P23 acts as a negative regulator of ferroptosis in NSCLC by blocking GPX4 degradation via chaperone-mediated autophagy

Abstract

Ferroptosis has been identified as a tumor-inhibiting event in a variety of cancers; however, its molecular basis in non-small cell lung cancer (NSCLC) has not been completely elucidated. Notably, glutathione peroxidase 4 (GPX4) plays a crucial role in ferroptosis. Our previous research revealed that prostaglandin E synthase 3 (p23), a potential transcription factor, plays a crucial role in promoting cancer progression and metastasis through succinylation. Our study revealed a previously unknown antiferroptotic function of p23. Mechanistically, p23 stabilizes GPX4 by competitively binding heat shock cognate 71 kDa protein (HSC70) to suppress chaperone-mediated autophagy (CMA) activity, which subsequently inhibits ferroptosis and accelerates tumor growth. Notably, impairing p23 succinylation disrupts its interaction with HSC70, restoring CMA-mediated GPX4 degradation. Collectively, our findings suggest that targeting p23-regulated CMA pathways represents a potentially viable strategy to modulate ferroptosis in NSCLC.

Graphical Abstract

The role of p23 in competing with GPX4 for binding to HSC70, blocking CMA-mediated degradation of GPX4 and inhibiting ferroptosis

Supplementary Information

The online version contains supplementary material available at 10.1186/s12943-025-02439-y.

Introduction

Globally, the incidence and mortality rates of lung cancer are exceedingly high, making it the leading cause of death among patients with malignant tumors, with a notable increasing trend [1]. Among these cases, NSCLC has a relatively high mortality rate [2]. With developments in diagnosis, surgical techniques, and antitumor medications, the 5-year survival rate has significantly improved [3]. However, numerous clinical data suggest that the overall prognosis of NSCLC patients remains unfavorable [2, 4, 5]. Significant progress has been made over the past two decades in understanding disease biology, utilizing predictive biomarkers, and developing therapeutic strategies, all of which have shown great promise in improving patient outcomes [6, 7].

Ferroptosis, an iron-dependent mode of cell death, distinctly differs from other forms of programmed cell death and is characterized by excessive lipid peroxidation [8, 9]. When the extent of peroxidation exceeds the antioxidant capacity of the ferroptosis defense system, the accumulation of lipid peroxides on the cell membrane triggers membrane rupture and iron release, ultimately culminating in cell death [10]. Recent research highlights the pivotal role of ferroptosis in tumor suppression and cancer therapy resistance [11, 12] and underscores its potential as a therapeutic target for lung cancer treatment [13, 14]. Ferroptosis is modulated by multiple regulatory mechanisms. Among these pathways, the control of lipid peroxides via the GPX4–GSH pathway stands out as one of the most critical [15]. As the core regulator of ferroptosis, GPX4 maintains cell membrane homeostasis by scavenging lipid peroxides (L-OOH) [16]. By catalyzing the reduction of lipid hydroperoxides to nontoxic lipid alcohols (L-OH) through glutathione (GSH), GPX4 blocks the lipid radical chain reaction and thereby inhibits ferroptosis [17, 18]. The unique three-dimensional structure of GPX4, especially the selenocysteine in the active center, endows it with the specific ability to clear membrane-bound phospholipid peroxides [19]. Studies have shown that intervention with pharmacological inhibitors (such as RSL3) or genetic knockout of GPX4 significantly increases the sensitivity of cells to ferroptosis, confirming the irreplaceable role of GPX4 in the regulation of ferroptosis [20, 21].

Recent studies have made significant progress in exploring The regulatory mechanisms of glutathione peroxidase 4 (GPX4) stability, revealing various strategies for its effective regulation. At the protein degradation level, proteolysis-targeting chimeras (such as compound 18a) facilitate GPX4 degradation through ubiquitination-mediated proteasomal pathways, demonstrating remarkable antitumor activity in fibrosarcoma cells [22]. The small-molecule degrader N6F11 specifically induces GPX4 degradation in tumor cells via TRIM25-mediated ubiquitination while preserving immune cell function, showing potential for synergistic effects with immunotherapy [23]. At the posttranslational modification level, ZDHHC20-mediated palmitoylation acts on cysteine 66 of GPX4 to promote its stabilization. Conversely, inhibiting this modification with 2-bromopalmitate increases cellular susceptibility to ferroptosis and suppresses tumor progression [24]. Moreover, multiple studies have revealed the close link between GPX4 level regulation and tumor development from different aspects. Cheng L. et al. reported that serum/glucocorticoid-regulated kinase 2 (SGK2) affects prostate cancer cell metastasis by regulating GPX4 levels [25]. USP8 can stabilize GPX4 through deubiquitination, and inhibiting USP8 significantly enhances the sensitivity of colorectal cancer model cells to ferroptosis [26]. Zhu S. et al. reported that after heat shock 70 kDa protein 5 (HSPA5) interacts with GPX4, it reduces GPX4 degradation, protecting human pancreatic ductal adenocarcinoma (PDAC) cells from ferroptosis [27]. Together, these findings reveal the complexity of the GPX4 degradation regulatory network, highlighting its great potential value in cancer treatment. Targeting GPX4 has emerged as a promising and effective strategy for selectively inducing ferroptosis in tumor cells [28]. Therefore, exploring the underlying molecular mechanisms regulating GPX4 has significant implications for cancer prevention and treatment.

As a highly conserved protein, p23 functions as a molecular cochaperone within the HSP90 complex, contributing to the stabilization of multiple client proteins [29]. Research has demonstrated that p23 is highly expressed in various tumors, including NSCLC, prostate cancer, and breast cancer, where it promotes tumor metastasis and drug resistance [30–32]. Increasing evidence suggests that p23 may perform certain functions independent of HSP90, including serving as a potential transcription factor. Our previous research revealed that p23 knockdown in NSCLC significantly induced cell death [32]. To investigate the underlying mechanism, we applied various cell death inhibitors, including ferroptosis inhibitors, and found that a ferroptosis inhibitor (Fer-1) markedly reversed p23 knockdown-induced cell death (Figure S1A), suggesting that p23 might play a crucial role in the ferroptosis process of tumor cells. Moreover, previous studies have reported a potential connection between p23 and ferroptosis [33], which further piqued our interest. Therefore, we selected p23 as our research target. However, the potential relationships among p23, GPX4 and ferroptosis, as well as the molecular mechanism by which p23 regulates ferroptosis to influence tumor progression, remain to be elucidated.

CMA is a key autophagic pathway involved in selective protein degradation [34]. This progression involves the recognition of cytoplasmic proteins carrying KFERQ motifs by HSC70, which subsequently direct these proteins to the lysosomal membrane, where they bind to LAMP2A [35]. At this stage, LAMP2A undergoes polymerization into tetramers to facilitate the internalization of substrates [36]. Growing evidence suggests a fundamental connection between CMA and cancer-related ferroptosis [37–39]. Thus, targeting CMA substrate proteins has become a promising strategy in cancer therapy, as they share a common degradation pathway. However, the precise mechanisms require further investigation. In this study, we identified a novel role of p23 in suppressing ferroptosis. Further research revealed that p23 hinders GPX4 degradation by interacting with HSC70 in the CMA pathway without affecting endogenous GPX4 gene expression. Our findings, for the first time, reveal the essential role of p23 in the CMA pathway and elucidate the mechanisms through which p23-mediated ferroptosis inhibition promotes NSCLC progression.

Materials and methods

Mice and xenograft model

6- to 8-week-old non-transgenic (NTG) mice were procured from the spfbiotech corporation. The use of animals in this study was approved by the Institutional Animal Care and Use Committee of Dalian Medical University (approval number: AEE19048). A total of 1 × 10^7 A549 cells were subcutaneously injected into The NTG mice. Tumor sizes were measured every three days. The experiment was continued until the volume of the xenograft tumors reached 150 mm³. Thirty-one days after the injection, the mice were euthanized, and the tumors were harvested for subsequent experiments.

Cell lines

The human A549, H1299, HEK293T, HLF, H460, and H358 cell Lines were sourced from the ATCC. These cell Lines were cultured and maintained in a humidified incubator at 37 °C with 5% CO₂. For cell cultivation, either RPMI 1640 or Dulbecco’s Modified Eagle Medium (DMEM) was employed, each supplemented with 10% fetal bovine serum (FBS). To ensure their identity and purity, the authenticity of all cell lines was rigorously verified using genomic short tandem repeat (STR) profiling analysis.

Human experimental samples and study details

Clinical sample collections were approved by the Ethics Committee of the First Affiliated Hospital of Dalian Medical University (PJ-KS-KY-2023-362). Clinical samples were obtained from 10 patients with NSCLC. Twenty milligrams of The samples were ground and lysed for 2 h. The supernatant was centrifuged at 20,000 g for subsequent WB and mRNA detection. Tissue microarrays were purchased from Shanghai Outdo Biotech Company and approved by the Ethics Committee of Shanghai Outdo Biotech Company (SHYJS-CP-1904014, SHYJS-CP-1701003). The expression levels of p23 and GPX4 were quantified following histochemical staining (IHC) for p23 and GPX4 in two respective tissue microarrays. Survival analysis was conducted using the “Log-rank test” based on the quantitative expression levels of these markers.

Plasmids and lentivirus

3^∗Flag-p23, pCDH-3^∗Flag-GPX4, pCDH-3^∗Flag-Asp60A, pCDH-3^∗Flag-Glu81A, pCDH-3^∗Flag-Ser85A, pCDH-3^∗Flag-Phe56A, pCDH-3^∗Flag-Arg12A, pCDH-3^∗Flag-Arg93A, pCDH-3^∗Flag-His57A, pCDH-3^∗Flag-All mut, pCDH-3^∗Flag-K7/K33 and pLKO.1-RFP-shp23 constructs were developed following standardized protocols. HEK293T cells were seeded at the 50% density for transfection for lentivirus production. The next day, the medium was removed, and cells were washed with PBS and incubated with serum-free DMEM for a 3-hour starvation period. Transfection was performed at a fixed ratio for 6 h, and The media was replaced with fresh complete medium. After 48 h, The supernatant was collected, passed through a 0.22 µM filter, and concentrated using PEG 8000. The viral precipitate was re-suspended in a complete medium, and Polybrene was added to a final concentration of 8 µg/mL before being applied to target cells. After 48 h, target cells were subjected to puromycin selection. After three days, surviving cells were passaged normally, with puromycin maintained for continued selection. All experiments involving GPX4 expression were conducted under conditions with sodium selenite (200 nM) to eliminate its potential interference with the experimental outcomes.

Cell viability assay

After normal digestion, cells in each group were counted and diluted to an average density of 6000 cells per well in 96-well, and treated with DMSO, erastin (20 µM) or RSL3(10 µM) for 36 h. Cell viability was then evaluated using the CCK-8 assay. The CCK-8 solution was diluted 1:10 with fresh blank medium to form a working solution, and 100 µl of The working solution was added to the well. The plates were then placed into a cell incubator for 1 to 3 h at 37 °C, and absorbance was measured at 450 nm using an enzyme-labeled instrument.

Lipid ROS and ROS assay using flow cytometer

Cells were plated at a density of 2.5 × 10^5 cells and cultured overnight. After 36 h of treatment with DMSO, erastin (20 µM) or RSL3(10 µM), The medium was removed, and cells were collected into 1.5 mL tubes. Each tube was supplemented with 500 µL of serum-free medium containing either 5 µM BODIPY™ 581/591 C11 or 10 µM DCFH-DA probes and incubated for 30 min, with gentle inversion every 5 to 10 min. Cells were washed twice with PBS, resuspended in PBS, and passed through a 0.4 μm nylon mesh. Intracellular Lipid peroxidation and total ROS levels were assessed by flow cytometry using a BD FACS Aria cytometer, analyzing a minimum of 10,000 cells per condition.

Western blot analysis

Cells were lysed using Western and IP cell lysis buffers, and The lysates were subjected to SDS-PAGE for protein separation. The proteins were then transferred onto a PVDF membrane and blocked with 5% skim milk in TBST at room temperature for at least one hour. The membrane was incubated overnight at 4 °C with specific primary antibodies. The next day, after washing with TBST, it was incubated with the secondary antibody at room temperature for two hours. Protein detection was performed using an ECL luminescent solution.

Silver staining of protein

The kit used for silver staining of proteins was purchased from Beyotime. Cells were plated at a density of 2.5 × 10^5 cells cultured overnight, and were transfected with control or sip23 next day. After 48 h, cells from both groups were collected for lysis and SDS-PAGE protein separation. The WB gel was placed in an appropriate volume of fixing solution and incubated overnight at room temperature with gentle shaking. The next day, the fixing solution was removed, and the gel was treated with 30% ethanol, shaking at 60–70 rpm for 10 min at room temperature. The gel was incubated with a silver dye sensitizing solution, silver solution, silver dye coloring solution, and silver dye stopping solution. The silver-stained gel was preserved in three changes of distilled water at 4 °C and prepared for subsequent MS identification.

RNA extraction, cDNA synthesis, and RT-PCR analysis

Total RNA was reverse-transcribed into cDNA using the TaqMan Reverse Transcription Reagents Kit. The fold change was determined using the comparative CT method and normalized to β-actin expression. The primer sequences used for PCR were: p23 forward, 5′-cagtcatggccaaggttaaca-3′ and p23 reverse, 5′-tcctcatcaccacccatgtt-3′; GPX4 forward, 5′-tcagcaagatctgcgtgaac-3′ and GPX4 reverse, 5′-atagtggggcaggtccttct-3′; HSC70 forward, 5′-gcctaccttgggaagactgt-3′ and HSC70 reverse, 5′-gcacgtttctttctgctcc-3′; LRRK2 forward, 5′-tgatgatgagggggaagaag-3′ and LRRK2 reverse, 5′-tccctatgagctgggaaatg-3′; MEF2D forward, 5′-cagcagccagcactacagag-3′ and MEF2D reverse, 5′-cagcagccagcactacagag-3′; LAMP2A forward, 5′- cgttctggtctgcctagtcc-3′ and LAMP2A reverse, 5′-cagtgccatggtctgaaatg-3′; LAMP1 forward, 5′-ctgcctttaaagctgccaac-3′ and LAMP1 reverse, 5′-tgttctcgtccagcagacac-3′; GAPDH forward, 5′-gt cagtggtggacctgacct-3′ and GAPDH reverse, 5′-tgctgtagccaaattcgttg-3′; β-actin forward 5′-gctcgtcgtcgacaacggct-3′ and β-actin reverse, 5′-caaacatgatctggctcatcttctc-3′.

RTCA real-time labeling free cell analysis

After normal digestion, cells in each group were counted and diluted to an average density of 3000 cells per well. After The analyzer was opened, the experiment was named, and various parameters were set. 50 µL of The corresponding blank medium was added to the special orifice plate to remove the background value. The pore plate was removed, and then 100 µL of medium-diluted cell suspension was added to The pore plate, which was placed in the incubator for more than 30 min, and then the pore plate was placed in a real-time labeled unlabeled cell analyzer to monitor cell proliferation. After the experiment, the cell proliferation curve was obtained, and the influence of gene recombination expression construction on cell proliferation was analyzed.

Immunohistochemistry (IHC)

We used the kit (PV-9000 and ZLI-90118) from ORIGENE for IHC detection. Tissue sections from Mouse models were placed in an electric oven at 65 °C for at least 2 h for dewaxing. Next, The sections were immersed in xylene for 15 min twice, using fresh xylene for The second immersion. Then, sequentially immerse them in alcohol of varying concentrations for 5 min each. Finally, the sections were washed twice with PBS. Antigen retrieval was performed using the boiling method. Tissue sections were placed in a citric acid repair buffer (pH 6) and heated to a boil over medium heat for 5 min, repeated four times. After naturally cooling to room temperature, The sections were washed twice with PBS for 5 min each. To block endogenous peroxidase activity, enough 3% H₂O₂ was applied to fully cover The tissue and incubated in the dark at room temperature for 10 min. The sections were then washed three times with PBS. The sections were covered with 10% goat serum drops for at least 30 min. The sections were not washed and dried. An appropriate amount of p23, GPX4, and other corresponding antibodies diluted with goat serum were added to The drops and incubated at 4 °C overnight. On The next day, the slides were cleaned three times with PBS solution, then the appropriate amount of HRP-labeled secondary antibody was added, incubated at room temperature for at least 1 h, and washed three times with PBS. A few drops of DAB solution were applied to The tissue sections in a dark environment, allowing color development at room temperature. The reaction was then promptly stopped by rinsing with distilled water. After drying, hematoxylin was applied to stain the nuclei for approximately 10 s, followed by immediate rinsing with distilled water to stop The reaction. A differentiation solution was then applied for 2 s, and the reaction was quickly stopped. Finally, the sections were sealed, observed, and photographed under a microscope.

Immunofluorescence (IF)

The cover-slip was sterilized with 100% ethanol, dried, placed in a six-well plate, and The cells were seeded at 2× 10⁵ cells per well onto it to attach to The surface. After washing with PBS, the cells were fixed with 4% paraformaldehyde at room temperature for 30 min and Then permeabilized with 0.1% Triton X-100 for 5 min. Following PBS washing, overnight incubation at 4 °C with The corresponding concentration of specific primary antibody or probes. The next day, the samples were incubated with the corresponding fluorescent secondary antibody at room temperature for 2 h (this step was not required for the probe). Subsequently, The slides were stained with DAPI for 5 min and washed with PBS. The slides were sealed with Mounting Medium, antifading, and imaged using a laser confocal microscope.

Frozen tissue Sect. (10 μm thick) were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and hybridized with BODIPY™ 581/591 C11 prob diluted in hybridization buffer 30 min at 37 °C. After washing with SSC buffer to remove unbound probes, nuclei were counterstained with DAPI. Sections were mounted with anti-fade mounting medium and imaged using a laser scanning confocal microscope with appropriate excitation/emission wavelengths for each fluorophore.

Isolation of lysosomes

Lysis and density gradient solutions were prepared according to The required dilution ratios, including 250 mM CaCl₂, 1 × extraction buffer, and 1 × OptiPrep density gradient medium. At least 2 × 10⁸ control and transfected cells were harvested, digested, and centrifuged at 600 g for 5 min to collect The cell pellet and washed with PBS. The pellet was resuspended in 2.7 mL of 1 × extraction buffer and gently mixed. The cells were then homogenized twice using a Dounce homogenizer to ensure proper lysis and supernatant centrifuged at 20,000 g for 20 min at 4 °C to obtain crude lysosomal pellets.

Native gel electrophoresis

Loading buffer without SDS or other denaturing agents was added to the lysosome complex without prior boiling. Natural gel separation adhesive and concentration adhesive (without SDS) of 8% were prepared according to the molecular weight of LAMP2A. After loading The samples, electrophoresis was carried out at a constant voltage of 80–120 V and The gel was transferred onto PVDF membrane at 40 V in an ice bath for 1.5 h. After The transfer, the membrane was incubated in an 8% acetic acid solution for 15 min and washed three times with TBST at room temperature. Next, The PVDF membrane was blocked in a 5% BSA skim milk solution at room temperature for at least 1 h and incubated LAMP2A antibodies overnight. The next day, the membrane was incubated with the second antibody at room temperature for 2 h, washed with TBST, and revealed by ECL luminescence in a gel imager.

Measurement of CMA activity in intact cells

Cells transfected with The stable pSIN-PAmCherry-KFERQ-NE plasmid were subjected to the appropriate treatments and then excited at 405 nm for at least 5 min. On The second day, after washing the medium with PBS, the cells were fixed with 4% paraformaldehyde for 30 min. Subsequently, The slides were stained with DAPI for 5 min and washed with PBS. The slides were sealed with Mounting Medium, antifading, and imaged using a laser confocal microscope.

Co-immunoprecipitation (Co-IP)

The IP binding buffer was prepared with 150 mM NaCl, 50 mM Tris-HCl, 0.2% Triton X-100, and 0.2% NP-40. At least 400 µg of protein was lysed, and The total volume was adjusted to 500 µL using The binding buffer. Pre-treated IP magnetic beads were added, followed by incubation with the appropriate antibodies while shaking at 4 °C overnight. The next day, the mixture was washed four times with the binding buffer using a magnetic rack. Finally, the samples were boiled at 95 °C for 10 min for subsequent Western blot analysis.

7.5ug p23 pure protein with Succinyl CoA sodium salt (20 mA) in Buffer (Tris-HCl 8.0 50mM, KCl 50mM, EDTA 0.1mM, DTT 1mM, PMSF 1mM, sodium butyrate 50mM, Nicotinamide 10mM,) for 20 min at 37 ° C, purified p23 and succinylated p23 were incubated with whole cell lysates and then co-immunoprecipitated with HSC70 antibody (Co-IP).

Mass spectrometry analysis

The gel strips were enzymatically digested by adding 100 µL of solution containing 100 mM TEAB, 1 µg trypsin, and 1/300 CaCl2, followed by overnight incubation at 37 °C. After low-speed centrifugation, The supernatant was collected, mixed with 200 µL acetonitrile, and vortexed. Peptides were extracted with 100 µL of 0.1% formic acid, pooled, and centrifuged at 12,000 g for 5 min at room temperature before freeze-drying. The resulting powder was dissolved in 0.1% formic acid, loaded onto a C18 desalting column, washed three times with 0.1% formic acid (3% acetonitrile), and eluted twice with 0.1% formic acid (70% acetonitrile). The eluate was lyophilized and analyzed using a Q ExactiveTM HF-X mass spectrometer equipped with a Nanospray Flex™ (ESI) ion source (2.1 kV, 320 °C capillary temperature). The scanning range was m/z 350–1500 with a resolution of 60,000 (m/z 200), AGC target of 3 × 10^6, and ion injection time of 20 ms. The top 40 precursors from each full scan were selected for HCD fragmentation in MS/MS mode (resolution 15,000, AGC target 1 × 10^5, maximum injection time 45 ms, normalized collision energy 27%, intensity threshold 2.2 × 10^4, dynamic exclusion 20 s).

Bioinformatics analysis

A total of 146 proteins (15 to 20 kDa) were identified in specific bands using mass spectrometry. Will these proteins associated with ferroptosis pathways of protein (http://www.zhounan.org/ferrdb/current/) intersection, can find is regulated by the sip23 iron death related genes, and draw the corresponding Wayne figure.

Co-IP samples were subjected to SDS-PAGE for protein separation. After 1 cm of separation, the strips were identified by mass spectrometry. The resulting spectra from each fraction were searched separately again ‘’homo_sapiens_uniprot ‘’ database by the search engines: Proteome Discoverer.

Results

p23 negatively regulates ferroptosis in NSCLC

To explore the correlation between p23 and ferroptosis in NSCLC, we initially assessed mitochondrial alterations following the siRNA-mediated knockdown of p23 (sip23) using transmission electron microscopy (TEM). The results indicated that sip23 (Fig. 1A) disrupted the mitochondrial outer membrane, caused mitochondrial shrinkage, and led to the loss of mitochondrial cristae (Figures B and S1B-C, red arrow). Furthermore, we used MitoTracker staining to observe the impact of p23 knockdown on mitochondrial structure through confocal microscopy. The results revealed that p23 knockdown significantly promoted mitochondrial fragmentation and structural changes (Figure S1D). In addition, sip23 facilitated erastin-induced generation of reactive lipid peroxides and reactive oxygen species (ROS) and suppressed A549 and H1299 cell viability (Fig. 1C-E). This was accompanied by key ferroptosis-related events, including increased oxidized glutathione (GSSG), depletion of glutathione (GSH), accumulation of ferrous ions, and elevated malondialdehyde (MDA) (Fig. 1F-I). Conversely, stable p23 overexpression (OEp23) (Fig. 1J) reversed ferroptosis-associated indicators in NSCLC, exhibiting effects opposite those of sip23 (Fig. 1K-Q). To further validate the role of p23 in ferroptosis, we assessed H350 and H460 cells and obtained similar results (Figure S2). These findings suggest that p23 exerts a negative regulatory effect on ferroptosis in NSCLC.

Fig. 1.

p23 negatively regulates erastin-induced ferroptosis in NSCLC. A Western blot (WB) analysis of p23 expression in whole-cell lysates (WCLs) from IN cells compared with β-actin levels; n = 3. B TEM scanning of mitochondrial morphology in the control group and sip23 group. C and D NC or sip23 cells were treated with erastin (20 µM, 36 h). Lipid ROS (C) and total ROS (D) levels were Then measured using flow cytometry with BODIPY™ 581/591 C11 and DCFH-DA probes. E NC or sip23 cells were treated with erastin (0–40 µM, 36 h), after which a CCK-8 kit was used to assess cell viability; n = 3. F-I The concentrations of GSSG (F), GSH (G), Fe2+ (H), and MDA (I) in NC or sip23 cells were measured using the appropriate assay kits; n = 3. J WB analysis of p23 expression in vector or OEp23 cells compared with that of β-actin; n = 3. K and L Vector or OEp23 cells were treated with erastin (20 µM, 36 h). Lipid ROS (C) and total ROS (D) levels were Then measured using flow cytometry with BODIPY™ 581/591 C11 and DCFH-DA probes. (M) Vector or OEp23 cells were treated with erastin (0–40 µM, 36 h), after which a CCK-8 kit was used to assess cell viability. N-Q The concentrations of GSSG (N), GSH (O), Fe2+ (P), and MDA (Q) in vector or OEp23 cells were measured using the appropriate assay kits; n = 3. Quantification graphs: The data represent the mean ± standard deviation from at least three independent experiments. Statistical analyses were performed using t tests (A, F-I, J, N-Q) or two-way ANOVA (E, M). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001

p23 deficiency induces GPX4 downregulation and inhibits lung cancer development

To explore the molecular mechanism by which p23 regulates ferroptosis, we utilized polypropylene gel electrophoresis combined with silver staining. This analysis revealed a significantly reduced protein band in the 15–20 kDa range following sip23 treatment (Fig. 2A). A total of 146 proteins in this specific band were identified through mass spectrometry (MS). Notably, three of these proteins (GPX4, CISD2, and MGST1) are associated with negative regulation of ferroptosis (ferroptosis suppressor database originated from http://www.zhounan.org/ferrdb/current/) (Fig. 2B). Western blot analysis further confirmed that GPX4 is a potential target of p23 in the negative regulation of ferroptosis. Knockdown of p23 led to a significant decrease in GPX4 levels, whereas CISD2 and MGST1 levels remained unchanged in A549 and H1299 cells (Figs. 2C and S3A) as well as in H358 and H460 cells (Figure S3B). Additionally, OEp23 increased GPX4 protein levels, whereas sip23 reduced GPX4 protein levels across all four cell lines (Figs. 2D-E and S3C-E). Moreover, we observed significantly greater expression of p23 and GPX4 in the lung cancer cell lines than in the normal cell lines (Figs. 2F and S3F). Therefore, we analyzed p23 and GPX4 expression in tumors and adjacent tissues from 10 lung cancer patients (Figs. 2G and S3G). The results revealed that both proteins were predominantly expressed in tumor tissues, suggesting a potential correlation. To further examine this relationship, we performed immunohistochemistry (IHC) on a commercial tissue microarray (n = 98) (Figs. 2H and S3H). Our analysis revealed a significant positive correlation between GPX4 and p23 expression levels, as summarized in Table 1. Moreover, further statistical analysis revealed that patients with low GPX4 and p23 expression levels had a better survival rate, whereas those with elevated GPX4 and p23 expression had a worse prognosis. Intermediate survival outcomes were observed when one factor exhibited high expression and the other exhibited low expression (Fig. 2I). These findings indicate that p23 positively regulates GPX4 protein levels and inhibits ferroptosis. Consequently, it is hypothesized that p23-mediated regulation of ferroptosis via GPX4 is one of the critical factors influencing NSCLC tumor growth.

Fig. 2.

p23 deficiency induced GPX4 downregulation and inhibited lung tumor development. A Silver staining showing the differentially expressed proteins in NC or sip23 A549 cells. B Venn diagram showing the overlapping proteins linked to ferroptosis between the MS results and the database search. C WB analysis of GPX4, CISD2, and MGST1 expression in NC or sip23 A549 and H1299 cells. D WB analysis of p23 and GPX4 expression in vector or OEp23 A549 and H1299 cells. E WB analysis of p23 and GPX4 expression in NC or sip23 A549 and H1299 cells. F WB analysis of p23 and GPX4 expression in normal or NSCLC cells. G WB analysis of p23 and GPX4 expression in NSCLC clinical samples from tumor tissues and adjacent normal tissues; n = 3. H Representative graphs of p23 and GPX4 expression in the tissue microarray according to NSCLC tissues paired with adjacent normal tissues. I Survival analysis results of p23 and GPX4 expression levels in patients. J WB analysis of p23 and GPX4 expression in A549 cells stably expressing shp23 or/and OEGPX4. K The effects of p23 and GPX4 on cell proliferation were detected using the xCELLigence assay. L-N A549 cells stably expressing shp23 or/and OEGPX4 were subcutaneously injected into NTG mice (n = 6), and the subcutaneous tumor size (L), volume (M), and weight (N) were measured. O WB analysis of p23 and GPX4 expression in subcutaneous tumors from different groups. P Representative IHC images of p23 and GPX4 expression in the indicated subcutaneous tumors (left) and their quantitative analysis (right); n = 3. Q Laser confocal microscopy analysis of subcutaneous xenograft tumors generated with BODIPY™ 581/591 C11 (left) and quantitative analysis (right); n = 3. R-U The concentrations of GSSG (R), GSH (S), Fe2+ (T), and MDA (U) in the indicated subcutaneous tumors were measured using the appropriate assay kits; n = 3. Quantification graphs: The data represent the mean ± standard deviation from at least three independent experiments. Statistical analyses were performed using t tests (M, N) or one-way ANOVA (P-U). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; n.s., not significant

Table 1.

Positive correlation between p23 and GPX4 protein expression in lung adenocarcinoma specimens

		GPX4		P-Value	R-value
		Lowa	higha
P23	low	17	11	0.0009	0.3825
	high	10	34

^aIn the present study, ImageJ was used to analyze two tissue chips infected with p23 or GPX4. A positive area ratio exceeding 60% was considered indicative of high expression, whereas a ratio below 60% was categorized as low expression

To validate this hypothesis, we conducted a rescue experiment in which GPX4 was overexpressed in p23-knockdown cells to determine whether the oncogenic effects of p23 are mediated through GPX4 (Figs. 2J and S3I). Knockdown of p23 (shp23) significantly suppressed A549 cell growth, and this effect was partially reversed by GPX4 overexpression (OEGPX4) (Fig. 2K). Consistent with the in vitro results, OEGPX4 also partially reversed the shp23-induced inhibition of A549 xenograft tumor proliferation (Fig. 2L-N). Western blotting and IHC staining further confirmed the upregulation of GPX4 in OEGPX4 xenograft tumor tissues, as well as the reversal of GPX4 expression by shp23 treatment (Figs. 2O-P and S3J). In addition, a reduction in lipid peroxidation and ferroptotic events, including the consumption of GSSG and the generation of GSH, as well as the depletion of iron and MDA, was also observed in OEGPX4 xenograft tumor tissues. These effects were effectively reversed by shp23 (Figs. 2Q-U). Overall, these findings revealed that the regulation of ferroptosis by p23 is mediated mainly through GPX4.

To further demonstrate that p23 regulates NSCLC proliferation and ferroptosis via GPX4, we employed RSL3, a specific inhibitor of GPX4. As expected, p23 knockdown (sip23) increased lipid ROS accumulation, although this effect was less pronounced than that induced by saturated RSL3 treatment. Notably, combining sip23 with RSL3 did not further increase lipid ROS levels beyond those achieved by saturating RSL3 alone, suggesting that both perturbations converge on the same pathway (Figure S3K). This pattern was corroborated by cell viability assays (Figure S3L) and MDA measurements (Figure S3M), which revealed analogous trends. Conversely, p23 overexpression (OEp23) attenuated the expression of ferroptosis markers, yielding effects that were diametrically opposed to those of sip23 (Figures S3N‒P). These data strongly demonstrate that p23 governs ferroptosis in NSCLC primarily through GPX4-dependent mechanisms. Moreover, as selenite can upregulate GPX4 expression [40, 41], we further determined whether selenite supplementation could restore GPX4 expression levels under conditions of p23 knockdown. Our findings revealed that selenite mildly increased GPX4 protein levels. Moreover, cell viability exhibited a parallel trend of change (Figures S3Q and S3R). These results provide evidence that p23 primarily governs ferroptosis in NSCLC via a GPX4-dependent regulatory pathway.

p23 knockdown mediates GPX4 downregulation by activating the CMA pathway

Given the universal role of p23 in regulating ferroptosis in lung cancer cells through GPX4, we selected A549 sip23 cells as the representative knockdown model and H1299 OEp23 cells as the representative overexpression model. These models were used to further explore the specific mechanism by which p23 regulates GPX4 protein levels. We initially assessed the mRNA levels of GPX4 following the overexpression or knockdown of PTGES3 (p23). The results revealed that GPX4 mRNA levels were unaltered upon PTGES3 (p23) overexpression or knockdown (Figures S4A and S4B), suggesting that p23 likely regulated GPX4 through posttranslational mechanisms—such as modulating protein stability—rather than via transcriptional control. In addition, as a selenoprotein, GPX4 expression levels are also regulated by selenium intake and the associated translational machinery [40, 41]. To eliminate the potential influence of this factor on our experimental outcomes, we further examined the effects of p23 on the xCT system (SLC7A11 and SLC3A2). This system facilitates selenium uptake, thereby promoting selenocysteine synthesis and subsequently enhancing GPX4 translation. As shown in Figure S4C, neither p23 knockdown nor overexpression had a significant effect on the expression of the xCT system components SLC7A11 and SLC3A2. These findings indicate that the upregulation of GPX4 protein levels by p23 is mediated mainly through the regulation of its protein stability. A protein stability assay revealed that sip23 treatment significantly shortened the half-life of GPX4 (Fig. 3A). To further investigate the pathway through which p23 affects GPX4 protein stability, we utilized two common inhibitors: CQ (a lysosomal inhibitor) and MG132 (a proteasome inhibitor). The results demonstrated that CQ was more efficient at preventing GPX4 degradation, suggesting a potential association between p23-mediated GPX4 degradation and the lysosomal pathway (Fig. 3B). This finding was further supported by immunofluorescence (IF) analysis, which revealed that sip23 induced the colocalization of GPX4 and lysosomes (Fig. 3C). Furthermore, AR7, a CMA activator (Figures S4D-F), effectively increased the degradation of GPX4 induced by sip23 (Figs. 3D and S4G). In addition, we used siRNA to interfere with LAMP2A to specifically inhibit the CMA pathway and found that in the CMA inhibition group (siLAMP2A), the degradation of GPX4 induced by p23 knockdown (sip23) was significantly inhibited (Figs. 3E and S4H). These results indicate that p23 affects GPX4 protein degradation through the CMA pathway in lysosomes. Moreover, the p23-knockdown group presented an increase in the number of autophagic lysosomes and chamber expansion (Figure S4I, red arrow).

Fig. 3.

p23 knockdown mediates GPX4 downregulation by activating the CMA pathway. A WB analysis of the effect of p23 on GPX4 protein levels in NC or sip23 A549 cells treated with CHX (10 µM) at the indicated time points; n = 3. B WB analysis of the effect of p23 on GPX4 protein levels in sip23 A549 cells treated with CHX (10 µM), CQ (20 µM), or MG132 (10 mg/mL) at the indicated time points; n = 3. C NC or sip23 A549 cells were treated with CQ (20 µM, 36 h), and the colocalization of GPX4 and lysosomes was detected using a confocal laser. The graph on the right presents the results of the colocalization analysis, where the overlap coefficient “Person r” between the GPX4 (green) labeling and lysosomal (red) labeling was calculated using the ImageJ software. D WB analysis of p23 and GPX4 expression in NC or sip23 A549 cells treated with or without AR-7 (20 µM, 36 h). E WB analysis of LAMP2A, GPX4, and p23 expression in A549 cells transfected with or without p23 or LAMP2A siRNA. F NC or sip23 A549 cells were photoactivated after transfection with the pSIN-PAmCherry-KFERQ-NE plasmid. Images (light) and quantification (right) of pSIN-PAmCherry-KFERQ puncta are shown; n = 3. G WB analysis of p23, GAPDH, GPX4, MEF2D, LAMP2A, and LRRK2 expression in NC or sip23 A549 and H1299 cells. H WB analysis of p23, GAPDH, GPX4, MEF2D, LAMP2A, and LRRK2 expression in Vector or OEp23 H1299 cells. I Native gel electrophoresis and WB analysis of LAMP2A in lysosomes purified from Vector or OEp23 H1299 cells. J WB analysis of LAMP2A, MEF2D, and GPX4 expression in lysosomes purified from Vector or OEp23 H1299 cells compared with LAMP1 levels. K WB analysis of p23, GAPDH, GPX4, MEF2D, LAMP2A, and LRRK2 expression in the indicated subcutaneous tumors. L Representative IHC images of MEF2D and LRRK2 expression in the indicated subcutaneous tumors. The quantified image is shown on the right; n = 3. M Native gel electrophoresis and WB analysis of LAMP2A in lysosomes purified from the indicated subcutaneous tumors. Quantification graphs: The data represent the mean ± standard deviation from at least three independent experiments. Statistical analyses were performed using t tests (F, L) or two-way ANOVA (A, B). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; n.s., not significant

To further investigate the regulatory role of p23 in the CMA pathway, we constructed the PAmCherry-KFERQ vector, a prototypical substrate protein carrier used in the CMA pathway. In A549 cells stably transfected with PAmCherry-KFERQ, a substrate protein was engineered to emit red fluorescence upon exposure to 405 nm light. The transition of red fluorescence from diffuse (cytosolic) to punctate (lysosomal) indicates an increase in chaperone-mediated autophagy (CMA) activity. As shown in Fig. 3F, the sip23 group presented a notable increase in red fluorescent puncta, indicating that sip23 strongly activated the CMA pathway. This observation underscores the universal role of p23 in CMA pathway regulation. Additionally, sip23 decreased the protein levels of CMA substrates (such as LRRK2, MEF2D, and GAPDH), whereas OEp23 upregulated these protein levels (Figs. 3G-H and S4J-K) without affecting their mRNA levels (Figure S4L). Furthermore, in A549 cells, OEp23 was observed to reduce LAMP2A multimerization and its protein levels in the lysosomal fraction, highlighting its role in regulating the rate-limiting step of the CMA pathway [42], without altering mRNA levels (Figs. 3I-J and S4M-O). Similarly, the downregulation of CMA substrates and the decrease in LAMP2A multimerization were enhanced in shp23 xenograft tumor tissues (Figs. 3K-M and S4P-Q). Overall, the results revealed that p23 knockdown activated the CMA pathway.

The interaction of p23 with HSC70 is required for CMA activation

As p23 is known to act as a cochaperone to maintain the stability of other proteins, we next investigated whether p23 directly interacts with GPX4. Co-IP and laser confocal microscopy revealed no significant interaction between p23 and GPX4 (Figs. 4A and S5A), suggesting the involvement of an additional protein in p23-mediated GPX4 stability through the CMA pathway. By combining Co-IP/MS and bioinformatics analysis, we determined that HSC70 interacts with p23 (Fig. 4B). Co-IP assays also confirmed that p23 interacts with HSC70 in both cell and mouse tumor tissue lysates (Fig. 4C and D). These results were supported by p23 and HSC70 colocalization in A549 cells (Fig. 4E). We showed that PTGES3 (p23) does not affect HSPA8 (HSC70) protein (Figs. 4F and S5B) or mRNA (Fig. 4G) levels. Previous studies have established that GPX4 harbors three functional KFERQ-like pentapeptide motifs (¹²⁴NVKFD¹²⁸, ¹⁶⁹LIDKN¹⁷³, and ¹⁸⁷QVIEK¹⁹¹) that serve as recognition sites for HSC70. Importantly, mutagenesis studies have confirmed that these specific sequences are essential for the HSC70-GPX4 interaction, as their disruption completely abolishes binding capacity [39]. The co-IP results also revealed that HSC70 can interact with both GPX4 and p23, and p23 overexpression decreased the interaction between HSC70 and GPX4 (Figs. 4H and S5C). Furthermore, GPX4 overexpression attenuated the binding between HSC70 and p23, whereas GPX4 knockdown enhanced this interaction (Figures S5D and S5E), supporting the notion that p23 competes with GPX4 for binding to HSC70. Additionally, our data demonstrated that neither GPX4 knockdown nor overexpression significantly altered p23 protein stability (Figures S5F and S5G). Moreover, the impact of p23 on GPX4 stability was largely prevented by HSC70 knockdown in A549 cells (Figs. 4I and S5H). We found that HSC70 overexpression reduced GPX4 protein levels, whereas HSC70 knockdown moderately elevated GPX4 expression (Figure S5I). Notably, even in the absence of p23, HSC70 maintained its interaction with GPX4. Furthermore, reexpression of p23 attenuated the binding between HSC70 and GPX4 (Figure S5J). These findings suggested that HSC70 possessed an intrinsic capacity to regulate GPX4 stability, with p23 serving as a modulatory factor in this regulatory mechanism. In addition, we detected changes in ferroptosis-related indicators after HSC70 knockdown (siHSC70) and found that the ferroptosis effect induced by sip23 was significantly inhibited (Figures S5K-O), suggesting that HSC70 plays a crucial role in the regulatory effect of p23 on GPX4 stability.

Fig. 4.

The interaction of p23 with HSC70 restrains CMA activation. A Co-IP analysis of the direct interaction between p23 and GPX4. B Schematic illustrating the detection of CMA-associated proteins that interact with p23. C and D Co-IP analysis of the direct interaction between p23 and HSC70 in HEK293T cells (C) and mouse subcutaneous tumors (D). E Representative IF images showing p23 and HSC70 colocalization. F WB analysis of p23 and HSC70 expression in the indicated cells. G HSPA8 (HSC70) and PTGES3 (p23) mRNA expression was detected by qPCR in NC or sip23 A549 cells and Vector or OEp23 H1299 cells; n = 3. H Co-IP analysis of the influence of p23 on the interaction between GPX4 and HSC70. I WB analysis of HSC70, GPX4, and p23 expression in A549 cells transfected with or without p23 or HSC70 siRNA. J Molecular docking simulation of interaction sites between full-length p23 and HSC70 299–509. K WB analysis of HSC70 and p23 expression in anti-FLAG immunoprecipitates from OEp23 or the indicated mutant HEK293T cells. L Co-IP analysis of the influence of wild-type or mutant p23 on the interaction between GPX4 and HSC70. M WB analysis of the influence of wild-type or mutant p23 on HSC70, GPX4, and LAMP2A expression. N Native gel electrophoresis analysis of LAMP2A in lysosomes purified from Vector, OEp23, or All mut HEK293T cells. O WB analysis of GPX4 and LAMP2A expression in lysosomes purified from Vector, OEp23, or All mut HEK293T cells compared with LAMP1 levels. Quantification graphs: The data represent the mean ± standard deviation from at least three independent experiments. Statistical analyses were performed using t tests (G). ∗∗∗p < 0.001; n.s. not significant

To gain deeper insight into how p23 regulates HSC70 function, we explored the specific regions of p23 involved in its interaction with HSC70. We adopted computer molecular docking to simulate the interaction between p23 (PDB:1EJF) and HSC70 (PDB:4KBQ), as neither the nucleotide-binding domain (NBD) nor the C-terminal protein substrate-binding domain (SBD) of HSC70 interacts with full-length p23 (Figures S5P and S5Q). Molecular docking studies revealed that the Asp60A, Glu81A, His57A, Phe56A, Arg12A, and Arg93 residues significantly contributed to the binding of HSC70 (Fig. 4J). subsequent mutation and co-IP assays further confirmed that Glu81, Arg12, and Arg93 of p23 are crucial for the interaction with HSC70 (Figs. 4K and S5R). Additionally, all mutations of these amino acid sites in p23 significantly activated CMA and impaired GPX4 stability (Figs. 4L-O and S5S-V). Collectively, these findings suggest that p23 facilitates CMA by interacting with HSC70.

Given that p23 has been previously identified as a cochaperone of HSP90 [43, 44], we investigated whether p23-mediated inactivation of the CMA pathway is dependent on HSP90. However, mutation of the critical binding sites of p23 (W106A/D108A) and HSP90 did not affect the interaction between HSC70 and p23 (Figures S5W and S5X) or LAMP2A, MEF2D, and LRRK2 protein levels (Figure S5Y), suggesting that p23 regulates inactivation of the CMA pathway independently of HSP90.

p23 succinylation enhances its affinity for HSC70 and suppresses CMA activation

Protein posttranslational modifications (PTMs) play crucial roles in the transduction of biological signals, protein stabilization, and recognition of interactions between proteins [45, 46]. Our previous research demonstrated that succinylation significantly impacts p23 function [32]. Therefore, we investigated whether succinylation affects the p23-mediated CMA pathway. Treatment with succinic acid to increase p23 succinylation (Fig. 5A) resulted in increased protein levels associated with CMA markers (Figs. 5B and S6A). In addition, with the increase in p23 succinylation, there was a marked increase in the interaction between p23 and HSC70 (Fig. 5C), whereas the interactions between HSC70 and GPX4, as well as between HSC70 and CMA substrates (such as LRRK2, MEF2D, and GAPDH), were notably reduced. (Figs. 5D and S6B). Accordingly, the half-lives of the GPX4 and CMA substrates were significantly prolonged (Figs. 5E and S6C). As expected, LAMP2A multimerization was reduced (Figs. 5F and S6D), indicating that the increase in p23 succinylation resulted in CMA inhibition. Moreover, p23 succinylation significantly inhibited ferroptotic events, including lipid ROS and ROS reduction, GSSG depletion, GSH production, and ferrous ion and MDA reduction (Figs. 5G-L). To illustrate the significance of p23 succinylation in mediating the CMA pathway, we evaluated the p23 plasmid with mutations at the succinylation site at K7, K33, and K79. However, our experimental data demonstrate that the K79 residue plays an essential structural role in maintaining p23 protein stability. As evidenced in Figure S6E, all K79 mutants (including K79R, K7R/K79R, K33R/K79R, and K7R/K33R/K79R) exhibited significantly reduced protein expression levels compared to wild-type p23 or other mutant variants. To ensure our findings specifically reflect succinylation-mediated regulation rather than confounding effects of protein instability, we focused on the K7 and K33 mutants in assessing p23’s role in modulating ferroptosis and CMA pathways. Compared with wild-type p23, the inhibition of p23 succinylation (K7/K33R) led to a decrease in the protein levels of CMA substrates (Fig. 5M and N and S6F), a weakened interaction between p23 and HSC70 (Fig. 5O), and increased binding of CMA substrate proteins to HSC70 (Figs. 5P and S6G). Furthermore, we conducted co-IP experiments using His-tagged p23 protein (1) and His-tagged succinylated p23 protein (2). The results demonstrated that only succinylated p23 could bind to HSC70 (Fig. 5Q) and that succinylated p23 reduced the binding of HSC70 to CMA substrate proteins, including GPX4 (Figs. 5R and S6H). These findings collectively support the conclusion that the alteration of p23 succinylation levels is a critical factor in its mediation of the CMA process. In addition, the inhibition of p23 succinylation impaired the interaction between p23 and HSC70, promoted the CMA pathway, and ultimately enhanced GPX4 degradation (Figs. 5S-V and S6I-L). We also examined the role of succinylation-defective p23 in ferroptosis. Compared with wild-type p23, the succinylation-defective mutant lost the ability to inhibit ferroptosis (Figures S6M-Q). These results further illustrate the functional significance of p23 succinylation in regulating ferroptosis. These findings collectively indicate that p23 succinylation suppresses CMA pathway activation by enhancing its interaction with HSC70, thereby hindering ferroptosis.

Fig. 5.

p23 succinylation enhances its affinity for HSC70 and suppresses CMA activation. A and B WB analysis of p23 as well as the succinylation levels of p23 (A), GAPDH, MEF2D, LRRK2, GPX4, and HSC70 (B) in A549 and H1299 cells treated with or without succinic acid (200 µM, 24 h). C and D A549 and H1299 cells treated with or without succinic acid (200 µM, 24 h). Co-IP analysis of the influence of p23 succinylation on its interaction with HSC70 (C) and the interaction between HSC70 and target proteins (GAPDH, GPX4, MEF2D, and LRRK2) (D). E A549 cells with or without succinic acid treatment (200 µM, 24 h) were treated with CHX (10 µM) for the indicated times, and GPX4 and MEF2D expression was detected using WB analysis; n = 3. F Native gel electrophoresis analysis of LAMP2A in lysosomes purified from A549 cells treated with or without succinic acid (200 µM, 24 h). G and H A549 and H1299 cells treated with or without succinic acid (200 µM, 24 h) were exposed to erastin (20 µM, 36 h). Lipid ROS (E) and total ROS (F) levels were measured with flow cytometry using BODIPY™ 581/591 C11 and DCFH-DA probes, respectively. I-L After treatment with succinic acid (200 µM, 24 h), the concentrations of GSSG (I), GSH (J), Fe2+ (K), and MDA (L) were measured in A549 and H1299 cells using the appropriate assay kits; n = 3. M and N WB analysis of p23 as well as the succinylation levels of p23 (M), GAPDH, MEF2D, LRRK2, GPX4, and HSC70 (N) in OEp23 or K7/K33R HEK293T cells treated with or without succinic acid (200 µM, 24 h). O and P Co-IP analysis of the influence of p23 succinylation on its interaction with HSC70 (O) and HSC70 combined with target proteins (GAPDH, GPX4, MEF2D, and LRRK2) (P) in OEp23 or K7/K33R HEK293T cells treated with or without succinic acid (200 µM, 24 h). Q and R Co-IP experiments were performed using His-tagged p23 pure protein (1) and His-tagged succinylated p23 pure protein (2). The ability of isolated succinylated p23 to bind HSC70 (Q) and the ability of HSC70 to bind CMA substrate proteins (R) were measured. S Co-IP analysis of the interaction between HSC70 and succinylated p23 or total p23. T WB analysis of the impact of p23 succinylation site mutation on HSC70, GPX4, and LAMP2A protein levels. U Native gel electrophoresis analysis of the impact of p23 succinylation site mutation on LAMP2A in lysosomes. V WB analysis of the impact of p23 succinylation site mutation on GPX4 and LAMP2A expression in lysosomes compared with LAMP1 levels. Quantification graphs: The data represent the mean ± standard deviation from at least three independent experiments. Statistical analyses included t tests (I‒L) or two-way ANOVA (E). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; n.s., not significant

Discussion

P23, a known HSP90 cochaperone, performs several essential functions independent of HSP90, including transcriptional activity [32]. In this study, we revealed a novel role of p23 in suppressing ferroptosis. First, we found that p23 knockdown significantly promoted ferroptosis-related events both in vitro and in vivo, thereby exerting a suppressive effect on tumor growth. Second, using protein MS identification and bioinformatics analysis, we observed that p23 knockdown significantly decreased the protein level of GPX4, a critical protein involved in ferroptosis. Moreover, our co-IP experiments revealed that p23 does not directly bind to GPX4, indicating that the effect of p23 on GPX4 expression and tumor growth is indirect. Further molecular biological results suggest that p23 succinylation enhances its interaction with HSC70, thereby reducing GPX4 degradation via the CMA pathway. Consequently, this impedes ferroptosis in NSCLC. This study identified p23 as a novel molecular target that specifically regulates ferroptosis in tumors, offering new insights into its role in the intricate crosstalk between CMA and ferroptosis. Although previous studies have reported a direct interaction between p23 and GPX4 [33], these findings do not contradict our findings. The two studies differ significantly in terms of the disease context (cancer vs. cerebral ischemia/reperfusion injury), tissue type (NSCLC vs. brain microvascular endothelial cells), and proposed mechanism (degradation via the CMA pathway through complex formation with HSC70 vs. ubiquitin-mediated degradation via complex formation with HSP90). Therefore, these two studies highlighted distinct aspects of p23 function and should be viewed as complementary rather than contradictory.

Given that Fe²⁺ is one of the key markers of ferroptosis, we also measured its changes in our experimental system. Notably, modulation of GPX4 protein levels led to corresponding changes in ferrous iron levels (Figure S7A). This finding is consistent with those of previous studies [47, 48], which demonstrated that GPX4 degradation intensifies ferroptosis and concurrently triggers an increase in Fe²⁺ levels. Nevertheless, the precise mechanism remains unclear. We hypothesize that p23 knockdown-induced GPX4 degradation results in the accumulation of lipid peroxides, thereby activating the Fenton reaction (Fe²⁺-catalyzed conversion of H₂O₂ to ·OH) and establishing a self-amplifying oxidative cycle [49]. Additionally, GPX4 deficiency may deplete glutathione (GSH), inhibit the cystine/glutamate antiporter SLC7A11/xCT, impair cystine uptake, and ultimately indirectly increase Fe²⁺ bioavailability [50]. Overall, alterations in GPX4 levels may indirectly affect iron homeostasis, but further experimental validation is needed. To further investigate the relationships among p23, iron homeostasis, and ferroptosis, we examined whether fluctuations in free iron influence p23 expression and ferroptotic responses. Our results revealed that ferrous ions did not alter p23 protein levels but significantly exacerbated the sip23-induced increase in lipid ROS and reduction in cell viability (Figures S7B-D). Furthermore, treatment with the iron chelator deferoxamine (DFO) effectively rescued the elevated lipid ROS levels and decreased the viability of the cells mediated by sip23 (Figures S7E and S7F). We hypothesize that sip23 impairs the clearance of lipid peroxides, whereas supplementation with free iron accelerates lipid peroxide generation. This combined effect leads to rapid lipid peroxide accumulation and robust activation of ferroptosis. By chelating free iron and inhibiting the Fenton reaction, DFO partially reverses the ferroptotic phenotype observed in p23-deficient cells. These findings suggest that although there may not be a direct regulatory link between p23 and iron, an indirect relationship likely contributes to the ferroptotic response.

As a key inhibitor of ferroptosis, GPX4 mitigates ferroptosis in human cancers by converting membrane lipid peroxides into nontoxic lipid alcohols more efficiently. The consumption of GPX4 leads to lipid peroxidation, resulting in ferroptosis in cells or tissues [51, 52]. Compared with that in adjacent tissues, the GPX4 level in lung cancer tissues was significantly elevated, and it was positively correlated with the degree of tumor malignancy in our study (Figure S7G and Table 2). These findings elucidate the regulatory mechanism of GPX4 in tumors and offer valuable insights for tumor treatment and prevention. The current investigation illustrated, for the first time, that p23 inhibits CMA pathway activation, thereby maintaining GPX4 protein stability and preventing ferroptosis in NSCLC.

Table 2.

Positive correlation between tumor pathological grade and GPX4 protein expression in lung adenocarcinoma specimens

	GPX4		P-Value	R-value
	low	high
Stage I	2	0	0.0158	0.9439
Stage I-II	4	1
Stage II	15	27
Stage II-III	6	8
Stage III	2	9

CMA modulates the onset and progression of tumors under both physiological and pathological conditions [38]. HSC70 is implicated in the recognition of cytoplasmic proteins bearing the KFERQ-like motif, followed by their targeting to lysosomes, thereby playing a crucial role in regulating CMA [53, 54]. CMA activation is influenced by HSC70 protein levels and the interactions between HSC70 and other functional proteins. Research has suggested that TPD52 activates CMA by interacting with HSC70 [55]. In the present study, p23 did not affect the expression level of HSC70; however, it inhibited CMA pathway activity and prevented the degradation of substrates, such as GPX4, by binding to HSC70. Indeed, in addition to GPX4, other CMA substrate proteins were also regulated by p23 in NSCLC, indicating the universal role of p23 in regulating the CMA pathway (Figs. 3G-L). Interestingly, in normal lung cells, p23 also interacts with HSC70 without affecting the degradation of CMA substrate proteins (Figures S7H and S7I). We hypothesize that the function of p23 in tumor cells differs from that in normal cells, and further investigation is needed to elucidate the underlying mechanism involved. Therefore, the impact of p23 on the interaction between HSC70 and other functional proteins is the primary factor contributing to its ability to inhibit CMA. In the present study, we demonstrated that p23 succinylation enhanced its interaction with HSC70, thereby impeding the binding of CMA substrates to HSC70 and hindering CMA pathway activation. As it is widely recognized that HSC70 plays a role in microautophagy, the impact of microautophagy on this system is noteworthy. However, our investigation into the knockdown of VPS4, a key factor in microautophagy, revealed that it did not restore the reduced GPX4 protein levels caused by sip23 (Figure S7J). This finding suggests that the occurrence of this system remains unaffected even when microautophagy is blocked.

PTMs are widely utilized posttranslational processes that play critical roles in regulating protein function. These modifications include the regulation of target protein localization, stability, activity, and physical interaction [45, 56, 57]. Numerous studies have demonstrated a close association between PTMs and ferroptosis. For example, Nedd4 ubiquitylation of VDAC2/3 inhibits ferroptosis in melanoma [58], SOCS2-enhanced ubiquitination of SLC7A11 can facilitate ferroptosis [59], and IL-1β enhances lysine (K) 1042 (NNT K1042ac) acetylation of nicotinamide nucleotide transhydrogenase (NNT) to protect tumor cells from ferroptosis [60]. In the present study, we demonstrated that p23 inhibits CMA and that this process is mediated by its succinylation. In addition, p23 physically interacts with HSC70, thereby inhibiting HSC70 binding to CMA substrate proteins (including GPX4) and consequently impeding the CMA pathway. This process is enhanced by p23 succinylation.

Conclusions

In summary, our study revealed the molecular mechanism by which p23 suppresses ferroptosis in NSCLC. The succinylation of p23 strengthened its interaction with HSC70, weakened the binding between GPX4 and HSC70, and prevented GPX4 degradation through the CMA pathway, ultimately inhibiting ferroptosis in NSCLC. Our study revealed the crucial roles of p23 in regulating both the CMA pathway and ferroptosis, expanding the identification of molecular targets for ferroptosis research in tumors. Additionally, these findings offer theoretical support for the potential development of p23 as a therapeutic target in cancer treatment.

Abbreviations

NSCLC

Non-small cell lung cancer

GPX4

Glutathione Peroxidase 4

p23

Prostaglandin E synthase 3

HSC70

Heat shock cognate 71 kDa protein

CMA

Chaperone-mediated autophagy

LAMP2A

Lysosomal associated membrane protein 2 A. SGK2:serum/glucocorticoid-regulated kinase 2

HSPA5

Heat shock 70-kDa protein 5

PDAC

Pancreatic ductal adenocarcinoma

Authors’ contributions

Z.L., G.L., Z.Y. and X.M. conceived and designed the study. J.C., Y.P., M.Z. and Y.C. contributed to performing the experiments and developing methodology. J.C., Y.P., Z.L., G.L., Z.Y. and X.M. contributed to the writing, reviewing, and revision of the paper. Y.C., S.Z., C.H., and W.Z. contributed to the clinical sample collection. M.Z., X.T, WH.Z. and X.H. provided acquisition, analysis, and interpretation of data and statistical analysis. MH.Z. and Y.W. provided instrumentation and material support. All authors read and approved the final version of the manuscript.

Funding

The authors thank National Natural Science Foundation of China (82004089 and 82103667), Liaoning Province science and technology plan joint project fund (2023JH2/101700082), Natural Science Foundation Program of Liaoning Province (2024-MS-040), “1 + X” Research Project of the Second Hospital of Dalian Medical University (LYYH2024002) and Liaoning Province Natural hospital cultivation of the Second hospital of Dalian Medical University (YJ20250034).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

All animal experiments were approved by the Animal Experimental Ethical Inspection Association of Dalian Medical University (AEE19048). Clinical sample collections were approved by Ethnics Committee of the First Affiliated Hospital of Dalian Medical University (PJ-KS-KY-2023-362). Tissue microarrays were purchased from Shanghai Outdo Biotech Company, and approved by Ethnics Committee of Shanghai Outdo Biotech Company (SHYJS-CP-1904014, SHYJS-CP-1701003).

Consent for publication

The manuscript is approved by all authors for publication.

Competing interests

The authors declare no competing interests.