Shikonin alleviates rotenone-induced Parkinson’s disease neuroinflammation by targeting PKM2-mediated glycolytic MG-Hs production

Abstract

Background

In Parkinson’s disease (PD), microglial activation is driven by metabolic reprogramming toward aerobic glycolysis, a shift regulated by pyruvate kinase M2 (PKM2). While the environmental toxin rotenone is a recognized PD risk factor, the precise glycolytic mechanism linking it to microglial neuroinflammation remains unclear, and the therapeutic potential of targeting this axis is largely unexplored.

Purpose

We sought to elucidate the specific glycolytic pathway by which rotenone induces microglial activation and to investigate whether shikonin, a natural PKM2 inhibitor, could attenuate neuroinflammation by targeting this metabolic mechanism.

Methods

Using rotenone (250 nM)-treated BV2 microglia, we assessed glycolytic function (lactate production, glucose consumption) and quantified the formation of methylglyoxal-derived hydroimidazolones (MG-Hs), key pro-inflammatory glycation adducts. NF-κB pathway activation and inflammatory cytokine release were evaluated. The inhibitory effects of shikonin on this cascade were systematically examined.

Results

We identified a novel mechanistic pathway: rotenone promotes PKM2-mediated glycolytic flux, leading to accumulation of the cytotoxic metabolite methylglyoxal (MG) and its derived MG-Hs. These MG-Hs function as critical signaling mediators that directly activate the NF-κB pathway, fueling neuroinflammation. Shikonin effectively disrupted this cascade at its source by inhibiting PKM2, thereby normalizing glycolytic activity, reducing MG-Hs formation, and subsequently suppressing NF-κB activation and the release of pro-inflammatory factors.

Conclusion

This study delineates a complete PKM2-glycolysis-MG-Hs-NF-κB axis as a fundamental mechanism in rotenone-induced neuroinflammation. Our results provide compelling preclinical evidence that shikonin exerts its neuroprotective effects by specifically targeting this metabolic-inflammatory pathway, positioning it as a highly promising disease-modifying therapeutic candidate for PD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-025-02542-z.

Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative disorder pathologically characterized by the selective loss of dopaminergic neurons in the substantia nigra pars compacta and the accumulation of α-synuclein-positive Lewy bodies. Accumulating evidence indicates that microglia-mediated neuroinflammation plays a critical role in PD pathogenesis, wherein activated microglia exacerbate neuronal damage through the sustained release of reactive oxygen species and pro-inflammatory cytokines [1–3]. As the resident immune cells of the central nervous system, microglia undergo dynamic metabolic reprogramming in response to pathological stimuli—shifting from oxidative phosphorylation under physiological conditions to aerobic glycolysis upon activation by signals such as LPS or α-synuclein [4, 5]. This metabolic switch, characterized by increased glucose uptake and lactate production, supports their pro-inflammatory phenotype [4, 5]. A key mediator of this process is pyruvate kinase M2 (PKM2), which enhances glycolytic flux and interfaces with inflammatory pathways such as NF-κB [6, 7]. However, the specific mechanisms by which environmental toxins such as rotenone trigger microglial metabolic reprogramming in PD remain poorly defined.

The rotenone (Rot)-induced PD model is widely employed as it recapitulates key clinical features of the disease [8, 9]. Although rotenone has been shown to activate MAPK signaling and promote NF-κB nuclear translocation in microglia [10], its capacity to directly induce pro-inflammatory cytokine production remains controversial. For example, Klintworth et al. reported that, unlike LPS, rotenone does not elevate TNF-α or IL-1β levels [11], suggesting that its microglial activation may involve alternative or indirect pathways. Resolving this discrepancy is essential for understanding PD etiology and developing targeted anti-inflammatory therapies.

Given the limitations of current symptomatic treatments for PD, there is an urgent need for disease-modifying strategies. Shikonin, a natural naphthoquinone derived from Lithospermum erythrorhizon, has garnered attention as a multi-target neuroprotective agent with antioxidant, anti-inflammatory, and anti-apoptotic properties. Experimental studies have demonstrated that shikonin attenuates oxidative stress in MPTP-induced PD models by enhancing SOD and GSH activities while reducing MDA levels [12]. It also suppresses neuroinflammation by downregulating pro-inflammatory cytokines such as TNF-α, IL-1β, and IL-6, inhibiting astrocyte activation, and blocking NF-κB nuclear translocation [12]. Moreover, shikonin modulates key signaling pathways by promoting Akt and ERK phosphorylation while suppressing JNK and NF-κB activation, thereby balancing neuronal survival and inflammatory responses [12]. Its anti-apoptotic effects are mediated through the reduction of caspase-3 and caspase-9 activities, regulation of Bcl-2/Bax ratio, and interaction with pro-apoptotic protein BAD [12, 13]. Recently, shikonin was shown to inhibit pyruvate kinase M2 (PKM2), thereby reducing lactate accumulation and contributing to metabolic regulation in PD [14].It has been shown to mitigate oxidative stress in PD models by enhancing antioxidant enzyme activities and suppressing lipid peroxidation [12]. Additionally, shikonin inhibits neuroinflammation through downregulation of pro-inflammatory cytokines and blockade of NF-κB signaling [12]. Recent studies further indicate that shikonin acts as a PKM2 inhibitor, implicating a role in metabolic regulation [14]. These multi-faceted actions position shikonin as a promising candidate for intervening in the metabolic-inflammatory axis of PD.

In this study, we investigate the glycolytic mechanisms underlying rotenone-induced neuroinflammation in BV2 microglia. We demonstrate that rotenone enhances glycolytic flux, leading to the accumulation of methylglyoxal-derived hydroimidazolones (MG-Hs)—reactive adducts formed from glycolytic byproducts [15–17]. We further show that shikonin, a known PKM2 inhibitor [18], attenuates rotenone-driven neuroinflammation through dual mechanisms: suppressing glycolysis and reducing MG-Hs formation. These effects correlate with significant inhibition of the microglial pro-inflammatory cascade. Our findings reveal a previously unrecognized glycolysis-mediated pathway in rotenone-induced neuroinflammation and provide mechanistic support for shikonin as a potential therapeutic agent targeting the metabolic-inflammatory network in Parkinson’s disease.

Materials and methods

Cell culture

The BV2 microglial and PC12 neuronal cell lines were procured from the American Type Culture Collection (ATCC, Manassas, VA, USA). BV2 cells were maintained in high-glucose Dulbecco’s Modified Eagle Medium (HDMEM), while PC12 cells were cultured in RPMI-1640 medium. Both media were supplemented with 10% heat-inactivated fetal bovine serum (FBS), 50 U/mL penicillin, and 50 µg/mL streptomycin (Gibco/Thermo Fisher Scientific). Cells were maintained at 37 °C in a humidified atmosphere containing 5% CO₂. All experiments were conducted using cells between passage numbers 4 and 15 to ensure phenotypic stability.

Cell differentiation

PC12 cells were differentiated in RPMI-1640 medium containing L-glutamine and supplemented with 1% donor horse serum (DHS; Thermo Fisher Scientific), 100 ng/mL nerve growth factor (NGF; Sigma-Aldrich, Cat# N2513), 50 µg/mL gentamicin, and 0.25 µg/mL amphotericin B. The differentiation medium was replaced every 48 h throughout the experimental period [19], with differentiation status routinely verified by assessing characteristic neurite outgrowth morphology prior to experiments.

Cell viability assay

BV2 microglial cells were seeded in 96-well plates at a density of 8,000 cells/well and cultured for 24 h to allow adherence. Cells were then treated with rotenone (Rot; Sigma-Aldrich, Cat# R8875), 2-deoxy-D-glucose (2-DG; Sigma-Aldrich, Cat# 25972-M), aminoguanidine (AG; Sigma-Aldrich, Cat# 396494), methylglyoxal (MG; Sigma-Aldrich, Cat# M-0252), or shikonin (Selleck, Cat# HY-N0822) at specified concentrations for 24 h. In parallel experiments, PC12 cells were plated in 96-well plates and allowed to adhere for 24 h prior to differentiation with 100 ng/mL nerve growth factor (NGF) for 72 h. Differentiated PC12 cells were then exposed to conditioned media from treated BV2 cells for 24 h. Cell viability for both cell types was determined using the sulforhodamine B (SRB; Sigma-Aldrich, Cat# 230162) colorimetric assay, with viability expressed as a percentage relative to vehicle-treated controls (absorbance of treated cells/absorbance of DMSO controls × 100%). All treatments were performed in triplicate wells and independently replicated in three separate experiments.

Preparation and application of Microglial-Conditioned Medium (MCM)

After intervention, the MCM was collected and centrifuged at 12,000 × g for 10 min at 4 °C to remove cellular debris. To eliminate any residual rotenone and other small-molecule compounds, the supernatant was further processed using ultrafiltration centrifugal devices with a 3-kDa molecular weight cut-off (UFC5003, Merck Millipore) according to the manufacturer’s instructions. The resulting drug-free CM was aliquoted and stored at −80 °C until use.

Metabolite analysis

The extracellular lactate and glucose concentrations in cell culture supernatants were quantitatively determined using commercial colorimetric assay kits (lactate: A019-2-1; glucose: F006-1-1; Nanjing Jiancheng Bioengineering Institute, China) following the manufacturer’s standardized protocols, with absorbance measurements performed at the specified wavelengths for each assay. Glucose consumption was calculated by subtracting the glucose concentration in the conditioned medium from that in fresh, unused medium after normalizing to total cellular protein content.

Measurement of TNF-α, IL-1β and IL-6

Following experimental treatments, cell culture supernatants were collected and centrifuged at 1,000 × g for 10 min to remove cellular debris. Subsequently, the concentrations of TNF-α (ml064292; Shanghai Meilian Biotechnology Co., Ltd.), IL-1β (ml106733; Shanghai Meilian Biotechnology Co., Ltd.), and IL-6 (ml064292; Shanghai Meilian Biotechnology Co., Ltd.) were measured using commercial enzyme-linked immunosorbent assay (ELISA) kits, strictly in accordance with the manufacturer’s instructions. Absorbance was read at 450 nm using a microplate reader.

Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling (TUNEL) staining

After drug treatment, the cells present in the supernatant were pooled with trypsinized adherent cells, centrifuged, and resuspended in PBS. Aliquots of this suspensionwere deposited on slides using a cytospin centrifuge. The cells were fixed with ice-cold 4% paraformaldehyde for 10 min and washed twice with PBS. They were then subjected to TUNEL (40308ES20; Yeasen Biotechnology Co., Ltd.) according to the manufacturer’s instructions. Next, 4 0,6-diamidino-2-phenylindole (DAPI) (1/1000 in PBS) was added to the cells for 5 min at 25℃. Images were obtained using a fluorescence microscope (Nikon ECLIPSE 50i).

Assessment of apoptosis by flow cytometry

PC12 cell apoptosis was quantitatively analyzed by flow cytometry using a commercial Annexin V-FITC/PI apoptosis detection kit (C1062L; Beyotime Institute of Biotechnology). Following experimental treatments, cells were harvested, washed twice with cold phosphate-buffered saline (PBS), and resuspended in 1× binding buffer at a density of 1 × 10^6 cells/mL. Cell suspensions were then dual-stained with fluorescein isothiocyanate (FITC)-conjugated Annexin V and propidium iodide (PI) for 15 min at room temperature in the dark, according to the manufacturer’s protocol. Samples were immediately analyzed on a Beckman CytoFLEX LX flow cytometer, with fluorescence signals collected through appropriate optical channels (FITC: 488 nm excitation/525 nm emission; PI: 535 nm excitation/617 nm emission). Data acquisition and analysis were performed using CytExpert software (v2.4; Beckman Coulter), with apoptotic populations identified based on established Annexin V/PI gating strategies.

PKM2 knockdown

BV2 cells were transfected with PKM2 siRNA^1# (5’-CUGGCAUCAUUUGUACCAUTT-3’), PKM2 siRNA^2# (5’-CAGAGACCAUCAAGAAUGUTT-3’), and PKM2 siRNA^3# (5’-CAUGCUGUCUGGAGAAACATT − 3’) (GenePharma, Shanghai, China) using UltraFection 3.0 (FXP135-020; Beijing 4 A Biotech Co., Ltd., Beijing, China) for 24 h, and then analyzed by Western blotting (Fig. S6A, see Fig. S13C for full immunoblots).

Plasmid transfection

For PKM2 overexpression, BV2 cells at 60%–70% confluence were transfected with either a control vector (pCDNA3.1-EGFP-3xLinker-MCS) or an PKM2 overexpression construct (pCMV-EGFP-Pkm(mouse)−2-Neo) using UltraFection 3.0 transfection reagent (Elji Biotechnology). The DNA- lipid complex (1:2DNA- to- reagent ratio) was incubated with cells in serum- free medium for 6 h at 37 °C, followed by replacement with complete growth medium. Transfection efficiency was assessed 24 h post- transfection via Western blotting (Fig. S5A, see Fig. S13A for full immunoblots).

Western blotting

Following experimental treatments, cells were lysed in RIPA buffer (Cell Signaling Technology, Beverly, MA, USA) supplemented with protease and phosphatase inhibitors, and protein concentrations were determined using the Bradford assay (Bio-Rad Laboratories, Hercules, CA, USA). Equal protein amounts (20 µg per lane) were resolved by SDS-PAGE and subsequently transferred to polyvinylidene difluoride (PVDF) membranes. After blocking with 5% non-fat milk in Tris-buffered saline containing 0.1% Tween-20 (TBST) for 1 h at room temperature, membranes were incubated overnight at 4 °C with the following primary antibodies: PFKFB3 (1:1000, 13763-1-AP, Proteintech), PKM2 (1:1000, 15822-1-AP, Proteintech), LDHA (1:1000, 19987-1-AP, Proteintech), MG-Hs (1:500, STA-011, Cell Biolabs), phospho-NF-κB p65 (1:1000, 3033 S, Cell Signaling Technology), NF-κB p65 (1:1000, 80979-1-RR, Proteintech), cleaved PARP (1:1000, #9542, Cell Signaling Technology), and β-actin (1:5000, 20536-1-AP, Proteintech) as a loading control. Following three washes with TBST, membranes were incubated with appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5000) for 1 h at room temperature. Protein bands were visualized using enhanced chemiluminescence (Ultra ECL Western Blotting Detection Reagent) and imaged using a ChemiDoc MP Imaging System (Bio-Rad). Band intensities were quantified using ImageJ software (v1.46r, NIH), with target protein expression levels normalized to β-actin.

Statistical analyses

R software (version 3.6.3) was used to analyze the data, which were expressed as the mean ± SD, by t-test or one-way ANOVA. A P value less than 0.05 was defined as a significant difference and expressed as *p < 0.05, **p < 0.01, and ***p < 0.001.

Results

Rot enhanced cell damage and inflammation in BV2 microglia

To establish an in vitro neuroinflammatory model, we first determined the optimal concentration of Rot for BV2 microglial activation. Cell viability assays using the SRB method revealed a concentration-dependent decrease in BV2 cell survival following Rot treatment, with a calculated half-maximal inhibitory concentration (IC₅₀) of approximately 250 nM (Supplementary Fig. 1 and Fig. 1A). This concentration was subsequently selected for subsequent experiments to induce microglial activation while maintaining measurable cell viability.

Fig. 1.

Rotenone induces cytotoxic and pro-inflammatory responses in BV2 microglia. A Cell viability was assessed using the sulforhodamine B (SRB) assay after 24 h of treatment with 250 nM rotenone (Rot). B Extracellular levels of TNF-α, IL-1β, and IL-6 were measured by ELISA in culture supernatants from BV2 microglia exposed to 250 nM Rot for 24 h. Data are presented as mean ± SD (n = 5 biologically independent experiments). ^***P < 0.001

Given the established correlation between inflammation and neurodegenerative disease progression [20], we next evaluated the secretion of pro-inflammatory factors in Rot-stimulated BV2 microglia. Enzyme-linked immunosorbent assay revealed a marked increase in extracellular levels of IL-1β and IL-6 in Rot-treated cells compared to vehicle controls (Fig. 1B), indicating that Rot promotes pro-inflammatory activation of microglia. Although no statistically significant change was observed in extracellular TNF-α levels following Rot treatment (Fig. 1B), the overall findings demonstrate that Rot elicits both cytotoxic and inflammatory responses in BV2 microglia, supporting its relevance as an in vitro model of neuroinflammation.

Rot caused inflammation by activating glycolysis in BV2 microglia

Microglia exhibit remarkable plasticity, capable of adopting distinct polarization states including resting (M0), pro-inflammatory (M1), and anti-inflammatory (M2) phenotypes [21]. In PD, activated microglia predominantly assume the M1 state, characterized by neurotoxic effects and a metabolic shift toward hyperactivated glycolysis [22, 23]. Based on preliminary observations that a 24-hour intervention was optimal for Rot-induced glycolytic activation in BV2 cells (Supplementary Fig. 2), a systematic evaluation of glycolysis was conducted following this treatment period. The results demonstrated that Rot significantly enhanced glycolytic activity in BV2 microglia. This was evidenced by increased glucose uptake, lactate production, lactate-to-glucose ratio, and upregulation of key glycolytic enzymes (Fig. 2A–C), aligning with the metabolic profile associated with neuroinflammation.

Fig. 2.

Rotenone promotes inflammatory responses via glycolytic activation in BV-2 microglia. A Representative immunoblots of glycolytic enzymes (PFKFB3, PKM2, LDHA) and β-actin (loading control) after 24 h treatment with 250 nM Rot. B Quantitative analysis of protein levels in (A) normalized to β-actin (n = 5). C Glycolytic parameters (glucose consumption, lactate production, lactate/glucose ratio) measured after Rot treatment (n = 5). D Representative immunoblots of MG-Hs and phospho-NF-κB p65 (p-p65) with β-actin. E Quantification of MG-Hs and p-p65 levels from (D) normalized to β-actin (n = 5). F Representative immunoblots of NF-κB p65 in cytoplasmic (β-actin control) and nuclear (Lamin B1 control) fractions. G Quantitative analysis of cytoplasmic p65 (normalized to β-actin) and nuclear p65 (normalized to Lamin B1) (n = 5). H Glycolytic parameters following pretreatment with 20 µM 2-DG for 2 h before Rot exposure (n = 5). I Representative immunoblots of MG-Hs and p-p65 with β-actin after 2-DG and Rot treatment. J Quantification of protein levels in (I) normalized to β-actin (n = 5). K Representative immunoblots of NF-κB p65 in cytoplasmic and nuclear fractions after intervention. L Quantitative analysis of p65 levels in both fractions (n = 5). M Extracellular levels of TNF-α, IL-1β, and IL-6 measured by ELISA (n = 5). Data are presented as mean ± SD; ^*P < 0.05, ^**P < 0.01, ^***P < 0.001

The glycolytic byproduct methylglyoxal can non-enzymatically modify arginine residues to form methylglyoxal-derived hydroimidazolones (MG-Hs), which are known activators of NF-κB-mediated inflammation [24–26]. In Rot-treated BV2 microglia, we observed a concurrent upregulation of MG-Hs (Fig. 2D, E) and activation of NF-κB p65, as indicated by its increased nuclear translocation (Fig. 2D–G), suggesting a potential mechanistic link between glycolytic activation and inflammatory signaling. To test this hypothesis, we used the glycolytic inhibitor 2-deoxy-D-glucose (2-DG) at a concentration (20 µM) that effectively suppressed Rot-induced glycolysis (Fig. 2H) without affecting cell viability (Supplementary Fig. 3 A). Treatment with this concentration of 2-DG alone did not significantly alter basal glycolysis in BV2 cells (Supplementary Fig. 3B–D), confirming its safety under these conditions. Notably, 2-DG treatment attenuated the Rot-induced increases in MG-Hs levels (Fig. 2I, J), NF-κB p65 activation (Fig. 2I–L), and subsequent secretion of IL-1β and IL-6 (Fig. 2M). These results support the conclusion that Rot promotes NF-κB-mediated inflammation in microglia through a glycolysis-dependent mechanism involving MG-Hs accumulation.

Rot caused inflammation via triggering the accumulation of MG-Hs in BV2 microglia

To establish the specific role of MG-Hs in NF-κB p65 activation, we employed AG, a potent MG scavenger [27]. Remarkably, AG treatment effectively attenuated MG-Hs accumulation and activation of NF-κB p65 in Rot-stimulated BV2 microglia (Fig. 3A-D), while having no discernible effect on glycolytic activity (Supplementary Fig. 4A-C). Consistent with these findings, AG significantly suppressed Rot-induced IL-1β, and IL-6 secretion (Fig. 3E), demonstrating that MG derivatives, rather than other glycolysis-related factors, primarily mediate the inflammatory response.

Fig. 3.

MG-Hs mediate rotenone-induced inflammatory responses in BV2 microglia. A Representative immunoblots of MG-Hs and phospho-NF-κB p65 (p-p65) in cells pretreated with 100 µM AG for 2 h followed by 250 nM Rot for 24 h (β-actin loading control). B Quantification of MG-Hs and p-p65 levels normalized to β-actin (n = 5). C Representative immunoblots of NF-κB p65 in cytoplasmic (β-actin control) and nuclear (Lamin B1 control) fractions. D Quantification of cytoplasmic (normalized to β-actin) and nuclear (normalized to Lamin B1) p65 levels (n = 5). E Extracellular levels of TNF-α, IL-1β, and IL-6 measured by ELISA (n = 5). F Representative immunoblots of MG-Hs and p-p65 in cells pretreated with 25 µM MG for 2 h followed by 250 nM Rot for 24 h (β-actin loading control). G Quantification of MG-Hs and p-p65 levels normalized to β-actin (n = 5). H Representative immunoblots of NF-κB p65 in cytoplasmic and nuclear fractions. I Quantification of p65 levels in both fractions (n = 5). J Extracellular cytokine levels (TNF-α, IL-1β, IL-6) measured by ELISA (n = 5). Data are presented as mean ± SD; ^*P < 0.05, ^**P < 0.01, ^***P < 0.001

To further substantiate this mechanism, we examined the consequences of exogenous MG administration. Complementing our pharmacological inhibition studies, MG supplementation exacerbated Rot-induced MG-Hs formation, NF-κB p65 activation, IL-1β and IL-6 production (Fig. 3F-J). Collectively, these results establish a clear causal relationship wherein Rot-induced glycolytic activation leads to MG-Hs accumulation, which in turn drives NF-κB-mediated neuroinflammation in microglia.

Shikonin inhibited the hyperglycolysis and MG-Hs in Rot-induced BV2 microglia

Our findings demonstrate that Rot-induced neuroinflammation in BV2 microglia is primarily mediated through glycolytic activation. This suggests that pharmacological modulation of glycolysis could represent a viable therapeutic strategy. Shikonin, a naturally occurring naphthoquinone derivative isolated from Lithospermum erythrorhizon (Boraginaceae), has been identified as a selective inhibitor of PKM2 isoform [28], known to shift cellular metabolism from glycolysis to oxidative phosphorylation [29, 30].

In Rot-stimulated BV2 microglia, shikonin treatment effectively suppressed hyperactivated glycolysis, as evidenced by reduced expression of key glycolytic enzymes (PFKFB3, PKM2, and LDHA; Fig. 4 A-B). This metabolic modulation was further confirmed by decreased glucose consumption, lactate production, lactate-to-glucose ratio, and MG-Hs accumulation (Fig. 4 C-G). These results collectively demonstrate that shikonin attenuates Rot-induced metabolic dysregulation in microglia by inhibiting PKM2-mediated glycolysis and subsequent formation of pro-inflammatory MG-Hs.

Fig. 4.

Shikonin attenuates rotenone-induced hyperglycolysis and MG-Hs formation in BV2 microglia. A Representative immunoblots of glycolytic enzymes (PFKFB3, PKM2, LDHA) and β-actin (loading control) following pretreatment with 50 nM shikonin (SK) for 1 h and subsequent treatment with 250 nM rotenone (Rot) for 24 h. B Quantitative analysis of glycolytic enzyme expression normalized to β-actin (n = 5). C-E Metabolic parameters measured by colorimetric assays: (C) glucose consumption, (D) lactate production, and (E) lactate-to-glucose ratio (n = 5). F Representative immunoblots of MG-Hs with β-actin loading control.G Quantitative analysis of MG-Hs levels normalized to β-actin (n = 5). Data represent mean ± SD from five biologically independent experiments. Statistical significance was determined by one-way ANOVA with Tukey's post-hoc test ^***P < 0.001

Shikonin inhibited inflammation in Rot-induced BV2 microglia via PKM2

To further characterize the anti-inflammatory effects of shikonin, we examined its impact on NF-κB activation and pro-inflammatory cytokine production in Rot-stimulated BV2 microglia. Consistent with its metabolic regulatory role, shikonin significantly attenuated both NF-κB p65 phosphorylation and the secretion of IL-1β and IL-6 (Fig. 5 A–C). To determine whether this effect is mediated through the inhibition of PKM2, we performed gain- and loss-of-function experiments. Notably, overexpression of PKM2 partially reversed shikonin’s inhibitory effects on Rot-induced MG-Hs accumulation (Supplementary Fig. 5B-C), NF-κB activation (Fig. 5D–E), and inflammatory cytokine release (Fig. 5 F). Conversely, siRNA-mediated silencing of PKM2 alone markedly attenuated these inflammatory responses (Supplementary Fig. 6B-C). Together, these results support the conclusion that PKM2 inhibition alleviates Rot-triggered neuroinflammation and that shikonin exerts its anti-inflammatory effects primarily by targeting PKM2. Collectively, our findings demonstrate that Rot promotes neuroinflammation via a glycolysis–MG-Hs–NF-κB axis, while shikonin protects by suppressing this cascade at the level of PKM2-mediated glycolysis.

Fig. 5.

Shikonin attenuates rotenone-induced neuroinflammation in BV2 microglia by inhibiting PKM2-mediated NF-κB signaling. A Representative immunoblots of phospho-NF-κB p65 (p-p65) and β-actin in cells pretreated with 50 nM SK for 1 h followed by 250 nM rotenone (Rot) for 24 h. B Quantitative analysis of p-p65 levels normalized to β-actin (n = 5). C Extracellular TNF-α, IL-1β, and IL-6 levels measured by ELISA (n = 5). D Representative immunoblots of NF-κB p65 in cytoplasmic (β-actin control) and nuclear. (Lamin B1 control) fractions from cells with control (OE-Con) or PKM2 overexpression (OE-PKM2), after SK and Rot treatments. E Quantification of NF-κB p65 levels in cytoplasmic and nuclear fractions (n = 5). F Cytokine secretion under corresponding conditions (n = 5). Data are shown as mean ± SD from five independent experiments. ^***P < 0.001

Neuroprotective effects of Shikonin against microglia-mediated neuronal injury via PKM2 Inhibition

Given the established correlation between pro-inflammatory factors and dopaminergic neuron degeneration in PD pathogenesis [20], we investigated whether shikonin could attenuate microglia-mediated neurotoxicity using an in vitro model in which NGF-differentiated PC12 neurons were exposed to conditioned media from Rot-treated BV2 microglia (Fig. 6 A). Neuronal viability was significantly higher in PC12 cells treated with conditioned media from shikonin-pretreated, Rot-stimulated BV2 cells (Rot + SK-MCM) than in those exposed to media from Rot-treated microglia alone (Rot-MCM) (Fig. 6B). This protective effect was further supported by reduced cleavage of PARP, an indicator of apoptosis, in the Rot + SK-MCM group (Fig .6 C–D). Morphological evaluation using TUNEL staining showed fewer TUNEL-positive neurons after Rot + SK-MCM treatment compared to Rot-MCM (Fig. 6E). Flow cytometric analysis confirmed a significant decrease in apoptotic cells under the Rot + SK-MCM condition (Fig. 6 F–G). Together, these results indicate that shikonin protects neurons by suppressing microglial-mediated neurotoxicity resulting from Rot-induced activation.

Fig. 6.

Shikonin attenuates microglia-mediated neuronal injury in a rotenone-induced Parkinson’s disease model. A Experimental design schematic: PC12 neurons differentiated with NGF were treated for 24 h with conditioned medium from BV2 microglia exposed to either: (i) 250 nM rotenone (Rot-MCM) or (ii) 250 nM rotenone plus 50 nM shikonin (Rot + SK-MCM). B Neuronal viability assessed by SRB assay (n = 5). C Representative immunoblots of cleaved PARP apoptosis marker with β-actin loading control. D Quantitative analysis of cleaved PARP expression normalized to β-actin (n = 5). E TUNEL staining (red, scale bar = 50 μm). F Representative flow cytometry dot plots of annexin V/PI staining. G Quantification of viable (PI⁻/annexin V⁻) and apoptotic (PI⁺/annexin V⁺ and PI⁻/annexin V⁺) cell populations (n = 5). Data represent mean ± SD from five biologically independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001

To examine whether this neuroprotection involves PKM2 inhibition, we performed genetic manipulation experiments. Overexpression of PKM2 partially reversed the beneficial effect of shikonin on Rot-induced microglial neurotoxicity (Supplementary Fig. 7 A). Conditioned medium from BV2 cells with normal PKM2 expression following Rot treatment induced pronounced PC12 cell damage and apoptosis, as observed in TUNEL staining (Supplementary Fig. 7B). In contrast, conditioned medium from PKM2-silenced BV2 cells exposed to Rot caused markedly less damage to PC12 neurons (Supplementary Fig. 7B). Together with the earlier finding that PKM2 silencing suppresses Rot-induced inflammatory responses in BV2 cells, these results demonstrate that PKM2 inhibition alleviates microglial-mediated neurotoxicity, and further support the conclusion that shikonin acts through PKM2 to exert its neuroprotective effects (Fig. 7).

Fig. 7.

Proposed mechanism of shikonin-mediated neuroprotection against rotenone-induced neuroinflammation. The schematic illustrates shikonin’s dual action in attenuating microglia-mediated neuroinflammation through: (1) direct inhibition of PKM2, suppressing glycolytic flux and subsequent MG-Hs formation; and (2) downstream reduction of NF-κB p65 activation and pro-inflammatory cytokine release. This metabolic intervention disrupts the vicious cycle between microglial hyperactivation and neuronal injury in Parkinson’s disease models

Discussion

Growing evidence underscores the critical role of neuroinflammation in PD pathogenesis, with microglial hyperactivation being a key driver of disease progression. While the inflammatory phenotype of microglia is well-recognized, its intimate link to metabolic reprogramming has only recently been appreciated. Our study reveals a novel metabolic-inflammatory axis in Rot-induced microglial activation, demonstrating that Rot promotes glycolysis-dependent accumulation of MG-Hs, which subsequently activate NF-κB-mediated inflammatory signaling (Fig. 7). This pathway diverges from the classical LPS-induced activation model and provides a mechanistic explanation for prior controversies. Importantly, we identify shikonin, a natural PKM2 inhibitor, as an effective modulator of this pathway, capable of attenuating both glycolytic flux and NF-κB activation by reducing MG-Hs formation. These findings not only elucidate a previously unrecognized mechanism linking metabolic reprogramming to neuroinflammation but also highlight the therapeutic potential of targeting the PKM2–glycolysis–MG-Hs axis.

A central controversy in the field concerns whether Rot directly activates microglia to drive PD pathogenesis. Earlier work by Klintworth et al. demonstrated that, unlike LPS, Rot treatment failed to elevate TNF-α or IL-1β, suggesting an indirect activation mechanism [11]. In contrast, Gao et al. reported that Rot activates MAPK signaling and induces nuclear translocation of NF-κB in microglia [10]. Our findings help resolve this discrepancy by identifying critical experimental variables, particularly concentration dependency. While previous studies using lower concentrations (e.g., 50 nM) showed minimal inflammatory response, our application of 250 nM Rot resulted in a marked increase in inflammatory factor secretion and NF-κB p65 activation, unequivocally demonstrating a dose-dependent direct activation of microglia. This concentration-dependent effect underscores the importance of experimental parameters in interpreting Rot’s actions and aligns with the graded nature of metabolic stress responses.

Beyond reconciling existing controversies, our study provides a significant advance by delineating the specific metabolic sequence underlying Rot-induced neuroinflammation. We establish that Rot-enhanced glycolytic flux leads to the accumulation of MG, which forms MG-Hs adducts that potently activate NF-κB p65 signaling. This glycolysis–MG-Hs–NF-κB axis represents a previously unidentified pathway that distinguishes Rot-induced neuroinflammation from other triggers. While the pro-inflammatory shift to glycolysis is a recognized hallmark of M1-polarized microglia, the causative role of MG-Hs in directly activating NF-κB is a novel insight. This mechanism expands upon the established understanding of PKM2’s role in glycolysis [6] by introducing MG-Hs as a critical downstream effector linking metabolic dysregulation to inflammatory signaling.

Notwithstanding these findings, our study has certain limitations that warrant consideration. First, the relatively low concentration of 2-DG (20 µM) found effective in our BV2 model, compared to the millimolar ranges often cited, may reflect cell type-specific and metabolic context-dependent sensitivity, which, while justified by our viability and glycolytic readouts, suggests a need for caution in extrapolating this concentration universally. Second, although the use of exogenous MG supplementation robustly supported the specific role of MG-Hs, the known multi-target nature of the scavenger aminoguanidine (AG) means we cannot fully discount potential contributions from its iNOS inhibitory activity. The inclusion of a second, structurally distinct carbonyl scavenger in future work would further solidify the conclusion. Third, while our gain- and loss-of-function experiments genetically establish PKM2’s central role in this cascade, the use of a specific PKM2 activator (e.g., TEPP-46) or a shikonin-insensitive PKM2 mutant would be required to definitively rule out any pleiotropic or redox-related off-target effects of shikonin. Finally, while our evidence for NF-κB activation—encompassing phosphorylation, nuclear translocation, and downstream cytokine production—is compelling, direct measurement of transcriptional activity via a NF-κB-reporter assay or analysis of classical NF-κB target genes like iNOS and COX-2 would provide an even more comprehensive view of the signaling linkage.

Despite these limitations, the therapeutic implications of targeting this axis are substantial. Our demonstration that shikonin suppresses both MG-Hs accumulation and NF-κB activation through PKM2 inhibition positions metabolic intervention as a viable strategy. This approach is consistent with the growing recognition of microglial metabolic reprogramming as a therapeutic target, as evidenced by other strategies such as targeting glucose transporters with levistilide A or activating AMPK with metformin [31, 32]. However, our work specifically identifies the PKM2–MG-Hs node as a druggable pathway within this framework. The multi-faceted actions of shikonin, coupled with its efficacy in normalizing this novel pathway, strongly support its potential as a disease-modifying agent for PD and warrant further translational investigation.

Conclusion

In conclusion, our study elucidates a novel glycolysis–MG-Hs–NF-κB signaling axis that directly links rotenone-induced metabolic reprogramming to microglial neuroinflammation. By demonstrating that shikonin, a PKM2 inhibitor, effectively attenuates this cascade through dual suppression of glycolytic flux and MG-Hs formation, we not only resolve prior controversies regarding rotenone’s direct inflammatory effects but also establish the therapeutic relevance of targeting microglial metabolism. These findings position shikonin as a promising disease-modifying candidate and underscore the broader potential of modulating the metabolic-inflammatory axis in Parkinson’s disease.

Authors’ contributions

Ya Zhao: Conceptualization, Funding acquisition, Methodology, Project administration, Writing – original draft; Dan Wang: Data curation, Formal analysis, Funding acquisition, Investigation, Methodology; Dan Mu: Visualization, Writing – review & editing; Lang Qu: Funding acquisition, Resources; Rong Li: Funding acquisition, Writing – review & editing.

Funding

This work was supported by Sichuan Science and Technology Program (2024YFFK0169), the Natural Science Foundation of Sichuan Province (2024NSFSC1731, 2025ZNSFSC0894), the Science and Technology Program of Shaanxi Province (2024SF-YBXM-324), the Science and technology project of Sichuan Provincial Health Commission (24WXXT13), the Open Fund of Key Laboratory of Preclinical Study for New Drugs of Gansu Province (GSKFKT-2301) and the Doctoral Scientific Research Fund of North Sichuan Medical College (CBY23-QDA05, CBY24-QDA01, CBY24-QDA03, CBY24-QDA05).

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.