SIRT3/AARS2 regulates SOD2 lactylation to determine neuronal fate in TBI

Abstract

Lactate accumulation is strongly associated with poor neurological outcomes in traumatic brain injury (TBI), creating a “lactate paradox” given its role as an energy substrate in the early stages of trauma. Imbalanced reactive oxygen species (ROS) act as cell injury factors throughout the pathological progression of TBI. This study aims to elucidate the key mechanisms connecting dysregulated lactate metabolism and cellular damage. We discovered that the high-lactate environment induced by TBI drives lysine lactylation of the mitochondrial antioxidant enzyme superoxide dismutase 2 (SOD2), inhibiting its enzymatic activity and leading to mitochondrial ROS (mtROS) accumulation. Mechanistically, aminoacyl-tRNA synthetase 2 (AARS2) and NAD⁺-dependent deacetylase sirtuin 3 (SIRT3) coordinate SOD2 lactylation through “resident sensor-writer” and “dynamic patrol-eraser” modes, respectively. Proteomic analysis revealed that SOD2 lactylation triggers a reprogramming of its interaction network, shifting its interactome away from proteins involved in energy metabolism and toward those associated with proteostasis. In a mouse model of TBI, activating SIRT3 reversed SOD2 lactylation, restored its enzymatic function, and reduced neuronal apoptosis in the injured area. This study clarifies how AARS2/SIRT3-regulated SOD2 lactylation influences neuronal fate, providing potential targets for treating secondary injury in TBI.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12035-026-05931-8.

Introduction

Traumatic brain injury (TBI) remains a leading cause of global mortality and disability, with severe cases lacking effective pharmacological interventions [1]. While primary mechanical injury is irreversible, the secondary injury cascade—characterized by neuroinflammation, blood–brain barrier disruption, and metabolic disturbances—offers a critical therapeutic window [2–4]. Within this complex pathological network, the loss of intracellular redox homeostasis, particularly the profound accumulation of mitochondrial reactive oxygen species (mtROS), acts as a primary driver of secondary damage [5, 6]. Following TBI, oxidative stress rapidly overwhelms the endogenous antioxidant defense system, leading to mitochondrial depolarization, lipid peroxidation, and ultimately neuronal apoptosis [7, 8]. Consequently, re-establishing mtROS homeostasis, rather than indiscriminately scavenging ROS, has emerged as a pivotal strategy for neuroprotection and improving functional outcomes [9–11].

In the pathological landscape of TBI, lactate has long been recognized as a highly debated metabolic player. Traditionally, the astrocyte-neuron lactate shuttle (ANLS) paradigm posits that early TBI-induced lactate accumulation serves as a vital alternative energy substrate, compensating for neuronal energetic deficits [12–14]. However, this seemingly benign compensatory mechanism sharply contradicts extensive clinical observations: a surge in brain or cerebrospinal fluid (CSF) lactate during the acute phase of TBI is consistently identified as one of the strongest independent predictors of poor neurological prognosis and increased mortality [15, 16]. This striking lactate paradox—where an ostensibly protective fuel source correlates positively with neurotoxicity—strongly implies that at pathologically high concentrations, lactate exerts a more deleterious, non-metabolic role in exacerbating TBI outcomes.

Recent advances in high-throughput proteomics have revolutionized our understanding of lactate, elevating it from a simple metabolic end-product to an active signaling molecule via a novel post-translational modification (PTM): lysine lactylation [17, 18]. Given the metabolic reprogramming and profound lactate accumulation in the TBI microenvironment, we hypothesize that this metabolite surge drives widespread protein lactylation, linking metabolic crisis directly to organelle dysfunction [19]. As the first line of mitochondrial antioxidant defense, superoxide dismutase 2 (SOD2) is a prime candidate for such regulation. SOD2 activity is highly sensitive to post-translational modifications (PTMs), most notably acetylation, which inhibits its capacity to clear superoxide anions—a process reversibly regulated by the mitochondrial deacetylase SIRT3 [20–23]. We posit that under the overwhelming lactate surge in TBI, lactylation may compete for these critical regulatory sites on SOD2, neutralizing its protective function, while SIRT3 might dually serve as a de-lactylase (eraser) to rescue this aberrant modification [24–26].

Based on this logical framework, our study proposes a core scientific hypothesis: TBI-induced high lactate concentrations directly inhibit SOD2 enzymatic activity by driving its aberrant lactylation, thereby crippling the mtROS defense system and precipitating neuronal apoptosis. To unbiasedly map the dynamic regulatory network of SOD2 under TBI pathology and avoid the spatiotemporal limitations of traditional affinity purification, we employed APEX2-mediated in vivo proximity labeling [27, 28]. By reconstructing the mitochondrial SOD2 interactome, we aimed to identify the upstream regulators of SOD2 lactylation. Unraveling this novel lactate-SOD2-ROS regulatory axis will not only resolve the long-standing lactate paradox but also provide precise molecular targets for mitigating TBI-induced secondary injury.

Methods and Materials

Animal Experiment Ethical Statement

Six-week-old male C57BL/6J mice were used in this study, purchased from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and supplied by the local distributor Shudaqiang (Chengdu) Biotechnology Co., Ltd. (Chengdu, China). All mice were housed at the Experimental Animal Center of the Western Theater General Hospital of the People’s Liberation Army. All animal experimental procedures were conducted in strict accordance with experimental animal welfare and ethical guidelines and were approved by the Ethics Committee of the Western Theater General Hospital (Approval Number: 2025EC1-ky019).

Animal Housing and Management

The housing environment was specific pathogen-free (SPF) grade, maintained under specific conditions: temperature 22 ± 2 °C, relative humidity 40–60%, and a 12-h light/dark cycle (light period from 07:00–19:00). Mice were group-housed in static micro-isolator cages (3–5 mice per cage), equipped with filter-top lids. Bedding consisted of a mixture of corncob and poplar shavings and was changed 1–2 times per week. Environmental enrichment items were provided, and mice had ad libitum access to standard chow and water. To minimize stress-induced interference with the experimental results, all mice underwent an acclimation period of at least 14 days prior to the start of the experiments. All major experimental procedures were performed during the mid-light phase to control for potential variables arising from circadian rhythms.

TBI Animal Model, Grouping, and Treatment Protocol

Animals and Experimental Groups

After a 1-week acclimation period, adult male C57BL/6 mice were randomly assigned to one of five experimental groups (n = 10 per group). To ensure optimal utilization of animals while adhering to the 3Rs principle, the 10 mice in each group were allocated as follows: n = 3 for Western blotting (e.g., SOD2 lactylation, Cleaved Caspase-3), n = 3 for biochemical assays (e.g., SOD2 activity, GSH/GSSG ratio), and n = 4 for histological analysis (e.g., TUNEL staining).

Sham Group (Sham): Underwent craniotomy without the cortical impact and received vehicle injections.

TBI Group (TBI): Subjected to CCI-induced TBI and received vehicle injections.

TBI + Lactate Group (TBI + Lac): Subjected to TBI and treated with L-Lactic acid and the vehicle for HKL/3-TYP.

TBI + Lactate + HKL Group (TBI + Lac + HKL): Subjected to TBI and treated with L-Lactic acid and Honokiol (HKL).

TBI + Lactate + 3-TYP Group (TBI + Lac + 3-TYP): Subjected to TBI and treated with L-Lactic acid and 3-TYP.

Controlled Cortical Impact (CCI) Procedure

On day 3 of the experimental protocol, mice designated for injury were anesthetized (2–3% isoflurane for induction, 1–1.5% for maintenance, 792632, Sigma-Aldrich) and fixed in a stereotaxic frame. The depth of anesthesia was monitored by confirming the absence of a pedal withdrawal reflex (toe pinch), and body temperature was maintained at 37 °C using a heating pad throughout the procedure. A craniotomy of approximately 4 mm in diameter was performed over the right parietal cortex (centered at 2.0 mm posterior to bregma, 2.0 mm lateral to the midline). Subsequently, the exposed cortex was subjected to a single, controlled impact using a CCI device (PinPoint PCI3000, Hatteras Instruments) with the following parameters: a 2 mm diameter impactor tip, an impact velocity of 4 m/s, an impact depth of 1.5 mm, and a dwell time of 100 ms. Sham-operated mice underwent the same anesthetic and surgical procedures, including craniotomy, but did not receive the cortical impact.

Drug Preparation and Administration Protocol

All drugs and vehicles were administered via intraperitoneal (i.p.) injection. Two distinct vehicles were used: a 20% fat emulsion (Intralipid®) for HKL and 3-TYP, and phosphate-buffered saline (PBS, pH 7.4) for L-Lactic acid.

Honokiol (HKL; HY-N0003, MCE) and 3-TYP (HY-108331, MCE) were prepared as homogeneous suspensions by mixing with the fat emulsion vehicle (1:14 v/v) and sonicating, following established protocols [29, 30]. L-Lactic acid (CAS: 79–33-4, Sangon Biotech) was dissolved in PBS, and the pH was precisely adjusted to 7.4 with NaOH.

The treatment schedule was as follows:

• HKL/3-TYP Pre-treatment: Starting on day 0, mice in the TBI + Lac + HKL and TBI + Lac + 3-TYP groups received daily injections of HKL (10 mg/kg) or 3-TYP (50 mg/kg) [29, 30], respectively, for seven consecutive days (from day 0 to day 6). This pre-treatment strategy aimed to effectively modulate SIRT3 activity prior to injury. The dosages were selected based on previous studies demonstrating their efficacy in modulating SIRT3 activity in vivo.

• Lactate Post-treatment: Starting immediately after TBI induction on day 3, mice in the TBI + Lac, TBI + Lac + HKL, and TBI + Lac + 3-TYP groups received daily injections of L-Lactic acid (54 mg/kg) for four consecutive days (from day 3 to day 6). The dose of L-Lactic acid (54 mg/kg) was chosen to model the systemic hyperlactatemia (~ 10 mmol/L) reported in severe TBI [31]. This dose, calculated to achieve the target concentration based on estimated mouse blood volume, was confirmed to be well-tolerated in a preliminary study.

• Vehicle Administration: To ensure rigorous control, vehicle-only groups received injections corresponding to the active treatments. The Sham and TBI groups received daily i.p. injections of the fat emulsion vehicle from day 0 to day 6, as well as daily i.p. injections of the PBS vehicle from day 3 to day 6. The TBI + Lac group received the fat emulsion vehicle daily from day 0 to day 6.

At the designated experimental endpoint, mice were euthanized. Euthanasia was induced with a lethal overdose of pentobarbital sodium (150 mg/kg, i.p. P3761, Sigma-Aldrich). Following the confirmed loss of consciousness, a terminal procedure was performed. For biochemical analyses (e.g., Western blotting, SOD2 activity assays), mice were exsanguinated via transcardial perfusion with ice-cold saline, after which brains were rapidly dissected and stored at −80 °C. For histological analyses (e.g., apoptosis detection), mice were perfused first with ice-cold saline, followed by 4% paraformaldehyde (PFA) to ensure tissue fixation. This two-step method ensures deep unconsciousness before death.

Cell Culture and Plasmid Transfection

Cell Culture: Human embryonic kidney cells (HEK293T, CL0005, PunoSci, China) and mouse neuroblastoma cells (N2a, CL-0168, PunoSci, China) were both purchased from PunoSci Biotechnology Co., Ltd. All cells were routinely cultured in high-glucose DMEM (C11995500BT, Gibco, China) supplemented with 10% fetal bovine serum (FBS, FSP500, China) and 1% penicillin–streptomycin solution (C0222, Beyotime, China) in a humidified incubator at 37 °C with 5% CO₂. In this study, HEK293T cells were utilized for the initial mapping of protein interactions (e.g., Co-IP and APEX2) due to their high transfection efficiency [32] and the conserved nature of protein lactylation across cell types [33]. However, to ensure physiological relevance to TBI, the neuronal N2a cell line was strictly used as the primary in vitro model for all subsequent functional validations.

Plasmid Construction and Source: The pCMV-SOD2-V5-APEX2 and pCMV-SIRT3-Flag expression plasmids were constructed by BioSune (China). Briefly, the CDS regions of mouse Sod2 and Sirt3 were amplified by PCR and cloned into the expression vector using homologous recombination technology (In-Fusion, 638947, Takara, Japan). After transformation of the recombinant products into DH5α, positive clones were selected, and plasmids were extracted using an endotoxin-free plasmid maxi kit (DP117, Tiangen, China). To verify the expression efficiency of the plasmids, they were transiently transfected into HEK293T cells, and Western blot analysis successfully detected target protein expression of the expected size, while the control group showed no such band (Fig. S1A). The validated plasmids were used for all subsequent experiments. To specifically knock down the expression of AARS2, a validated shRNA targeting AARS2 (shRNA1-AARS2) plasmid was commercially constructed and synthesized by Miaoling Biology (Wuhan, China). The remaining plasmids were kindly provided by Professor Tao Kai of the Western Theater General Hospital.

Primers used for constructing pCMV-SOD2-V5-APEX2:

• Forward: 5′-CCCAAGCTTGCGGCCGCcaccATGTTGTGTCGGGCGGCG-3′

• Reverse: 5′-GGGCTTGCCTGCTAGCttCTTCTTGCAAGCTGTGTATCTTTC-3′

Primers used for constructing pCMV-SIRT3-Flag:

• Forward: 5′-TGACGATGACAAGCTTATGGTGGGGGCCGGCATC-3′

• Reverse: 5′-GCCACTGTGCTGGATATCTTATCTGTCCTGTCCATCCAGCTTG-3′

Transient Plasmid Transfection: Transfection was performed using Lipo8000™ transfection reagent (C0533, Beyotime, China). Following the manufacturer's instructions, plasmid DNA and transfection reagent were pre-incubated in Opti-MEM I Reduced Serum Medium (31985062, Gibco, USA) before being added to the cells to be transfected.

In Vitro Simulation of the TBI Pathological Microenvironment

To simulate the complex pathological microenvironment after TBI in vitro, we applied the following treatments to cells with appropriate adjustments to the treatment time based on published literature:

• Hypoxia treatment (simulating ischemia and hypoxia) [34]: The experimental group (Hypoxia) was placed in a tri-gas incubator with a gas composition of 2% O₂, 5% CO₂, and 93% N₂ for 6 h to simulate the hypoxic state of brain tissue after TBI. The control group was cultured synchronously under standard culture conditions (21% O₂, 5% CO₂).

• High lactate treatment (simulating lactic acidosis) [35]: The experimental group was treated with sodium lactate (Sodium Lactate, 71,718, Sigma-Aldrich, USA) added to the culture medium to a final concentration of 50 mM for 24 h to simulate the abnormal accumulation of lactate in local tissues after TBI. The control group received treatment with an equal volume of solvent (e.g., PBS or culture medium).

• High glucose treatment (simulating early-stage high glycolytic flux and promoting lactate flux after TBI) [36]: The purpose of using high glucose treatment in this study was not to investigate hyperglycemia per se, but to simulate the drastic high glycolytic state in the brain during the early stages of TBI, thereby actively promoting the production flux of endogenous lactate and further aggravating the lactate accumulation environment of TBI. The experimental group (High Glucose, HG) was treated with glucose solution added to the basal medium to a final concentration of 33 mM for 24 h. The normal control group maintained conventional culture conditions.

APEX2-Mediated Proximity Biotinylation

To map the highly dynamic and potentially transient interaction network of SOD2 within the living cellular microenvironment, we employed the engineered ascorbate peroxidase (APEX2) proximity labeling system. Unlike traditional Co-Immunoprecipitation (Co-IP), which often fails to capture weak or transient enzyme–substrate interactions due to stringent lysis and washing steps, APEX2 allows for high-spatiotemporal resolution labeling of the local proximitome in cells, it biotinylates all spatially proximal proteins around the target protein (SOD2) within living cells with extremely high spatiotemporal resolution (< 1 ms, < 20 nm radius).

Live cell proximity labeling: Cells were incubated with culture medium containing 500 μM biotin-phenol (HY-125658S, MCE) at 37 °C for 30 min. The reaction was initiated by adding H₂O₂ (final concentration 1 mM) and allowed to proceed at room temperature for 1 min. The reaction was immediately quenched by washing three times with pre-chilled DPBS quenching buffer containing 10 mM sodium azide, 10 mM sodium ascorbate (HY-B0166A, MCE), and 5 mM Trolox (HY-101445, MCE).

Streptavidin pulldown enrichment and Western blot analysis: Cells were collected, and protein lysates were prepared. Biotinylated proteins were captured using streptavidin magnetic beads (HY-K0208, MCE). After thorough washing, the captured proteins were eluted using a high-concentration urea elution buffer (containing 8 M urea). The eluate was separated by SDS-PAGE, transferred to a membrane, and chemiluminescence detection and semi-quantitative analysis were performed using horseradish peroxidase (HRP)-conjugated streptavidin (SA00001-0, Proteintech, 1:10,000).

Protein Blotting Analysis (Western Blotting)

The quantified protein samples were mixed with loading buffer and denatured at 95 °C for 10 min. 20–30 µg of protein was loaded onto 10% or 12% SDS-PAGE gels (ET15010/ET15012, ACE) for electrophoresis, followed by wet transfer to PVDF membranes (IPVH00010, Millipore). The membranes were blocked with 5% skim milk in TBST at room temperature for 1 h and incubated with the corresponding primary antibody at 4 °C overnight. The next day, after washing with TBST, the membranes were incubated with HRP-conjugated secondary antibody at room temperature for 1 h. The signal was detected using ECL reagent, and ImageJ software (v1. 54p) was used to analyze band gray values, which were then normalized to the internal control protein (β-actin or GAPDH).

Immunoprecipitation (IP) for specific PTM detection: To precisely evaluate the specific lactylation status of SOD2, a rigorous reciprocal immunoprecipitation strategy was employed. For all IP assays, total cellular or tissue lysates were strictly partitioned: precisely 10% of the total lysate volume was reserved as the whole-cell Input control to monitor basal protein expression and ensure equal loading, while the remaining 90% was subjected to target enrichment. Specifically, the 90% IP fractions were incubated with specific anti-SOD2 antibodies, anti-tag antibodies (e.g., anti-HA or anti-V5), or a pan-anti-lactyl-lysine (Pan-Klac) antibody at 4 °C overnight, followed by capture with Protein A/G magnetic beads (HY-K0202, MCE). The purified immune complexes were subsequently resolved via SDS-PAGE and subjected to immunoblotting. Crucially, to systematically eliminate the masking effect of the ~ 25 kDa precipitating antibody light chains against our target SOD2, a rigorous cross-species IP-immunoblotting strategy was prioritized where feasible (e.g., performing IP with a rabbit host antibody followed by immunoblotting with a mouse primary antibody). Furthermore, conformation-specific secondary antibodies—which exclusively recognize natively folded primary antibodies and disregard denatured IgG chains—were utilized during chemiluminescent detection, thereby ensuring unambiguous identification of the 25 kDa lactylated SOD2 band (for detailed information on all Western blots and fluorescent antibodies, please refer to Supplementary Table S1.)

Immunofluorescence Staining and Confocal Analysis

Cells were cultured on sterile coverslips. After treatment, they were fixed with 4% PFA at room temperature for 20 min and permeabilized with 0. 3% Triton X-100 (GC204003, Servicebio) for 10 min. After blocking with 2% BSA for 1 h, they were incubated with the corresponding primary antibody at 4 °C overnight. The next day, after washing, they were incubated with a fluorescent secondary antibody at room temperature in the dark for 1 h. The cell nuclei were counterstained with Hoechst 33342 (C1027, Beyotime), and images were acquired using a Nikon A1R HD confocal microscope after mounting. The Coloc 2 plugin of ImageJ (v1. 54p) was used to analyze protein colocalization and calculate Pearson’s correlation coefficient.

Superoxide Dismutase (SOD) Activity Assay

SOD activity in brain tissue homogenates was measured using a CuZn/Mn-SOD Activity Assay Kit (S0103, Beyotime). After normalizing for total protein concentration using a BCA assay kit (B6167, uelandy), the procedure was performed strictly according to the manufacturer’s instructions. To specifically detect SOD2 activity, a Cu/Zn-SOD inhibitor was added. Absorbance was measured at a wavelength of 450 nm, and enzyme activity was calculated (see supplementary information table S2 for detailed data).

mtROS Measurement

Fluorescence quantification: mtROS levels in cells were quantified using a Mitochondrial Superoxide Detection Kit (Mito NeoD, S0062, Beyotime). According to the instructions, cells were incubated with the Mito NeoD probe, and fluorescence intensity was measured using a multi-well plate reader (Ex/Em = 544/605 nm) (see supplementary information table S3 for detailed data).

Fluorescence microscopy imaging: For visualization, cells were incubated with the Mito NeoD probe, then labeled with MitoTracker Green (C1048, Beyotime) to visualize mitochondria, and cell nuclei were counterstained with Hoechst 33342 (C1027, Beyotime). Images were acquired using a ZEISS Axio Observer fluorescence microscope. In ImageJ, mitochondrial regions of interest (ROIs) were defined based on the green fluorescence signal, and the mean fluorescence intensity (MFI) of the red fluorescence within the ROIs was measured for quantification.

Importantly, to rigorously validate the dynamic range, sensitivity, and specificity of the MitoNeoD fluorescent probe, a standardized positive control was incorporated across all independent replicates. Cells pre-treated with mSoxUp—a potent, mitochondria-targeted superoxide inducer provided directly within the MitoNeoD assay kit—served as the specific positive control to ensure that the captured signals accurately reflected authentic intracellular mtROS fluctuations.

Histological Analysis (TUNEL Staining)

Tissue perfusion and preparation: After deep anesthesia, mice were transcardially perfused with physiological saline and 4% PFA. Brains were harvested and post-fixed in 4% PFA for 24 h, followed by gradient dehydration using 20% and 30% sucrose solutions.

Cryosectioning and TUNEL staining: Dehydrated brain tissues were embedded in OCT, and 30 μm thick coronal sections were prepared using a Leica CM1950 cryostat. Staining was performed strictly according to the instructions of the One Step TUNEL Apoptosis Assay Kit (Green Fluorescence) (C1088, Beyotime), and cell nuclei were counterstained with Hoechst 33342.

Image acquisition and analysis: Images were acquired using an Olympus VS200 fully automated digital slide scanning system. In QuPath software (v0. 6. 0), ROIs were drawn in the injured cortical region, and TUNEL-positive cells and total cell nuclei were automatically identified and counted. Apoptosis density (cells/mm²) and apoptosis rate (%) were calculated.

Proteomics Analysis

APEX2 proximity labeling proteomics: A quantitative mass spectrometry (MS) analysis workflow consisting of three control groups. The workflow was designed in two steps: First, defining the interaction background: By comparing the SOD2-V5-APEX2 group with the negative control GFP group, we precisely screened out a high-confidence SOD2 proximity proteome, thoroughly eliminating interference from non-specific background proteins. Second, quantifying dynamic changes: Only within this high-confidence SOD2 proximity proteome list did we compare the SOD2 + NaLac treatment group with the SOD2 experimental group, thereby quantifying the true dynamic changes induced by lactate in the SOD2 proximity proteome with an extremely high signal-to-noise ratio. Cell groups were treated according to the experimental plan, and cells were collected after proximity labeling. Cells were lysed with IP lysis buffer. The supernatant was incubated with streptavidin magnetic beads at 4 °C overnight. After stringent washing, the magnetic bead-protein complex was sent to Shanghai Applied Protein Technology Co., Ltd. for in situ trypsin digestion and LC–MS/MS identification. Volcano plots, heatmaps, GO, and KEGG enrichment analyses were performed on the returned results using R software (v4. 4. 2) (data and related R code are in supplementary information Note S2.)

Lactyl-lysine modification proteomics: At 12 h post-TBI, brain tissue samples (~ 100 mg/sample) were collected from the injured cortex, snap-frozen in liquid nitrogen, and stored at −80 °C. Samples were sent to PTM Bio for analysis. The process included protein extraction, digestion, lactyl-lysine peptide immunoprecipitation enrichment, and LC–MS/MS identification and bioinformatics analysis.

Quantification of Tissue L-Lactate, GSH/GSSG Ratio, and Apoptosis

For the rigorous quantification of metabolic alterations, oxidative stress markers, and apoptotic execution, identical ultra-low temperature (−80 °C) archived brain tissues from the original animal cohort were utilized, thereby ensuring absolute biological consistency across all supplementary evaluations.

Endogenous L-lactate concentrations in the brain tissues were measured utilizing the highly sensitive L-Lactate Assay Kit (S0227S, Beyotime, Shanghai, China). Tissues were homogenized in the provided BeyoLysis™ Buffer A. Following the enzymatic colorimetric reaction, the precise L-lactate levels were quantified by measuring the absorbance at 570 nm (A570) using a microplate reader, strictly adhering to the manufacturer’s optimized protocol.

Furthermore, to dynamically evaluate the tissue oxidative stress status, the ratio of reduced to oxidized glutathione (GSH/GSSG) was determined using the GSH and GSSG Assay Kit (S0053, Beyotime, Shanghai, China). Tissues were instantly homogenized in the dedicated Protein Removal Reagent M to prevent artefactual oxidation. Total glutathione and GSSG levels were independently quantified via the DTNB chromogenic reaction measured at 412 nm. The specific GSH content was subsequently calculated by subtracting twice the GSSG concentration from the total glutathione amount.

Additionally, to evaluate the execution phase of cellular apoptosis within the injured parenchyma, the identically archived tissues were subjected to cleaved caspase-3 assessment. Total proteins were extracted from the tissue homogenates using RIPA lysis buffer, and the active cleaved caspase-3 fragments (17/19 kDa) were subsequently detected and quantified via immunoblotting, as detailed in the “Protein Blotting Analysis (Western Blotting)” section.

Neurological Function Assessment

To evaluate the longitudinal recovery of neurological function following TBI, a modified Neurological Severity Score (mNSS) assessment was performed on all 50 mice (n = 10 per group) across the five experimental cohorts. The mNSS is a composite test grading motor, sensory, reflex, and balance functions on an 18-point scale, where a higher score indicates more severe neurological deficits. The scoring criteria included 11 specific parameters: forelimb flexion (0–1), hindlimb flexion (0–1), head twisting (0–1), walking on a flat surface (0–3), placing reflex (0–1), proprioception (0–1), beam balance (0–6), absence of pinna reflex (0–1), absence of corneal reflex (0–1), absence of startle reflex (0–1), and the presence of abnormal signs (0–1).

All behavioral assessments were conducted by two independent investigators blinded to the experimental groupings. Specifically, all mice were uniquely identified and randomized using an ear-punching coding system managed by a separate researcher who did not participate in the behavioral testing. The two evaluating investigators performed the modified Neurological Severity Score (mNSS) assessments independently at five designated time points: immediately prior to the TBI induction (baseline) and subsequently at 24, 48, 72, and 96 h post-injury. The final mNSS score for each animal was defined as the average of the scores provided by these two blind observers, and the coding assignment was only revealed after the completion of all data analysis.

Statistical Analysis

All quantitative data were statistically analyzed and plotted using GraphPad Prism 9 software. For comparisons between two independent groups, an unpaired Student’s t-test was performed. For comparisons among three or more groups, data were analyzed using one-way or two-way analysis of variance (ANOVA), followed by Tukey’s post hoc test for multiple comparisons. All molecular and biochemical data are presented as mean ± standard deviation (SD). For longitudinal behavioral assessments, such as the modified Neurological Severity Score (mNSS), a repeated-measures two-way ANOVA followed by Tukey’s post hoc test was utilized to account for both treatment and time factors. These behavioral data are expressed as mean ± standard error of the mean (SEM). Statistical significance was set at P < 0.05, with asterisks indicating the significance level: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Results

TBI Is Accompanied by SOD2 Lactylation Modification

TBI-induced tissue ischemia-hypoxia and increased lactate metabolic flux promote the production of lactyl-CoA, which in turn drives protein lactylation [37–39].

To reveal the impact of TBI on the protein lactylation (Kla) landscape, we performed an unbiased quantitative proteomic analysis on brain tissues from sham and TBI mice (Fig. 1A). The analysis revealed that TBI caused widespread and drastic remodeling of protein lactylation levels in brain tissue. Compared to the sham group, the TBI group had 45 proteins with significantly upregulated lactylation levels and 229 proteins with significantly downregulated lactylation levels (Fig. 1B). Notably, 11% of the lactylation-upregulated proteins were localized to mitochondria. Since mitochondrial proteins typically account for approximately 7–14% of the total cellular proteome, this distribution indicates that TBI-induced protein lactylation is a widespread, pan-cellular response to damage rather than being strictly confined to a specific organelle. However, given the central role of metabolic crisis and oxidative stress in acute TBI pathology, we specifically directed our focus toward the mitochondrial network. Among these localized targets, further site-level analysis drew our attention to a core mitochondrial antioxidant enzyme, SOD2, specifically including K68, K75, K114, K122, and K130 (Fig. 1C––G). Crucially, 3D molecular modeling (Fig. 1H, I) visualized the spatial distribution of these newly identified lactylation sites on both the SOD2 monomer and its functional tetrameric complex. Previous studies have well-documented that reversible acetylation at specific lysine residues—particularly K68 and K122—critically governs SOD2 enzymatic activity [23, 40]. Strikingly, our mass spectrometry data demonstrate that TBI induces robust lactylation at these exact same functional hotspots (K68 and K122), as well as adjacent residues.

Fig. 1.

(A) Schematic of the experimental design for proteomic analysis of TBI-induced protein lactylation. (B) Venn diagram of differentially lactylated proteins in TBI versus sham groups. Pie charts show subcellular localization for proteins with upregulated (45) and downregulated (229) lactylation. (C-G) Representative high-resolution LC–MS/MS spectra confirming the identification of novel SOD2 lactylation sites at K68 (C), K75 (D), K114 (E), K122 (F), and K130 (G). (H-I) 3D molecular modeling illustrating the spatial distribution of the identified lactylation sites mapped onto the structure of the SOD2 monomer (H) and the active SOD2 tetramer (I). Statistical analysis in the proteomic screening was performed using a two-tailed Student's t-test; P < 0.05 was considered significant. Abbreviations in pie charts: mito, mitochondria; cyto, cytoplasm; extra, extracellular; pero, peroxisome; PM, plasma membrane; ER, endoplasmic reticulum

This exact site overlap strongly suggests that under the high-lactate pathology of TBI, lactylation may hijack these critical regulatory sites, acting as a novel and potentially more disruptive mechanism to inhibit SOD2 function by mimicking or outcompeting acetylation. Therefore, we decided to conduct in-depth experimental validation of the functional consequences of this modification.

Simulating TBI Microenvironment Increases SOD2 Lactylation

To investigate the effect of the TBI-characteristic microenvironment on SOD2 lactylation (Kla), we established in vitro models using human embryonic kidney (293T) and neuroblastoma (N2a) cells, respectively. We simulated this pathological hypoxic and high-lactate microenvironment by inducing lactate flux through hypoxic treatment, high-lactate culture, and high glucose.

To specifically assess the lactylation status of SOD2, cells were transfected with an HA-tagged SOD2 plasmid. We then performed immunoprecipitation (IP) using a pan-lactyllysine (Pan-Kla) antibody to enrich the global lactylated proteome, followed by immunoblotting against the HA tag. The results showed that in 293 T cells, SOD2 Kla levels were significantly upregulated under all three conditions (Fig. 2A-C). This lactate-driven modification was further confirmed in N2a neuronal cells following direct sodium lactate supplementation (Fig. 2D), indicating a direct causal relationship between the TBI pathological environment and SOD2 lactylation. Collectively, these data establish a stable in vitro model of TBI-mimicking conditions, laying a solid foundation for subsequent functional investigations into SOD2 lactylation.

Fig. 2.

TBI-mimicking conditions increase SOD2 lactylation. (A-D) Immunoblotting analysis of SOD2 lactylation in 293 T cells (A-C) and N2a cells (D) subjected to hypoxia (Hypo), high glucose (HG), or sodium lactate (NaLac) treatments. The relative lactylation level of SOD2 was quantified by calculating the ratio of the immunoprecipitated lactylated SOD2 (IP: Pan-Kla, IB: HA) to the total expressed SOD2 (Input: HA). Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using an unpaired two-tailed Student's t-test. *P < 0.05, ****P < 0.0001

SOD2 Lactylation Reduces SOD2 Activity and Increases mtROS

Having established that the TBI microenvironment induces SOD2 lactylation, and given that SOD2 is a core antioxidant enzyme for scavenging mitochondrial superoxide [41, 42], we next investigated whether this modification directly impairs its enzymatic activity.

The results clearly showed that in N2a and 293 T cells, the high lactylation state induced by the simulated TBI environment directly led to a significant decrease in the activity of endogenous SOD2 dismutase (Fig. 3A). To further confirm that the modification itself, rather than other factors, was inhibiting enzyme activity, we performed a rescue experiment involving SOD2 overexpression. Remarkably, although SOD2 overexpression effectively increased the basal enzyme activity of cells, this protective effect was significantly attenuated following sodium lactate supplementation (Fig. 3B). This indicates that lactylation modification can inhibit the catalytic function of SOD2.

Fig. 3.

SOD2 Lactylation Impairs Its Enzymatic Activity and Increases mtROS. (A) SOD2 enzymatic activity in 293 T and N2a cells under simulated TBI conditions. (B) SOD2 activity in wild-type (WT) SOD2-overexpressing cells treated with or without sodium lactate (NaLac). (C) mtROS levels in 293 T and N2a cells following NaLac treatment, measured by MitoSOX Red fluorescence. (D) mtROS levels in WT SOD2-overexpressing cells following NaLac treatment. (E) Representative confocal images of NaLac-treated N2a cells stained with MitoSOX Red (mtROS), MitoTracker Green (mitochondria), and DAPI (nuclei). Scale bars: 50 μm (left panels) and 10 μm (Zoom panels). For (C-E), N denotes negative control; P denotes positive control (treated with mSoxUp, a specific mitochondrial superoxide inducer provided within the kit). Quantification of the mtROS signal from (E), presented as the ratio of MitoSOX Red to MitoTracker Green fluorescence intensity. All data are presented as mean ± SD from three independent experiments. Statistical significance was determined by one-way ANOVA followed by Tukey's post-hoc test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant)

The consequence of the loss of enzyme activity is the accumulation of substrate (superoxide). We quantitatively measured the levels of mtROS using the MitoSOX Red fluorescent probe. The data confirmed that high sodium lactate treatment significantly increased the intracellular mtROS levels (Fig. 3C). Crucially, even in cells overexpressing more SOD2 protein, the high-lactate environment still led to a significant accumulation of mtROS (Fig. 3D), which again confirmed that the newly expressed SOD2 protein was also inhibited by lactylation. Through fluorescence microscopy imaging, we visually confirmed this robust increase in mtROS (Fig. 3E).

Together, these data delineate a clear pathological cascade: TBI-induced lactylation inhibits SOD2 enzymatic activity, thereby triggering a profound accumulation of mtROS. This finding prompted us to investigate the specific acyltransferase (“writer”) responsible for executing this modification on SOD2.

AARS2 Mediates SOD2 Lactylation to Impair Enzymatic Activity and Promote mtROS Accumulation

Having identified the pathological function of SOD2 lactylation, our next key question was: which writer is executing this modification? Considering the specific compartment of the mitochondrial matrix, although studies have reported that various acetyltransferases such as P300 possess lactyltransferase activity, their specificity in mitochondria is unclear. Notably, recent studies have identified aminoacyl-tRNA synthetase AARS2 as a key lysine lactyltransferase in mitochondria [43, 44], This non-classical function of AARS2 is thought to stem from the structural similarity between lactic acid and alanine, allowing it to compete for lactyl-tRNA as a donor to catalyze protein lactylation. Therefore, we hypothesized that AARS2 mediates SOD2 lactylation.

To validate AARS2 as an in vivo interacting partner of SOD2, we performed unbiased proximity labeling using APEX2. In cells expressing a SOD2-V5-APEX2 fusion protein, we observed a significant enrichment of AARS2 compared to controls (Log2FC = 0.55, P = 0.0029; Fig. S1B). This finding led us to prioritize AARS2 for subsequent functional characterization. However, further investigation into whether a high-lactate environment enhances SOD2 modification by recruiting AARS2 yielded a paradoxical result. APEX2 proximity labeling revealed that treatment with high sodium lactate did not alter the abundance of AARS2 in the SOD2 proximity interactome (Log2FC = 0.002, P = 0.97; Fig. S1C). However, Co-IP assays revealed that under high sodium lactate conditions, the physical co-precipitation of AARS2 with SOD2 was significantly enhanced (Fig. 4A, Fig. S1D), as was their cellular colocalization (Fig. 4C, D). These seemingly divergent results likely reflect the different nature of the two assays: APEX2 detects proteins in broad spatial proximity (~ 20 nm), while Co-IP captures stable, high-affinity interactions. Therefore, our data suggest that lactate does not recruit AARS2, but rather strengthens its pre-existing, stable binding to SOD2.

Fig. 4.

AARS2 Promotes Lactate-Induced SOD2 Lactylation and Dysfunction. (A) Co-IP assay and corresponding quantitative analysis showing the physical interaction between SOD2 and AARS2 in N2a cells treated with or without NaLac. The relative interaction was quantified as the ratio of immunoprecipitated AARS2 to SOD2. (B) SOD2 enzyme activity in vector control or AARS2-overexpressing N2a cells ± NaLac. (C-D) Representative immunofluorescence images and 2D intensity scatter plots (C) showing the colocalization of AARS2 (red) and SOD2 (green) in NaLac-treated N2a cells, alongside quantitative Pearson Correlation Coefficient analysis (D). Nuclei were stained with DAPI (blue). (E) Immunoblot analysis and quantification of SOD2 lactylation (Kla) levels following Co-IP in AARS2-overexpressing N2a cells ± NaLac. (F-G) Co-IP assay (F) and quantification (G) demonstrating that knockdown of AARS2 (shAARS2) abolishes NaLac-induced SOD2 lactylation, compared to the negative control (ShNC). (H-I) Representative fluorescence images (H) and quantification (I) of mtROS levels, detected by MitoSOX Red (red). Mitochondria were labeled with MitoTracker Green (green). For H and I, N denotes negative control; P denotes positive control (treated with mSoxUp, a specific mitochondrial superoxide inducer provided within the kit). Statistical significance was determined using an unpaired two-tailed Student's t-test for two-group comparisons (D), or one-way ANOVA followed by Tukey's post-hoc test for multiple comparisons (A, B, E, G, I), n = 3, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant

We then performed a series of functional experiments to confirm AARS2’s role as a lactylation writer. The results showed that overexpression of AARS2 alone did not affect SOD2 lactylation, enzyme activity, or mtROS levels. However, when combined with high sodium lactate (NaLac), AARS2 overexpression significantly increased SOD2 lactylation (Fig. 4E), robustly inhibited SOD2 dismutase activity (Fig. 4B), and ultimately led to a substantial accumulation of mtROS (Fig. 4H, I, Fig. S1E).

Crucially, to address whether high lactate levels are sufficient to promote SOD2 lactylation independently of AARS2, we performed loss-of-function experiments using AARS2 knockdown (shAARS2). As expected, NaLac exposure significantly increased SOD2 lactylation in the control group (shNC). Strikingly, depletion of AARS2 almost completely abolished this NaLac-induced SOD2 lactylation (Fig. 4F, G).

This lactate-dependent experimental evidence demonstrates that AARS2 is the indispensable writer mediating SOD2 lactylation, and that its catalytic function is strictly contingent upon the pathological metabolic environment of high lactate.

SIRT3 Enhances SOD2 Activity and Reduces mtROS by Lowering SOD2 Lactylation Levels

Having identified AARS2 as the lactylation “writer”, we next investigated the potential “eraser” enzyme responsible for removing this modification. SIRT3, a major mitochondrial deacetylase already established to regulate SOD2 acetylation [45], emerged as our primary candidate.

To validate this hypothesis, we performed a series of functional rescue experiments in N2a cells subjected to high-lactate treatment. Overexpression of SIRT3 significantly reversed the lactate-induced hyperlactylation of SOD2, restoring it to near-basal levels (Fig. 5A). This reversal of the modification led to a direct functional recovery, as SIRT3 overexpression also successfully rescued the SOD2 dismutase activity that was inhibited by high lactate (Fig. 5B).

Fig. 5.

SIRT3 Rescues SOD2 Activity by Counteracting Lactate-Induced Lactylation. (A) SOD2 lactylation levels in N2a cells overexpressing SIRT3 under high-lactate (NaLac) conditions. (B) SOD2 enzyme activity in SIRT3-overexpressing N2a cells treated with NaLac. (C) mtROS levels in NaLac-treated N2a cells overexpressing SIRT3. mtROS was detected with MitoSOX Red (red) and mitochondria with MitoTracker Green (green). N denotes negative control; P denotes positive control. Scale bars: 100 μm (left panels) and 10 μm (Zoom panels). (D) Co-IP analysis of the SIRT3-SOD2 interaction in 293 T and N2a cells treated with or without NaLac. (E) Immunofluorescence showing colocalization of SIRT3 (green) and SOD2 (red) in NaLac-treated N2a cells. Nuclei are stained with DAPI (blue).Data are presented as mean ± SD from three independent experiments, Statistical significance was determined using an unpaired two-tailed Student's t-test for two-group comparisons (E), or one-way ANOVA followed by Tukey's post-hoc test for multiple comparisons (A-D),*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant

Ultimately, the SIRT3-mediated restoration of SOD2 enzymatic activity effectively reduced mitochondrial oxidative stress. Both quantitative MitoSOX analysis and fluorescence microscopy confirmed that SIRT3 expression prevented the accumulation of mtROS under high-lactate conditions (Fig. 5C, Fig. S1F). Together, these results delineate a clear protective pathway wherein SIRT3 counteracts the detrimental effects of high lactate by delactylating SOD2 and restoring its enzymatic function.

Next, we investigated the physical interaction between SIRT3 and SOD2. Notably, the lactate-induced enhancement of this interaction appeared to be cell-type specific. While a significant enhancement in both Co-IP and co-localization was observed in the neuronal N2a cell line (Fig. 5D, E), no statistically significant change in Co-IP was detected in HEK293T cells (Fig. S1G). This suggests that in response to metabolic stress, neurons—a cell type particularly vulnerable to such insults—may possess a more robust mechanism for the dynamic recruitment of SIRT3.

Notably, and in stark contrast to the resident writer AARS2 that was stably present in the SOD2 proximal proteome, SIRT3 was not detected in our APEX2 screen (Note S1). The combination of a positive Co-IP result (indicating stable binding) with a negative APEX2 result (suggesting a lack of close spatial proximity) implies that SIRT3 may employ a transient or highly dynamic “hit-and-run” catalytic mechanism [46, 47]. We will explore this model in greater detail in the “Discussion” section.

SOD2 Lactylation Remodels Its Proximal Interactome, Shifting Cellular Focus from Metabolism to Proteostasis

To map the global changes in the SOD2-proximal proteome induced by high lactate, we employed APEX2-based proximity labeling coupled with quantitative mass spectrometry (MS). Prior to MS analysis, we validated the labeling system. We confirmed the labeling efficiency of the SOD2-V5-APEX2 fusion protein (Fig. S1H) and observed no significant differences in total biotinylated protein levels or the spatial co-localization of biotin and SOD2-V5 signals following lactate treatment (Fig. 6A–C). These crucial control experiments indicated that high lactate does not cause a global shift in the SOD2 proximitome, but rather induces a specific remodeling of its interactome.

Fig. 6.

SOD2 Lactylation Remodels Its Interactome to Reprogram Cellular Function. (A) Co-IP of the SOD2-V5-APEX2 fusion protein and its proximal interactome in N2a cells treated with or without sodium lactate (NaLac). (B-C) Quantitative analysis of Pearson Correlation Coefficients (B) and representative immunofluorescence images (C) showing the colocalization of SOD2 (green) and APEX2-biotinylated proximal proteins (streptavidin, red) in N2a cells ± NaLac. Nuclei were stained with DAPI (blue). (D) Volcano plot of proteins showing differential interaction with SOD2 in NaLac-treated versus control cells. Up-regulated (red) and down-regulated (blue) interactors are highlighted. (E) Heatmap of the top 30 proteins with altered SOD2 interaction (15 upregulated, 15 downregulated) between the NaLac-treated and control groups. Proteomics analysis was performed on three independent biological replicates(n = 3). Statistical test results are indicated within the plots. *P < 0.05, **P < 0.01, ns, not significant

To characterize these compositional shifts, we performed quantitative mass spectrometry (MS) analysis. Our experimental design (see “Methods and Materials”) allowed us to first define the high-confidence SOD2 proximal proteome by filtering against a negative control, and then to specifically quantify lactate-induced changes within this defined interactome. This rigorous analysis revealed a significant reprogramming of the SOD2 interaction network (Fig. 6D). Specifically, we observed a decreased association of SOD2 with proteins central to mitochondrial bioenergetics. Key downregulated interactors included subunits of the respiratory chain complex I, the TCA cycle-related protein ACOT13, and the sulfurtransferase TST. Conversely, there was a significantly increased association with proteins involved in proteostasis and cellular stress responses. Key upregulated interactors included the components of the unfolded protein response (UPR) and disulfide bond formation pathways (e.g., Ero1a, Txndc family) (Fig. 6E).

These data establish a novel link between a specific post-translational modification on SOD2 and the broader cellular protein homeostasis machinery (Fig. S2). In response to lactate-induced modification, the SOD2 interactome undergoes a functional switch, disengaging from metabolic modules and engaging with protein quality control networks.

In Vivo SIRT3 Modulation Is Associated with Preserved SOD2 Activity and Reduced Neuronal Apoptosis Following TBI

To further examine the in vivo relevance of our findings, we employed a mouse CCI model to test whether SIRT3 modulation could influence pathological outcomes within a high-lactate microenvironment (Fig. 7A).

Fig. 7.

Pharmacological modulation of SIRT3 mitigates lactate-driven SOD2 lactylation, oxidative stress, and neurological deficits in vivo. (A) Schematic of the experimental design, including TBI induction and pharmacological treatments. (B) Longitudinal assessment of neurological deficits. Neurological impairment was assessed via the Modified Neurological Severity Score (mNSS) at indicated hours post-injury. n = 10 independent animals per group, Data for this panel are presented as mean ± SEM and analyzed by two-way repeated-measures ANOVA followed by multiple comparisons. & P < 0.05, && P < 0.01, &&&& P < 0.0001, TBI + Lac vs. TBI; * P < 0.05, ** P < 0.01, ****P < 0.0001, TBI + Lac + HKL vs. TBI + Lac; # P < 0.05, ## P < 0.01, ### P < 0.001, #### P < 0.0001 TBI + Lac + 3-TYP vs. TBI + Lac). (C) Quantification of L-lactate concentration (μmol/g tissue) in perilesional brain tissues, confirming the robust metabolic modeling, n = 3 independent animals per group. (D) Representative immunoblots of SOD2 lactylation in brain tissues following Co-IP with anti-Kla antibody, n = 3 independent animals per group. (E) Quantification of SOD2 enzymatic activity in perilesional brain lysates across treatment groups, n = 3 independent animals per group. (F) Quantitative analysis of the GSH/GSSG ratio, shown on a Log10 scale, n = 3 independent animals per group. (G) Representative Western blots and quantification of Cleaved Caspase-3 normalized to GAPDH, n = 3 independent animals per group. (H-I) Representative immunofluorescence images (H) and quantitative analysis (I) of TUNEL-positive apoptotic cells (green) in the perilesional cortex. Nuclei were counterstained with DAPI (blue). Scale bars are indicated in the panels, n = 4 independent animals per group. For biochemical assays, Western blot/Co-IP analyses, and histological quantification, each “n” represents one independent animal. For histological quantification, multiple fields or sections from the same animal were averaged to generate one biological replicate. Data are expressed as mean ± SD except for Panel B and were analyzed by one-way ANOVA followed by multiple comparisons. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant

Our initial observations suggested that SIRT3 activity is associated with functional recovery. Longitudinal mNSS assessments revealed that the neurological deficits induced by TBI were significantly potentiated by exogenous lactate (TBI + Lac). Notably, pharmacological activation of SIRT3 with HKL was associated with improved mNSS scores, whereas its inhibition with 3-TYP further aggravated neurobehavioral impairment over 96 h (Fig. 7B). Importantly, these effects did not appear to be due to changes in the metabolic load itself; tissue L-lactate levels remained consistently elevated in all lactate-supplemented groups regardless of SIRT3 modulation (Fig. 7 C), while post-mortem analysis confirmed target engagement of HKL and 3-TYP on SIRT3 expression (Fig. S1I-J).

Mechanistically, the suggested “functional switch” of SOD2 appeared to be bidirectionally controlled by SIRT3 in vivo. In the TBI + Lac group, the surge in SOD2 lactylation correlated with a profound reduction in its enzymatic activity (Fig. 7D, E). SIRT3 activation by HKL partially reversed this hyperlactylation and was associated with restored scavenging function. Conversely, 3-TYP treatment was linked a more severe molecular phenotype, driving SOD2 toward functional failure (Fig. 7D, E).

This molecular restoration (or failure) potentially influenced the redox balance and neuronal fate. The “lactylation-high/activity-low” state in the TBI + Lac group resulted in observed depletion of the GSH/GSSG antioxidant pool and apoptotic execution (Fig. 7F–I). The data suggests that by de-lactylating SOD2, HKL may help replenish antioxidant defenses and blunt the apoptotic response, evidenced by a reduction in Cleaved Caspase-3 levels and TUNEL-positive neurons (Fig. 7G–I). These results were mirrored by the 3-TYP group, which exhibited the most catastrophic oxidative damage and neuronal loss (Fig. 7F–I). Together, these findings provide supportive in vivo evidence for the involvement of SIRT3 in the lactate-associated regulation of SOD2 lactylation, antioxidant activity, and neuronal apoptosis after TBI.

Discussion

This study systematically elucidates a novel pathological regulatory axis driven by metabolic dysfunction in TBI. We demonstrate that the TBI-induced high-lactate microenvironment is not merely a passive indicator of energy crisis, but an active pathological signal that drives the robust lactylation of the core mitochondrial antioxidant SOD2. This modification fundamentally neutralizes SOD2’s ability to scavenge superoxide anions, precipitating a catastrophic accumulation of mtROS. Crucially, we identify that this process is dynamically governed by a pair of functionally antagonistic enzymes: the resident writer AARS2 and the dynamic eraser SIRT3 (Fig. 8). Together, in vitro proteomics and in vivo pharmacological interventions depict how SOD2 lactylation acts as a molecular switch, reprogramming its interaction network from energy metabolism to protein quality control. These findings mechanistically link metabolic imbalance, PTMs, and cell fate determination, providing a profound new framework for understanding TBI pathophysiology.

Fig. 8.

Mechanism diagram of SOD2 Lactylation in TBI. TBI induces an increase in lactate (Lac) production, and lactate enters cells through monocarboxylate transporters (MCTs). Under the catalysis of AARS2, lactate undergoes lactylation modification with SOD2, inhibiting its ability to clear mtROS. The accumulation of mtROS further damages mitochondria, ultimately leading to cell death. SIRT3 can remove the lactyl groups on SOD2, restoring its enzymatic activity and thus protecting cells from damage

By identifying the precise molecular targets of lactate, our findings provide a direct mechanistic resolution to the “lactate paradox” outlined earlier. The magnitude of the post-TBI lactate surge confers a critical stoichiometric advantage that fundamentally alters the cellular PTM landscape. Under physiological conditions, the concentration of the classical acetylation donor, acetyl-CoA, ranges from ~ 10–500 μM [48], while the methylation donor SAM is ~ 1–30 μM [49]. In stark contrast, intracellular lactate conventionally exists at 500–5000 μM [50] and spikes exponentially within mitochondria following mechanical trauma. Given that lactylation and acetylation often target identical lysine residues (such as K68 and K122 on SOD2), this overwhelming concentration advantage strongly suggests a potential competitive crosstalk. We hypothesize that in the TBI microenvironment, the massive lactate pool allows lactylation to kinetically overwhelm and potentially mask classical PTM sites. While previous investigations into TBI-induced oxidative stress have predominantly attributed SOD2 inactivation to classical PTMs (e.g., acetylation or tyrosine nitration) under steady-state or milder stress [51, 52], our findings highlight lactylation as a substantial and non-negligible PTM force specifically within the lactate-enriched microenvironment that frequently accompanies severe TBI. This potential PTM crosstalk not only offers a plausible explanation for why conventional anti-acetylation therapies often fall short in acute TBI models but also highlights SOD2 lactylation as a critical driver of the acute oxidative crisis.

A key mechanistic breakthrough of this study lies in uncovering the distinct kinetic modalities governing SOD2 lactylation. By integrating APEX2 proximity labeling with traditional Co-IP, we propose a dual-regulatory model: AARS2 functions as a resident writer, constitutively positioned within the SOD2 proximitome and remaining highly sensitive to local lactate fluctuations. Conversely, SIRT3 operates as a dynamic eraser, employing a transient, hit-and-run catalytic mechanism [53, 54]. Following TBI, the explosive accumulation of lactate forces the AARS2-mediated writing process into catalytic overdrive. This process operates through simple mass action, utilizing the vast lactate pool to rapidly hyper-lactylate SOD2. In stark contrast, the transient erasing capacity of SIRT3 is intrinsically limited by its maximal catalytic velocity (V_max). Consequently, the sheer stoichiometry of the substrate-driven lactylation wave persistently outpaces SIRT3’s dynamic recruitment, leaving SOD2 trapped in an enzymatically paralyzed state and continuously draining the cellular antioxidant reserves (GSH/GSSG).

Beyond localized oxidative stress, proximity proteomics revealed that SOD2 lactylation triggers a profound cellular disaster response. We observed that structurally impaired SOD2 actively disengages from core bioenergetic pathways (e.g., oxidative phosphorylation and the TCA cycle) while simultaneously recruiting a cohort of protein quality control machinery, represented by HSPD1 and endoplasmic reticulum (ER)-associated proteins. This “Functional Triage” perfectly echoes recent conceptual advances in the field. For instance, elegant studies have recently demonstrated that following brain damage, cellular networks undergo a fundamental shift from energy metabolism toward proteostasis and autophagic clearance to mitigate acute stress [55, 56]. Our data provides the critical upstream mitochondrial trigger for this phenomenological shift: the functional reprogramming of SOD2 via lactylation prioritizes emergency damage control over long-term energy production.

Therapeutically, our in vivo data demonstrate that pharmacological activation of SIRT3 via HKL effectively reverses SOD2 lactylation and significantly blunts terminal neuronal apoptosis. While prior work in TBI and metabolic stress models has largely attributed the neuroprotective effects of SIRT3 modulators to their canonical deacetylase activity [57, 58], our study significantly expands this regulatory paradigm. We demonstrate that under the extreme metabolic crisis of TBI, mitigating the overriding lactylation storm is an equally critical, albeit highly challenging, therapeutic mechanism for SIRT3. However, we acknowledge several limitations in our study. First, due to the requirement for rapid tissue snap-freezing to preserve labile post-translational modifications and metabolites, we evaluated neuroprotection primarily at the cellular and functional levels rather than utilizing gross anatomical lesion volume assessments. Second, the number of independent animals used for several downstream molecular, biochemical, and histological analyses was relatively limited. Although 10 mice per group were included in the behavioral assessments, subsets of animals from the same cohort were allocated to different endpoint analyses, resulting in n = 3 independent animals per group for biochemical assays and Western blot/Co-IP analyses and n = 4 independent animals per group for histological quantification. This endpoint-specific allocation strategy was intended to maximize the information obtained from each animal while reducing the total number of animals used, where possible [59, 60], and similar multi-endpoint allocation strategies have been used in comparable preclinical studies [61–64]. Nevertheless, we acknowledge that these sample sizes limit the statistical power and robustness of inference for individual molecular, biochemical, and histological endpoints. Therefore, although the consistency across behavioral, biochemical, molecular, and histological readouts supports the proposed biological interpretation, these findings should be interpreted as supportive rather than definitive evidence. Future studies using larger independent animal cohorts are required to further validate the proposed SIRT3–SOD2 lactylation axis in vivo. Third, there remains a translational gap between preclinical success and clinical reality. Our pharmacological strategy relied on a pre-treatment paradigm [24], which prophylactically equips the SIRT3 pool to counter the incoming lactylation wave. Yet, in clinical post-injury administration, the AARS2-mediated writing process has already utilized the massive lactate pool to hyper-lactylate SOD2. Because SIRT3’s erasing capacity V_max is intrinsically limited, it struggles to reverse this pre-established accumulation. This profound kinetic mismatch explains why SIRT3 modulators often show robust efficacy in pre-treatment animal models but fall short in delayed clinical trials.

Despite these mechanistic insights, our study leaves several critical questions unanswered, mapping the trajectory for future investigations. First, while we identified AARS2 as a direct writer for SOD2 lactylation, it remains unclear whether AARS2 functions as a target-specific or a more universal lactylation writer. Given that other enzymes, such as p300 [33, 65], have also been reported to mediate this modification, future studies must delineate the exact division of labor and substrate specificity among these writers. Second, although our mass spectrometry identified five distinct lactylation sites on SOD2, we have yet to perform site-directed mutagenesis (e.g., lysine-to-arginine/glutamine mutants) to definitively isolate which specific residue—or combination thereof—acts as the primary driver of SOD2’s functional paralysis. Third, the reliance of SIRT3 on NAD⁺ presents a complex variable in the energy-depleted TBI brain. It remains to be determined whether the erasing bottleneck is solely constrained by SIRT3’s inherent enzyme kinetics V_max, or if it is fundamentally starved by acute post-injury NAD⁺ depletion. Consequently, exploring whether co-supplementing NAD⁺ precursors could synergistically rescue the de-lactylation process, and investigating the potential existence of other undiscovered de-lactylases, remain urgent translational hypotheses. Ultimately, while our data clearly highlights the “functional triage” of SOD2, our initial proteomic screening actually revealed a global post-injury surge in pan-cellular protein lactylation. Whether the ultimate determination of neuronal fate following TBI hinges predominantly on a single critical node (like SOD2) or is the emergent outcome of an extensively reprogrammed global protein network remains a profound open question. Addressing these multi-layered challenges—from pinpointing precise catalytic residues to deciphering global network crosstalk—will be crucial for translating this novel metabolic-PTM axis into viable clinical therapeutics.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contribution

J.S., J.L., and K.T. conceived and designed the study; J.S. performed data curation, formal analysis, methodology development, software analysis, validation, visualization, and wrote the original draft of the manuscript; J.L. contributed to conceptualization, investigation, methodology, validation, visualization, and writing of the original draft; W.F. contributed to formal analysis, investigation, validation, visualization, and manuscript revision; F.G. contributed to validation and manuscript revision; H.Y. contributed resources and manuscript revision; P.W., X.Y., and K.T. supervised the study, contributed to project administration and validation, and revised the manuscript; K.T. also acquired funding and provided resources; all authors reviewed the manuscript and approved the final version; J.S., J.L., and W.F. contributed equally to this work; P.W., X.Y., and K.T. are corresponding authors.

Funding

This work was supported by the National Natural Science Foundation of China (grant number 82101467); the Youth Project of the Department of Science and Technology of Sichuan Province (grant number 23NSFSC2757); and the General Hospital of Western Theater Command Hospital Management Project (grant number 2024-YGJC-A03).

Data Availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Material. Additional data related to this paper may be requested from the authors.

Declarations

Competing Interests

The authors declare no competing interests.