ROS-regulated SUR1-TRPM4 drives persistent activation of NLRP3 inflammasome in microglia after whole-brain radiation

Abstract

Delayed radiation-induced brain injury (RIBI) characterized by progressive cognitive decline significantly impacts patient outcomes after radiotherapy. The activation of NLRP3 inflammasome within microglia after brain radiation is involved in the progression of RIBI by mediating inflammatory responses. We have previously shown that sulfonylurea receptor 1-transient receptor potential M4 (SUR1-TRPM4) mediates microglial NLRP3-related inflammation following global brain ischemia. However, the role of SUR1-TRPM4 in RIBI remains unclear. Here, we found that whole-brain radiation induced up-regulation and assembly of SUR1-TRPM4, which further activated the NLRP3 inflammasome in microglia and caused persistent neuroinflammation in mice. Blocking SUR1-TRPM4 by glibenclamide or gene deletion of Trpm4 effectively prevented NLRP3-mediated neuroinflammation and alleviated RIBI. Utilizing the mouse model of RIBI and irradiated BV2 cells, we further demonstrated that irradiation caused mitochondrial damage to microglia, leading to violent release of reactive oxygen species (ROS), which enhanced the transcription of SUR1, TRPM4, and NLRP3 inflammasome-related molecules. Moreover, ROS up-regulated ten-eleven translocation 2 (TET2) to enhance TRPM4 expression by mediating the demethylation of the gene promoter, thereby facilitating the assembly of SUR1-TRPM4 in microglia. In summary, this study deciphers that SUR1-TRPM4 crucially mediates the persistent activation of microglial NLRP3 inflammasome under the action of ROS after whole-brain radiation, offering novel therapeutic strategies for delayed RIBI as well as other NLRP3-related neurological disorders involving excessive ROS production.

Introduction

Radiotherapy, vital for head and neck tumors, targets tumors but inevitably harms adjacent brain tissue [1, 2], causing common central nervous system (CNS) complications like radiation-induced brain injury (RIBI) [3]. Advancements in radiotherapy techniques have significantly reduced acute RIBI involving increased intracranial pressure and acute neurological syndromes. However, the incidence of delayed RIBI—brain damage induced by ionizing radiation that manifests several months to years after the completion of radiotherapy, characterized by cognitive decline and radioactive necrosis of the brain, often progressive and irreversible in nature—has been steadily rising [4, 5]. RIBI-induced cognitive decline has emerged as a crucial factor influencing radiotherapy patients’ quality of life and survival [4, 6, 7]. Currently, delayed RIBI lacks a fully understood pathogenesis, leading to limited clinical options and severe adverse effects [4, 8, 9]. Therefore, delving into the pathophysiological underpinnings of delayed RIBI and uncovering innovative therapeutic targets assume paramount clinical significance.

Tissue damage-induced chronic neuroinflammation is considered the key mechanism of delayed RIBI, driving its onset and progression [4, 10]. Microglia, the resident immune sentinels of the CNS, exhibit a heightened sensitivity to radiation and can be activated sustainedly for over 9 months after whole-brain radiation, thereby precipitating neuroinflammation by releasing inflammatory mediators [11,12,13,14]. The chronic inflammatory environment further ignites neuronal and astrocytic dysfunction, thereby constituting the primary driver of the progressive worsening of delayed RIBI [4, 10]. Notably, this neuroinflammatory milieu also impedes the differentiation of neural precursor cells into neurons, ultimately compromising neurogenesis, now recognized as the pivotal mechanism underlying chronic cognitive decline associated with RIBI [13, 14]. Numerous animal studies have consistently demonstrated that inhibiting microglia-mediated neuroinflammation yields significant improvements in functional outcomes in RIBI [14,15,16], underscoring the therapeutic potential of targeting microglia-related chronic neuroinflammation to mitigate the detrimental effects of RIBI.

The pivotal protein assemblies within the host defense arsenal, inflammasomes, orchestrate the innate inflammatory responses to tissue damage-liberated danger-associated molecular patterns (DAMPs) [17]. Among these intricate complexes, the canonical inflammasome characterized by nucleotide-binding oligomerization domain-like receptor containing pyrin domain 3 (NLRP3), triggers the conversion of the caspase-1 precursor (pro-caspase-1) into its active form, which catalyzes the proteolytic cleavage of interleukin-1β (IL-1β) and fuels excessive inflammation [18]. It has been documented that NLRP3 inflammasome activation is implicated in the radiation-induced injury of various organs by mediating inflammatory reactions [19]. Qin et al. further validate the activation of NLRP3 inflammasome in microglia after brain irradiation, establishing a strong correlation with RIBI [20]. Our previous studies have unveiled the microglial NLRP3 activation by sulfonylurea receptor 1-transient receptor potential M4 (SUR1-TRPM4) after global brain ischemia, and blocking it can effectively alleviate brain injury [21, 22]. However, the role of SUR1-TRPM4 in microglial NLRP3 inflammasome activation in the aftermath of irradiation is yet to be elucidated.

Beyond utilizing SUR1-TRPM4 blockade to curb NLRP3 activation, modulation of pertinent protein levels at appropriate timings emerges as a potentially effective therapeutic strategy. Radiation frequently leads to a surge in the production of reactive oxygen species (ROS) within tissues [19, 23], which is recognized as one of DAMPs capable of priming the NLRP3 inflammasome [24]. Furthermore, ROS up-regulate the expression of TRPM4 within the lesions [25]. Hence, we speculated that eliminating ROS in the irradiated brain might be a potential strategy to refine prevention and treatment against RIBI. DNA methylation, a pivotal epigenetic modification for transcriptional repression and genome stability, adds a methyl group to cytosine’s 5-carbon, forming 5-methylcytosine (5mC) [26, 27]. Prior studies have found that the transcriptional level of TRPM4 in diseases such as colorectal cancer is modulated by the epigenetic methylation of its gene promoter [28]. However, whether a similar regulatory mechanism exists for TRPM4 expression in the RIBI model remains unclear. TET2, a member of the ten-eleven translocation (TET) family as dioxygenase erasing gene methylation marks, is up-regulated in lipopolysaccharide-treated microglia and mediates the expression of pro-inflammatory genes [26, 27, 29]. Therefore, we hypothesized that TRPM4 expression in microglia after brain radiation might also be regulated by TET2-mediated DNA demethylation, thereby maintaining high expression for an extended period post-radiation to mediate persistent microglial NLRP3 activation.

In this study, we affirmed that SUR1-TRPM4 mediated the unremitting activation of NLRP3 inflammasome in microglia after whole-brain radiation, and blocking SUR1-TRPM4 potently alleviated delayed RIBI. Cerebral ROS after radiation stimulated the nuclear translocation of transcription factors, inducing the expression of SUR1-TRPM4 complex and NLRP3 inflammasome-related proteins in microglia. Additionally, the up-regulated TET2 by ROS formed a complex with signal transducer and activator of transcription 5 (STAT5) and bound to DNA for catalyzing promoter demethylation, enhancing the expression of TRPM4 and the assembly of SUR1-TRPM4 in microglia after whole-brain radiation. Our data shed new light on the pivotal role of SUR1-TRPM4 in microglia-mediated persistent neuroinflammation involving NLRP3 activation after whole-brain radiation. This understanding paves the way for innovative therapeutic strategies aimed at modulating the expression and activity of microglial SUR1-TRPM4, thereby offering promising avenues for treating delayed RIBI.

Materials and methods

Animals

All animal experiments have been approved by the Animal Care and Use Committee of Nanfang Hospital, Southern Medical University (Guangzhou, China), and complied with Animal Research: Reporting of In Vivo Experiments guidelines. We used male Trpm4^−/− mice on C57BL/6 background (8–10 weeks old, 20–25 g, Model Organisms Center, Shanghai, China) and wild-type littermates. Wild-type and Trpm4^−/− mice were both confirmed by extracting and separating tail DNA in agarose gel. All animals in current study were housed in a specific pathogen-free facility under a strict 12-h light/dark cycle with free access to food and water.

Cell lines culture and treatments

Murine BV2 microglial cells were purchased from American Type Culture Collection (Cat. # Bio-73434; VA, USA) and cultured in high-glucose Dulbecco’s modified Eagle’s medium (DMEM; Cat. # 12491015; Gibco, NY, USA) containing 10% fetal bovine serum (FBS; Cat. # A5256701; Gibco), 100 IU/mL penicillin (Cat. # ST486; Beyotime, Shanghai, China), and 100 mg/mL streptomycin (Cat. # ST487; Beyotime), under a humidified atmosphere of 5% CO₂ at 37 °C.

When BV2 cells reached 80-90% confluence, they were primed via radiation. The 0.5-h incubation with adenosine triphosphate (ATP, 5 mmol/L; Cat. # 20–306; Sigma-Aldrich, MO, USA) was conducted 6 h after the radiation. Additionally, the following reagents were added into the medium 0.5 h preceding the intervention with ATP: MCC950 (10 µmol/L; Cat. # HY-12815; MCE, NJ, USA) against NLRP3, 9-phenanthrol (9-Ph, 50 µmol/L; Cat. # 484-17-3; Sigma-Aldrich) against TRPM4, and sulfonylureas (100 µmol/L; Sigma-Aldrich) including glibenclamide (GLB; Cat. # 10238-21-8), gliclazide (GLZ; Cat. # 21187-98-4) and glimepiride (GLM; Cat. # 93479-97-1) against SUR1. The cells were pre-treated with N-acetylcysteine (NAC, 1 mmol/L; Cat. # S1623; Selleck, shanghai, China), the ROS scavenger, for 30 min before irradiation. Particularly, cells were pre-incubated with transcription factor inhibitors such as Mithramycin A (2 µmol/L; Cat. # HY-A0122; MCE) blocking specificity protein 1 (Sp-1) and JSH23 (30 µmol/L; Cat. # HY-13982; MCE) against nuclear factor kappa-B (NF-κB) for 1 h before challenged by radiation.

siRNA transfection

The siRNA transfection was performed as per the previous study with some modifications [29]. BV2 cells were cultured as described above. When cells reached 70-90% confluence, the pre-designed Tet2-siRNA (Cat. # S103016; Thermo Fisher Scientific, MA, USA) or non-targeting siRNA (Cat. # D-001210-02; Dharmacon, USA) was mixed with Opti-MEM™ medium (Cat. # 31985070; Thermo Fisher Scientific) thoroughly to prepare the premix solution. The premix solution was then added to the already diluted Lipofectamine™ 3000 reagent (Cat. # L3000015; Thermo Fisher Scientific) in a 1:1 ratio, followed by the incubation for 10 min. After adding the siRNA-lipid complexes to the cells, the cells were incubated at 37 °C for transfection.

RIBI Model in mice and treatments

Male C57BL/6 mice weighing 20–25 g were randomly divided into groups. The RIBI model was conducted as described in the previous study [13]. After anesthetization, the head of a prone mouse was placed in a radiation field (1 × 1 cm [2]) from the post-canthus line to the post-aurem line. Afterward, radiation was administered with a single dose of 30 Gy (3 Gy/min) employing the small animal/cell X-ray MultiRad 225 irradiator (Faxitron, AZ, USA), characterized by a source-to-skin distance of 58 cm. The mice in the sham group received anesthetic procedures at the same time as those irradiated but not whole-brain radiation.

The irradiated mice were intraperitoneally administrated 10 µg of GLB (100 µg/mL; Cat. # 10238-21-8; Sigma-Aldrich) immediately after completing the brain irradiation, followed by the maintenance dose of 10 µg once daily until euthanized [30]. In contrast, irradiated mice in the vehicle group received an equivalent volume of dimethyl sulfoxide (DMSO; Cat. # ST038; Beyotime) and saline (Fig. S1A). Bobcat339 (Cat. # HY-111558; MCE), an effective and selective cytosine-based inhibitor of TET enzymes, was dissolved in DMSO and diluted in saline at 1 mg/mL [31]. Bobcat339 (10 mg/kg), or the same volume of vehicle, was administrated by tail vein injection three days before irradiation, followed by daily infusion once a day until euthanization (Fig. S1A). NAC (300 mg/kg; Cat. # S1623; Selleck) or vehicle was administrated intraperitoneally 30 min before irradiation [32], followed by daily infusion once a day until euthanization (Fig. S1A).

Cell irradiation model

The BV2 cells were seeded into a 6-well plate and irradiated with a single dose of 10 Gy (1 Gy/min) using the X-ray MultiRad 225 irradiator, since earlier studies reported that 10 Gy was the optimal radiation dose to activate microglia [33]. After irradiation, cells were returned to the 5% CO₂ incubator. The control cells received sham-irradiation. Cells or culture supernatants were collected at indicated time points post-irradiation.

Brain tissue single-cell suspension Preparation and Gradient Centrifugation

The brain tissue single-cell suspension preparation and gradient centrifugation was performed as per the previous studies with some modifications [34, 35]. Mice were euthanized via cervical dislocation. The mouse brains were extracted and carefully moved to a Petri dish after removal of cerebellum and olfactory bulbs. The tissues were chopped into small pieces with a scissor and then incubated in 1 × Hank’s Balanced Salt Solution (HBSS; Cat. # H1025; Solarbio, Beijing, China) containing 1 mg/mL IV collagenase (Cat. # 40510ES60; Yeaseen, Shanghai, China), 3% FBS (Cat. # A5256701; Gibco) and 5 mg/mL DNase I (Cat. # D8072; Solarbio) for 45 min at 37 °C. After the termination of digestion with cold phosphate buffered saline (PBS; Cat. # P1020; Solarbio) containing 3% FBS, the cell suspension was filtered through a 70-µm cell strainer (Cat. # BS-70-CS; Biosharp, Hefei, China), followed by centrifugation at 1500 rpm for 7 min to obtain the precipitate.

The above-mentioned single-cell suspension of mouse brain was purified using a Percoll gradient centrifugation. We added 10 × PBS (Cat. # P1022; Solarbio), Percoll stock solution (Cat. # 17-0891-01; GE Healthcare, CT, USA), and RPMI medium 1640 (Cat. # 11875119; Gibco) into a centrifuge tube in a ratio of 1: 3.3: 5.7, respectively, and thoroughly mixed them to obtain a 33% (v/v) isotonic Percoll solution. The cells were resuspended in 7 mL of 33% Percoll solution and subsequently centrifuged (Centrifuge 5810R; Eppendorf, Hamburg, Germany) at 2300 rpm, with acceleration and deceleration settings both at 1, for 25 min at 25 °C, without engaging the brake. The precipitate was resuspended in 2 mL of the 1 × red blood cell lysis buffer which was prepared by diluting the 10 × red blood cell lysis buffer (Cat. # 420301; BioLegend, CA, USA) with deionized water, and then incubated at room temperature for 4 min to remove the remaining red blood cells. Finally, the reaction was terminated using 1 × PBS containing 3% FBS.

Flow Cytometry and fluorescence-activated cell sorting (FACS)

The intracellular staining for flow cytometry was performed as per the previous study with some modifications [36]. The above-mentioned rodent purified cells were pelleted at 1500 rpm for 7 min at 4 °C and resuspended in the cold buffer (1 × PBS containing 3% FBS). After a non-specific block with CD16/CD32 antibody (1: 100; Cat. # 101302; clone 93; BioLegend) for 30 min, the cell viability was confirmed by Zombie NIR™ Fixable Viability Dye (FVD, 1: 300; Cat. # 423106; BioLegend) staining for 30 min, followed by the 30-min surface labeling with specific antibodies as follows: CD45-APC (1: 100; Cat. # 103112; clone 30-F11; BioLegend), CD11b-FITC (1: 200; Cat. # 101205; clone M1/70; BioLegend), and TMEM119-PE (1: 40; Cat. # 12-6119-82; clone V3RT1Gosz; Thermo Fisher Scientific). Notably, all incubation steps during the flow cytometry process were conducted at room temperature, and the cells were washed at least once after each incubation step to remove any unbound dye from the system.

After centrifugation, the cells were resuspended in 100 µL of working solution for fixation and permeabilization, which was prepared by mixing the concentrate and diluent of a commercial kit (Cat. # 00-5521-00; Thermo Fisher Scientific) in a ratio of 1: 3. Following a 30-min fixation and permeabilization, the cells were incubated for 30 min with a mixture containing the following two primary antibodies, rabbit anti-NLRP3 (1: 50; Cat. # ab263899; clone EPR23094-1; Abcam, Cambridge, UK) and mouse anti-caspase-1 p20 (1: 100; Cat. # sc-398715; clone D-4; Santa Cruz, CA, USA). Finally, the cells were incubated for 30 min with the secondary antibodies conjugated with Qdot™ 605 (1: 50; Cat. # Q11402MP; Thermo Fisher Scientific) and BV421 (1: 50; Cat. # 405317; clone Poly4053; BioLegend), respectively. The samples were analyzed on an LSRII/Fortessa flow cytometer (BD Biosciences, Heidelberg, Germany). Gates were established using antibody isotype controls (provided by manufacturers) and fluorescence minus one (FMO) controls. The data analysis was performed using FlowJo V10 software (TreeStar, OR, USA).

FACS was also conducted following the aforementioned gating strategy. Utilizing a specially ordered BD FACSAria III with a 100 μm nozzle and operating at 30 psi, the CD45⁺CD11b⁺TMEM119⁺ cells from mouse brains were sorted for subsequent co-immunoprecipitation (Co-IP), western blotting, and quantitative real-time polymerase chain reaction (qRT-PCR). Approximately 200,000 to 300,000 cells were sorted from the brain tissue of each mouse. Cells derived from 3 to 5 mice within the same group were pooled to form a single sample, containing approximately 900,000 cells. The protein concentration obtained from these samples was roughly 1–2 µg/µL, while the RNA concentration was approximately 20–40 ng/µL.

Western blotting

Western blotting was routinely performed as previously reported [37]. Rodent brain tissues, sorted cells or BV2 cells were homogenized in RIPA lysis buffer (Cat. # P0013; Beyotime) containing protease inhibitor cocktail. After denatured in loading buffer, the samples were subjected to SDS-PAGE and then transferred to PVDF membranes (Millipore, MA, USA). After blocked by 5% non-fat milk, the membranes were incubated overnight at 4 °C with the primary antibodies as below: Mouse anti-β-actin (1: 20000; Cat. # 66009-1-Ig; clone 2D4H5; Proteintech, IL, USA), rabbit anti-GAPDH (1: 20000; Cat. # 81640-5-RR; clone 1H18; Proteintech), rabbit anti-Histone H3 (1: 2500; Cat. # ab1791; Abcam), rabbit anti-TRPM4 (1: 500; Cat. # SAB2102584; Sigma-Aldrich), rabbit anti-SUR1 (1: 500; Cat. # ab217633; Abcam), rabbit anti-NLRP3 (1: 500; Cat. # NBP2-12446; Novus, CO, USA), rabbit anti-pro-caspase-1 (1: 500; Cat. # ab138483; Abcam), rabbit anti-caspase-1 p20 (1: 500; Cat. # bs-10442R; Bioss, Beijing, China), rabbit anti-precursor of IL-1β (pro-IL-1β; 1: 1000; Cat. # 26048-1-AP; Proteintech), rabbit anti-IL-1β p17 (1: 1000; Cat. # NB600-633; Novus), rabbit anti-TET2 (1: 1000; Cat. # ab309481; clone EPR26694-93; Abcam), rabbit anti-Sp-1 (1: 2000; Cat. # 21962-1-AP; Proteintech), and rabbit anti-NF-κB p65 (1: 3000; Cat. # 80979-1-RR; clone 4C7; Proteintech). After washing, the membranes were incubated for 60 min with secondary antibodies including goat anti-mouse IgG (1: 10000; Cat. # 91196 S; CST, MA, USA) or goat anti-rabbit IgG conjugated with HRP (1: 10000; Cat. # 7074P2; CST). The densities of protein blots were quantified by ImageJ software (NIH, MD, USA) and normalized to the level of β-actin, GAPDH or Histone H3.

Measurement of gene expression

The mRNA levels of SUR1, TRPM4, NLRP3, caspase-1, TET2, Tumor necrosis factor alpha (TNF-α), IL-1β, IL-6, IL-18, inwardly rectifying potassium channel 6.1 (Kir6.1), Kir6.2, and GAPDH were routinely measured by qRT-PCR [38]. Briefly, total RNA was isolated using Trizol Reagent (Cat. # 15596018CN; Thermo Fisher Scientific) and reverse transcribed to cDNA with the PrimeScript™ RT Master Mix Kit (Cat. # RR036A; Takara, Dalian, China) according to the manufacturer’s instructions. qRT-PCR was performed using the SYBR Green master mixes (Cat. # A25778; Takara) and Roche LightCycler480 System. Relative changes of mRNA expression were normalized to the level of GAPDH. The specific primers for qRT-PCR were listed in Table S1.

Co-IP

This assay was performed in line with our previous study [22]. Cells were lysed in moderate lysis buffer (Cat. # E125-01; Genstar, Beijing, China) containing phenylmethylsulfonyl fluoride. After centrifuged at 14,000 g for 15 min, the supernatant was obtained and quantified, of which the portion containing 500 µg protein was immunoprecipitated with 1 µg rabbit anti-NLRP3 (1: 30; Cat. # ab263899; clone EPR23094-1; Abcam) or rabbit anti-TET2 (1: 30; Cat. # ab309481; clone EPR26694-93; Abcam) under rotation overnight at 4 °C. Following the rotary incubation with 30 µL protein A/G magnetic beads (Cat. # B23202; Bimake, TX, USA) for another 20 min at room temperature and subsequent magnetic separation, the complexes were washed 5 times, and resuspended in 40 µL loading buffer prior to denatured at 100 °C. Finally, the protein complexes were subjected to western blotting as described above, and the employed antibodies were as follows: rat anti-NLRP3 (1: 250; Cat. # MA5-23919; clone 768319; Thermo Fisher Scientific), rat anti-pro-caspase-1 (1: 500; Cat. # 14-9832-82; clone 5B10; Thermo Fisher Scientific), rabbit anti-TET2 (1: 1000; Cat. # ab309481; clone EPR26694-93; Abcam), and rabbit anti-phospho-STAT5 (pSTAT5; 1: 1000; Cat. # 44-390G; Thermo Fisher Scientific). Whole cell lysates were used as an input control, and homophytic IgG was used as a negative control.

Caspase-1 activity detection

BV2 cells were seeded in 96-well plates, of which the activated caspase-1 was evaluated utilizing a FAM-FLICA detection kit (Cat. # ICT-98; Immunochemistry Technologies, MN, USA) as described previously with some adjustments [39]. The fluorescent probe FAM-YVAD-FMK (FLICA) was employed to irreversibly label in situ activated caspase-1 in living cells, and the green fluorescent signal directly reflected the caspase-1 activity at the time the reagent was added. Briefly, following the reconstitution with DMSO, FLICA was diluted with PBS, added to each sample and incubated for 1 h prior to the nuclear staining with Hoechst. FLICA excited from 490 to 495 nm and emitted from 515 to 525 nm. The images were captured with a fluorescence microscope and the fluorescence intensity of cleaved caspase-1 was quantified by the multiscan spectrum (BMG, Offenburg, Germany).

Biochemical analysis

After irradiation, the mouse brain tissues or BV2 cells were homogenized in ice-cold normal saline, frozen at -20℃ for 5 min, and centrifuged at 4000 g for 15 min. Thereafter, the supernatant was collected into fresh tubes for evaluation. The following data were obtained using commercial assay kits (Nanjing Jiancheng Biological Product, Nanjing, China) as previously described: the content of lipid peroxidation product malondialdehyde (MDA; Cat. # A003-1) and the activity of superoxide dismutase (SOD; Cat. # A001-3) and catalase (CAT; Cat. # A007-1), reflecting the level of intracellular oxidative stress damage [40,41,42].

Measurement of ROS production

After single-cell suspension preparation and gradient centrifugation, the purified cells from mouse brain were incubated in 100 µL of diluted DCFH-DA fluorescence probe (10 µM; Cat. # S0033S; Beyotime) for 30 min at 37℃, with an addition of multiple antibodies against surface markers, including CD45-APC (1: 100; Cat. # 103112; clone 30-F11; BioLegend), CD11b-FITC (1: 200; Cat. # 101205; clone M1/70; BioLegend), and TMEM119-PE (1: 40; Cat. # 12-6119-82; clone V3RT1Gosz; Thermo Fisher Scientific). Finally, the samples were analyzed on an LSRII/Fortessa flow cytometer (BD Biosciences).

BV2 cells were seeded on 6-well plates and pre-incubated for 24 h. After radiation and drug treatment, the cell culture medium was removed, and the DCFH-DA working solution was added to stain the cells for 20 min at 37℃. The cells were then washed three times and observed under a fluorescence microscope (Olympus, Tokyo, Japan). Quantification was performed in images after appropriate thresholding.

Evaluation of mitochondrial state

BV2 cells were seeded on 6-well plates and pre-incubated for 24 h. MitoTracker^® Green FM (Cat. # 9074 S; CST) was reconstituted with DMSO to prepare 1 mM stock solution, and further diluted with medium to prepare 100 nM working solution containing 10 µg/mL Hoechst 33,342 (Cat. # C0030; Solarbio). After radiation, the culture medium was removed, and the working solution was added at indicated time points to stain the cells for 20 min at 37℃. Finally, the cells were washed and observed under confocal microscope (Olympus). Quantification was performed in images after appropriate thresholding using the ImageJ software.

Detection of mitochondrial membrane potential

BV2 cells were seeded on 6-well plates and pre-incubated for 24 h. JC-1 staining (Cat. # C2006; Beyotime) was performed to evaluate changes of mitochondrial membrane potential. The JC-1 staining working solution was prepared by adding 5 µL of 200 × JC-1 and 0.2 mL of 5 × staining buffer to 0.8 mL of ultrapure water, with an addition of 2 µL of Hoechst 33,342 (Cat. # C0030; Solarbio). After radiation, the culture medium was removed at indicated time points. 1 mL of culture medium and 1 mL of JC-1 staining working solution were added and mixed thoroughly to stain the cells for 20 min at 37℃. During incubation, 10 mL of washing buffer was prepared by diluting 5 × staining buffer and then placed on ice. After staining, the cells were washed twice and observed under confocal microscope (Olympus). Quantification was performed in images after appropriate thresholding using the ImageJ software.

Detection of mitochondrial ROS (mROS)

MitoSOX™ staining (Cat. # M36008; Thermo Fisher Scientific) was performed for detecting the production of mROS. In brief, MitoSOX™ was reconstituted with DMSO to prepare 5mM stock solution. 1 µL of stock solution and 2 µL of Hoechst 33,342 (Cat. # C0030; Solarbio) were then added to 2 mL of HBSS to obtain the staining working solution. After radiation, the culture medium was removed at indicated time points. 1 mL of staining working solution was added to stain the cells for 30 min at 37℃. Finally, the cells were washed twice with HBSS and then observed under confocal microscope (Olympus). Quantification was performed in images after appropriate thresholding using the ImageJ software.

Measurement of intracellular ATP

BV2 cells were cultured in a 96-well plate and allowed to grow overnight. After treatment, the intracellular ATP level was measured using CellTiter-Glo luminescent cell viability assay kit (Cat. # G7571; Promega, WI, USA) as previously reported [43]. The luminescence was measured to calculate the ATP concentration using the standard curve.

Histological examination

Mice were euthanized 1 week, 4 weeks, or 8 weeks post-radiation, respectively. Following fixation with 4% paraformaldehyde (PFA; Cat. # P0099; Beyotime), brains were immersed in 15% and 30% sucrose (PFA; Cat. # ST1672; Beyotime) at 4 °C for cryoprotection. With regard to the BV2 cells, they were fixed with 2% PFA for 10 min at room temperature. For detecting neuronal loss and the levels of certain markers, the mouse brain slides were incubated overnight at 4 °C with antibodies against neuronal nuclei (NeuN; 1: 50; Cat. # 24307T; clone D4G4O; CST) for neurons, glial fibrillary acidic protein (GFAP; 1: 250; Cat. # ab68428; clone EPR1034Y; Abcam) for astrocytes, ionized calcium-binding adapter molecule-1 (Iba-1; 1: 200; Cat. # NBP2-19019; Novus) for microglia, and doublecortin (DCX; 1: 250; Cat. # ab207175; clone EPR19997; Abcam) for immature neurons.

To evaluate the polarization of microglia and the localization of SUR1-TRPM4 in microglia after irradiation, the slides of brain sections and cells were incubated overnight at 4 °C with antibodies against inducible nitric oxide synthase (iNOS; 1: 200; Cat. # 18985-1-AP; Proteintech), CD206 (1: 200; Cat. # 18704-1-AP; Proteintech), TRPM4 (1: 50; Cat. # PA5-116483; Thermo Fisher Scientific), SUR1 (1: 50; Cat. # MA5-27660; clone N289/16; Thermo Fisher Scientific), and Iba-1 (1: 50; Cat. # PA5-18039; Thermo Fisher Scientific).

After diluting the secondary antibodies at a 1:200 ratio, the slides were thoroughly washed and then incubated for 60 min at 37 °C with goat anti-rabbit IgG (Cat. # ab150077; conjugation Alexa Fluor^® 488; Abcam), donkey anti-rabbit IgG (Cat. # ab150075; conjugation Alexa Fluor^® 647; Abcam), donkey anti-goat IgG (Cat. # ab150131; conjugation Alexa Fluor^® 647; Abcam), donkey anti-goat IgG (Cat. # ab150129; conjugation Alexa Fluor^® 488; Abcam), or goat anti-mouse IgG (Cat. # ab6563; conjugation Cy5 ^®; Abcam). Subsequently, the slides were counterstained with DAPI in the dark. Finally, fluorescent signaling was observed under confocal microscope (Olympus).

Magnetic resonance imaging (MRI) scanning

The 7.0 T nuclear magnetic resonance scanner (Bruker Biospin, Ettlingen, Germany) was utilized to scan the consecutive 0.5-mm thick coronal and horizontal T2-weighted images (T2WI) of mouse brain. Mice were anesthetized with isoflurane (5% induction, 1–2% maintenance) before the prostration on a custom-made holder with strapping to minimize head motion. T2WI were obtained using a 2-dimensional turbo spin echo sequence (repetition time/echo time, 1600 ms /125 ms; flip angle, 90°; field of view, 40 mm; matrix, 200/400 r; echo planar imaging factor, 1; turbo spin echo factor, 10; number of signal averages, 14; total scan time, 11 min) [13].

The presence or absence of brain swelling was assessed by observing any possible structural change, midline shift and signal changes on coronal T2WI. In addition, an enlarged lateral ventricle indicates cerebral cortex atrophy [13]. Therefore, we measured the total volume of the lateral ventricles on horizontal T2WI [44]. Briefly, we manually segmented the bilateral ventricles from T2WI. The bilateral ventricles were outlined on each slice, and the areas were measured using ImageJ software. Ventricular volume was calculated by summing the ventricle areas across all slices showing the lateral ventricles and multiplying by the slice thickness (0.5 mm). All measurements were repeated three times, and the mean value was used for the final calculation.

Evans blue extraction test

Blood-brain barrier (BBB) permeability was assessed utilizing Evans blue (Cat. # 314-13-6; Sigma Aldrich), a diazo blue dye with a molecular weight of 960 D [45]. When injected intravenously, Evans blue binds to plasma albumins (molecular weight, ≈ 68 kDa) and becomes fluorescent. While small-molecule tracers like fluorescein are indeed more sensitive to changes in BBB permeability and can detect subtle alterations reflecting early BBB changes, studies have found that the exudation of sodium fluorescein into the brain tissue largely depends on its plasma concentration, posing a potential risk for objectively assessing alterations in BBB permeability [46,47,48]. Furthermore, the leakage of small-molecule tracers may not necessarily be accompanied by significant cerebral edema, as such tracers can pass through the BBB even when it is slightly altered or even normal [48, 49]. In contrast, Evans blue leakage often indicates more severe BBB damage and is often accompanied by evident vasogenic edema, as the dye, once bound to plasma albumin to form a high molecular complex, cannot easily cross the BBB. Since our aim in this study was to investigate whether there was significant cerebral edema after whole-brain radiation, thereby determining whether cerebral edema plays a crucial role in the pathophysiological process of RIBI, we chose Evans blue as the tracer to assess whether there was notable BBB damage and cerebral edema in the irradiated mouse brain.

Evans blue (2% in normal saline, 4 mL/kg) was injected through tail vein at 1 h before sacrifice. 20 min post-injection, the mice were deeply anesthetized with 1% sodium pentobarbital and perfused transcardially with normal saline to eliminate blood and intravascular Evans blue dye until the drainage was colorless. After that, the hemispheres were excised, weighed, and then minced into small pieces. These pieces were immersed in a formamide (Cat. # 75-12-7; Sigma Aldrich) solution at 60 °C for 24 h to extract Evans blue. Following extraction, the brain-containing solution was centrifuged at 2000 rpm for 10 min to obtain the supernatant. The Evans blue concentration in the supernatant was calculated using a microplate reader (SpectraMax M5, Molecular Devices, USA) at a wavelength of 632 nm. The degree of Evans blue leakage was expressed as µg per gram brain tissue.

Cerebral water content calculation

After irradiation, brain samples were weighed and dried in an oven at 60 °C for 7 days and weighed to determine the dry weight. Brain water content (%) was calculated as (wet weight-dry weight) / wet weight × 100%.

Behavioral testing

The Morris Water Maze was carried out to evaluate short-term spatial learning and memory as described previously [50]. The water maze apparatus is a circular tank filled with opaque water (divided into four quadrants, called Q1, Q2, Q3, Q4), and a hidden platform is submerged 1 cm below the water surface and not visible to the rodents. Firstly, the rodents were trained to search for the platform for 5 consecutive days after 8 weeks post-radiation at a frequency of four trials/day, orienting by referencing 3 external cues surrounding the tank. If the rodents did not find the platform in 60 s, they were manually placed on it for 15 s. Rodents’ movements were tracked by TSE VideoMot2 tracking system (Bad Homburg, Germany) to record the path and latency time taken to escape from 4 randomly assigned locations. After the training termed as acquisition trial, the probe trial was performed on the following day, when the rodents were allowed 60 s to explore the platform which has been removed. The percentage of total time that rodents spent in the target quadrant and the number of platform location crossings were recorded and analyzed.

Dot blot

Genomic DNA was isolated using the QIAamp DNA Mini Kit (Cat. # 51304; Qiagen, USA), quantified with the NanoDrop 2000 (Thermo Fisher Scientific), and then diluted to the same concentration with ddH₂O. The extracted DNA was denatured at 95 °C for 10 min and carefully dropped onto NC membranes (Cat. # 66485; Pall Corporation, New York, USA). After drying at 80 °C for 15 min, each side of the membranes was cross-linked with UV irradiation for 15 min. The blots were blocked with 5% Bovine Serum Albumin (BSA; Cat. # ST2249; Beyotime) at room temperature for 1 h, followed by overnight incubation with anti-5-mC (1: 10000; Cat. # 39649; clone 33D3; Active Motif, USA) or anti-5-hydroxymethylcytosine (5-hmC; 1: 10000; Cat. # 39092; Active Motif) at 4 °C. Subsequently, the blots were incubated for 2 h with the secondary antibodies used in western blotting as described above. DNA was detected using a Bio-Rad gel imaging system (Bio-Rad, USA).

Database

JASPAR database (http://jaspar.genereg.net) was used to predict the potential transcription factors that mediated the transcription of Tet2 or Trpm4 [51], showing the possible binding sites for transcription factors to the promoters of genes.

Chromatin immunoprecipitation (ChIP)

ChIP was carried out with the SimpleChIP^® Plus Enzymatic Chromatin IP Kit (Cat. # 9005 S; CST), following the manufacturer’s recommended instructions. Genomic DNA and proteins were cross-linked by adding formaldehyde to a final concentration of 1% for 10 min at room temperature. After glycine treatment, cell chromatin was sonicated to generate fragments ranging from 150 to 900 bp. These fragments were then immunoprecipitated using antibodies targeting TET2 (1: 50; Cat. # 92529 S; clone D9K3E; CST), NF-κB p65 (1: 100; Cat. # 8242 S; clone D14E12; CST), pSTAT5 (1: 100; Cat. # 44-390G; Thermo Fisher Scientific), or an isotype control normal rabbit IgG. The ChIP-enriched DNA was subsequently analyzed by qRT-PCR. The specific primers for ChIP were listed in Table S1.

Statistical analysis

Preliminary analysis of data normality was performed with Shapiro–Wilk’s test. All data were presented as means ± SD. The comparison of data among multiple groups was performed by one-way ANOVA followed by Tukey’s post hoc multiple comparison tests. The data of escape latency in the water maze test were analyzed with repeated-measures ANOVA comprising treatment groups, time points, and groups × time interaction, followed by Tukey’s post hoc multiple comparison tests. SPSS 20.0 (IBM, NY, USA) and GraphPad Prism 8.0 (GraphPad, CA, USA) were used for statistical analyses. P < 0.05 was considered statistically significant.

Results

RIBI may be closely related to the inordinate activation of pro-inflammatory microglia

We first evaluated the level of neuroinflammation after whole-brain radiation in the RIBI mouse model. According to the different experimental purposes, mice subjected to RIBI modeling (n = 235) or sham surgery (n = 70) were randomly grouped (Fig. S1B). Continuous neuronal loss in the CA1 region of the hippocampus was observed after whole-brain radiation and only a small number of neurons remained at 2 months after irradiation (Fig. 1A). However, no apparent changes in signal on T2WI or brain water content were found, indicating no significant brain edema observed within 2 months after irradiation (Fig. 1B-C). Furthermore, no significant disruption of BBB was observed within 2 months after whole-brain radiation, as indicated by the results of Evans blue leakage detection in the brain (Fig. 1D). These results suggest that RIBI, at least within 2 months post-radiation, may be attributed to other pathophysiological processes beyond brain edema.

Fig. 1

Test of brain edema and BBB permeability in the RIBI model. (A) The neuropathological damage in the CA1 region of the hippocampus after whole-brain radiation, characterized by the changes in immunofluorescent staining for NeuN, Iba-1, and GFAP. (B) MRI scanning between sham and whole-brain radiation group. (C) Cerebral water content calculation between sham and whole-brain radiation group. (D) Evans blue extraction test between sham and whole-brain radiation group. 7 d, 7 days. 1 m, 1 month. 2 m, 2 months. Scale bar = 100 μm. Data are represented as mean ± SD. *P < 0.05. n = 5 per group

Extensive pieces of evidence underscore that ionizing radiation triggers the sustained activation of microglia in the brain and the release of inflammatory mediators, and consequently perpetuates neuroinflammation, ultimately culminating in cell death or functional abnormalities within the brain [4, 23]. In line with previous studies, our findings revealed that within several months following RIBI modeling, there was a marked activation of astrocytes and microglia in the hippocampus of mice (Fig. 1A). Notably, microglia within the hippocampus exhibited a marked upregulation in the expression of the pro-inflammatory marker iNOS but not anti-inflammatory marker CD206, accompanied by a synchronized up-regulation of various inflammatory factors including TNF-α, IL-1β, IL-6, and IL-18 within the brain (Fig. 2A-C). Concordantly, irradiated BV2 cells also displayed a pronounced pro-inflammatory phenotype (iNOS⁺) in vitro, with the up-regulation of multiple inflammatory factors (Fig. 2D-E). In sum, the aforementioned results imply that whole-brain radiation can prompt the persistent transition of microglia towards a pro-inflammatory state and thereby elicit unremitting neuroinflammation, which is intimately associated with RIBI.

Fig. 2

Whole-brain radiation prompts the transition of microglia towards a pro-inflammatory state and elicits neuroinflammation. (A) Confocal analysis to detect the co-localization of pro-inflammatory or anti-inflammatory markers with Iba-1 within hippocampus after RIBI modeling. (B-C) qRT-PCR results showing the mRNA levels of various inflammatory factors including TNF-α, IL-1β, IL-6, and IL-18 within the hippocampus or cortex after radiation. (D) Confocal analysis to detect the expression of pro-inflammatory or anti-inflammatory markers in the irradiated BV2 cells. (E) qRT-PCR results showing the mRNA levels of various inflammatory factors including TNF-α, IL-1β, IL-6, and IL-18 within irradiated BV2 cells. 1 d, 1 day. 3 d, 3 days. 7 d, 7 days. 1 m, 1 month. 2 m, 2 months. Scale bar = 50 μm. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 versus sham or control. n = 3 or 5 per group

SUR1-TRPM4 mediates the activation of NLRP3 inflammasome in microglia after brain radiation

To delve into the role of NLRP3 inflammasome in inflammatory brain injury after radiation, we first verified whether NLRP3 inflammasome activation occurred following radiation, by measuring the levels of NLRP3, caspase-1 p20 and IL-1β p17 fragments. The results demonstrated that the levels of NLRP3, pro-caspase-1, and caspase-1 p20 from irradiated mouse brains were significantly elevated at 6 h and plateaued for at least 7 days after radiation, while the levels of pro-IL-1β and IL-1β p17 peaked on day 1 after radiation, followed by a gradual decline until day 7 when the post-radiation levels remained remarkably augmented in comparison with the sham group (Fig. 3A-B). The flow cytometry results also verified that the expression levels of NLRP3 and caspase-1 p20 in microglia within the brain of post-RIBI mice were significantly increased (Fig. 3C-D). Moreover, Co-IP further confirmed the distinct interaction between NLRP3 and pro-caspase-1 in microglia within the brain of post-modeling mice (Fig. 3E). These changes imply that whole-brain radiation contributes to the assembly and activation of NLRP3 inflammasome in microglia.

Our recent research has unveiled a pivotal driving role of SUR1-TRPM4 in the activation of NLRP3 inflammasomes within microglia following cerebral ischemia [21]. Interestingly, the protein levels of SUR1 and TRPM4 were also significantly up-regulated after whole-brain radiation, and the trends were consistent with that of the NLRP3 inflammasome (Fig. 3A-B). In addition, fluorescent staining of mice hippocampus performed on day 7 post-radiation showed that SUR1 and TRPM4 localized massively in Iba-1-positive cells (Fig. 3F), suggesting a mass of assembly of SUR1-TRPM4 complex in microglia after irradiation. The levels of inflammasome-related proteins in mouse brains were further tested following genetic silencing of Trpm4 or pharmacological blockade of the SUR1-TRPM4 channel by GLB (Fig. 3G-H). As expected, we observed that the levels of caspase-1 p20 and IL-1β p17 were dramatically augmented on day 7 post-radiation but recovered in the presence of GLB or in Trpm4^−/− mice. Notably, no hypoglycemia was detected and there was also no significant difference in blood glucose levels among the 4 experimental groups at baseline and at indicated time points after whole-brain radiation (Table S2). In sum, the above results elucidate that brain radiation fuels the activation of microglial NLRP3 inflammasome, which can be reversed by blocking SUR1-TRPM4.

Fig. 3

SUR1-TRPM4 mediates the activation of NLRP3 inflammasome in microglia after brain radiation. (A-B) Western blotting results of SUR1, TRPM4, and NLRP3 inflammasome-related molecules within the brain tissue of sham-operated and irradiated mice at different time points after brain radiation. (C-D) The gating strategy (C) and histogram (D) of flow cytometry to indicate the dynamic changes in the expression of NLRP3 and caspase-1 p20 within microglia (CD45⁺CD11b⁺TMEM119⁺) after RIBI modeling. The FMO negative controls consisted of CD45⁺CD11b⁺TMEM119⁺ cells from a sample in which only the NLRP3 or caspase-1 p20 antibody was omitted. (E) The association of NLRP3 and caspase-1 detected by Co-IP 7 days after brain radiation, in the sorted microglia within brain by FACS according to the gating strategy in (C). Samples are immunoprecipitated with isotype IgG as a control. (F) Confocal analysis of TRPM4, SUR1, and Iba-1 staining within hippocampus 7 days after brain radiation. (G-H) Western blotting results of TRPM4, caspase-1 p20, and IL-1β p17 from wild type or Trpm4^−/− mouse brain on day 7 after undergoing different treatments. 6 h, 6 h. 1 d, 1 day. 3 d, 3 days. 7 d, 7 days. MFI, mean fluorescence intensity. Rad, radiation. Scale bar = 50 μm. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 versus sham. #P < 0.05, ##P < 0.01, ###P < 0.001 versus wild type vehicle. n = 5 per group

SUR1-TRPM4 mediates the activation of NLRP3 inflammasome in BV2 cells challenged with irradiation

We then sought to validate the connection between SUR1-TRPM4 and NLRP3 inflammasome in radiation-primed BV2 cells. Our recent studies, leveraging an in vitro model of ischemia-reperfusion injury coupled with ATP stimulation, have demonstrated that while damage stimuli serve as the priming signal to promote the expression of NLRP3 inflammasome-associated proteins, ATP acts as the activating signal to induce the assembly and activation of the SUR1-TRPM4-mediated NLRP3 inflammasome [21]. Consequently, in this research, we have continued to utilize the in vitro model of irradiation injury combined with ATP stimulation for further exploration. The fluorescence in Fig. 4A showed the obvious formation of the SUR1-TRPM4 channel after modeling, as evidenced by the highly consistent localization of SUR1 and TRPM4. We also found a distinct interaction between NLRP3 and pro-caspase-1 in irradiated BV2 cells treated with ATP (Fig. 4B). Afterwards, the increment of caspase-1 activity was reflected by the higher fluorescent intensity and increased caspase-1 p20 in post-modeling BV2 cells treated with ATP (Fig. 4C-F). Likewise, the ATP-induced release of IL-1β into supernatant was also observed via Western blotting. Nevertheless, the up-regulation of caspase-1 cleavage and mature IL-1β release was eliminated through intervening in NLRP3 by MCC950 (Fig. 4C-F), suggesting that ATP mainly activates canonical NLRP3 inflammasome rather than non-NLRP3 inflammasomes. To ascertain whether ATP-induced NLRP3 activation is associated with its function in regulating the opening of SUR1-TRPM4, which has been previously reported by us [21], we detected the expression of relevant molecules in ATP-treated BV2 cells challenged with radiation when exposed to the SUR1 or TRPM4 inhibitors. The results revealed that the NLRP3 inflammasome activation was abolished by 9-Ph or sulfonylureas, including GLB, GLM, and GLZ, as evaluated by the level of caspase-1 p20 and supernatant IL-1β returning to baseline (Fig. 4G-H). Taken together, these findings indicate the regulatory role of SUR1-TRPM4 in ATP-induced NLRP3 inflammasome activation in irradiated microglia, further suggesting that blocking SUR1-TRPM4 may be an effective strategy for inhibiting unremitting microglial NLRP3 activation after radiation.

Fig. 4

SUR1-TRPM4 mediates the activation of NLRP3 inflammasome in BV2 cells challenged with irradiation. (A) Confocal analysis of TRPM4 and SUR1 staining in radiation-primed BV2 cells. (B) The association of NLRP3 and caspase-1 detected by Co-IP, in radiation-primed BV2 cells with or without ATP. Samples are immunoprecipitated with isotype IgG as a control. (C-D) The immunofluorescence reflecting the caspase-1 activity in radiation-primed BV2 cells challenged with ATP with or without MCC950. (E-F) The protein expression levels of caspase-1 p20 and supernatant IL-1β reflecting the NLRP3 activation in radiation-primed BV2 cells challenged with ATP with or without MCC950. (G-H) Western blotting results manifesting the levels of NLRP3 inflammasome activation in radiation-primed BV2 cells challenged with ATP, in the presence of indicated SUR1 or TRPM4 inhibitors. Rad, radiation. Scale bar = 50 μm. Data are represented as mean ± SD. ***P < 0.001 versus control. ###P < 0.001 versus radiation-primed BV2 cells challenged with ATP. n = 3 per group

Inhibiting SUR1-TRPM4 with GLB improves delayed brain atrophy and cognitive impairment after radiation

A wealth of evidence indicates that persistent neuroinflammation serves as the primary cause of delayed RIBI [4]. Consequently, we further investigated whether the anti-inflammatory effects of GLB ameliorated post-radiation brain damage. We first conducted MRI scanning to evaluate the severity of cerebral cortex atrophy by measuring the volume of lateral ventricles of indicated groups 8 weeks post-radiation [13]. As illustrated in Fig. 5A, the lateral ventricles were prominently enlarged in irradiated mice compared to the sham group, but the dilated lateral ventricles were partly restored in the irradiated Trpm4^−/− mice with or without GLB, or the irradiated wild-type mice treated with GLB, indicating that the radiation-induced brain atrophy is mitigated via closing the SUR1-TRPM4 channel with GLB.

The mice above were next used to perform the Morris Water Maze test. During the acquisition trial, the escape latency (F = 98.459, P < 0.001) and the distance to reach the platform (F = 67.979, P < 0.001) became progressively shorter in all groups over time (Fig. 5B), and repeated-measures ANOVA revealed that the escape latency (F = 43.511, P < 0.001) and the distance to reach the platform (F = 46.882, P < 0.001) differed significantly among the groups, with no significant interaction between groups and time points. Furthermore, Tukey’s post hoc analysis revealed that pharmacological or genetic suppression of SUR1-TRPM4 significantly reduced both the latency and distance required to locate the hidden platform during the final 2 training days. Likewise, the frequency of crossing the platform area in the probe trial was increased when SUR1-TRPM4 was inhibited, in comparison with the irradiated placebo-control mice (Fig. 5C-D). Overall, these findings render SUR1-TRPM4 a key trigger of post-radiation brain atrophy and cognitive impairment, suggesting anti-inflammatory neuroprotection of GLB on delayed RIBI.

Targeting SUR1-TRPM4 with GLB ameliorates neuropathologic injury in the RIBI model

The sustained activation of microglia, which creates a chronic neuroinflammatory microenvironment, leads to dysfunctional neurons and astrocytes, serving as the primary reason for the progressive exacerbation of delayed RIBI [4, 10]. Notably, this inflammatory microenvironment in the hippocampus inhibits the differentiation of neural precursor cells into neurons, resulting in impaired neural regeneration, a mechanism currently regarded as the main contributor to chronic cognitive decline associated with RIBI [52]. Here, immunofluorescent staining in mice hippocampal DG region and cortex showed significantly more Iba-1-positive cells 8 weeks post-radiation, compared to the sham controls (Fig. 5E and G). However, Iba-1-positive cells were dramatically reduced in the irradiated Trpm4^−/− mice in the presence or absence of GLB, or irradiated wild-type mice with GLB. Likewise, intervening in SUR1-TRPM4 prevented the upward tendency of GFAP-positive cells in the hippocampus after radiation. Moreover, more DCX⁺ immature neurons in the hippocampal DG region were reserved by abrogating the SUR1-TRPM4 channel in the RIBI model (Fig. 5F-G). Taken together, these data indicate that GLB can effectively inhibit glial cell activation and rescue immature neurons after whole-brain radiation by blocking SUR1-TRPM4, which may underlie the potential mechanism of GLB to improve delayed cognitive dysfunction through its anti-inflammatory effects.

Fig. 5

Inhibiting SUR1-TRPM4 with GLB ameliorates brain atrophy, cognitive impairment and neuropathological damage in RIBI model. (A) The MRI scanning results to evaluate the severity of brain atrophy and the volume of lateral ventricles of indicated groups 8 weeks post-radiation. (B-D) The effect of blocking SUR1-TRPM4 on post-brain radiation spatial memory and learning deficits assessed by Morris Water Maze analysis 8 weeks post-radiation, including the mean latency and distance to reach the hiding platform (B), and the frequency of crossing the platform area (C) during the probe trial (D). (E-G) The effect of blocking SUR1-TRPM4 on post-brain radiation neuropathological damage characterized by the changes in immunofluorescent staining for Iba-1 in hippocampal DG region and cortex (E), DCX in hippocampal DG region, and GFAP in hippocampus (F). Scale bar = 100 μm. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 versus wild type sham; #P < 0.05, ##P < 0.01, ###P < 0.001 versus wild type vehicle. n = 5 or 10 per group

Microglia may trigger oxidative stress by producing ROS in the RIBI model

We endeavored to delve deeper into the specific mechanisms underlying the radiation-induced up-regulation of SUR1, TRPM4, and NLRP3 inflammasome-related proteins in microglia. We initially assessed the level of oxidative stress in the brain of post-RIBI mice. As shown in Fig. 6A-C, the MDA content and the activity of SOD and CAT in the brain of irradiated mice were remarkably increased, suggesting enhanced tissue oxidative stress after radiation. The results of flow cytometry on the brain further discovered the noticeably increased intracellular ROS level of microglia (CD45⁺CD11b⁺TMEM119⁺) from irradiated mice (Fig. 6D-F). These data imply that microglia may trigger oxidative stress by producing ROS after radiation.

Fig. 6

Microglia may trigger oxidative stress by producing ROS in the RIBI model. (A-C) Quantification of MDA content, SOD activity, and CAT activity in the brain tissues to evaluate the level of oxidative stress after whole-brain radiation. (D-F) The gating strategy (D) and flow cytometry plots (E) to indicate the dynamic changes in the ROS content within microglia (CD45⁺CD11b⁺TMEM119⁺) after RIBI modeling. The FMO negative control consisted of CD45⁺CD11b⁺TMEM119⁺ cells from a sample in which only the DCFH-DA fluorescence probe was omitted. 1 d, 1 day. 3 d, 3 days. 7 d, 7 days. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 versus sham. n = 5 per group

Radiation leads to mitochondrial dysfunction and increased mROS release in BV2 cells

Elevated oxidative stress levels and ROS overproduction were also found in irradiated BV2 cells (Fig. 7A-D). It is generally believed that mitochondria are the main source of intracellular ROS, and the increased release of mROS often indicates mitochondrial damage under various stress conditions [53,54,55]. Therefore, we evaluated mitochondrial dysfunction in irradiated BV2 cells. Compared to the control group, irradiation resulted in lessened mitochondrial mass, increased mitochondrial oxidants, depressed mitochondrial membrane potential, and reduced intracellular ATP (Fig. 7E-J), indicating the anabatic structural damage and dysfunction of mitochondria in irradiated microglia and thereby releasing excessive mROS.

Fig. 7

Radiation leads to mitochondrial dysfunction and increased mROS release in BV2 cells. (A-C) Quantification of MDA content, SOD activity, and CAT activity in the irradiated BV2 cells to evaluate the level of oxidative stress. (D-F) ROS generation was detected by DCFH-DA probe (D), mitochondria were marked by MitoTracker staining (E), and mROS were measured by MitoSox staining (F). (G) Quantification graph for mean fluorescence intensity. (H) JC-1 staining for mitochondrial membrane potential measurement. (I) Quantification of membrane potential by measuring the ratio of red and green fluorescence intensity. JC-1 is an ideal fluorescent probe widely utilized for detecting mitochondrial membrane potential. When the mitochondrial membrane potential is high, JC-1 accumulates within the mitochondrial matrix, forming aggregates that emit red fluorescence. Conversely, at lower mitochondrial membrane potentials, JC-1 fails to accumulate in the matrix and remains as monomers, emitting green fluorescence. This allows for a straightforward detection of changes in mitochondrial membrane potential through the transition in fluorescence color. (J) The histogram to show the level of intracellular ATP in indicated groups. 1 d, 1 day. 3 d, 3 days. Scale bar = 50 μm. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 versus control. n = 3 per group

Post-radiation overproduction of ROS induces the transcription of SUR1-TRPM4 and NLRP3 inflammasome-related molecules in microglia

We proceeded to investigate whether the excessive production of ROS in the brain post-radiation was involved in up-regulating the expression of SUR1-TRPM4 and NLRP3 inflammasome-related proteins in microglia. After eliminating cerebral ROS with NAC (Fig. 8A-B), the declined mRNA levels of SUR1 (encoded by Abcc8), TRPM4, NLRP3, caspase-1, and IL-1β were found in sorted microglia from irradiated mice (Fig. 8C). After NAC intervention, similar changes were also observed in the irradiated BV2 cells (Fig. 8D-F). DAMPs such as ROS are known to prime the expression of NLRP3 inflammasome mediated by the nuclear translocation of transcription factors under stress conditions [24, 54, 56, 57]. We found that the nuclear translocation of Sp-1 and NF-κB p65 was markedly promoted in irradiated BV2 cells but largely reversed by NAC, the ROS scavenger (Fig. 8G-H). Furthermore, the mRNA and protein levels of key components of SUR1-TRPM4 and NLRP3 inflammasome were remarkably elevated in post-modeling cells but reversed by inhibiting Sp-1 and NF-κB p65 (Fig. 8I-K). However, no alteration was found in the levels of Kir6.1 and Kir6.2 (encoded by Kcnj8 and Kcnj11, respectively), which functioned as the components of the K_ATP (SUR1/2-Kir6.x) channel, another specific target of sulfonylureas. In short, these data indicate that radiation induces the overproduction of ROS which act as a priming signal for key molecules of SUR1-TRPM4 and NLRP3 inflammasome in microglia, with the assistance of transcriptional factors like Sp-1 and NF-κB.

Fig. 8

Post-radiation overproduction of ROS induces the transcription of SUR1-TRPM4 and NLRP3 inflammasome-related molecules in microglia. (A-B) The flow cytometry plots (A) to indicate the changes in the ROS content within microglia (CD45⁺CD11b⁺TMEM119⁺) from post-modeling mice after eliminating ROS by NAC. The FMO negative control consisted of CD45⁺CD11b⁺TMEM119⁺ cells from a sample in which only the DCFH-DA fluorescence probe was omitted. (C) qRT-PCR results showing the mRNA levels of SUR1, TRPM4, NLRP3, caspase-1, and IL-1β within the sorted microglia from post-RIBI mouse brain after eliminating ROS by NAC. (D-E) ROS content in irradiated BV2 cells was detected by DCFH-DA probe. (F) qRT-PCR results showing the mRNA levels of SUR1, TRPM4, NLRP3, caspase-1, and IL-1β within the irradiated BV2 cells. (G-H) Western blotting manifesting the nuclear translocation of NF-κB p65 and SP-1 in the irradiated BV2 cells after eliminating ROS by NAC. (I) The heatmap reflecting the mRNA levels of indicated genes in radiation-primed BV2 cells with or without transcriptional factor inhibitors. (J-K) Western blotting results of SUR1, TRPM4, and NLRP3 in radiation-primed BV2 cells with or without transcriptional factor inhibitors. Mith A, Mithramycin A. Scale bar = 50 μm. Data are represented as mean ± SD. **P < 0.01, ***P < 0.001 versus sham or control. #P < 0.05, ##P < 0.01, ###P < 0.001 versus vehicle. n = 3 or 5 per group

ROS induce the transcription of TET2 in microglia after radiation

As mentioned above, we observed that the expression of SUR1-TRPM4 in the mouse brains after radiation remained significantly higher than the sham group at 7 days post-modeling and tended to increase further (Fig. 3A-B). Hence, we further explored whether TET2, beyond ROS, maintained high expression of TRPM4 in microglia post-irradiation by mediating the demethylation of gene promoter. We first found that the expression of TET2 was markedly increased in the microglia sorted from irradiated mouse brains (Fig. 9A-B), the elevated mRNA level of which after radiation was reversed in the presence of NAC (Fig. 9C). The similar changes were also observed in the irradiated BV2 cells and implied the induction of microglial TET2 transcription by ROS after radiation (Fig. 9D-F). Using the JASPAR database, we predicted that NF-κB might bind to the promoter of Tet2 (Fig. 9G), as further confirmed by the enriched NF-κB p65 on the Tet2 promoter (Fig. 9H-I). In sum, these results suggest that ROS may promote the NF-κB p65-mediated transcription of TET2 in microglia after radiation.

Fig. 9

ROS induce the expression of TET2 in microglia after radiation. (A-B) Western blotting results of TET2 within the sorted microglia from post-RIBI mouse brain. (C) qRT-PCR results showing the mRNA level of TET2 within the sorted microglia from post-RIBI mouse brain after eliminating ROS by NAC. (D-E) Western blotting results of TET2 within the irradiated BV2 cells. (F) qRT-PCR results showing the mRNA level of TET2 within the irradiated BV2 cells after eliminating ROS by NAC. (G) Prediction of binding site for NF-κB p65 to the promoter of TET2 by JASPAR database. (H-I) ChIP-qPCR to verify the binding of NF-κB p65 to the TET2 promoter. IgG antibodies were used as the negative control. Rad, radiation. 1 d, 1 day. 3 d, 3 days. 7 d, 7 days. Data are represented as mean ± SD. **P < 0.01, ***P < 0.001 versus sham or control. ##P < 0.01, ###P < 0.001 versus vehicle. n = 3 or 5 per group

TET2 participates in the regulation of TRPM4 expression and its association with SUR1 within microglia following radiation exposure

The effects of up-regulated TET2 on the SUR1-TRPM4 complex in the microglia after radiation were evaluated by the pharmacological or genetic inhibition of TET2. As depicted in Fig. 10A-C, the upsurge in mRNA and protein expression levels of TRPM4 within microglia of post-RIBI mice underwent a pronounced reversal upon inhibition of TET2 by Bobcat339. The immunofluorescence results revealed that treatment with Bobcat339 also suppressed the colocalization of SUR1 and TRPM4 in microglia within the brains of irradiated mice (Fig. 10D). After Tet2 gene knockdown, a notable down-regulation trend was also observed in the expression of TRPM4 and the distribution of SUR1-TRPM4 in irradiated BV2 cells (Fig. 10E-H). Taken together, these data indicate that TET2 plays a pivotal role in modulating the expression of TRPM4 and the assembly of SUR1-TRPM4 complex in microglia after radiation.

Fig. 10

TET2 participates in the regulation of TRPM4 expression and its association with SUR1 within microglia following radiation. (A-C) qRT-PCR (A) and western blotting (B-C) results of TRPM4 within the sorted microglia from post-RIBI mouse brain after the pharmacological inhibition of TET2. (D) Confocal analysis of TRPM4, SUR1, and Iba-1 staining within hippocampus of post-RIBI mice. (E-G) qRT-PCR (E) and western blotting (F-G) results of TRPM4 within the irradiated BV2 cells after Tet2 gene knockdown. (H) Confocal analysis of TRPM4 and SUR1 staining within the irradiated BV2 cells after Tet2 gene knockdown. Scale bar = 50 μm. Data are represented as mean ± SD. ***P < 0.001 versus sham or control. ##P < 0.01, ###P < 0.001 versus vehicle or siControl. n = 3 or 5 per group

TET2-pSTAT5 complex facilitates TRPM4 expression in irradiated microglia by enhancing the demethylation of Trpm4 promoter

Subsequently, we embarked on a more profound exploration to ascertain whether TET2’s enhancement of TRPM4 expression in microglia following radiation exposure was intricately tied to the erasure of methylation marks, which were pivotal for epigenetic regulation. Using dot-blot, we observed a decrease in 5-mC but an increase in 5-hmC of the irradiated BV2 cells, suggesting an enhancement in the overall DNA demethylation levels within irradiated BV2 cells (Fig. 11A). Generally, TET2 induces DNA demethylation and transcriptional activation of specific genes by forming a complex with phosphorylated STAT family members, the upstream transcription factors, which are translocated to the nucleus and bind to specific DNA sequences [58,59,60]. Hence, we predicted that STAT5 might bind to the promoter of Trpm4 by using the JASPAR database (Fig. 11B), as further confirmed by the enriched pSTAT5 on the Trpm4 promoter in the irradiated BV2 cells (Fig. 11C). The results of Co-IP further revealed a remarkable interaction between TET2 and pSTAT5, indicating that TET2 was indeed recruited by phosphorylated STAT5 in the BV2 cells after radiation (Fig. 11D). Moreover, Trpm4 promoter DNA sequences were significantly pulled down by TET2 antibodies in irradiated BV2 cells (Fig. 11E), demonstrating a tight binding of TET2 and Trpm4 promoter. Finally, we directly detected the levels of 5-hmC on the Trpm4 promoters, which exhibited a significant increase in irradiated BV2 cells, but Tet2 knockdown broke this trend (Fig. 11F). In sum, these results manifest that the TET2-pSTAT5 complex in irradiated microglia enters the nucleus and binds to the Trpm4 promoter, facilitating TRPM4 expression by catalyzing demethylation.

Fig. 11

TET2-STAT5 complex facilitates TRPM4 expression in irradiated microglia by enhancing the demethylation of Trpm4 promoter. (A) Global 5-mC and 5-hmC levels in irradiated BV2 cells. (B) Prediction of binding site for STAT5 to the promoter of TRPM4 by JASPAR database. (C) ChIP-qPCR to verify the binding of pSTAT5 to the TRPM4 promoter in irradiated BV2 cells. IgG antibodies were used as the negative control. (D) The association of TET2 and pSTAT5 detected by Co-IP, in irradiated BV2 cells. Samples are immunoprecipitated with isotype IgG as a control. (E-F) ChIP-qPCR to verify the binding of TET2 (E) or 5-hmC (F) to the TRPM4 promoter in irradiated BV2 cells. IgG antibodies were used as the negative control. Rad, radiation. Data are represented as mean ± SD. ***P < 0.001 versus control. ###P < 0.001 versus siControl. n = 3 per group

Discussion

The sustained activation of microglia and the ensuing chronic neuroinflammation play pivotal roles in delayed RIBI [12]. Upon activated following radiation exposure, microglia undergo proliferation and directed migration towards sites of cellular damage. This orchestrated response involves their engagement in phagocytosis, efficiently eliminating dead cells and thereby fostering tissue regeneration and repair processes [61]. Simultaneously, inflammatory signaling pathways within microglia are ignited, prompting the up-regulation of inflammatory cytokines and chemokines, which, in turn, stimulate downstream inflammatory cascades, triggering neuroinflammatory responses [11, 62]. Thus, the role of microglia in RIBI exhibits a complex and nuanced dual nature, where the delicate balance between facilitating recovery and modulating the inflammatory microenvironment, once disrupted, can potentially lead over time to chronic neuroinflammation lasting for years after irradiation [63]. In vitro and in vivo experiments, we found that a large number of activated microglia in the hippocampus exhibited a pronounced pro-inflammatory phenotype, accompanied by an up-regulation of various inflammatory cytokines after irradiation, suggesting the formation of an inflammatory storm dominated by microglia after whole-brain radiation. Uncovering the molecular mechanisms underlying the sustained activation of microglia and the promotion of neuroinflammation holds promise for identifying novel therapeutic targets to prevent and treat delayed RIBI.

TRPM4 is a non-selective monovalent cation channel and becomes activated when sensing intracellular [ATP] decreases or [Ca²⁺] increases [64]. After CNS injury, TRPM4 channels form a novel SUR1-TRPM4 channel with the regulatory subunit SUR1, imparting new characteristics: (1) increased sensitivity to changes in [Ca²⁺]; (2) prolonged open duration; (3) block ability by sulfonylureas like GLB [65]. It is noteworthy that SUR1-TRPM4, being absent from normal brain tissue, represents an exceptionally promising target for disease intervention. Our researches and those of others, conducted in models of neurological diseases such as ischemic stroke [66], traumatic brain injury [67], cardiac arrest [68, 69], subarachnoid hemorrhage [70], and status epilepticus [71], have revealed that GLB exerts potent anti-cerebral edema and BBB protective effects, potentially through blocking the SUR1-TRPM4 complex in neurovascular units that allows the dramatic influx of Na⁺ and water.

Intriguingly, several clinical studies indicate that sulfonylureas not only reduce brain edema but also improve neurological function in some patients without obvious brain edema [72,73,74]. Animal studies have further demonstrated that GLB exerts favorable influences on brain injuries primarily marked by neuroinflammation, underscoring an additional anti-inflammatory neuroprotection of sulfonylureas [75, 76]. Activated microglia have been reported by us to highly express SUR1-TRPM4 under pathological conditions [66, 68], and thereby modulate downstream inflammatory pathways [75], suggesting that SUR1-TRPM4 in microglia may serve as a potential target for the anti-inflammatory effects of sulfonylureas. Our recent findings further disclose that SUR1-TRPM4 functions as a crucial K⁺ efflux enhancer in microglia after cardiac arrest, playing a pivotal role in activating the NLRP3 inflammasome by gating the downstream depolarization-sensitive K⁺ channels, and underscore that blocking SUR1-TRPM4 in microglia is a central mechanism by which sulfonylureas such as GLB exert their anti-inflammatory effects [21].

Previous reports have documented the involvement of NLRP3 inflammasome in inflammatory responses observed in radiation injury models of various organs [19]. Moreover, the activation of NLRP3 inflammasome in microglia has been confirmed to exist in both post-radiation mouse brains and in vitro irradiated microglia [52]. Consistently, we demonstrated the elevated expression and assembly of NLRP3 inflammasome in the microglia after irradiation. The expression of SUR1-TRPM4 was synchronously up-regulated with NLRP3 inflammasome-related proteins, and the activation of NLRP3 inflammasome could be abolished by sulfonylureas or TRPM4 inhibitors. Notably, treatment with GLB in irradiated Trpm4^−/− mice showed no better performance in abrogating NLRP3 activation, compared to the irradiated Trpm4^−/− mice without GLB, suggesting that SUR1-TRPM4 indeed acts as the specific target of sulfonylureas such as GLB. Furthermore, in the irradiated BV2 cells to mimic the pathophysiologic process in RIBI model, no changes were found in the expression levels of Kir6.1 and Kir6.2, indicating that the K_ATP (SUR1/2-Kir6.x) channel is not the target of sulfonylureas after radiation. To our knowledge, this is the first study clarifying the regulatory role of SUR1-TRPM4 in the activation of microglial NLRP3 inflammasome after brain radiation. Although both this study and our previous research on global cerebral ischemia [21] have confirmed the role of SUR1-TRPM4 in mediating NLRP3 inflammasome activation in microglia, thereby amplifying neuroinflammation, there are notable differences between the two studies. Firstly, global cerebral ischemia represents an acute brain injury, whereas delayed RIBI is a progressively developing chronic brain injury. The significant value of this study lies in further expanding the applicability of the finding that SUR1-TRPM4 is involved in brain injury by mediating NLRP3 inflammasome activation, thereby offering new insights into the clinical diagnosis and treatment of various chronic brain injuries. Secondly, our previous research on global cerebral ischemia did not elucidate the specific mechanisms regulating the expression of SUR1-TRPM4 and NLRP3 inflammasome proteins. In this study, we clarified the pivotal role of ROS following whole-brain radiation, which modulates SUR1-TRPM4 expression through multiple pathways, driving sustained inflammation mediated by microglial NLRP3. Lastly, for the first time in this study, we unveiled the epigenetic regulatory mechanism of TRPM4 expression and highlighted the crucial role of TET2 therein, providing a novel target for blocking SUR1-TRPM4-mediated persistent neuroinflammation after radiation.

Interestingly, direct irradiation of BV2 cells was found to solely upregulate the expression of NLRP3 inflammasome-related proteins in this study. Further promotion of the assembly and activation of the NLRP3 inflammasome occurred only when ATP stimulation was concurrently applied. However, in the brains of irradiated mice, significant NLRP3 inflammasome activation was observed without additional ATP stimulation. We speculate that the primary reason for this discrepancy is the presence of numerous other cell types in addition to microglia within the mouse brain. Notably, neurons have been proven to release substantial amounts of ATP upon irradiation-induced damage, thereby activating microglia [77]. In contrast, the BV2 cells cultured in vitro are of a single type and limited in number, which is insufficient to replicate the ATP release conditions observed in vivo. Consequently, supplementary ATP is required to activate the NLRP3 inflammasome in BV2 cells.

After whole-brain radiation, nearly 90% of patients encounter memory and executive function impairments, with symptoms appearing within 3–4 months post-radiotherapy [78]. The number of activated microglia in the hippocampus is inversely correlated with neurogenesis [16], indicating that the microglia-mediated persistent chronic inflammation also contributes to abnormal differentiation of neural precursor cells, a critical mechanism underpinning cognitive impairment in RIBI [4, 10]. Thus, inhibiting destructive inflammatory response fueled by microglia may be of therapeutic potential for the treatment of RIBI. Suppressing microglial inflammatory factors or depleting microglia has been reported to mitigate cognitive impairment and restore neurogenesis in post-radiation mouse [14,15,16]. Similarly, we confirmed that blocking SUR1-TRPM4 with GLB could significantly inhibit the activation of microglia and astrocytes in the hippocampus, accordingly rescuing hippocampal immature neurons and improving deficits in hippocampus-dependent functions such as learning, memory, and spatial information processing in RIBI model. Notably, 2 months after whole-brain radiation, the persistent transition of microglia towards a pro-inflammatory state and corresponding neuroinflammation was evident, yet no significant cerebral edema was observed at the same time point, thereby excluding the possibility that intervening with SUR1-TRPM4 exerts neuroprotective effects by reducing cerebral edema within 2 months after irradiation. However, we observed Evans blue leakage 2 months post-radiation. We speculate that microglia-mediated neuroinflammation predominantly drives the process of brain injury for a period of time after radiation, and sustained inflammatory stimulation leads to the disruption of BBB 2 months after radiation, although cerebral edema has not yet formed at this stage. Over time, cerebral edema may subsequently emerge. This further suggests that as time elapses post-radiation, the pathophysiological mechanisms underlying brain injury grow increasingly intricate, potentially involving a synergistic action between neuroinflammation and cerebral edema. Notably, we have not provided direct evidence to indicate whether SUR1-TRPM4 is specifically expressed in microglia after radiation. Hence, we cannot rule out the possibility that SUR1-TRPM4 is also expressed within the neurovascular unit post-radiation and contributes to the later stages of cerebral edema formation. This awaits further clarification through subsequent research.

In addition to inhibiting unremitting NLRP3 activation through blocking SUR1-TRPM4, timely modulation of related protein expression represents another promising therapeutic avenue. The priming of the inflammasome machinery prompts a transcriptional surge in the expression of NLRP3 and pro-IL-1β. Among the prevalent priming stimuli explored in the context of sterile injury, the recognition of DAMPs stands out as a pivotal event [24]. Radiation causes extensive decomposition of water within tissues and cells, ultimately leading to the generation of ROS [19], which is recognized as one of DAMPs capable of priming the NLRP3 inflammasome [24]. We verified that microglia might be involved in oxidative stress within the brain by generating excessive ROS after radiation, on account of severe mitochondrial dysfunction. In a microenvironment where the redox balance is disrupted, ROS-induced sustained activity and overexpression of TRPM4 mediate cell death and tissue damage [25]. However, we found that irradiation alone, in the absence of ATP stimulation, merely led to a minor assembly and activation of NLRP3 inflammasomes within microglia, suggesting that radiation-induced ROS may not be sufficient as a direct trigger to cause the opening of SUR1-TRPM4 and subsequently initiate sustained and extensive activation of NLRP3 inflammasomes. Meanwhile, our data revealed that ROS functioned as a pivotal priming signal for the expression of SUR1-TRPM4 and NLRP3 inflammasome components after radiation, facilitated by toll-like receptors and downstream transcriptional factors [19, 24]. It is noteworthy that ROS can also be excessively produced under violent Ca²⁺ influx, the crucial signal for TRPM4 opening [79], implying a significant risk of a vicious cycle forming between the priming signal and activation signal for SUR1-TRPM4-mediated NLRP3 activation in microglia within the brain after irradiation. Multiple threads within the intricate network of SUR1-TRPM4 regulation in microglial NLRP3 inflammasome activation post-irradiation converge on ROS as a pivotal node, suggesting that scavenging ROS may be a crucial strategy to mitigate RIBI.

DNA methylation typically governs diverse biological processes and controls the expression of tissue-specific genes [10]. Radiation induces DNA hypomethylation, with mechanisms involving sustained induction of ROS [10]. In microglia after radiation, we found that ROS up-regulated the NF-κB p65-mediated expression of TET2, a methylcytosine dioxygenase catalyzing the demethylation within gene promoter regions and thereby causing transcriptional disinhibition [26]. Moreover, we found that TET2 elevated the expression of TRPM4 and the assembly of SUR1-TRPM4 in irradiated microglia by increasing 5-hmC on the Trpm4 promoters, uncovering a novel, previously unreported mechanism whereby radiation regulates the inflammatory state of microglia through epigenetic modifications. Typically, STAT family play a crucial role in TET2-mediated demethylation relying on the nuclear translocation in response to phosphorylation and dimerization [58,59,60]. We elucidate that, for the first time, TET2 achieves nuclear entry in the form of TET2-pSTAT5 complex, thereby exerting its regulatory influence on the expression of TRPM4 in irradiated microglia.

This study has several limitations. Firstly, despite our efforts to use microglia sorted from irradiated mice in the majority of our experiments to bolster the reliability of our in vitro findings, BV2 cells still do not adequately represent the functional state of microglia in vivo, when compared to primary microglia. Hence, further experiments employing primary microglia are necessary to validate the novel mechanisms uncovered in this study. Secondly, Evans blue leakage observed 2 months post-radiation suggests a subsequent risk of cerebral edema. However, due to the insufficient observation period in this study, cerebral edema was not observed. Additionally, there was a lack of exploration into the mechanisms by which neuroinflammation and potential cerebral edema, beyond the 2-month mark, collectively contribute to the progression of RIBI. This warrants further investigation in subsequent studies.

Conclusions

This study unveils the pivotal role of SUR1-TRPM4 in driving microglial NLRP3-mediated persistent neuroinflammation following whole-brain radiation. By blocking SUR1-TRPM4, we demonstrate a significant alleviation of delayed RIBI and an improvement in cognitive dysfunction. Furthermore, we have elucidated the pivotal role of ROS after whole-brain irradiation, where it regulates the expression of SUR1-TRPM4 through multiple pathways to drive persistent inflammation mediated by microglial NLRP3. Notably, inducibly expressed TET2 by ROS forms a complex with pSTAT5, which subsequently enters the nucleus to mediate demethylation in gene promoters, thereby enhancing TRPM4 expression in irradiated microglia (Fig. 12). In summary, this study will certainly be insightful for deciphering the priming and activation signals of SUR1-TRPM4-mediated persistent NLRP3 activation within microglia post whole-brain radiation, which offers novel therapeutic strategies for clinical treatment of delayed RIBI as well as other NLRP3-related inflammatory neurological disorders involving excessive ROS production, although clinical translation still requires time.

Fig. 12

Schematic representation of the regulatory effects of SUR1-TRPM4 on microglial NLRP3-mediated neuroinflammation after whole-brain radiation. This study unveils the pivotal role and neuroprotective potential of SUR1-TRPM4 in driving microglial NLRP3-mediated persistent neuroinflammation following whole-brain radiation. The specific mechanism by which SUR1-TRPM4 regulates NLRP3 inflammasome activation in microglia has been reported in our previous research [21]. Here, we have further demonstrated that irradiation causes mitochondrial damage to microglia, leading to violent release of ROS which enhance the transcription of SUR1, TRPM4, and NLRP3 inflammasome-related molecules. Moreover, inducibly expressed TET2 by ROS forms a complex with pSTAT5, which subsequently enters the nucleus to mediate demethylation in gene promoters, thereby enhancing TRPM4 expression in irradiated microglia. To sum up, our findings decipher that SUR1-TRPM4 crucially mediates the persistent activation of microglial NLRP3 inflammasome under the action of ROS after whole-brain radiation, offering novel therapeutic strategies for delayed RIBI as well as other NLRP3-related neurological disorders involving excessive ROS production