocam 305 color camera; Carl Zeiss AG), at 10% intensity. Upon acquiring the images, the slices were quantified in area coverage (%) for myelin abundance in the cortex tissue in both the ipsilateral cortex, as well as, at the contralateral side. Additionally, the degree of myelination (%) of the corpus callosum was measured in both sides. In addition, the myelin debris found at the immediate proximity of the infarct core was quantified using Fiji by ImageJ.

2.9. Microglia morphology

Images were skeletonized to extract cell complexity metrics (e.g., branch number, process length, endpoints) from Iba1‐stained cells using the AnalyzeSkeleton plugin in ImageJ. Due to the dense clustering and amoeboid microglia in the infarct border zone (located subjacent to the infarct core) this region could not be reliably analyzed for morphology or cell counting. Morphological analysis was therefore conducted in the neighboring lateral zone with well‐separated cells, though with a higher branching complexity, located adjacent to, but outside the dense reactive core, while keeping to the immediate vicinity of the infarct border. At the size of a 300 × 300 µm, the region was quantified in terms of total cells analyzed, and standard complexity metrics from individual skeletal analysis. For each animal, a total of 45 cells (15 cells per slice, 3 sections per animal) within the region were selected at random by a macro script (with a minimum 9 × 9 pixel threshold to avoid selection of smaller fragments & noise) and skeletonized and analyzed one by one, with manual confirmation of cell singularity. Of these cells analyzed, a threshold of <5 endpoints were set to classify microglia under less branched, reactive states, and the results were compared in both groups with standard statistical approaches.

2.10. Statistical analysis

All statistical analysis were performed using GraphPad Prism (v10.4.2, www.graphpad.com) and RStudio (version 4.4.3, R Core Team, https://cran.rstudio.com/bin/windows/base/). GraphPad Prism was used for correlation analysis and generation of linear regression plots, and RStudio to run aligned rank transform for analysis of variance (ART‐ANOVA) and analysis of covariance (ANCOVA) analysis (using a linear model including the interaction between variables). Specific statistical tests can be found in figure legends, and full ANCOVA reports can be found in Supporting Information S1: Table 1. For all rounds of analysis, a significance threshold of p < 0.05 was applied for the statistical tests.

3. RESULTS

3.1. Environmental enrichment improves and sustains sensorimotor recovery after photothrombotic stroke

We first assessed sensorimotor function in mice housed in EE (EE) or SE following PT stroke. Paw placement, foot fault, and limb symmetry tests were performed at 7, 14, and 21 days poststroke (Figure 1A). Mice in the EE group showed significantly improved performance across all three behavioral tests compared with SE mice, and these benefits were sustained up to 21 days poststroke (Figure 1B–D; Housing effect: p < 0.001 and p = 0.003). To integrate behavioral performance into a single metric, the three parameters were combined into a composite “Neuroscore.” Area under the curve (AUC) analysis of the Neuroscore confirmed a significant overall improvement in EE mice relative to SE mice (Figure 1E; Housing effect: p = 0.003; Figure 1F: p = 0.003).

Figure 1.

Sensorimotor testing and composite Neuroscore analysis after photothrombotic stroke in standard environment (SE) and environmental enrichment (EE) mice. (A) Schematic experimental outline of housing environment as well as behavioral testing. (B) Paw placement scores at Days 7, 14, and 21 post‐photothrombotic (PT). (C) Foot fault on grid scores at Days 7, 14, and 21 post‐PT. (D) Symmetry of limbs scores at Days 7, 14, and 21 post‐PT. (E) Neuroscore at Days 7, 14, and 21 post‐PT (addition of all three behavioral tests). (F) Area under the curve (AUC) of the composite Neuroscore. Values are expressed as individual experimental replicates with mean ± SEM. In (B–E) nonparametric two‐way analysis of variance (aligned rank transform for analysis of variance [ART‐ANOVA]) with post hoc comparisons using ART‐based contrasts was performed, with p‐values adjusted for multiple comparisons using the Holm method. Significant time/housing effects are shown. In (F) Mann–Whitney test was performed. n = 4 mice in SE and n = 7 mice in EE. Significant p‐values are expressed with 3 decimals.

3.2. Enriched environment changes the relationship between infarct size and microglial activation

We next examined whether the behavioral improvements in EE mice were accompanied by changes in the composition of microglial cell populations around the infarct. Immunofluorescence staining was performed for Iba1 (pan‐microglial marker), Gal3 (pro‐inflammatory marker), and P2RY12 (homeostatic marker). The peri‐infarct region was defined as the 400 μm zone surrounding the infarct core (Figure 2A,B). Consistent with previous studies, infarct size did not differ between SE and EE groups (Figure 2C). ⁸ , ¹⁶ Also, no significant correlation with behavioral performance (represented by Neuroscore's AUC) was found (Figure 2D). Likewise, quantification of Iba1, Gal3, and P2RY12 coverage in the peri‐infarct region revealed no significant differences between housing conditions (Figure 2E,G,I). Notably, Iba1 and Gal3 coverage correlated positively with infarct area in SE mice, but not in EE mice (Figure 2F: r = 0.847, p = 0.033, and r = 0.219, p = 0.636; Figure 2H: r = 0.937, p = 0.006 and r = −0.028, p = 0.953). ANCOVA analysis statistically confirmed the positive correlation between Iba1 and Gal3 with infarct sizes and revealed a differential group effect in P2RY12 staining (Supporting Information S1: Table 1: infarct_area vs. peri_iba1, infarct_area p = 0.002; infarct_area vs. peri_gal3, infarct_area p < 0.001; infarct_area vs. peri_p2y12, group p = 0.031). In order to characterize remote brain regions from the lesion, striatum area was also analyzed (Supporting Information S1: Figure 1B–I). While there was no difference between SE and EE groups in the coverage of Iba1, Gal3, and P2RY12, there was a positive correlation between Iba1 and Gal3 with infarct size in SE (Supporting Information S1: Figure 1F: r = 0.891, p = 0.017; Supporting Information S1: Figure 1G: r = 0.968, p = 0.002), but not for P2RY12.

Figure 2.

Infarct size correlations with peri‐infarct immunofluorescence analysis of ionized calcium binding adaptor molecule 1 (Iba1), galectin‐3 (Gal3), and purinergic receptor P2Y12 (P2RY12). (A) Representative immunostaining of 4′,6‐diamidino‐2‐phenylindole (DAPI), Iba1, P2RY12, and Gal3 in a standard environment (SE) mouse at peri‐infarct area (at ×20 magnification). (B) Representative immunostaining of DAPI, Iba1, P2RY12, and Gal3 in an enriched environment (EE) mouse at peri‐infarct area (at ×20 magnification). (C) Quantification of indirect infarct area measurements. (D) Correlation of infarct area with Neuroscore per group. (E) Iba1 coverage quantification measured as the percentage of image covered by Iba1 area (%area). (F) Correlation of infarct area with Iba1 coverage. (G) Gal3 coverage quantification measured as the percentage of image covered by Iba1⁺Gal3⁺ area (%area). (H) Correlation of infarct area with Gal3 coverage. (I) P2RY12 coverage quantification measured as the percentage of image covered by Iba1⁺P2RY12⁺ area (%area). (J) Correlation of infarct area with P2RY12 coverage. Peri‐infarct area is shown as dashed red lines in A and B. In (C, E, G, and I), values are expressed as individual experimental replicates with mean ± SEM. In (D, F, H, and J), values are expressed as individual experimental replicates with simple linear regression lines; Pearson correlation's r value with p‐value are also shown. In (C, E, G, and I), unpaired t‐test was performed. n = 4/6 mice in SE and n = 7 mice in EE (2 mice in SE were not behaviorally characterized). P‐values and r values are expressed with 3 decimals. P‐values were not corrected for multiple comparisons.

Microglial phagocytic activity was assessed using CD68, a lysosomal marker, and TREM2, a receptor implicated in microglial phenotypic regulation in neurodegeneration). Quantification in the peri‐infarct region revealed no significant differences in CD68 or TREM2 coverage between SE and EE mice (Figure 3A). However, CD68 coverage correlated positively with infarct size in SE mice, whereas TREM2 did not (Figure 3B: r = 0.845, p = 0.034 and Figure 3C: r = −0.651, p = 0.161). In this case, ANCOVA also showed a positive effect of infarct area and CD68 coverage, but not with TREM2 (Supporting Information S1: Table 1: infarct_area vs. peri_cd68, infarct_area p = 0.002; infarct_area vs. peri_trem2, infarct_area p = 0.145). Bodipy was also used to quantify lipid accumulation within CD68‐positive cells, but no statistically significant relationship was found (Supporting Information S1: Figure 2A–C).

Figure 3.

Infarct size correlations with peri‐infarct immunofluorescence analysis of cluster of differentiation 68 (CD68), triggering receptor expressed on myeloid cells 2 (TREM2), and 4,4‐difluoro‐4‐bora‐3a,4a‐diaza‐s‐indacene (Bodipy), and microglial morphology analysis. (A) Representative immunostaining of ionized calcium binding adaptor molecule 1 (Iba1), Bodipy, CD68, and TREM2 in standard and enriched environment mice at peri‐infarct area (at ×20 magnification), with CD68 and TREM2 coverage quantification measured as the percentage of image covered by CD68 or TREM2 area (%area). (B) Correlation of infarct area with CD68 coverage. (C) Correlation of infarct area with TREM2 coverage. (D) Example of skeletonization of a microglial cell for morphology analysis. (E) Quantification of number of microglial endpoints per cell. (F) Quantification of average branch length per cell. (G) Percentage of microglial cells per mouse presenting ameboid (0–5 branches per cell), transitional (6–12 branches per cell), or ramified (>12 branches per cell) morphology. In (A, G), values are expressed as individual experimental replicates with mean ± SEM. In (B, C, G) values are expressed as individual experimental replicates. In (A, B) simple linear regression lines and Pearson correlation's r value with p‐value are also shown. In (E, F) values are expressed as violin plots, and each value corresponds to a microglial cell (45 cells per mouse). In (A, G), unpaired t‐test was performed. In (E, F) Kolmogorov–Smirnov test was performed. n = 6 mice in standard environment and n = 7 mice in enriched environment. p‐Values and r values are expressed with 3 decimals. p‐Values were not corrected for multiple comparisons.

To further characterize microglial activation, we analyzed the morphology of individual microglial cells randomly sampled from the peri‐infarct region (Figure 3D). Neither the number of endpoints nor the total process length differed between housing conditions (Figure 3E,F). Notably, classification of cells by branching complexity revealed a significant reduction in amoeboid‐shaped microglia in EE mice (Figure 3G: p = 0.045), suggesting a functional shift of these cells, that is, due to reduced microglial activation or reduced infiltration of hematogenous Iba‐1‐expressing immune cells from the periphery (Figure 3G). These data suggest that EE may modulate the relationship between infarct size and microglial activation, reflecting a distinct inflammatory response profile in the peri‐infarct region.

3.3. Enriched housing modulates microglial‐infarct correlations in white matter regions

White matter injury and degeneration is a common consequence of ischemic stroke and microglia are key players in this process. ²⁷ , ²⁸ Therefore, we next examined the microglial response in white matter regions at 21 days post‐PT. Specifically, we quantified Iba1, Gal3, and P2RY12 in the corpus callosum and external capsule ipsilateral to the lesion (Figure 4A,B). As in the peri‐infarct area, marker coverage did not differ directly between SE and EE mice (Figure 4C–E). However, Iba1 and Gal3 displayed significant positive correlations with infarct size in SE mice (Figure 4F: r = 0.969, p = 0.001 and Figure 4G: r = 0.923, p = 0.009). Importantly, ANCOVA revealed both infarct size‐dependent effects of these markers and housing group‐specific influences (Supporting Information S1: Table 2: infarct_area vs. cc_iba1, infarct_area p < 0.001 and group p = 0.012; infarct_area vs cc_gal3, infarct_area p < 0.001 and group p = 0.043). Moreover, a significant interaction between infarct size and housing was observed for Iba1, with a nonsignificant trend also present for Gal3 (Supporting Information S1: Table 2: infarct_area vs. cc_gal3, infarct_area:group p = 0.123). CD68 expression in white matter revealed a pattern similar to that observed in the peri‐infarct region, with a positive correlation with infarct size detected only in SE mice (Figure 5C: r = 0.921, p = 0.009). In contrast, TREM2 displayed a distinct response in white matter, showing a positive correlation with infarct size in SE animals (Figure 5E: r = 0.829, p = 0.041), differing from the peri‐infarct findings. ANCOVA analysis underscored this divergence: in white matter, CD68 showed a non‐significant trend in housing group effect, while TREM2 coverage was influenced by infarct size (Supporting Information S1: Table 2: infarct_area vs. cc_cd68, group p = 0.064; infarct_area vs. cc_trem2, infarct_area p = 0.022), in contrast to the patterns observed in the peri‐infarct region.

Figure 4.

Infarct size correlations with white matter immunofluorescence analysis of ionized calcium binding adaptor molecule 1 (Iba1), galectin‐3 (Gal3), and purinergic receptor P2Y12 (P2RY12). (A) Representative immunostaining of 4′,6‐diamidino‐2‐phenylindole (DAPI), Iba1, P2RY12, and Gal3 (at ×20 magnification) in a standard environment (SE) mouse at white matter area (corpus callosum + external capsule). (B) Representative immunostaining of DAPI, Iba1, P2RY12, and Gal3 in an enriched environment (EE) mouse at white matter area (at ×20 magnification). (C) Iba1 coverage quantification measured as the percentage of image covered by Iba1 area (%area). (D) Gal3 coverage quantification measured as the percentage of image covered by Iba1⁺Gal3⁺ area (%area). (E) P2RY12 coverage quantification measured as the percentage of image covered by Iba1⁺P2RY12⁺ area (%area). (F) Correlation of infarct area with Iba1 coverage. (G) Correlation of infarct area with Gal3 coverage. (H) Correlation of infarct area with P2RY12 coverage. White matter area is shown as dashed red lines in (A, B). In (C–E) values are expressed as individual experimental replicates with mean ± SEM. In (F–H) values are expressed as individual experimental replicates with simple linear regression lines; Pearson correlation's r value with p‐value are also shown. In (C–E) unpaired t‐test was performed n = 6 mice in SE and n = 7 mice in EE. p‐Values and r values are expressed with 3 decimals. p‐Values were not corrected for multiple comparisons.

Figure 5.

Infarct size correlations with white matter immunofluorescence analysis of cluster of differentiation 68 (CD68) and triggering receptor expressed on myeloid cells 2 (TREM2). (A) Representative immunostaining of CD68 and TREM2 in standard and enriched environment mice at white matter area (at ×20 magnification). (B) CD68 coverage quantification measured as the percentage of image covered by CD68 area (%area). (C) Correlation of infarct area with CD68 coverage. (D) TREM2 coverage quantification measured as the percentage of image covered by TREM2 area (%area) (E) Correlation of infarct area with TREM2 coverage. In (B, D) values are expressed as individual experimental replicates with mean ± SEM. In (C, E) values are expressed as individual experimental replicates with simple linear regression lines; Pearson correlation's r value with p‐value are also shown. In (B, D) unpaired t‐test was performed. n = 6 mice in standard environment and n = 7 mice in enriched environment. p‐Values and r values are expressed with 3 decimals. p‐Values were not corrected for multiple comparisons.

3.4. Myelin debris and myelin loss correlate with inflammation and infarct size depending on the housing condition

To further assess the impact of EE on white matter integrity 21 days after PT, we stained myelin fibers using Black Gold Myelin (Figure 6A–D). In the peri‐infarct region, staining revealed puncta‐like structures absent in the contralateral hemisphere (Figure 6B,D), consistent with local myelin debris. Quantification showed no difference between SE and EE (Figure 6E), but debris levels positively correlated with infarct size only in SE mice (Figure 6F: r = 0.865, p = 0.026). In white matter regions (corpus callosum and external capsule), total myelin coverage relative to the contralateral side (myelin loss) was similarly unaffected by housing (Figure 6G) yet correlated with infarct size exclusively in SE animals (Figure 6H: r = −0.907, p = 0.013). To explore the relationship between these myelin‐related measures and inflammation, we correlated peri‐infarct myelin debris with microglial markers in this region (Figure 6I), and myelin loss with markers in white matter (Figure 6J). Myelin debris correlated significantly with Gal3 in SE (Figure 6I Gal3: r = 0.82, p = 0.048), and ANCOVA confirmed a housing group effect specific to this marker (Supporting Information S1: Table 3: myelin_debris vs. peri_gal3, group p = 0.027). Interestingly, myelin loss correlated with all measured microglial markers in SE mice, suggesting a stronger microglial involvement in this group. However, ANCOVA revealed no group effect, although myelin loss itself exerted a highly significant influence across markers, including interaction effects (Supporting Information S1: Table 3).

Figure 6.

Quantification of peri‐infarct myelin debris, white matter myelin loss, and their correlations with microglial markers. (A) Representative myelin staining in a standard environment mouse (at ×20 magnification). (B) Enlarged views of infarct contralateral cortical (green square) and peri‐infarct (red square) myelin in a standard environment (SE) mouse. (C) Representative myelin staining in an enriched environment mouse (at ×20 magnification). (D) Enlarged views of infarct contralateral cortical (green square) and peri‐infarct (red square) myelin in an enriched environment (EE) mouse. (E) Myelin debris coverage quantification measured as the percentage of peri‐infarct image covered by Black Gold Myelin dark debris area (%area). (F) Correlation of infarct area with myelin debris coverage. (G) Myelin loss quantification measured as the percentage of myelin lost at corpus callosum in ipsilateral versus contralateral infarct. (H) Correlation of infarct area with myelin loss. (I) Correlations of myelin debris with ionized calcium binding adaptor molecule 1 (Iba1), galectin‐3 (Gal3), purinergic receptor P2Y12 (P2RY12), cluster of differentiation 68 (CD68), and triggering receptor expressed on myeloid cells 2 (TREM2) coverages at peri‐infarct. (J) Correlations of myelin loss with Iba1, Gal3, P2RY12, CD68, and TREM2 coverages at white matter. In (E, G) values are expressed as individual experimental replicates with mean ± SEM. In (F, H, I, J), values are expressed as individual experimental replicates with simple linear regression lines; Pearson correlation's r value with p‐value are also shown. In (E, G) unpaired t‐test was performed. n = 6 mice in SE and n = 7 mice in EE. p‐Values and r values are expressed with 3 decimals. p‐Values were not corrected for multiple comparisons.

3.5. White matter TREM2 links enriched housing to improved behavioral recovery

Finally, we investigated whether behavioral recovery, as measured by Neuroscore, was associated with microglial or myelin parameters. Strikingly, both correlation and ANCOVA analyses identified a significant relationship only for TREM2 coverage in white matter (Figure 7A–C; Supporting Information S1: Table 4). In EE mice, TREM2 area correlated positively with Neuroscore (Figure 7B: r = 0.890, p = 0.007), and ANCOVA confirmed a significant housing group effect (Figure 7C, Supporting Information S1: Table 4: auc vs. cc_trem2, group p = 0.039). No other inflammatory or myelin markers showed robust associations with behaviour, with the exception of a nonsignificant trend for myelin debris predicting Neuroscore (Supporting Information S1: Table 4). These findings highlight TREM2‐positive microglia in white matter as a potential key mediator linking enriched housing to improved functional recovery after stroke.

Figure 7.

Correlation of Neuroscore with microglial and myelin markers, highlighting a white matter‐specific role of triggering receptor expressed on myeloid cells 2 (TREM2). (A) Correlation of Neuroscore's area under the curve (AUC) with TREM2 coverage at peri‐infarct area. (B) Correlation of Neuroscore's AUC with TREM2 coverage at white matter area (corpus callosum+external capsule). (C) Analysis of covariance (ANCOVA) table result of Neuroscore's AUC with TREM2 at white matter. In a and b, values are expressed as individual experimental replicates with simple linear regression lines; Pearson correlation's r value with p‐value are also shown. n = 4 mice in standard environment and n = 7 mice in enriched environment. p‐Values were not corrected for multiple comparisons.

4. DISCUSSION

In this study, we investigated how EE (EE) influences chronic neuroinflammation and white matter pathology following PT stroke in mice. By analyzing brains 21 days postinjury, we focused on the late recovery phase, when acute inflammation has subsided and tissue reorganization processes begin to dominate. ²⁹ Our findings demonstrate that microglial activation, myelin debris, and myelin loss remain pronounced at this stage, yet their relationship to lesion size and behavioral outcome is strongly modulated by the housing condition.

EE has long been shown to exert beneficial effects on neuroplasticity and recovery after stroke, ⁷ , ⁸ , ¹⁶ , ²² , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ as well as to modulate the inflammatory response following stroke. ⁹ , ¹² , ¹⁴ , ³⁷ , ³⁸ In parallel, considerable attention has been directed toward microglial characterization following stroke, ³⁹ , ⁴⁰ , ⁴¹ , ⁴² including their role in white matter repair. ⁴³ Nevertheless, a study systematically addressing microglial phenotypes during the recovery period after PT stroke under enriched housing conditions (together with the extent of accompanying myelin injury) has so far been lacking.

4.1. The role of Galectin‐3 in mechanisms of stroke recovery

In the present investigation, a central observation was that Gal3, a marker found in proinflammatory microglia, consistently increased with infarct size in SE mice, both in peri‐infarct and white matter regions. This supports Gal3's established role as a key driver of postischemic microglial activation possible through interleukin‐6/Janus kinase/signal transducer and activator of transcription (IL‐6/JAK/STAT) signalling. ⁴⁴ We have previously found Gal3 to be related to neuronal cell death and memory impairment following global brain ischemia, and binding to the TLR4 receptor, ¹⁸ as well as having a detrimental role in models of Parkinson's ⁴⁵ , ⁴⁶ and Alzheimer's disease. ⁴⁷ , ⁴⁸ On the other hand, previous studies have shown that Gal3 exerts protective effects shifting microglia morphology and function during the early phase following transient occlusion of the middle cerebral artery. ⁴⁹ Thus, actions of Gal3 may differ on which type of microglial cells is affected as well as on the spatio‐temporal context of microglia modulation following injury. In addition, Gal3 actions may also be dependent on the type of ischemic injury and include both pro‐inflammatory and potential reparative processes. Additional mechanistic studies will be required to resolve this dual role in different stroke models.

Importantly, in the present study we found that Gal3 also correlated with the accumulation of peri‐infarct myelin debris, implicating that microglia expressing Gal3 either participate in degradation of myelin sheaths or are involved in tissue cleaning processes, that is, tissue recycling. In contrast, inflammatory markers were not tightly coupled to lesion size in EE animals. Instead, their response appeared more consistent throughout the cohort, with individual animals clustering closer together and showing less variability across different infarct sizes. This harmonization likely underlies why correlations between inflammation, myelin debris/loss, and infarct size were observed in SE but not in EE. These findings suggest that enrichment dampens the injury‐driven pro‐inflammatory activity, consistent with the idea that Gal3 sustains maladaptive states at later stages of stroke recovery. ⁵⁰ , ⁵¹

4.2. TREM2 and reparative microglia

In parallel, we observed differences in TREM2 dynamics. While peri‐infarct TREM2 expression was limited, white matter regions displayed robust coverage, and here its role diverged from Gal3. In SE mice, TREM2 scaled with infarct size, but critically, in EE mice, white matter TREM2 correlated positively with behavioral outcome as measured by Neuroscore. ANCOVA confirmed a significant housing effect, suggesting that enrichment potentiates TREM2‐related pathways that support recovery. TREM2 has been implicated in lipid handling, phagocytosis, and remyelination, ²⁸ , ⁵² processes essential for white matter repair, ⁵³ and is a central receptor in the microglial response in Alzheimer's where we have found TREM2 can be a receptor for Gal3. ⁴⁸

It is important to note that these findings are correlative, and we cannot conclude that TREM2 directly mediates recovery. Future studies using TREM2 knockout models or pharmacological modulation are necessary to test causality and explore whether enhancing TREM2 function can promote white matter repair.

The association between TREM2 and functional recovery specifically in EE animals suggests that this receptor could be a key mediator linking environmental modulation to adaptive microglial responses and behavioral benefit.

Additional markers further support our findings that EE weakens the link between infarct size and detrimental neuroinflammation. CD68, a lysosomal protein linked to phagocytosis, showed positive correlations with infarct size in SE but not EE mice, both peri‐infarct and in white matter. Similarly, both peri‐infarct myelin debris and white matter myelin loss followed a lesion‐driven trajectory in SE, with infarct size predicting both debris accumulation and loss of myelin coverage. In EE, these associations were attenuated, suggesting that enrichment provides resilience against damage‐driven processes. Notably, peri‐infarct myelin debris correlated exclusively with Gal3, further highlighting its role in sustaining a pro‐inflammatory response. ⁵⁴ By contrast, no consistent effects were observed for P2RY12, consistent with previous reports that homeostatic microglia are largely absent in chronic post‐stroke tissue. ¹⁹

Taken together, these results reveal a coherent pattern: in SE animals, inflammatory and degenerative responses scale proportionally with infarct size. In EE, these responses appear attenuated across animals independent from lesion size, suggesting a regulated and resilient neuroinflammatory milieu involved in promoting recovery. A tissue environment has been created in which maladaptive scaling of Gal3‐driven inflammation is suppressed, while TREM2‐driven reparative pathways in white matter become linked to functional recovery.

4.3. Study limitations

While our findings highlight distinct inflammatory patterns linked to housing conditions, several limitations should be acknowledged. First, the assumption that Iba1 is expressed exclusively by microglia represents an important limitation. Specifically, while our analysis used Iba1 as a pan‐microglial marker, it is important to acknowledge that Iba1 also labels infiltrating monocyte‐derived macrophages, particularly within and around the infarct core. To the best of our knowledge, resident microglia and infiltrating monocytes/macrophages positive for Iba1 cannot be distinguished by immunofluorescence means. At Day 21 post‐PT stroke, infiltrating macrophages are substantially reduced compared to the acute phase (Days 3–7), yet a small population persists within the lesion core, often adopting repair‐associated phenotypes. ⁵⁵ In parallel, resident microglia repopulate the ischemic territory and increasingly express markers such as CD206, arginase‐1, IL‐10, and transforming growth factor‐beta, reflecting a transition toward a reparative milieu. ⁵⁶ This implies that the Iba1‐positive population at this stage represents a mixed but predominantly microglial composition, with a minor contribution from lingering infiltrating macrophages that likely participate in remodeling rather than acute inflammation. Consequently, the interpretation of our Iba1‐based findings has been performed in light of this cellular heterogeneity. Second, our analysis focused exclusively on microglia/macrophages, and thus we cannot exclude contributions from other cells such astrocytes, oligodendrocytes, or other peripheral immune cells involved in inflammatory cascades after PT. Third, all analyses were performed at a late recovery stage (21 days poststroke), which allowed us to capture a chronic response but did not address how EE may influence the earlier inflammatory processes. Fourth, our study is inherently correlative: although we observed associations between TREM2, Gal3, myelin debris/loss, and behavioral recovery, a causality cannot be inferred without targeted these pathways. Fifth, the EE paradigm itself is multifaceted, combining social, physical, and sensory stimulation; our design does not disentangle which components are most critical for the observed effects. Sixth, it also must be addressed that some of the apparent uniformity in inflammatory responses in EE mice may partly reflect the influence of infarct size itself, with markers tending to respond similarly across animals. Nevertheless, even within this relatively consistent response, we still observed distinct changes, such as the existence of TREM2‐associated reparative activity in white matter, highlighting that enrichment can modulate specific aspects of post‐stroke inflammation and recovery. Seventh, as summarized previously, ⁵⁷ microglial cells show a sexual dimorphism under physiological conditions and after stroke. Differences include divergent phenotypes, differences in the expression of inflammatory markers and differences in the response to sexual hormones. As an exploratory study, this investigation only included male mice; further larger studies will be required to determine differences in the inflammatory response between male and female mice following stroke and housing in different environments. Eight, the purpose of this study was to investigate the effect of poststroke housing conditions on inflammation. This kind of response is not found in no‐stroked animals. Therefore, sham‐operated mice were not included. Taken together, these limitations indicate that while EE promotes a shift towards a regulated inflammatory milieu after stroke, further mechanistic and longitudinal studies are needed to fully define the cellular and molecular drivers of this process.

4.4. Implications for rehabilitation and translational relevance

From a translational perspective, these findings identify Gal3 and TREM2 as divergent but complementary targets in chronic stroke recovery. Gal3 may represent a biomarker or therapeutic target for maladaptive inflammation and myelin‐related injury, while TREM2 emerges as a candidate mediator of white matter repair and functional improvement under enriched conditions. Importantly, the dual and context‐dependent roles of Gal3, along with the correlative nature of TREM2 associations, should be considered when designing interventions. Future mechanistic experiments including targeted knockouts or pharmacological modulation of TREM2 and Gal3, would strengthen causal interpretation and translational relevance. On top of that, future studies are warranted to explore whether enrichment directly suppresses Gal3 signalling or enhances TREM2 pathways, and if these two populations of microglia distinct in their spatial distribution and function could be modulated separately.

While enrichment cannot reverse severe pathological processes, our results reinforce the concept that it shapes the balance between pro‐inflammatory and reparative microglial functions. Multi‐modal stimulation by EE appears to create a tissue milieu that is in favor of recovery. This suggests that interventions targeting Gal3 suppression and TREM2 potentiation may contribute to post‐stroke white matter repair and functional outcomes.

AUTHOR CONTRIBUTIONS

Lluís Camprubí‐Ferrer: Conceptualization (lead); data curation (lead); formal analysis (lead); investigation (lead); methodology (lead); supervision (equal); visualization (lead); writing—original draft (lead); writing—review and editing (lead). Ömer K. Kolbay: Data curation (lead); formal analysis (lead); investigation (equal); methodology (equal). Miriana J. Quattromani: Conceptualization (equal); data curation (equal); formal analysis (equal); investigation (equal); methodology (lead). Tadeusz Wieloch: Conceptualization (equal); supervision (equal); writing—original draft (equal); writing—review and editing (equal). Karsten Ruscher: Conceptualization (equal); investigation (equal); supervision (lead); writing—original draft (equal); writing—review and editing (lead). Tomas Deierborg: Conceptualization (equal); funding acquisition (lead); supervision (lead); writing—review and editing (equal).

CONFLICT OF INTEREST STATEMENT

Karsten Ruscher is a member of Neuroprotection editorial board and is therefore excluded from the peer‐review process and all editorial decisions related to the publication of this manuscript. All other authors declare no competing interests.

ETHICS STATEMENT

Ethical approval for animal experiments was obtained from the Malmö‐Lund review committee (ethical permit numbers M 50‐15 and M 25‐12). All animal procedures were conducted in accordance with applicable national and institutional guidelines and complied with Wiley's policies on animal research and reporting. Not applicable for this animal study.

Supporting information

Supplementary Material.

Camprubí‐Ferrer L, Kolbay ÖK, Quattromani MJ, Wieloch T, Ruscher K, Deierborg T. Environmental enrichment modulates chronic poststroke inflammation and links white matter TREM2‐positive microglia in recovery in mice. Neuroprotection. 2026;4:65‐79. 10.1002/nep3.70028

DATA AVAILABILITY STATEMENT

Data are available upon request from the authors. The data that support the findings of this study are available from the corresponding author upon reasonable request.