How Lowering PRNP Levels May Make Pancreatic Cancer Cells More Sensitive to the Drug Gemcitabine

Single-cell RNA sequencing (scRNA-seq) datasets of pancreatic cancer were downloaded from the Gene Expression Omnibus (GEO) database under the accession numbers GSE186960↗, GSE189753↗, GSE222952↗, and GSE205013↗. These datasets were processed using the Seurat package. High-quality cells were meticulously filtered on the basis of criteria including nFeature > 500, min.cells > 3, and mitochondrial gene expression < 20%. Subsequent data normalization was performed, and the top 2,000 most variable genes were selected for further analysis. To standardize expression levels and mitigate batch effects, the ScaleData function and Harmony algorithm were applied. Classical marker genes were used to annotate distinct cell populations within the dataset. Differential genes were identified using the following criteria: adjusted P value < 0.05, absolute average log2-fold change > 1, and a significant percentage difference between subclusters > 0.1.

Microarray and transcriptome datasets from pancreatic cancer cell lines, bearing the following accession numbers, were harnessed: GSE105083↗, GSE106336↗, GSE140077↗, GSE152121↗, GSE152123↗, GSE153460↗, GSE172303↗, GSE223303↗, GSE228106↗, GSE249302↗, GSE35141↗, GSE78982↗, GSE79953↗, GSE80617↗, and GSE97766↗. To guarantee precise gene evaluation, these bulk datasets were analyzed individually rather than collectively. For the RNA-seq data, specifically SRP303224, quality control was rigorously conducted via FastQC. Read trimming was then executed with Fastp, followed by alignment to the GRCh38 reference genome using HISAT2. Finally, read counting was performed using FeatureCounts. Differential genes in bulk RNA-seq data were identified using the criteria: adjusted P value < 0.05 and absolute log₂-fold change > 0.5.

Using data from The Cancer Genome Atlas (TCGA) on pancreatic cancer, patients were categorized into high-PRNP expression and low-PRNP expression groups on the basis of the median expression level of the PRNP to explore its role in expression. To deduce stromal cell type scores within both the high- and low-expression patient groups, we employed the EPIC, xCell, MCPcounter and estimate packages.

The single-cell spatial transcriptomics dataset, with the accession number GSE235452↗, was processed via the Seurat package. Data normalization and batch effect correction were executed via the SCTransform function. Clustering was then conducted via the FindNeighbors and FindClusters functions, culminating in cell type scoring, which was accomplished via the AddModuleScore function.

Proteomic data were retrieved from the ProteomeXchange database (proteomexchange.org) under the accession number PXD030861. For proteomic identification and quantification, MaxQuant (version 2.4.2) was used with default settings, referencing the human Swiss-Prot protein database.

Pseudotime trajectories for single cells were analyzed via the Monocle2 package. The functions newCellDataSet, estimateSizeFactors, and estimateDispersions were leveraged to carry out these analyses. The detectGenes function was utilized to filter out low-quality cells, applying a min_expr threshold of 0.1.

The protein–protein interaction (PPI) network for PRNP was constructed via the GeneMANIA database, complemented by enrichment analysis. To delve into the molecules associated with PRNP, the ClusterProfiler package was used for comprehensive Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Set Enrichment Analysis (GSEA). Furthermore, Gene Set Variation Analysis (GSVA) was implemented through the dedicated GSVA package, while visualizations were refined via the Gseavis package for enhanced clarity.

Three machine learning algorithms LASSO, SVM-RFE and RF were employed to explore disease states. The LASSO algorithm was utilized for variable selection and complexity regularization. SVM-RFE was implemented to identify the most relevant key genes through recursive feature elimination. For the RF algorithm, its principle involved deriving reliable results from a large ensemble of underlying tree models. Finally, the optimal key genes were determined by intersecting the common features identified across all three algorithms. This integrative approach ensured the selection of robust biomarkers with cross-method validation. The predictive value of core genes is assessed by ROC. This allows the prediction of core genes as biomarkers as well as diagnostic capabilities.

The diagnostic potential of candidate genes in pancreatic adenocarcinoma was validated using the GEPIA2 platform (http://gepia2.cancer-pku.cn/↗), which integrates standardized RNA-seq data from TCGA-PAAD (179 tumors) and GTEx-Pancreas (171 normals).

Gemcitabine sensitivity in pancreatic cancer was investigated by leveraging the Genomics of Drug Sensitivity in Cancer (GDSC) database to estimate individual patient responses. The R package oncoPredict was subsequently used to calculate Gemcitabine sensitivity scores. Concurrently, the Tumor Immune Dysfunction and Exclusion (TIDE) algorithm was applied to predict immunotherapy response and evaluate immune escape potential.

Gemcitabin (Macklin) has a purity of > 95.0%. The human pancreatic cancer cell lines PANC-1 and ASPC-1 were kindly provided by Professor Yongsu Zhen. The cell lines were maintained in a 37 °C incubator under 5% CO₂ conditions. While ASPC-1 cells thrived in RPMI-1640 medium supplemented with 10% FBS, PANC-1 cells exhibited optimal growth in DMEM under the same serum conditions. The cells were seeded at a density of 5×10⁵ in 6-well dishes and transfected with 100 pmol specific Small interfering RNAs (siRNAs) for 48h in the presence of 4 μL Lipo8000 (Beyotime) transfection reagent. siRNAs were purchased from GenePharma. The following antibodies were used: E-cadherin(Huabio); Slug(Wanleibio); Snail(Wanleibio); PRNP(UpingBio); P53(Huabio); BAX(Huabio); Bcl2(Huabio); β-actin(ABclonal); SLC7A11(Biodragon); GAPDH(ABclonal); GPX4(Huabio). Ferroptosis inhibitors were purchased from Cayman.

Total RNA was isolated using the Zsgentech 6-min Rapid RNA Extraction Kit and reverse-transcribed to cDNA. Relative mRNA quantification was performed via the 2^-ΔΔCt method with GAPDH as the endogenous control. The primers used are shown in Table 1. The siRNA sequences are shown in Table 2.

Pancreatic cancer cells, ranging from 3000 to 4000 per well, were seeded in 96-well plates and cultured for 24 hours. They were then treated with GEM in serum-free media for 24–48 hours. Following treatment, MTT was added, and the cells were incubated for an additional 4 hours. DMSO was subsequently added to dissolve the formed crystals, and the absorbance was measured at 570 nm. The IC50 values were calculated via GraphPad 9.0.

Cell viability was evaluated via the trypan blue exclusion method. The cells were plated in 6-well plates at a density of 5×10⁵ cells per well in serum-free medium supplemented with GEM and incubated at 37°C with 5% CO₂ for 48 hours. All the cells from each well were subsequently centrifuged into a tube and stained with a 0.5% trypan blue solution for 3 minutes. The stained cells were observed under an inverted microscope, and viability was determined by counting the viable cells.

To assess the cell proliferation capacity via the clonogenic survival assay, pretreated cells were plated in 6-well plates. Once adhered, the cells were categorized into six groups on the basis of experimental requirements: Control (C), GEM (G), siPRNP (SI), siPRNP+GEM (SI+G), Ferrostatin-1(Fer-1) and siPRNP+Ferrostatin-1(SI+Fer-1). Each group included at least three replicate wells. The cells were incubated until most of the cell clusters comprised approximately 50 cells. The samples were then fixed with paraformaldehyde and stained with crystal violet for 20 minutes. The formed cell colonies were observed under an inverted microscope.

Cells in optimal growth conditions were seeded into 6-well plates. Upon adherence, a scratch was introduced into the monolayer via a 100 µl pipette tip. Following a rinse with PBS, the cells were treated according to the experimental protocol, with each group featuring at least three replicate wells. Culture medium was added, and the scratch area was monitored at various time points via an inverted microscope.

For the invasion assay, 100 µL of Matrigel was added to the chamber. A complete medium containing 20% FBS was placed in the lower chamber of a 24-well plates, and 100 µL of the treated cell suspension was evenly distributed into the upper chamber. After the specified incubation period, the medium was discarded, and the cells were gently rinsed twice with PBS. The cells were subsequently fixed with 4% paraformaldehyde for 10 minutes and stained with crystal violet for another 10 minutes. Nonmigrated cells on the upper surface were carefully removed with a wet cotton swab, and the cells were observed and photographed under an inverted microscope.

For the migration assay, a similar procedure was used. Specifically, 500 µL of complete medium containing 20% FBS was added to the lower chamber of a 24-well plates, and 100 µL of the treated cell suspension was evenly placed into the upper chamber. The subsequent steps mirrored those of the Transwell invasion assay.

JC-1 solution was added to the 6-well plates at a concentration of 10 μM, followed by incubation in the dark at 37°C for 15 minutes. During this period, the plate was gently shaken every 4 min to prevent dye agglomeration. After incubation, the dye was aspirated, and the cells were thoroughly washed twice with PBS to eliminate any unbound JC-1 dye. The cells were then examined and photographed under an inverted fluorescence microscope.

For Reactive Oxygen Species (ROS) detection, the instructions provided by the ROS detection kit from Report Biotech were followed meticulously. Specifically, DCFH-DA reagent (10 mM) was diluted 1000 times with serum-free medium to prepare the working solution. The supernatant from the 6-well plates was aspirated and discarded, and 1 mL of the DCFH-DA working solution was added to each well. After staining, the supernatant was removed, and the cells were rinsed three times with serum-free cell culture medium to ensure the complete removal of unbound DCFH-DA. The cells were subsequently observed and photographed under an inverted fluorescence microscope. Additionally, flow cytometry was employed to quantify the fluorescence intensity before and after transfection or drug stimulation by selecting the appropriate fluorescence channel, with an excitation wavelength of 488 nm and an emission wavelength of 525 nm.

Pancreatic cancer cells in the logarithmic growth phase were plated into 6-well plates. Upon adherence and reaching a growth density exceeding 80%, the medium was exchanged with medium containing the drug for a 48-hour treatment period. Following treatment, the cells were harvested, and detection reagents, prepared according to the instructions of the GSH level detection kit, were added. The absorbance of the samples at a wavelength of 412 nm was then measured via a microplate reader. The relative GSH levels of each group were subsequently calculated, statistically analyzed, and graphically represented.

Pancreatic cancer cells were lysed in an ice bath environment, and the control solution, standard and sample were added to the centrifuge tube according to the standard and working solution configured in the MDA kit (Boxbio) instructions, and then the working solution was added. After mixing, heat at 100 °C for 15 mins. After cooling to room temperature, centrifuge, the supernatant was added to a 96-well plates, and the absorbance of each well (532 nm) was measured by a microplate reader to calculate the MDA concentration.

Acridine orange (AO) and ethidium bromide (EB) were combined in a 1:1 ratio to create the working solution. This solution was dispensed into each well, and the cells were subsequently incubated at 37°C for 15 minutes. Following incubation, alterations in nuclear morphology were observed under a fluorescence microscope.

Pancreatic cancer cells with good growth status were inoculated into 6-well plates, and after the cells were treated according to the experimental requirements, the cells were collected, the proteins in the cells were extracted, and then the protein expression was detected. The steps include: reagent preparation, protein sample preparation, determination of protein content, SDS-PAGE electrophoresis, membrane transfer, immunoreaction, chemical reflectance development and fixation, and gel image analysis.

Pancreatic cancer cells were cultivated to the logarithmic growth phase and subsequently treated with drugs for 48 hours, while a negative control was established. The cells were then washed twice with the prepared PBS buffer and resuspended in 1× Buffer to a density of 1×10⁶/mL. Using a flow cytometry kit, the resuspended cells were stained with 5 μL of each Annexin V-FITC solution and PI solution for 15 minutes at room temperature. After staining, the cells were washed with PBS buffer and analyzed for apoptosis via flow cytometry. The results were further analyzed with FlowJo software.

In this study, we utilized the R language (version 4.4.0) for analysis. Each experiment was independently replicated three times or more, and the results are presented as the means ± standard deviations. A t test was used for comparisons between two groups, whereas one-way ANOVA was used for comparisons among multiple groups. The statistical analysis and plotting of group values were conducted via ImageJ and GraphPad Prism 9.0 software. p < 0.05 indicated that the difference was statistically significant.

The PANC-1 cell line is widely used in pancreatic cancer research; thus, we prioritized PANC-1 data for our investigations. By analyzing scRNA-seq data from PANC-1 cells treated with gemcitabine for 24 hours, we generated a uniform manifold approximation and projection (UMAP) plot to visualize cellular distribution (Figure 1A). At the single-cell resolution, we excluded PANC-1 cells that exhibited no response to gemcitabine. Through GSVA enrichment analysis, clusters 5 and 8 were associated with gemcitabine stimulation, highlighting the upregulation of signaling pathways, including the TGF-β, WNT, and ECM receptor pathways. This upregulation facilitated the fibrotic transformation of cancer cells, thereby contributing to gemcitabine resistance (Figure 1B). To delve deeper into the genetic alterations within these two subclusters, we conducted differential analysis between subclusters 2 and 7 from the control group and subclusters 5 and 8. Stringent criteria were applied—adjusted P value < 0.05, absolute average log2-fold change > 1, and a significant percentage difference between subclusters > 0.1. We identified 14 downregulated genes and 737 upregulated genes (Figure 1C). Furthermore, via proteomics technology, we identified 39 proteins that demonstrated remarkable stability in gemcitabine-resistant PANC-1 cells (Figure 1D).

Furthermore, to identify genes inherently resistant to gemcitabine-induced cell death, we integrated two gene expression datasets, GSE80617↗ and GSE153460↗, via the SVA package and conducted three rigorous rounds of differential analysis. By applying stringent criteria of an adjusted P value < 0.05 and an absolute log2-fold change > 0.5, we identified 39 genes that were upregulated upon drug stimulation and played pivotal roles in regulating the emergence of drug-resistant phenotypes (Figure 1E). These genes included ABL2, AIDA, AMOTL2, AURKB, BIRC2, CDC42EP1, DDAH1, DIAPH3, FGF2, FHL2, GPRC5A, HDAC9, HK2, IFRD1, MICAL3, NFKBIE, OSBPL6, OXCT1, POLA2, PRNP, PRTFDC1, RAB32, RAD54B, RELB, RFTN1, RRM1, RRM2, RRM2B, SAMD4A, SBDS, SLC1A4, SLC39A14, SLC7A1, SNX24, SOAT1, SYNE1, SYT1, TAX1BP3 and UAP1.

To evaluate the adaptability of a gene set across diverse pancreatic cancer cell lines and effectively filter out false positives, transcriptome and microarray datasets encompassing both GEM-treated groups and GEM-resistant pancreatic cancer cell lines were meticulously selected. The expression data for the 39 genes were extracted from each dataset, normalized, and subsequently integrated into a heatmap for comprehensive visualization (Figure 1F).

To systematically identify clinically relevant genes associated with gemcitabine response in pancreatic cancer, we implemented a multi-algorithm integration approach starting with 39 candidate genes. The LASSO regression model identified 26 prognostic genes, while the random forest algorithm selected 9 key features. Support vector machine recursive feature elimination (SVM-RFE) refined the list to 6 core genes (Figures 1G–I). Venn diagram analysis of the three algorithm outputs revealed FHL2, PRNP, and RRM as consistently selected candidates (Figure 2A). These genes exhibited predictive value for gemcitabine sensitivity, with ROC analysis showing area under the curve (AUC) values > 0.7 (Figure 2B). Validation using the TCGA-PAAD cohort (n=178) confirmed significant tumor-specific overexpression compared to normal controls (p < 0.05) (Figure 2C). Survival analysis revealed that elevated expression of all three genes correlated with poorer overall survival (OS) and disease-free survival (DFS) (p < 0.05) (Figure 2D). Notably, PRNP showed the strongest positive correlation with RRM1 (Pearson’s R = 0.37, p < 0.05), supporting its prioritization as a potential therapeutic target for overcoming gemcitabine resistance in pancreatic cancer (Figure 2E).

Next, KEGG and GO enrichment analyses were performed on the selected gene set to pinpoint PRNP-associated genes and signaling pathways. The results revealed that the PRNP is intricately linked with multiple pivotal pathways, including ferroptosis, neurodegenerative diseases, actin filament binding, and microtubule binding signaling pathways. These findings underscore the strong correlation between PRNP expression and vital biological processes such as cell proliferation, cell migration, and cell apoptosis (Figure 2F).

Using transcriptome data from the TCGA-PAAD cohort, we stratified patients into high-PRNP and low-PRNP expression groups based on the median PRNP expression level. This stratification allowed us to systematically analyze the relationship between PRNP expression and both drug sensitivity and immune cell infiltration.

Drug sensitivity analysis using the oncoPredict algorithm revealed that patients with high PRNP expression exhibited significant resistance to gemcitabine (Figure 3A). Furthermore, we assessed the correlation between PRNP expression and TIDE, a predictor of immunotherapy response. The High group demonstrated a significantly elevated TIDE score compared to the Low group. Consequently, the proportion of patients predicted as non-responders to immunotherapy was substantially higher in the High group (80.3%) than in the Low group (54.2%). These TIDE results suggest that high PRNP expression promotes an immunosuppressive tumor microenvironment conducive to immune escape (Figures 3B, C).

To further elucidate the association between PRNP and the tumor immune microenvironment, we analyzed the composition of infiltrating immune cell populations. Using the xCell algorithm to estimate relative abundance, we found that the High group was significantly enriched for immunosuppressive cells, including Th2 cells, iDC, B cells, Tregs, and M2 macrophages. Conversely, the Low group exhibited higher scores for immune-active populations such as CD4+Tcm, NKT cells, and Th1 cells. These results indicate that high PRNP expression is associated with an immunosuppressive microenvironment, whereas low PRNP expression correlates with a profile suggestive of cellular immunity and immune activation (Figure 3D).

To systematically investigate the regulatory role of PRNP in the tumor immune microenvironment, we constructed correlation heatmaps of immune cell interactions. Analysis revealed a more complex and enhanced cooperative network among various immune cells in the High group compared to the Low group. Critically, a strong and concentrated positive correlation was observed between immunosuppressive cells, notably Tregs and M2 macrophages, suggesting a coordinated immunosuppressive network under high PRNP conditions. Concurrently, association analysis between the gemcitabine resistance score and the immune microenvironment revealed distinct infiltration patterns. In the low-score group, the resistance score correlated positively only with Th2 cells, indicating a limited immune shift in early resistance. In stark contrast, the high-score group showed significant negative correlations with cytotoxic/effector populations like NKT cells, CD4+Tem, and CD8+ naive T cells. This aligns with our xCell findings, jointly indicating that gemcitabine resistance is associated with a loss of cytotoxic potential and impaired recruitment of immune cell reserves.

Furthermore, correlating the TIDE score with immune cell infiltration elucidated the immunological basis of its predictive power. In the Low group, the score positively correlated with Monocytes, CD8+ naive T, Pro-B, and CD4+ T cells, but negatively with CD4+ Tem, suggesting initial immune recruitment alongside early effector cell suppression. Conversely, the High group showed positive correlations with B cells, Pro-B cells, and CD4+ naive T cells, while maintaining a strong negative correlation with CD4+ Tem. This pattern reflects a profoundly dysregulated microenvironment, explaining from a cellular perspective why a high TIDE score predicts poor immunotherapy response and robust immune escape (Figures 3E, F).

To investigate the potential role of PRNP in promoting CAF accumulation, we applied four distinct deconvolution algorithms. These analyses consistently revealed a significant positive correlation between PRNP expression and CAF abundance, suggesting that PRNP may drive the formation of a CAF-rich tumor microenvironment. Additionally, the level of PRNP expression influenced the expression profiles of various genes within the genes, notably RRM1 gene expression was increased in the PRNP high group, while FHL2 gene expression was not significantly correlated. Finally, GSVA and GSEA analyses demonstrated that high PRNP expression significantly increased the expression of genes involved in epithelial–mesenchymal transition related pathways, such as the TGF-β signaling pathway and the ECM receptor interaction pathway (p < 0.05). These findings indicate that PRNP may play a pivotal role in promoting EMT in pancreatic cancer patients and is closely associated with drug resistance (Figures 4A, B).

In the single-cell model dataset GSE189753↗, which focused on the impact of gemcitabine on mouse pancreatic cancer, particular emphasis was placed on the expression patterns and regulatory mechanisms of Prnp. On the basis of classification criteria drawn from the literature, eight distinct cell clusters were identified and named according to their primary cellular markers: Monocytes/Macrophages, CAFs, T cells, B cells, Neutrophils, Dendritic cells, Mast cells, and Cancer cells. (Supplementary Figure S1) Notably, Prnp gene expression was predominantly observed in the CAF subpopulation, with statistical significance (p < 0.05). Further subdivision of the CAFs subpopulation revealed that gemcitabine treatment led to an increase in the proportion of CAF2 cell subpopulations in the mouse model (Figures 4C, D). CAF2 cells are intimately associated with gemcitabine-induced fibrosis and fall within the inflammatory CAF (iCAF) subtype.

To explore the temporal dynamics of Prnp expression during CAF development, we performed cell trajectory analysis using Monocle2. The results indicated that Prnp facilitated the generation of CAF2 subpopulations (Figures 4E–H). KEGG pathway enrichment analysis showed that gemcitabine not only promotes the formation of CAF2 subpopulations but also induces EMT and upregulates pathways associated with ferroptosis resistance (Figures 4I, J). Collectively, these findings suggest that gemcitabine promotes the generation of pancreatic cancer-associated fibroblast subpopulations, particularly CAF2, which are closely correlated with Prnp expression.

To validate PRNP expression in chemotherapy-treated patients, we examined genes expression within the PRJNA694728 gemcitabine-resistant pancreatic cancer patient-derived xenograft (PDX) model transcriptome dataset. The results demonstrated that the genes presented elevated expression in resistant samples following gemcitabine treatment. Notably, PRNP expression increased in some sensitive patients after chemotherapy (Figure 5A).

To further explore PRNP expression during chemotherapy, we analyzed the GSE205013↗ scRNA-seq dataset of chemotherapy-treated patients (Supplementary Figures S2C–E), which included samples from 27 patients. Among them, 7 patients had undergone chemotherapy (4 on the FOLFIRINOX regimen and 3 on the gemcitabine/Abraxane regimen) prior to tissue collection, while the remaining 20 patients were untreated at the time of specimen collection. By isolating CAF subpopulations from single-cell data, we confirmed that the PRNP was highly expressed in both iCAFs and Schwann cells post-chemotherapy (p < 0.05). Additionally, GSVA analysis revealed significant upregulation of EMT-related pathways in iCAF subpopulations, further confirming that the PRNP promotes EMT in chemotherapy-treated populations. Additionally, PRNP expression was increased in Schwann cells after chemotherapy (Figures 5E–G). Similarly, in the gemcitabine chemotherapy GSE222952↗ dataset (Supplementary Figures S2A, B), PRNP expression tended to increase in CAF subpopulations after GEM treatment (Figures 5B–D).

In summary, PRNP gene expression is elevated in CAF subpopulations of pancreatic cancer patients post-chemotherapy, fostering the emergence of EMT-related genes and drug resistance mechanisms. Based on previous findings, we confirmed that gemcitabine-induced PRNP upregulation facilitates the generation of iCAFs in pancreatic cancer. To delve deeper into the expression patterns of PRNP in pancreatic cancer samples and the primary cell types that impact them, we conducted an analysis utilizing spatial transcriptomics technology. By integrating data from seven samples, we assessed cell types and identified various populations, including tumor cells, cancer-associated fibroblasts, immune cells, Schwann cells, endothelial cells, and pericytes. Notably, PRNP expression was particularly prominent in CAFs, with a strong emphasis on the iCAF subpopulation, as per our previous research. In sample GSM7498817↗, we observed that iCAFs encircled Schwann cells, suggesting a potential synergistic interaction between these two subpopulations, which may collectively expedite the progression of pancreatic ductal adenocarcinoma (Figure 5H, Supplementary Figure S3).

Bioinformatics analysis revealed that gemcitabine treatment of pancreatic cancer cells increases PRNP expression while inducing EMT. To validate this finding, we conducted in vitro cellular experiments. We selected the human pancreatic cancer cell lines ASPC-1 and PANC-1 as our study subjects. The MTT method was used to evaluate the impact of gemcitabine on the survival rates of these two pancreatic cancer cell lines (Figure 6A). Pancreatic cancer cells were exposed to different concentrations of gemcitabine for varying durations. Compared to the control group, gemcitabine significantly inhibited cell growth in a concentration- and time-dependent manner. After 48 hours of treatment, the half-maximal inhibitory concentrations (IC50) for PANC-1 and ASPC-1 cells were determined to be 48.79 μM and 8.69 μM, respectively. These IC50 values served as the benchmark for guiding subsequent experimental steps.

qRT–PCR and Western blot were employed to confirm that GEM enhances PRNP gene expression in pancreatic cancer cells (p < 0.05) (Figures 6D, E). To delve deeper into the function of PRNP, we designed three non-overlapping siRNA sequences targeting PRNP using Gemma. Through qRT–PCR screening, si1031 emerged as the most effective siRNA for knocking down PRNP in ASPC-1 and PANC-1 cells (p < 0.05). Thus, si1031 was used to establish PRNP silenced ASPC-1 and PANC-1 cell lines for subsequent experiments (Figure 6B).

ASPC-1 and PANC-1 cells were categorized into a negative control group (NC) and a PRNP gene knockdown group (siPRNP). After these two groups were exposed to gemcitabine for 48 hours, the results indicated that, at an equivalent drug concentration, the survival rate of the cells in the siPRNP group was markedly lower than that in the NC group. Additionally, the half-maximal inhibitory concentration values for PANC-1 and ASPC-1 cells in the siPRNP group were 27.69 μM and 0.79 μM, respectively, which were significantly lower than those of cells without PRNP knockdown (Figure 6C).

Cell colony formation and trypan blue staining experiments were conducted, and the silencing of PRNP enhanced the toxic effect of gemcitabine on pancreatic cancer cells (Figures 6F–H). To further explore the relationships between the PRNP and biological processes such as cell apoptosis, migration, invasion, autophagy, and ferroptosis, we selected a series of commonly related markers, including P53, BAX, and BCL2 (related to cell apoptosis); GPX4 and SCL7A11 (related to ferroptosis); and Snail and Slug (related to migration and invasion). Construction of a gene interaction network using GeneMANIA revealed direct associations between PRNP and these markers (Figure 6I), suggesting that PRNP may directly regulate the occurrence and development of these biological pathways.

A wound-healing assay was performed to evaluate cell migration ability. Compared with those in the control group and the GEM group, the cell migration rates in both the siPRNP and the siPRNP+GEM groups gradually decreased (Figures 7A, B). Furthermore, the cell migration ability of the siPRNP+GEM group was notably lower than that of the GEM group (p < 0.05). These findings indicate that the knockdown of PRNP can potentiate the inhibitory effect of gemcitabine on the migration of pancreatic cancer cells.

Transwell migration assay to confirm alterations in cell migration ability, we discovered that the number of cells that crossed the Transwell chamber in the GEM group was markedly lower than that in the control group, indicating the inhibitory effect of gemcitabine on pancreatic cancer cell migration. Furthermore, the number of cells that passed through the chamber in the siPRNP+GEM group was even greater than that in the GEM group, suggesting that PRNP knockdown augments the inhibitory effect of gemcitabine on the migration of human pancreatic cancer cells (p < 0.05). The results of the Transwell invasion assay aligned with those of the migration assay (p < 0.05), demonstrating that silencing PRNP enhances the inhibitory effect of gemcitabine on pancreatic cancer cell invasion (Figures 7C–F).

After 48 hours of gemcitabine treatment, we observed an increasing trend in the protein expression of E-cadherin, whereas the protein expression of Snail was increased and the protein expression of Slug was decreased (p < 0.05). After the PRNP gene was silenced, the promotion of E-cadherin protein expression by GEM and the decrease in the protein expression of Snail and Slug were significantly increased (p < 0.05) (Figures 7G, H). Taken together, the results of the wound-healing and Transwell assays indicate that GEM affects the EMT pathway to inhibit the migration and invasion abilities of pancreatic cancer cells and that silencing the PRNP gene synergistically inhibits the EMT process in pancreatic cancer cells with GEM.

Ferroptosis, a form of programmed cell death reliant on iron and ROS, plays a crucial role in tumorigenesis and progression and is tightly linked to drug resistance in cancer. To explore the role of PRNP in ferroptosis, we employed the DCFH-DA ROS fluorescent probe to monitor ROS production in pancreatic cancer cells after GEM treatment. Upon silencing the PRNP gene, GEM markedly augmented ROS generation in pancreatic cancer cells (Figure 8A).

The JC-1 fluorescent probe is a key indicator of ferroptosis, as ferroptosis is associated with specific changes in mitochondrial membrane potential. A decrease in mitochondrial membrane potential causes the JC-1 probe to shift from red to green fluorescence. We observed that, upon silencing the PRNP gene, GEM significantly accelerated the reduction in the mitochondrial membrane potential in pancreatic cancer cells (Figure 8B), suggesting that silencing PRNP facilitates ferroptosis induction.

From a molecular standpoint, upon the introduction of GEM, the expression levels of the key ferroptosis-related proteins SLC7A11 and GPX4 increase, conferring resistance to ferroptosis. However, when the PRNP was silenced, the expression of these two proteins was suppressed, thereby promoting the induction of ferroptosis (p < 0.05) (Figures 8C, D). Further exploration of alterations in the content of glutathione (GSH) and malondialdehyde (MDA), revealed that PRNP gene silencing led to significant depletion of intracellular GSH and MDA levels rise (p < 0.05). With GSH depletion and MDA rise, cells become more susceptible to ROS-induced apoptosis (Figures 8E, F). Flow cytometry-based ROS detection further confirmed a marked increase in intracellular ROS levels following PRNP silencing (Figure 8G). These findings indicate that PRNP silencing enhances gemcitabine-induced ferroptosis in pancreatic cancer cells.

To determine whether PRNP deficiency-induced ferroptosis activation specifically relies on the downregulation of SLC7A11 and GPX4 protein levels, this study conducted ferroptosis inhibitor rescue experiments in PRNP knockdown pancreatic cancer cells.

MTT analysis demonstrated that adding 1 μM Ferrostatin-1 to pancreatic cancer cells with silenced PRNP gene significantly suppressed the cell death process, initially suggesting an intrinsic link between Ferrostatin-1 and PRNP gene expression regulation (Figure 9A). Western blot analysis further revealed that, compared to PRNP silenced pancreatic cancer cells, the PRNP protein expression level in Ferrostatin-1 - treated silenced cells exhibited an upward trend, providing robust protein level evidence for the relationship between Ferrostatin-1 and PRNP gene expression (Figures 9B, C).

To explore whether Ferrostatin-1 and PRNP in silenced pancreatic cancer cells are involved in the ferroptosis pathway, we assessed key ferroptosis related indicators. The results showed that after Ferrostatin-1 treatment, intracellular GSH levels significantly increased, indicating enhanced cellular antioxidant capacity and alleviated oxidative stress - induced damage. Concurrently, MDA levels were markedly inhibited, suggesting reduced intracellular lipid peroxidation and blocked lipid peroxidation chain reactions (p < 0.05) (Figures 9D, E). Additionally, ROS fluorescent probe detection showed a decrease in intracellular ROS fluorescence intensity, indicating reduced ROS levels (Figure 9F). Collectively, these findings imply that Ferrostatin-1 may inhibit ferroptosis by regulating cellular redox balance, and this process may be associated with PRNP.

Simultaneously, depth exploration was conducted into the molecular mechanism underlying ferroptosis. Western blot analysis that treatment with ferroptosis inhibitors exerted a significant reversal effect on the down - regulation of SLC7A11 and GPX4 protein expression induced by PRNP knockdown (Figures 9G, H). Specifically, following the administration of ferroptosis inhibitors, the previously diminished levels of SLC7A11 and GPX4 proteins, which had been suppressed due to PRNP silencing, were restored to a notable extent. This finding provides crucial insights into the interplay between PRNP, ferroptosis inhibitors, and the key ferroptosis - related proteins SLC7A11 and GPX4.

Integrating the results of PRNP silencing and pharmacological interventions, we conclude that PRNP exerts a critical ferroptosis-suppressing role in pancreatic cancer cells by positively regulating SLC7A11 and GPX4 protein expression. Notably, PRNP silenced pancreatic cancer cells exhibit significantly enhanced ferroptosis responses when exposed to gemcitabine, suggesting that PRNP dysfunction may sensitize pancreatic cancer cells to gemcitabine-induced ferroptosis.

To investigate the synergistic apoptotic effect of gemcitabine and PRNP silencing in pancreatic cancer cells, we first used MTT assays and colony formation analysis to quantify cell viability and clonogenic survival. Subsequent AO/EB fluorescence staining revealed that gemcitabine alone induces apoptosis, and PRNP silencing also exerts a pro-apoptotic effect. Importantly, the combination of gemcitabine and siPRNP produced a synergistic apoptotic response, indicating that PRNP silencing enhances gemcitabine-induced apoptosis in pancreatic cancer cells (Figures 10A, B).

Perform Western blot analysis to evaluate the expression levels of key apoptosis regulatory factors P53, BCL2, and BAX. In pancreatic cancer cells treated with GEM, we observed that compared with the control group, P53 and BAX protein expressions were significantly up-regulated, while BCL2 protein levels were down regulated (p < 0.05). When PRNP is silenced alone, BCL2 protein levels are downregulated (p < 0.05), while increasing BAX levels does not affect TP53. It is worth noting that compared with the single treatment group, the combination therapy of siPRNP and GEM showed a synergistic effect, leading to an increase in P53 and BAX expression, while BCL2 protein expression was inhibited (p < 0.05). These protein expression patterns collectively indicate that PRNP silencing enhances GEM induced cell apoptosis (Figures 10C, D).

Finally, flow cytometry-based apoptosis detection revealed a significant increase in the proportion of apoptotic cells in the siPRNP+GEM group (p < 0.05), further confirming that the combination of GEM and PRNP silencing promoted apoptosis (Figures 10E, F).

The dataset used in this study can be found in an online repository. All datasets are publicly available.The name of the repository and the access number can be found in the methodology section of the article.