Acidic bile salts induce APE1-dependent PRDX2 activation to drive oxaliplatin resistance via ferroptosis inhibition

Abstract

Background

The incidence of esophageal adenocarcinoma (EAC) is rapidly increasing in Western countries. Gastroesophageal reflux, containing acidic bile salts (ABS), is the main risk factor for EAC. Cancer cells develop adaptive abilities to recalibrate redox balance via hijacking the antioxidant systems to maintain reactive oxygen species (ROS) below lethal levels. We investigated the role of PRDX2 and its regulation in EAC chemoresistance under reflux conditions.

Methods

We analyzed public databases to identify PRDX2's aberrant overexpression and potential role in chemoresistance in EAC. To model acute and chronic GERD in vitro, we applied transient and repeated ABS exposures. Using 2D and 3D organotypic culture models of EAC cells, we investigated PRDX2's anti-ferroptosis, pro-chemoresistance functions, and regulatory mechanisms. In addition, we also utilized patient derived organoids and xenografts, cell line-derived xenografts, and human EAC tissue microarrays.

Results

Aberrant expression of PRDX2 was detected in both human EAC tissues and cell lines under ABS exposure. PRDX2 was regulated via an APE1-redox-dependent transcription activation of NF-kB. The knockdown of PRDX2 impaired the recovery of EAC cells from ABS-induced ROS and ROS-dependent lipid peroxidation. Silencing PRDX2 sensitized the chemo-resistant EAC cells to oxaliplatin by enhancing ferroptosis. Mechanistically, we found that PRDX2 inhibits ferroptosis by stabilizing GPX4, a crucial ferroptosis suppressor. PRDX2 enhances GPX4 stability through OTUB1-dependent deubiquitination, preventing its degradation. The APE1-redox inhibitor APX2009 significantly sensitized EAC cells to oxaliplatin by downregulating PRDX2 and inducing ferroptosis. Importantly, the combination of oxaliplatin and APX2009 showed synergistic tumor-suppressive effects in xenograft models.

Conclusions

Our findings revealed a novel link between reflux-induced redox rebalance and ferroptosis-related chemoresistance in EAC via APE1-redox/NF-kB/PRDX2/OUTB1/GPX4 signaling cascade. Targeting redox with APE1-redox-specific inhibitors is a potential novel strategy for drug combination in refractory EAC, via inhibition of PRDX2.

Graphical abstract

Highlights

• •Exposure to acidic bile salts, a mimic of gastroesophageal reflux disease, induces PRDX2 expression to protect esophageal cells from ferroptotic cell death.
• •PRDX2 modulates ferroptosis through OTUB1-dependent protein stabilization of GPX4.
• •Acidic bile salts-induced PRDX2 promotes chemoresistance by inhibiting ferroptosis.
• •APE1 redox/NF-κB axis mediates transcriptional regulation of PRDX2 induction.
• •Pharmacological inhibition of APE1 redox function overcomes oxaliplatin resistance.

1. Introduction

Esophageal cancer is the eighth most common cancer and the sixth leading cause of cancer-related mortality globally [1]. In the U.S., there are an estimated 22,370 new cases and 16,130 deaths in 2024 [2]. Esophageal adenocarcinoma (EAC) is the predominant subtype of esophageal cancer in North America, Europe and Oceania with a rapidly rising incidence [3,4]. Chronic gastroesophageal reflux disease (GERD) is characterized by prolonged exposure of the lower esophagus to acidic gastric contents and bile salts. It is recognized as the primary risk factor for metaplastic Barrett's esophagus (BE) and a driving factor for the progression to EAC [5,6]. Unfortunately, the overall 5-year survival rate remains only 22%, plummeting to 5% in patients with distant metastasis [2]. One of the primary challenges in treating EAC is the resistance to chemotherapy and several targeted therapy regimens, calling for the development of new therapeutics, particularly for refractory EAC patients.

Peroxiredoxins (Prxs or PRDXs) belong to a superfamily of small non-seleno peroxidases [7]. They are distinguished members of the antioxidant defense system, hydrolyzing H₂O₂ through thiol-dependent oxidation, thus balancing intracellular reactive oxygen species (ROS) and protecting cells from ROS-mediated oxidative stress [8,9]. Based on the position and number of conserved Cys, six isoforms of mammalian Prxs are classified into three types: 2-Cys Prxs (Prx1-4), atypical 2-Cys Prx (Prx5), and 1-Cys Prx (Prx6) [10]. Among them, Prx2 (or PRDX2) is the most abundantly expressed isoform with dominant peroxidase activity [11] and the highest sensitivity to hyperoxidation [12,13]. More importantly, PRDX2 can regulate redox-dependent signaling via protein interaction with various partners, playing critical roles in cell proliferation, apoptosis, and tumorigenesis [[14], [15], [16]]. Aberrant upregulation of PRDX2 has been reported in several cancer types, including gastric [17], colorectal [18], pancreatic [19], and breast cancer [20]. This dysregulation contributes to tumor progression, lymph node metastasis, radio-resistance and chemoresistance. However, the role of PRDX2 in EAC tumorigenesis remains largely unexplored.

Ferroptosis refers to an iron-dependent form of programmed cell death mediated by intracellular lethal accumulation of lipid peroxidation [21]. It is molecularly and mechanistically distinct from other types of cell death, such as apoptosis, necrosis, and autophagy [21,22].

Ferroptosis involves an imbalance between ferroptosis metabolites, mainly peroxidation of polyunsaturated fatty acid (PUFA)- containing phospholipids (PLs), and the buffering capability of ferroptosis defense systems [23]. Central regulators of ferroptosis include System Xc−, a membrane-located cysteine/glutamate antiporter with a core subunit named SLC7A11 (xCT),

and glutathione peroxidase 4 (GPX4). SLC7A11 imports extracellular cystine, which is then converted into glutathione (GSH) as a cofactor of GPX4. GPX4 is capable of detoxifying excessive PL hydroperoxides by utilizing GSH, thus suppressing ferroptosis [[24], [25], [26]]. Inhibition of SLC7A11 and GPX4, whether in genetic or pharmacological ways, leads to uncontrolled burden of lipid peroxidation and initiates ferroptosis [27]. Recent lines of evidence have revealed the tumor-suppressive effects of ferroptosis, while its inactivation facilitates tumor development and progression [[28], [29], [30], [31]]. Recent studies highlight the role of ferroptosis in enhancing response to cancer treatment with unexpected sensitivity of therapy-refractory cancer cells to ferroptosis [32,33]. For example, pharmacological inhibition of SLC7A11 by Erastin, a class I FIN (ferroptosis inducer), sensitized resistant head and neck cancer cell lines to cisplatin in vitro and in vivo by boosting ferroptotic cell death [34]. Unfortunately, the in vivo or clinical usage of most FINs has been constrained due to their poor metabolic stability, low solubility and suboptimal pharmacokinetics [32].

In this study, we demonstrate that upregulation of PRDX2 in response to reflux conditions, protects cancer cells from ferroptosis cell death and subsequently facilitates chemo-resistance through GPX4-mediated ferroptosis suppression. We also reveal that the induction of PRDX2 is regulated by ABS-induced APE1-redox-dependent NF-κB activation. Targeting APE1 redox function re-sensitizes resistant EAC cells to oxaliplatin treatment in vivo.

2. Methods

2.1. Bioinformatics analysis of public databases

We conducted analysis using publicly accessible databases from The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO). RNA expression profiles and related clinical data of EAC were obtained from the TCGA official website (https://portal.gdc.cancer.gov/repository), which includes 79 EAC samples and 9 normal esophageal tissue samples. Comprehensive analyses, including genome-wide gene expression profiling, Pearson correlation analysis, Gene Set Enrichment Analysis (GSEA), and Gene Set Variation Analysis (GSVA), were performed. Additionally, two GEO datasets, GSE1420 and GSE165252, were retrieved from the National Center for Biotechnology Information (NCBI) GEO database (https://www.ncbi.nlm.nih.gov/). The GSE1420 dataset [35] encompassed genome-wide gene expression profiling of surgical specimens from 8 EAC cases and 8 normal esophageal epithelial tissues. The GSE165252 dataset [36] is composed of biopsy/surgical samples from EAC patients before the start of treatment (n = 32), on-treatment (n = 29) and after transthoracic esophagectomy (n = 10). The samples were further divided into responders and non-responders based on patients' reactions to a combination therapy of neoadjuvant chemoradiotherapy and a PD-L1 inhibitor (atezolizumab). Genome-wide gene expression profiling and Pearson correlation analysis were applied to this dataset. The expression data were processed and analyzed using the R programming environment (version 4.3.3). Visualization of data distributions was achieved through the generation of violin plots utilizing ggplot2 package (version 3.5.1). Pearson correlation analysis was conducted employing the ‘Hmisc’ package (version 5.1-2) in R. Differentially expressed genes were identified using the Limma package (version 3.58.1), with a significance threshold set at a p-value <0.05 and a log2 fold-change (log2FC) cutoff defined as log2FC > mean(|log2FC|) + 2 × standard deviation (|log2FC|). Gene Set Enrichment Analysis and Gene Set Variation Analysis were performed by R packages ‘GSEABase’ (version 1.64.0) and ‘GSVA’ (version 1.50.1), respectively.

In addition, two online databases, BioGRID (https://thebiogrid.org/) and IntAct (https://www.ebi.ac.uk/intact/search?query=EBI-2510844) were used to screen protein interactions.

2.2. Human samples

A tissue microarray (TMA) containing esophageal adenocarcinoma and normal tissues was purchased from TissueArray.com (ES8011b). This TMA contains 35 cases of EAC and 5 cases of normal esophageal tissues in duplicates.

2.3. Cell culture and reagents

Human EAC cell lines included OE33, SKGT4, OE19, and FLO-1. The OE33 cell line was kindly provided by Dr. David Beer (University of Michigan, Ann Arbor, MI), while SKGT4 was a generous gift from Dr. Xiaochun Xu (MD Anderson, Houston, TX). OE19 and FLO-1 cell lines were commercially acquired from Sigma-Aldrich (St. Louis, MO, USA). In addition, the human esophageal fibroblasts (hEF), purchased from ScienCell Research Laboratories (Carlsbad, CA, USA), was used for some experiments. The OE19 and OE33 cells were cultured in RPMI-1640 medium (GIBCO, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS) (Invitrogen Life Technologies, Carlsbad, CA, USA) and 1% penicillin/streptomycin (GIBCO). The FLO-1 and SKGT4 cells were maintained in DMEM (GIBCO) supplemented with 10% FBS and 1% penicillin/streptomycin. All cell lines were routinely screened for Mycoplasma contamination using a Mycoplasma Detection Kit (PCR) from Southern Biotech (Birmingham, AL). Additionally, the cell lines were verified to retain their original in vitro morphological characteristics and were periodically authenticated via short tandem repeat (STR) profiling performed by Labcorp (Burlington, NC). Comprehensive details regarding the reagents and kits used in this study are provided in Supplementary Table S1, and the antibodies utilized are listed in Supplementary Table S2.

2.4. Acidic bile salts (ABS) exposure

We used an acidic bile salts (ABS) cocktail that replicates the composition of bile acids present in the distal esophagus under clinical reflux conditions [37]. The ABS cocktail consists of an equimolar mixture of deoxycholic acid, glycocholic acid, glycodeoxycholic acid, glycochenodeoxycholic acid, and taurocholic acid, dissolved in serum-free medium adjusted to pH 4.0. A final concentration of 200 μM (40 μM of each bile salt) was used to treat EAC cells for 20 min, followed by recovery in complete media, unless otherwise specified. For repeated ABS exposure experiments, cells were treated daily with 200 μM ABS at pH 5.5 for 20 min over a period of 14 days, as previously described [37,38].

2.5. Gene expression and silence

To establish stable cell lines, lentiviral particles expressing PRDX2 shRNA, APE1 shRNA, or control shRNA were produced by VectorBuilder Inc. (Santa Clara, CA, USA) and transduced into OE33 cells using standard protocols. The specific targeting sequences are provided in Supplementary Table S3. Stable cell populations were selected over a 10-day period using puromycin (1 μg/ml) in growth medium supplemented with 10% FBS.

Overexpression of APE1 and NF-kB-p65 followed previously established methodologies [37]. To be specific, the flag-tagged coding sequence of wild type APE1 and the redox-deficient mutant of APE1 (C65A) were cloned into the pcDNA3.1 mammalian expression vector (Invitrogen, Carlsbad, CA, USA). The redox-deficient mutant of APE1 (C65A) was generated using the QuickChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA). The coding sequence of NF-κB-p65 was cloned into the pCMV mammalian expression plasmid (Invitrogen). Additionally, the flag-tagged human PRDX2-pCMV plasmid (Vector ID: VB210709-1193scq) and OTUB1 wild-type (WT)-pcDNA3.1 plasmids (Catalog#: 118209) were obtained from VectorBuilder Inc. (Santa Clara, CA, USA) and Addgene (Watertown, MA, USA), respectively. PolyJet reagent (SignaGen Laboratories, Rockville, MD, USA) was applied for transient gene overexpression transfection.

For transient knockdown experiments, APE1, PRDX2, and OTUB1 siRNAs were transfected into EAC cells using LipoJet reagent (SignaGen Laboratories) according to the manufacturer's protocol. The sequences for control siRNA, APE1 siRNA, PRDX2 siRNA, and OTUB1 siRNA are provided in Supplementary Table S3.

2.6. CellTiter-Glo cell viability assay

Cell viability was measured using the CellTiter-Glo cell viability assay (Promega, Madison, WI, USA) in accordance with the manufacturer's protocol. To investigate the impact of transient ABS exposure on cells with PRDX2 depletion compared to control cells, OE33 and OE19 cells were plated at a density of 1000 cells per well in 96-well plates, followed by transient knockdown of PRDX2 using siRNA. ABS exposure began 24 h post-transfection (day 0). Cell viability was monitored on the fourth day (day 3).

To determine the half-maximal inhibitory concentration (IC50) of oxaliplatin under various conditions, cells were seeded at 1000 cells per well in 96-well plates and treated with oxaliplatin, Erastin, APX2009, a combination of oxaliplatin and Erastin and a combination of oxaliplatin and APX2009, or PBS (as a control) for five days. Luminescence measurements were conducted using the CLARIOstar® Plus multi-mode microplate reader (BMG Labtech, Ortenberg, Germany).

2.7. ROS and lipid peroxidation assay

Cells with PRDX2 knockdown or control cells were plated in triplicates in 12-well plates and cultured for 24 h before exposure to ABS. Following treatments, cells were maintained in complete media for recovery at designated time points. For intracellular ROS detection, cells were incubated with 4 μM H2DCFDA (Thermo Fisher Scientific) in fresh medium at room temperature for 30 min. For lipid peroxidation, cells were incubated with 5 μM BODIPY 581/591 C11 dye (Invitrogen) at 37°C for 45 min. After incubation, cells were harvested as a single-cell suspension by trypsinization, washed with PBS, and analyzed using a Beckman Coulter flow cytometer (Brea, CA, USA) for ROS and lipid peroxidation levels following the manufacturer's protocol.

2.8. RNA isolation and quantitative RT-PCR

Total RNA was extracted using TRIzol reagent (Invitrogen) and cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA), following the manufacturer's instructions. Quantitative RT-PCR was conducted using SYBR® Select Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) on a CFX Connect real-time system (Bio-Rad). Primers were obtained from Integrated DNA Technologies (IDT, Coralville, IA, USA). The threshold cycle (Ct) values were determined by Bio-Rad CFX Manager software (version 3.0) and averaged across triplicated reactions. Melting curve analysis was performed to assess the dissociation properties of double-stranded DNA templates during thermal denaturation. The sequences of the primers used are provided in Supplementary Table S4 mRNA expression levels for all samples were normalized to the housekeeping gene HPRT1 [39] [40], and the results were expressed as relative fold changes.

2.9. Immunoprecipitation and western blot analysis

Cells were lysed using RIPA buffer (Santa Cruz Biotechnology, Dallas, TX, USA) supplemented with 1× Halt protease inhibitor cocktail and 1× Halt phosphatase inhibitor cocktail (Thermo Fisher Scientific). The lysates were then sonicated thoroughly and centrifuged at 12,000 rpm for 15 min at 4°C. Protein concentrations were quantified using the Pierce BCA Protein Assay (Thermo Fisher Scientific). For immunoprecipitation assays, cell lysates were incubated with protein A/G beads (MilliporeSigma, Burlington, MA, USA) and 2 μg of specific antibody or control IgG overnight at 4°C under constant rotation. The immunocomplexes were subsequently washed eight times with PBST buffer (PBS (GIBCO) containing 0.1% Tween® 20 (Sigma-Aldrich)) and then eluted and analyzed by Western blot. For Western blot analysis, protein samples were prepared by mixing cell lysates with 4X sample buffer (Invitrogen) and denatured at 85°C for 10 min. Proteins were separated by electrophoresis on 10% SDS-PAGE gels and transferred onto 0.45 μm nitrocellulose membranes (Bio-Rad). The membranes were probed with the indicated primary antibodies followed by HRP-conjugated secondary antibodies. β-actin was used as a loading control. Finally, protein bands were visualized using Immobilon Western Chemiluminescent HRP Substrate detection reagents (MilliporeSigma and Thermo Fisher Scientific, Burlington, MA, USA). Images were acquired using the Bio-Rad ChemiDoc XRS + Imaging System. Band intensities were quantified with Image Lab software (version 6.0.1, Bio-Rad). The antibodies utilized are listed in Supplementary Table S2.

2.10. Chromatin immunoprecipitation (ChIP) assay

The chromatin immunoprecipitation (ChIP) assay was conducted using the ChIP-IT Express Enzymatic Kit (Active Motif, Carlsbad, CA, USA) referring to the manufacturer's instructions. Briefly, cells were cultured in 100 mm dishes until they reached 60–70% confluence. Following a 3-h recovery period following ABS treatment, cells were cross-linked with 1% formaldehyde (Sigma-Aldrich) and then washed with ice-cold PBS. The cross-linking reaction was quenched using Glycine Stop-Fix Solution, followed by an additional PBS wash. Cells were harvested using Cell Scraping Solution containing phenylmethanesulfonyl fluoride (PMSF) and pelleted by centrifugation. The cell pellets were resuspended in lysis buffer supplemented with protease inhibitor cocktail (PIC) and PMSF. The protein-chromatin complexes were further lysed using a Dounce homogenizer and enzymatically sheared using the Enzymatic Shearing Cocktail to generate chromatin fragments ranging from 200 bp to 1000 bp in size. The supernatants containing sheared chromatin were collected and incubated overnight at 4°C with an immunoprecipitation antibody targeting NF-κB-p65 and Protein G magnetic beads on an end-to-end rotator. The precipitated protein-DNA complexes were then washed, eluted, and de-crosslinked at 95°C for 15 min. Magnetic beads were removed using a magnetic stand, and the supernatants containing purified DNA were analyzed by quantitative RT-PCR to assess NF-κB binding to the PRDX2 promoter. Three pairs of primers spanning the predicted NF-κB-p65 binding sites in the PRDX2 promoter were designed for this analysis (Supplementary Table S4).

2.11. Three-dimensional organotypic culture

Three-dimensional (3D) organotypic cultures (OTC) of OE33 cells were established using a previously described protocol [41]. Human esophageal fibroblasts (hEF) were embedded into 3D matrixes containing collagen I (high-concentration rat-tail collagen, Corning, Incorporated, Corning, NY, USA) and Matrigel (BD Biosciences, Franklin Lakes, NJ, USA) at a density of 75,000 cells per well and cultured for 7 days. OE33 cells were then seeded onto the fibroblast-containing matrixes at a density of 500,000 cells per well and incubated for an additional 7 days. Prior to harvesting, the ABS cocktail (100 μM, 30 min) was applied to the OTC cultures on top of the inserted chambers, followed by a 6-h recovery period in complete medium. The samples then underwent fixation in 70% ethanol and paraffin embedding, followed by hematoxylin and eosin (H&E) staining, and immunofluorescence analysis. The OTC samples that were not subjected to any treatment served as the control group.

2.12. Human EAC organoid culture

Human EAC organoid cultures derived from de-identified patient-derived xenografts (PDXs) were established following a previously described protocol [42] [43]. In summary, PDX tissues were dissected into small fragments measuring 0.1–0.5 mm in diameter and rinsed with culture medium. The tissue fragments were then digested using 2.5 mg/mL Collagenase A in advanced DMEM-F12 (adDMEM/F12), followed by further digestion with TrypLE™. The resulting cell suspension was filtered through a 40-μm cell strainer to obtain single-cell suspensions. Subsequently, 20,000 cells washed with ice-cold PBS were resuspended in 50 μL of 90% Matrigel. The cell-Matrigel mixture was plated in the center of each well of a 24-well culture plate and inverted at 37°C to allow the Matrigel to solidify. Prewarmed IntestiCult™ Organoid Growth Medium (human) was then added to each well to support organoid growth. After a 14-day culture period, the organoids were randomly divided into four experimental groups and subjected to lentivirus transduction and treatments as follows: (1) control shRNA with PBS; (2) PRDX2 shRNA with PBS; (3) control shRNA with 5 μM oxaliplatin; and (4) PRDX2 shRNA with 5 μM oxaliplatin. Following 72 h of treatment, bright-field images of the organoids were captured using microscopes. The samples were then harvested and sent to the Cancer Modeling Shared Resource at the University of Miami for paraffin embedding and sectioning.

2.13. Immunofluorescence staining

Immunofluorescence (IF) staining was used to evaluate the subcellular localization and expression of PRDX2, following established protocols [37,43]. Briefly, OE33 and SKGT4 cells were fixed with freshly prepared 4% paraformaldehyde solution for 45 min at room temperature (RT) after recovery from ABS or control treatment. The cells were then permeabilized using PBS containing 0.25% Triton™ X-100 (Sigma-Aldrich). Primary antibodies targeting Ki-67, APE1, PRDX2, and GPX4 were applied overnight at 4°C in a humidified chamber. The slides were then incubated with Alexa Fluor conjugated secondary antibodies (Alexa Fluor 488 and/or 568) for 1.5 h at RT in the dark. Finally, the samples were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) and visualized using a Keyence BZ-X710 fluorescence microscope (Keyence Co., Osaka, Osaka, Japan).

For formalin fixed paraffin-embedded (FFPE) tissue sections, slides were deparaffinized and rehydrated as previously described [37]. Antigen retrieval was performed using Tris-EDTA Buffer, pH 9.0, 1X (Genemed Biotechnologies Inc, South San Francisco, CA, USA) for 20 min. The slides were then blocked with 10% non-immune goat serum (Thermo Fisher Scientific) for at least 30 min at RT. Primary antibodies targeting PRDX2, APE1, GPX4, and Ki-67 were applied overnight at 4°C in a humidified chamber. The slides were then incubated with Alexa Fluor conjugated secondary antibodies (Alexa Fluor 488 and/or 568) for 1 h at RT in the dark. Finally, the samples were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) and visualized using a Keyence BZ-X710 fluorescence microscope (Keyence Co.)

2.14. Immunohistochemistry

Slides from the FFPE TMA were subjected to standard deparaffinization and rehydration procedures, followed by antigen unmasking in sub-boiled Tris-EDTA Buffer, pH 9.0, 1X (Genemed Biotechnologies Inc) for 20 min. Immunohistochemistry (IHC) staining was performed using the IHC Select® Immunoperoxidase Secondary Detection system (Millipore Sigma) in accordance with the manufacturer's protocol. The slides were first incubated with Blocking Reagent for a minimum of 60 min RT, followed by primary antibodies targeting PRDX2 or a negative control reagent for 1 h at RT, Secondary Antibody for 30 min at RT, and Streptavidin HRP for 10 min at RT. After rinsing thoroughly, the slides were counterstained with Hematoxylin Counter Stain Solution for 10 min at RT and dehydrated using a graded series of ethanol and xylene (Thermo Fisher Scientific). The intensity and frequency of IHC staining were quantified using a scoring system as previously described [44]. Index scores, as outlined in prior studies [40], were used to statistically compare PRDX2 protein expression levels between esophageal adenocarcinoma (EAC) tissues and normal esophageal epithelium.

2.15. Cycloheximide (CHX) chase assay

The cycloheximide (CHX) chase assay was utilized to evaluate the intracellular protein degradation dynamics and quantify protein half-life by inhibiting de novo protein synthesis [45], as previously described [46]. In brief, parental OE33 cells and cells stably transfected with PRDX2-targeting or control shRNA were pretreated with 100 μg/ml CHX for 2 h prior to ABS exposure. Cells were recovered in complete medium supplemented with 100 μg/ml CHX. Whole-cell lysates were harvested at indicated time points post-recovery. GPX4 and β-actin protein levels were analyzed via Western blot, and band intensities were quantified using Image Lab software (Bio-Rad). The GPX4/β-actin intensity ratio was determined for each time point to evaluate protein degradation kinetics.

2.16. Ubiquitination assay

To determine the ubiquitination levels of GPX4 protein, cells were co-transfected with an HA-tagged ubiquitin plasmid using PolyJet transfection reagent, along with other specified experimental treatments. To inhibit proteasomal degradation, 10 μM MG132 (MilliporeSigma) was administered 6 h before harvesting. Cell lysates were collected and immunoprecipitated with an anti-GPX4 antibody. Subsequently, the ubiquitination levels of the target proteins were analyzed by Western blot analysis using an anti-HA antibody.

2.17. Animal experiments

All animal experiments were performed in compliance with the approved animal protocol (23-110) by the Institutional Animal Care and Use Committee (IACUC) of the University of Miami. The ARRIVE reporting guidelines were followed to ensure transparency and reproducibility in the animal studies [36]. Tumor volume was determined using the formula: tumor volume (mm³) = 1/2 × (major axis) × (minor axis)² [47]. Mice were humanely euthanized when tumor size reached 1000 mm³.

In this study, de-identified PDXs (ID: GTR0165) originating from the human gastro-esophageal junction were utilized, as previously described [48]. Samples were chopped into uniform sizes and subcutaneously implanted into the bilateral flanks of seven-week-old NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ immune deficient mice (NSG™, Jackson Laboratory, Bar Harbor, ME). The mice were randomly divided into two groups: an untreated control group (PBS, n = 8) and a treatment group receiving APE1-redox inhibitor (APX2009, n = 8) at a dose of 20 mg/kg daily through intraperitoneal injection.

To evaluate the effectiveness of the combination therapy, 2×10^6 OE19 cells, a cell line exhibiting high resistance to oxaliplatin, were suspended in a 1:1 mixture of Matrigel and phosphate-buffered saline (PBS) and subcutaneously implanted into the bilateral flanks of NOD/SCID mice of 6 weeks of age. The mice were then divided into four groups: untreated control (PBS) (n = 8); oxaliplatin (n = 8), 1 mg/kg, twice weekly; APE1-redox inhibitor (APX2009) (n = 8), 20 mg/kg, daily; and a combination of oxaliplatin and APX2009 (n = 8). For both experiments, tumor growth was monitored twice weekly, and treatments were initiated once the tumors reached approximately 150 mm³. Tumor tissues from two randomly selected mice per group were collected after 14 days of treatment. The remaining six mice per group continued treatment for 28 days, with body weight and tumor mass measured twice weekly. Survival was monitored until the end of the study period. Statistical differences in tumor sizes were determined with Student's t-test using Prism 10 (GraphPad Software, San Diego, CA, USA). Additionally, Kaplan–Meier survival analysis and log-rank tests were performed using Prism 10 to determine statistical significance in survival outcomes.

2.18. Statistical analysis

All statistical analyses were conducted using Prism 10. Data were quantified and presented as mean ± standard error of the mean (SEM). The unpaired Student's t-test was applied to evaluate differences between two independent experimental variables, and to determine statistical significance of Pearson correlation coefficients between two continuous variables. Wilcoxon test was applied for calculating statistical significance for two group comparisons in the public datasets. Extra sum-of-squares F-test was used to compare the two dose-response curves. Survival outcomes were evaluated via Kaplan-Meier survival curves, and statistical significance was determined using the log-rank test. A p-value <0.05 was considered statistically significant.

3. Results

3.1. PRDX2 is overexpressed and induced by ABS in human EAC

PRDX2 exhibits tissue-specific expression patterns in various human cancers, yet its role in esophageal adenocarcinoma (EAC) remains unexplored. To fill this gap, we analyzed esophageal cancer data from the TCGA, and GSE1420 datasets. Our analysis revealed significantly elevated levels of PRDX2 in EAC compared to normal esophageal epithelium in both TCGA (Fig. 1A) and GSE1420 datasets (Fig. 1B). To validate these findings at the protein level, IHC staining was performed on a human tissue microarray (TMA). This analysis confirmed a significant overexpression of PRDX2 in EAC tissues compared to normal esophageal samples (Fig. 1C).

Fig. 1.

PRDX2 is overexpressed in human esophageal adenocarcinoma and is induced by acidic bile salts. (A and B) Violin plots show gene expression of PRDX2 in esophageal adenocarcinoma (EAC) and normal esophagus (NE) tissues in TCGA-EAC database (A) and GSE1420 dataset (B), respectively. (C) Representative immunohistochemistry (IHC) staining images for PRDX2 protein expression in EAC and NE (left) and the corresponding quantification of index scores based on the staining of a human EAC tissue microarray. (D and E) qRT-PCR shows mRNA levels of PRDX2 in OE33 (D) and SKGT4 (E) cells during indicated recovery time courses after ABS exposure. (F and G) Western blotting shows protein levels of PRDX2 and β-actin in OE33 (F) and SKGT4 (G) cells during indicated recovery time courses after ABS exposure. (H and I) Representative immunofluorescent staining images for PRDX2 protein in OE33 (H) and SKGT4 (I) cells with ABS treatment. UT, untreated as control. Mean fluorescent intensity per cell was quantified using ImageJ from three independent fields. DAPI was used for nuclear staining. (J) Representative immunofluorescent staining images for PRDX2 in OE33-derived 3D organotypic culture with ABS treatment. UT, untreated as control. Statistical data are shown as mean ± SEM. Wilcoxon test was applied for calculating statistical significance for two group comparisons in the public datasets. t-test was performed to analyze the experimental data for two group comparisons. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To assess whether reflux conditions in patients with GERD influence PRDX2 regulation, we modeled GERD episodes using ABS cocktail exposure for 20 min followed by a recovery period, mimicking a GERD episode in patients. In OE33 and SKGT4 cells, we observed a significant increase in PRDX2 mRNA (Fig. 1D and E) and protein levels (Fig. 1F and G) over a 6-h recovery period. Consistent with these observations, immunofluorescence staining revealed remarkable enhanced fluorescence intensity of PRDX2 in 2D cultures of ABS-treated OE33 (Fig. 1H) and SKGT4 (Fig. 1I) cells. This observation was also recapitulated in a 3D OTC model of reflux conditions (Fig. 1J). Together, these findings indicate that PRDX2 is overexpressed in human EAC tissues and can be transcriptionally induced by ABS exposure, mimicking reflux conditions.

3.2. PRDX2 protects esophageal cells from ABS-induced lipid peroxidation and ferroptosis

Previous studies have shown that exposure to ABS increases the ROS generation and oxidative stress [49] [50]. In this study, we measured intracellular ROS levels in OE33 cells using a specific fluorogenic dye, CM-H2DCFDA, followed by flow cytometry analysis. Notably, ROS levels surged significantly within 20 min after ABS exposure but returned to baseline after 3 h, suggesting a rapid cellular recovery capacity (Fig. 2A). A similar response was observed in FLO-1 cells (Supplementary Fig. 1A). Since ROS can trigger lipid peroxidation, a known driver of ferroptosis, we then assessed dynamic changes in lipid peroxidation using the C11-BODIPY fluorescent probe. Consistent with the ROS patterns, lipid peroxidation levels increased at 20 min post-ABS exposure but resolved by 3 h in both OE33 (Fig. 2B) and FLO-1 (Supplementary Fig. 1B) cells. We also examined the protein levels of 4-hydroxynonenal (4-HNE), one of the major end products of lipid peroxidation, by Western Blot. 4-HNE protein was greatly induced at 20 min after ABS exposure but dropped to basal levels after 3 h in both OE33 (Fig. 2C) and FLO-1 (Supplementary Fig. 1C) cells. To confirm the prerequisite role of ROS in mediating these effects, we pretreated cells with a ROS scavenger, N-acetylcysteine (NAC), 1 h prior to ABS exposure. NAC pretreatment markedly suppressed ABS-induced accumulation of lipid peroxidation at 20 min (Fig. 2D), underscoring ROS as the primary instigator of this oxidative cascade.

Fig. 2.

PRDX2 protects esophageal cells from ABS-induced lipid peroxidation and ferroptosis. (A and B) ROS levels (A) and lipid peroxidation levels (B) were detected by flow cytometry using CM-H2DCFDA and C11 BODIPY, respectively, in OE33 cells during 3h-recovery after ABS exposure. The representative flow cytometry profile overlap is shown on the left and the quantification of mean fluorescence intensity is shown on the right. (C) Western blots show protein levels of 4-HNE and β-actin in OE33 cells at 20min- and 3h-recovery time points after ABS exposure. (D) Lipid peroxidation levels were detected by flow cytometry using C11 BODIPY assay as in (B). OE33 cells with or without NAC pretreatment followed by ABS exposure were shown. (E and F) OE19 cells were transfected with PRDX2 specific siRNA (siPRDX2) or control siRNA (siCtrl). ROS levels (E) and lipid peroxidation levels (F) were measured by flow cytometry using CM-H₂DCFDA and C11 BODIPY respectively, during 3h-recovery after ABS exposure. The representative flow cytometry profile overlaps are shown on the left and the quantifications of mean fluorescence intensity are shown on the right. (G) Western blots show protein levels of 4-HNE and β-actin in OE19 cells with PRDX2 knockdown siRNA or control siRNA, at 20min- and 3h-recovery time points after ABS exposure. Statistical data are shown as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and NS, no significance as calculated by t-test for two group comparisons. UT, untreated.

PRDX2, a member of the antioxidant enzyme family, is crucial for balancing intracellular levels of ROS and mitigating oxidative stress. Therefore, we hypothesized that PRDX2 may protect cells from reflux-induced damage by modulating ROS and lipid peroxidation. To test this hypothesis, we exposed PRDX2-silenced (via siRNA) and control cells to ABS. We measured ROS and lipid peroxidation levels at baseline and during recovery using flow cytometry. As expected, control cells exhibited transient ROS accumulation at 20 min post-ABS exposure, followed by full recovery at 3 h. In contrast, PRDX2-silenced cells retained high ROS levels at 3 h (Fig. 2E and Supplementary Fig. 1D). Consequently, PRDX2 silencing impaired the resolution of ABS-induced lipid peroxidation by 3 h, unlike control cells which showed full recovery (Fig. 2F and Supplementary Fig. 1E). In agreement with this finding, 4-HNE protein levels in PRDX2-depleted cells remained elevated at 3 h post-ABS exposure, while levels in control cells returned to baseline by 3 h (Fig. 2G and Supplementary Fig. 1F).

3.3. PRDX2 alleviates ferroptosis-related cell death under acidic bile salts (ABS) through altering GPX4 protein half-life

To assess the impact of PRDX2 depletion on cell survival under reflux conditions, we conducted CellTiter-Glo cell viability assays. PRDX2-silenced cells showed significantly reduced viability compared to controls at 72 h post-ABS exposure (Fig. 3A and B). To delineate ferroptosis involvement, we introduced Ferrostatin-1 (Fer-1), a potent and selective inhibitor of lipid peroxidation, into the cell viability assays. Application of Fer-1 significantly enhanced the viability of PRDX2-silenced cells under reflux conditions, indicating that ferroptosis plays a role in ABS-induced cell death in PRDX2 silenced cells (Fig. 3C and D). Furthermore, we screened multiple key ferroptosis regulators (GPX4, xCT, TFR1, FTL, FSP1 and ACSL4) via Western blot in the conditions of PRDX2 knockdown and ABS exposure. PRDX2 silencing suppressed ABS-induced upregulation of GPX4—a negative regulator of lipid peroxidation (Fig. 3E and F). No obvious or consistent changes were observed in the expression of xCT, TFR1, FTL and FSP1 when comparing PRDX2 knockdown to control (Supplementary Fig. 2A and 2B). Meanwhile, ectopic overexpression of PRDX2 significantly increased GPX4 protein levels compared to controls (Fig. 3G and H), without affecting the levels of xCT, TFR1, FTL and FSP1 (Supplementary Fig. 2C and 2D). These results underscore the essential role of PRDX2 in maintaining oxidative homeostasis and shielding cells from ABS-induced lipid peroxidation and ferroptosis.

Fig. 3.

PRDX2 alleviates ferroptosis-related cell death under acidic bile salts (ABS) through altering GPX4 protein half-life. (A and B) OE33 (A) and OE19 (B) cells were transfected with control siRNA or PRDX2-knockdown siRNA, followed by ABS exposure; Relative cell viability measured every 24 h by CellTiter-Glo assay are shown during 72-h recovery (d1-d3, day1-day3). (C and D) OE33 (C) and OE19 (D) cells transfected with control siRNA or PRDX2-knockdown siRNA were exposed to ABS, followed by treatment with DMSO or Ferrostatin-1 (Fer-1), a potent and selective inhibitor of lipid peroxidation. Relative cell viability measured every 24 h by CellTiter-Glo assay are shown during 72-h recovery after ABS exposure (d1-d3, day1-day3). (E and F) OE33 (E) and OE19 (F) cells were transfected with control siRNA or PRDX2-knockdown siRNA, followed by ABS exposure; protein levels of PRDX2, GPX4 and β-actin were examined by Western blot analysis. (G and H) Western blots show upregulation of GPX4 protein level after ectopic PRDX2 overexpression in OE33 (G) and SKGT4 (H) cells. (I, J and K) GPX4 protein degradation curves under cycloheximide (CHX) treatment (100 μg/ml) following the indicated time points in OE33 cells; ABS exposure verse untreated control (UT)(I), or PRDX2 depletion (shPRDX2) verse shRNA control (shCtrl) (J), or shPRDX2 verse shCtrl under ABS condition (K). The quantification of GPX4 protein levels was based on the measurements of band intensity of western blotting shown in Supplementary Fig. 3, panel 3A-1 to 3A-3, 3C-1 to 3C-3 and 3D-1 to 3D-3, respectively, normalized to the band intensity of β-actin of the same sample. Statistical data are shown as mean ± SEM. ∗p < 0.05, and ∗∗p < 0.01 as calculated by t-test for two group comparisons. UT, untreated.

To further explore the regulatory relationship between PRDX2 and GPX4 under reflux conditions, we examined GPX4 mRNA levels by qRT-PCR. While changes in protein levels were obvious, GPX4 mRNA levels remained unchanged after PRDX2 silencing or ectopic expression (Supplementary Fig. 2E–2H), suggesting post-transcriptional regulations. We then assessed GPX4 protein stability using CHX chase assays. ABS exposure significantly increased GPX4 protein half-life time from 3.148 h to 6.769 h (Fig. 3I) based on the Western blots’ bands intensities shown in Supplementary Fig. 3A–1 to 3A-3. To investigate the effect of PRDX2 depletion on GPX4 protein stabilization, OE33 cells with stable PRDX2 knockdown were generated using lentiviral shRNA. Western blot confirmed the efficient PRDX2 silencing (Supplementary Fig. 3B). CHX chase assays revealed that PRDX2 depletion decreased GPX4 protein half-life time from 3.433 h to 1.746 h (Fig. 3J and Supplementary Fig. 3C–1 to 3C-3) and reduced ABS-induced GPX4 stability, with half-life time from 6.999 h to 3.043 h (Fig. 3K and Supplementary Fig. 3D–1 to 3D-3). These findings prompted us to investigate the mechanism by which PRDX2 regulates GPX4 protein stability.

3.4. PRDX2 stabilizes GPX4 protein by regulating OTUB1-dependent deubiquitination

To identify potential interaction partners of PRDX2 and GPX4 proteins, we searched BioGRID (https://thebiogrid.org/) and IntAct (https://www.ebi.ac.uk/intact/search?query=EBI-2510844). OTUB1 was notably identified as the only candidate partner with overlap in this analysis (Fig. 4A). OTUB1 is a deubiquitinase that plays a critical role in the ubiquitin-proteasome system [51]. Gene Set Enrichment Analysis (GSEA) of TCGA-EAC database, comparing PRDX2-high and PRDX2-low groups, showed a trend for enrichment of the deubiquitination pathway in the PRDX2-high group, despite a non-significant p-value. (Supplementary Fig. 4A), suggesting PRDX2 may be involved in protein deubiquitination. Experimental validation confirmed interactions among PRDX2, OTUB1, and GPX4 in OE33 cells, with or without ABS exposure (Fig. 4B). Furthermore, silencing OTUB1 via siRNA diminished ABS-induced GPX4 protein accumulation, demonstrated the critical roles of OTUB1 in GPX4 protein regulation (Supplementary Fig. 4B).

Fig. 4.

PRDX2 stabilizes GPX4 protein by regulating OTUB1-dependent deubiquitination. (A) Screening of the public databases (BioGRID and IntAct) identified OTUB1 as the only one partner that interacts with both PRDX2 and GPX4. (B) Immunoprecipitation (IP) of GPX4 protein using an anti-GPX4 antibody in OE33 cells with/without ABS exposure; IgG was used as a negative control. Western blotting with GPX4, PRDX2 and OTUB1 are shown. (C) For analysis of GPX4 ubiquitination, OE33 cells were transfected with control siRNA or OTUB1 siRNA together with Ub-HA plasmid followed by ABS exposure under MG132 presence. IP of GPX4 protein was performed and western blotting with Ub-HA, GPX4 and OTUB1 are shown. GPX4 ubiquitination was evaluated by immunoblots with anti-HA antibody. IgG was used as a negative control. (D) OE33 cells were transfected with control shRNA or PRDX2 shRNA with Ub-HA plasmid followed by transfection of PcDNA control or Flag-OTUB1 plasmids in shPRDX2 cells. IP of GPX4 protein was performed and western blotting with Ub-HA, PRDX2, GPX4 and OTUB1 are shown. GPX4 ubiquitination was evaluated by immunoblots with anti-HA antibody. IgG was used as a negative control. (E) OE33 cells were transfected with control vector or PRDX2 expression plasmid, together with Ub-HA plasmid. IP of GPX4 protein was performed and western blotting with Ub-HA, PRDX2, GPX4 and OTUB1 are shown. GPX4 ubiquitination was evaluated by immunoblots with anti-HA antibody. (F and G) OE33 cells were transfected with control vector or PRDX2 expression plasmid, IP of PRDX2 protein (F) and OTUB1 protein (G) were performed using an anti-PRDX2 antibody or an anti-OTUB1 antibody, respectively. IgG was used as a negative control. Western blotting with PRDX2, OTUB1, GPX4 and β-actin are shown.

To investigate the effects of OTUB1 on the ubiquitination levels of endogenous GPX4 under reflux conditions, we immunoprecipitated GPX4 in OE33 cells with OTUB1 silencing or control cells, in the absence or presence of ABS exposure. The ubiquitination levels were then detected by Western blot. The results demonstrated an increased ubiquitination of GPX4 after OTUB1 knockdown, compared to the negative control. This increase in GPX4 ubiquitination was detected both at the basal level (comparing lane 2 to lane 1) and under ABS conditions (comparing lane 4 to lane 3) (Fig. 4C). To further determine OTUB1's involvement in PRDX2-dependent regulation of GPX4 protein, we transfected PRDX2-depleted cells with OTUB1-WT overexpression plasmid or control vector. The protein level of GPX4 that decreased by PRDX2 depletion (comparing lane 2 to lane 1) was rescued by ectopic OTUB1 overexpression (comparing lane 4 to lane 2) (Supplementary Fig. 4C). More importantly, GPX4-ubiquitination was increased after PRDX2 depletion (comparing lane 2 to lane 1), which was diminished by ectopic OTUB1 overexpression (comparing lane 4 to lane 2) but not vector control (comparing lane 3 to lane 2) (Fig. 4D). Moreover, PRDX2 overexpression greatly relieved the ubiquitination level of GPX4 (comparing lane 4 to lane 2) (Fig. 4E). While GPX4-OTUB1 binding was unaffected, OTUB1 expression increased in Input samples following ectopic PRDX2 overexpression. This elevated OTUB1 may mediate the observed reduction in GPX4 ubiquitination.

To further clarify the essential involvement of OTUB1 in PRDX2-mediated regulation of GPX4 ubiquitination, we performed Co-IP assays in PRDX2-overexpressing cells using anti-PRDX2 antibody (Fig. 4F) and anti-OUTB1 antibody (Fig. 4G), respectively. Our findings confirmed that PRDX2 recruits OTUB1 to GPX4. In addition, ectopic PRDX2 overexpression increased OTUB1 protein levels, which is consistent with previous findings. Taken together, these findings indicate that, under reflux conditions, PRDX2 is essential for stabilizing GPX4 protein by alleviating its ubiquitination in an OTUB1-dependent manner.

3.5. ABS-induced PRDX2 promotes chemoresistance in EAC cells via inhibiting ferroptosis

Chronic reflux with ABS exposure is the main risk factor for EAC that remains while patients are under chemotherapeutic treatment. Ferroptosis suppression has been recently recognized as an important mechanism of resistance to chemotherapy. Given the preceding observations, we hypothesized that ABS-induced PRDX2 contributes to chemoresistance in EAC cells by inhibiting ferroptosis. Initial experiments determined the IC50s of oxaliplatin across four EAC cell lines: OE19, SKGT4, FLO-1, and OE33(Supplementary Fig. 5A). We found OE19 (IC50 = 46.539 μM) and SKGT4 (IC50 = 61.238 μM) cells are intrinsically resistant to oxaliplatin, while FLO-1 (IC50 = 1.913 μM) and OE33 (IC50 = 0.856 μM) are sensitive. To investigate the effect of chronic ABS exposure on chemoresistance, we exposed OE33 and FLO-1 cells (both are sensitive to oxaliplatin) to ABS (20 min per day) for two weeks, and examined the IC50 to oxaliplatin. Notably, the repeated ABS (rABS) exposures significantly increased oxaliplatin IC50 in OE33 (Fig. 5A) and FLO-1 (Supplementary Fig. 5B) cells as compared to parental cells, suggesting that cells adapted to chronic ABS become refractory to oxaliplatin. We detected increased protein levels of PRDX2 and GPX4 in these rABS treated cells as compared to parental cells (Fig. 5B and Supplementary Fig. 5C).

Fig. 5.

ABS-induced PRDX2 promotes chemoresistance in EAC cells via inhibiting ferroptosis. (A) CellTiter-Glo cell viability assay shows IC50s of oxaliplatin treatment in OE33 cells with or without repeated ABS exposure (rABS). (B) Western blots show PRDX2, GPX4 and β-actin levels in OE33 cells with or without repeated ABS exposure (rABS). (C) Violin plot shows gene expression of PRDX2 in non-responding (NR) and responding (R) resectable EAC in GSE165252 dataset. (D) Pearson correlation analysis shows significant positive correlation (R = 0.36, P = 0.00055) between PRDX2 gene expression and single-sample gene set enrichment score of multiple drug resistance pathway signaling in TCGA-EAC database. (E) CellTiter-Glo cell viability assay shows IC50s of oxaliplatin treatment in OE19 cells with PRDX2 stable silencing (shPRDX2) comparing to control cells (shCtrl). (F) CellTiter-Glo cell viability assay shows IC50s of oxaliplatin treatment in OE19 cells with or without Erastin at indicated concentrations. (G) Western blots show GPX4, PRDX2, and β-actin levels in OE19 cells with oxaliplatin, Erastin, or combination treatment of both and untreated controls. (H) Patient derived EAC organoids were subjected to stable PRDX2 silencing (shPRDX2) or scramble shRNA, followed by oxaliplatin (oxp) treatment or vehicle; representative images of the organoids under white light are shown. (I and J) Quantifications of the size (I) and the number (J) of the organoids in panel (H) are shown. (K) Representative immunofluorescent staining images of PRDX2 (green) and GPX4 (red) from the organoids in panel (H). Statistical data are shown as mean ± SEM. Extra sum-of-squares F-test was used to compare the two dose-response curves in (A). Wilcoxon test was applied for calculating statistical significance for two group comparisons in the public dataset in (C). t-Test was employed to determine statistical significance of Pearson correlation coefficients between two continuous variables in (D). t-Test was also performed to analyze the experimental data for two group comparisons. ∗p < 0.05.

To explore the clinical relevance of these findings, gene expression analysis was performed on the GSE165252 dataset, which includes patients’ responses to neoadjuvant chemoradiotherapy in combination with a PD-L1 inhibitor (atezolizumab). Significantly higher PRDX2 levels were observed in non-responding EAC patients, compared to responders (Fig. 5C). Additionally, a significant positive correlation between PRDX2 expression and the multi-drug resistance gene signature was identified in the TCGA-EAC database using GSVA (Fig. 5D). These data strongly suggest that ABS-induced PRDX2 may contribute to oxaliplatin resistance in EAC cells. To confirm these observations in cellular models, OE19 and SKGT4 cells, two oxaliplatin-resistant EAC cell lines, were transduced with lentiviral shRNA for stable PRDX2 knockdown. Western blot confirmed the knockdown efficiency (Supplementary Fig. 5D and 5E). PRDX2 depletion significantly increased oxaliplatin sensitivity showing significantly reduced IC50 values (Fig. 5E and Supplementary Fig. 5F). To investigate the role of ferroptosis in oxaliplatin resistance, we used Erastin, a ferroptosis inducer. We found that Erastin significantly increased oxaliplatin sensitivity in OE19 cells (Fig. 5F), accompanied by decreased GPX4 protein levels, confirming ferroptosis induction (Fig. 5G). To further validate the vital role of PRDX2 in mediating chemoresistance, human EAC organoids were transduced with lentiviral shRNA targeting PRDX2 or a control shRNA, followed by treatment with or without oxaliplatin (Fig. 5H). Stable PRDX2 silencing combined with oxaliplatin resulted in a remarkable decrease in organoid number and size compared with the other three groups (Fig. 5I and J). Dual Immunofluorescent staining of PRDX2 and GPX4 confirmed their co-localization, mainly in the cytosol (Fig. 5K). Notably, the combination of PRDX2 depletion and oxaliplatin treatment decreased GPX4 intensity to the largest extend, indicating induction of ferroptotic cell death (Fig. 5K, Supplementary Fig. 5G and 5H). Pearson correlation analysis revealed strong positive association between PRDX2 and GPX4 protein levels in human EAC organoids (Supplementary Fig. 5I, R² = 0.6557, p < 0.0001). Taken together, these findings demonstrate that inhibition of ferroptosis plays a significant role in maintaining chemoresistance in EAC, with PRDX2 being a key contributor. Targeting PRDX2 re-sensitized EAC cells to oxaliplatin treatment by promoting ferroptotic cell death.

3.6. PRDX2 is regulated by APE1 in response to ABS exposure

Next, we investigated the mechanisms driving ABS-induced PRDX2 upregulation. Previously, we reported that ABS exposure induces APE1 protein which mitigates reflux-associated ROS accumulation and oxidative DNA damage in BE and EAC [44]. We identified significant positive correlations between APE1 and PRDX2 gene expression in the TCGA-EAC cohort (Fig. 6A) and GSE165252 dataset (Fig. 6B). We hypothesized that APE1 regulates PRDX2 expression under reflux conditions. Ectopic overexpression of APE1 in OE33 cells markedly increased PRDX2 levels at both mRNA (Fig. 6C) and protein levels (Fig. 6D). Similar findings were achieved in SKGT4 cells (Supplementary Fig. 6A and 6B). In contrast, APE1 silencing attenuated ABS-induced PRDX2 transcripts and protein in OE33 (Fig. 6E and F) and SKGT4 cells (Supplementary Fig. 6C and 6D). Moreover, in response to ABS exposure, we noted simultaneous upregulation of APE1 (green) and PRDX2 (red) proteins in OE33-derived 3D organotypic cultures, as confirmed by dual immunofluorescence co-staining (Fig. 6G). This phenomenon was mirrored in human EAC tissues, where overexpression of both proteins was similarly detected (Fig. 6H).

Fig. 6.

PRDX2 is regulated by APE1 in response to ABS exposure. (A and B) Pearson correlation analysis of APE1 and PRDX2 at transcription level in TCGA-EAC database (A) and the GSE165252 dataset (B). (C) qRT-PCR shows PRDX2 mRNA expression was significantly elevated in the APE1 overexpressed cells compared to vector control cells. (D) Western blots show the protein levels of APE1, PRDX2 and β-actin in OE33 cells with or without ectopic APE1 overexpression. (E) qRT-PCR shows PRDX2 mRNA levels in transient APE1 knockdown (siAPE1) or control siRNA (siCtrl), with or without ABS exposure. (F) Western blots show the protein levels of APE1, PRDX2 and β-actin in OE33 cells in transient APE1 knockdown (siAPE1) or control siRNA (siCtrl), with or without ABS exposure. (G and H) Representative immunofluorescent staining of APE1 (green) and PRDX2 (red) in OE33-derived 3D organotypic culture with or without ABS treatment (G) and human samples of EAC and normal esophagus (NE) (H). Statistical data are shown as mean ± SEM. t-Test was applied to determine the statistical significance of Pearson correlation coefficients between two continuous variables, and to analyze the experimental data for two group comparisons. ∗p < 0.05, ∗∗p < 0.01.

3.7. APE1 redox function is critical for ABS-induced, NF-kB-dependent PRDX2 activation

A previous study demonstrated that NF-κB can bind to the PRDX2 promoter, inducing its transcription during gastric tumorigenesis [17]. We identified 12 predicted NF-κB binding sites on the PRDX2 promoter and categorized these into three clusters (Cluster A to C, Supplementary Fig. 7A). To investigate whether NF-kB binds to PRDX2 in EAC cells, we performed ChIP assays using a phospho–NF–κB-p65 antibody, followed by qPCR with primers spanning all three clusters. Primer set 1, targeting the region from −1037 to −1024 bp, showed significant amplification in ChIP samples with phospho–NF–κB-p65 antibody. Notably, this binding was significantly enhanced after ABS exposure compared to untreated controls (Supplementary Fig. 7A). In line with these findings, ectopic overexpression of NF-κB-p65 in OE33 (Supplementary Fig. 7B and 7D) and SKGT4 (Supplementary Fig. 7C and 7E) cells increased PRDX2 mRNA and protein levels. Conversely, the NF-κB inhibitor BAY 11-7082 suppressed TNF-α-induced phosphorylation of NF-κB p65 and reduced PRDX2 expression (Supplementary Fig. 7F and 7G). These results demonstrate that NF-κB directly regulates PRDX2 transcription through promoter binding, an effect that is further enhanced by ABS exposure.

Previous studies by our team and others have established that APE1 regulates multiple redox-dependent transcription factors (TFs), including NF-κB, to promote gene transcription [37,[52], [53], [54]]. This function relies on the N-terminal Cys65 residue of APE1 [55], which keeps critical cysteines of TFs in a reduced state to enhance their DNA-binding affinity [56,57]. Given APE1's functions in redox-mediated transcriptional regulation, we hypothesized that APE1 regulates ABS's induction of PRDX2 through APE1-redox mediated NF-κB activation. Western blot analysis demonstrated increased protein levels of APE1, phospho-p65, and PRDX2 after transient (Fig. 7A) and repeated ABS treatments in OE33 cells (Fig. 7B), and in SKGT4 cells (Supplementary Fig. 7H and 7I). Notably, BAY 11-7082 blocked APE1 overexpression-induced phospho-p65 activation and PRDX2 upregulation (Fig. 7C; Supplementary Fig. 7J), confirming that APE1's regulation of PRDX2 is NF-κB-dependent.

Fig. 7.

PRDX2 is regulated by APE1 redox function in an NF-κB dependent manner. (A) Western blots show the protein levels of APE1, phosphor-p65, total p65, PRDX2, and β-actin at different time points after a transient ABS exposure in OE33 cells. (B) Western blots show the protein levels of APE1, phosphor-p65, total p65, PRDX2, and β-actin in OE33 cells after 14 days repeated ABS exposure (rABS). (C) OE33 cells were transfected with APE1 overexpression plasmid or control vector, followed by treatment with BAY 11-7082, an inhibitor of canonical NF-κB pathway. Western blots show the protein levels of APE1, phosphor-p65, total p65, PRDX2, and β-actin. (D) OE33 cells with APE1 stable silencing (shAPE1) were transfected with APE1-wild-type (WT) expression plasmid or APE1-redox-deficient-mutant (C65A) plasmid, followed by ABS exposure or vehicle control. Western blots show the protein levels of APE1, phosphor-p65, total p65, PRDX2, and β-actin. (E) Western blots show the protein levels of APE1, phosphor-p65, total p65, PRDX2, GPX4 and β-actin in OE33 cells with a transient ABS exposure in the presence or absence of APX2009 treatment (5 μM and 10 μM), an APE1 redox-specific inhibitor. (F) qRT-PCR shows PRDX2 mRNA level change in OE33 cells after ABS exposure, with or without APX2009 administration (10 μM). (G) Representative immunofluorescent images of APE1 (green) and PRDX2 (red), alongside with H&E staining and immunohistochemistry staining for 4-HNE (right panel), in tissue samples derived from a PDX mouse model with or without APX2009 treatment. Statistical data are shown as mean ± SEM. t-Test was used for two group comparisons using experimental data. ∗∗p < 0.01.

To evaluate the requirement of APE1's redox activity in the ABS-induced NF-κB/PRDX2 pathway, we reconstituted either wild-type APE1 (WT) or a redox-deficient mutant (C65A) plasmid into APE1-silenced OE33 cells (Fig. 7D). WT APE1 restored phospho-p65 and PRDX2 levels, whereas the C65A mutant did not. Furthermore, cells expressing the C65A mutant failed to respond to ABS stimulation (Fig. 7D). To further validate these findings, we found that the APE1 redox inhibitor APX2009 blocked ABS-induced upregulation of phospho-p65, PRDX2 and GPX4 at protein level (Fig. 7E; Supplementary Fig. 7K) and PRDX2 mRNA levels (Fig. 7F; Supplementary Fig. 7L). Finally, in a PDX mouse model, APX2009 treatment reduced PRDX2 immunostaining, as compared to untreated controls (Fig. 7G). In contrast, the 4-HNE level estimated by IHC staining was significantly increased after APX2009 treatment comparing to the untreated control (Fig. 7G and Supplementary Fig. 7M). In summary, these results demonstrate that APE1's redox function is essential for ABS-induced, NF-κB-dependent PRDX2 regulation.

3.8. Inhibition of APE1 redox function reverses oxaliplatin resistance

Collectively, our results demonstrate that ABS-induced upregulation of PRDX2 suppresses ferroptosis by increasing expression of its negative regulator, GPX4, thereby promoting chemoresistance. Importantly, this mechanism is dependent on APE1's redox activity, indicating that pharmacological inhibition of APE1 redox function—such as with APX2009—may help overcome chemoresistance to agents like oxaliplatin in EAC. To test this, we treated oxaliplatin-resistant OE19 and SKGT4 cells with two concentrations of APX2009 and assessed their sensitivity to oxaliplatin. APX2009 synergistically decreased the IC50 of oxaliplatin in both cell lines: from 46.54 μM to 15.68 μM in OE19 (with 10 μM APX2009; Fig. 8A) and from 61.24 μM to 15.17 μM in SKGT4 (with 10 μM APX2009; Fig. 8B).

Fig. 8.

Inhibition of APE1 redox function reverses oxaliplatin resistance. (A and B) CellTiter-Glo cell viability assay shows IC50s of oxaliplatin treatment in OE19 (A) and SKGT4 (B) cells with or without APX2009 presence (5 μM and 10 μM). (C) Tumor growth curve using average tumor volume of OE19-derived xenografts with or without oxaliplatin 1 mg/kg, APX2009 20 mg/kg, or combination treatments for at most 28 days. (D) Kaplan–Meier survival curve for the xenografts in (C) following the treatment endpoint. (E) Representative IHC images of Ki-67, immunofluorescent images of APE1 (green) and PRDX2 (Red) using the tissues from xenografts in (C).

To validate these findings in vivo and assess the therapeutic potential of APX2009, we established tumor xenografts derived from OE19, an EAC cell line that is intrinsically resistant to oxaliplatin. Combination therapy with oxaliplatin and APX2009 resulted in significantly slower tumor growth (Fig. 8C) and prolonged survival (Fig. 8D) compared to either monotherapy or control groups. Immunohistochemical staining for the proliferation marker Ki-67 showed a marked reduction in the combination group relative to the others (Fig. 8E and Supplementary Fig. 8A). Additionally, immunofluorescence analysis revealed significant decreases in APE1, PRDX2 (Fig. 8E), and GPX4 and 4-HNE expression (Supplementary Fig. 8B) in tumors from mice receiving combination therapy. Notably, APE1 and PRDX2 were co-expressed, with APE1 localized predominantly in the nucleus and PRDX2 mainly in the cytoplasm (Fig. 8E). Using IHC, we observed significant upregulations of 4-HNE levels in mice receiving combination therapy, in comparison with the other groups, indicating an increase in lipid peroxidation and ferroptosis, contributing to the observed tumor suppression (Supplementary Fig. 8B and 8C).

4. Discussion

EAC is an aggressive malignancy that has shown a seven-fold increase in incidence in the last few decades, especially in the United States and western countries [58]. Chronic GERD is the main risk factor for the development of a pre-neoplastic condition known as Barrett's esophagus (BE) and its progression to EAC [5] [6]. GERD leads to the accumulation of high levels of reactive oxygen species and oxidative stress with aberrant cellular redox homeostasis. This leads to increase in the levels of oxidative DNA damage and activation of several pro-tumorigenic signaling pathways [38,52,59]. As EAC tumors arise and progress in the presence of highly oxidative environment, cancer cells must develop adaptative antioxidant pro-survival mechanisms. These inherent adaptive mechanisms attribute to the poor responsiveness to current chemotherapeutic strategies in EAC [60], resulting in poor clinical outcomes [2]. In this study, we report on the cross talk between intracellular redox status, antioxidant defense system, ferroptotic cell death and chemoresistance in EAC. We introduce a novel APE1/NF-κB/PRDX2/GPX4 signaling axis in EAC.

Prx2 (PRDX2) is one of the most abundantly expressed peroxiredoxin members with dominant peroxidase activity in multiple cell types [11,61]. Aberrant upregulation of PRDX2 has been reported in several cancer types, including gastric [17], colorectal [18], pancreatic cancer [19], contributing to tumor progression, lymph node metastasis, radio-resistance and chemoresistance. In this study, we detected PRDX2 overexpression in human EAC tissues and observed a significant increase in PRDX2 levels following ABS exposure, mimicking GERD in vitro. Our findings illustrate the role of PRDX2 in resolving oxidative stress and lipid peroxidation in GERD's highly oxidative environment associated with EAC tumorigenesis and progression. We found that PRDX2 is not only crucial for regulating oxidative stress but also for protecting cells against ferroptosis under stress conditions. Interestingly, we observed that knockdown of PRDX2 alone did not induce significant ROS surge (Fig. 2E and Supplementary Fig. 1D). However, ABS exposures induced significant ROS surge, where PRDX2 silencing failed to scavenge the excess ROS, resulting in cell death. This finding explains why PRDX2 silencing alone had minimal effects on cell viability, whereas its knockdown following ABS exposure promotes cell death. Taken together, we can conclude that PRDX2 supports cancer cell survival, particularly under oxidative stress conditions.

Ferroptosis is a unique type of programmed cell death. It is strongly linked to dysfunction of antioxidant systems and iron accumulation, which favors excessive production of lipid peroxidation, damage to cell membranes, and finally, cell death [62]. Canonical key regulators of ferroptosis include System Xc−, a membrane-located cysteine/glutamate antiporter with a core subunit named SLC7A11 (xCT), and glutathione peroxidase 4 (GPX4). Our data supported that GPX4 is a major player in ABS-modulated ferroptosis. Mechanistically, we found that PRDX2 regulates GPX4 protein stability through modulating ubiquitination-mediated protein degradation. We further reported protein interactions among PRDX2, GPX4 and OTUB1. OTUB1 is a founding member of OTU-domain deubiquitinases (DUBs) that belongs to the proteosome superfamily. Canonically, OTUB1 relieves protein degradation by catalyzing polyubiquitin chains of substrate proteins, with a special preference for K48-linked chains [63,64]. On the other hand, a non-canonical function of OTUB1 relies on that it inhibits ubiquitin transfer onto a protein substrate by interacting with the ubiquitin-charged E2 and locking it in an inactive conformation [51,65]. We discovered that GPX4 is regulated by OTUB1 in a PRDX2-dependent manner where PRDX2 plays a critical role in the interaction of GPX4 and OTUB1, relieving GPX4 ubiquitination under reflux conditions. Accumulation of GPX4 protein protected cancer cells from ferroptotic cell death under stress. However, the detailed mechanism of whether OUTB1 stabilizes GPX4 through its canonical or non-canonical functions needs further investigation.

We also observed that PRDX2 recruits OTUB1 under reflux-induced oxidative stress conditions. PRDX2 is known to serve dual roles in cellular biology, functioning both as a high-efficiency antioxidant enzyme and as a signaling regulator through its scaffolding and redox relay capabilities [66]. Interestingly, many DUBs are reported to be susceptible to ROS-mediated oxidation because of the cysteine amino acid residues in their active sites [67]. Under redox stress conditions, OTUB1 was subjected to S-nitrolyzation at C23 and C91, thus impairing its native structure and protein stability [68]. These findings suggest that over-oxidation of OTUB1 catalytic cysteine not only jeopardized its catalytic function, but destroyed the protein stability. Considering the highly effective antioxidant capability of PRDX2, we hypothesize that under ABS-induced oxidative stress, PRDX2 interacts with OTUB1 to relieve the oxidation status of OTUB1, thus protecting its native structure and promoting its protein stability. However, experimental validation of this hypothesis is needed.

EAC is an aggressive malignancy with a poor clinical outcome largely due to resistance to chemotherapeutics [60]. Accumulating evidence supports the role of anti-ferroptosis in chemoresistance in several cancer types [32]. In this study, we demonstrated that repeated ABS exposure, mimicking chronic GERD conditions in patients with EAC, rendered EAC cells more resistant to oxaliplatin treatment, consistent with clinical data showing frequent resistance to chemotherapeutics in EAC [60]. EAC cells exposed to reflux conditions were more resistant to oxaliplatin and showed higher levels of PRDX2 and GPX4 with suppression of ferroptosis. PRDX2 knockdown or the use of Erastin, a widely used class I ferroptosis inducer (FIN) [69], restored sensitivity to oxaliplatin in cells with constitutive resistance to oxaliplatin, further confirming the role of PRDX2 and ferroptosis in chemotherapeutic response in EAC.

Targeting ferroptosis using FINs has generated considerable interest in cancer research due to their enormous therapeutic potential. However, the clinical application of FINs has been largely hindered by issues of metabolic stability, potency, solubility, and pharmacokinetics [[69], [70], [71]]. In addition, Conoidin A—currently the only specific inhibitor of PRDX2—irreversibly binds to the catalytic cysteine residues, thereby inactivating its antioxidant enzymatic activity. However, it remains at the research stage and has not yet entered clinical trials [72,73]. Our study identified APE1 as a key druggable upstream regulator of PRDX2 and ferroptosis, highlighting its critical roles as a potential therapeutic target. Increased levels of APE1 (also known as REF [1]) have been reported in EAC [44,52,74]. APE1 regulates PRDX2 transcripts via its redox function in an NF-κB-dependent manner. This regulatory signaling axis between APE1 and PRDX is consistent with earlier work in gastric cancer [17]. Inhibition of APE1's redox activity, using APX2009 overcame chemoresistance by sequentially downregulating PRDX2 and GPX4 levels, thereby enhancing ferroptosis in inherently resistant EAC cells. Meanwhile, our mechanistic findings call for further development of novel PRDX2 inhibitors that can be applied to clinical to overcome chemotherapy resistance in EAC patients.

5. Conclusions

In summary, PRDX2 plays a critical role in the development of the cellular protective adaptation mechanisms in response to GERD or chemotherapy through ferroptosis inhibition. This provides cancer cells with an inherent capacity to resist chemotherapeutics. Furthermore, our findings highlight a novel APE1-NF-kB-PRDX2-GPX4 signaling axis in EAC, emphasizing the clinical potential of APE1 redox inhibitors and PRDX2 inhibitors for overcoming chemoresistance and improving outcomes for EAC patients.

Ethics approval and consent to participate

All animal experiments were approved by the Institutional Animal Care and Use Committee of University of Miami (IACUC 110-23).

Funding:

This study was supported by the National Cancer Institute (NCI) under award number P01CA268991, and the Sylvester Comprehensive Cancer Center support fund TSPA-2023-03. The NCI-funded Sylvester Comprehensive Cancer Center (P30CA240139) supported the use of flow cytometry (SCR_022501), cancer modeling (SCR_022891), and biostatistics (SCR_022890) shared resources. The content is solely the authors’ responsibility and does not necessarily represent the official views of the National Institutes of Health or the University of Miami.

CRediT authorship contribution statement

Lei Chen: Conceptualization, Formal analysis, Investigation, Methodology. Farah Ballout: Methodology, Resources, Validation. Zheng Chen: Data curation, Methodology, Resources. Krishnapriya Thangaretnam: Methodology, Resources. Jianwen Que: Conceptualization, Methodology, Resources. Xi Steven Chen: Data curation, Formal analysis, Methodology. Oliver Gene McDonald: Formal analysis, Methodology, Resources. Alexander Zaika: Conceptualization, Resources. Alan Livingstone: Resources. Silvia Giordano: Resources. Ramin Shiekhattar: Methodology, Writing – review & editing. Heng Lu: Conceptualization, Methodology, Supervision, Writing – review & editing. Dunfa Peng: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing. Wael El-Rifai: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Abbreviations:

EAC, esophageal adenocarcinoma; BE, Barrett's esophagus; GERD, gastroesophageal reflux disease; ABS, acidic bile salts; rABS, repeated acidic bile salts; ROS, reactive oxygen species; PRDXs/Prxs, peroxiredoxins; GPX4, glutathione peroxidase 4; GSH, glutathione; PUFA, polyunsaturated fatty acid; PLs, phospholipids; FIN, ferroptosis inducer; TCGA, The Cancer Genome Atlas; GEO, Gene Expression Omnibus; GSEA, Gene Set Enrichment Analysis; GSVA, Gene Set Variation Analysis; NCBI, National Center for Biotechnology Information; hEF, human esophageal fibroblast; FBS, fetal bovine serum; PBST, PBS containing 0.1% Tween® 20; IC50, half-maximal inhibitory concentration; OTC, organotypic culture, PDX, patient-derived xenografts; CDX, cell line-derived xenografts; TMA, tissue microarray; Ct, threshold cycle; ChIP, chromatin immunoprecipitation; PMSF, phenylmethanesulfonyl fluoride; PIC, protease inhibitor cocktail; H&E, hematoxylin and eosin; RT, room temperature; FFPE, formalin fixed paraffin-embedded; DAPI, 4′,6-diamidino-2-phenylindole; IF, immunofluorescence; IHC, immunohistochemistry; CHX, cycloheximide; IACUC, Institutional Animal Care and Use Committee; SEM, standard error of the mean; NAC, N-acetylcysteine; DUBs, deubiquitinases.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on reasonable request.