Selenium-Thioredoxin Axis Contributes to Ferroptosis Resistance in Pancreatic Cancer Cells

TL;DR

The study finds that selenium and thioredoxin work together to protect some pancreatic cancer cells from a type of cell death called ferroptosis, suggesting a new way to treat these cancers.

Contribution

The novel contribution is the discovery of a selenium–thioredoxin redox axis that provides ferroptosis resistance in epithelial-type pancreatic cancer cells, independent of GPX4.

Findings

Epithelial-type PDAC cells resist ferroptosis via a selenium–thioredoxin axis, independent of GPX4.

Chemical inhibition of thioredoxin reductases sensitizes epithelial cells to ferroptosis inducers.

Selenium supplementation protects cells from ferroptosis even when glutathione is depleted.

Abstract

Pancreatic ductal adenocarcinoma (PDAC) shows substantial heterogeneity in cysteine dependence and ferroptosis sensitivity. We identify two PDAC subtypes distinguished by EMT status: mesenchymal-like cells are highly cysteine-dependent and rapidly undergo ferroptosis upon cystine deprivation or system xc− inhibition, whereas epithelial-type cells are ferroptosis-resistant. Selenium supplementation protects cells from erastin-induced ferroptosis, and this protection persists even when intracellular glutathione (GSH) is depleted, supporting an additional GPX4-independent protective mechanism. Sepp1 knockdown does not alter sensitivity, indicating that selenium’s protective effect is independent of Sepp1. Instead, epithelial-type cells rely on both cytosolic and mitochondrial thioredoxin reductases (TrxR1 and TrxR2) to maintain ferroptosis resistance. Chemical inhibition of thioredoxin reductases abolishes selenium-mediated protection and sensitizes epithelial cells to ferroptosis inducers, while dual genetic suppression of TrxR1 and TrxR2 similarly restores ferroptosis sensitivity. These findings uncover a selenium–thioredoxin redox axis that functions independently of GPX4 and contributes ferroptosis resistance in epithelial-type PDAC cells. Co-targeting cysteine metabolism and thioredoxin reductases may therefore represent a rational strategy to overcome ferroptosis resistance in some PDAC subtypes.

Figures

Full text

1. Introduction

Pancreatic ductal adenocarcinoma (PDAC) remains one of the deadliest malignancies, with a five-year survival rate below 12% despite major advances in molecular characterization and treatment strategies [1,2,3,4]. PDAC is defined by profound metabolic rewiring driven in part by oncogenic KRAS, which is mutated in more than 90% of PDAC cases, promoting elevated reactive oxygen species (ROS), altered redox homeostasis, and increased reliance on antioxidant pathways [5,6,7,8,9,10]. Among these vulnerabilities, ferroptosis, an iron-dependent form of cell death driven by uncontrolled lipid peroxidation, has emerged as a promising therapeutic avenue for PDAC and other solid tumors [11]. Recent reviews highlight the therapeutic potential of ferroptosis induction through system xc^−^ inhibition, cysteine depletion, nanoparticle-based delivery of ferroptosis inducers, and metabolic combination therapies [12,13,14,15,16].

Cysteine metabolism is central to maintaining redox homeostasis, serving as a precursor for glutathione (GSH) synthesis and other thiol-based antioxidants [14,17,18,19]. PDAC cells exhibit strong cystine/cysteine dependence, and inhibition of system xc^−^ or cysteine depletion induces rapid ferroptosis in a subset of PDAC models [20,21,22,23]. However, ferroptosis sensitivity is highly heterogeneous, and many tumor cells utilize parallel antioxidant systems, including FSP1, peroxiredoxins, and the thioredoxin (Trx) pathway, to limit lipid peroxidation independently of GPX4 [24,25,26].

Emerging evidence also links epithelial–mesenchymal transition (EMT) programs to ferroptosis sensitivity. EMT alters lipid composition, ROS [27,28], and metabolic wiring, and cancers including PDAC show mesenchymal-like cells can be more ferroptosis-prone under cysteine stress [8,19,27,28,29,30,31,32]. Transcriptional profiling has classified PDAC into epithelial and mesenchymal subtypes with distinct vulnerabilities. These observations raise the possibility that EMT state may serve as a functional determinant of ferroptosis sensitivity, despite the underlying mechanisms remaining incompletely understood [28].

Selenium metabolism adds another layer of ferroptosis regulation [33]. Through selenocysteine incorporation, selenium supports antioxidant selenoproteins that suppress ferroptosis [34,35,36]. Particularly, selenium availability has been observed to inversely correlate with cancer progression, suggesting a protective role in some context [36,37]. GPX4, one of twenty-five human selenoproteins, reduces lipid hydroperoxides using glutathione (GSH) as a cofactor and is a key ferroptosis regulator [34,38]. Inhibition of GPX4 function by either blocking cysteine uptake (erastin) or its direct inhibitor (RSL-3) triggers ferroptosis [38,39].

The plasma selenium transporter Sepp1 delivers selenium to peripheral tissues and influences redox buffering, metabolic remodeling, and clinical outcome in cancers [37,40]. Recent studies have revealed that selenium does not uniformly support GPX4, but instead is differentially partitioned among selenoproteins, including PRDX6 and TrxR enzymes, in ways that can alter ferroptosis sensitivity independent of GPX4 [37,41,42]. These insights highlight the importance of understanding selenium-driven, GPX4-independent antioxidant mechanisms in PDAC. However, the specific contribution of Sepp1 or selenium to PDAC ferroptosis sensitivity remains unclear.

The thioredoxin (Trx) system, comprising thioredoxin (Trx), thioredoxin reductases (TrxR1/TrxR2), and NADPH, is a major parallel antioxidant pathway that detoxifies peroxides independently of the GSH–GPX4 axis [37,43,44]. Both TrxR1 and TrxR2 are frequently elevated in PDAC and other cancers, supporting survival under oxidative stress [44,45,46,47,48]. Being a selenite reducing enzyme as well as a selenol-containing enzyme, TrxR plays a central role in selenium pathophysiology [46,49]. Growing evidence now links thioredoxin reductases to GPX4-independent ferroptosis resistance, and inhibition of TrxR1 sensitizes diverse cancer cells to system xc^−^ blockade, glutathione depletion, or cysteine stress in some tumor cells [44,49,50,51].

Here, we investigate how cysteine dependence, selenium utilization, and the thioredoxin system collectively shape ferroptosis sensitivity across PDAC subtypes. We identify two phenotypically distinct PDAC groups differing in ferroptosis susceptibility, refine the contribution of selenium beyond GPX4, and uncover a previously underappreciated role for the thioredoxin system as a GPX4-independent ferroptosis-defense axis. Together, these findings reveal a targetable metabolic–redox interaction in PDAC and motivate deeper mechanistic and therapeutic exploration.

2. Results

2.1. Pancreatic Cancer Cells Exhibit Differential Cysteine-Dependence

Ferroptosis inducers that block cysteine uptake (erastin) or directly inhibit GPX4 (RSL-3) can eliminate many different types of cancer cells effectively [20,21,52]. To evaluate ferroptosis sensitivity in pancreatic cancer, we examined the response of six PDAC cell lines to erastin treatment. PANC-1 and MIAPaCa-2 cells exhibited marked sensitivity, whereas the epithelial primary tumor lines CAPAN-2 and BXPC-3 and epithelial metastatic lines CFPAC-1 and ASPC-1 were highly resistant (Figure 1A and Figure S1A). Cell death occurred rapidly (within 14–18 h), accompanied by characteristic membrane rupture (Figure S1B). Consistent with loss of membrane integrity, elevated extracellular protease activity was detected in the culture media of sensitive cells (Figure 1B).

To determine whether ferroptosis was driving this differential response, we assessed lipid peroxidation, a hallmark and trigger of ferroptosis. Erastin-induced robust lipid peroxidation exclusively in the eratin-sensitive PANC-1 cells, whereas erastin-resistant Capan-2 cells showed minimal or undetectable lipid peroxidation (Figure 1C,D). In agreement with these phenotypes, Western blot analysis revealed activation of stress-associated death signaling (p-p38) and DNA damage (p-H2AX) only in erastin-sensitive cells (Figure 1E). Similar patterns were observed with RSL3, a GPX4 inhibitor, which likewise produced clear sensitivity differences across PDAC lines (Figure S1C). The erastin-induced cell death and death markers were significantly prevented by the non-specific necrosis inhibitor Necrostatin-1 and the ferroptosis inhibitor Ferrostatin-1, but not by the broad-spectrum caspase inhibitor Q-Vad (Figure 1F,G and Figure S1D). Although the phosphorylation of p38 and γH2AX are not specific markers of ferroptosis, their selective induction in erastin-sensitive cells and suppression by Ferrostatin-1 support their use here as contextual indicators of ferroptosis. Together with robust lipid peroxidation and caspase-independent, Ferrostatin-1-reversible cell death, these data support ferroptosis as the predominant mode of death in the erastin-sensitive cells.

Together, these findings indicate potential inter-tumoral heterogeneity in cysteine dependence and ferroptosis susceptibility among PDAC. This prompted us to define the molecular determinants of this variability and to test whether distinct cellular programs underlie the observed ferroptosis phenotypes.

2.2. Cysteine Dependence Is Associated with the EMT Status of Pancreatic Cancer Cells

Two ferroptosis inducers, erastin and RSL3, were originally identified for their ability to selectively kill oncogenic RAS-mutant cells in vitro [39,53]. Because KRAS mutations are a hallmark of PDAC, we first assessed whether KRAS status influenced ferroptosis sensitivity. Five of six PDAC cell lines harbored oncogenic KRAS mutations (Figure S2A), yet mutation status did not correlate with erastin sensitivity, indicating that factors beyond RAS activation govern cysteine dependence. In contrast, Smad4, an essential mediator of TGF-β signaling and EMT [54,55], was mutated or lost in four erastin-resistant lines but retained in both erastin-sensitive lines (Figure S2A). This pattern suggested a possible association between Smad4 status and ferroptosis sensitivity. However, forced Smad4 expression in resistant cells induced only minor changes in EMT marker levels (Figure S2C) and failed to alter erastin sensitivity (Figure S2B), indicating that Smad4 loss alone does not drive ferroptosis resistance in PDAC.

Previous studies have linked phenotypic cysteine dependence to EMT status in breast cancer [19,27]. To determine whether EMT status similarly correlates ferroptosis sensitivity in PDAC, we performed unsupervised clustering and gene set enrichment analysis (GSEA) using the transcriptomic profiles (GSE21456) of all six PDAC cell lines used in this study. Cluster analysis grouped the cell lines into two distinct subgroups (Figure 2A), which aligned precisely with their erastin sensitivity (Figure 1A). The erastin-sensitive subgroup (PANC-1 and MIAPaCa-2) exhibited strong mesenchymal features, including elevated expression of Vimentin and ZEB1. SMAD9 also showed increased expression in this group, as identified through our unbiased clustering analysis. In contrast, the four erastin-resistant lines displayed epithelial characteristics, such as high E-cadherin, along with increased GATA3 and KLF5, which were also identified from this analysis despite not being classical epithelial markers. GSEA further confirmed enrichment of a mesenchymal gene signature, defined previously in breast cancer, in the erastin-sensitive PDAC subgroup (Figure 2B). Expression of representative EMT markers was validated at both the RNA (Figure 2C) and protein level (Figure 2E). Together, these results indicate that cysteine dependence and ferroptosis sensitivity in PDAC cells are closely associated with EMT status. Importantly, we interpret EMT markers here as phenotypic indicators of ferroptosis vulnerability rather than direct mechanistic drivers.

2.3. Selenium Protects Cells from Erastin-Induced Ferroptosis

From our transcriptomic analysis, two selenoproteins, Sepp1 and SelM, emerged as differentially expressed genes between PDAC subtypes (Figure 2A). These genes displayed nearly opposite expression patterns (Figure 2D). Given that Sepp1 functions as a selenium transport and storage protein, its reduced expression in erastin-sensitive cells could reflect limited selenium availability and altered selenoprotein metabolism in this subgroup.

To determine whether selenium influences ferroptosis sensitivity, we cotreated cells with sodium selenite and assessed their response to erastin. Selenite markedly protected erastin-sensitive PDAC cells from ferroptosis in a dose-dependent manner (Figure 3A,B and Figure S3A,B). Consistent with this protective effect, selenite suppressed erastin-induced stress signaling (p-p38), DNA damage (p-H2AX), lipid peroxidation, and expression of death-associated genes (Figure 3C–E and Figure S3D). In parallel, selenite increased GPX4 protein expression (Figure 3C). Selenite also rescued cells from RSL3-induced death, indicating a broader protective effect against ferroptosis (Figure S3C). Selenite had no observable effect on cell growth or viability in erastin-resistant cells, either treated alone or in combination with erastin.

However, despite increased GPX4 protein levels, intracellular glutathione (GSH) remained severely depleted under erastin plus selenite treatment (Figure 3F). Because GPX4 enzymatic activity strictly depends on the reduced GSH, these findings strongly suggest that GPX4 activity remains functionally impaired under these conditions despite elevated protein level. Although GPX4 activity and redox state were not directly assessed in this study, the depletion of intracellular GSH under these conditions suggests that selenium-mediated protection is largely independent of GPX4 activity.

2.4. Inhibition of the Thioredoxin System but Not Sepp1 Abolishes the Protective Effects of Selenium

Sepp1, a selenium transporter, was highly expressed in erastin-resistant PDAC cells but nearly absent in erastin-sensitive cells (Figure 2D,E). This inverse correlation suggested a potential role for Sepp1 in regulating cysteine dependence or oxidative stress in PDAC. Previous studies report that Sepp1 deficiency elevates ROS levels and promotes stem cell-like properties [56], suggesting the possibility that loss of Sepp1 in erastin-sensitive cells might impair selenium delivery and exacerbate ROS accumulation. To test whether Sepp1 contributes to ferroptosis resistance, we knocked down Sepp1 in erastin-resistant Capan-2 cells, confirming efficient depletion at the RNA expression level (Figure 4A). Unexpectedly, Sepp1 knockdown failed to increase erastin sensitivity, as cell viability and lipid peroxidation remained unchanged relative to controls (Figure 4B). These findings indicate that Sepp1 is not a key determinant of ferroptosis resistance or cysteine dependence in PDAC cells.

We next examined the thioredoxin (Trx) system, which includes the selenoproteins thioredoxin reductase 1 and 2 (TrxR1/TrxR2). The Trx system functions in parallel with the GPX4/GSH pathway to detoxify ROS [57], and selenium supplementation can enhance TrxR activity, suggesting that selenite may also protect cells through TrxR in addition to GPX4. To test this hypothesis, we used a thioredoxin reductase inhibitor D9. Increasing doses of D9 progressively abolished the protective effect of selenite against erastin-induced ferroptosis (Figure 4C,D). Immunoblotting confirmed that D9 restored erastin-induced stress signaling (p-p38) and DNA damage (p-H2AX), both of which were suppressed by selenite alone (Figure 4E). In addition, D9 cotreatment reduced TrxR1 and TrxR2 protein expression, confirming effective inhibition of the Trx system (Figure 4E). We also observed intrinsic variability in D9 sensitivity across PDAC cell lines. Mesenchymal/erastin-sensitive cells were markedly vulnerable to D9, whereas epithelial/erastin-resistant cells exhibited relative resistance (Figure 4F). These data suggest both pathways appear to act synergistically and erastin-resistant cells may rely on a more efficient thioredoxin system to escape cysteine depletion.

Collectively, these results demonstrate that selenium-dependent protection from ferroptosis requires an intact thioredoxin reduction system, not Sepp1.

2.5. Inhibition of the Thioredoxin System Sensitizes Epithelial-Type Pancreatic Tumor Cells to Erastin

As shown above, selenite-mediated protection from ferroptosis requires an intact thioredoxin (Trx) system, suggesting that this pathway may also confer intrinsic erastin resistance in epithelial-type PDAC cells. To test this hypothesis, we treated four epithelial-type/erastin-resistant PDAC cell lines with erastin in combination with the TrxR inhibitors. D9 markedly enhanced erastin-induced cell death in a dose-dependent manner, whereas either D9 or erastin alone produced minimal toxicity (Figure 5A,B and Figure S4A,B). D9 cotreatment also restored erastin-induced death signaling (p-p38), DNA damage (p-H2AX), and expression of death-associated genes, all of which were suppressed when erastin was used alone in resistant cells (Figure 5C and Figure S4C). Similarly, auranofin (Au), another TrxR inhibitor, produced similar sensitizing effects, confirming that TrxR blockade broadly enhances ferroptosis induction in epithelial-type PDAC cells (Figure 5D–F and Figure S4D,E).

Because the pharmacologic studies implicated TrxR activity in ferroptosis resistance, we next dissected the individual contributions of TrxR1 and TrxR2 enzymes, which localize to the cytosol and mitochondria, respectively. We used shRNA-mediated knockdown of each enzyme independently and confirmed efficient depletion in erastin-resistant cells (Figure 6B,D). Single knockdown of either TrxR1 or TrxR2 failed to restore erastin sensitivity, as cell viability and activation of death marker remained unchanged (Figure 6A–D). These results suggest functional redundancy between the cytosolic and mitochondrial thioredoxin pathways.

To determine whether combined inhibition is required to overcome resistance, we generated a double knockdown (dKD) of both TrxR1 and TrxR2 (Figure 6F). Strikingly, dKD rendered erastin-resistant cells highly sensitive to erastin, producing robust cell death (Figure 6E).

Thioredoxin reductase activity and thioredoxin redox state were not directly measured in these experiments; therefore, we interpreted TrxR involvement based on functional perturbation using pharmacologic inhibition and dual TrxR1/TrxR2 suppression. Together, these findings demonstrate that epithelial-type PDAC cells rely on both TrxR1 and TrxR2 to maintain resistance to erastin-induced ferroptosis. Dual inhibition of cytosolic and mitochondrial thioredoxin pathways is therefore required to overcome this resistance and restore ferroptosis vulnerability.

3. Discussion

Ferroptosis has emerged as a metabolic vulnerability in multiple tumor types, including PDAC. The present study identifies two phenotypically distinct PDAC subgroups with differential dependence on cysteine and divergent susceptibility to ferroptosis, demonstrating that ferroptosis sensitivity is not uniformly distributed across pancreatic tumor cells. Our findings build on recent reports underscoring the metabolic heterogeneity of PDAC and extend prior work by linking EMT state, selenium utilization, and thioredoxin reductase function to ferroptosis resistance [12,13,24,48,55].

We first show that mesenchymal-like PDAC cells exhibit pronounced cysteine dependence and rapid ferroptotic death upon system xc^−^ inhibition or direct GPX4 blockade, whereas epithelial-type cells remain largely resistant under the same conditions. These data support an emerging view that EMT status is associated with ferroptosis sensitivity, consistent with observations in breast cancer and other tumors [19,27,31]. Our transcriptomic clustering and GSEA analyses reinforce this association, showing that ferroptosis-sensitive PDAC cells are enriched mesenchymal gene signatures, while ferroptosis-resistant cells predominantly express epithelial markers. This aligns with recent PDAC-focused ferroptosis reviews suggesting rewiring of redox metabolism and lipid composition during EMT may dictate ferroptosis sensitivity [24,27,32]. However, our findings emphasize that EMT status functions primarily as a correlational biomarker rather than a direct mechanistic driver, as modulating Smad4 alone was insufficient to alter ferroptosis phenotypes. Direct manipulation of EMT inducers such as ZEB1, SNAIL, or TWIST, together with functional analysis of EMT-associated lipid remodeling pathways, will be required to determine whether EMT programs actively dictate ferroptosis susceptibility. At present, EMT status is best viewed as a phenotypic biomarker that distinguishes differences in ferroptosis sensitivity rather than serving as a direct mechanistic driver.

Because EMT and TGF-β signaling influence selenium metabolism and redox balance, Sepp1 emerged from our gene expression analysis as a candidate modulator. However, functional testing revealed that Sepp1 is not required for ferroptosis resistance, despite its differential expression between PDAC subtypes. This is an important clarification, as prior studies linked Sepp1 deficiency to increased ROS production and stem-like phenotypes [56], suggesting a potential regulatory role. Our results demonstrate that Sepp1 does not determine cysteine dependence, ferroptosis sensitivity, or erastin response in epithelial-type of PDAC cells, suggesting that selenium-mediated protection is mediated through other components of the antioxidant network. This conclusion is further supported by recent studies demonstrating that selenium availability alone does not determine GPX4 activity; instead, selenium is redistributed among selenoproteins in a cell state-dependent manner [41,42].

We next addressed how selenium supplementation robustly protects cells from ferroptosis. Although selenium increased GPX4 protein expression, intracellular GSH remained severely depleted under erastin plus selenium, rendering GPX4 catalytically inactive. This observation strongly argues that selenium may protect PDAC cells through GPX4-independent mechanisms as well, consistent with emerging reports showing that alternative selenoproteins, including PRDX6 and noncanonical selenium-utilization pathways, can modulate ferroptosis independently of GPX4 [26,33,41,42,58]. Notably, recent work demonstrated that PRDX6 functions as a selenium gatekeeper, determining whether selenium is allocated to GPX4 versus other selenoproteins, and thereby controlling ferroptosis sensitivity under cysteine stress [41,42]. Although we did not directly examine PRDX6 in this study, the selective preservation of thioredoxin reductase function under selenium treatment suggests that epithelial PDAC cells may prioritize selenium toward the thioredoxin axis rather than GPX4. This hypothesis is worth investigating in future.

Our pharmacologic and genetic data identify the thioredoxin system, specifically thioredoxin reductases TrxR1 and TrxR2, as the key selenium-dependent ferroptosis defense mechanism in these cells. TrxRs are selenoproteins whose activity can increase with selenium availability, and the Trx/TrxR axis operates parallel to the GPX4/GSH pathway to detoxify ROS [26,49,51]. Inhibition of TrxR activity using D9 or Auranofin abolished selenium-mediated protection and restored ferroptosis sensitivity, reactivating oxidative stress signaling, DNA damage, and lipid peroxidation. These results are consistent with recent studies showing that TrxR1 inhibition sensitizes multiple cancer types to ferroptosis via GPX4-independent redox collapse [26,51]. Importantly, our identification of dual TrxR1/TrxR2 dependency highlights a cooperative cytosolic–mitochondrial redox axis, suggesting that epithelial-type PDAC cells rely on multi-compartment thioredoxin buffering to withstand cysteine deprivation. This is consistent with newly proposed models in which mitochondrial lipid peroxidation and mitochondrial–cytosolic ROS exchange act as amplifiers of ferroptotic stress [26]. Further biochemical assessment of TrxR enzymatic activity, thioredoxin redox state, and subcellular ROS distribution would offer a more comprehensive understanding of the underlying redox mechanisms.

Importantly, single knockdown of TrxR1 or TrxR2 alone did not restore ferroptosis sensitivity, revealing functional redundancy between the cytosolic and mitochondrial TrxR isoforms. Double knockdown (TrxR1/TrxR2) was required to induce ferroptotic death in response to erastin. Although we did not directly quantify TrxR activity or thioredoxin redox state and recognize that D9 lacks complete selectivity and may exert off target effects, the agreement between pharmacologic inhibition (D9 and auranofin) and dual TrxR1/TrxR2 suppression supports an on-target requirement for the thioredoxin system. This finding highlights that epithelial PDAC cells rely on both cytosolic and mitochondrial thioredoxin pathways to maintain redox homeostasis under cysteine-deprived conditions, underscoring the cooperative requirement for these enzymes in ferroptosis resistance. This redundancy mirrors patterns observed in other cancer contexts in which mitochondria-derived hydroperoxides act as initiating or amplifying signals for ferroptotic cell death, requiring suppression at both the mitochondrial and cytosolic levels to fully restore susceptibility [26,49].

From a translational standpoint, our work suggests that EMT markers and thioredoxin pathway components may serve as functional biomarkers to identify patients for ferroptosis-based therapies, complementing recent analyses of ferroptosis gene signatures in PDAC [15,28]. Combination strategies that co-target cysteine metabolism and thioredoxin reductases may help overcome ferroptosis resistance and exploit subtype-specific metabolic vulnerabilities [11]. Moreover, recent ferroptosis guidelines emphasize the need to therapeutically exploit cofactor limitations, such as GSH depletion and selenium competition, to force cells into redox collapse [16,48]. Our data indicate that selenium’s protective capacity depends on preserving TrxR activity in addition to maintaining GPX4 expression, highlighting thioredoxin reductases as a rational therapeutic vulnerability in epithelial PDAC. However, our experiments in this study were conducted in vitro. Future studies extending these findings to three-dimensional culture systems, patient-derived organoids, and in vivo animal models will be necessary to validate the physiological relevance of the selenium–thioredoxin axis and to define its therapeutic potential in PDAC.

Overall, these data uncover a selenium–thioredoxin redox axis that contributes to ferroptosis resistance in epithelial PDAC cells (Figure 7) and provide a mechanistic framework for developing precision ferroptosis therapies tailored to PDAC subtype, EMT status, and antioxidant pathway usage. Future work integrating selenium-partitioning regulators such as PRDX6, EMT-linked lipidomic remodeling, and redox-cofactor dynamics will further clarify how subsets of PDAC cells sustain resistance to ferroptosis.

4. Materials and Methods

4.1. Cell Culture and Reagents

Pancreatic tumor cell lines and HEK293T cells were obtained from ATCC (Manassas, VA, USA) and cultured according to standard protocols in a humidified incubator (95% humidity, 5% CO_2_, 37 °C). All cell lines were maintained in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 1% penicillin–streptomycin, unless otherwise specified. Cysteine-deficient medium was prepared according to a previously published protocol [19]. Death inhibitors Q-VD-OPH, Necrostatin-1, and Ferrostatin-1, as well as the ferroptosis inducer erastin, were purchased from Cayman Chemical (Ann Arbor, MI, USA). Sodium selenite (selenium source) was obtained from Sigma-Aldrich (St. Louis, MO, USA). Thioredoxin reductase inhibitors D9 and Auranofin were purchased from MedChem Express (South Brunswick, NJ, USA). All compounds were applied either individually or simultaneously as co-treatments, without any pretreatment. A complete list of antibodies (Cell Signaling Technologies, Danvers, MA, USA) used in this study is provided in Supplementary Table S1.

4.2. Lentiviral Cell Infection

Lentiviral particles were produced in HEK293T cells by transfecting lentiviral packaging plasmids together with the respective shRNA constructs using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA, USA). Viral supernatants were collected 48 h after transfection, clarified by centrifugation, and used immediately for infection or stored at −80 °C. Target cells were infected with viral supernatant for 48 h, followed by antibiotic selection with either puromycin or blasticidin (Cayman Chemical) to establish stable knockdown lines. shRNAs targeting Sepp1, TrxR1, and TrxR2 were obtained from Sigma-Aldrich, and the empty pLKO.1 vector served as the infection control.

4.3. Heatmap Cluster Analysis and Pathway Enrichment Analysis

Gene expression profiling data from six pancreatic cancer cell lines were retrieved from the GEO database (accession number GSE21654). Probe intensities were normalized using the Robust Multi-array Average (RMA) algorithm. Gene expression changes were calculated as Δlog_2_ (zero-transformed) values relative to the corresponding reference samples. Probe sets displaying at least a twofold change were selected for hierarchical clustering to visualize differential gene expression patterns across the cell lines. Pathway enrichment analysis was performed using Gene Set Enrichment Analysis (GSEA) with the G2 annotated gene sets and default settings, including 1000 permutations, to identify significantly enriched biological pathways.

4.4. RNA Extraction and Real-Time RT-PCR

Total RNA was isolated using the PureLink™ RNA Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Two micrograms of RNA were reverse-transcribed into cDNA, and quantitative real-time PCR (qRT-PCR) was performed using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). mRNA expression levels were normalized to ActB (β-actin) using the ΔΔCt method. All primer sequences used in this study are provided in Supplementary Table S2.

4.5. Immunoblotting Analysis

Cells were lysed in RIPA buffer (Sigma-Aldrich) supplemented with protease and phosphatase inhibitor cocktails (Thermo Fisher Scientific). Protein concentrations were quantified using the BCA assay, and equal amounts of total protein were resolved by SDS-PAGE and transferred to PVDF membranes. Immunoreactive bands were detected using the ECL Plus chemiluminescence system (GE Healthcare, Piscataway, NJ, USA) and visualized with the LAS-4000 imaging analyzer (GE Healthcare).

4.6. Cell Viability and Cytotoxicity

Cell viability was assessed either by measuring intracellular ATP levels using the CellTiter-Glo Assay Kit (Promega, Madison, WI, USA) or by Crystal Violet staining. Cells were seeded and treated in triplicate for all experiments. Cytotoxicity was evaluated using the CytoTox-Fluor Assay Kit (Promega), which quantifies protease activity released from damaged or membrane-compromised cells. Luminescence and fluorescence signals were recorded using a Synergy LX multimode plate reader (BioTek, Winooski, VM, USA).

4.7. Detection of Cellular Lipid Peroxidation, H2O2, and Glutathione

Lipid peroxidation was measured using the Image-iT™ Lipid Peroxidation Kit (Invitrogen). Cells were seeded in 96-well plates and treated in triplicate. After treatment, the kit reagent was added and incubated for 30 min, followed by replacement with Live Cell Imaging Solution (Thermo Fisher Scientific). Fluorescence was recorded at ~590 nm (oxidized, red) and ~510 nm (reduced, green) using a Synergy LX multimode reader (BioTek). The red/green ratio was calculated and normalized to total cellular protein or ATP content. In addition, fluorescent images were captured using the ZOE™ Fluorescent Cell Imager (Bio-Rad, Hercules, CA, USA).

Intracellular H_2_O_2_ levels were quantified using the ROS-Glo™ H_2_O_2_ Assay Kit (Promega), and glutathione (GSH) levels were measured using the GSH-Glo™ Glutathione Assay Kit (Promega) according to the manufacturer’s instructions. Luminescence signals were collected using a Synergy LX multimode reader (BioTek) and normalized to total protein content.

4.8. Statistical Analysis

All experiments were performed with at least three biological replicates unless otherwise specified, and results were confirmed in a minimum of two independent experiments. Statistical significance between groups was assessed using unpaired two-tailed Student’s t-tests, unless otherwise specified. All analyses were performed using GraphPad Prism 8.0, and p-values < 0.05 were considered statistically significant. Data are presented as mean ± standard deviation (SD) from at least three independent replicates.