C3aR inhibition alleviates cerebral ischemia-reperfusion injury via STING-mediated ferroptosis through SLC7A11/GPX4

Summary

Cerebral ischemia-reperfusion injury couples complement activation with innate immune signaling and ferroptotic lipid peroxidation. We investigated whether C3aR signaling modulates stimulator of interferon genes (STING)-associated ferroptosis through the solute carrier family 7 member 11 (SLC7A11)/glutathione peroxidase 4 (GPX4) anti-ferroptotic pathway. In rat middle cerebral artery occlusion/reperfusion (MCAO/R) and BV2 oxygen-glucose deprivation/reperfusion (OGD/R) models, C3aR blockade with SB290157 was examined with 2′3′-cGAMP-mediated STING activation, immunofluorescence, western blotting, reverse transcription quantitative PCR (RT-qPCR), ferroptosis assays, and molecular docking, molecular dynamics (MD), and molecular mechanics/generalized Born surface area (MM/GBSA) analyses. C3aR inhibition improved neurological outcomes, limited infarction and edema, preserved mitochondrial membrane potential, reduced microglia-associated STING activation and ferroptosis-related Fe²⁺, reactive oxygen species (ROS), and 4-hydroxynonenal (4-HNE) accumulation, and restored SLC7A11 and GPX4 expression; these effects were partly reversed by STING activation. Structural analyses provided supportive evidence for potential STING-SLC7A11/GPX4 interactions. These findings position C3aR-STING signaling as a complement-linked mechanism regulating ferroptosis and suggest a potential therapeutic direction for limiting reperfusion-associated injury after ischemic stroke.

Graphical abstract

Highlights

• •C3aR inhibition attenuates cerebral ischemia-reperfusion injury
• •SB290157 suppresses microglial STING activation and preserves mitochondrial function
• •STING activation promotes ferroptosis by downregulating SLC7A11 and GPX4
• •2′3′-cGAMP partially reverses the protective effects of C3aR inhibition

Immunology; Biological sciences; Cell biology

Introduction

Cerebral ischemia and hypoxia are critical pathological conditions that cause substantial neuronal damage. In recent decades, the global burden of ischemic stroke has risen markedly, with the number of affected individuals reaching approximately 7.8 million—an increase of nearly 88% compared to three decades ago.¹ Clinically, the prompt restoration of cerebral blood flow through vascular recanalization—via intravenous thrombolysis or mechanical thrombectomy—is the cornerstone of acute ischemic stroke treatment.^2,3 However, reperfusion itself can paradoxically result in further neuronal injury, a phenomenon known as CI/RI.⁴ The underlying mechanisms include oxidative stress, neuroinflammation, autophagy dysregulation, and notably, ferroptosis—an iron-dependent form of regulated cell death.

The cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway, a central component of the innate immune response, has been shown to be upregulated in CI/RI.⁵ Aberrant activation of this pathway exacerbates neuroinflammation, oxidative stress, and ferroptosis, thereby contributing to neuronal dysfunction. Ferroptosis is primarily driven by lipid peroxidation in an iron-dependent manner. STING activation enhances reactive oxygen species (ROS) production through mitochondrial DNA (mtDNA) leakage and oxidative stress, catalyzing lipid peroxidation and compromising membrane integrity.⁶ Glutathione peroxidase 4 (GPX4), a key antioxidant enzyme, normally inhibits lipid peroxidation.⁷ However, STING-mediated oxidative stress can deplete intracellular glutathione (GSH), inactivate GPX4, and result in the accumulation of lipid peroxides, ultimately triggering ferroptosis.⁸

The solute carrier family 7 member 11 (SLC7A11) encodes a light-chain subunit of the system Xc⁻ antiporter, which facilitates cystine uptake and glutamate efflux. Cystine is vital for GSH synthesis and the maintenance of redox homeostasis.⁹ Downregulation of SLC7A11 reduces intracellular cystine and GSH, thereby impairing GPX4 activity and enhancing susceptibility to ferroptosis.¹⁰ In vitro models of ischemic injury, such as oxygen-glucose deprivation/reperfusion (OGD/R), demonstrate reduced expression of both SLC7A11 and GPX4, promoting ferroptotic cell death.¹¹ Li et al. reported that STING may indirectly suppress SLC7A11 by activating autophagy, thereby facilitating ferroptosis through increased lipid peroxidation.¹²

Complement component 3 (C3) is another critical mediator of innate immunity. Upon activation, C3 is cleaved into C3a, which binds to its cognate G protein-coupled receptor (GPCR) C3aR.¹³ C3aR is highly expressed in astrocytes, microglia, and neural stem cells. During the acute phase post-stroke, C3aR activation exacerbates neuroinflammation by increasing vascular permeability, promoting endothelial activation, and facilitating infiltration of peripheral immune cells, thereby compromising the blood-brain barrier (BBB).^14,15 In renal ischemia-reperfusion injury, C3a/C3aR activation has been shown to induce ERK signaling and ROS production in neutrophils, resulting in tissue damage.¹⁶ Moreover, C3a-C3aR signaling enhances ferroptosis-driven neurodegeneration in Parkinson’s disease models.¹⁷ Intriguingly, in infectious models, C3a has been reported to inhibit STING pathway activation via C3aR, suppressing downstream phosphorylation of TBK1 and p38 mitogen-activated protein kinase (MAPK).¹⁸

Despite these findings, the role of C3a/C3aR in modulating ferroptosis during cerebral ischemia-reperfusion remains poorly understood. It is unclear whether C3a/C3aR exerts its effects by regulating ferroptosis through the cGAS-STING pathway or via modulation of ROS. The precise molecular interactions between these pathways warrant further investigation.

In this study, we employed a rat model of CI/RI to investigate the regulatory relationship between C3aR and the cGAS-STING pathway. Pharmacological inhibition of C3aR using SB290157 and activation of STING with 2′3′-cGAMP were applied to explore their respective roles. Our findings reveal that CI/RI activates C3aR and induces STING signaling, leading to the suppression of SLC7A11 and GPX4 expression and the promotion of ferroptosis. Mechanistically, inhibition of C3aR attenuates cGAS-STING activation and ROS production, thereby reversing ferroptosis through upregulation of the SLC7A11/GPX4 axis. These results provide novel insights into the role of the C3aR-STING-ferroptosis axis in ischemic brain injury and suggest potential therapeutic targets for CI/RI.

Results

Cerebral blood flow alterations confirm successful ischemia-reperfusion modeling

Relative cerebral blood flow (rCBF) was assessed by laser speckle contrast imaging (LSCI) to verify the successful establishment of the MCAO/R model. As shown in Figure 1, rCBF in the left hemisphere was significantly reduced after ischemia induction compared with baseline levels (p = 0.0008). Following reperfusion, rCBF was partially restored and significantly increased relative to the post-ischemic level (p = 0.0297), confirming the successful induction of cerebral ischemia-reperfusion in rats.

Figure 1.

Changes in rCBF before and after cerebral ischemia-reperfusion

(A) LSCI of cerebral blood flow in the normal rat brain (baseline), during ischemia, and 24 h after reperfusion.

(B) Representative photographs and corresponding LSCI scans of the ischemic brain. Regions of interest (ROIs) were used for perfusion analysis and are expressed in perfusion units (PUs).

(C) Quantitative analysis of rCBF at three time points. Data are presented as mean ± SD (n = 3). Statistical analysis was performed using one-way repeated-measures ANOVA followed by Holm-Sidak multiple comparisons test. ∗p < 0.05, ∗∗∗p < 0.001.

Inhibition of C3aR suppresses cGAS-STING pathway activation

To investigate whether C3aR inhibition preserves mitochondrial integrity and suppresses STING activation, Δψm was first assessed in BV2 microglia using JC-1 staining. OGD/R markedly decreased Δψm compared with the control group, indicating significant mitochondrial dysfunction (p = 0.0003). In contrast, treatment with SB290157 significantly restored Δψm levels (p = 0.0005), suggesting improved mitochondrial integrity (Figures 2A and 2B).

Figure 2.

C3aR inhibition preserves mitochondrial integrity and suppresses microglia-associated STING activation after MCAO/R

(A) Representative fluorescence images of Δψm in BV2 microglia under normal conditions, after OGD/R, and following OGD/R + SB290157 (1 μg/μL) treatment. Scale bars, 150 μm.

(B) Quantitative analysis of Δψm in the three groups.

(C) Representative double immunofluorescence images of STING (red) and Iba1 (green) in the ipsilateral peri-infarct cortical region from the sham, MCAO/R, MCAO/R + SB290157, and MCAO/R + SB290157 + 2′3′-cGAMP groups. Nuclei were counterstained with DAPI (blue). Scale bars, 20 μm.

(D) Quantitative analysis of STING fluorescence intensity in the four groups.

(E) Representative western blot images showing C3aR, cGAS, and STING expression in the ipsilateral peri-infarct cortical region.

(F–H) Quantitative analysis of C3aR, cGAS, and STING protein levels.

Data are expressed as mean ± SD (n = 3 independent biological replicates for western blot; n = 5 for Δψm and immunofluorescence). Statistical analysis was performed using one-way ANOVA followed by Holm-Sidak multiple comparisons test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To further define the cellular localization of STING activation in vivo, double immunofluorescence staining for STING and Iba1 was performed in the ipsilateral peri-infarct cortical region. Compared with the sham group, MCAO/R markedly increased STING fluorescence intensity in Iba1-positive microglia, accompanied by enhanced co-localization signals, indicating microglia-associated STING activation following cerebral ischemia-reperfusion injury (p = 0.0004). Treatment with SB290157 significantly reduced STING fluorescence intensity and weakened its co-localization with Iba1-positive cells (p = 0.0007), whereas 2′3′-cGAMP partially reversed this inhibitory effect and restored STING expression in microglia (p = 0.0065) (Figures 2C and 2D).

Consistent with these observations, western blot analysis further supported that MCAO/R was associated with increased expression of C3aR, cGAS, and STING in the ischemic penumbra (p = 0.0135 for C3aR, p = 0.0231 for cGAS, and p = 0.0038 for STING), whereas SB290157 treatment was associated with reduced expression of these proteins (p = 0.0432 for C3aR, p = 0.0455 for cGAS, and p = 0.0017 for STING) (Figures 2E–2H). These findings suggest that C3aR inhibition may preserve mitochondrial integrity and suppress microglia-associated cGAS-STING pathway activation following MCAO/R.

C3aR inhibition alleviates ischemia-reperfusion injury via STING modulation

To evaluate the neuroprotective effect of C3aR inhibition, 2,3,5-triphenyltetrazolium chloride (TTC) staining was performed to assess infarct size. The MCAO/R group exhibited a markedly increased infarct area compared with the sham group (p = 0.0027), whereas SB290157 treatment significantly reduced infarct volume (p = 0.0008). This protective effect was partially reversed by co-administration of the STING agonist 2′3′-cGAMP, which significantly increased infarct size (p = 0.0041) (Figures 3A and 3B).

Figure 3.

Evaluation of infarct volume and neurological deficits following ischemia-reperfusion using TTC staining and mNSS

(A) Representative TTC-stained brain sections from the four experimental groups, with infarcted regions indicated in white.

(B) Quantitative analysis of infarct volume.

(C) mNSS at 24, 48, and 72 h post-reperfusion.

(D) Quantification of brain water content at day 3 post-reperfusion.

Data are expressed as mean ± SD (n = 3 for infarct volume, n = 5 for neurological scores and brain water content). Statistical analysis for (B) and (D) was performed using one-way ANOVA followed by Holm-Sidak multiple comparisons test. Statistical analysis for (C) was performed using two-way repeated-measures ANOVA followed by Holm-Sidak post hoc test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Consistent with the TTC findings, brain water content was significantly elevated in the MCAO/R group compared with the sham group (p = 0.0051) and was reduced following SB290157 treatment (p = 0.0486). The addition of 2′3′-cGAMP partially reversed this improvement, resulting in aggravated cerebral edema (p = 0.0073) (Figure 3D).

Neurological deficits were further evaluated using modified neurological severity score (mNSS) at 24, 48, and 72 h post-reperfusion. The sham group showed no detectable deficits, whereas the MCAO/R group exhibited the highest scores, indicating severe neurological impairment. In contrast, SB290157 significantly improved neurological function, while co-administration of 2′3′-cGAMP partially abolished these beneficial effects (Figure 3C). These findings suggest that C3aR inhibition exerts neuroprotective effects, at least in part through modulation of STING signaling.

C3aR antagonism attenuates neuronal injury following MCAO/R

Histological analysis of the ipsilateral peri-infarct cortical region using H&E and Nissl staining revealed well-preserved neuronal morphology in the sham group, characterized by intact cellular architecture and abundant cytoplasm. In contrast, the MCAO/R group exhibited marked neuronal degeneration, including cellular shrinkage, disorganized arrangement, cytoplasmic vacuolization, and loss of Nissl bodies (p = 0.0053 for H&E and p = 0.0002 for Nissl).

Treatment with SB290157 significantly attenuated these pathological changes and preserved neuronal integrity (p = 0.0113 for H&E and p = 0.0172 for Nissl). Following administration of 2′3′-cGAMP, this neuroprotective effect was partially attenuated and accompanied by a trend toward increased neuronal injury (p = 0.0341 for H&E and p = 0.0417 for Nissl) (Figure 4).

Figure 4.

Histological analysis of neuronal damage in the ipsilateral peri-infarct cortex by H&E and Nissl staining

(A) Representative images of H&E and Nissl staining in the ipsilateral peri-infarct cortical region (magnification ×40; scale bars, 0.020 mm).

(B) Quantification of surviving neurons based on H&E staining.

(C) Quantification of Nissl-positive neurons.

All data are presented as mean ± SD with individual data points shown (n = 3). Statistical analysis was performed using one-way ANOVA followed by Holm-Sidak multiple comparisons test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

These findings suggest that C3aR antagonism alleviates neuronal injury in the ischemic penumbra, at least in part through suppression of STING signaling.

C3aR inhibition mitigates ferroptosis and improves cerebral ischemia-reperfusion outcomes

Ferroptosis, a form of iron-dependent cell death driven by lipid peroxidation, was evaluated by quantifying Fe²⁺ levels in the ischemic penumbra. Compared with the sham group, the MCAO/R group exhibited significant Fe²⁺ accumulation (p = 0.0004). Treatment with SB290157 significantly decreased Fe²⁺ content (p = 0.0010), whereas co-treatment with 2′3′-cGAMP partially reversed this effect (p = 0.0014) (Figure 5A).

Figure 5.

Expression of ferroptosis markers (Fe²⁺, ROS, and 4-HNE) in the ipsilateral peri-infarct cortex of rats

(A) Quantitative analysis of Fe²⁺ levels.

(B) Quantitative analysis of ROS levels.

(C) Quantitative analysis of 4-HNE levels.

All data are expressed as mean ± SD (n = 3). Statistical analysis was performed using one-way ANOVA followed by Holm-Sidak multiple comparisons test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

In addition, ROS and 4-HNE, two key indicators of lipid peroxidation, were markedly elevated in MCAO/R rats (p = 0.0006 for ROS and p = 0.0004 for 4-HNE). These oxidative stress markers were significantly reduced following SB290157 treatment (p = 0.0015 for ROS and p = 0.0007 for 4-HNE), whereas 2′3′-cGAMP partially abolished these protective effects (Figures 5B and 5C). These findings suggest that C3aR inhibition may alleviate ferroptosis by attenuating oxidative stress, at least in part through STING suppression.

C3aR inhibition regulates ferroptosis via the STING-SLC7A11/GPX4 axis

To further explore the molecular mechanism by which C3aR regulates ferroptosis, we examined the expression of STING, SLC7A11, and GPX4 in the ischemic penumbra.

Western blot analysis suggested that MCAO/R was associated with increased STING expression and reduced SLC7A11 and GPX4 levels (p = 0.0038 for STING, p = 0.0041 for SLC7A11, and p = 0.0112 for GPX4). In contrast, treatment with SB290157 was associated with decreased STING expression and partial restoration of SLC7A11 and GPX4 levels (p = 0.0017 for STING, p = 0.0360 for SLC7A11, and p = 0.0374 for GPX4) (Figures 6A–6D).

Figure 6.

Expression of STING, SLC7A11, and GPX4 in the ipsilateral peri-infarct cortex of rats and protein docking analysis

(A) Representative western blot images of STING, SLC7A11, and GPX4 expression in the sham, MCAO/R, and SB290157 groups.

(B–D) Quantitative analysis of STING, SLC7A11, and GPX4 protein levels across the three groups (n = 3 independent biological replicates per group).

(E) Western blot images of STING, SLC7A11, and GPX4 expression in the sham, MCAO/R, SB290157, and SB290157 + 2′3′-cGAMP groups.

(F–H) Corresponding quantitative analysis of protein expression in these four groups (n = 3 independent biological replicates per group).

(I–K) Quantitative RT-qPCR analysis of STING, SLC7A11, and GPX4 mRNA levels (n = 5 independent biological replicates per group).

(L) STING-SLC7A11 protein docking model, including stick (top), cartoon (middle), and surface (bottom) representations, showing key amino acid interactions.

(M) STING-GPX4 protein docking model with corresponding visualization.

All quantitative data are expressed as mean ± SD. Statistical analysis for (B)–(K) was performed using one-way ANOVA followed by Holm-Sidak multiple comparisons test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. NS indicates no statistically significant difference.

Consistent with these observations, co-treatment with 2′3′-cGAMP was associated with increased STING expression and decreased SLC7A11 and GPX4 levels, partially counteracting the effects of SB290157 (p = 0.0409 for STING, p = 0.0372 for SLC7A11, and p = 0.0009 for GPX4) (Figures 6E–6H).

RT-qPCR analysis using the primers listed in Table 1 showed a similar trend, with STING mRNA expression significantly increased and SLC7A11 and GPX4 mRNA expression significantly decreased in the 2′3′-cGAMP group (p = 0.0014 for STING, p = 0.0027 for SLC7A11, and p = 0.0042 for GPX4) (Figures 6I–6K), further supporting the protein-level findings.

Table 1.

Primers used for quantitative real-time PCR

Gene	Forward primer (5′–3′)	Reverse primer (5′–3′)
STING	CCACTGCCGCCTCATTGTCTA	GCCACTCATGGTAACTTCCTCCTT
SLC7A11	AACTGCTGGTAATACGCCCC	CATCACCACAGTGATGCCCA
GPX4	GCAGGAGCCAGGAAGTAATCAAG	ACAGTGGGTGGGCATCGTC
GAPDH	GCAAGAGAGAGGCCCTCAG	TGTGAGGGAGATGCTCAGTG

To further investigate the potential protein-protein interactions involved in this signaling axis, molecular docking analysis was performed. Structural models revealed plausible binding interfaces between STING and SLC7A11 (Figure 6L) as well as STING and GPX4 (Figure 6M), based on stick, cartoon, and surface representations. The predicted binding sites exhibited multiple hydrogen-bonding and electrostatic interaction regions, suggesting structural compatibility between STING and these ferroptosis-related proteins.

To further validate the stability and persistence of these predicted interactions, molecular dynamics (MD) simulations were subsequently performed.

MD simulation confirms stable interactions between STING and SLC7A11/GPX4

To further validate the protein-protein interactions predicted by molecular docking, 100 ns MD simulations were performed for the STING-SLC7A11 and STING-GPX4 complexes. As shown in Figure 7A, the root-mean-square deviation (RMSD) trajectories indicated that both complexes gradually reached equilibrium during the simulation. The STING-SLC7A11 complex converged earlier, reaching a stable plateau after approximately 20 ns, with RMSD fluctuations maintained around 1.0 nm. In contrast, the STING-GPX4 complex underwent greater conformational rearrangement during the initial phase and stabilized after approximately 80–90 ns, with RMSD values fluctuating around 2.0 nm. These findings indicate that both complexes formed stable conformations, with the STING-SLC7A11 complex showing relatively greater structural stability.

Figure 7.

MD simulation profiles of STING complexes with GPX4 and SLC7A11

(A) Root-mean-square deviation (RMSD).

(B and C) Root-mean-square fluctuation (RMSF).

(D) Radius of gyration (Rg).

(E) Number of intermolecular hydrogen bonds.

(F) Solvent-accessible surface area (SASA).

(G) Free energy landscape (FEL) analysis of the SLC7A11-STING complex.

(H) FEL analysis of the GPX4-STING complex.

(I) MM/GBSA binding free energy decomposition.

Both STING complexes exhibited favorable dynamic stability, persistent intermolecular interactions, and energetically favorable binding profiles throughout the 100 ns simulation, supporting the structural feasibility of STING interaction with GPX4 and SLC7A11.

Residue-level flexibility was further evaluated by root-mean-square fluctuation (RMSF) analysis. As shown in Figures 7B and 7C, both systems exhibited generally low residue fluctuations, suggesting limited structural perturbation during the simulation. Higher RMSF values were mainly observed in terminal and loop regions, consistent with their intrinsic flexibility and without affecting the core binding interface.

The structural compactness of the complexes was assessed by the radius of gyration (Rg). As shown in Figure 7D, the Rg values of the STING-SLC7A11 and STING-GPX4 complexes remained stable throughout the 100 ns simulation, fluctuating around 3.5 and 2.1 nm, respectively, indicating no significant structural collapse or unfolding.

Hydrogen bond analysis further demonstrated stable intermolecular interactions. As shown in Figure 7E, both complexes maintained a relatively constant number of hydrogen bonds throughout the simulation, despite minor dynamic fluctuations, indicating persistent and stable protein-protein interactions.

Consistently, solvent-accessible surface area (SASA) analysis showed no marked changes during the simulation (Figure 7F), with values fluctuating around 470 nm² for the STING-SLC7A11 complex and 165 nm² for the STING-GPX4 complex, further supporting stable folded conformations.

Free energy landscape (FEL) analysis revealed well-defined energy minima in both systems (Figures 7G and 7H), indicating sufficient conformational sampling and stable low-energy states.

To quantitatively assess binding affinity, MM/GBSA free energy calculations were performed using the equilibrated trajectories from 90 to 100 ns. To further validate the protein-protein interactions predicted by molecular docking, 100 ns MD simulations were performed for the STING-SLC7A11 and STING-GPX4 complexes.

As shown in Figure 7A, the RMSD trajectories indicated that both complexes gradually reached equilibrium during the simulation. The STING-SLC7A11 complex converged earlier, reaching a stable plateau after approximately 20 ns, with RMSD fluctuations maintained around 1.0 nm. In contrast, the STING-GPX4 complex underwent greater conformational rearrangement during the initial phase and stabilized after approximately 80–90 ns, with RMSD values fluctuating around 2.0 nm. These findings indicate that both complexes adopted stable conformations, with the STING-SLC7A11 complex exhibiting relatively greater structural stability.

Residue-level flexibility was further evaluated by RMSF analysis. As shown in Figures 7B and 7C, both systems exhibited generally low residue fluctuations, suggesting limited structural perturbation during the simulation. Higher RMSF values were mainly observed in the terminal and loop regions, consistent with their intrinsic flexibility and without affecting the core binding interface.

The structural compactness of the complexes was further assessed by the Rg. As shown in Figure 7D, the Rg values of the STING-SLC7A11 and STING-GPX4 complexes remained stable throughout the 100 ns simulation, fluctuating around 3.5 and 2.1 nm, respectively, indicating no significant structural collapse or unfolding.

Hydrogen bond analysis further demonstrated stable intermolecular interactions. As shown in Figure 7E, both complexes maintained a relatively constant number of hydrogen bonds throughout the simulation despite minor dynamic fluctuations, indicating persistent protein-protein interactions.

Consistently, SASA analysis showed no marked changes during the simulation (Figure 7F), with values fluctuating around 470 nm² for the STING-SLC7A11 complex and 165 nm² for the STING-GPX4 complex, further supporting the maintenance of stable folded conformations. In addition, FEL analysis revealed well-defined energy minima in both systems (Figures 7G and 7H), indicating sufficient conformational sampling and stable low-energy states.

To quantitatively assess binding affinity, MM/GBSA free energy calculations were performed using the equilibrated trajectories from 90 to 100 ns. As shown in Figure 7I and Table 2, the total binding free energy was −116.38 kcal/mol for the STING-SLC7A11 complex and −53.95 kcal/mol for the STING-GPX4 complex, indicating thermodynamically favorable binding in both systems. In both cases, van der Waals interactions were the dominant energetic contributors, while electrostatic interactions also contributed favorably.

Table 2.

The binding free energy by MM/GBSA (kcal/mol)

Binding energy	GPX4-STING (kcal/mol)	SLC7A11-STING (kcal/mol)
ΔEvdw	−55.78 ± 6.79	−123.24 ± 7.57
ΔEelec	−5.16 ± 6.78	−35.66 ± 16.82
ΔEGB	14.38 ± 6.36	60.1 ± 14.95
ΔEsurf	−7.39 ± 0.81	−17.58 ± 1.24
ΔEGas	−60.94 ± 9.26	−158.9 ± 16.92
ΔEsolv	6.99 ± 6.13	42.52 ± 14.8
ΔEBind	−53.95 ± 6.43	−116.38 ± 7.3

ΔEvdw, van der Waals energy; ΔEelec, electrostatic energy; ΔEGB, polar solvation energy; ΔEsurf, nonpolar solvation energy; ΔEGas, gas-phase free energy; ΔEsolv, solvation free energy; ΔEBind, total binding free energy.

These MD simulation results provide supportive structural and thermodynamic evidence for stable interactions between STING and SLC7A11/GPX4, further supporting the potential involvement of the STING-SLC7A11/GPX4 axis in ferroptosis regulation.

Discussion

While C3aR has been implicated in stroke pathology,¹⁹ the underlying mechanisms remain incompletely understood. Unlike traditional forms of cell death, ferroptosis—an iron-dependent, lipid peroxidation-driven process—has emerged as a critical contributor to CI/RI.²⁰ Previous studies have demonstrated that ferroptosis is activated following ischemic stroke and that pharmacological inhibition of ferroptosis can improve neuronal survival and neurological function in both in vivo and in vitro models.²¹

C3aR is a member of the GPCR family. Upon activation, it recruits β-arrestin to mediate receptor internalization and desensitization, while also serving as a scaffold for alternative signaling pathways.²² Following cerebral ischemia-reperfusion, complement protein C1q promotes the cleavage of C3 into C3a and C3b. C3a, an anaphylatoxin, binds to C3aR to trigger downstream inflammatory signaling.^22,23 C3aR activation enhances microglial activation and initiates the NF-κB pathway, thereby increasing the transcription and release of pro-inflammatory cytokines such as TNF-α and IL-6. This cascade ultimately leads to neutrophil infiltration, disruption of the BBB, and exacerbation of neuronal damage.¹⁹ Zhang et al. reported that in chronic cerebral hypoperfusion, activated C3aR induces STAT3 signaling in microglia, promoting a pro-inflammatory phenotype and contributing to white matter damage through increased myelin phagocytosis and abnormal cell migration.²⁴

The cGAS-STING pathway is a key driver of inflammation, metabolic imbalance, and cell death following CI/RI.²⁵ Nuclear and mtDNA, released from necrotic neurons after brain injury, function as damage-associated molecular patterns (DAMPs) that activate cGAS.²⁶ Upon binding cytosolic double-stranded DNA (dsDNA), cGAS synthesizes the second messenger 2′3′-cGAMP, which activates STING. STING subsequently promotes microglial M1 polarization, NLRP3 inflammasome activation, and mitochondrial autophagy imbalance, thereby exacerbating injury.²⁷

Our findings suggest that C3aR activation contributes to microglial mitochondrial dysfunction, leading to mtDNA leakage and subsequent activation of the cGAS-STING pathway.^28,29 In the MCAO/R rat model, C3aR expression was markedly elevated, while BV2 microglia subjected to OGD/R exhibited a significant reduction in Δψm, indicating mitochondrial injury.³⁰ These alterations were substantially attenuated by SB290157, a C3aR antagonist, which preserved Δψm and suppressed the expression of cGAS and STING. These findings are consistent with the study by Ishii et al., who reported that C3aR translocates to mitochondria under oxidative stress conditions, thereby disrupting calcium homeostasis and impairing respiratory chain function.²⁸ Importantly, the double immunofluorescence staining for STING and Iba1 further strengthens the cellular interpretation of our results. The increased co-localization of STING within Iba1-positive microglia after MCAO/R, together with its attenuation by SB290157 and partial restoration by 2′3′-cGAMP, provides direct in vivo evidence that microglia-associated STING activation participates in C3aR-mediated ischemic injury. Similarly, Ding et al. reported that STING was predominantly localized in activated microglia (Iba1⁺) within the injured cortical region, as demonstrated by double immunofluorescence staining.³¹ This observation further supports a mechanistic link between complement-mediated neuroinflammation and microglia-associated ferroptotic signaling.

Moreover, Chi et al. demonstrated that C3aR antagonism reduced neuronal injury and α-synuclein pathology in a Parkinson’s disease model,³² suggesting that the neuroprotective role of C3aR inhibition may extend beyond ischemic injury to other neuroinflammatory disorders. In agreement with these findings, our H&E and Nissl staining results demonstrated markedly reduced neuronal damage in SB290157-treated rats, further supporting the neuroprotective effect of C3aR inhibition.

We propose that the neuroprotective effect of C3aR inhibition is mediated, at least in part, through suppression of STING-dependent ferroptosis. Previous studies have demonstrated that STING can interact with NCOA4, thereby promoting ferritinophagy and intracellular iron release, ultimately triggering ferroptotic cell death.³³ In addition, cGAS-mediated sensing of mtDNA activates the STING-TBK1-IRF3 signaling axis, which has been reported to suppress GPX4 and FTH1 expression while upregulating ACSL4 and COX-2, thereby promoting lipid peroxidation and iron overload.³⁴

Consistent with these findings, SB290157 significantly reduced the levels of Fe²⁺, ROS, and 4-HNE in the ischemic penumbra, all of which are established hallmarks of ferroptosis. Notably, these protective effects were partially reversed by co-administration of 2′3′-cGAMP, a STING agonist, further supporting the notion that STING functions as a key downstream mediator of ferroptotic injury following C3aR activation.

At the molecular level, we observed that MCAO/R markedly decreased the expression of SLC7A11 and GPX4, two central regulators of the anti-ferroptotic defense system. In contrast, SB290157 treatment restored the expression of both proteins while concomitantly reducing STING levels. Mechanistically, SLC7A11, a functional subunit of the system Xc⁻ cystine/glutamate antiporter, mediates the uptake of extracellular cystine for conversion into cysteine, which serves as a precursor for GSH synthesis.³⁵ GSH is a critical intracellular antioxidant and an essential cofactor for GPX4, which detoxifies lipid peroxides and prevents ferroptotic membrane damage. Accordingly, GPX4 inactivation results in the accumulation of toxic lipid peroxides and ultimately induces ferroptosis.³⁶

Previous studies have further emphasized the importance of this pathway. For example, kaempferol has been shown to alleviate ferroptosis and iron overload through upregulation of the SLC7A11/GPX4 axis,¹¹ whereas aconitine suppresses this pathway, leading to mitochondrial dysfunction and ferroptotic cell death.³⁷ Importantly, in the SB290157 + 2′3′-cGAMP group, STING expression was restored, accompanied by a secondary decline in SLC7A11 and GPX4 levels, which is consistent with STING-mediated suppression of the anti-ferroptotic axis. Together, these findings support the existence of a C3aR-STING-SLC7A11/GPX4 regulatory pathway that contributes to ferroptosis during cerebral ischemia-reperfusion injury.

The protein docking, MD simulations, and MM/GBSA analyses of the anti-ferroptotic axis provide additional structural support for the proposed STING-SLC7A11/GPX4 regulatory pathway. Compared with static docking alone, the stable RMSD convergence, low residue fluctuations outside the flexible terminal regions, persistent intermolecular hydrogen bonding, and favorable binding free energies collectively support the structural feasibility and dynamic stability of STING interactions with both SLC7A11 and GPX4. Although these computational findings do not by themselves establish direct binding in vivo, they provide important thermodynamic and conformational evidence supporting a mechanistic link between STING activation and ferroptosis regulation.

Importantly, the interpretation of our findings should also be considered in the context of different experimental stroke models. The present study employed a transient MCAO/R model, which is widely recognized as one of the most clinically relevant preclinical paradigms for investigating ischemia followed by reperfusion, particularly in the setting of intravenous thrombolysis or mechanical thrombectomy.³⁸ Compared with permanent MCAO, transient MCAO/R more directly captures reperfusion-driven pathological processes, including mitochondrial dysfunction, oxidative stress, sterile inflammation, BBB disruption, and ferroptosis-related secondary injury.³⁹ In contrast, photothrombotic stroke models generate highly reproducible and anatomically well-defined cortical lesions, making them advantageous for studies of cortical circuit remodeling and localized histopathology; however, they generally lack a true reperfusion phase and may therefore underrepresent reperfusion-associated mechanisms.⁴⁰ Moreover, embolic or thromboembolic models provide additional translational value because they more closely mimic clot-dependent vascular occlusion and recanalization therapy, although they often introduce greater heterogeneity in infarct evolution and reperfusion dynamics.⁴¹ Accordingly, the C3aR-STING-SLC7A11/GPX4 axis identified in the present study may be particularly relevant to reperfusion-dominant injury states, while future validation in permanent, photothrombotic, and embolic stroke models will be essential to define the broader generalizability and translational relevance of this mechanism.

These model-dependent differences also have important implications for the interpretation of microglial responses. Emerging evidence suggests that microglia do not exert uniform effects across different stroke paradigms; rather, their role is influenced by lesion size, anatomical location, temporal stage, and the presence or absence of reperfusion. For example, microglial responses may differ substantially between permanent MCAO and photothrombotic stroke, highlighting that the net effect of inflammatory cells is highly context dependent rather than universal.⁴² This consideration is particularly relevant to our study, as the newly added STING/Iba1 co-localization data indicate that microglia are an important cellular contributor to STING signaling in transient MCAO/R. Nevertheless, these findings do not exclude potential contributions from astrocytes, neurons, endothelial cells, or infiltrating myeloid cells under other ischemic conditions. These findings further support the concept that C3aR inhibition attenuates cerebral ischemia-reperfusion injury by suppressing microglia-associated STING activation and downstream ferroptotic signaling, while also highlighting the broader relevance of this mechanism across reperfusion-dominant stroke models.

In summary, the present study demonstrates that C3aR plays an important role in the pathogenesis of cerebral ischemia-reperfusion injury by modulating the cGAS-STING signaling pathway and ferroptosis-related processes. Pharmacological inhibition of C3aR attenuated STING activation, preserved mitochondrial function, and restored the SLC7A11/GPX4 anti-ferroptotic axis, thereby alleviating ferroptosis-associated neuronal injury and improving neurological outcomes following CI/RI.

Collectively, these findings identify the C3aR-STING-SLC7A11/GPX4 axis as a promising mechanistic pathway and a potential therapeutic target for mitigating ischemia-reperfusion-induced brain injury.

Limitations of the study

Several limitations of this study should be acknowledged. First, the sample size for the in vivo western blot analyses was relatively limited, which may affect the statistical robustness of protein-level quantification. Second, although molecular docking, MD simulations, and MM/GBSA analyses provided supportive structural evidence for potential interactions between STING and SLC7A11/GPX4, these computational findings do not establish direct binding or regulatory interactions in vivo. Future studies using co-immunoprecipitation, proximity ligation assays, and cell-type-specific loss-of-function approaches are needed to validate these interactions and clarify their spatial and temporal coordination within the post-ischemic microenvironment. Third, the in vitro experiments were performed using BV2 microglial cells, which may not fully recapitulate the phenotype and functional heterogeneity of primary microglia after ischemia-reperfusion. Fourth, the present study relied mainly on pharmacological inhibition of C3aR and activation of STING; therefore, genetic approaches such as C3aR knockdown, STING knockdown, or conditional cell-specific manipulation would further strengthen causal inference. Finally, as with many preclinical stroke studies, the use of young, otherwise healthy male animals, potential anesthesia-related effects, and reliance on a transient MCAO/R model may limit the direct extrapolation of these findings to clinically heterogeneous stroke populations. Future validation in additional stroke models, aged animals, females, and animals with vascular risk factors will be important to assess the broader translational relevance of the C3aR-STING-SLC7A11/GPX4 axis.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Yanqiang Wang (wangyq6@alumni.sysu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •All data reported in this article will be shared by the lead contact upon request.
• •This article does not report original code.
• •Any additional information is available from the lead contact upon request.

Acknowledgments

We thank all members who were involved in this research. This work was supported by the Yuan Du Scholars, Affiliated Hospital of Shandong Second Medical University Horizontal Project (grant nos. WYFYKY-HX202307 and WYFYKY-HX202201); the National Natural Science Foundation of China (grant no. 81870943); the Health China · BuChang ZhiYuan Public Welfare Project for Heart and Brain Health (grant no. HIGHER2023072); and the Shandong Second Medical University Affiliated Hospital Technology Development Project (grant nos. 2023FYM001 and 2023FYM006).

Author contributions

Y.Z., Y.X., and Y.W. conceived and designed this study. Y.Z., Y.X., Z.C., S.H., J.L., and M.P. performed the experiments. Y.Z. and Y.X. performed the statistical analysis. Y.Z. wrote the manuscript. Y.W. revised the manuscript. All authors read and approved the final version of the manuscript.

Declaration of interests

The authors declare no conflict of interest.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-C3aR antibody, 1:2000	Affinity	Cat# DF10198; RRID: AB_2840777
Anti-cGAS antibody, 1:1000	Proteintech	Cat# 84045-1-RR; RRID: AB_3671612
Anti-STING antibody, 1:2000	Proteintech	Cat# 19851-1-AP; RRID: AB_10665370
Anti-GPX4 antibody, 1:1000	Abmart	Cat# T56959; RRID: AB_2936461
Anti-SLC7A11 antibody, 1:1000	Servicebio	Cat# GB115276
Anti-β-actin antibody, 1:5000	Abways	Cat# CY1132
Primary antibody against STING for immunofluorescence	Servicebio	Cat# GB150063-100
Primary antibody against Iba1 for immunofluorescence	Servicebio	Cat# GB15105-100; RRID: AB_3663047
Chemicals, peptides, and recombinant proteins
SB290157	Merck	559410-10 MG
2′3′-cGAMP	InvivoGen	VAC-NACGA23
Dimethyl sulfoxide (DMSO)	Servicebio	GC203006-10 mL
Sterile phosphate-buffered saline (PBS)	Servicebio	G4202-500 ML
2,3,5-Triphenyltetrazolium chloride (TTC)	Solarbio	G3005
DAPI	Servicebio	G1012-10 ML
DMEM	Servicebio	G4511DLEU-500 ML
Glucose-free DMEM	Servicebio	G4528-500 ML
Fetal bovine serum	Thermo Fisher	A5256701
Penicillin-streptomycin	Thermo Fisher	15140148
RIPA buffer	Beyotime	P0013B
PVDF membranes	Beyotime	FFP77
ECL substrate	Beyotime	P0018AS
TRIzol reagent	Thermo Fisher	15596026CN
PrimeScript™ RT Master Mix	Takara	RR036A
TB Green Premix Ex Taq II	Takara	RR091Q
Critical commercial assays
JC-1 detection kit	Beyotime	C2006
4-HNE assay kit	Affandi	A117371
Fe2+ assay kit	Nanjing Jiancheng	H129-1-2
Experimental models: Cell lines
BV2 murine microglial cells	Procell	CL-0493
Experimental models: Organisms
Male Sprague-Dawley rats	Jinan Pengyue Experimental	PY-C-202
Oligonucleotides
STING qPCR primers	This study	Forward: CCACTGCCGCCTCATTGTCTA; Reverse: GCCACTCATGGTAACTTCCTCCTT
SLC7A11 qPCR primers	This study	Forward: AACTGCTGGTAATACGCCCC; Reverse: CATCACCACAGTGATGCCCA
GPX4 qPCR primers	This study	Forward: GCAGGAGCCAGGAAGTAATCAAG; Reverse: ACAGTGGGTGGGCATCGTC
GAPDH qPCR primers	This study	Forward: GCAAGAGAGAGGCCCTCAG; Reverse: TGTGAGGGAGATGCTCAGTG
Software and algorithms
ImageJ software	National institutes of health	http://ImageJ.nih.gov/ij
GraphPad Prism	GraphPad Software	Version 10.1.2
Instruments and equipment
Laser speckle contrast imaging system	Moor Instruments	moorO2FloMeas
Fluorescence microscope	Thermo Fisher Scientific	EVOS™ M7000

Experimental model and study participant details

Animals

Male Sprague-Dawley (SD) rats (7–8 weeks old, 280–320 g) were obtained from Jinan Pengyue Experimental Animal Center (License No. SCXK (Lu) 20220006). All animals were housed in a specific pathogen-free (SPF) environment under controlled conditions (temperature: 22 ± 2°C; humidity: 40–50%; 12-h light/dark cycle), with ad libitum access to food and purified water. Following a 1-week acclimatization period, rats were fasted for 12 h prior to surgery.⁴³ All animal procedures conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of Shandong Second Medical University (Approval No. 2024SDL669).

Middle cerebral artery occlusion (MCAO) model, experimental groups, and drug administration

Rats were anesthetized with intraperitoneal injection of 1% sodium pentobarbital (50 mg/kg). A midline cervical incision was made to expose the left common, internal, and external carotid arteries. The external carotid artery was ligated, and a nylon suture was inserted through the common carotid artery to occlude the internal carotid artery. After 90 min of ischemia, the suture was withdrawn to initiate reperfusion. Successful ischemia was confirmed by ≥ 70% reduction in cerebral blood flow using laser Doppler flowmetry. Sham-operated rats underwent identical procedures without suture insertion.⁴⁴

Animals were randomly divided into four groups (n = 16 per group):

• 1Sham;
• 2MCAO/Reperfusion (MCAO/R);
• 3MCAO/R + SB290157 (20 mg/kg);

• 4 MCAO/R + SB290157 + 2′3′-cGAMP.

SB290157 (Merck, Germany) was dissolved in DMSO to prepare a 100 mg/mL stock solution and diluted with sterile PBS to a final concentration of 20 mg/kg in 100 μL.⁴⁵ SB290157 was administered intravenously via the tail vein 2 h after reperfusion. The 2′3′-cGAMP (500 μg/kg; InvivoGen, France) was administered intranasally 2 h post-reperfusion.³¹ Sham and MCAO/R groups received an equivalent volume of 0.9% saline. All treatments were administered once daily for three consecutive days.

Cell culture

BV2 murine microglial cells (Procell, Wuhan, China) were used as an in vitro microglial model and cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin at 37°C in a humidified incubator with 5% CO₂.

Method details

Laser speckle contrast imaging (LSCI)

Cerebral blood flow was monitored in real time using a laser speckle contrast imaging system (moorO2FloMeas, Moor Instruments). After anesthesia, the scalp was incised to expose the skull. Images were acquired at 1 Hz to assess relative cerebral blood flow (rCBF) during ischemia and reperfusion.⁴⁶

TTC staining and infarct volume quantification

Coronal brain slices were incubated in 2% 2,3,5-triphenyltetrazolium chloride (TTC; Solarbio, China) at 37°C for 30 min.⁴⁷ After staining, slices were photographed, and infarct volume was analyzed using ImageJ software:

$$\text{Infarct}\text{volume}(\%)=\left[\left(\text{volume}\text{of}\text{contralateral}\text{hemisphere}-\text{non}-\text{infarct}\text{volume}\text{of}\text{ipsilateral}\text{hemisphere}\right)/\text{volume}\text{of}\text{contralateral}\text{hemisphere}\right]\times 100\%.$$

Neurological function assessment

Neurological function was assessed using the modified neurological severity score (mNSS) at 24, 48, and 72 h post-reperfusion. The evaluation included sensory, motor, reflex, and balance tests. A score of 0 represents normal function, and 18 indicates maximal neurological deficit. Rats were acclimated to the test environment before each evaluation.⁴⁸

Brain water content

Brain edema was assessed using the wet-dry weight method. On day 3 post-MCAO, brains were dissected into left and right hemispheres. Wet weights were recorded immediately, and dry weights were obtained after overnight dehydration at 100°C.⁴⁶ Water content was calculated as:

$$\text{Brain}\text{water}\text{content}(\%)=\left[\left(\text{wet}\text{weight}-\text{dry}\text{weight}\right)/\text{wet}\text{weight}\right]\times 100\%\text{.}$$

Histological staining

Rats were perfused with 0.9% saline followed by 4% paraformaldehyde via the left ventricle. Brains were harvested, fixed, paraffin-embedded, and coronally sectioned at 5 μm thickness. Coronal sections corresponding to the ipsilateral peri-infarct cortical region (ischemic penumbra) were selected according to the Paxinos and Watson rat brain atlas and subjected to hematoxylin–eosin (HE) and Nissl staining for microscopic analysis.⁴⁹

Oxygen-glucose deprivation/reoxygenation (OGD/R) model

For establishment of the OGD/R model, cells were incubated in glucose-free DMEM under hypoxic conditions (93% N₂, 2% O₂, 5% CO₂) for 3 h, followed by reoxygenation in normoxic complete DMEM for 24 h as previously described.⁵⁰ Treatment groups included Sham, OGD/R, and OGD/R + SB290157 (1 μg/μL).¹⁷

Mitochondrial membrane potential (Δψm) assay

Mitochondrial membrane potential was assessed using the JC-1 detection kit (Beyotime, China). Cells were stained and observed under a fluorescence microscope (EVOSTM M7000, Thermo Fisher Scientific). The ratio of JC-1 monomers to aggregates was quantified using ImageJ software.⁵¹

Immunofluorescence staining

Paraffin-embedded coronal brain sections (5 μm) from the ipsilateral peri-infarct cortical region were deparaffinized, rehydrated, and subjected to antigen retrieval in citrate buffer. After blocking with 5% bovine serum albumin (BSA) for 1 h at room temperature, sections were incubated overnight at 4°C with primary antibodies against STING and Iba1. After washing, the sections were incubated with the corresponding fluorescent secondary antibodies for 1 h at room temperature in the dark. Nuclei were counterstained with DAPI. Fluorescence images were captured using a fluorescence microscope, and STING fluorescence intensity was quantitatively analyzed using ImageJ software.⁵²

Western blotting

Protein was extracted from the ipsilateral peri-infarct cortical region (ischemic penumbra) of rat brains at the indicated time point after reperfusion, according to the Paxinos and Watson rat brain atlas. Tissue samples were homogenized in RIPA buffer (Beyotime, China), and total protein was isolated by centrifugation. Equal amounts of protein were separated by SDS-PAGE and transferred onto PVDF membranes. After blocking, membranes were incubated overnight at 4°C with the following primary antibodies: anti-C3aR (1:2000, Affinity, DF10198), anti-STING (1:2000, Proteintech, 19851-1-AP), anti-GPX4 (1:1000, Abmart, T56959), anti-SLC7A11 (1:1000, Servicebio, GB115276), and anti-β-actin (1:5000, Abways CY1132). After incubation with HRP-conjugated secondary antibodies, protein bands were visualized using ECL substrate (Beyotime, China).⁴⁸

Quantitative real-time PCR (RT-qPCR)

Total RNA was isolated using TRIzol reagent (Thermo Fisher, USA), and cDNA was synthesized with PrimeScript RT Master Mix (Takara). RT-qPCR was performed using TB Green Premix Ex Taq II (Takara), with GAPDH as the internal control.⁴⁹ Primer sequences are listed in Table 1.

Measurement of ROS and 4-HNE

On day 3 post-reperfusion, the ipsilateral peri-infarct cortical region (ischemic penumbra) was rapidly dissected on ice according to the Paxinos and Watson rat brain atlas. Tissue samples were homogenized and centrifuged, and the resulting supernatants were used to determine ROS and 4-HNE levels using commercial assay kits (Affandi, China) according to the manufacturer’s instructions.⁴⁶

Fe²⁺ quantification

Fe²⁺ content in the ipsilateral peri-infarct cortical tissue corresponding to the ischemic penumbra was measured using a commercial assay kit (Nanjing Jiancheng, China). Brain tissues were rapidly dissected on ice, ultrasonically homogenized, and centrifuged, and the supernatant was processed according to the manufacturer’s instructions.⁸

Protein docking analysis

The 3D structures of STING, SLC7A11, and GPX4 were retrieved from the RCSB Protein DataBank. Protein-protein docking was conducted using HADDOCK 2.0 to evaluate STING–SLC7A11 and STING–GPX4 interactions. The docking complexes were visualized using PyMOL software.⁵³

Molecular dynamics simulation and MM/GBSA analysis

To further validate the stability of the predicted protein–protein interactions, 100 ns molecular dynamics (MD) simulations were performed for the STING–SLC7A11 and STING–GPX4 complexes based on the optimal docking conformations. Simulations were carried out using GROMACS 2023.4 with the CHARMM36 force field. Each complex was solvated in a TIP3P water box, neutralized with counterions, and subjected to energy minimization, followed by NVT and NPT equilibration at 300 K and 1 atm.⁵⁴

Production runs were performed for 100 ns with a 2 fs time step. Trajectory analyses included RMSD, RMSF, radius of gyration (Rg), hydrogen bond occupancy, solvent-accessible surface area (SASA), and free energy landscape (FEL). Binding free energies were further calculated using the MM/GBSA method based on the equilibrated trajectories from 90 to 100 ns, including van der Waals, electrostatic, polar solvation, and nonpolar solvation energy components.⁵⁵

Quantification and statistical analysis

All data are presented as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism 10.1.2. Comparisons between two groups were performed using an unpaired Student’s t test. For comparisons among three or more independent groups, one-way ANOVA followed by Holm–Sidak multiple comparisons test was used. For repeated measurements over different time points, such as neurological function scores, two-way repeated-measures ANOVA followed by Holm–Sidak post hoc test was applied. The specific statistical test used for each experiment is indicated in the corresponding figure legends. A p value <0.05 was considered statistically significant.³³