Microglial CR3 promotes neuron ferroptosis via NOX2-mediated iron deposition in rotenone-induced experimental models of Parkinson's disease

Abstract

The activation of complement receptor 3 (CR3) in microglia contributes to neurodegeneration in neurological disorders, including Parkinson's disease (PD). However, it remains unclear for mechanistic knowledge on how CR3 mediates neuronal damage. In this study, the expression of CR3 and its ligands iC3b and ICAM-1 was found to be up-regulated in the midbrain of rotenone PD mice, which was associated with elevation of iron content and disruption of balance of iron metabolism proteins. Interestingly, genetic deletion of CR3 blunted iron accumulation and recovered the expression of iron metabolism markers in response to rotenone. Furthermore, reduced lipid peroxidation, ferroptosis of dopaminergic neurons and neuroinflammation were detected in rotenone-lesioned CR3^−/− mice compared with WT mice. The regulatory effect of CR3 on ferroptotic death of dopaminergic neurons was also mirrored in vitro. Mechanistic study revealed that iron accumulation in neuron but not the physiological contact between microglia and neurons was essential for microglial CR3-regulated neuronal ferroptosis. In a cell-culture system, microglial CR3 silence significantly dampened iron deposition in neuron in response to rotenone, which was accompanied by mitigated lipid peroxidation and neurodegeneration. Furthermore, ROS released from activated microglia via NOX2 was identified to couple microglial CR3-mediated iron accumulation and subsequent neuronal ferroptosis. Finally, supplementation with exogenous iron was found to recover the sensitivity of CR3^−/− mice to rotenone-induced neuronal ferroptosis. Altogether, our findings suggested that microglial CR3 regulates neuron ferroptosis through NOX2 -mediated iron accumulation in experimental Parkinsonism, providing novel points of the immunopathogenesis of neurological disorders.

Highlights

• •The expression of CR3 is elevated in rotenone PD mice.
• •CR3 ablation attenuates rotenone-induced iron deposition in mice.
• •CR3 ablation ameliorates rotenone-induced neuron ferroptosis.
• •Microglial NOX2-derived superoxide contributes to CR3-regulated neuron ferroptosis.
• •Exogenous iron recovers the sensitivity of CR3^−/− mice to rotenone-induced ferroptosis.

1. Introduction

Complement receptor 3 (CR3) is a principal component of the complement system that displays both beneficial and detrimental effects in the nervous system depending on different pathological conditions [1,2]. Microglia are innate immune cells in the brain and express high levels of CR3 [3]. The expression of CR3 was elevated in the postmortem brains of patients with Alzheimer's disease (AD) and Parkinson's disease (PD) as well as related animal models [[4], [5], [6]]. Recent evidence suggests that abnormal expression and activation of CR3 in microglia contributes to the pathological alterations of neurological disorders [[7], [8], [9]]. Czirr et al. found that CR3 was involved in regulating phagocytosis and clearance of β-amyloid (Aβ), a main component of plaque senile in AD, by microglia [10]. CR3 can also interact with aggregated α-synuclein to elicit a proinflammatory response and production of free radicals in microglia [8,11]. We found that CR3 deletion significantly blocked the activation of microglia and related neurodegeneration in mouse PD models [[12], [13], [14]]. However, the molecular mechanisms by which microglial CR3 regulates neuronal damage remain elusive currently.

Ferroptosis is a newly recognized form of cell death caused by excessive iron deposition and subsequent lipid peroxidation [15]. The disruption of iron homeostasis is an essential inducer of neuronal ferroptosis in neuropathological conditions. The aberrant iron accumulation, abnormal levels of iron regulatory molecules and related ferroptosis of neurons in the brain are closely linked with the initiation and progression of multiple neurological disorders, including PD [16,17]. Iron chelators displayed potent protective efficacy on dopaminergic neurons and ameliorated neurobehavioral deficits in PD mouse models [18,19]. The regulatory mechanism of neuronal ferroptosis has been a hot topic in the past few years. However, most research has focused on iron-related metabolism proteins and redox signaling in neurons. Recent evidence suggests that non-cell-autonomous mechanisms could exist to regulate iron homeostasis in neuron [20,21]. Microglial activation is a crucial event and plays important roles in the pathogenesis in neurodegenerative processes [[22], [23], [24]]. Microglia-mediated neuroinflammation can induce iron imbalance in neurons, although the regulatory efficacy of neuroinflammation on neuronal ferroptosis remains unknown. Huo et al. reported that intra-globuspallidus injection of lipopolysaccharide (LPS) stimulated activation of microglia and iron accumulation in the brains of rats [25]. Furthermore, the pro-inflammatory factors interleukin-1β (IL-1β), IL-6 and tumor necrosis factor α (TNFα) produced by activated microglia were found to promote iron influx and simultaneously reduce iron efflux in primary midbrain neurons through upregulation of divalent metal transporter 1 (DMT1) and downregulation of ferroportin 1 (FPN1) [26,27]. In midbrain neuron-glia cultures, genetic deletion of CR3 reduced the contents of TNFα and IL-1β in response to high-mobility group box 1 (HMGB1) [28]. Reduced TNFα, compared to that in the wild type (WT) group, was also detected in LPS-treated primary CR3^−/− cultures [29], suggesting that CR3 could regulate production of proinflammatory factors in microglia. However, whether microglial CR3 could regulate neuronal ferroptosis via microglial activation and underlying mechanisms have not been reported.

In this study, we explored the role of microglial CR3 in ferroptosis of neurons by using mouse and cell culture PD models generated by rotenone as well as CR3^−/− mice and siRNA-mediated knockdown of CR3 cell culture systems. We hypothesized that CR3 might contribute to ferroptosis of dopaminergic neurons via microglia-mediated neuroinflammation and subsequent disruption of iron homeostasis in neurons.

2. Materials and methods

2.1. Reagents

Rotenone (R9975) and anti-phospho-p47phox (pSer345) antibody (SAB4504721) were purchased from Sigma-Aldrich, Inc (St. Louis, MO, USA). The antibodies against CR3, C3, tyrosine hydroxylase (TH), ICAM-1, FPN1, transferrin receptor (TFR), glutathione peroxidase 4 (GPX4), xCT, Cyclooxygenase-2 (COX-2), and GAPDH and iron assay kit were provided by Abcam (Cambridge, MA, USA). Antibody against ionized calcium binding adaptor molecule-1 (Iba-1) was provided by Wako Chemicals (Richmond, VA, USA). The BODIPY 581/591C11 and Ferrorange (F374) were provided by Thermo Fisher Scientific (Waltham, MA, USA) and Dojindo Laboratories (Kumamoto, Japan), respectively. The malondialdehyde (MDA) and glutathione (GSH) assay kits were provided by Beyotime Biotechnology (Shanghai, China).

2.2. Animal treatment

Adult male WT (C57BL/6) and CR3^−/− mice were randomly assigned to the control and rotenone groups (n = 20–25 for each group). For mice in the rotenone group, 1.5 mg/kg/day rotenone was administered (i.p.) for 21 days to generate the PD model [24,30]. Control mice received equal amounts of vehicle. To investigate the role of iron in CR3-mediated neurodegeneration, another batch of CR3^−/− mice was used and randomly separated into control, rotenone and rotenone + iron dextran groups (n = 10 for each group). The treatment regimen of rotenone was the same as above. Mice in the rotenone + iron dextran group received (i.p.) iron dextran (100 mg/kg/day) after 14 days of initial rotenone injection. The dosage of iron dextran was chosen based on previous report [31].

2.3. Ethics statement

All experiments involving animals were conducted according to the ethical policies and procedures approved by the Institutional Animal Care and Use Committee of Dalian Medical University, China (Approval no. AEE21016).

2.4. Gait assay

The gait of mice was measured based on our previous report [32]. Briefly, mice were permitted to walk in a closed box (50 cm long and 10 cm wide) with white paper on the bottom. All mice were trained three times before the formal experiment. In formal experiment, the paws of mice were painted with ink and then were allowed to walk through the darkness box. The strides of mice were recorded and the parameters, including the stride length and stride distance of the forelimbs and hindlimbs were analyzed.

2.5. Iron assay

The assay of iron content was performed as we previously reported [33,34]. Briefly, the midbrain tissues of mice with PBS perfusion were homogenized in 4–10 vol of iron assay buffer on ice and then were centrifuged at 16,000×g for 10 min at 4 °C. After quantification of the protein concentration of the supernatant, 100 μl samples (equal amount of total protein) or standard were added to a 96-well plate, followed by 5 μl of iron assay buffer. After mixing well using a horizontal shaker, the plate was incubated at 25 °C for 30 min (avoid light). After that, 100 μl iron probe was added to each well and were incubated at 25 °C for additional 60 min (avoid light). Then, the absorbance at 593 nm was measured. The levels of iron were calculated according to the standard curve.

2.6. MDA, GSH and GSSG assay

The midbrain tissues of mice with PBS perfusion were homogenized in ice-cold lysis buffer containing 20 mM Tris (pH 7.5), 150 mM NaCl, 1 % Triton X-100 and protease inhibitor mixture [24]. The homogenates were centrifuged for 10 min at 10,000×g at 4 °C. The levels of MDA, GSH and GSSG were measured in the supernatant using assay kits [24]. The principle for MDA determination was based on the reaction of MDA and thiobarbituric acid (TBA) to produce a red product that has an absorption maximum at 535 nm. Similarly, GSH can react with the chromogenic substrate DTNB to produce yellow TNB that displays an absorption maximum at 412 nm. The absorbance at 535 and 412 nm was measured for MDA and GSH assays, respectively. For the GSSG assay, GSH was initially removed from the sample and then GSSG was reduced to GSH via glutathione reductase. The absorption maximum at 412 nm was recorded based on the interaction of GSG and DTNB.

2.7. RNA sequencing and bioinformatics analysis

Mice in each group were transcardially perfused with PBS and then the mid-brain tissues were dissected. Total mRNA was subsequently extracted to construct the cDNA library and the library was then sequenced using pair-end 150bp reads on the Illumina HiSeq 4000 platform. The bioinformatics analysis was carried out following the pipeline outlined in our previous publication [35]. Differing somewhat from our previous approach, the differential expression genes (DEGs) between CR3KO Rot and WT Rot group were annotated in the KEGG database using the R package ClusterProfiler [36]. To identify genes related to Ferroptosis, we used FerrDb [37], one database describing the Ferroptosis regulators and Ferroptosis-disease associations, to screen DEGs (CR3KO Rot vs. WT Rot) related to Ferroptosis and categorize them according to the classification provided by FerrDb.

2.8. Midbrain neuron-glia culture

The primary cultures were done using the midbrain tissues of embryonic E15 rats [12,38]. All the embryonic midbrain samples were dissociated into a single suspension in ice-cold MEM medium. After that, the samples were centrifuged at 1200 rpm for 10 min and the pellet was re-suspended in MEM containing heat-inactivated FBS (10 %), heat-inactivated horse serum (10 %), glucose (1 g/L), l-glutamine (2 mM), sodium pyruvate (1 mM), nonessential amino acids (100 μM), penicillin (50 U/ml), and streptomycin (50 μg/ml). The cells were seeded into poly-d-lysine-coated 24-well plates at 5.5 × 10⁵ per well and then were maintained at 37 °C supplemented with 5 % CO₂ for 7 days.

2.9. BV2 and SH-SY5Y cells

The BV2 microglia and SH-SY5Y neuroblastoma cells were regularly maintained in DMEM containing FBS (10 %), penicillin–streptomycin (1 %), non-essential amino acids (1 %), sodium-pyruvate (1 %) and l-glutamine (1 %) at 37 °C supplemented with 5 % CO₂. Cells were used until passage ten.

2.10. Transfection of siRNA

SiRNAs transfection was performed as described in previous studies [8,39]. In brief, BV2 microglia were seeded on 6-well plates and were transfected with CR3-specific siRNA using Lipo3000 for 48h based on manufacturer's protocol (GenePharma, Shanghai, China).

2.11. MTT assay

This test was performed in SY-SY5Y cells in a 96-well plate [12,40]. In brief, cell cultures were treated with MTT solution for 4 h. Then, we removed the supernatant, and DMSO was added to the cells for an additional 10 min. To measure the survival of cells, the absorbance of the 96-well plate was read by a plate reader at 540 nm.

2.12. Superoxide assay

Superoxide production was assessed by measuring the SOD-inhibitable reduction of the tetrazolium salt WST-1 [8,12,41]. Briefly, BV2 microglial cells were washed with HBSS (no phenol red) and then were treated with rotenone with or without CR3 blocking antibody. Subsequently, 50 μl SOD (50 U/ml) or HBSS was added to each well along with 50 μl WST-1 (1 mM). A microplate spectrophotometer was used to measure the absorbance at 450 nm and the difference of absorbance between the wells with or without SOD represented the amount of superoxide.

2.13. C11-BODIPY staining

The C11-BODIPY staining was performed based on previous report [42,43]. In brief, the cells were incubated with C11-BODIPY581/591 reagent at a concentration of 10 μM after washing 3 times with serum-free medium. One hour after incubation at 37 °C in the dark, the cells were washed with serum-free culture medium. Then, the immunofluorescence images of cells were taken using a fluorescence microscope at 488 nm (for oxidized lipid lipid) and 561 nm (for non-oxidized lipid) excitation length. The fluorescence intensity was measured by ImageJ to calculate the rate of oxidation of C11-BODIPY581/591.

2.14. Ferroorange staining

The ferroorange staining was performed based on a previous report [44]. In brief, the cells were washed with serum-free medium and then 1 μM ferroOrange solution was added to the culture. One hour after incubation at 37 °C, the cells were washed with serum-free cell culture medium again and then the immunofluorescence pictures were taken using a fluorescence microscope.

2.15. Co-IP

Co-IP was conducted based on previous reports [45,46]. Briefly, the cell samples were homogenized in RIPA lysis buffer, followed by centrifugation at 12000g for 10 min. The supernatant fraction (100 μg of protein) was pre-cleared for 1 h with 10 μg of A/G agarose. After removing the A/G agarose, the pre-cleared lysate was incubated with anti-gp91^phox, anti-p47^phox antibody (10 μg) or control IgG overnight. Then the samples were incubated with 50 μg A/G agarose beads for an additional 2 h at RT. After washing, the beads were eluted and the samples were analyzed by immunoblotting.

2.16. Western blot

The brain and cell samples were homogenized in RIPA lysis buffer that was supplemented with protease inhibitors on ice. After removing the insoluble fractions through centrifugation, the supernatant was collected for analysis. For Western blot, equal amounts of proteins from supernatant samples were separated by 4–12 % Bis-Trispolyacrylamide electrophoresis and then were transferred to PVDF membranes. The membranes were probed with primary antibodies overnight at 4 °C after 1h of incubation with 5 % nonfat milk. On the second day, the membranes were washed with PBST and then were probed with a suitable secondary antibody for 1 h. The infrared band signals were detected using ECL reagents.

2.17. QRT-PCR

The RNA was purified from brain or cell samples using TRIzol reagent and the concentrations were quantified using a Nanodrop 2000, UV–vis spectrophotometer. Two micrograms RNA for each sample was reverse transcribed with MuLV reverse transcriptase and oligo dT primers [24,30,47]. SYBR Green Premix was used for real-time PCR amplification. The conditions was set up as 95 °C for 10 s, 55 °C for 30 s, and 72 °C for 30 s for up to 40 cycles. The mRNA transcripts of all genes were normalized to GAPDH using the 2^−ΔΔCt method.

2.18. Immunohistochemistry and immunocytochemistry

For immunohistochemistry, the brain samples were dissected from mice that were perfused with 4 % PFA transcardially. The brains were cryoprotected in 30 % sucrose before serial 35 μm coronal sections were collected with a cryostat. For immunocytochemistry, cell cultures were washed, followed by fixation with 4 % PFA for 10mins. The endogenous peroxidase in sections and cells were quenched by 3 % H₂O₂ in 0.01 M PBS for 15 min. Subsequently, the brain sections and cells were blocked with 4 % goat serum (diluted in 0.25 % Triton/PBS) for 2 h, followed by incubation with primary antibody against TH or Iba-1 in PBS containing Triton X-100 (0.1 %) at 4 °C overnight. Then, the sections or cells were washed again before probing with an appropriate secondary antibody. Immunostaining was visualized using DAB. The counts of THir neurons were performed by two individuals that were blinded to the treatment.

2.19. Double immunofluorescence staining

This experiment was performed as reported previously [8,48]. Briefly, the sample sections were initially blocked with 4 % goat serum for 2 h and then were probed with antibody against CD68 (1:500) at 4 °C overnight. After that, the sections were washed with PBS and then were probed with antibody against Iba-1 (1:1000) at 4 °C. After washing three times with PBS, the sections were probed with appropriate Alexa-488 (green) or Alexa-594 (red) labeled secondary antibodies for 2h at RT. The double immunofluorescence staining was observed using a confocal microscope.

2.20. Statistical analysis

The statistical analyses were done using SPSS22.0. Data were presented as the mean ± SEM. The student t-test was used to compare means between two groups. The Shapiro-Wilk and Bartlett's tests were used to differentiate normal distribution and equal variance of data, respectively. Comparisons of more than two groups or more than two parameters were performed using one-way or two-way ANOVA, followed by Tukey's post hoc test for normally distributed data; otherwise, a nonparametric test was employed. A p value < 0.05 was considered statistically significant.

3. Results

3.1. The expression of CR3 is increased in rotenone PD mice

To determine whether ferroptosis is involved in CR3-mediated neuronal damage, alterations in the CR3 protein were initially evaluated in rotenone PD mice. CR3 was up-regulated in the midbrain of rotenone-treated mice compared with vehicle controls (Fig. 1A and B). ICAM-1 and iC3b are two ligands of CR3 that induce its activation [49]. Significant increases in both ICAM-1 and iC3b expression were also observed in rotenone-treated mice in comparison to control mice (Fig. 1A–C and D). The expression of ICAM-1 and iC3b were only mildly increased in CR3^−/− mice after rotenone treatment and the magnitude of increase of both proteins is much smaller than that in rotenone-treated WT mice (Supplementary Fig. 1).

Fig. 1.

CR3 deficiency reverses rotenone-induced alterations of gene expression in mice. (A) Representative blots of CR3, ICAM-1 and iC3b in the midbrain region of mice. (B–D) Quantification of CR3, ICAM-1 and iC3b blots. (E) The comparison of the number and Volcano plot of DEGs between CR3^−/−-Rot & WT-Rot groups. (F) The RNA expression profiles of DEGs between CR3^−/−-Rot & WT-Rot groups. (G) The KEGG pathways enriched by DEGs (CR3^−/−-Rot & WT-Rot groups). (H) The fold change and classification of DEGs (CR3^−/−-Rot & WT-Rot groups) related to Ferroptosis using FerrDb. n = 4. ∗p < 0.05, ∗∗p < 0.01.

Then, the transcriptome changes were compared between WT and CR3^−/− mice after rotenone exposure and an analysis of DEGs among groups was conducted. There were 737 DEGs between CR3^−/−-Rot & WT-Rot groups, in which 436 DEGs were up-regulated and 301 DEGs were down-regulated (Fig. 1E). The expression profiles of DEGs between WT-Rot and CR3^−/−-Rot groups were displayed in Fig. 1F. Then, by analyzing the KEGG pathways enriched by DEGs (CR3^−/−-Rot & WT-Rot), we have identified the Ferroptosis pathway as one of the enriched pathways (Fig. 1G). DEGs related to Ferroptosis were screened and categorized using the FerrDb database (Fig. 1H). CR3 deficiency reduced gene expression of CHAC1, HSPB1 and PTGS2, three markers of ferroptosis, and ferroptotic drivers, such as Alox12B, PPKCA, SAT1 and TRF as well as elevated JUN, one suppressor of ferroptosis, in response to rotenone (Fig. 1H). In contrast, the ferroptosis was not identified as a enriched pathway by analyzing the KEGG pathways enriched by DEGs between CR3^−/−-Con and WT-Con (Supplementary Fig. 2 and supplementary Table). These results suggested that CR3 might play a role in ferroptosis in rotenone-treated mice.

3.2. Genetic ablation of CR3 reduces iron content and attenuates alterations in iron metabolism proteins in rotenone PD mice

Iron is a key factor in initiating ferroptosis. We therefore investigated the effects of CR3 on iron deposition in mice by using CR3^−/− mice. A significant increase in iron content was detected in the midbrain of WT mice lesioned by rotenone, but this increase was not detected in rotenone-treated CR3^−/− mice (Fig. 2A). In agreement with iron elevation, the expression of ferritin was reduced in rotenone-treated WT mice, which was blunted in CR3^−/− mice (Fig. 2B and C). CR3 deficiency also reversed rotenone-induced decrease of gene expression of FTH-1, the main subunit of ferritin, in mice (Fig. 2D). Double-immunofluorescence staining using antibodies against TH and ferritin revealed that rotenone reduced ferritin expression in THir neurons, which was significantly reversed by CR3 deficiency (Fig. 2E and F), suggesting that CR3 reduced iron contents in dopaminergic neurons in rotenone PD mice.

Fig. 2.

Rotenone increases iron content and alters iron metabolism markers in WT, but not in CR3^−/− mice. (A) Iron content in the midbrain of WT and CR3^−/− mice. (B) Representative blots of ferritin, Fpn-1, TFR, DMT1, IRP1 and IRP2 in WT and CR3^−/− mice. (C) Quantification of ferritin blots. (D) The mRNA levels of FTH-1 in the midbrain of WT and CR3^−/− mice. (E) The representative images of double fluorescence staining using anti-ferritin and TH antibodies. (F) The quantification of TH⁺Ferritin⁺/TH⁺ cells. (G) Quantification of Fpn-1, TFR, DMT1, IRP1 and IRP2 blots in the midbrain of WT and CR3^−/− mice. (H) The gene expression of DMT1, IRP1 and IRP2 in the midbrain of WT and CR3^−/− mice. n = 4–6. ∗p < 0.05, ∗∗p < 0.01.

Consistently, the alterations in proteins involved in iron influx, FPN-1, and efflux, TfR, in rotenone-treated WT mice were also significantly reversed in CR3^−/− mice (Fig. 2B and G). In addition to FPN-1 and TFR, DMT1, iron regulatory protein 1 (IRP1) and IRP2 are also involved in regulating iron homeostasis [50]. Rotenone exposure resulted in significant increase of DMT1, IRP1 and IRP2 expression in WT mice, which was significantly mitigated in CR3^−/− mice (Fig. 2B and G). In agreement with protein expression, rotenone-induced alterations of mRNA levels of iron-metabolism proteins, including DMT1, IRP1 and IRP2, were suppressed by CR3 deficiency (Fig. 2H). These results suggested that CR3 plays a key role in rotenone-induced iron deposits.

3.3. Genetic ablation of CR3 blunts lipid peroxidation and ferroptosis in midbrain in rotenone PD mice

Iron accumulation induces ROS production through the Fenton reaction to elicit lipid peroxidation, another event required for the occurrence of ferroptosis [51]. Then, we evaluated whether CR3-mediated iron deposition triggered lipid peroxidation or not. The content of the lipid peroxidation product MDA was firstly determined. Increased levels of lipid peroxidation product MDA and 4-HNE, an oxidative stress-induced lipid peroxidation marker [52], were observed in rotenone-treated WT mice in comparison to WT vehicle controls (Fig. 3A and B). In contrast, rotenone-induced elevation of MDA and 4-HNE was not detected in CR3−/− mice (Fig. 3A and B), suggesting that CR3 deficiency mitigates rotenone-induced lipid peroxidation. Furthermore, recovered levels of GSH and the ratio of GSH/GSSG and diminished GSSG contents were observed in rotenone-treated CR3^−/− mice compared with rotenone-challenged WT mice (Fig. 3C–E)Here, we measured increased protein and mRNA levels of xCT, a subunit of the xc-system responsible for maintaining intracellular GSH content, in rotenone-treated WT mice (Fig. 3F–H), which might be a complementary response in response to reduced GSH. Genetic ablation of CR3 blunted rotenone-induced increase in xCT in mice (Fig. 3F–H).

Fig. 3.

Genetic ablation of CR3 blunts lipid peroxidation and ferroptosis in rotenone PD mice. (A–D) The concentrations of MDA, 4-HNE, GSH and GSSG in the midbrain of WT and CR3^−/− mice. n = 3–5. (E) The ratio of GSH/GSSG. (F) Representative blots of xCT, GPX4, COX-2 and ACSL4 in the midbrain of WT and CR3^−/− mice. (G) Quantification of xCT, GPX4, COX-2 and ACSL4 blots. (H) The gene expression of SLC7A11, GPX4, PTGS2 and ACSL4 in the midbrain of WT and CR3^−/− mice. n = 4. (I) Representative images of double fluorescence staining using anti-GPX4 and TH antibodies. (J) The quantification of GPX4⁺TH⁺/TH⁺ cells. n = 3. (K, L) Quantification of nigral THir neuron number and optic density of striatal TH staining in mice. n = 6–8. Scale bar = 200 μm; ∗p < 0.05, ∗∗p < 0.01.

Recent studies have shown that inhibition or deletion of GPX4 directly leads to ferroptosis due to lipid peroxide accumulation [53]. Rotenone injection resulted in a decrease in GPX4 expression in WT mice but not in CR3^−/− mice (Fig. 3F and G). In addition to GPX4, COX-2 and acyl-CoA synthetase long-chain family member 4 (ACSL4) are elevated in ferroptosis and have also been considered as indicators of ferroptosis [54]. Elevation of COX2 and ACSL4 expression was detected in rotenone-challenged WT mice (Fig. 3F and G). Genetic ablation of CR3 markedly blunted rotenone-induced increase in COX2 in mice (Fig. 3F and G). In agreement with protein levels, the alterations in the gene expression of GPX4, PTGS2 and ACSL4 in rotenone-treated mice were also blunted by CR3 deficiency (Fig. 3H).

Double-fluorescence staining using antibodies against TH and GPX4, a marker of ferroptosis, revealed that rotenone reduced GPX4 expression in THir neurons, which was significantly reversed by CR3 deficiency ( Fig. 3I and J), suggesting that CR3 reduced ferroptosis in dopaminergic neurons in rotenone PD mice. As ferroptosis leads to neuronal degeneration, the effects of genetic ablation of CR3 on dopaminergic neurodegeneration elicited by rotenone were further evaluated. Consistently, rotenone-induced loss of dopaminergic neurons in the SNpc and reduction of axon fibers were also markedly mitigated in CR3^−/− mice (Fig. 3K and L). These results revealed that CR3 mediated iron accumulation and lipid peroxidation, and ultimately, ferroptosis in rotenone-treated mice.

3.4. Genetic ablation of CR3 improves brain inflammatory markers in rotenone PD mice

Considering the tight regulation between iron deposits and neuroinflammation, microglial and astroglial activation were detected by immunohistochemistry using antibodies against Iba-1/CD68 and GFAP, respectively (Fig. 4A and B). WT mice injected with rotenone displayed larger cell bodies and higher intensity of Iba-1 and CD68 staining than WT vehicle mice (Fig. 4A and B). Quantitative analysis revealed high density of Iba-1 immunostaining and number of CD68⁺Iba-1⁺ microglia in WT mice treated with rotenone in comparison to vehicle mice, indicating activation of microglia (Fig. 4A–D). In contrast, microglia in CR3^−/− mice challenged with rotenone showed morphology similar to that of vehicle-treated CR3^−/− mice (Fig. 4A–D). No obvious difference in Iba-1/CD68 staining density and CD68⁺Iba-1⁺ microglia number in CR3^−/− mice treated with vehicle and rotenone was observed (Fig. 4A–D). Consistently, genetic ablation of CR3 also suppressed rotenone-induced astroglial activation as shown by the reduced GFAP⁺ cell number in rotenone-treated CR3^−/− mice in comparison to rotenone-treated WT mice (Fig. 4A and E).

Fig. 4.

CR3 deficiency improves rotenone-induced brain inflammatory markers in mice. (A) Representative graphs of Iba-1 and GFAP staining in WT and CR3^−/− mice. (B) Representative images of double fluorescence staining using anti-Iba-1 and CD68 antibodies. (C) Quantification of Iba-1 staining density in mice. (D) Quantification of CD68⁺Iba-1⁺/Iba-1⁺ cells. (E) Quantification of GFAP⁺ cell number in mice among groups. n = 4–6. (F–H) The mRNA levels of iNOS (E), TNFα (F) and IL-1β (G) in WT and CR3^−/− mice. n = 3. Scale bar = 50 μm; ∗p < 0.05, ∗∗p < 0.01.

Furthermore, in comparison with vehicle group, increased expression of iNOS, TNFα and IL-1β was detected in WT mice treated with rotenone, which was also significantly prevented in rotenone-treated CR3^−/− mice (Fig. 4F–H). These results suggested that CR3-mediated ferroptosis was accompanied by increased neuroinflammation in rotenone-challenged mice.

3.5. Microglial CR3 regulates neuron ferroptosis through iron accumulation but not microglia-neuron physiological contacts

Next, we investigated the mechanisms by which CR3 regulates neuronal ferroptosis by using an in vitro system. We initially determined the impact of blocking ferroptosis by liproxstatin-1 (Lipo-1) and ferrostatin-1 (Fer-1) on dopaminergic neuron survival in rotenone-treated cell culture (Fig. 5A). In agreement with a previous report [38], rotenone reduced the survival of dopaminergic neurons by reducing the TH⁺ cell number in primary midbrain neuron-glial cultures compared with the vehicle control (Fig. 5A and B). Treatment with ferroptosis inhibitor, liproxstatin-1 (Lipo-1) and ferrostatin-1 (Fer-1) partially restored the survival of dopaminergic neurons in cultures exposed to rotenone (Fig. 5A and B), suggesting that ferroptosis represents a part of rotenone-induced cell death.

Fig. 5.

Microglia-neuron contact is not required for CR3-regulated neuronal ferroptosis. (A) Representative images of TH immunostaining in midbrain neuron-glial cultures treated with rotenone with or without Lipo-1 or Fer-1 pre-treatment. (B) Quantification of THir neuron number in different groups. (C) Quantification of THir neuron number in midbrain neuron-glial cultures treated with rotenone with or without CR3 blocking antibody and FeCl₂ pre-treatment. (D) BV2 microglia were transfected with CR3-siRNA for 24h, followed by rotenone treatment for additional 24h. The collected conditioned medium was used to treat SH-SY5Y cells for 24h and cell viability was evaluated. (E) BODIPY 581/591C¹¹ and Ferrorange staining in SH-SY5Y cells treated with conditioned medium prepared from rotenone-treated Ctrl and CR3-siRNA-transfected BV2 microglia. (F) The quantification of rate of oxidation of BODIPY 581/591C¹¹ among groups. (G) Representative blots and quantification of GPX4 in SH-SY5Y cells treated with conditioned medium derived from rotenone-treated Ctrl and CR3-siRNA-transfected BV2 microglia. n = 3–5. Scale bar = 100 μm; ∗p < 0.05, ∗∗p < 0.01.

Furthermore, the impact of CR3 on rotenone-induced ferroptosis was explored. Primary cultures were pre-treated with a CR3-blocking antibody (anti-CR3 Ab) to block CR3. As shown in Fig. 5C, rotenone-induced dopaminergic neurodegeneration was significantly blocked by anti-CR3 Ab antibody. Notably, the restoration of neuron survival induced by anti-CR3 Ab was comparable to that of the ferroptosis inhibitor. Furthermore, CR3-elevated neuronal survival against rotenone-induced toxicity was almost abolished by adding exogenous iron (Fig. 5C), indicating that CR3 mediates neuronal ferroptosis in response to rotenone through iron deposition.

CR3 is mainly expressed in immune cells, i.e., microglia in the brain. Subsequently, the role of microglia-neuron contact in CR3-mediated ferroptosis was investigated. BV2 microglial cells were transfected with CR3-specific siRNA to knockdown the expression of CR3 [14]. The conditioned medium (CM) prepared from rotenone-treated con-siRNA-transfected BV2 cells reduced the viability of SH-SY5Y cells compared with vehicle con-siRNA CM (Fig. 5D and Supplementary Fig. 3). However, no significant difference in cell viability was observed in SH-SY5Y cells treated with CM derived from vehicle or rotenone-treated CR3-siRNA-treasfected BV2 cells (Fig. 5D and Supplementary Fig. 3). The reduced neurotoxicity in SH-SY5Y cells treated with rotenone/CR3-siRNA CM was accompanied by a decrease of iron content and lipid peroxidation as measured by Ferroorange and BODIPY581/591C11 dyes, respectively (Fig. 5E). Quantification of rate of oxidation of BODIPY 581/591C¹¹ further supported the immunofluorescence observation (Fig. 5 and F). Moreover, rotenone-induced reduction in GPX4, a marker of ferroptosis, was also recovered in SH-SY5Y cells treated with CR3-siRNA CM (Fig. 5G). These findings suggested that microglia-neuron contacts are not required for CR3-mediated ferroptosis in response to rotenone.

3.6. NOX2 is involved in CR3-mediated neuron ferroptosis

The above results inspired us to hypothesize that CR3-mediated ferroptosis in neurons might be due to cytotoxic factor release by microglia. Previous studies indicated that proinflammatory factors and superoxide that could be released from activated microglia are capable of regulating iron balance and subsequent ferroptosis [27,55,56]. We initially measured the contents of TNFα and IL-1β in the supernatant of rotenone-treated BV2 microglia transfected with con or CR3-siRNA. However, no detectable levels of these proinflammatory factors were found (Data not shown), suggesting that proinflammatory factor production might not be critical for CR3-mediated ferroptosis.

Next, we explored the role of superoxide produced by microglia in CR3-mediated ferroptosis. Compared with the vehicle control, rotenone treatment induced superoxide production. The CR3 blocking antibody markedly reduced superoxide production in response to rotenone (Fig. 6A). NADPH oxidase 2 (NOX2) and Nrf2 are two key factors that regulate ROS production in microglia. The phosphorylation of p47^phox at the Ser345 site is essential for its membrane translocation and NOX2 activation [57]. As illustrated in Fig. 6B, rotenone-induced expression of pSer345-p47^phox was abrogated by CR3 blocking antibody, indicating that CR3 mediates p47^phox phosphorylation at the Ser345 site. To further confirm the role of CR3 in rotenone-induced NOX2 activation and superoxide production, the effects of CR3 on p47^phox membrane translocation were determined. In control siRNA-transfected microglia, rotenone treatment significantly elevated the expression of p47^phox in the membrane fraction and simultaneously, reduced the level of p47^phox in the cytosol (Fig. 6C and D). Rotenone-induced p47^phox membrane translocation was significantly attenuated by CR3 silencing since similar p47^phox expression levels in the membrane and cytosol fractions were detected between the vehicle and rotenone groups (Fig. 6C and D). The p47^phox interacts with gp91^phox in the membrane to assemble the NOX2 complex to produce superoxide. As seen in Fig. 6E, compared with vehicle, rotenone exposure elevated the interaction of p47^phox and gp91^phox, which was significantly reduced by CR3 knockdown. NOX2-derived extracellular superoxide can be changed into H₂O₂, resulting in an oxidative stress microenvironment [[58], [59], [60], [61]]. We therefore measured the concentration of H₂O₂ in CM. CR3 knockdown also reduced the levels of H₂O₂ in CM of rotenone-treated microglia (Supplementary Fig. 4). These results suggested that CR3 mediated H₂O₂ production through NOX2 activation via p47^phox Ser345 phosphorylation and subsequent membrane translocation. In contrast, no significant difference in Nrf2 and downstream HO-1 expression was observed among the groups (Data not shown).

Fig. 6.

CR3 regulates ferroptotic death of dopaminergic neurons through NOX2. (A) Superoxide production in rotenone-treated microglia with or without CR3 blocking antibody was measured using the SOD-inhibitable reduction of WST-1 experiment. (B) Representative blots and quantification of Ser345-phosphorylated p47^phox (Ser345-p-p47^phox) in rotenone-treated microglia with or without CR3 blocking antibody. (C) Representative blots of p47^phox in the membrane and cytosol fractions derived from rotenone-treated Ctrl and CR3-siRNA-transfected BV2 microglia. Gp91phox was used for internal controls of membrane fractions [41,76]. (D) Quantification of p47^phox in the membrane and cytosol fractions. (E) Representative blots of CO-IP experiments between p47^phox and gp91^phox. (F) BV2 microglia were treated with rotenone with or without GSK2795039 and catalase (CAT). The collected conditioned medium was used to treat SH-SY5Y cells for 24h and then BODIPY 581/591C¹¹ and Ferrorange staining were performed to measure lipid peroxidation and iron, respectively. (G) Quantification of THir neuron number in rotenone-treated midbrain neuron-glial cultures with or without GSK2795039, CR3 blocking antibody, or PMA. (H) BODIPY 581/591C¹¹ and Ferrorange staining were performed in H₂O₂-treated SH-SY5Y cells. (I, J) The mRNA levels of IRP2 and TFR in H₂O₂-treated SH-SY5Y cells. (K) Quantification of THir neuron number in midbrain neuron-glial cultures treated with different concentrations of H₂O₂ with or without Fer-1. n = 3–6. Scale bar = 100 μm; ∗p < 0.05, ∗∗p < 0.01.

To further assess the contribution of NOX2 and related superoxide, GSK2795039, a specific inhibitor of NOX2 and the H₂O₂ scavenger catalase (CAT) were used to block NOX2 activation and deplete H₂O₂, respectively, in rotenone-treated BV2 microglia and then the CM was collected to treat SH-SY5Y cells. Interestingly, the elevated iron content and lipid peroxidation in SH-SY5Y cells treated with rotenone CM were significantly reduced by CM prepared from GSK2795039+rotenone or CAT + rotenone-treated microglia (Fig. 6F). Importantly, inhibition of NOX2 also attenuated rotenone-induced dopaminergic neurodegeneration in midbrain primary cultures. Notably, the neuroprotective efficacy of NOX2 inhibition was comparable to that of CR3 blocking antibody. Moreover, stimulating NOX2 activation by PMA abolished CR3 blocking antibody-afforded neuroprotection (Fig. 6G). To further verify NOX2-derived superoxide contributes to CR3-mediated neuronal ferroptosis, H₂O₂ was used to mimic the oxidative environment. As seen in Fig. 6H, H₂O₂ induced iron accumulation and lipid peroxidation in SH-SY5Y cells, which were significantly dampened by iron chelator, DPO. Furthermore, the gene expression of the iron-metabolism proteins TFR and IRP2 was also elevated in SH-SY5Y cells in response to H₂O₂ challenge (Fig. 6I and J). Furthermore, we found that H₂O₂ dose-dependently induced loss of dopaminergic neurons in primary cultures, which was significantly dampened by an inhibitor of ferroptosis, Fer-1 (Fig. 6K). These results suggested that NOX2 activation contributes to CR3-regulated iron deposition, lipid peroxidation and subsequent neuronal ferroptosis in response to rotenone.

3.7. Exogenous iron recovers the sensitivity of CR3^−/− mice in response to rotenone-induced ferroptosis

To further corroborate that reduced iron accumulation is related to CR3 ablation-afforded protection, CR3^−/− mice were supplemented with iron dextran 2 weeks after the initial rotenone injection for 7 consecutive days (Fig. 7A). We hypothesized that iron supplementation could counter the protective effects of CR3 ablation in vivo. As seen in Fig. 7B–E, iron supplementation significantly increased the gene expression of SLC7A11, COX2 and ACSL4 and reduced GPX4 mRNA levels in rotenone-treated CR3^−/− mice. Furthermore, iron supplementation greatly decreased the survival of dopaminergic neurons in rotenone-treated CR3^−/− mice (Fig. 7F–H), which was accompanied with exacerbation of iNOS, TNFα and IL-1β transcription (Fig. 7I–K). Taken together, our data showed that iron supplementation recovered the sensitivity of CR3^−/− mice in response to rotenone-induced ferroptotic death of dopaminergic neurons.

Fig. 7.

Supplementation with exogenous iron recovers the ferroptosis of dopaminergic neurons in rotenone-treated CR3^−/− mice. (A) Experimental illustration. Rotenone-lesioned mice were treated with iron dextran after 2 weeks of initial rotenone injection for one week. (B–E) The mRNA levels of SLC7A11, GPX4, COX-2 and ACSL4 in rotenone-treated CR3^−/− mice with supplemented with iron dextran. n = 6. (F) Representative images of TH immunostaining in rotenone-treated CR3^−/− mice with supplemented with iron dextran. (G,H) Quantification of nigral THir neuron number (G) and striatal TH immunostaining density (H) in rotenone-treated CR3^−/− mice supplemented with iron dextran. n = 5. (I–K) The mRNA levels of iNOS, TNFα and IL-1β in rotenone-treated CR3^−/− mice supplemented with iron dextran. n = 6. Scale bar = 200 μm; ∗p < 0.05, ∗∗p < 0.01.

Consistent with attenuated dopaminergic neurodegeneration, rotenone-induced gait abnormalities in WT mice was abrogated in CR3^−/− mice since a similar pattern of gait performance between vehicle and rotenone-treated CR3^−/− mice was observed (Fig. 8A–F). Furthermore, we found that iron supplementation significantly exaggerated gait abnormalities in CR3^−/− mice by showing shorter stride lengths and wider stride distances in the rotenone group than in rotenone and iron dextran cotreated mice (Fig. 8G–L). These results suggested that iron supplementation recovered the sensitivity of CR3^−/− mice in response to rotenone-induced neuron ferroptosis and motor deficits.

Fig. 8.

Supplementation with exogenous iron promotes gait abnormalities in rotenone-treated CR3^−/− mice. (A–D) The stride length between subsequent limbs in WT and CR3^−/− mice. (E, F) The stride distance between limb placements in WT and CR3^−/− mice. n = 20–25. (G–J) The stride length between subsequent limbs in CR3^−/− mice supplemented with iron dextran. (K, L) The stride distance between limb placements in CR3^−/− mice supplemented with iron dextran. n = 10. ∗p < 0.05, ∗∗p < 0.01.

4. Discussion

CR3 is abundantly expressed in immune cells, including microglia to mediate adhesion, chemotaxis and phagocytosis [1]. Recently, CR3 has gradually become appreciated for its important role in neurological disorders. Hong et al. found that in an AD mouse model, inhibition of microglial CR3 reduced the number of phagocytic microglia, and the degree of synapse loss [62]. Deletion of TYROBP, an adapter protein for CR3, was beneficial to neurons in an early Alzheimer's pathology mouse model [63]. Blocking CR3 also displayed potent dopaminergic neuroprotection since CR3 deficiency attenuated MPTP and LPS-elicited lesion of dopaminergic neurons in mice [64,65]. However, the regulatory mechanisms for CR3-mediated neuronal damage remain to be investigated. Here, we extended a previous study and identified CR3 as a critical factor in promoting iron dyshomeostasis in neurons and subsequent ferroptosis (Fig. 9). Iron dyshomeostasis and related iron deposition in neurons play important roles in the initiation of ferroptosis [51]. Genetic deletion of CR3 mitigated iron deposition and disrupted the balance of iron metabolism proteins in the brains of rotenone-treated mice. In a cell culture system, siRNA-mediated CR3 silencing in microglia decreased the iron content in neurons in response to rotenone. Iron overload-induced lipid hydroperoxides accumulate to toxic levels in neurons, resulting in lipid peroxidation and subsequent ferroptosis [51]. Consistently, CR3 deletion or silence reduced lipid peroxidation in neurons in response to rotenone. Moreover, CR3 deletion or silence also significantly dampened the alterations in ferroptosis-related markers and degeneration of dopaminergic neurons. These results showed that CR3 regulates the ferroptotic death of neurons by promoting iron accumulation in a rotenone experimental PD model (Fig. 9). Consistent with our results, Fernández-Mendívil et al. reported that genetic deletion of heme oxygenase-1 (HO-1), an anti-inflammatory and antioxidative enzyme, in microglia prevented iron deposits, oxidative stress and neuron ferroptosis, which was associated with cognitive decline in aged mice [66].

Fig. 9.

Proposed model showing how CR3 mediates neuron ferroptosis in experimental PD. Rotenone exposure elevates CR3 expression and genetic deletion of CR3 attenuates neuron ferroptosis in rotenone-induced experimental PD model, indicating a CR3-dependent pathway. Mechanistically, CR3 mediates NOX2 through p47phox phosphorylation and subsequent membrane translocation. Superoxide production from activated NOX2 contributes to iron elevation and lipid peroxidation in neuron, resulting in neuron ferroptosis.

Mechanistically, the most critical question to answer is how CR3 regulates iron accumulation and subsequent neuronal ferroptosis. We initially proposed that CR3-mediated production of proinflammatory factors might be the main reason to induce iron dysfunction in neurons. A previous study proved that proinflammatory signals, including TNFα, IL-6 and LPS, could elevate iron contents in neurons [26]. Moreover, blocking proinflammatory cytokine production and lipid peroxidation through knockdown of ACSL4 prevented neuronal ferroptosis in an ischemic stroke model [67]. However, in our study, the levels of proinflammatory mediators, such as TNFα and IL-1β, were too low to be detected, although the mRNA levels of these factors were elevated in rotenone-treated microglia, suggesting that proinflammatory factors mediated by CR3 might not be critical for iron dysfunction and subsequent neuronal ferroptosis in response to rotenone.

In addition to inflammation, oxidative stress also plays a role in promoting iron accumulation in neurons. Gao et al. reported that oxidative stress induced by H₂O₂ elevated iron levels in a neuroblastoma cell line [68]. Consistently, Dev et al. also found that extracellular H₂O₂ treatment led to a time-dependent increase in the cellular labile iron pool in neurons, which was associated with alterations in iron transport proteins [69]. Furthermore, superoxide-mediated release of iron from ferritin resulted in iron-related lipid peroxidation, a key feature of ferroptosis [70]. In this study, we found that CR3 was capable of enhancing H₂O₂ production from microglia that could form an oxidative microenvironment and subsequently result in iron accumulation and subsequent lipid peroxidation and ferroptosis in neurons. This conclusion was supported by the following experimental evidence. First, siRNA-mediated CR3 silencing significantly reduced rotenone-induced H₂O₂ production in microglia. Second, scavenging H₂O₂ by CAT in microglia suppressed iron accumulation and lipid peroxidation in neurons in response to rotenone challenge. Third, H₂O₂ alone was capable of elevating the iron content and disrupting the balance of iron metabolism proteins in neurons. Fourth, an iron chelator mitigated H₂O₂-induced lipid peroxidation. Fifth, H₂O₂-induced dopaminergic neurotoxicity was attenuated by iron chelators in primary neuron-glial cultures.

NOX2 is a superoxide-producing enzyme in microglia. Whereas, Nrf2 is a master regulator of antioxidant cellular response in a various of cell types [71]. The production of superoxide from activated microglia can be regulated by both NOX and Nrf2 signaling [72]. Liu et al. reported that in LPS-treated microglia, deficiency of BAP31, an integral ER membrane protein, elevated LPS-induced superoxide production through NOX activation and inhibition of Nrf2 signaling [73]. Lee et al. found that urolithin B, a metabolite of ellagitannins, could dampen LPS or poly I:C-induced superoxide production through inhibition of NOX and simultaneous upregulation of HO-1, a downstream target of Nrf2 [74]. In this study, in BV2 microglia, rotenone induced H₂O₂ production, which could be mitigated by CR3 knockdown. Furthermore, we found that CR3 deficiency blocked rotenone-induced NOX2 activation and no significant effects on Nrf2 and HO-1 expression were detected, indicating that NOX2 activation was the main source of H₂O₂ production mediated by CR3. Similar regulatory effects of CR3 on NOX2 activation were reported in microglia treated with α-synuclein [8] and the environmental toxins paraquan and maneb [12]. Importantly, the activation of NOX2 was reported to be involved in neuronal ferroptosis. Cao et al. found that in a rat model of myocardial ischemia-reperfusion, lysine-specific methyltransferase 2B (KMT2B) exacerbated ferroptosis by activating TNFα/NOX2 pathway [75]. Furthermore, inhibition of NOX2 by apocynin abrogated iron elevation and ferroptotic death of dopaminergic neurons in a mouse PD model [33].

5. Conclusion

Taken together, CR3 is overexpressed in microglia in a rotenone-induced mouse PD model, resulting in iron deposition, lipid peroxidation and ferroptosis of dopaminergic neurons as well as neurobehavioral deficits. Genetic deletion or silencing of CR3 attenuated these toxic effects both in vivo and in vitro, revealing a critical role of microglial CR3 in iron-related pathology and progression in PD. Therefore, strategies aiming to prevent CR3 activation and iron deposition might be effective therapeutic approaches to retard neurodegenerative processes in patients suffering from PD.

CRediT authorship contribution statement

Qinghui Wang: Investigation, Data curation. Jianing Liu: Methodology, Investigation, Data curation. Yu Zhang: Investigation. Zhen Li: Investigation. Zirui Zhao: Investigation. Wanwei Jiang: Writing – review & editing. Jie Zhao: Writing – review & editing. Liyan Hou: Writing – review & editing, Supervision, Formal analysis, Conceptualization. Qingshan Wang: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors have declared no conflict of interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.