Microglial Inhibition Promotes Proliferation and Differentiation of Neural Stem Cells via STAT3/SDF‐1/CXCR4 Signaling Pathway in Hypoxic–Ischemic Encephalopathy

ABSTRACT

Neonatal hypoxic–ischemic encephalopathy (HIE) may induce substantial neuronal damage. In particular, an overactivation of microglia following HIE represents a pathogenically important process. Previous studies have shown that microglial inhibitors can exert neuroprotective effects in HIE; however, the specific mechanisms underlying these effects have not yet been elucidated. Ligation of the left common carotid artery and exposure to 5% O₂ were utilized to produce an HIE model in rats. A number of experimental approaches were then used to determine the effect of a microglial inhibition, achieved via the administration of GW2580, a Csf1r inhibitor, and investigate the mechanisms involved. Our HIE models exhibited substantial brain infarction and were significantly impaired in motor functions (p < 0.01–0.001, in all tests examined). In the infarction areas, the number of microglia, macrophages, and neural stem cells (NSCs) was all dramatically increased over that in sham‐injured rats, respectively (p < 0.05–0.001). The administration of GW2580 significantly reduced the numbers of microglia and macrophages, but increased the number of NSCs when compared to those in vehicle‐treated HIE models (p < 0.05–0.001). Furthermore, GW2580 significantly ameliorated both the histological and behavioral phenotypes in HIE rats and increased STAT phosphorylation (p < 0.05–0.001). Finally, the inhibition or activation of STAT3 respectively decreased or increased the neuroprotective effects of GW2580 (p < 0.05–0.001). Collectively, our findings demonstrate that the STAT3 signaling pathway plays a critical role in the neuroprotective effects of microglial inhibition and may facilitate the development of novel therapeutic strategies to treat stroke.

Partial microglial depletion indirectly promoted the activity of the STAT3 signaling pathway in neural cells and increased the proliferation and/or differentiation of NSCs in areas of infarction to exert neuroprotective effects via the paracrine effects of NSCs and immature neurons in HIE.

1. Introduction

Neonatal hypoxic–ischemic encephalopathy (HIE), a common cause of death or long‐term disabilities in survivors, is caused by a lack of oxygen (hypoxia) or a deficiency in blood supply (ischemia) in the brain during a neonatal period. The most common neurological sequelae of HIE include cognitive deficits, seizures, and motor disabilities (Konrad et al. 2025). Despite extensive studies, however, the mechanisms underlying these neurological deficits have not yet been determined, although recent evidence has highlighted the critical role of neuroinflammation in these pathological processes.

Brain injuries, including neuronal and non‐neuronal events, occur immediately following hypoxia and/or ischemia during the initial stages of HIE. In order to protect itself against these injuries, the body may immediately initiate a series of responses, especially inflammatory and immunological responses within or around the infarction areas, or even at the whole body level (Pimentel‐Coelho 2025; Toorell et al. 2024). These responses lead to the recruitment of both local microglia and immune cells from the blood, such as macrophages, to the site of injury. Within a timeframe of minutes to hours, these cells become activated, changing from a homeostatic status towards a pro‐inflammatory status. This process initiates the release of a number of pro‐inflammatory cytokines which subsequently facilitate the induction of inflammatory neural responses which help tissues to clear up harmful materials (Mesquida‐Veny et al. 2021; Pimentel‐Coelho 2023). In the peripheral tissues of the body, the microenvironment may shift from a pro‐inflammatory to an anti‐inflammatory status, in a manner that is predominantly driven by the activation of macrophages. These anti‐inflammatory responses are considered to represent a major pathological process.

However, unlike the peripheral tissues, immune cells in the brain do not undergo transformation from a pro‐inflammatory towards an anti‐inflammatory status during the pathological processes of HIE, especially during the sub‐acute to chronic stages following the initial hypoxic–ischemic insult. Thus, the persistent activation of either microglia or macrophages in the brain may induce a chronic inflammatory process that is characterized by a sustained pro‐inflammatory microenvironment in which these cells may not only lose their protective effects but may also cause further damage to neurons (Hillman et al. 2025; Quan and Zhang 2023). Therefore, the inhibition or depletion of microglia has become a potential therapeutic strategy for HIE (Pimentel‐Coelho 2023; Shen et al. 2024). However, since microglia perform a range of vital functions, a complete depletion of these key structures is not an ideal approach, since this may cause further detriment to the microenvironment (Jin et al. 2017). Therefore, there is an urgent need to develop a strategy that could partially deplete microglia in the brain. More critically, there is a clear need to investigate how microglial inhibition may attenuate the pathological process associated with HIE.

Colony‐stimulating factor 1 receptor, also referred to as CD115, is a tyrosine kinase receptor that plays important roles in proliferation, differentiation, and migration of cells, and also in colonization of microglia and macrophages to exert effect upon inflammatory processes (Tarale and Alam 2022; Xiang et al. 2023). Csf1r is highly conserved across all vertebrates (Huang et al. 2024), and is predominantly expressed in microglia within the central nervous system (CNS) and/or rarely, in neural stem cells (NSCs) and neurons (Nandi et al. 2012; Rojo et al. 2019). Although the precise role of Csf1r in intracerebral hypoxia and/or ischemia has yet to be defined, its microglial localization and established role in microglial proliferation suggest that Csf1r must play a role in regulating inflammatory responses following HIE. In a previous study, researchers found that knockout of Csf1r in the micelled to a complete depletion of microglia and macrophages, concurrent with severe growth and developmental abnormalities (Patkar et al. 2021). Further evidence revealed that the antagonism of Csf1r exerted neuroprotective effects on neuronal damage in various conditions of brain injury (Henry et al. 2020; Neal et al. 2020; Peng et al. 2025). Therefore, the administration of Csf1r inhibitors, such as GW2580, may represent an effective approach to inhibit microglia. In particular, it would be highly prudent to investigate how Csf1r inhibitors could exert their therapeutic effects on the clinical symptoms associated with HIE. Notably, GW2580 is known to inhibit microglia by exerting effect on the Csf1r signaling pathway; this inhibitory effect leads to a neuroprotective response via remodeling of the immunological microenvironment within or around the hypoxic and/or/ischemic injury sites.

NSCs, which reside in the periventricular zone of the brain, may migrate to areas of infarction following hypoxic–ischemic insults, and may further differentiate into different neural subtypes in an attempt to initiate an endogenous repair mechanism (Arvidsson et al. 2002; Hassanzadeh et al. 2025; Williamson et al. 2023). Recent studies identified an increased population of NSCs within areas of infarction and reported that this increase was due to the expression of SDF‐1/CXCR4 (Hu et al. 2019; Wilson et al. 2024), a functional and bi‐directional regulator of the STAT3 signaling pathway (Ma et al. 2024). Consequently, it would be highly advantageous to investigate whether the administration of GW2580 could alter the distribution of NSCs, and whether this effect could be regulated by changes in the activity of the STAT3 signaling pathway or SDF‐1/CXCR4 axis.

In the present study, we first confirmed that the neuroprotective effects of GW2580 in a rat model of HIE were mediated by microglial inhibition and a shift of microglial mediators of microglia from a pro‐inflammatory condition (such as the expression of TNF‐α and iNOS) to an anti‐inflammatory condition (such as the expression of IL‐10 and Arg‐1). Next, we investigated whether the neuroprotective effects of GW2580 were associated with an activation of the STAT3 signaling pathway to drive the proliferation and differentiation of NSCs into immature neurons. These NSCs potentially promote the neuroprotection via paracrine effects. Our results, for the first time, established a molecular basis for the neuroprotective effects of microglial inhibition on HIE‐related neuronal injury, thus providing a new concept for the development of novel therapeutics for HIE.

2. Methods

2.1. Animals

All animal experiments here were conducted in accordance with the provisions for animal care and use described in ‘Guidance for the Care and Use of Laboratory Animals’ issued by the National Science Foundation of China (NSFc), and all procedures were approved by the IACUC in Guangzhou Women and Children's Medical Center (GWCMC) (Ethics Number: 2019‐23001). This study was also conducted in accordance with the ARRIVE guidelines. Male and female Sprague–Dawley (SD) rats were purchased from Charles River Laboratories (Beijing, China) and raised in the animal facility at GWCMC (Baiyun) under standard conditions: a 12/12 h light/dark cycle, a temperature of 20°C–22°C, a humidity of 60%, and an ad libitum supply of food and water. All animals included in experiments were at 3 weeks‐of‐age.

2.2. HIE Modeling

Surgery was performed following anesthesia with a mixture of isoflurane and O₂. Each rat was subjected to permanent ligation of the left common carotid artery (LCCA) and was immediately exposed to a 5% O₂ hypoxic environment for 2 h, as described previously (Ziabska et al. 2025). Rats that underwent the same surgical procedures, but without the artery ligation, were used as sham‐injured controls. These animals were randomly assigned into groups by researchers who were not involved in behavioral testing.

2.3. Magnetic Resonance Imaging

Brain infarction was evaluated by magnetic resonance imaging (MRI, Magnetom Prisma 3T, SIEMENS) under anesthesia 24 h after HIE. Each animal received a coronal T2‐weighted scan covering the entire brain with a resolution of 0.3 mm × 0.3 mm × 0.9 mm. Infarction areas were analyzed by ImageJ.

2.4. Behavioral Assessments

Researchers who conducted the behavioral tests were blinded to treatment groupings until the data analysis stage. All tests were carried out between 9:00 and 18:00 h in a sound‐ and light‐proof behavioral testing room. Four behavioral tests were performed (Longa test, beam walking test, grid walking test, and rotarod test), as described below.

Longa test: Neurological deficits were evaluated using the Zea Longa scoring system (Longa et al. 1989). In brief, rats were individually placed in an open‐field area, and motor function was assessed according to the following 5‐point scaling: score 0, no neurological deficit or normal locomotion; score 1, failure to fully extend the contralateral forelimb; score 2, contralateral circling during spontaneous movement; score 3. contralateral falling during ambulation; and score 4, no spontaneous locomotion, depressed consciousness or coma. Higher scores were considered to indicate more severe neurological dysfunction.

Beam walking test: Motor coordination was evaluated by an established protocol (Du et al. 2017). In brief, rats were individually placed on a narrow beam (14 mm in diameter) with an enclosed cage at the distal end, and performance was assessed according to the following 7‐point neurological scaling: score 0, normal locomotion, bilateral paws maintained plantigrade stepping on the beam surface with ≤ 2 ft slips and no lateral grasping; score 1, successful traversal with the affected limbs engaged in > 50% of steps; score 2, successful traversal with the affected limbs used in < 50% of steps; score 3, complete traversal but with ≥ 1 instance of affected limb placement on the beam surface; score 4, progression achieved by dragging the contralateral hindlimbs; score 5, unable to traverse but maintained contralateral limb contact for ≥ 5 s; and score 6, complete inability to traverse and no functional limb placement. A higher score was considered to indicate a more severe motor deficit.

Grid walking test: The procedures were described previously (Liu et al. 2012). In brief, rats were individually placed on an elevated metallic grid under standardized lighting for 2 min. While moving around on the grid, animals typically placed their paws on the wireframe for foot holds. Missed steps were defined as limbs (forelimb or hindlimb) passing fully through the grid openings during placement. The total number of steps (limb lifts typically placed on the wireframe) and missed steps were quantified. The missed step ratio was calculated separately for the forelimbs and hindlimbs.

Rotarod test: The procedures were described previously (Tanaka et al. 2025). In brief, rats were individually placed on a motorized cylinder, and the length of time that each animal remained on the cylinder was measured. The rotation speed was gradually accelerated from 4 to 40 rpm over a period of 5 min. The experiment was terminated once an animal fell off the rungs or gripped the device and spun around for two consecutive revolutions without attempting to walk on the rungs.

2.5. Chemical Administration

GW2580 (MedChemExpress, HY‐10917; 40 mg/kg; Conway et al. 2008), stattic (MedChemExpress, HY‐138185; 5 mg/kg; Alhazzani et al. 2021), and colivelin (MedChemExpress, HY‐P1061; 1 mg/kg; Zhao et al. 2019) were dissolved in a vehicle composed by 10% DMSO, 40% PEG300, 5% Tween‐80, and 45% saline. All compounds were administered by a single oral gavage commencing at 2 h post‐HIE; thereafter, this treatment was carried out daily until the day before tissue collection (Figure 2A or 6A). BrdU (MedChemExpress, HY‐15910; 75 mg/kg) was dissolved in sterilized saline and administered by intraperitoneal (i.p.) injection concurrently with either GW2580 or vehicle.

FIGURE 2.

GW2580 partially depleted microglia and macrophages in brains of HIE rats. (A) Schematic diagram showing experimental procedures. Rats were treated with GW2580 or a vehicle 2 h after HIE and then subjected to MRI 24 h after HIE. Behavioral tests were conducted on days 1, 2, and 3 after HIE. Sample collection was conducted on days 3, 4, and 7 after HIE. (B, C) Representative images for the co‐immunofluostaining of Iba‐1 (red) together with DAPI (blue) on day 3 (B), and their quantitative analysis (C) (n = 6 per group). (D) FC successfully differentiated microglia (CD11b+/CD45^Low) from macrophages (CD11b+/CD45^High) on day 3. Quantitative analysis of FC data are shown for microglia (E) and macrophages (F). n = 6 per group, and data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

FIGURE 6.

The STAT3 signaling pathway mediated GW2580‐induced neuroprotection and regulated SDF‐1/CXCR4 expression in HIE. (A) Schematic diagram showing experimental procedures. Rats were treated with GW2580, colivelin, stattic, or vehicle 2 h following HIE. MRI was conducted 24 h after HIE, and then animals were evaluated by three behavioral tests over three consecutive days. Sample collection was conducted on days 3, 4, and 7 after HIE. (B) MRI revealed infarction areas in three slices with an interval of 0.3 mm for every two slices and (B) quantitative analyses (ratio of infarction area to the brain area on the same slice), n = 5 per group. Arrows point to the brain infarction areas. (D) Behavioral responses in three behavioral tests. Data show mean behavioral scores for the Longa test (left) or beam walking test (middle), along with staying time (duration) in the rotarod test (right). In all these tests, n = 8–11. Only data from Day 3 is presented. (E) Quantitative analyses of SDF‐1 expression, as determined by ELISA (n = 6 in each group). (F) Representative immunoblotting for CXCR4 and (G) quantitative analyses of expression levels. In all tests, n = 6. Data are presented as mean ± SEM; and *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

Clodronate liposomes (10 mg/100 μL; Liposome, C‐005) or control liposomes (10 mg/100 μL; Liposome, P‐005) were intravenously (i.v.) injected at 2 h after HIE and treatment was repeated every 3 days. When clodronate liposomes and GW2580 were co‐administered, the initial dosing timepoint (2 h post‐HIE) and routes (oral gavage for GW2580; i.v. for clodronate liposomes) were maintained, with GW2580 administered daily and clodronate liposomes every 3 days.

2.6. 2,3,5‐Triphenyltetrazolium Chloride Staining

The procedures used for 2,3,5‐triphenyltetrazolium chloride (TTC) staining were described previously (Lalancette‐Hebert et al. 2007). In brief, rats were heavily anesthetized with sodium pentobarbital (200 mg/kg) and rapidly decapitated. Brains were rapidly removed and frozen with powdered dry ice. Then, we used a rodent brain matrix to prepare coronal sections (2 mm thick). Sections were immersed in 1% TTC staining solution (PH1205, Thermo Fisher Scientific) at 37°C for 1 h in the dark, followed by fixation in 4% paraformaldehyde (PFA) for 1 h at 4°C. Stained sections were sequentially photographed, and areas of infarction were quantified by using ImageJ.

2.7. Nissl Staining

The procedures were described in our previous publication (Li et al. 2023). In brief, rats were heavily anesthetized with sodium pentobarbital (40 mg/kg) and trans‐cardially perfused with 0.9% saline followed by 4% PFA. Brains were then post‐fixed and equilibrated in 4% PFA in 30% sucrose phosphate buffer overnight and were then dehydrated with a graded series of ethanol concentrations before being embedded in paraffin. Coronal sections (5 μm) were prepared with a microtome (Laica RM 2135). Sections were deparaffinized in xylene and rehydrated through a descending series of ethanol concentrations to distilled water. All sections were then immersed in pre‐warmed Nissl staining solution (Solarbio; G1436) at 60°C for 30 mins. After a brief rinse in distilled water, slides were soaked in 95% ethanol for 10 s and dehydrated through an ascending series of ethanol concentrations. Finally, sections were cleared in xylene, mounted with neutral resin, and images were captured by light microscopy (Leica SP4).

2.8. Immunofluorescence Staining

The procedures of immunofluorescence staining were described in our previous publication (Li et al. 2023). In brief, rats were heavily anesthetized and received transcranial perfusion, as described above. Brains were then post‐fixed and equilibrated in 4% PFA and 30% sucrose phosphate buffer overnight. Coronal brain sections (5 μm) were prepared by using a Cryostat (Leica 3050S). Sections were permeabilized with 0.3% Triton X‐100 in PBS and blocked with 5% goat serum in PBS containing 0.1% Triton X‐100 for 1 h. Subsequently, sections were incubated with primary antibodies in PBS at 4°C overnight, followed by incubation with fluorescent secondary antibodies for 1 h at room temperature (RT). After washing, slides were mounted with an anti‐fade mounting medium (Electron Microscopy Sciences), and fluorescent images were acquired by microscopy (Leica SP8).

The following primary antibodies and kits were used: Iba‐1 antibody (1:1000; Abcam, ab178846), NeuN antibody (1:1000; Abcam, ab104224), SOX‐2 antibody (1:500; Abcam, ab97959), DCX antibody (1:200; Abcam, ab207175), Ki67 antibody (1:500; Abcam, ab279653), One‐step TUNEL In Situ Apoptosis Kit (elabscience, E‐CK‐A320), BrdU Detection Kit (Servicebio, G4102), GFAP antibody (1:500; CST, #3670T), and Phospho‐STAT3 (Tyr705) (1:100; CST, #9145T). We also used the following secondary antibodies: goat anti‐rabbit IgG (1:1000; Alexa Fluor 488; Abcam 150,081), goat anti‐rabbit IgG (1:1000; Alexa Fluor 594; Abcam 150,080), goat anti‐mouse IgG (1:1000; Alexa Fluor 488; Abcam 150,113), and goat anti‐mouse IgG (1:1000; Alexa Fluor 594; Abcam 150,116).

2.9. Western Blotting

Western blotting was used to determine the expression levels of target proteins, as described in our previous publication (Li et al. 2023). In brief, areas of infarction from brains of HIE rats, and the corresponding regions from sham‐injured rats, were collected from animals that had been heavily anesthetized and decapitated, as described above. These tissues were homogenized in RIPA lysis buffer (Beyotime) supplemented with protease inhibitors to extract total proteins, which were subsequently quantified. A total of 20 μg protein per sample was separated by 10% SDS‐PAGE and then transferred onto PVDF membranes (Immobilon‐P, Millipore, Bedford). Subsequently, membranes were blocked with 5% skimmed milk in PBS containing 0.1% Tween‐20 for 1 h at RT, and were then incubated overnight with primary antibodies at 4°C, followed by incubation with HRP‐conjugated secondary antibodies for 1 h at RT. Positive signals were visualized by an ECL system (Immobilon Crescendo Western HRP Substrate). To normalize protein loading, membranes were re‐probed with an anti‐β‐actin antibody. Densitometry was performed by ImageJ to determine expression levels. The primary antibodies used for western blotting were as follows: STAT3 antibody (1:1000; Abcam, ab68153), Phospho‐STAT3 (Tyr705) antibody (1:1000; CST, #9145T), CXCR4 antibody (1:1000; Abcam, ab124824), iNOS antibody (1:1000; Abcam, ab283655), Arg‐1 antibody (1:1000; Abcam, ab233548), β‐actin (1:1000; CST, #4970). Secondary antibodies included HRP‐goat anti‐rabbit IgG (1:5000; Biodragon, BF03008), and HRP‐goat anti‐mouse IgG (1:5000; Biodragon, BF03009).

2.10. Enzyme‐Linked Immunosorbent Assay (ELIS)

EELISA was performed in accordance with the methodology described in our previous publication (Joseph et al. 2013). In brief, brain tissues from the areas of infarction in HIE rats, and the corresponding regions from sham‐injured rats, were collected as described above and were homogenized in RIPA lysis buffer (Beyotime) containing protease inhibitors. Protein lysates or standards (100 μL per well) were loaded into 96‐well plates pre‐coated with target‐specific antibodies. Plates were incubated at 37°C for 60 min, followed by the addition of 100 μL of detection solution A and further incubation at 37°C for 60 min. After three washes, 100 μL of detection solution B was added and incubated at 37°C for 30 min. Following five additional washes, 90 μL of TMB substrate was added and incubated at 37°C for 15 min in the dark. Reactions were terminated with 50 μL of stop solution, and absorbance was immediately measured at 450 nm using a microplate reader (Varioskan LUX, Thermo Fisher). Rat cytokine ELISA kits were used to detect the levels of TNF‐α (Cloud‐Clone, SEA133Ra), IL‐10 (Cloud‐Clone, SEA056Ra), and SDF‐1 (Cloud‐Clone, SEA122Ra).

2.11. Flow Cytometry (FC)

The procedures were described previously (Rajan et al. 2019). In brief, brain tissues from the areas of infarction in HIE rats, and the corresponding areas in sham‐injured rats, were collected as described above, and were dissociated into single‐cell suspensions. Cells were resuspended in staining buffer (PBS containing 2% FBS and 1 mM EDTA) and the number of cells was determined. Aliquots of 1 × 10⁶ cells per tube were then incubated with fluorochrome‐conjugated antibodies (Anti‐CD45 antibody, FITC; 1:100; BD Biosciences, 554,877; Anti‐CD11b antibody, PE; 1:200; BD Biosciences, 562,105) for 30 min at RT in the dark. After staining, cells were washed twice with staining buffer, filtered through a 70 μm cell strainer, and resuspended in 1 mL of PBS. Samples were analyzed by a BD FACSCanto II flow cytometer (BD Biosciences, 657,338) with FACSDiva software (v8.0.1). Data analysis was performed using FlowJo software (vX.7.0).

2.12. Statistical Analysis

SPSS version 22.0 software was used for all statistical analyses and data are expressed as mean ± standard error of the mean (S.E.M). Comparisons between two groups or multiple comparisons between groups was performed by the Student's t‐test or by one‐way analysis of variance (ANOVA) followed by a post hoc Duncan's test, respectively. p < 0.05 was considered to be statistically significant.

3. Results

3.1. Establishment of HIE Rat Models

MRI brain imaging, TTC staining, histological staining (Nissl staining), and behavioral tests were used to validate our HIE model as detailed in Figure S1A. As shown in Figure S1B, a hyperintense shadow, which indicates the infarction areas, on T2‐weighted MR imaging was obviously noted in the brains of HIE animals, and a quantitative analysis revealed a highly significant difference when compared to that in sham‐injury (p < 0.001). TTC staining revealed distinct infarction areas in the left hemisphere of HIE animals, yielding results consistent with those in MRI studies (p < 0.001; Figure S1C). Furthermore, Nissl staining showed a dramatic loss of neurons in the infarction areas in the brain of HIE rats, and a time‐course study indicated that this neuronal loss kept at a highly significant difference at least from day 1 up to day 7 following injury (all p < 0.001; Figure S1D). In behavioral evaluations, a highly significant difference was observed between HIE and sham‐injured groups in the average score in the Longa test (p < 0.001; Figure S1E), the average score in the beam walking test (p < 0.001; Figure S1F), the misstep rate in the grid walking test (p < 0.01; Figure S1G), and the staying duration in the rotarod test (p < 0.001; Figure S1H). All of these results indicated that we had successfully generated a rat model of HIE.

3.2. Microglia and Macrophages Were Significantly Recruited to Areas of Infarction in Brains of a Rat Model of HIE

Ionized calcium binding adaptor molecule 1 (Iba‐1) is an established biomarker for both microglia and macrophages and is known to be significantly up‐regulated following brain injury (Chen et al. 2025; Shkirkova et al. 2024). In order to confirm whether this was true in our HIE rat models, we performed a time‐course study from day 1 to day 7 post‐HIE (Figure 1A). Analysis showed that the expression levels of Iba‐1 in the areas of infarction in HIE rats were significantly increased from day 3 to day 7, but not on days 1 or 2, when compared to those in sham‐injured rats (p < 0.001 for each timepoint from days 3 to 7; Figure 1A,B). Considering that microglia and macrophages are two distinctive cell populations and play different roles in the regulation of immunological/inflammatory events in response to injuries (Mesquida‐Veny et al. 2021), we employed FC to distinguish these two cell populations based on their different CD45 expression profiles (CD11b+/CD45^Low for microglia and CD11b+/CD45^High for macrophages) (Rajan et al. 2019). Given our Iba‐1 immunostaining results, we performed a time‐course experiment from days 3 to 7 post‐HIE. Analysis revealed that the numbers of microglia (Figure 1C,D) and macrophages (Figure 1C,E) were significantly up‐regulated at all timepoints examined when compared to those in sham‐injured rats, respectively (p < 0.001 for each cell type at each timepoint). Collectively, these results indicated that microglia and macrophages were significantly recruited into the areas of infarction in the brain following HIE.

FIGURE 1.

Both microglia and macrophages were significantly recruited into areas of infarction in brains of HIE rats. (A) Representative images for co‐immunostaining of Iba‐1 (red) together with DAPI (blue) and (B) quantitative analysis of staining signals (n = 6 in each group). (C) FC successfully differentiated microglia (CD11b+/CD45^Low) from macrophages (CD11b+/CD45^High). Quantitative analysis of FC data for microglia (D) and macrophages (E). n = 6 per group; data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

3.3. GW2580, Partially, but Significantly, Diminished the Recruitment of Microglia and Macrophages to Areas of Infarction in Brains of a Rat Model of HIE

Rats were first subjected to either HIE or sham‐injury, and then treated with a vehicle or GW280 2 h later. Behavioral tests were conducted on days 1 and 3, and brain tissues were collected on days 3, 4 and 7 after HIE or sham‐injury (Figure 2A). Iba‐1 immunostaining (Figure 2B) and quantitative analyses (Figure 2C) indicated that the proportion of Iba‐1⁺cells was significantly increased in the areas of infarction in brains of HIE rats when compared to those in the sham‐injured rats (p < 0.001 at each timepoint), thus indicating that Iba‐1⁺ cells had been recruited to the areas of infarction. The administration of GW2580 significantly reduced the numbers of Iba‐1⁺cells in both HIE and sham‐injured rats when compared to those in animals‐treated with vehicle (p < 0.05, sham‐vehicle vs. sham‐GW2580; p < 0.05–0.001, HIE‐vehicle vs. HIE‐GW2580). Collectively, these data indicated that GW2580 significantly reduced the number of Iba‐1⁺ cells in brains of both HIE and sham‐injured rats, but particularly in HIE rats.

Next, we used FC to distinguish the effects of GW2580 in different types of cells. Given that both microglia and macrophages exhibited the most significant changes between days 3 to 7, we primarily present our FC data on days 3 and 7 post‐HIE. FC revealed that both of the number of microglia and macrophages were significantly increased in the areas of infarction in HIE rats (Figure 2D). Quantitative analyses revealed a significant difference on day 3 (Figure 2E) and day 7 (Figure 2F) after HIE, when compared to the sham‐vehicle (p < 0.001). However, the administration of GW2580 significantly reduced the number of microglia in HIE rats when compared to rats receiving the vehicle (p < 0.001); similar results were observed when comparing sham‐injured rats receiving GW2580 and those receiving vehicle (p < 0.001) on either day 3 (Figure 2E) or day 7 (Figure 2F). For macrophages, however, this effect was only evident in HIE rats on either day 3 (p < 0.01; Figure 2E) or day 7 (p < 0.001; Figure 2F), but not in sham‐injured animals on either day (Figure 2E,F); macrophages were not detectable in the brains of rats undergoing sham‐injury. Collectively, these data indicate that GW2580 significantly reduced the number of microglia in the brains of either HIE or sham‐injured rats, and reduced the number of macrophages in the brains of rats following HIE, and that this effect occurred due to inhibition of the Csf1r signaling pathway.

3.4. GW2580 Induced Clear Pathological and Neurobehavioral Changes in HIE Rats

Next, we investigated whether GW2580 could rescue the neuropathological and neurobehavioral phenotypes in HIE rats by performing MRI brain imaging, histological and immunological staining, and behavioral tests. In sham‐injured animals, T2‐weighted MRI and quantitative analyses did not reveal any significant differences between sham‐vehicle and sham‐GW2580 rats (Figure 3A). However, a hyper‐intense shadow was clearly evident in the brains of HIE rats receiving the vehicle; quantitative analyses revealed there was a highly significant difference between sham‐vehicle and HIE‐vehicle rats (p < 0.001). The intensity of this shadow was significantly reduced in HIE rats receiving GW2580 when compared to those receiving the vehicle (p < 0.001), thus indicating that GW2580 could effectively ameliorate brain injury caused by HIE.

FIGURE 3.

Effects of GW2580 on the pathological and neurobehavioral changes observed in HIE rats. (A) MRI revealed infarction areas in three slices with an interval of 0.3 mm for every two slices (left); quantitative analyses (right; ratio of infarction area to the brain area on the same slice); n = 5 in each group. The arrows point to infarction areas in the brain. (B) Representative images of Nissl staining on day 4 (left) along with quantitative analysis (right; n = 6 per group). The position of the dashed line depicts the boundary of the infarction area. (C) Representative images for the co‐immunofluorescence staining of NeuN (red) together with BrdU (green) and DAPI (blue) on day 4 (left) along with quantitative analysis (right; n = 6 per group). (D) Representative images for the co‐immunofluorescence staining of TUNEL (red) together with DAPI (blue) on day 4 (left) along with quantitative analysis (right; n = 6 in each group). (E–G) Behavioral responses in three behavioral tests. Mean behavioral score for the Longa test (E) and the beam walking test (F), and staying time (duration) in the rotarod test (G). In all these tests, n = 8–14. Data are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

Next, we used three different staining methods to further characterize the changes at the histological level. Nissl staining revealed a significant reduction in cell number following HIE (p < 0.001) and that this reduction was significantly rescued by GW2580 treatment (p < 0.001) on days 3, 4, and 7 (Figure 3B). The results of NeuN staining were very similar to those for Nissl staining on days 3, 4, and 7 (p < 0.001, sham‐vehicle vs. HIE‐vehicle; p < 0.001, HIE‐vehicle vs. HIE‐GW2580; Figure 3C). TUNEL staining revealed that the number of apoptotic cells was significantly increased in the areas of infarction in HIE rats on days 3, 4, and 7 (p < 0.001); the administration of GW2580 only reduced the number of apoptotic cells on days 3 and 4 after HIE (p < 0.001), but not on day 7. The ineffective response on day 7 may be due to that the basal line of apoptosis was already very low at this timepoint. Significant co‐staining of BrdU and NeuN was not observed in any of the experimental groups (data was not shown), thus indicating the lack of newly generated neurons. To verify that the absence of BrdU/NeuN co‐staining was not due to experimental limitations, we tested immature neuron generation using doublecortin (DCX, a marker for neural progenitor cells/immature neurons) in combination with BrdU. As shown in Figure S2, the proportion of BrdU⁺/DCX⁺ cells was significantly increased, thus indicating that our BrdU labeling strategy was effective.

To further investigate whether or how these neuropathological changes were associated with behavioral changes, we employed three behavioral tests to evaluate the behavioral response over days 1–3 after GW2580 administration. As shown in Figure 3E–G, on day 1, a significant difference was observed in every behavioral test between the sham‐vehicle and HIE‐vehicle groups (p < 0.001, respectively); however, no significant differences were observed in any of the behavioral tests when compared between the HIE‐vehicle and HIE‐GW2580 groups (Figure 3E–G). On day 2, we detected a significant effect for GW2580 on behavioral responses in the Longa test (p < 0.001; Figure 3E) and beam walking test (p < 0.05; Figure 3F); from day 3, a significant effect was consistently detected for all behavioral tests (p < 0.01–0.001; Figure 3E–G). Collectively, these results indicated that GW2580 had a significant neuroprotective effect on HIE‐induced defects, and that this effect was not dependent on the generation of new neurons.

3.5. GW2580 Induced Change in Microenvironment From a Pro‐Inflammatory to an Anti‐Inflammatory Status

Our previous experiments had revealed that GW2580 induced striking effects in HIE rats; however, the specific mechanisms for these effects remained unknown. As both microglia and macrophages are implicated in the immune‐microenvironment (Fitch and Silver 2008; Mesquida‐Veny et al. 2021), we next determined whether GW2580 could induce change in the microenvironment. Accordingly, we selected TNF‐α and iNOS as markers of a pro‐inflammatory status, while IL‐10 and Arg‐1 were chosen as indicators of an anti‐inflammatory status. Since our previous experiments had not resulted in any significant effect of GW2580 on any indices in sham‐injured rats, we did not include sham‐GW2580 rats in our subsequent experiments.

ELISA revealed that the expression levels of TNF‐α (Figure 4A) and IL‐10 (Figure 4B) in the areas of infarction in HIE rats were significantly increased when compared to those in sham‐injured rats (p < 0.01–0.001); this was the case at all three timepoints, except for the levels of IL‐10 on day 7. GW2580 significantly suppressed the expression levels of TNF‐α on all 3 days examined when compared to those in the HIE‐vehicle group (p < 0.01–0.001; Figure 4A), but significantly elevated the expression levels of IL‐10 on all 3 days when compared to those in the HIE‐vehicle group (p < 0.05–0.01; Figure 4B).

FIGURE 4.

GW2580 facilitated changes of the microenvironment from a pro‐inflammatory status to an anti‐inflammatory status. Quantitative analyses of (A) TNF‐α and (B) IL‐10 expression, as determined by ELISA (n = 6 per group). (C) Representative immunoblotting for iNOS and Arg‐1. Quantitative analyses of iNOS (D) and Arg‐1 expression levels (E). n = 6 per group. Data are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

Furthermore, Western blots (Figure 4C) revealed a significant increase in the expression levels of both iNOS (p < 0.01–0.001; Figure 4D) and Arg‐1 (p < 0.05–0.001; Figure 4E) in the areas of infarction in HIE rats on all 3 days when compared to those in sham‐injured rats, except that the levels of Arg‐1 were significantly lower in HIE rats on day 7 (p < 0.05). Similarly, GW2580 significantly suppressed the expression levels of iNOS on all 3 days when compared to those in the HIE‐vehicle group (p < 0.01 for all three timepoints), but significantly elevated the expression levels of Arg‐1 on all 3 days when compared to those in the HIE‐vehicle group (p < 0.01 for all three timepoints).

Collectively, these results demonstrated that GW2580 inhibited the Csf1r signaling pathway causing a shift in the inflammatory microenvironment in the infarction area from a pro‐inflammatory status towards an anti‐inflammatory status. This change in the inflammatory microenvironment may have contributed to a synergistic effect of GW2580 on HIE‐induced inflammatory responses.

3.6. GW2580 Facilitated the Recruitment of NSCs to Areas of Infarction in Brains of HIE Animals

Previous reports suggested that the migration of NSCs from the periventricular zone to the areas of infarction following HIE might facilitate endogenous repair mechanisms (Arvidsson et al. 2002; Shahid and Begum 2025). Therefore, we next investigated whether the effects of GW2580 were associated with an increased population of NSCs at the area of infarction.

We performed immunostaining with SOX‐2, a known marker for NSCs (Han et al. 2025), to quantitatively evaluate the spatiotemporal distribution of NSCs following HIE and GW2580 treatment (Figure 5A). Although a gradual increase in NSCs was detected in the areas of infarction in all rats following HIE, a significantly lower proportion of NSCs was consistently observed in HIE rats when compared to that in sham controls (p < 0.05–0.001; Figure 5B). In addition, the treatment of HIE rats with GW2580 significantly increased the number of NSCs when compared to those in the HIE‐vehicle group (p < 0.05–0.001; Figure 5B).

FIGURE 5.

Effects of GW2580 on the number of NSCs were dependent on the activation of STAT3 and expression of the SDF‐1/CXCR4 axis in HIE rats. (A) Representative images for the co‐immunofluorescence staining of SOX‐2 (red) together with DAPI (blue) (A) on day 4, along with quantitative analysis (B) (n = 6 per group). (C–E). Representative immunoblotting for total STAT3 protein and phosphorylated STAT3‐Y705 (p‐STAT3) (C) and quantitative analyses of the expression levels of total STATs (D) or p‐STAT3 (E). n = 6 per group. (F) Quantitative analyses of SDF‐1 expression, as determined by ELISA (n = 6). Representative immunoblotting for CXCR4 (G) and quantitative analyses of expression levels (H). n = 6 for all tests. Data are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

These data indicate that the inhibition of the Csf1r signaling reduced the number of microglia, leading to a shift in the microenvironment towards an anti‐inflammatory status and an increased number of NSCs within the area of infarction, and thus suggest that GW2580 may promote tissue repair by regulating the dynamics of NSCs.

3.7. GW2580 Enhanced STAT3 Phosphorylation and Regulated SDF‐1/CXCR4 Expression in Infarction Area of Brains in HIE Animals

To elucidate the mechanisms underlying the increase of NSCs within the infarction areas, we investigated the STAT3 signaling pathway and the SDF‐1/CXCR4 chemokine axis as potential regulatory targets (Carbajal et al. 2010; Ma et al. 2024). Figure 5C shows the representative immunoblotting of both the total STAT3 and its Y705‐site phosphorylated protein following HIE or the treatment, and quantitative analyses revealed that, compared to those in sham‐injured controls, the total STAT3 level (p < 0.05–0.001; Figure 5D) and the Y705‐site phosphorylated STAT3 level (p < 0.05–0.001; Figure 5E) were both significantly increased in HIE rats on all days examined. Surprisingly, the treatment with GW2580 did not alter the total STAT3 level (Figure 5D) but significantly enhanced the phosphorylated STAT3 level in HIE rats compared to those in HIE‐vehicle rats, respectively, on all days examined (p < 0.05–0.01; Figure 5E). As shown in Figure 5F, the expression level of SDF‐1 was remarkably upregulated by HIE rather than by sham injury on all 3 days examined (p < 0.001, respectively); and moreover, GW2580 could further upregulate the expression level rather than the vehicle in HIE rats on all 3 days examined (p < 0.001, respectively). As shown in Figure 5G, immunoblotting showed an increased expression level of CXCR4 following HIE or the treatment, and quantitative analyses revealed that a significantly increased expression level was noted on day 3 (p < 0.001), but neither day 4 nor day 7 following HIE in comparison with that on the same day following the sham injury. Notably, GW2580 could even further upregulate the expression level of CXCR4 on all days examined compared to that in HIE‐vehicle rats (p < 0.001 in all three time points; Figure 5H), suggesting that GW2580 could enhance the STAT3 signaling pathway and promote the SDF‐1/CXCR4 axis activation in the infarction areas. Taken together, all these findings indicated that Csf1r inhibition could regulate NSCs dynamics through the STAT3 signaling pathway as well as the SDF‐1/CXCR4 axis, pointing to a potential mechanism for GW2580 effects.

3.8. Activation of STAT3 Signaling Pathway Was Implicated in GW2580‐Mediated Neuroprotective Effects

Our previous analyses showed that GW2580 increased the recruitment of NSCs, the phosphorylation of STAT3, and the activation of in the areas of infarction in HIE rats. The SDF‐1/CXCR4 axis, as a bidirectional regulator of STAT3, is known to mediate the recruitment of NSCs (Carbajal et al. 2010; Merino et al. 2015). Notably, IL‐10, an anti‐inflammatory factor, has been shown to be able to directly activate the STAT3 signaling pathway (Sharma et al. 2011), which could also upregulate the SDF‐1/CXCR4 axis (Junior et al. 2023). Moreover, our data suggested that GW2580 could shift the microenvironment from a pro‐inflammatory status to an anti‐inflammatory status in the infarction areas, thus indicating a pivotal role for STAT3 activation in mediating the neuroprotective effects of GW2580. To test this hypothesis, we next compared the effects of both colivelin (a STAT3 agonist) and stattic (a STAT3 inhibitor). As shown in Figure 6A, HIE animals received GW2580, colivelin, stattic, or vehicle 2 h post‐HIE, followed by MRI (day 1), behavioral tests (days 1 and 3), and tissue collection (days 3, 4, 7); a sham‐injured group was used as a baseline control. As shown in Figure 6B,C, the results in sham‐injured rats, HIE‐vehicle rats, and HIE‐GW2580 rats were very similar to those described earlier (Figure 3). MR T2‐weighted imaging (Figure 6B) and quantitative analyses (Figure 6C) revealed that the co‐administration of stattic and GW2580 completely removed the neuroprotective effect of GW2580 when compared to that in the HIE‐GW2580 group (p < 0.001), and when compared to that in the HIE‐vehicle group (p < 0.05, Figure 6C). The administration of stattic alone led to a worse outcome, with a significant expansion of the infarction areas when compared to those in the HIE‐vehicle group (p < 0.01; Figure 6C). The administration of colivelin alone to HIE animals achieved an effect that was very similar to that of GW2580, and there was a highly significant difference when compared to the HIE‐vehicle group (p < 0.001; Figure 6B,C); there was no significant difference between the HIE‐colivelin and HIE‐GW2580 groups. As shown in Figure 6D, behavioral experiments on day 3 (Longa test, beam walking test, and rotarod test) demonstrated that HIE rats consistently exhibited a significant deficit in all behavioral tests when compared sham‐injured rats (p < 0.001, for each test), and that the co‐administration of stattic and GW2580 completely removed the neuroprotective effects of GW2580 when compared to the HIE‐GW2580 group (p < 0.001). Although stattic alone had no significant effect on HIE‐induced behavioral deficits in both the Longa test and beam walking test, a significantly impaired effect was detected in the rotarod test when compared to the HIE‐vehicle group (p < 0.001; Figure 6D). Similarly, the use of colivelin in HIE animals had a very similar effect to GW2580; we detected a highly significant difference when compared to the HIE‐vehicle group (p < 0.01–0.001; in all behavioral tests); there was no significant difference between the HIE‐colivelin and HIE‐GW2580 groups.

Collectively, these results suggested that the inhibition of STAT3 reversed the neuroprotective effects of GW2580, or even led to a worse response, whereas the activation of STAT3 exerted neuroprotective effects that were similar to those induced by GW2580. Thus, the neuroprotective effects of GW2580 in HIE were likely attributed to activation of the STAT3 signaling pathway.

3.9. Role of STAT3 Signaling Pathway in Regulating SDF‐1/CXCR4 Expression in Areas of Infarction in Brains of HIE Rats

The SDF‐1/CXCR4 axis is a well‐known bidirectional regulator of the STAT3 signaling pathway (Ma et al. 2024). However, the role of STAT3 in regulating SDF‐1/CXCR4 expression in HIE, particularly in an anti‐inflammatory microenvironment mediated by GW2580, remains unknown. Based on our previous results, the most significant changes in SDF‐1/CXCR4 expression were observed on day 3 post‐HIE; therefore, we primarily focused on changes in SDF‐1/CXCR4 at this particular timepoint. As shown in Figure 6E–G, the co‐administration of stattic with GW2580 effectively reversed the GW2580‐induced upregulation in SDF‐1 and CXCR4 when compared to that in the HIE‐GW2580 group (p < 0.001) and HIE‐vehicle group (p < 0.05 for SDF‐1 expression and p < 0.01 for CXCR4 expression). The administration of colivelin alone significantly increased the expression levels of both SDF‐1 (p < 0.01) and CXCR4 (p < 0.05) when compared to the HIE‐vehicle group, but with no significant differences compared to the HIE‐GW2580 group. The administration of stattic alone to HIE rats significantly inhibited the expression levels of both SDF‐1 and CXCR4 when compared to the HIE‐GW2580 group (p < 0.001, respectively) and the HIE‐vehicle group (p < 0.05 for SDF‐1, p < 0.001 for CXCR4).

These the results suggested that the inhibition of the STAT3 signaling pathway in HIE could reverse the ability of GW2580 to up‐regulate the expression of SDF‐1/CXCR4, whereas activation of the STAT3 signaling pathway produced an effect that was similar to that induced by GW2580. In contrast, inhibition of the STAT3 signaling pathway alone also suppressed the expression of SDF‐1/CXCR4. Collectively, these results indicated that the STAT3 signaling pathway was a key regulator of SDF‐1/CXCR4 expression in HIE.

3.10. The Recruitment of NSCs to Infarction Areas Following GW2580 Treatment Was Dependent on STAT3 Signaling Pathway

The neuroprotective effects of GW2580 were associated with the recruitment of NSCs into the areas of infarction in a manner that could potentially be regulated by the upregulation of SDF‐1/CXCR4 (Carbajal et al. 2010; Merino et al. 2015). We found that GW2580‐induced STAT3 activation promoted SDF‐1/CXCR4 expression, thus suggesting that STAT3 may act as a key regulator for the increase of NSCs. Given that we observed the most significant changes in the numbers of NSCs on day 4 post‐HIE, we specifically focused on this timepoint in our next investigations. As shown in Figure 7A,B, the results in sham, HIE, HIE‐vehicle, and HIE‐GW2580 rats were consistent with those in Figure 5A,B. The co‐administration of stattic and GW2580 partially diminished the GW2580‐induced increase in NSCs when compared to the HIE‐GW2580 group (p < 0.05) but also when compared to the HIE‐vehicle group (p < 0.001). Colivelin alone significantly elevated the number of NSCs when compared to the HIE‐vehicle group (p < 0.001) in a manner that was similar to the HIE‐GW2580 group. The administration of stattic alone significantly increased the recruitment of NSCs to the areas of infarction when compared to the HIE‐vehicle group (p < 0.05), although this effect was significantly weaker than that of GW2580 (p < 0.01). Collectively, these findings indicated that the ability of GW2580 to facilitate the recruitment of NSCs to the infarction areas was dependent upon the STAT3 signaling pathway.

FIGURE 7.

STAT3 signaling pathway mediated an increase in the numbers of NSCs in areas of infarction following GW2580 treatment. (A) Representative images for the co‐immunofluorescence staining of SOX‐2 (red) together with Ki67 (green) and DAPI (blue). (B) Quantification of SOX‐2⁺/Ki67⁺ co‐stained cells (n = 6 per group). Only data from day 4 are presented. Data are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

3.11. Activation of STAT3 Signaling Pathway Facilitated Proliferation of NSCs in Infarction Areas Following Treatment With GW2580

Previous reports showed that an increased number of NSCs might be due to either recruitment or proliferation, both of which are associated with the upregulation of SDF‐1/CXCR4 (Carbajal et al. 2010; Merino et al. 2015). Thus, it is possible that the STAT3 signaling pathway might also play a critical role in these processes. Due to the difficulty in distinguishing resident versus recruited NSCs in vivo, we decided to focus on proliferation in our next experiments. We used immunofluorescent labeling of NSCs along with a cell cycle marker, Ki67 (Askew et al. 2017), and investigated whether the activity of the STAT3 signaling pathway was able to regulate the proliferation of NSCs, and hence the overall dynamics of NSCs in the areas of infarction in HIE animals. Co‐staining of both SOX‐2 and Ki67 revealed that following HIE, the proportion of Ki67⁺/SOX‐2⁺ co‐stained cells was significantly higher than in sham‐injured animals (p < 0.01), and that this injury effect was further enhanced by GW2580 when compared to the HIE‐vehicle group (p < 0.001). The co‐administration of stattic and GW2580 largely abolished the effect of GW2580 when compared to the HIE‐GW2580 group (p < 0.05), although the co‐stained cell population was still significantly higher than in the HIE‐vehicle group (p < 0.001). The administration of colivelin alone significantly increased the double‐stained cell population when compared to the HIE‐vehicle group (p < 0.001), although there was no significant difference when compared to the HIE‐GW2580 group. The administration of stattic alone significantly reduced the population of Ki67⁺/SOX‐2⁺ cells in the infarction areas when compared to the HIE‐GW2580 group (p < 0.01), although this population was significantly higher than that in the HIE‐vehicle group (p < 0.05).

Collectively, these data confirmed that activation of the STAT3 signaling pathway contributed to the increased number of NSCs by promoting the proliferation of NSCs in the areas of infarction in HIE animals.

3.12. Activation of STAT3 Signaling Pathway Promoted Differentiation of NSCs Into Immature Neurons and Enhanced Proliferation of Immature Neurons

Previous experiments revealed that the neuroprotective effect of GW2580 was independent of the generation of new neurons, thus suggesting that the increase in NSCs did not directly contribute to tissue repair through their differentiation into neurons, but possibly through paracrine effects mediated by cytokines secreted by NSCs and immature neurons (Moretti et al. 2025; Xu et al. 2022). The STAT3 signaling pathway clearly played a pivotal role in the proliferation of NSCs within the areas of infarction; however, the potential effects of this pathway on the proliferation and differentiation of NSCs into immature neurons, remained unclear. To address this issue, we performed co‐immunostaining of DCX (a marker for neural progenitor cells/immature neurons) and Ki67. As shown in Figure 8A,B, the population of DCX⁺ and Ki67⁺/DCX+ double‐stained cells significantly increased after HIE when compared to sham‐injured rats (p < 0.05–0.01); this increase was further elevated by GW2580 when compared to HIE‐vehicle rats (p < 0.01). The co‐administration of stattic and GW2580 almost completely diminished the effects of GW2580 when compared to the HIE‐GW2580 group (p < 0.01 for both cell populations) with no significant difference to the HIE‐vehicle group. The administration of colivelin significantly increased the population of DCX⁺ cells (p < 0.05) and double‐stained cells (p < 0.01), when compared to the HIE‐vehicle group, with no significant difference when compared to the HIE‐GW2580 group. The administration of stattic to HIE rats significantly reduced both cell populations when compared to the HIE‐GW2580 group (p < 0.01, respectively) with no significant difference to the HIE‐vehicle group.

FIGURE 8.

STAT3 activation promoted the differentiation of NSCs into immature neurons and enhanced the proliferation of immature neurons. (A) Representative images for the co‐immunofluorescence staining of DCX (red) together with Ki67 (green) and DAPI (blue). (B) Quantification of DCX⁺/ Ki67⁺ co‐stained cells (n = 6 in each group). Only data from day 4 is presented. All data in the bar figures are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

Collectively, these results suggested that activation of the STAT3 signaling pathway promoted the differentiation of NSCs into immature neurons and also facilitated the proliferation of immature neurons; or that GW2580 promoted the proliferation of NSCs via activation of the STAT3 signaling pathway, thus increasing the numbers of NSCs in areas of infarction. In addition, by enhancing the STAT3‐driven differentiation of NSCs into immature neurons and promoting their proliferation, GW2580 was able to further increase the population of immature neurons. These findings demonstrated that STAT3 acts as a key regulator for the proliferation and differentiation of NSCs and the increase in immature neurons within the areas of infarction in HIE.

3.13. Cellular Localization of STAT3 Activation in Areas of Infarction in HIE Following GW2580 Treatment

Our previous results validated the role of STAT3/SDF‐1/CXCR4 activation in regulating cellular changes in the areas of infarction in HIE. However, there are many different types of cells in the brain; thus, we considered it important to specify the localization of key signaling molecules within the brains of experimental animals. Given the important role of the STAT3 signaling pathway in regulating these cell behaviors, we next performed co‐immunostaining of p‐STAT3 with known markers for different cell types, including Iba‐1 for microglia/macrophages, GFAP for astrocytes, SOX‐2 for NSCs, and NeuN for mature neurons. We focused these analyses on day 4 post‐HIE. As shown in Figure 9, p‐STAT3 (green staining) was not observed in brain sections from any sham animals. However, following HIE, there was a significant increase in p‐STAT3 staining in all cell types in areas of infarction, including microglia/macrophages (p < 0.001; Figure 9A,B), NSCs (p < 0.001; Figure 9E,F), and mature neurons (p < 0.001; Figure 9G,H), but not GFAP cells (Figure 9C,D). Following GW2580 treatment, the population of Iba‐1⁺/p‐STAT3⁺ cells was significantly reduced when compared to the HIE‐vehicle group (p < 0.05; Figure 9A,B). In contrast, GW2580 significantly increased the area of GFAP⁺/STAT3⁺ cells in the infarction areas in HIE animals when compared to animals treated with vehicle (p < 0.001; Figure 9C,D). Furthermore, the proportions of SOX‐2⁺/p‐STAT3⁺ cells (Figure 9E,F) or NeuN⁺/p‐STAT3⁺ cells (Figure 9G,H) were significantly increased following GW2580 treatment when compared to the HIE‐vehicle group (p < 0.001 for both).

FIGURE 9.

Cellular localization of STAT3 in infarction areas following GW2580 treatment. (A) Representative images for the co‐immunofluorescence staining of Iba‐1 (red) together with p‐STAT3 (green) and DAPI (blue). (B) Quantification of Iba‐1⁺/p‐STAT3⁺ co‐stained cells (n = 6 per group). (C) Representative images for the co‐immunofluorescence staining of GFAP (red) together with p‐STAT3 (green) and DAPI (blue). (D) Quantification of the co‐localization area for GFAP⁺/p‐STAT3⁺ (n = 6 per group). (E) Representative images for the co‐immunofluorescence staining of SOX‐2 (red) together with p‐STAT3 (green) and DAPI (blue). (F) Quantification of SOX‐2⁺/p‐STAT3⁺ co‐stained cells (n = 6 per group). (G) Representative images for the co‐ immunofluorescence staining of NeuN (red) together with p‐STAT3 (green) and DAPI (blue). (H) Quantification of NeuN⁺/p‐STAT3⁺ co‐stained cells (n = 6 per group). Only data from day 4 are presented. Data are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

Collectively, these results demonstrated that GW2580 inhibited the effects of microglia/macrophages on STAT3 activation in the areas of infarction, while the increased activation of the STAT3 signaling pathway following GW2580 treatment primarily originated from neurons, astrocytes, and NSCs. Specifically, the activation of newly emerged astrocytes, along with increased numbers of neurons and NSCs following treatment, are of particular importance. These results led to the identification of key intercellular communication mechanisms, in which the GW2580‐induced inhibition of microglia may alter the immune status of the microenvironment, leading to STAT3 activation in other neural cell types, thus inducing the activation of repair processes.

3.14. Microglia, Not Macrophages, Mediate Neuroprotective Effects of GW2580

Our previous analyses demonstrated that GW2580 significantly attenuated HIE‐induced neuropathology and neurobehavioral deficits, and that the treatment of HIE animals with GW2580 influenced both microglia and macrophages, two distinctive cell populations that play different roles in HIE‐induced brain pathology. However, the individual contributions of these cell populations to the neuroprotective effects of GW2580 remained unclear. In peripheral tissues, macrophages are well known for their role in driving the transition from a pro‐inflammatory to an anti‐inflammatory microenvironment. However, this shift does not occur in the central nervous system, although our current FC results did reveal a transient increase in macrophages on day 3 post‐HIE which was potentially linked to their role in promoting the anti‐inflammatory microenvironment (Mesquida‐Veny et al. 2021). Accordingly, we next investigated whether macrophages played a similar role in the GW2580‐mediated neuroprotective effect in HIE. To this end, we used clodronate liposomes, a macrophage scavenger that selectively induces apoptosis in macrophages, to determine whether effects were macrophage‐dependent. These liposomes do not cross the blood–brain barrier and, therefore, would have no effect on the microglia. We utilized Iba‐1 immunostaining and FC to validate the effect of clodronate liposomes. Our time‐course experiment found no significant change in the number of Iba‐1⁺ cells when compared between HIE‐vehicle and HIE‐clodronate liposomes groups of rats. FC analyses confirmed that clodronate liposomes significantly reduced CD11b+/CD45^High macrophages without altering CD11b+/CD45^Low microglial proportions (Figure S3). To assess whether macrophages contributed to the neuroprotection effects of GW2580, we co‐administered clodronate liposomes and GW2580 and then employed MRI, Nissl staining, and behavioral tests to evaluate experimental outcomes. When administered alone, clodronate liposomes did not significantly reduce the area of infarction when compared to that in HIE‐control liposome rats, while GW2580 alone led to a significant reduction in the area of infarction when compared to HIE‐control liposome rats (p < 0.001). The combination of GW2580 and clodronate liposomes also led to a significant reduction in the area of infarction when compared to the HIE‐control liposome groups (p < 0.001), with no significant difference between the HIE‐GW2580 and HIE‐GW2580 + clodronate liposome groups (Figure 10A,B). Nissl staining from Day 4 showed no significant difference between the HIE‐control liposome and HIE‐clodronate liposome groups. A significant increase in cell number was observed in both the HIE‐GW2580 group and the HIE‐GW2580 + clodronate liposome group when compared to the HIE‐control liposome group (p < 0.001 for each comparison). No significant difference was found between the HIE‐GW2580 and HIE‐GW2580 + clodronate liposome groups (Figure 10C,D). Behavioral tests from day 3 were consistent with these findings. Administration of clodronate liposomes did not improve behavioral performance in any of the behavioral tests. However, GW2580 alone, and the co‐administration of GW2580 and clodronate liposomes, led to significant improvements in performance in all three behavioral tests (p < 0.001 in the Longa test (Figure 10E), beam walking test (Figure 10F), and p < 0.01 in the rotarod test (Figure 10G). No significant differences were detected in any of the three tests when compared between the HIE‐GW2580 + clodronate and HIE‐GW2580 groups. In summary, clodronate liposome‐mediated macrophage depletion in HIE did not influence the population of microglia, while macrophage depletion alone did not result in any neuroprotective effects. However, the co‐administration of clodronate liposomes and GW2580 achieved effects that were equivalent to GW2580 treatment.

FIGURE 10.

Microglia, not macrophages, were found to mediate the neuroprotective effects of GW2580. (A) MRI showing infarction areas in three slices with an interval of 0.3 mm for every two slices and (B) quantitative analyses (ratio of infarction area to the brain area on the same slice), n = 5 per group. Arrows point to brain infarction area. (C) Representative images for Nissl staining on day 4 and (D) quantitative analysis. n = 6 per group. Behavioral responses in three behavioral tests. Data show mean behavioral score for the Longa test (E), beam walking test (F), and staying time (duration) in the rotarod test (G). Only data from day 3 are presented. In all these tests, n = 8–11. Data are presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001, Student's t test.

Collectively, these findings suggested that macrophages were not the primary cell population responsible for the neuroprotective effects of GW2580 in HIE. Instead, the neuroprotective effects were primarily mediated by the ability of GW2580 to inhibit 5Csf1r signaling, thus leading to a reduction in the number of microglia and an alteration of the microenvironment towards an anti‐inflammatory status.

4. Discussion

In this study, we identified, for the first time, a potential molecular mechanism that underlies the neuroprotective effects of GW2580, a 5Csf1r inhibitor, in a rat model of HIE. Our results revealed that GW2580 promoted the proliferation of NSCs in areas of infarction by inhibiting the 5Csf1r signaling pathway, thus suppressing microglial expansion and facilitating a shift in the microenvironment towards an anti‐inflammatory status. Subsequent activation of the STAT3 pathway then influenced the levels of SDF‐1/CXCR4 expression. We also found that the increased proliferation of NSCs did not lead to the significant generation of new neurons but did inhibit apoptosis in neurons that were resident in the areas of infarction in HIE rats. These findings define a new role for microglia in neuronal repair associated with HIE and further refine the significance of an increased number of NSCs in the brain following HIE. Collectively, these new insights are important to our translational efforts to treat HIE.

There are several merits to our current study. First, multiple approaches, including histological staining, brain imaging, and behavioral tests, were used to evaluate a rat model of HIE, thus leading to robust validation. Second, comprehensive efforts were employed to identify the potential molecular pathways underlying our experimental observations. In total, almost 20 molecules were examined, and at least three signaling pathways were explored. Our investigations identified a new role for microglia in neuronal repair. Furthermore, the neuroprotective effects of GW2580 were demonstrated at different levels, including histology, behavior, and neuronal imaging. Most importantly, this effect was confirmed by identifying the signaling pathway involved, thus providing strong evidence to define a new role of microglia in neuronal repair. Of particular importance is the fact that MRI, performed after GW2580 treatment, revealed significant recovery, which could be attributed to the reduction of edema as evidenced by reduced hyper‐intense shadow in T2‐weighted MR imaging.

Our most significant finding was the elucidation of a specific molecular cascade following the treatment of HIE rats with GW2580. As a microglial inhibitor, GW2580 might exert effects directly by inhibiting the microglia and causing changes in key microglial mediators from pro‐inflammatory factors (e.g., TNF‐α, iNOS) to anti‐inflammatory factors (e.g., IL‐10, Arg‐1) that are associated with HIE. However, our results demonstrated that GW2580 may exhibit other functional roles. Previous studies revealed that STAT3 played a role in modulating neuroinflammation and neural repair (Yang et al. 2017; Zhao et al. 2019). In the present study, we demonstrated that the neuroprotective effects of GW2580 were associated with enhancement of the STAT3 signaling pathway. Furthermore, STAT3 inhibitors reversed these neuroprotective effects, whereas STAT3 agonists exhibited similar protective actions as GW2580. These results provided critical evidence for involvement of the STAT3 activity in GW2580‐mediated neuroprotective effects. In addition, our findings also demonstrated that dynamic changes in NSCs were associated with activation of the STAT3 signaling pathway. Previous studies revealed that following HIE, NSCs from the periventricular zone could migrate to the area of infarction and differentiate into various neural subtypes to initiate endogenous repair (Arvidsson et al. 2002; Li et al. 2022; Merino et al. 2015). In addition, evidence also indicates that the recruitment of NSCs to the area of infarction was driven by expression of the SDF‐1/CXCR4 axis (Carbajal et al. 2010; Merino et al. 2015), a bidirectional regulator of the STAT3 signaling pathway (Ma et al. 2024).

Notably, we found that GW2580 promoted a shift in microglial mediators from pro‐inflammatory factors (e.g., TNF‐α, iNOS) to anti‐inflammatory factors (e.g., IL‐10, Arg‐1). These anti‐inflammatory mediators (e.g., IL‐10) might directly activateSTAT3 (Sharma et al. 2011) to promote up‐regulation of the SDF‐1/CXCR4 axis (Junior et al. 2023), thus suggesting a pivotal role for STAT3. According to our results, the effects of GW2580 were associated with an increased population of NSCs, the phosphorylation of STAT3, and upregulation of the SDF‐1/CXCR4 axis in the areas of infarction in HIE rats. STAT3 inhibitors and agonists reduced and increased, respectively, the number of NSCs and their proliferation in areas of infarction, thus suggesting that activation of the STAT3 signaling pathway contributed to the proliferation of NSCs. Existing evidence also shows that the SDF‐1/CXCR4 axis also facilitates the recruitment of NSCs, and that this axis might enhance the recruitment of NSCs after GW2580 treatment (Carbajal et al. 2010; Merino et al. 2015). In the present study, GW2580‐mediated STAT3 activation promoted the differentiation of NSCs into immature neurons and their subsequent proliferation, thereby increasing the population of immature neurons. Moreover, significant functional recovery was observed as early as day 3 following GW2580 treatment. Given that neuronal maturation, as indicated by the expression of NeuN requires a longer period (e.g., 14 days) (Arvidsson et al. 2002), the absence of BrdU/NeuN co‐staining at this early time point may indicate that the neuroprotective effects of GW2580 were independent of the generation of new neurons. Collectively, these results indicate that increased population of NSCs did not contribute directly to tissue repair via differentiation into mature neurons, but possibly through paracrine effects mediated by cytokines secreted by NSCs and immature neurons (Moretti et al. 2025; Xu et al. 2022). This interpretation is consistent with our present data, thus suggesting that future research should aim to identify specific mediators, particularly those related to NSCs and immature neurons.

Another important finding of this study was the demonstration that GW2580 promoted a shift from a pro‐inflammatory status (e.g., TNF‐α, iNOS) to an anti‐inflammatory status (e.g., IL‐10, Arg‐1). According to previous studies, a pro‐inflammatory microenvironment persists over time following injury to the CNS, resulting in the insufficient transition of microglia and macrophages towards an anti‐inflammatory microenvironment; this imbalance exacerbates neuroinflammation and neuronal apoptosis following injury (Mesquida‐Veny et al. 2021). In the present study, GW2580 significantly suppressed the expression of pro‐inflammatory markers (TNF‐α and iNOS) while enhancing the expression of anti‐inflammatory markers (IL‐10 and Arg‐1) and increasing the number of neurons retained in the area of infarction. These results indicated that the shift towards an anti‐inflammatory microenvironment could promote healing and reduce the tissue damage caused by excessive inflammation. Notably, anti‐inflammatory cytokines, such as IL‐10, have been shown to activate the STAT3 signaling pathway (Ma et al. 2024; Sharma et al. 2011). Collectively, these results suggested that the GW2580‐induced alteration of the microenvironment towards an anti‐inflammatory status may enhance STAT3 activation via cytokine‐mediated mechanisms, thereby establishing a link between our two key findings. It is important to note that the inflammatory mediators utilized in this study were measured in homogenates prepared from areas of infarction, thus reflecting a net change in the local brain microenvironment but not distinguishing the cellular source. Interestingly, we also identified distinct responses of microglia and macrophages in terms of GW2580‐mediated neuroprotection, highlighting regulatory dynamics that were specific to certain cell types. To directly determine whether GW2580 alters the intrinsic phenotypic state of these cells, future studies should analyze isolated microglia and macrophages by utilizing techniques such as FC or single‐cell RNA sequencing. In addition, further experiments that target specific anti‐inflammatory factors (e.g., IL‐10, Arg‐1) will help to elucidate their individual contributions to STAT3 activation and functional recovery.

In peripheral tissues, it has been well‐established that macrophages drive the transition to an anti‐inflammatory microenvironment as pathology progresses. However, this shift does not typically occur in the CNS (Mesquida‐Veny et al. 2021). Nevertheless, our FC results revealed a transient increase in macrophages on day 3 post‐HIE that was potentially associated with their role in the shift towards an anti‐inflammatory microenvironment. Therefore, we employed a macrophage scavenger to determine whether macrophages played a role in GW2580‐mediated neuroprotection in HIE that was similar to their function in peripheral tissues. Analysis revealed that combining the scavenger with GW2580 achieved effects that were similar to those derived from GW2580 treatment, while the sole depletion of macrophages did not result in any neuroprotective effects. These findings suggested that GW2580‐induced neuroprotective effects were not mediated by macrophage‐dependent mechanisms but rather through the direct modulation of microglia.

It is important to consider the mechanism by which GW2580, a Csf1r inhibitor, can promote a tissue profile that is rich in anti‐inflammatory mediators. Our data demonstrate a reduction in pro‐inflammatory microglial cells alongside differential modulation of the profile of inflammatory mediators in the area of infarction by inhibiting Csf1r signaling, promoting the expression of anti‐inflammatory markers (IL‐10, Arg‐1) and suppressing pro‐inflammatory markers (TNF‐α, iNOS). A key question is whether the shift in inflammatory mediators represents an indirect improvement in the inflammatory microenvironment or the direct effect of Csf1r signaling on cellular phenotype. Our results demonstrate that treatment with GW2580 reduced the contribution of microglia/macrophages to STAT3 activation in the area of infarction, while the overall STAT3 activation detected after GW2580 treatment primarily originated from neurons, astrocytes, and NSCs. This key finding suggests that the altered inflammatory status observed in our experiments was largely an indirect, microenvironment‐mediated consequence of reducing the microglia/macrophage cell population, rather than a direct reprogramming effect. Future studies employing the detection of cell‐specific inflammatory cytokines or conditional genetic models could further distinguish these direct and indirect effects. Nonetheless, our findings highlight that modulation of the microglia/macrophage population could effectively reshape the inflammatory microenvironment and activate endogenous repair mechanisms in the brain.

With regards to our animal model, we utilized three‐week‐old rats to generate a model of HIE. According to the findings in many reports, such as from Dr. Noble‐Haeusslein (Semple et al. 2013), at this stage of development, rat brains are almost the same size as adults while maintaining only 50% synaptic density and undergoing peak myelination, suggesting ongoing neural development. This indicated that rats at this age could provide a critical research window for evaluating early nerve damage and therapeutic intervention.

5. Conclusions

In summary, we demonstrated that the partial depletion of microglia and the activation of the STAT3 signaling pathway significantly enhanced the recovery in a neonatal rat model of HIE. The mechanisms underlying these effects involved a shift in microglial mediators from pro‐inflammatory factors (e.g., TNF‐α, iNOS) to anti‐inflammatory factors (e.g., IL‐10, Arg‐1) factors. This shift of the microenvironment promoted the proliferation and differentiation of NSCs and likely involved beneficial paracrine effects of NSCs and immature neurons. Further research is now necessary to fully elucidate the mechanisms by which GW2580 and similar interventions exert their beneficial effects; such research could optimize treatment protocols for clinical use.

Author Contributions

Ao Ding, Guiqin Duan, Mingwei Zhu, and Li Chen contributed equally to this work. The order of authorship among these authors was determined based on the extent of contribution to experimental work and manuscript preparation. Ao Ding designed and performed the experiments, analyzed the data, and wrote the manuscript. Guiqin Duan provided methodological and technical assistances, helped to the experimental design and data analyses, and played a fundamental role in revising the manuscript. Mingwei Zhu and Li Chen contributed to experimental design, performed experiments, analyzed data, and edited the manuscript. Mingwei Zhu also provided partial financial support. Jie Luo, Ting Tan, Zilin Li, and Wenhui Wang contributed to specific experimental procedures and data acquisition. Jun Wang provided methodological and technical assistance. Yuan Chen and Ya‐Ping Tang supervised the study. Ya‐Ping Tang also contributed to the experimental design and secured the financial supports.

Funding

This work was supported by Guangzhou Health Care Cooperative Innovation Major Project (201704020221) and the Natural Science Foundation of Guangdong Province (2022A1515012506), both to M.Z.; and the National Natural Science Foundation of China (NSFc) (31671098 and 81620108021), Key Scientific and Technological Project of Guangzhou City (202007030002) and Guangdong Province (2018B030335001), and Key Technologies R&D Program of Sichuan Province of China (14ZC0054), all to Y‐P.T.

Ethics Statement

All the experiments on animals were conducted in accordance with the provisions for animal care and use described in “Guidance for the Care and Use of Laboratory Animal” issued by NSFC, and were pre‐approved by the IACUC in GWCMC (Ethics No. 2019‐23001).

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: HIE rat models. (A). Schematic diagram of experimental procedures. Three‐week‐old SD rats were subjected to either HIE or sham‐surgery. MRI and four behavioral tests were conducted 24 h after HIE. Sample collection was carried out in each day from day 1 to day 7 after HIE. (B). MRI shows the infarction areas in five slices with an interval of 0.3 mm for every two slices (left) and their quantitative analyses (right, ratio of infarction areas to the brain areas on the same slice; n = 5 in each group). The arrows point to the brain infarction areas. (C). TTC staining shows the infarction areas in five slices with an interval of 2 mm for every two slices (left) and their quantitative analyses (right, ratio of infarction area to the brain area on the same slice; n = 3 in each group). (D). Representative pictures of Nissl staining (left) and their quantitative analysis (right) (n = 6 in each group). The position of the dashed line is the boundary of the infarction area. (E‐H). Behavioral tests: Average behavioral score in Longa test (E) and beam walking test (F), misstep rate in grid walking test (G), and staying time (duration) in rotarod test (H). In all these tests, n = 8–10. All data are presented as mean ± SEM, and *, p < 0.05; **, p < 0.01; ***, p < 0.001, all Student's t test. For TTC staining, however, the non‐parametric Mann–Whitney U test was used due to the limited sample size, with multiple comparisons adjusted using the Bonferroni method.

Figure S2: Immature neurons increase in the infarction area following GW2580 treatment. (A) Representative pictures of co‐immunostaining of DCX (red) together with BrdU (green) and DAPI (blue). (B) Quantitative analyses of the DCX+/BrdU+ co‐stained cells (n = 6 in each group). Only the data in day 3 was presented. Data were presented as mean ± SEM, and *, p < 0.05; **, p < 0.01; ***, p < 0.001, all Student's t test.

Figure S3: Clodronate‐mediated macrophage depletion in HIE did not alter microglial proportions. (A) Schematic diagram of experimental procedures. Rats were treated with GW2580, clodronate liposomes or control liposomes at 2 h following HIE. MRI was conducted 24 h after HIE, and then animals were evaluated by three behavioral tests in consecutive 3 days. Sample collection was conducted in day 3, day 4, and day 7 after HIE. (B and C) Representative pictures of co‐immunostaining of Iba‐1 (red) together with DAPI (blue) in day 3 (B) and their quantitative analysis (C; n = 6 in each group). (D‐F) FC differentiates microglia (CD11b+/CD45^Low) and macrophages (CD11b+/CD45^High) (D) and their quantitative analysis at day 3 (E) or day 7 (F), n = 6 in each group. Data were expressed as mean ± SEM, and *, p < 0.05; **, p < 0.01; ***, p < 0.001, all Student's t test.

Data Availability Statement

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