ROS-responsive hydrogel microspheres deliver shTMEM2-loaded liposomes to suppress pyroptosis and ameliorate intervertebral disc degeneration Authors: Zhiqiang Xu, Hanmo Fang, Chengrui Peng, Zhiwei Liao, Guoyu Yang, Hu Qian, Jun Ao, Xinghuo Wu Published: 2026 ABSTRACT AbstractLow back pain (LBP) is a highly prevalent musculoskeletal disorder worldwide. Nucleus pulposus cell (NPC) pyroptosis contributes to intervertebral disc degeneration (IVDD), but the underlying mechanisms remain unclear. We identified TMEM2, a transmembrane protein not previously linked to IVDD, as markedly upregulated in degenerated NPCs by single‑cell RNA sequencing. TMEM2 expression correlated positively with IVDD severity and NPC pyroptosis, and its knockdown alleviated pyroptosis. Mechanistically, TMEM2 binds BAX to promote mPTP opening, leading to mtDNA release and activation of the STING–TBK1–NF‑κB pathway, ultimately triggering NLRP3 inflammasome‑mediated pyroptosis. To address the avascular nature of the disc, we developed an injectable ROS‑responsive hydrogel microsphere system encapsulating shTMEM2‑loaded liposomes (Lipo/shTMEM2@MS). This platform enables sustained release, on‑demand degradation, and single‑dose intradiscal administration. In a rat model, Lipo/shTMEM2@MS effectively ameliorated disc degeneration. Collectively, these findings identify TMEM2 as a key driver of NPC pyroptosis and establish a ROS‑responsive microsphere strategy as a promising translational approach for IVDD.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12951-026-04579-9.Keywords: Low back pain, Intervertebral disc degeneration, TMEM2, Mitochondrial DNA, Mitochondrial permeability transition pore, Pyroptosis Introduction Low back pain (LBP) is a major global health issue, representing one of the most prevalent musculoskeletal disorders and a leading cause of work disability and reduced quality of life [1–3]. Although the etiology of LBP is multifactorial, intervertebral disc degeneration (IVDD) is recognized as one of its primary causes [4–6]. Dysfunction of nucleus pulposus cells (NPCs) is considered an initiating event in IVDD [7, 8]. The intervertebral disc is the largest avascular tissue in the human body, and consequently, the nucleus pulposus resides in a chronically hypoxic, low-nutrient, and weakly acidic microenvironment [9–11]. This unique physiological feature makes it particularly vulnerable to injury and inflammatory stimuli, yet simultaneously poses a major challenge for local drug delivery. Therefore, the development of drug delivery systems that enable sustained local release and respond to the pathological microenvironment holds great promise for IVDD therapy. Mitochondrial dysfunction is a key pathological feature of IVDD [12–14]. The mitochondrial permeability transition pore (mPTP), a multiprotein complex located at the interface between the inner and outer mitochondrial membranes, opens transiently under physiological conditions to maintain mitochondrial calcium homeostasis and membrane potential balance [15, 16]. However, under pathological conditions such as oxidative stress, the mPTP undergoes abnormally sustained opening, leading to collapse of the mitochondrial membrane potential, excessive production of reactive oxygen species (ROS), and leakage of mitochondrial DNA (mtDNA) into the cytosol [17, 18]. The leaked mtDNA acts as a damage-associated molecular pattern (DAMP) and is recognized by cytosolic cyclic GMP-AMP synthase (cGAS), thereby activating the STING signaling pathway [19–21]. This in turn leads to TBK1-mediated phosphorylation of nuclear factor-κB (NF-κB) and subsequent transcription of pro-inflammatory cytokines [22, 23]. In recent years, pyroptosis, an inflammatory form of programmed cell death, has gained considerable attention in the context of IVDD [24, 25]. Pyroptosis is mediated by inflammasome activation, in which caspase-1 cleaves Gasdermin D to form membrane pores, resulting in the release of cellular contents and a robust inflammatory response [26–28]. Nevertheless, the regulatory mechanism linking the mPTP–mtDNA–STING axis to NPCs pyroptosis still remains poorly understood in the setting of IVDD. TMEM2 is a cell surface transmembrane protein initially recognized for its ability to degrade hyaluronic acid (HA) in the extracellular matrix [29]. Recent studies have revealed that TMEM2 is also involved in maintaining mitochondrial homeostasis by regulating mitochondrial fission and fusion [30]. In various tumor models and endoplasmic reticulum stress paradigms, TMEM2 is closely associated with cell survival and inflammatory responses [31–33]. However, the role of TMEM2 in IVDD and its potential link to NPCs pyroptosis have not yet been reported. In this study, single-cell RNA sequencing (scRNA-seq) analysis of human degenerated nucleus pulposus tissues revealed a marked upregulation of TMEM2 in degenerated NPCs. Further mechanistic investigation demonstrated that TMEM2 interacts with BAX to promote mPTP opening, leading to the release of mtDNA into the cytosol and subsequent activation of the STING–TBK1–NF-κB signaling axis, ultimately inducing NPCs pyroptosis. In view of the avascular nature of the intervertebral disc and the challenges associated with sustained local drug delivery, we designed an injectable ROS-responsive nanocomposite hydrogel microsphere system (Lipo/shTMEM2@MS, Scheme 1), consisting of shTMEM2-loaded liposomes encapsulated within microspheres. The system is built on dynamic boronate ester bonds formed between phenylboronic acid-modified alginate and dextran, enabling on-demand degradation in response to ROS in the inflammatory microenvironment. In a rat model of disc needle puncture, Lipo/shTMEM2@MS effectively retarded the progression of disc degeneration. Collectively, this study not only reveals a critical regulatory role of TMEM2 in NPCs pyroptosis but also, based on this mechanism, develops a ROS-responsive composite microsphere delivery system, thereby providing a novel target and strategy for the precise treatment of intervertebral disc degeneration. Schematic illustration of the preparation route and mechanism of action of Lipo/shTMEM2@MS. a, b, Microfluidic fabrication of Lipo/shTMEM2@MS composite microspheres. c, Therapeutic mechanism after intradiscal injection: the Lipo/shTMEM2@MS release platform exerts its therapeutic effects Materials and methods Ethics statement This study was performed in line with the principles of the Declaration of Helsinki. All procedures involving human tissues and animals were approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology (approval No. 2025IACUC:4953). Study subjects Patients undergoing lumbar disc herniation surgery were recruited. Exclusion criteria included infections, tumors, or endocrine disorders. Sixteen nucleus pulposus (NP) samples were collected from 7 males and 5 females (age range: 18–62 years). IVDD severity was graded by the Pfirrmann system, with 3 cases each for grades II, III, IV, and V. Culture of human and rat NPCs Human NP tissues were collected in PBS. Cells were isolated with type II collagenase and cultured in DMEM/F12 (1:1) containing 15% FBS (Gibco, USA) and 1% penicillin/streptomycin (Servicebio) at 37 °C in 5% CO₂. Third-passage cells were used for experiments. Oxidative stress was induced by 50–200 µM TBHP for 24 h. Rat NP tissues were harvested from five male SD rats (200–230 g). Cells were digested with 0.1% collagenase and 2 U/mL hyaluronidase (Sigma) for 2 h, then cultured in DMEM/F12 with 15% FBS and 1% penicillin/streptomycin for one week. Third-passage cells were used for experiments. For STING pathway activation, NPCs were treated with 10 µM STING agonist‑1 (MedChemExpress, HY‑19711) for 24 h. In co‑treatment experiments, STING agonist‑1 was added together with 100 µM TBHP 24 h after transfection, and cells were incubated for an additional 24 h. Statistical analysis Data are expressed as mean ± SD. Statistical analysis and graphing were performed with GraphPad Prism 10. One-way ANOVA was applied for multiple group comparisons, and unpaired Student’s t-test for two-group comparisons. A p-value < 0.05 was considered statistically significant. Supplementary methods and materials Detailed experimental procedures, materials, and data analyses are provided in the Supplementary Materials. Results TMEM2 is upregulated in human degenerative intervertebral discs To investigate the correlation between TMEM2 and IVDD, we re-analyzed single-cell transcriptomic data from eight human NP tissues with different Pfirrmann grades retrieved from the GEO database (GSE165722) based on a previous study [4]. Cells were annotated using cell-specific biomarkers (Supplementary Fig. S1 a), ultimately identifying seven major cell types: NPCs, neutrophils, T cells, neutrophil precursor cells, macrophages, erythrocytes, and erythroid progenitor cells (Fig. 1a). Based on TMEM2 expression across all cells, we categorized them into TMEM2⁺ and TMEM2⁻ groups. Notably, TMEM2⁺ cells were specifically enriched in nucleus pulposus cells (Fig. 1b). We further performed dimensionality reduction and clustering on NPCs and identified their subpopulations using specific marker genes (Fig. 1c). Given our previous finding that NPCs degeneration is accompanied by fibrosis [34], we broadly classified NPCs into fibrotic, homeostatic, and other subtypes (Fig. 1d). Further analysis revealed significant differences in NPCs subpopulation composition between mildly (grades II/III) and severely (grades IV/V) degenerated tissues. As degeneration progressed, the proportion of fibrotic NPCs markedly increased, while homeostatic NPCs decreased correspondingly (Fig. 1e). Moreover, TMEM2 expression was significantly higher in severely degenerated NP tissues compared to mildly degenerated ones (Fig. 1f, g) and was specifically elevated in fibrotic NPCs, exceeding its levels in other NPCs subtypes (Fig. 1h, i). Gene Ontology (GO) and KEGG analyses comparing TMEM2-high and TMEM2-low expression groups showed that TMEM2 upregulation was significantly associated with cellular response to oxidative stress, positive regulation of fibroblast migration, and activation of DNA damage response pathways, highlighting TMEM2’s potential role in driving oxidative damage, pro-fibrotic processes, and DNA injury during IVDD (Supplementary Fig. S1 b).To validate these findings, we collected clinical NP tissue specimens with different degeneration grades classified according to the Pfirrmann MRI grading system. Due to differences in tissue water content, degenerated discs exhibited varying signal intensities on T2-weighted MRI. Compared to mildly degenerated NP tissues, severely degenerated tissues showed obvious volume shrinkage, leading to reduced disc height (Fig. 1j). With increasing degeneration grade, NP tissues exhibited nuclear clustering and karyorrhexis” (Fig. 1k). IHC staining indicated that TMEM2 levels increased with IVDD severity (Fig. 1l, m). Western blot analysis demonstrated higher TMEM2 protein expression in degenerated NP tissues (Fig. 1n, o). RT-qPCR analysis of TMEM2 mRNA expression yielded consistent results (Fig. 1p). We further performed IHC staining of GSDMD on clinical NP tissues. GSDMD expression was higher in severe degeneration and positively correlated with TMEM2 (Supplementary Fig. S2 a).Together, these data indicate that TMEM2 expression positively correlates with the severity of IVDD. TMEM2 is upregulated in human degenerative intervertebral discs. a, UMAP projection of single - cell RNA sequencing (scRNA - seq) data representing 29,441 cells derived from human nucleus pulposus (NP) tissues, categorized into seven major cell subtypes. b, UMAP plot illustrating the distribution of TMEM2 expression across the seven cell subtypes. c, Heatmap displaying marker genes for distinct sub‑clusters identified in the scRNA - seq data. d, UMAP visualization of NP cells (NPCs, 14,419 cells) further stratified into three subtypes. e, Proportions of NPC subtypes in mild (Pfirrmann grades II/III) versus severe (grades IV/V) intervertebral disc degeneration (IVDD). f, g, Violin plot and bubble plot comparing TMEM2 expression levels between mild and severe IVDD groups. h, i, Violin plot and bubble plot depicting TMEM2 expression across NPC subtypes. j, Representative T2 - weighted magnetic resonance imaging (MRI) images and corresponding NP tissue sections graded according to the Pfirrmann classification system. k, Hematoxylin and eosin (H&E) staining of NP tissues at different degenerative stages. l, m, Immunohistochemical (IHC) staining and quantitative analysis of TMEM2 in NP tissues with varying degrees of degeneration. n, o, Representative western blot images and quantification of TMEM2 protein expression in patient NP tissues across degeneration grades. p, Reverse transcription quantitative real - time PCR (RT - qPCR) analysis of TMEM2 mRNA expression in NP tissues from patients with different degenerative grades. Data are presented as mean ± standard deviation (SD) from at least three independent experiments. n.s., not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 NPCs degeneration is accompanied by pyroptosis and upregulated TMEM2 expression Previous studies have indicated that oxidative stress plays a significant role in the progression of IVDD [35]. To further investigate the degeneration mechanisms of NPCs, we established an oxidative stress-induced degeneration model using tert-butyl hydroperoxide (TBHP). NPCs were treated with different concentrations of TBHP for 24 h. Immunofluorescence staining showed that TMEM2 levels in NPCs increased in a dose-dependent manner after TBHP treatment (Fig. 2a, b). Consistently, RT-qPCR results indicated an elevation in TMEM2 mRNA expression following the same trend (Fig. 2c). Based on the CCK-8 assay, we selected 100 µM TBHP for subsequent experiments (Fig. 2d). Immunofluorescence staining further demonstrated increased co-localization and enhanced fluorescence intensity of GSDMD, CASP1, and NLRP3 in the cytoplasm after TBHP treatment (Fig. 2e–j). Western blot analysis confirmed that TBHP treatment for 24 h upregulated the expression levels of pyroptosis-associated proteins, including NLRP3, ASC, cleaved caspase-1, GSDMD, GSDMD-N, and TMEM2 (Fig. 2k–r). These findings suggest that NPCs degeneration is accompanied by pyroptosis and that TMEM2 is involved in this process. Human NPCs were treated with 50, 100, and 200 µM TBHP for 24 h. a, b, Representative immunofluorescence images and quantitative analysis of TMEM2 in TBHP - treated NPCs. c, RT - qPCR analysis of TMEM2 mRNA expression in TBHP‑treated NPCs. d, Viability of NPCs cultured with different concentrations of TBHP for 24 h.e, f, Representative immunofluorescence images and quantitative analysis of GSDMD in TBHP - treated NPCs. g, h, Representative immunofluorescence images and quantitative analysis of NLRP3 in TBHP - treated NPCs. i, j, Representative immunofluorescence images and quantitative analysis of Caspase‑1 in TBHP‑treated NPCs. k–r, Representative western blot images and quantitative analysis of pyroptosis‑associated proteins (NLRP3, ASC, cleaved caspase‑1, GSDMD, GSDMD‑N, and TMEM2) in TBHP - treated NPCs. Data are presented as mean ± SD from at least three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001 Knockdown of TMEM2 alleviates NP cell pyroptosis To further investigate the role of TMEM2 in IVDD, TMEM2 was knocked down in NPCs by siRNA transfection, as confirmed by Western blot and RT-qPCR (Fig. 3a–c). Treatment with TMEM2 siRNA significantly suppressed the TBHP-induced upregulation of pyroptosis-associated proteins (NLRP3, ASC, cleaved caspase-1, GSDMD, and GSDMD-N) in NPCs (Fig. 3d–j). Flow cytometry and TUNEL staining further demonstrated that only a small number of pyroptotic NPCs were observed in the control group, while TBHP treatment markedly increased the number of pyroptotic cells. Importantly, TMEM2 siRNA treatment exerted a marked protective effect against TBHP-induced NPCs pyroptosis, resulting in a significant reduction in the pyroptosis rate (Fig. 3k–n). Immunofluorescence showed that TBHP decreased COL II and increased MMP3, and these effects were reversed by TMEM2 siRNA (Supplementary Fig. S3 a, b). Collectively, these results indicate that knockdown of TMEM2 effectively protects NPCs from TBHP-induced pyroptosis. NPCs were transfected with siTMEM2 (50 nmol/L) for 24 h, followed by treatment with 100 µM TBHP for another 24 h. a, b, Representative western blot images and quantitative analysis of TMEM2 knockdown efficiency by siRNA. c, RT - qPCR analysis of TMEM2 mRNA expression after siRNA knockdown. d–j, Representative western blot images and quantitative analysis of NLRP3, ASC, pro_caspase_1, cleaved caspase_1, GSDMD, and GSDMD-N in different treatment groups. k, l, Flow cytometry analysis and quantification of pyroptosis in different treatment groups. m, n, TUNEL staining and quantitative analysis in different treatment groups. Data are presented as mean ± SD from at least three independent experiments. n.s., not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 TMEM2 knockdown attenuates the activation of the STING/TBK1/NF-κB axis and NPCs pyroptosis Our previous studies have demonstrated that STING acts as an upstream regulator of the NLRP3 inflammasome and participates in modulating NLRP3 inflammasome-mediated pyroptosis in NPCs [34, 36, 37]. TBK1 and NF-κB are key downstream signaling molecules of STING [38, 39]. Therefore, we further investigated whether TMEM2 regulates NPCs pyroptosis through the STING/TBK1/NF-κB signaling pathway. Western blot results showed that the TBHP-treated group exhibited elevated levels of TMEM2, STING, p-TBK1, and p-NF-κB. Treatment with TMEM2 siRNA significantly reduced the expression levels of TMEM2, STING, p-TBK1, and p-NF-κB. Notably, addition of a STING agonist upregulated the expression of TMEM2, STING, p-TBK1, and p-NF-κB (Fig. 4a–g). Subsequently, we examined the expression levels of pyroptosis-associated molecules by Western blot. The results indicated that TMEM2 siRNA treatment markedly suppressed the TBHP-induced increase in pyroptosis-associated proteins (NLRP3, ASC, cleaved caspase-1, GSDMD, and GSDMD-N), while the addition of a STING agonist reversed this inhibitory effect of TMEM2 knockdown on pyroptosis (Fig. 4h–n). We further evaluated the ultrastructure of NPCs by TEM. In TBHP-treated NPCs, extensive cell swelling was observed, a hallmark of GSDMD-mediated plasma membrane perforation that leads to pyroptosis [35]. TMEM2 knockdown alleviated TBHP-induced cell swelling; again, addition of a STING agonist counteracted the protective effect of TMEM2 knockdown on pyroptosis (Fig. 4o). Cell viability assessed by Hoechst and PI staining showed that TBHP treatment increased cell death, TMEM2 knockdown reduced TBHP-induced NPCs death, and the addition of a STING agonist elevated NPCs death (Fig. 4p, q). Taken together, these data indicate that TMEM2 regulates NPCs pyroptosis via the STING/TBK1/NF-κB pathway. NPCs were treated with siTMEM2 and STING agonist-1 plus 100 µM TBHP for 24 h. a–g, Representative western blot images and quantitative analysis of TMEM2, STING, TBK1, p‑TBK1, p65, and p‑p65 in different treatment groups. h–n, Representative western blot images and quantitative analysis of NLRP3, ASC, pro‑Caspase‑1, cleaved Caspase‑1, GSDMD, and GSDMD‑N in different treatment groups. o, TEM images of NPCs under different treatments. p, q, Representative fluorescence images and quantitative analysis of Hoechst/PI staining in NPCs across treatment groups. Data are presented as mean ± SD from at least three independent experiments. n.s., not significant; *p < 0.05; **p < 0.01; ***p < 0.001 TMEM2 interacts with BAX to regulate mPTP opening Mitochondrial homeostasis disruption is a key pathological feature of IVDD [40]. Under mitochondrial stress, aberrant mPTP opening facilitates the release of DAMPs, including mtDNA, into the cytoplasm, where mtDNA activates the STING pathway and promotes cell death [17, 41]. To assess whether TMEM2 regulates mPTP opening in TBHP-treated NPCs, we performed fluorescence microscopy using calcein-AM. TMEM2 knockdown significantly increased mitochondrial fluorescence intensity, indicating mPTP closure (Fig. 5a–c). To further determine whether TMEM2 affects mtDNA release via mPTP, we examined the co localization of the mtDNA binding protein TFAM with the mitochondrial marker TOM20 by confocal microscopy. TBHP treatment reduced TFAM TOM20 co localization, whereas TMEM2 knockdown restored it (Fig. 5d, e). Direct quantification of cytosolic mtDNA (ND1 qPCR) confirmed that TBHP increased mtDNA leakage, which was reversed by siTMEM2, suggesting that TMEM2 silencing prevents mtDNA leakage(Supplementary Fig. S4 a). To elucidate the molecular mechanism, we investigated whether TMEM2 interacts with mPTP components. The mPTP is a multi-protein complex spanning the inner and outer mitochondrial membranes [42], including VDAC, BAX, HK2 (outer membrane), ANT, PiC, ATPase, CypD (inner membrane), and CK (intermembrane space) [43–45]. Co-immunoprecipitation (co-IP) revealed that TMEM2 binds to BAX under oxidative stress (Fig. 5f). This interaction was further supported by confocal microscopy showing enhanced co-localization of TMEM2 with BAX upon TMEM2 overexpression (Fig. 5g, h). Molecular docking predicted a binding energy of − 65.36 kJ/mol for the TMEM2-BAX complex (Fig. 5i). To test whether BAX mediates TMEM2-regulated mPTP opening, we used the BAX inhibitor BAI1 (5 µM) and the BAX activator BTSA1 (15 µM). BAX inhibition increased calcein-AM fluorescence (mPTP closure), while BAX activation decreased it (mPTP opening). Importantly, TMEM2 knockdown suppressed oxidative stress-induced mPTP opening, and this effect was partially reversed by the BAX agonist (Fig. 5j–l). Collectively, these results demonstrate that TMEM2 interacts with BAX to regulate mPTP opening under oxidative stress conditions. TMEM2 interacts with BAX to regulate mPTP opening. a–c, Representative fluorescence images of calcein_AM/Co²⁺ quenching staining, quantitative analysis of fluorescence intensity, and mPTP opening rate in NPCs treated with TBHP or siTMEM2 and TBHP for 24 h compared with control (decreased fluorescence indicates increased mPTP opening). d, e, Representative confocal images showing mtDNA (green) and mitochondria (red), along with co_localization analysis in NP cells treated as indicated for 24 h. f, Co_IP results showing interaction between TMEM2 and BAX, but not with other mPTP_related proteins (PiC, HK2, CK, ANT, ATPase, VDAC, or CypD) upon TMEM2 overexpression. g, h, Representative confocal images and co_localization analysis of BAX (green) and TMEM2 (red). i, Molecular docking illustrating a strong interaction between TMEM2 and BAX. j–l, mPTP opening assessed by calcein_AM/Co²⁺ quenching staining in NP cells treated with TBHP under modulation of BAX or TMEM2. BAX inhibitor: BAI1 (5 µM); BAX activator: BTSA1 (15 µM). Quantitative analysis of fluorescence intensity and mPTP opening rate is shown. Data are presented as mean ± SD from at least three independent experiments. n.s., not significant; **p < 0.01; ***p < 0.001 Preparation and characterization of Lipo/shTMEM2@MS Free liposome/shRNA complexes injected into the intervertebral disc have low viscosity and tend to migrate through tissue spaces, potentially spreading to the spinal cord and causing nerve compression or spinal cord injury [46]. Therefore, a delivery system capable of long-term controlled release at the lesion site is urgently needed. Accordingly, this study did not further evaluate the therapeutic effect of non-encapsulated liposome/shRNA complexes on IVDD. Degenerated nucleus pulposus tissue exhibits irregular morphology and heterogeneous distribution. Thus, developing an intelligent responsive platform that recognizes the degenerative microenvironment, releases liposome/shRNA complexes specifically at the diseased site, and remains inert in healthy tissue is essential for precise on-demand shRNA delivery. Based on this concept, Alg, PBA, and dextran were selected as the matrix materials. First, PBA was covalently conjugated to Alg via an amide bond, and RGD peptide was further coupled to enhance the adhesion and spreading of nucleus pulposus cells, yielding bifunctional alginate (Alg-PBA-RGD). Dextran, rich in vicinal diol groups, was used as the crosslinking partner. Alg-PBA-RGD and dextran were interconnected by dynamic boronate ester bonds, which are readily cleaved in the presence of ROS, thereby constructing ROS-responsive hydrogel microspheres. Subsequently, Lipo/shTMEM2 was mixed with the hydrogel precursor solution and fabricated into composite microspheres encapsulating Lipo/shTMEM2 (Lipo/shTMEM2@MS) using microfluidic technology (Fig. 6a). TEM showed spherical liposomes with a diameter of approximately 80 nm (Fig. 6b). SEM revealed that the freeze-dried microspheres exhibited an irregular porous network structure, with pore size decreasing as crosslinking density increased (Fig. 6c). This structure facilitates nutrient transport, cell migration, and proliferation. When suspended in PBS, the microspheres rapidly settled to the bottom of an Eppendorf tube (Fig. 6d). The ¹H NMR spectrum of Alg-PBA displayed a new set of signals at δ 7.2–7.8 ppm, corresponding to the aromatic protons of phenylboronic acid, confirming successful conjugation of PBA to the alginate backbone (Fig. 6e). Dynamic light scattering (DLS) analysis showed that the average diameter of blank liposomes was approximately 74 nm, which increased to 85 nm after complexation with shTMEM2 (Fig. 6f, g). The microspheres prepared by microfluidic technology exhibited excellent size uniformity, with an average diameter of 100 μm (Supplementary Fig. S5), suitable for injectable delivery (Fig. 6h). Fluorescence microscopy using DiI-labeled liposomes revealed intense red fluorescence inside the microspheres (Fig. 6i). Quantitative analysis showed that the encapsulation efficiency of Lipo/shTMEM2@MS was 73.5 ± 3.7% and the loading capacity was 14.1 ± 0.4% (w/w) (Supplementary Fig. S6). The zeta potential decreased from + 35.4 mV to − 12.5 mV, indicating successful binding of the negatively charged shTMEM2 to the cationic liposomes (Fig. 6j). Agarose gel electrophoresis showed a clear band for free shTMEM2, whereas no band was observed for intact Lipo/shTMEM2@MS, confirming that shTMEM2 was completely encapsulated within the microspheres (Fig. 6k). To determine the optimal N/P ratio for Lipo/shTMEM2 complexation, we performed gel retardation assays at N/P ratios of 3:1, 5:1, and 7:1. Complete complexation (absence of free shRNA band) was achieved at N/P = 5:1 and 7:1, while a faint band remained at N/P = 3:1 (Supplementary Fig. S7a). These data support the selection of N/P = 5:1 for subsequent experiments. Collectively, these results demonstrate the successful loading of shTMEM2 into the microspheres. The remaining mass of Lipo/shTMEM2@MS decreased from 100% on day 1 to 27.8% on day 28, indicating steady degradation over four weeks (Fig. 6l). To evaluate the ROS-responsive release behavior of Lipo/shTMEM2@MS, the microspheres were incubated in PBS with or without 1 mM H₂O₂, and the fluorescence intensity of the released nanoparticles in the supernatant was measured by fluorescence spectrophotometry to calculate the cumulative release rate (Fig. 6m). In PBS, shRNA was slowly released to approximately 33% over 12 days, indicating good stability under normal conditions. In contrast, in the presence of 1 mM H₂O₂, the release rate was significantly accelerated, reaching a cumulative release of 70% at 12 days, confirming that oxidative stress triggers boronate ester bond cleavage and enables on-demand shRNA delivery in the inflammatory microenvironment. NPCs were incubated with DiI-labeled Lipo/shTMEM2 for 6 h. Confocal microscopy showed DiI fluorescence distributed in the cytoplasm surrounding DAPI-stained nuclei, indicating efficient uptake of the complexes by NPCs (Fig. 6n, o). Therefore, these characterization results demonstrate the successful preparation of Lipo/shTMEM2@MS. Preparation and characterization of Lipo/shTMEM2@MS. a, Schematic of microfluidic fabrication. b, TEM image of liposomes. c, SEM image of freeze‑dried microspheres. d, Photograph of microsphere sedimentation. e, ¹H NMR spectra of Alg and Alg‑PBA. f, DLS size distribution of blank liposomes. g, DLS size distribution of Lipo/shTMEM2. h, Bright‑field microscopy image of Lipo/shTMEM2@MS. i, Fluorescence microscopy image of DiI‑labeled liposomes inside microspheres (red). j, Zeta potentials of blank liposomes and Lipo/shTMEM2. k, Agarose gel electrophoresis. l, Degradation profile of microspheres in PBS. m, ROS‑responsive release profile.n, Confocal images of NPCs incubated with DiI‑labeled Lipo/shTMEM2 (red) and DAPI‑stained nuclei (blue). o, Colocalization analysis Biocompatibility of Lipo/shTMEM2@MS The biocompatibility of Lipo/shTMEM2@MS was evaluated using NPCs. After culturing NPCs on the surface of Lipo/shTMEM2@MS microspheres for 3 and 5 days, live/dead staining showed that the proportion of viable cells exceeded 90% in both groups, with no significant difference between them (Fig. 7a, b). Immunofluorescence staining revealed intact cytoskeletons and well‑spread cell morphology, confirming good cytocompatibility of the microspheres, and RGD modification further enhanced cell adhesion (Fig. 7c, d). Live/dead staining also demonstrated that cell viability remained above 90% on days 1, 3, and 5, consistent with the CCK‑8 assay results (Fig. 7e–g). Scratch wound healing assay showed that TBHP treatment significantly suppressed NPC migration (24 h closure rate: 54.6%), whereas the Lipo/shTMEM2@MS group exhibited a closure rate of 77.2% at 24 h, which was significantly higher than that of the TBHP group, indicating that the microsphere sustained‑release system effectively restored the migratory capacity of damaged cells (Fig. 7h, i). In vivo fluorescence imaging revealed that after intradiscal injection of DiR‑labeled Lipo/shTMEM2@MS into rat caudal discs, strong fluorescence signals were detected at the injection site on days 1, 7, and 14, persisting up to day 14, indicating excellent retention of the microspheres within the intervertebral disc (Fig. 7j). Collectively, these results demonstrate the favorable biocompatibility of Lipo/shTMEM2@MS. Biocompatibility of Lipo/shTMEM2@MS.a, b, Live/dead staining and quantitative analysis of cells cultured on microspheres.c, d, Immunofluorescence staining of cytoskeleton and quantitative analysis of cells on microspheres.e–g, Live/dead staining and CCK_8 results with quantitative analysis.h, i, Scratch wound healing assay and quantitative analysis.j, In vivo fluorescence imaging.Data are presented as mean ± SD from at least three independent experiments. n.s., not significant; ***p < 0.0001;****p < 0.0001 Lipo/shTMEM2@MS ameliorates IVDD in SD rats To evaluate the therapeutic efficacy of Lipo/shTMEM2@MS in a rat model of IVDD, a percutaneous needle puncture‑induced IVDD model was established using Sprague-Dawley rats. Rats were randomly divided into four groups: control (non‑punctured), model 1 (puncture + PBS), model 2 (puncture + Lipo/NCshRNA@MS, loaded with non-targeting shRNA), and model 3 (puncture + Lipo/shTMEM2@MS) (Fig. 8a). MRI examination at 4 weeks post-operation showed that the Pfirrmann grade score in the model 3 was significantly lower than that in the model 1 and model 2 (Fig. 8b, c). H&E and Safranin O‑fast green staining demonstrated that the model 3 exhibited markedly reduced NP loss and less severe histological degeneration compared with the model 1 and model 2 (Fig. 8d–f). IHC further revealed decreased protein expression levels of GSDMD, NLRP3, and Caspase‑1 in the model 3 (Fig. 8g–j). Collectively, these results indicate that Lipo/shTMEM2@MS effectively ameliorates disc degeneration in the puncture‑induced rat model. Lipo/shTMEM2@MS ameliorates IVDD in SD rats. a, Schematic diagram of the experimental design in the rat model (n = 9). b, c, Representative MRI images of rat tail discs at 4 weeks after puncture surgery and quantitative analysis based on the Pfirrmann grading system. d, Representative H&E staining and Safranin O-fast green staining of intervertebral discs from different groups at 4 weeks post-surgery. e, f, Histological grading scores for individual parameters and the total histological score at 4 weeks post-surgery. g-j, Representative immunohistochemical staining of Caspase-1, GSDMD, and NLRP3 in midsagittal sections of rat intervertebral discs. Data are presented as mean ± SD from at least three independent experiments. **p < 0.01; ***p < 0.001; ****p < 0.0001 Discussion Current therapeutic strategies for IVDD mainly consist of conservative management and surgical intervention, which mostly alleviate symptoms but fail to target the underlying biological mechanisms driving disease progression [47]. In this study, through scRNA‑seq of human degenerated nucleus pulposus tissues, we found that TMEM2 was markedly upregulated in degenerated NPCs and its expression positively correlated with pyroptosis‑related genes, suggesting that TMEM2 may be involved in regulating NPCs pyroptosis. Based on this, we further confirmed that TMEM2 is a key regulator of NPCs pyroptosis and developed a reactive ROS‑responsive microsphere system for sustained delivery of shTMEM2. Both in vitro and in vivo experiments demonstrated that this strategy effectively reduces NPCs pyroptosis, repairs damaged discs, and delays degenerative progression. TMEM2 is a transmembrane protein previously studied mainly for its role in HA degradation and tumor progression, where its expression correlates positively with invasiveness and poor prognosis [33, 48, 49]. A recent study reported that TMEM2 regulates mitochondrial homeostasis by activating the TGF-β/SMAD signaling pathway, upregulating mitochondrial fission-related genes, thereby promoting mitochondrial fragmentation, suppressing respiratory function, and exacerbating oxidative stress [30]. Mitochondrial morphological and functional disturbances are closely linked to various cell death programs, including pyroptosis [50]. In line with these observations, our data demonstrated that TMEM2 expression progressively increased with the degenerative grade of NPCs and was significantly upregulated under oxidative stress. Mechanistically, we found that TMEM2 promotes aberrant opening of the mPTP, leading to mtDNA leakage. Knockdown of TMEM2 improved mitochondrial function, attenuated oxidative stress, and significantly inhibited NPCs pyroptosis. These results directly link TMEM2 to mitochondria-dependent pyroptosis, suggesting that TMEM2 may act as a key driver of IVDD by affecting mPTP opening. Under oxidative stress, aberrant mPTP opening leads to mtDNA release into the cytosol [17]. The release of mtDNA involves multiple cellular processes, most importantly the opening of release channels. Previous studies have shown that mPTP opening is traditionally regulated by calcium overload and Bcl‑2 family proteins [51, 52]. Duan et al. found that under hypoxic conditions, mitochondrial Drp1 promotes excessive mPTP opening through interactions with BAX‑PiC and LRRK2‑HK2, thereby aggravating mitochondrial damage and cell death [53]. Importantly, we discovered that the interaction between TMEM2 and BAX also regulates mPTP opening, leading to mtDNA release. The leaked mtDNA acts as a damage‑associated molecular pattern (DAMP) that activates the STING signaling pathway, which in turn phosphorylates NF‑κB via TBK1, ultimately triggering NLRP3 inflammasome‑mediated NPC pyroptosis. Future studies should explore the structural basis of the TMEM2‑BAX interaction and whether post‑translational modifications regulate their binding, as this will help identify new therapeutic targets and provide theoretical foundations for more effective treatment strategies. The avascular nature of the intervertebral disc limits the retention of free liposome/shRNA complexes [54]. Previous attempts using non-responsive hydrogels or repeated injections of growth factors have shown limited efficacy due to rapid clearance [55–58]. We designed a ROS-responsive microsphere system based on dynamic boronate ester bonds between Alg-PBA-RGD and dextran, which degrade in the presence of H₂O₂, enabling on-demand release of Lipo/shTMEM2. Although various nanohydrogel platforms have been reported for scavenging ROS, suppressing inflammation, and reducing ECM degradation, the ROS-responsive microsphere platform developed in this study exhibited superior performance in increasing NPC viability and migration and reducing pyroptosis in vitro [59]. The RGD motif further promoted cell adhesion. Overall, Lipo/shTMEM2@MS shows therapeutic potential for IVDD and warrants further systematic preclinical studies. Although this study provides a new strategy for IVDD treatment, several limitations exist. First, lack of immunofluorescence co‑staining for TMEM2 and fibrotic markers in clinical samples is a limitation; however, our scRNA‑seq data support their association. Second, only TMEM2 knockdown experiments were performed to investigate nucleus pulposus cell pyroptosis, without corresponding overexpression experiments. Therefore, the necessity of TMEM2 in NPC pyroptosis and extracellular matrix changes has been demonstrated. Future studies with TMEM2 overexpression are warranted to further establish the causal relationship. Third, immunohistochemical staining for TMEM2 and key components of the STING/TBK1/NF‑κB pathway in rat disc samples was not carried out in this study. Future work should address this to confirm the in vivo relevance of the pathway. Fourth, the use of advanced three‑dimensional organoid models could provide a more physiologically relevant platform to study IVDD pathogenesis and therapeutic responses in future investigations [60].Fifth, the current rat tail needle puncture model, although widely used, does not fully recapitulate the chronic, progressive nature of human IVDD, which involves multifactorial etiologies such as aging, mechanical overload, and metabolic dysregulation. Future studies should consider more clinically relevant animal models (e.g., age‑related spontaneous degeneration or mechanically induced models) to better mimic the human disease process and improve translational predictability. In summary, this study reveals a novel mechanism by which TMEM2 drives NPC pyroptosis through regulation of mPTP opening, mtDNA release, and STING pathway activation. Based on this mechanism, we developed a ROS‑responsive composite microsphere delivery system (Lipo/shTMEM2@MS). This system exhibits good biocompatibility, on‑demand release of shTMEM2 in the inflammatory microenvironment, effective silencing of TMEM2, blockade of the pyroptosis pathway, and significant retardation of disc degeneration. Both in vitro and in vivo experiments confirmed the safety and efficacy of this strategy. Collectively, Lipo/shTMEM2@MS represents a promising translational approach for IVDD treatment, warranting further preclinical studies and long‑term safety assessments. Conclusion This study demonstrates for the first time that TMEM2 drives nucleus pulposus cell pyroptosis via the BAX-mPTP-mtDNA-STING signaling axis, establishing TMEM2 as a potential therapeutic target for IVDD. Based on this mechanism, we developed an injectable ROS-responsive hydrogel microsphere system (Lipo/shTMEM2@MS) for sustained local delivery of shTMEM2. The system exhibits excellent biocompatibility, ROS-triggered degradation, and sustained release properties, effectively silencing TMEM2 expression, suppressing pyroptosis, and significantly retarding disc degeneration in a rat model. Collectively, this study provides a translational strategy for IVDD therapy that combines TMEM2-targeted gene therapy with an intelligent drug delivery platform. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements The authors thank all study participants. Author contributions Z.X., Z.L. and X.W. conceived and designed the study. Z.X. and H.F. performed the in vitro experiments. Z.X. and H.F. assisted with the animal surgeries. Z.X. collected the clinical samples. C.P. and G.Y. analyzed the scRNA-seq data and provided interpretation. Z.X. wrote and edited the manuscript. H.Q., J.A. and X.W. reviewed and edited the manuscript. All authors read and approved the final manuscript. Funding This work was supported by the National Natural Science Foundation of China (NO.82572788). Data availability Data availability statementThe single‑cell RNA sequencing data supporting this study are available from the GEO database under accession number GSE165722. All other data generated or analyzed during this study are included in this published article and its supplementary information files. Raw data are available from the corresponding author upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Zhiqiang Xu and Hanmo Fang contributed equally to this work. Contributor Information Jun Ao, Email: ao00jun@163.com. Xinghuo Wu, Email: wuxinghuo@163.com. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement Data availability statementThe single‑cell RNA sequencing data supporting this study are available from the GEO database under accession number GSE165722. All other data generated or analyzed during this study are included in this published article and its supplementary information files. Raw data are available from the corresponding author upon reasonable request.