Sfrp2 in Microglia Inhibited S100a8-Mediated Neuroinflammation and Protected Neural Damage Following Retinal Ischemia-Reperfusion

Abstract

Purpose

The purpose of this study was to explore retinal Sfrp2 expression, functional roles, and underlying mechanism in retinal ischemia reperfusion (I/R).

Methods

We established an I/R mouse model in vivo and a lipopolysaccharide-stimulated BV2 model in vitro. Immunohistochemistry and Western blotting (WB) assessed Sfrp2 expression in human retinal tissue, and WB further evaluated its expression in I/R model. Immunofluorescence staining was used to define its localization. Sfrp2 expression was ablated by Cre-lox mediated conditional deletion in mouse retinal microglia and by siRNA-mediated knockdown in BV2 cells. Hematoxylin and eosin staining, immunofluorescence, and TUNEL staining were applied for evaluating retinal structure, survival of inner retinal neurons, and cell apoptosis, respectively. Retinal function was evaluated by electroretinography. RNA-sequencing, WB, and immunoprecipitation were used to elucidate the underlying mechanisms.

Results

Sfrp2 was highly expressed in the retina subjected to I/R injury. Sfrp2 was shown to localize in microglia in an I/R model. Correspondingly, microglial Sfrp2 deficiency exacerbated the retinal structural damage, inner retinal neuronal degeneration, visual dysfunction, and microglia-mediated inflammation in the I/R model. Mechanistically, Sfrp2 deficiency activated TRAF6-TAK1-NF-κB signaling cascade by promoting the TRAF6-TAK1 complex formation, subsequently increasing the expression of downstream inflammatory mediator S100a8. Importantly, S100a8 inhibitor could partially mitigate the retinal injury induced by microglial Sfrp2 deficiency following I/R.

Conclusions

Sfrp2 in microglia inhibits TRAF6-TAK1-NF-κB signaling pathway and downstream inflammatory mediator S100a8 thereby attenuating neuroinflammation and protecting the retina in I/R. Sfrp2 may represent a protective strategy for neuronal degeneration in ischemic retinopathy.

Retinal ischemia-reperfusion (I/R) is a pathophysiological process that has been found to be connected to a large number of retinal neurodegenerative diseases, such as glaucoma, diabetic retinopathy (DR), and age-related macular degeneration (AMD), which are major causes of visual impairment or blindness.^1–3 Neuroinflammation plays a pivotal role in I/R injury.^4,5 Pro-inflammatory cytokines and chemokines may trigger neuronal apoptosis and cell death, thereby accelerating the progression and deterioration of neurodegenerative diseases.^6–8 Despite extensive investigation into neuroinflammation in retinal degenerative diseases, no specific pharmacological treatment is available for retinal I/R injury. Current therapeutic strategies mainly rely on anti-inflammatory, antioxidant, and neuroprotective interventions, yet their clinical efficacy remains limited. Therefore, further exploration of neuroinflammatory mechanisms in retinal I/R injury remains warranted, as targeting neuroinflammation may help attenuate ongoing neural damage.

A hallmark of neuroinflammation is the activation of microglia. Persistent stress drives microglia toward a pro-inflammatory state, characterized by elevated phagocytic activity and augmented release of neurotoxic mediators.⁹ In response to I/R injury or lipopolysaccharide (LPS) stimulation, these microglia become activated, undergoing phenotypic and functional changes to shield the central nervous system (CNS) or retina from infection or damage.^10,11 In this process, a broad spectrum of inflammation-associated signaling pathway is triggered.^12,13 The TRAF6-TAK1 axis, a common downstream integrator of diverse inflammatory pathways, is activated in this context, thereby driving the subsequent activation of the NF-κB signaling cascade.^14,15 These processes amplify neuroinflammatory responses and ultimately exacerbate neuronal degeneration.¹⁶ Multiple studies have elucidated that TRAF6-TAK1 signaling plays a critical role in regulating inflammation in both the CNS and the retina.^17,18 Thus, suppressing the activation of TRAF6-TAK1 signaling pathway in microglia may represent a potential therapeutic target for modulating retinal neurodegenerative disease progression.

Secreted frizzled-related proteins (SFRPs) are glycoproteins characterized by a frizzled-like cysteine-rich domain, which allows them to interact with Wnt ligands or frizzled receptors, thus functioning as key modulators of Wnt signaling. Five members of the SFRP family (Sfrp1–Sfrp5) have been identified in mammals.¹⁹ As a member of the SFRP family, Sfrp2 plays an important role in inflammatory diseases. Sfrp2 has been shown to exacerbate airway inflammation in chronic obstructive pulmonary disease by regulating the canonical Wnt signaling pathway.²⁰ Moreover, it is reported that targeting elevated Sfrp2 expression in the dorsal root ganglion could alleviate gouty arthritis.²¹ Additionally, numerous studies have also elucidated that Sfrp2 was implicated in retinal diseases, such as uveitis,²² retinitis pigmentosa (RP),^23,24 and retinoblastoma.^25,26 It is also reported that Sfrp2 plays a crucial role in retinal neural regulation, encompassing processes such as retinal neurogenesis and axon guidance.^27–29 As noted above, Sfrp2 has been extensively studied in inflammation and retinal neural regulation. However, in the absence of effective therapies for retinal I/R injury, investigating the role of Sfrp2 in I/R-induced retinal neuroinflammation may provide mechanistic insights and support the development of future therapeutic strategies.

Retinal I/R-induced neuroinflammation is typically accompanied by the robust production of inflammatory mediators.³⁰ S100a8, a member of the Ca²⁺-binding S100 protein family, typically forms a stable heterodimer with S100a9 and has recently gained considerable attention as a critical alarmin that modulates inflammatory responses and participates in retinal neuroinflammation.^31,32 However, the precise mechanisms underlying the involvement of S100a8 in I/R, and the therapeutic potential of targeting S100a8, have yet to be elucidated.

Therefore, in this study, we identified the expression and localization of Sfrp2 in I/R. We explored the functional roles of Sfrp2 by conditional knockout mice via selective ablation of Sfrp2 in microglia and silencing Sfrp2 expression in BV2. Finally, we identified Sfrp2 as a microglial regulator that plays a non-redundant role in restraining neuroinflammation in I/R-induced retinal injury by inhibiting downstream inflammatory mediator S100a8.

Materials and Methods

Clinical Specimens

Human retinal tissues were obtained from organ donors with retinal ischemia (n = 3) and from healthy donors without retinal disease (n = 3). Patients with retinal ischemia were diagnosed with central retinal vein occlusion, ocular ischemic syndrome, and primary open-angle glaucoma, respectively. Participants with concomitant retinal diseases that could confound retinal inflammatory responses, were excluded from this study. All donors died as a result of traffic accidents. The postmortem interval to tissue collection was within 5 hours, and all experiments were initiated within 48 hours after death. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Committee of the People's Hospital of Guangxi Zhuang Autonomous Region (KY-GZR-2021-108). Informed consent was obtained from the families of deceased donors. All procedures involving human eye tissue complied with the ethical standards outlined in the ARVO Best Practices for Using Human Eye Tissue in Research. The baseline characteristics of the patients are listed in Supplementary Table S1.

Immunohistochemistry of Human Retinal Tissue

Immunohistochemistry (IHC) samples were fixed in 4% paraformaldehyde (PFA; Solarbio Life Sciences, Beijing, People's Republic of China), dehydrated, and cleared prior to paraffin embedding. Serial sections (10 µm) were prepared and subjected to antigen retrieval by microwave treatment. After blocking with 5% goat serum, the sections were incubated overnight at 4°C with primary antibodies against Sfrp2. Appropriate secondary antibodies were subsequently applied, and immunoreactivity was visualized using the avidin-biotin complex method with 3,3′-diaminobenzidine as the chromogen. Target proteins were identified by the presence of brown staining.

Animals

Male and female C57BL/6J mice and homozygous Sfrp2^−/− mice (Sfrp2^fl/flCx3cr1^CreERT2) were purchased from Cyagen Biotechnology Co., Ltd. and Shanghai Model Organisms. The mice were housed under a specific pathogen-free and temperature-controlled facility on a 12-hour light/dark cycle in Animal Laboratories of Zhongshan Ophthalmic Center. All animal procedures were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee of the Zhongshan Ophthalmic Center, Sun Yat-Sen University.

I/R Model and Drug Administration

The I/R model was established according to our prior study.^33,34 Briefly, anesthesia was administered via intraperitoneal injection of 1% pentobarbital solution (0.1 mL/10g; Sigma, St. Louis, MO, USA). The pupils were dilated by tropicamide phenylephrine and corneas were given anesthesia via 0.5% tetracaine. Sterile saline was infused into the anterior chamber of the right eye through a 32-gauge needle, maintaining the intraocular pressure (IOP) at 110 millimeters of mercury (mm Hg) for 60 minutes. The needle was then withdrawn to allow the IOP to return to normal. The contralateral eye without IOP elevation served as the non-I/R control.

To investigate the effect of S100a8 inhibitor, paquinimod (PAQ; Selleck Chemicals, Houston, TX, USA) was prepared in a stock solution containing 5% dimethyl sulfoxide (DMSO; Solarbio Life Sciences, Beijing, People's Republic of China) and 95% corn oil, and subsequently diluted in saline to a final concentration of 5% (v/v). The concentration of PAQ administrated intraperitoneally was determined based on the previous study.³⁵ After the establishment of the I/R model, PAQ was administered intraperitoneally at a dose of 10 mg/kg/day, once daily for 3 consecutive days. The vehicle control group received intraperitoneal injections of a 5% solution containing DMSO and 95% corn oil.

Western Blotting

Human or mouse retinas or cells were lysed in RIPA lysis buffer (Servicebio Technology, Hubei, People's Republic of China) supplemented with protease and phosphatase inhibitors (Sigma, St. Louis, MO, USA). The concentrations of retinal and cellular proteins were measured with BCA Protein assay kit (Beyotime, Beijing, People's Republic of China). Proteins were separated by SDS-PAGE and transferred onto polyvinylidene fluoride membranes, which were then blocked with 5% nonfat milk at room temperature for 2 hours. Membranes were incubated overnight at 4°C with primary antibodies, followed by incubation with the appropriate secondary antibodies for 2 hours at room temperature. Protein band intensities were quantified using ImageJ software (National Institutes of Health, Baltimore, MD, USA). Details of antibodies are listed in Supplementary Table S2. The original Western blotting (WB) images are shown in Supplementary Figure S1.

Immunofluorescence Staining

For cryosections, the eyeballs were dehydrated through a graded sucrose series and embedded in optimal cutting temperature (OCT; Sakura Finetek, Torrance, CA, USA) compound and then sliced transversely (8 µm) with a cryostat at –20°C. For retinal flat-mounts, the eyeballs were fixed in 4% PFA for 45 minutes at room temperature and the retinas were subsequently dissected out as cups. For cellular immunofluorescence (IF) staining, BV2 cells were fixed in 4% PFA for 30 minutes at room temperature and permeabilized with 0.2% Triton X-100 (Solarbio Life Sciences, Beijing, People's Republic of China) in PBS for 5 minutes. Cryosections, retinal cups and cells were incubated with primary antibodies, followed by incubation with the corresponding secondary antibody. Nuclei were counterstained with 4’,6-diamidino-2-phenylindole (DAPI, Servicebio Technology, Hubei, People's Republic of China). Confocal imaging was conducted using a ZEISS LSM 980 microscope (Carl Zeiss Microscopy, Oberkochen, Germany). Retinal whole-mounts were imaged at 20 × magnification, whereas retinal sections and cells were imaged at 40 × oil-immersion magnification. Image stacks were acquired with a z-step size of 0.8 µm. Laser power was set to 5% for 488 nm, 5% for 555 nm, and 1% for 405 nm excitation, with a pixel dwell time of 1.27 µs and a frame size of 1024 × 1024. All retinal images were systematically collected from the middle retina, and uniform post-processing, including background subtraction and contrast adjustment, was applied using Zeiss ZEN 3.7 software. Details of antibodies are listed in Supplementary Table S2.

Hematoxylin and Eosin Staining

Hematoxylin and eosin (H&E) staining was performed on paraffin-embedded retinal sections. Briefly, eyeballs were fixed in 4% neutral-buffered formalin, dehydrated through a graded ethanol series, cleared in xylene, and embedded in paraffin. Sections (5 µm) were deparaffinized, rehydrated, and sequentially stained with H&E. After dehydration and clearing, the sections were mounted with coverslips. Images of the stained sections were acquired using a Pannoramic MIDI scanner (3DHISTECH, Budapest, Hungary). For histological evaluation, six eyeballs were analyzed per experimental group. From each eyeball, three non-adjacent retinal sections passing through the optic nerve were randomly chosen. One region from each section was then selected for quantitative assessment. Accordingly, the final average value for each eyeball was calculated from measurements obtained from three regions across the three sections. The thicknesses of retinal sublayers including nerve fiber layer/ganglion cell layer (NFL/GCL), inner plexiform layer (IPL) and inner nuclear layer (INL), and total retinal thickness were quantified using ImageJ software (National Institutes of Health, Baltimore, MD, USA). Measurements were performed at 3 predefined retinal locations, situated approximately 500 µm (central), 1100 µm (middle), and 1700 µm (peripheral) from the optic nerve head.³⁴ A reference line was drawn orthogonally across retinal layers at each site to ensure consistent measurement. Layer boundaries were defined according to the established literature.^36,37

TUNEL Assay

TUNEL staining (In Situ Cell Death Detection Kit, Fluorescein; Roche, Indianapolis, IN, USA) was performed according to the manufacturer's instructions. The sections were stained with TUNEL for 1 hour at 37°C after fixing and blocking, and then mounted with anti-fade medium containing DAPI. Images were photographed using a confocal microscope (ZEISS LSM 980).

Electroretinography

The mice were dark-adapted overnight for 12 hours. After anesthesia, the pupils of the mice were dilated. Two gold ring active electrodes were positioned on the corneal surface, with the reference and ground electrodes placed subcutaneously in the cheek and tail, respectively. Scotopic electroretinography (ERG) and responses were elicited at light intensities of 0.01, 3.0, and 10.0 cd s/m². The photopic negative response (PhNR) was elicited with a 20-cd s/m² flash stimulus. Amplitudes of the scotopic a-wave, b-wave, oscillatory potentials (OPs), and PhNR were subsequently measured.

RNA Sequencing

RNA sequencing (RNA-seq) was carried out by Gene Denovo Biotechnology Co. (Guangzhou, People's Republic of China). Total RNAs were extracted from I/R and I/R + Sfrp2^–/– retinas at 3 days after I/R injury by Trizol reagent (Vazyme, Nanjing, People's Republic of China). RNA quality was assessed on an Agilent 2100 Bioanalyzer (Agilent Technologies). Poly(A)-enriched RNA libraries were constructed through standard cDNA synthesis and adaptor ligation workflows and sequenced on the Illumina NovaSeq 6000 platform. After quality assessment and library construction, the raw sequencing data were processed using the R software package DESeq.2 for normalization and identification of the differential expression genes (DEGs). DEGs were screened using |log2(fold change)| >1 and false discovery rate <0.05. DEGs were then obtained for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis and Gene Ontology (GO) analysis. Volcano plot was generated using the open sources R software language, and heat maps were produced with the Multi-Experiment Viewer (MEV) program.

Cell Culture and Treatment

BV2 microglial cells (Kunming Institute of Zoology, Chinese Academy of Sciences, Beijing, People's Republic of China) were maintained in DMEM (Thermo Fisher, Waltham, MA, USA) supplemented with 10% FBS and streptomycin-penicillin (Thermo Fisher, Waltham, MA, USA). The cells were identified using Iba-1. The Sfrp2-siRNA was purchased from GenePharma (Shanghai, People's Republic of China). For cell transfection, Sfrp2-siRNA (CCGAAAGGGACCTGAAGAATT) was diluted in buffer (Lipofectamine transfection reagent; Invitrogen, Carlsbad, CA, USA) and transfected into BV2 cells at a working concentration of 100 nM according to the manufacturer’s protocol prior to LPS (Sigma, St. Louis, MO, USA; 100 ng/mL) stimulation. LPS was added to the cultures for 6 hours, 12 hours, 24 hours, and 48 hours.

Immunoprecipitation

BV2 cells were stimulated with or without Sfrp2-siRNA and LPS (100 ng/mL, 24 hours) at the indicated time point. Retinal and cellular proteins were extracted and homogenized in RIPA lysis buffer.

Immunoprecipitation (IP) was performed according to the manufacturer’s instructions. For each sample, 5 µg of antibody was added to the protein solution, gently vortexed to mix, and incubated at 4°C for 12 to 16 hours overnight. Then, 20 µL of magnetic beads were prepared in a centrifuge tube, washed, and vortexed. On the next day, the incubated samples were added to the prepared magnetic beads and incubated at 4°C for 12 to 16 hours overnight. The supernatant was aspirated on a magnetic rack, and the magnetic beads were washed with IP lysis buffer, repeated three times. Subsequently, 50 µL of SDS-PAGE loading buffer was added to each tube and boiled at 99°C for 10 minutes. The supernatant containing loading buffer was retained following separation with a magnetic rack. Finally, the target proteins were analyzed by WB. Details of antibodies are listed in Supplementary Table S2.

Statistical Analysis

IHC, WB, and IP analyses were each performed in triplicate. Retinal whole-mount, frozen section, and cellular IF staining, H&E staining, TUNEL staining, and ERG were each conducted in six independent replicates. Statistical analyses were performed using unpaired two-tailed Student's t-test and 1-way ANOVA followed by Tukey's post hoc test. Data were presented as the mean ± standard error of mean (SEM). GraphPad Prism 9.2 (GraphPad Software, Boston, MA, USA) was used to perform the statistical analyses. Statistical significance was established when the P value was <0.05.

Results

The Expression of Sfrp2 Was Upregulated in Patients With Retinal Ischemia and in an I/R Mouse Model

To evaluate the expression pattern of Sfrp2 in patients with retinal ischemia, IHC and WB were performed. IHC analysis showed that Sfrp2 expression was upregulated in patients affected by retinal ischemia, whereas it was restricted to low expression in donors without retinal ischemia (Fig. 1A, 1C). WB also confirmed the result, showing the increased expression of Sfrp2 in patients with retinal ischemia (Fig. 1B, 1D). We then detected the temporal dynamics of Sfrp2 protein expression in an I/R mouse model at 6 hours, 12 hours, 1 day, 3 days, and 7 days post-injury. WB demonstrated that the protein level of Sfrp2 progressively increased in the retinas from 6 hours after injury, peaked at day 1 after induction, and dropped thereafter (Figs. 1E, 1F).

Figure 1.

Sfrp2 expression and cellular localization following I/R injury. (A) Representative IHC images showing Sfrp2 expression in retinal tissues from healthy controls and patients with retinal ischemia. Scale bar = 20 µm. (B) WB analysis of Sfrp2 expression in retinal tissues from healthy controls and patients with retinal ischemia. (C, D) Quantification of Sfrp2 expression in retinal tissues from healthy controls and patients with retinal ischemia in IHC and WB analyses, n = 3, *P < 0.05; **P < 0.01; 1-way ANOVA. (E, F) WB analysis and quantification of the time course of Sfrp2 in I/R model, n = 3, *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significance; 1-way ANOVA. (G, H) Representative images of mice frozen retinal sections showing co-labeling of Sfrp2 and iba1, along with quantification of expression levels of Sfrp2 positive microglia in each group, 1 day after I/R. Scale bar = 20 µm. Enlarged image is shown below the main figures. Scale bar = 4 µm. Blue indicates DAPI staining, n = 6, ***P < 0.001; Student's t-test. (I) Timeline of the experimental workflow for the mice models. Data are presented as mean ± SEM.

Given the critical role of microglia in I/R injury, we next performed double staining with Iba1, a marker for microglia, and Sfrp2 to investigate whether Sfrp2 was expressed in microglia. Co-immunostaining demonstrated that Sfrp2 was almost undetectable in the normal retina, whereas it was localized to microglia following I/R injury (Fig. 1G, 1H). Together, these data demonstrated that Sfrp2 expression was markedly increased after I/R and was localized to microglia, implying that Sfrp2 might contribute to the regulation of neuroinflammatory responses.

Sfrp2 Deficiency in Microglia Aggravated the Retinal Structural Damage and the Loss of Inner Retinal Neurons in an I/R Model

Degeneration of retinal neurons results in retinal thinning.³⁸ We next examined the retinal thickness to determine whether Sfrp2 influences neuronal degeneration in I/R. To explore the role of Sfrp2 in retinal structure, Sfrp2^–/– mice were established and then H&E staining was performed. We observed that the thicknesses of retinal sublayers (NFL/GCL, IPL, and INL) in the central, middle, and peripheral regions, and total retinal thickness across all regions did not differ between the Sfrp2^–/– and control groups. However, compared with the I/R group, the I/R + Sfrp2^–/– group exhibited a marked reduction in the thickness of different retinal sublayers (NFL/GCL, IPL, and INL) and total retinal thickness across all regions (Figs. 2A–E).

Figure 2.

Sfrp2 deficiency exacerbated retinal structural damage and reduced RGC survival following I/R injury. (A) Representative H&E images from the healthy control and I/R groups (7 days after I/R) with or without Sfrp2 deficiency. Scale bar = 20 µm. (B–E) Quantification of retinal sublayers and total thickness (n = 6). (F) Representative images of flat‐mounted retinas of RBPMS and β3‐Tublin in each group (7 days after I/R). Scale bar = 50 µm. (G) Quantification of RBPMS and β3‐Tubulin‐labeled RGCs in each group (n = 6). *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM.

Given the observed thinning of the inner retinal layers in the I/R + Sfrp2^–/– group, we next investigated which inner retinal neuronal populations were undergoing degeneration. To evaluate the effect of Sfrp2 on RGC survival, retinal whole flat‐mount IF staining was performed using RBPMS and β3-tubulin markers. We found that RGC density was comparable between the control and the Sfrp2^–/– groups. However, levels of these two markers were reduced in the I/R + Sfrp2^–/– group compared with the I/R group (Figs. 2F, 2G). Then, we investigated whether Sfrp2 acts on other types of retinal neuron cells. We performed frozen section IF staining with neuronal markers, including calretinin (amacrine cells), calbindin (horizontal cells), and PKC-α (bipolar cells). The control and Sfrp2^–/– groups displayed no discernible discrepancy in the quantity of amacrine (both in GCL and INL), horizontal and bipolar cells, and axonal length of bipolar cells. However, the I/R + Sfrp2^–/– group exhibited a more pronounced loss of amacrine cells (both in GCL and INL) compared with the I/R group. Although their numbers remained unchanged, bipolar cells in the I/R + Sfrp2^–/– group exhibited shortened axonal fibers compared with those in the I/R group (Figs. 3A–E). Additionally, to evaluate the effect of Sfrp2 on I/R-induced apoptosis of retinal neurons, TUNEL staining was applied to quantify apoptotic cells. TUNEL staining showed that Sfrp2 deficiency in microglia significantly increased I/R-induced apoptosis of RGCs and neurons in INL (Figs. 3F, 3G). These results showed that microglial Sfrp2 deficiency exacerbated retinal structural damage and inner retinal neuronal deterioration induced by I/R.

Figure 3.

Sfrp2 deficiency aggravated inner retinal neurons damage and apoptosis after I/R injury. (A) Representative immunofluorescence images of calretinin, calbindin and PKC-α from the healthy control and I/R groups with or without Sfrp2 deficiency (7 days after I/R). (B–E) Quantification of calretinin⁺ amacrine cells, calbindin⁺ horizontal cells, PKC-α⁺ bipolar cells, and axonal length of PKC-α⁺ bipolar cells (n = 6). (F, G) Representative images and quantification of TUNEL⁺ cells in each group (1 day after I/R, n = 6). *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM. Scale bar = 20 µm.

Sfrp2 Deficiency in Microglia Aggravated Visual Dysfunction Following I/R Injury

We further investigated whether Sfrp2 deficiency in microglia affects retinal visual function following I/R injury. ERG recordings were performed to measure electrical responses, including the scotopic a-wave under both 3.0 and 10.0 cd s/m² conditions (Figs. 4B, 4C, 4G, 4I), b-wave under 0.01, 3.0 and 10.0 cd s/m² conditions (Figs. 4A–C, 4F, 4H, 4J), OPs (Figs. 4D, 4K), and PhNR (Figs. 4E, 4L), which represented the activity of photoreceptors, bipolar cells, amacrine cells, and RGCs, respectively.^39,40 Our data showed no differences in the amplitudes of the a-wave, b-wave, OPs, or PhNR between the control and Sfrp2^–/– groups. These amplitudes were significantly reduced in the I/R group. Sfrp2 deficiency in microglia dramatically aggravated the reduction of these functional parameters after I/R injury. Altogether, these data provided evidence that microglial Sfrp2 deficiency exacerbated visual dysfunction following I/R injury.

Figure 4.

Sfrp2 deficiency impaired retinal visual function in the I/R group. (A–E) Representative waveforms of scotopic ERG, OPs, and PhNR from each group under different flashlight intensities (7 days after I/R). (F–L) Quantification of the amplitudes of a‐wave, b‐wave, OPs, and PhNR (n = 6). *P < 0.05; ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM.

Sfrp2 Deficiency Triggered Microglia-Mediated Inflammation After I/R Injury

As retinal microglia served as a potent inducer of neuroinflammation in the retina, we detected the activation of microglia following I/R. We further performed double immunostaining using Iba1 together with CD68 (Figs. 5A, 5B) and CD16 (Figs. 5C, 5D) to label activated and pro-inflammatory microglia, respectively. Our results showed that there were almost no changes of the numbers of CD68⁺ Iba1⁺ and CD16⁺ Iba1⁺ cells in the control group compared with Sfrp2^–/– retinas. The I/R + Sfrp2^–/– group exhibited increased numbers of CD68⁺ Iba1⁺ and CD16⁺ Iba1⁺ microglia cells compared with the I/R group, suggesting a more notable inflammatory response in I/R mice with microglial Sfrp2 deficiency. Thus, our data suggested that Sfrp2 deficiency in microglia exacerbated retinal inflammatory responses after I/R injury.

Figure 5.

Sfrp2 deficiency increased the activation of microglia in the I/R group. (A, B) Representative images showing co-labeling of Iba1 and CD68, and quantification of Iba1⁺ and CD68⁺ cells in each group (3 days after I/R, n = 6). (C) Representative immunofluorescence images of retinal sections stained for Iba1 and CD16 in each group. (D) Quantification of Iba1⁺ and CD16⁺ cells in each group (n = 6). ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM. Scale bar = 20 µm.

Sfrp2 Ablation in Microglia Modulated TRAF6-TAK1-NF-κB Signaling Pathway and Downstream Inflammatory Mediator S100a8 Following I/R Injury

To elucidate Sfrp2 regulation of inflammation, we first performed bulk RNA‐seq of retinas from wild type and Sfrp2^–/– mice at 3 days after I/R injury. RNA‐seq analysis revealed that microglia-specific Sfrp2 deletion resulted in significant upregulation of 124 genes and downregulation of 53 genes in the retina (Fig. 6A). KEGG analysis indicated enrichment in inflammation-associated pathways, including IL-17, NF-κB, and JAK-STAT signaling pathways (Fig. 6B). To further confirm this finding at the protein level, we examined the activation of inflammatory pathways by WB. Multiple inflammatory signaling pathways, including IL-17 and TLR4, have been shown to converge on downstream effectors such as TRAF6.^41–43 Autopolyubiquitination of TRAF6 promotes the recruitment of downstream regulators, including TAB2/3 and TAK1, leading to TAK1 phosphorylation and subsequent activation of NF-κB signaling.^15,44 Thus, we detected the expression levels of TRAF6, Phospho-TAK1, and Phospho-NF-κB-P65. Consistent with the bulk RNA-seq, WB showed that the protein levels of Phospho-TAK1 and Phospho-NF-κB-P65 were markedly upregulated in the I/R + Sfrp2^–/– group (Figs. 6C, 6D). However, the protein levels of TRAF6 and Phospho-STAT3 were comparable between the I/R and I/R + Sfrp2^–/– groups (see Figs. 6C, 6D). Additionally, GO analysis revealed that Sfrp2 deletion significantly enriched inflammation-related biological processes comprising 27 upregulated genes, including S100a8, a member of the Ca²⁺-binding S100-protein family, which has been extensively studied for its involvement in inflammatory responses^32,45 (Fig. 6E). WB analysis confirmed the bulk RNA-seq findings, demonstrating elevated S100a8 protein level in the I/R + Sfrp2^–/– group relative to the I/R group (Figs. 6F, 6G). Meanwhile, IF staining demonstrated that S100a8 was expressed in microglia, suggesting that S100a8 might participate in microglia-mediated inflammatory responses (Fig. 6H, 6I).

Figure 6.

RNA-seq revealed pathways underlying the neuroinflammatory effects of Sfrp2 deficiency. (A) Volcano plot presenting upregulated and downregulated genes between the I/R + WT and I/R + Sfrp2^–/– groups. (B) KEGG enrichment analysis of the DEGs illustrating enriched pathways. (C) WB results showing the expression changes of pathway proteins including TRAF6, p-TAK1, TAK1, p-NF-κB, NF-κB, p-STAT3, and STAT3 in the I/R and I/R + Sfrp2^–/– groups. (D) Quantification of the protein expression of TRAF6, p-TAK1, p-NF-κB, and p-STAT3 (n = 3). **P < 0.01; ns, no significance; 1-way ANOVA. (E) Heatmap showing the upregulated inflammatory genes involved in inflammatory response of GO enrichment analysis. (F) WB results illustrating the protein expression of IL-1β, S100a8, Ccl2, Ccl7, and Timp1. (G) Quantification of the protein expression of IL-1β, S100a8, Ccl2, Ccl7, and Timp1 (n = 3). *P < 0.05; ns, no significance; 1-way ANOVA. (H) Representative immunofluorescence images of retinal sections stained for Iba1 and S100a8 in each group (3 days after I/R). Scale bar = 20 µm. (I) Quantification of Iba1⁺ and S100a8⁺ cells in each group (n = 6). **P < 0.01; ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM.

These in vivo findings prompted us to investigate how Sfrp2 modulates the downstream pathways in vitro. BV2 cells were stimulated with LPS to establish an in vitro model of neuroinflammation.⁴⁶ We detected the temporal dynamics of Sfrp2 protein expression in the control group, and at 6 hours, 12 hours, 24 hours, and 48 hours after LPS stimulation. Consistent with the in vivo findings, the expression of Sfrp2 in BV2 cells also exhibited a time-dependent pattern, peaking at 1 day after LPS stimulation in WB (Figs. 7A, 7B). IF staining further confirmed the increased expression of Sfrp2 in BV2 cells following LPS stimulation (Fig. 7C). Consistently, Sfrp2 knockdown in BV2 cells markedly enhanced TAK1 and NF-κB phosphorylation and S100a8 expression upon LPS stimulation, suggesting activation of TRAF6-TAK1-NF-κB signaling pathway following stimulation (Figs. 7D, 7E). Collectively, transcriptomic pathway enrichment, together with increased phosphorylation of key signaling components indicated that microglial Sfrp2 deficiency might play a pivotal role in regulating the inflammatory response after I/R injury by activating the TRAF6-TAK1-NF-κB signaling pathway and inducing high expression of inflammatory mediator S100a8.

Figure 7.

Sfrp2 deficiency increased the interaction of proteins in TRAF6-TAK1 signaling pathway. (A, B) WB analysis and quantification showing the time-dependent expression of Sfrp2 in BV2 cells with or without LPS stimulation (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significance; 1-way ANOVA. (C) Representative immunofluorescence images of Iba1 and Sfrp2 in BV2 cells with or without LPS stimulation. Scale bar = 20 µm. (D) WB analysis depicting the expression of S100a8, p-TAK1, TAK1, p-NF-κB, NF-κB, and Sfrp2 in LPS-stimulated BV2 cells with or without the treatment of Sfrp2-siRNA. (E) Quantification of the expression of S100a8, p-TAK1, p-NF-κB, and Sfrp2 in each group (n = 3). *P < 0.05; **P < 0.01; 1-way ANOVA. (F) The interactions of Sfrp2 with TRAF6 and TAK1 in BV2 cells with or without LPS stimulation. (G) The interaction of TRAF6 with TAK1 in LPS-stimulated BV2 cells with or without the treatment of Sfrp2-siRNA. (H) Timeline of the experimental workflow for the mice models, including interventions and assessments. Data are presented as mean ± SEM.

LPS-Induced Inflammation in Sfrp2-Deficient Microglia Initiated Downstream Signaling Cascades by Enhancing TRAF6-TAK1 Complex Formation In Vitro

Because the TRAF6-TAK1 signaling pathway was activated in microglial Sfrp2 deficiency group following I/R injury, we further explored how Sfrp2 regulates the aforementioned signaling pathway by IP assays. To this end, we first investigated whether Sfrp2 can interact with TRAF6 or TAK1 following LPS stimulation. WB analysis revealed specific Sfrp2-conjugated TRAF6 and TAK1 bands that were detected in immunoprecipitated proteins with anti-Sfrp2 antibody, but not with control IgG following LPS stimulation (Fig. 7F). Likewise, downregulation of Sfrp2 in BV2 cells resulted in a specific TRAF6-conjugated TAK1 band in the proteins immunoprecipitated with anti-TRAF6 antibody, but not with control IgG following LPS stimulation (Fig. 7G). Taken together, these results indicated that Sfrp2 knockdown in BV2 cells promoted the interaction of TRAF6 and TAK1, thereby activating the downstream NF-κB signaling cascade and amplifying the neuroinflammatory response.

S100a8 Inhibitor Reduced Neuroinflammation Induced by Microglial Sfrp2 Deletion Following I/R Injury

Sfrp2 deficiency in microglia increased the expression of S100a8 and then exacerbated neuroinflammation following I/R injury. Given the pivotal role of S100a8 in this process mentioned above, we further investigated whether a S100a8 inhibitor could alleviate the resulting neuroinflammatory response. PAQ is an inhibitor of S100a8 and widely used in inflammatory diseases.^35,47 We evaluated the anti-inflammatory effects of PAQ by examining microglial activation at 3 days after I/R. Similarly, double immunostaining was performed for Iba1 in combination with CD68 (Figs. 8A, 8B) and CD16 (Figs. 8C, 8D). We found that PAQ treatment significantly mitigated the activation of activated and pro-inflammatory microglia in the I/R and I/R + Sfrp2^–/– groups. These data implicated that S100a8 inhibitor treatment effectively alleviated the exacerbated neuroinflammatory response caused by microglial Sfrp2 deletion during I/R injury.

Figure 8.

PAQ treatment reduced the activation of microglia cells induced by Sfrp2 deficiency in I/R injury (3 days after I/R). (A) Representative images showing co-labeling of Iba1 and CD68. Scale bar = 20 µm. (B) Quantification of Iba1⁺ and CD68⁺ cells in each group (n = 6). (C) Representative immunofluorescence images of retinal sections stained for Iba1 and CD16 in each group. Scale bar = 20 µm. (D) Quantification of Iba1⁺ and CD16⁺ cells in each group (n = 6). ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM.

S100a8 Inhibitor Rescued Retinal Thickness Reduction and Retinal Neurons Loss Caused by Microglial Sfrp2 Deletion in I/R

To evaluate the influence of S100a8 inhibitor on retinal structural defects and neurons loss induced by microglial Sfrp2 deficiency following I/R injury, H&E staining, retinal whole flat‐mount, and frozen section IF staining were performed at 7 days after I/R. H&E staining showed that PAQ treatment preserved retinal sublayer thickness (NFL/GCL, IPL, and INL) and total retinal thickness in I/R and I/R + Sfrp2^–/– mice (Figs. 9A–E). In addition, IF staining demonstrated that PAQ treatment also preserved inner retinal neurons, as evidenced by increased numbers of RGCs and amacrine cells, and a marked elongation of bipolar cells axonal fibers in the I/R + PAQ and I/R + Sfrp2^–/– + PAQ groups compared with the I/R and I/R + Sfrp2^–/– groups, respectively (Figs. 9F, 9G, 10A–E). However, horizontal cells showed no notable differences among different groups (see Figs. 10A, 10C). Collectively, these findings underscored the potent protective effects of the S100a8 inhibitor in mitigating retinal structural damage and preserving inner retinal neurons.

Figure 9.

PAQ treatment alleviated retinal structural damage and RGC loss induced by Sfrp2 deficiency following I/R injury (7 days after I/R). (A) Representative H&E images of retinas from the healthy control and I/R groups with or without Sfrp2 deficiency and PAQ treatment. Scale bar = 20 µm. (B–E) Quantification of retinal sublayers and total thickness (n = 6). (F) Representative images of flat‐mounted retinas of RBPMS and β3‐Tublin in each group. Scale bar = 50 µm. (G) Quantification of RBPMS and β3‐Tublin‐labeled RGCs in each group (n = 6). ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM.

Figure 10.

PAQ treatment attenuated inner retinal neurons damage induced by Sfrp2 deficiency following I/R injury (7 days after I/R). (A) Representative immunofluorescence images of calretinin, calbindin, and PKC-α from control and I/R groups with or without Sfrp2 deficiency and PAQ treatment. Scale bar = 20 µm. (B–E) Quantification of calretinin⁺ amacrine cells, calbindin⁺ horizontal cells, PKC-α⁺ bipolar cells, and axonal length of PKC-α⁺ bipolar cells (n = 6). **P < 0.01; ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM.

S100a8 Inhibitor Alleviated Visual Dysfunction Induced by Microglial Sfrp2 Deletion After I/R

Finally, we investigated the effect of S100a8 inhibitor on visual dysfunction induced by microglial Sfrp2 deficiency at 7 days after I/R. ERG recordings revealed that PAQ treatment robustly enhanced the amplitudes of the a-wave (Figs. 11B, 11C, 11G, 11I), b-wave (Figs. 11A–C, 11F, 11H, 11J), OPs (Figs. 11D, 11K), or PhNR (Figs. 11E, 11L) in the I/R + Sfrp2^–/– + PAQ and I/R + PAQ groups compared with their corresponding controls (I/R + Sfrp2^–/– and I/R groups, respectively). These data demonstrated that S100a8 inhibitor treatment could attenuate visual dysfunction induced by microglial Sfrp2 deficiency following I/R injury.

Figure 11.

PAQ treatment mitigated visual dysfunction induced by Sfrp2 deficiency following I/R injury (7 days after I/R). (A–E) Representative waveforms of scotopic ERG, OPs, and PhNR from each group under different flashlight intensities. (F–L) Quantification of the amplitudes of a‐wave, b‐wave, OPs, and PhNR (n = 6). ***P < 0.001; ns, no significance; 1-way ANOVA. Data are presented as mean ± SEM.

Discussion

Increasing evidence suggests that neuroinflammation plays a crucial role in exacerbating various neural injuries and degenerative diseases.^48–50 Extensive studies have highlighted the promising therapeutic potential of targeting neuroinflammation in I/R.^51,52 However, as some patients show limited responsiveness to existing treatments, elucidating the specific signaling pathways underlying neuroinflammation holds significant promise in exploring targeted therapeutic strategies for I/R. Our study, for the first time, reveals the pivotal role of Sfrp2 activation in TRAF6-TAK1-NF-κB signaling pathway in mitigating I/R injury. Particularly, the inhibition of the downstream inflammatory mediator S100a8 effectively alleviated retinal neuroinflammation induced by microglial Sfrp2 deficiency in I/R injury, indicating targeting Sfrp2 may be a promising avenue for novel neuroinflammation-targeted interventions in the management of I/R injury (Fig. 12).

Figure 12.

Schematic diagram. The deficiency of Sfrp2 in microglia enhanced the interaction of TRAF6 with TAK1, thereby activating the NF-κB pathway, upregulating the proinflammatory mediator S100a8, and ultimately exacerbating neuroinflammation following I/R injury. Treatment with PAQ, an inhibitor of S100a8, effectively modulated the inflammatory microenvironment and suppressed neuroinflammation following I/R injury.

Current therapeutic strategies for retinal degenerative diseases aim to slow disease progression and alleviate secondary complications. However, they largely fail to prevent progressive neuronal loss, and their clinical efficacy remains limited. Accumulating evidence indicates that neuroinflammation plays a central role in the pathogenesis of retinal degeneration, contributing to sustained neuronal injury and functional decline. Despite extensive investigation into inflammatory mechanisms in retinal disorders, effective pharmacological strategies specifically targeting retinal inflammation are still lacking. This underscores a critical need for more effective therapeutic interventions. The SFRP family serves as a critical regulator of Wnt signaling, an important pathway that orchestrates embryonic development, tissue regeneration, cell proliferation, and carcinogenesis. Sfrp2 is a member of SFRP family and can modulate inflammatory responses across a broad spectrum of diseases.^20,21 Notably, it is also reported that Sfrp2 is expressed in the mouse retina,⁵³ and participated in various retinal diseases, including uveitis, RP, and retinoblastoma.^22,23,25 Given the emerging involvement of Sfrp2 in multiple retinal disorders mentioned above, delineating its role in retinal neuroinflammatory diseases is both timely and of significant interest. In this study, we found that Sfrp2 expression became robustly induced in the retina in response to I/R injury. Additionally, we found that Sfrp2 was localized to microglia (see Fig 1). These results demonstrated that Sfrp2 was expressed in the retina, and its level was markedly upregulated following I/R injury, indicating that Sfrp2 might participate in microglia-mediated retinal inflammation. In addition, Sfrp2 has also been implicated in retinal neurodevelopment and axon formation.^27–29 In this regard, it is of particular interest to examine how Sfrp2 influences retinal neuronal integrity after I/R injury. Our study revealed that Sfrp2 plays a critical role in maintaining the integrity of retinal neurons. We provide the following evidence: (1) Sfrp2 conditional knockout in microglia severely disrupted retinal structure in I/R, resulting in a reduction of overall retinal thickness as well as the thickness of retinal sublayers, including the NFL/GCL, IPL, and INL. (2) Sfrp2 deficiency in microglia not only markedly deteriorated RGC survival and increased their apoptotic loss, but also substantially reduced the number or axonal length of other retinal neuronal populations, including amacrine, horizontal, and bipolar cells in I/R; and (3) moreover, microglial Sfrp2 deficiency markedly compromised visual function in I/R. Our study suggested that Sfrp2 exerted a protective effect on the retina, whereas its deficiency exacerbated retinal inflammation, thereby leading to the observed structural and functional retinal impairments following I/R injury.

Retinal injury induced by I/R is a common feature of retinal neurodegenerative diseases. Similarly, LPS, also known as endotoxins, can trigger a host inflammatory response that drives retinal damage, ultimately leading to retinal neurodegeneration.⁵⁴ Some important inflammatory events to consider in the context of retinal injury are the release of potentially toxic mediators by activated inflammatory cells and glial elements. Recent studies have revealed that microglia act as the principal resident immune cells and play a central role in mediating inflammatory responses during retinal injury.^55–57 As we know, the inflammation triggered by either I/R injury or LPS exposure is tightly linked to the activation of the TRAF6-TAK1 and downstream NF-κB signaling pathway.^44,58,59 In this process, TRAF6 acts as a key ubiquitin ligase required for NF-κB activation. Working in concert with the UBC13-UEV1A ubiquitin-conjugating enzyme complex, TRAF6 mediates Lys 63-linked polyubiquitination, which triggers activation of the TAK1 kinase complex. Activated TAK1 then phosphorylates IκB kinase, ultimately resulting in NF-κB activation. Parallel observations were obtained in this study, the RNA-seq and WB data revealed that TRAF6-TAK1-NF-κB signaling pathway was activated in microglial Sfrp2 deficiency mouse after I/R injury in vivo and in Sfrp2-siRNA treated BV2 after LPS stimulation in vitro. It is also reported that the inhibition of TRAF6-TAK1 signaling pathway could alleviate inflammation.^17,60 Similarly, we found that the suppression of S100a8, the downstream inflammatory mediators of the TRAF6-TAK1-NF-κB pathway, effectively attenuated microglial activation and mitigated the structural and functional retinal damage induced by microglial Sfrp2 deficiency following I/R injury. Given that Sfrp2 has been shown to negatively regulate the TRAF6-TAK1-NF-κB signaling pathway, we sought to further investigate how Sfrp2 modulates this pathway. We found that Sfrp2 deficiency in microglia augmented the TRAF6-TAK1 complex formation after LPS stimulation, suggesting that, at basal condition, Sfrp2 inhibited the interaction between TRAF6 and TAK1, thereby suppressing the inflammatory signaling cascade. Overall, our convergent in vitro and in vivo findings demonstrated that the TRAF6-TAK1-NF-κB signaling pathway played a critical role in microglia-mediated neuroinflammation. Sfrp2 negatively modulated the TRAF6-TAK1-NF-κB signaling pathway may suggest that it could emerge as a therapeutic target in I/R injury.

The S100 proteins comprise a family of proteins that exhibit a high degree of structural similarity, but are not functionally interchangeable. The structure and function of the S100 proteins are regulated by Ca²⁺ binding, allowing them to translate intracellular calcium fluctuations into precise cellular responses.^61,62 This protein family modulates diverse cellular processes such as proliferation, migration and/or invasion, inflammation, and differentiation.^63–65 As a member of the S100 protein family, S100a8 is involved in mediating inflammatory responses in various diseases, including rhinosinusitis,⁶⁶ myocardial infarction,⁶⁷ acute kidney injury⁶⁸ and so on. A previous study has also reported that silencing S100a8 in microglial cells could attenuate retinal damage induced by I/R injury.³² In this study, our results showed that Sfrp2 deficiency led to the activation of TRAF6-TAK1-NF-κB signaling, thus, upregulated the expression of inflammatory mediator including IL-1β, S100a8, Ccl2, Ccl7, and so on. KEGG enrichment analysis showed that Sfrp2 deficiency resulted in significant alterations of S100a8 within the IL-17 signaling pathway. Consistently, GO analysis revealed that S100a8 exhibited one of the most pronounced changes among all upregulated DEGs in inflammation-related biological processes. Given the established role of S100a8 in inflammatory responses reported in previous studies, S100a8 was therefore considered a promising candidate for further investigation. PAQ, a selective inhibitor of S100a8, has seen application in diverse animal disease models, including traumatic brain injury,³⁵ Sjögren's dry eye disease,⁶⁹ and Parkinson's disease.⁷⁰ PAQ exerted neuroprotective effects by ameliorating neurovascular injury in traumatic brain injury.³⁵ It attenuated the progression of Parkinson’s disease by suppressing astrocyte-mediated neuroinflammation.⁷⁰ In the aforementioned diseases, PAQ has been shown to exert anti-inflammatory and neuroprotective effects by suppressing S100a8. Therefore, it was examined for treating neuroinflammation in this study. Here, we found that PAQ could significantly suppress S100a8-mediated inflammatory responses, the loss of retinal neurons, and visual dysfunction, which shows congruence with the previous study mentioned above, indicating the inhibition of S100a8 could alleviate Sfrp2 deficiency induced neuroinflammation after I/R injury.

There are also some limitations in this study. First, the limited sample size of human retinal specimens in this study may restrict a comprehensive evaluation of the expression and potential role of Sfrp2 in human tissues. Further studies incorporating a larger number of human retinal samples may facilitate more robust comparison and validation of the results observed in the mouse model. Second, although the animal model offered important insights into the underlying pathophysiology, translating these findings into clinical practice remains challenging, particularly with respect to the interactions between Sfrp2-targeted therapies and existing treatments, as well as their safety, stability, and long-term outcomes in humans. Moreover, our findings indicated that Sfrp2 deletion was associated with aggravated inflammation following I/R injury. Future investigations incorporating recombinant SFRP2-based gain-of-function strategies will be pursued to further confirm and extend these results.

In summary, our findings demonstrated that Sfrp2 in microglia inhibited S100a8-mediated neuroinflammation and alleviated retinal damage following I/R injury. Collectively, these findings pivotally position Sfrp2 as a novel and promising target for neuroinflammatory retinal disorders.