Abstract
Background
Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal cancer, with drug resistance significantly impeding effective chemotherapy. The clinical importance of key genes in PDAC chemoresistance remains unclear.
Methods
Cytotoxicity assays were conducted on eight PDAC cell lines treated with gemcitabine, albumin paclitaxel, irinotecan, 5-FU, and cis-platinum. Based on GR50 values, cell lines were categorized as chemosensitive (CS) or chemoresistant (CR). RNA-seq data from the Cancer Cell Line Encyclopedia were analyzed for differential gene expression. Various statistical methods, including univariate Cox, LASSO, random forest, and multivariate Cox regression, were employed to construct a risk signature. This model’s effectiveness was validated using time-dependent ROC curves and Kaplan–Meier survival analysis in TCGA and GSE57495 + GSE28735 combined datasets. Downstream molecules of core genes were explored, and the core gene’s function was validated through in vivo and in vitro experiments.
Results
Drug resistance profiles were established based on GR50 values. EDC4 and USP20 were identified as key genes in the risk signature. qRT-PCR validated core gene expressions in PDAC cell lines. EDC4 was selected for further analysis due to its highest correlation coefficient in the multivariate Cox regression model. EDC4 knockdown increased proliferation and chemoresistance in MIA PaCa-2 cells, while overexpression inhibited these traits in AsPC-1 cells. MATN3 and SGCE were identified as downstream targets of EDC4. Immunohistochemistry of tissue microarrays confirmed that low EDC4 levels were associated with poor prognosis in PDAC, highlighting its potential as a therapeutic target.
Conclusion
PDAC cell lines exhibit distinct chemoresistance capabilities. EDC4 and its downstream targets MATN3 and SGCE play significant roles in PDAC multidrug chemoresistance, providing novel insights for treatment.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12935-025-04000-9.
Introduction
Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy of the digestive tract, accounting for approximately 4.7% of all cancer-related deaths. It is projected to become the third leading cause of cancer mortality by 2025 [1]. At diagnosis, around 80–85% of patients present with either unresectable tumors or distant metastases [2, 3]. For those with resectable disease, adjuvant chemotherapy is the standard treatment following surgery [4]. Regimens such as FOLFIRINOX (5-fluorouracil/leucovorin with oxaliplatin and irinotecan) and gemcitabine combined with albumin-bound paclitaxel have improved outcomes in advanced PDAC [5, 6]. However, the emergence of chemoresistance remains a major obstacle to long-term clinical benefit.
The limited efficacy of chemotherapy in PDAC is largely due to its high toxicity and the inherent resistance of tumor cells. Drug resistance is the leading cause of treatment failure [7]. Traditionally, drug sensitivity is measured using IC50 values (the concentration required to inhibit cell growth by 50%). However, IC50 is heavily influenced by the intrinsic proliferation rate of different cell lines. GR50 (Growth Rate 50), an alternative metric that adjusts for cell division rate, offers a more accurate assessment of drug efficacy by quantifying growth inhibition independent of baseline proliferation [8–10]. In this study, we employed GR50 to evaluate drug responses, ensuring comparable assessments across both fast- and slow-growing PDAC cell lines.
Our aim was to systematically characterize the chemosensitivity of eight PDAC cell lines to five commonly used chemotherapeutic agents. Based on GR50 values, we classified the cell lines into sensitive and resistant groups. We then identified key genes associated with multidrug sensitivity and validated their functional roles in vitro and in vivo.
Methods
Cell lines and culture condition
All pancreatic cancer cell lines were purchased from the American Type Culture Collection (ATCC), and identified by short tandem repeat (STR) sequencing and mycoplasma testing every two months. The HPAF2, PATU8988T, MIA PaCa-2, T3M4, and PANC-1 cell lines were cultured in high glucose Dulbecco’s modified Eagle’s medium (DMEM; CORNING, USA), and the AsPC-1, SW1990 and BxPC-3 cell lines were cultured in RPMI-1640 medium (CORNING, USA). All cell lines were placed in a humidified incubator at 37℃ with 5% carbon dioxide. To construct stable EDC4-overexpressing and EDC4-knockdown cell lines, EDC4 ORF sequences were cloned into Ubi-MCS-SV40-Puromycin lentiviral vectors and siEDC4-1 sequences were cloned into hU6-MCS-CMV-Puromycin lentiviral vectors (Shanghai Genechem, China). AsPC-1 and MIA PaCa-2 cells were infected with these lentiviruses for over 24 h. Then, puromycin (1 µg/ml, Sigma‒Aldrich, USA) was added to the medium to remove uninfected cells.
In vitro growth inhibition assays and GR50 (growth rate 50) calculation
Cells were seeded into 96-well plates and treated after 24 h with five chemotherapeutics. The drug concentration gradients were set at 0, 1 nM, 10 nM, 100 nM, 1 µM, 10 µM, 100 µM and 1 M. The SRB assay was performed after an additional 48 h of incubation to determine the cell viability and inhibition rate [11]. The previously published method was utilized to determine the growth rate (GR) inhibition metrics [8]. This approach is better suitable for testing cellular sensitivity and can potentially minimize confounding factors like the intrinsic cell proliferation.
Definition of chemosensitive (CS) and chemoresistant (CR) cell lines
The chemosensitivity of each cell line was defined based on GR50 values, which measure the drug concentration required to reduce cell growth by 50%. Eight PDAC cell lines were treated with five chemotherapeutic agents (gemcitabine, albumin-bound paclitaxel, irinotecan, 5-FU, and cisplatin), and GR50 values were calculated using dose-response curves. For each drug, cell lines were ranked based on their GR50 values. The four cell lines with the lowest GR50 values (indicating higher sensitivity) were classified as chemosensitive (CS), while the four cell lines with the highest GR50 values (indicating lower sensitivity) were classified as chemoresistant (CR). To identify cell lines with consistent sensitivity or resistance across multiple drugs, a Venn diagram analysis was performed to determine the overlap of CS and CR classifications.
Datasets sources and processing
RNA-seq data of the eight PDAC cell lines were obtained from the Cancer Cell Line Encyclopedia (CCLE) database. RNA-seq data for PDAC tissue with follow-up information was obtained from The Cancer Genome Atlas (TCGA) dataset in UCSC XENA website (https://xenabrowser.net/datapages/; up to March 1, 2022). Proteomic data of pancreatic cancer was obtained from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) database (https://www.linkedomics.org/) [12]. The DEGs between CS and CR cell lines were obtained by limma R package. log|Fold Change| >1 and P < 0.05 were set as cutoffs for the DEGs. The pheatmap, dplyr, and ggplot2 R package were used to create the heatmap and volcano plot. Further dimension reduction was performed on the survival-related DEGs using the LASSO algorithm (R package glmnet), random forest (R package randomforest), and bi-directional stepwise multivariate Cox analysis (R package rms) to yield the final signature and chemosensitivity score. According to gene expression and multivariate Cox coefficients, the signature formula was established as follows: (exprgene1 × Coefgene1) + (exprgene2 × Coefgene2). Furthermore, patients in the TCGA cohort and GSE57495 + GSE28735 cohort were divided into low-score and high-score groups based on the median chemosensitive score. This median cutoff was applied consistently across both datasets to ensure comparability and balance between groups. The efficiency and universality of the signature (Chemosensitive score) were assessed by Kaplan Meier (K–M) and time-dependent ROC curves in TCGA (n = 176) and GSE57495 + GSE28735 (n = 105) cohorts, respectively. The R package pROC was used to calculate the Area Under the Curve (AUC) for ROC analysis. We then developed a nomogram to predict overall survival in pancreatic cancer patients using a multivariate Cox proportional hazards model (R package rms). We included clinical and pathological variables such as age, gender, tumor grade, nodal status, tumor stage, and a chemosensitivity score derived from our gene expression analysis. Statistical analysis was performed using R software package (version 4.2.1). A two-sided p-value < 0.05 was considered statistically significant.
Batch effect correction for multi-dataset integration
For cross-dataset integration, batch effects were eliminated using the ComBat function from the sva package in R. The detailed procedure was as follows: Merged expression matrices from multiple datasets (CCLE, TCGA, GSE57495 + GSE28735) were read into R using cdata <- read.table(“merge.txt”, header = T, sep = “\t”, row.names = 1) and converted to matrix format (cdata <- as.matrix(cdata)); a custom batch information file (“sif.txt”) with sample names (consistent with those in the expression matrix) and batch labels (denoted as 1, 2, 3, etc. to distinguish different datasets) was imported via csif <- read.table(“sif.txt”, header = T, sep = “\t”, row.names = 1); a null model matrix was generated with modcombat = model.matrix(~ 1, data = csif) to account for baseline variations, followed by batch effect correction using combat_edata = ComBat(dat = cdata, batch = csif$batch, mod = modcombat, par.prior = TRUE, prior.plots = FALSE); corrected data was exported using write.table(combat_edata, “ComBat_data(all)(counts).txt”, sep = “\t”, quote = F), and the effectiveness of correction was validated by comparing boxplots of expression distributions before (boxplot(as.data.frame(cdata), main = “Original”)) and after (boxplot(as.data.frame(combat_edata), main = “Batch corrected”)) processing, confirming improved data consistency across datasets.
Transfection assay and western blotting
Transfections were conducted by using OPTI-MEM (Invitrogen, USA) and Lipofectamine 3000 (Invitrogen, USA). siEDC4-1 (sense: 5’-GAAGAGAUCUUGGUCCAUATT-3’; antisense: 5’-UAUGGACCAAGAUCUCUUCTT-3’), siEDC4-2 (sense: 5’-GAUGGAGAUCGGCAUAAUATT-3’; antisense: 5’-UAUUAUGCCGAUCUCCAUCTT-3’), siEDC4-3 (sense: 5’-GACCAGAACCGAGAGUUAATT-3’; antisense: 5’- UUAACUCUCGGUUCUGGUCTT-3’) and negative control (NC) were purchased from Tsingke (Beijing, China). The EDC4-overexpression plasmid (EDC4 OE) and negative control plasmid (Vector) were purchased from GeneChem (Shanghai, China). EDC4 cDNA was subcloned into a pcDNA3.1 vector. Western blotting assays were performed according to previously described protocols [13]. The primary antibodies used were anti-EDC4 polyclonal antibody (Proteintech, 17737-1-AP, 1:1000 dilution), anti-SGCE polyclonal antibody (Proteintech, 12407-1-AP, 1:2000 dilution), anti-MATN3 polyclonal antibody (Bioss, bs-18694R, 1:500 dilution), anti-GAPDH monoclonal antibody (Proteintech, 60004-1-Ig, 1:5000 dilution), and anti-β-actin polyclonal antibody (Proteintech, 20536-1-AP, 1:5000 dilution).
Cell proliferation experiment and cell viability assay after transfection
At 24 h after transfection, pancreatic cancer cells were seeded in the 96-well plates (3000 cells per well). Then, the cell proliferation was measured at 0 h, 24 h, 48 h, 72 h and 96 h using the Sulforhodamine B (SRB) assay, a colorimetric method for measuring cell density based on protein binding, was performed after an additional 48 h of incubation to determine cell viability and inhibition rate. In this assay, cells were fixed with trichloroacetic acid (TCA), stained with SRB dye, and subsequently washed to remove unbound dye. The optical density at a wavelength of 564 nm (OD564) was detected by a microplate reader (Synergy H1, Biotek, USA). The growth rate (GR) inhibition metrics were determined following a previously established protocol [11]. The 50% inhibitory concentration (IC50) values were defined as the drug concentration that inhibited cell growth by 50% compared to the untreated control. IC50 values were calculated using GraphPad Prism 9.0 software (La Jolla, USA).
RNA isolation and RT‒PCR
TRIzol reagent (Ambion, Life Technologies, 15596026) was performed to extract total RNA from treated pancreatic cancer cell lines. cDNA was synthesized by a cDNA Reverse Transcription kit (Thermo Scientific, K1622). Quantitative RT‒PCR was conducted using PowerUp™ SYBR™ Green Master Mix (Applied Biosystems, A25742) in StepOnePlus™ (Applied Biosystems) according to the manufacturer’s protocols. We purchased primers from Sangon Biotech (Shanghai, China) (Supplementary Table 1). The relative expression of mRNAs was calculated by the 2^(-ΔΔCT) method.
Immunohistochemistry and tissue microarray
Immunohistochemistry was performed following the previously described protocols [14]. For tissue microarray, a total of 83 PDAC patients (75 of them with survival information) underwent surgical resection at Peking Union Medical College Hospital were included. All patients involved in this study signed the informed consent. The Ethics Committee of Peking Union Medical College Hospital has approved this project (Approval number: I-23PJ1544). The results of staining were determined independently by two experienced pathologists. Immunohistochemical scoring was performed independently by two experienced pathologists in a blinded manner. Inter-observer consistency was evaluated, and cases with discrepant scores were jointly reviewed to reach a consensus. The staining score was calculated as the product of intensity and proportion scores. Staining score = tumor cell proportion × staining intensity. Proportion: 0 (no stained tumor cells), 1 (< 25% stained tumor cells), 2 (25–50% stained tumor cells), 3 (50–75% stained tumor cells), and 4 (≥ 75% stained tumor cells). Intensity: 0 (negative particles), 1 (lightly yellow particles), 2 (brownish-yellow particles) and 3 (brown particles). Patients were divided into high and low EDC4 expression groups according to the median staining score. Survival differences were assessed using Kaplan–Meier analysis with log-rank tests.
Animal experiments and drug administration protocol
All animal experiments in the current research were performed according to the guidelines of the Institutional Animal Care and Use Committee (Beijing, China). The relative control cells, EDC4-overexpressing AsPC-1 cells and EDC4-knockdown MIA PaCa-2 cells were injected into the right back of 6-week-old female BALB/c nude mice (Chinese Academy of Sciences, China) (5 × 106 cells in 300 µl of PBS per mouse). The mice were randomly divided into different groups, five in each group. The tumor sizes were measured and assessed according to the previously described method [11]. To confirm the role of EDC4 in inhibiting chemosensitivity in vivo, mice with EDC4-overexpression tumor and relative control tumor were administered modified FOLFIRINOX (mFOLFIRINOX) therapeutic regimen, and mice with EDC4-knockdown tumor and relative control tumor were administered gemcitabine/nab-paclitaxel (GEM/AB) therapeutic regimen for three weeks. The chemotherapy measurement and drug administration protocol were adapted from the literature by Delahoussaye et al. [15]. The drug administration protocol was as follows: The mFOLFIRINOX protocol included the administration of oxaliplatin at a dose of 10 mg/kg via tail vein injection, 5-fluorouracil (5-FU) at a dose of 35 mg/kg via intraperitoneal injection, and irinotecan at a dose of 15 mg/kg via intraperitoneal injection. These three drugs were administered on the same day, once a week. The GEM/AB protocol involved the administration of gemcitabine at a dose of 50 mg/kg via intraperitoneal injection, twice weekly. Nab-paclitaxel was administered at a dose of 30 mg/kg (calculated as paclitaxel) via intraperitoneal injection 1 h after the first weekly gemcitabine dose, once a week. These protocols were followed to ensure consistent dosing and effective drug delivery. Mice were euthanized after three weeks of administration. The xenografts were excised to perform further study in weight and morphology.
Results
Differential expressed genes in multi-drug chemosensitive and chemoresistance cell lines of PDAC
First, we constructed GR50 profiles of eight pancreatic cancer cell lines in five commonly used chemotherapeutic drugs including gemcitabine, albumin paclitaxel, irinotecan, 5-FU and cis-platinum (Figs. 1A-E). The results indicated that pancreatic cancer cell lines exhibit varied sensitivity and resistance to different chemotherapeutic agents. We defined four cell lines with the lower GR50 values for chemotherapeutic drug as chemosensitive (CS) cell lines and other four as chemoresistance (CR) cell lines. Venn diagrams highlighted the specific cell lines that are chemosensitive or chemoresistant to the drugs tested. After taking intersection, PATU8988T, BxPC-3, and MIA PaCa-2 were selected as multi-drug CS cell lines while PANC-1, AsPC-1, and SW1990 were selected as multi-drug CR cell lines (Fig. 1F, G). Then, we performed DEG analysis between CR and CS cell lines. DEG analysis in the CCLE (Fig. 1H) and hierarchical clustering (Fig. 1I) revealed significant gene expression changes. 210 DEGs among multi-drug CR and CS cell lines were obtained and used for subsequent analysis (Supplementary Table 2). After performing univariate Cox analyses with the follow-up information in TCGA database, there were 57 DEGs significantly related to pancreatic cancer patients’ survival (Supplementary Table 3). Among them, 30 DEGs were highly expressed in CR cell lines and related to poor prognosis (Fig. 1J), 27 DEGs were highly expressed in CS cell lines and related to better prognosis (Fig. 1K). Then, KEGG and GO analysis about 57 survival-related DEGs were performed. Most of them were enriched in “Regionalization” and “Development” pathways (Fig. 1L, M). Existing studies have confirmed that pancreatic cancer cells can acquire cancer stem cell (CSC) properties through epithelial-mesenchymal transition (EMT), thereby leading to chemoresistance and poor prognosis [16]. For example, high expression of ZEB1 during EMT in pancreatic cancer cells can induce CSC properties and further contribute to chemoresistance [17]. The process by which pancreatic cancer cells acquire CSC properties through EMT to achieve chemoresistance is a typical case that involves both Regionalization and Development pathways [16]. Consistently, the Regionalization pathway enriched in our study is closely associated with EMT (a key process for cells to gain mesenchymal characteristics and enhanced mobility), while the Development pathway is strongly linked to the maintenance of CSC properties (which support tumor self-renewal and drug resistance).
Fig. 1.
GR50 profiles in eight pancreatic cancer cell lines and DEGs screening. (A-E) GR50 Dose-Response Curves of Various Pancreatic Cancer Cell Lines to Chemotherapeutic Agents including Gemcitabine, Albumin Paclitaxel, Cis-Platinum, Irinotecan, and 5-FU respectivly. (F, G) Venn diagrams showing the overlap of chemosensitive (CS) and chemoresistant (CR) cell lines to different chemotherapeutic agents. (H) Volcano plot of differential gene expression in the Cancer Cell Line Encyclopedia (CCLE). (I) Heatmap showing the hierarchical clustering of gene expression profiles in the CCLE. (J-K) Dot plots showing the 30 and 27 survival-related DEGs in chemosensitive (J) and chemoresistant (K) cell lines. (L-M) Bubble plots of KEGG and GO analysis for 57 survival-related DEGs
Establishment and validation of multi-drug chemosensitive related gene signature
We conducted LASSO, random forest, and multivariate Cox analyses to further downscale the 57 survival-related genes (Figs. 2A-D). Two hub genes with P-value < 0.05 in multivariate Cox analyses (EDC4 and USP20) were screened out to construct the risk signature (Fig. 2D). Interestingly, EDC4 and USP20 were all associated with long-term prognosis and promote chemosensitivity. Therefore, we calculated Chemosensitive Score based on their coefficients: Chemosensitive Score = (Expression of EDC4 × 0.90691) + (Expression of USP20 × 0.76112). Furthermore, patients in the TCGA cohort and GSE57495 + GSE28735 cohort were divided into low-score and high-score groups based on the median chemosensitive score. Kaplan‒Meier survival analysis demonstrated that patients in the high-score group had significantly longer survival than those in the low-score group in these cohorts (Figs. 2E, G). Time-dependent ROC curves showed that our model had good predictive ability (Figs. 2F, H). Moreover, given the high predicting ability of the two-gene signature, the chemosensitive score and clinicopathological factors were incorporated to establish a nomogram in the TCGA cohort (Fig. 2I), with the ROC curve showing our nomogram had great predicting efficacy (Fig. 2J). Because EDC4 had the highest coefficient in the multivariate Cox regression model, we chose it for subsequent analysis.
Fig. 2.
Identification and Validation of a Prognostic Gene Signature. (A) LASSO coefficient profiles of the candidate genes. The optimal value of the tuning parameter (lambda) is identified through tenfold cross-validation. (B) LASSO coefficient profiles for different log(lambda) values. Each curve corresponds to a gene. (C) Venn diagram showing the intersection of genes selected by LASSO and Random Forest methods. (D) Forest plot displaying the Hazard Ratios (HR) and p-values for the identified genes in the prognostic model. (E) Kaplan-Meier survival curves comparing patients with high and low chemosensitive scores in the TCGA dataset. The log-rank test p-value indicates a significant difference between the groups. (F) Time-dependent ROC (Receiver Operating Characteristic) curves evaluating the prognostic performance of the two-gene signature in the TCGA dataset at 1, 3, and 5 years. (G) Kaplan-Meier survival curves comparing patients with high and low scores in the combined GSE57495 and GSE28735 datasets. (H) Time-dependent ROC curves evaluating the prognostic performance of the gene signature in the combined GSE57495 and GSE28735 datasets at 1, 3, and 5 years. (I) Nomogram for predicting 1-year, 3-year, and 5-year overall survival probabilities in patients based on the two-gene signature and clinical parameters (age, gender, grade, N stage, T stage, and chemosensitive score). (J) Time-dependent ROC curves for the nomogram predicting 1-year, 3-year, and 5-year overall survival probabilities
EDC4 regulated the proliferation and chemoresistance of PDAC cell lines
First, we analyzed the expression level of EDC4 in the above mentioned 3 CR (PANC-1, AsPC-1, SW1990) and 3 CS (PATU-8988, BxPC-3, MIA PaCa-2) cell lines by qRT-PCR (Fig. 3A) and WB (Fig. 3B). The expression of EDC4 was lower in AsPC-1 cells while higher in MIA PaCa-2 cells. Hence, we detected the efficiency of EDC4-siRNA in the MIA PaCa-2 cell line and EDC4 OE plasmid in the AsPC-1 cell line using qRT-PCR (Fig. 3C) and western blotting (Figs. 3D, E).
Fig. 3.
EDC4 Expression and Its Impact on Pancreatic Cancer Cell Lines. (A) Relative expression of EDC4 mRNA in different pancreatic cancer cell lines. (B) Protein levels of EDC4 in various pancreatic cancer cell lines, analyzed by Western blotting with β-actin as a loading control. (C) Knockdown efficiency of EDC4 in MIA PaCa-2 cells using three different siRNAs (siEDC4-1, siEDC4-2, siEDC4-3), compared to the negative control (NC). (D) Western blot analysis of EDC4 protein levels in MIA PaCa-2 cells following siRNA-mediated knockdown. (E) Western blot analysis of EDC4 protein levels in AsPC-1 cells overexpressing EDC4 (EDC4 OE) compared to the vector control. (F) Cell proliferation assay (measured by OD 565 nm) in MIA PaCa-2 cells with EDC4 knockdown over 4 days. (G) Cell proliferation assay in AsPC-1 cells with EDC4 overexpression over 4 days. (H-L) Cell viability assays calculated by IC50 in MIA PaCa-2 cells with EDC4 knockdown treated with different concentrations of gemcitabine (H), 5-FU (I), albumin-bound paclitaxel (J), irinotecan (K), and cisplatin (L). (M-Q) Cell viability assays calculated by IC50 in AsPC-1 cells with EDC4 overexpression treated with different concentrations of gemcitabine (M), 5-FU (N), albumin-bound paclitaxel (O), irinotecan (P), and cisplatin (Q). Significance levels: *P < 0.05, **P < 0.01, ***P < 0.001
EDC4 knockdown in MIA PaCa-2 cells significantly increased cell proliferation over 96 h compared to the control (Fig. 3F). Conversely, EDC4 overexpression in AsPC-1 cells significantly inhibited cell proliferation (Fig. 3G). We then performed cell viability assays in MIA PaCa-2 cells with EDC4 knockdown and in AsPC-1 cells with EDC4 overexpression treated with different concentrations of gemcitabine, 5-FU, albumin-bound paclitaxel, irinotecan, and cisplatin. Since we focused on a single cell line (MIA PaCa-2 and AsPC-1) under controlled conditions, where cell proliferation background was highly consistent. In this context, we used the more classic and traditional IC50 metric to assess drug sensitivity. EDC4 knockdown significantly decreased cell sensitivity to these chemotherapeutic agents, as indicated by reduced cell viability at various drug concentrations (Figs. 3H-L). Overexpression of EDC4 increased cell sensitivity to these drugs, as shown by higher cell viability at increasing drug concentrations (Figs. 3M-Q). These results suggest that EDC4 plays a significant role in modulating the sensitivity of pancreatic cancer cells to various chemotherapeutic agents.
High EDC4 expression is related to longer survival of PDAC patients in a PUMCH clinical cohort
We further included 83 PDAC samples and examined the EDC4 expression levels and clinicopathologic characteristics in PDAC tissues by tissue microarray and immunohistochemistry (Table 1) (Fig. 4A). Among them, 75 PDAC patients have complete follow-up information. The results showed that the PDAC patients in the high EDC4 expression group had significantly longer survival than the low EDC4 expression group (Fig. 4B) (Table 2). Therefore, Higher EDC4 expression might indicate multi-drug chemosensitivity of PDAC in the clinical practice.
Table 1.
Clinicopathological characteristics of PDAC patients stratified by EDC4 expression
| Variables | n | EDC4 expression | ||
|---|---|---|---|---|
| Low group (n=46) |
High group (n=37) |
P value | ||
| Gender | 0.302 | |||
| Male | 50 | 30 | 20 | |
| Female | 33 | 16 | 17 | |
| Age | 1 | |||
| <60 | 36 | 20 | 16 | |
| ≥60 | 47 | 26 | 21 | |
| Location | 0.372 | |||
| Head | 54 | 28 | 26 | |
| Body/Tail | 29 | 18 | 11 | |
| Differential degree | 0.174 | |||
| Low | 36 | 23 | 13 | |
| High/moderate | 47 | 23 | 24 | |
| T staging | 0.944 | |||
| T1/2 | 58 | 32 | 26 | |
| T3 | 25 | 14 | 11 | |
| N staging | 0.573 | |||
| N0 | 34 | 18 | 16 | |
| N1/N2 | 49 | 29 | 20 | |
| TNM staging | 0.561 | |||
| I/IIA | 33 | 17 | 16 | |
| IIB/III | 50 | 29 | 21 | |
Fig. 4.
EDC4 Expression and Overall Survival in a PUMCH Clinical Cohort. (A) Immunohistochemical staining of tumor tissue sections shows low EDC4 expression (left) and high EDC4 expression (right). The intensity of the staining indicates the level of EDC4 expression. (B) Kaplan-Meier survival curves comparing overall survival of patients with high versus low EDC4 expression. Patients with high EDC4 expression (red line, n = 33) have significantly poorer survival compared to those with low EDC4 expression (blue line, n = 42). Scale bar equals 50 μm
Table 2.
Univariate and multivariate Cox regression analyses of prognostic factors in PDAC patients
| Variables | n | Univariate | Multivariate | ||
|---|---|---|---|---|---|
| HR (95%CI) | P | HR (95%CI) | P | ||
| Gender | |||||
| Female | 32 | 1 | 0.351 | ||
| Male | 43 | 1.302 (0.748-2.264) | |||
| Age | |||||
| <60 | 31 | 1 | 0.085 | ||
| ≥60 | 44 | 1.636 (0.934-2.867) | |||
| Location | |||||
| Head | 49 | 1 | 0.282 | ||
| Body/Tail | 26 | 0.726 (0.404-1.302) | |||
| Differential degree | |||||
| Low | 34 | 1 | 0.0036** | 1 | 0.026* |
| High/moderate | 41 | 0.448 (0.261-0.770) | 0.533 (0.307-0.926) | ||
| T staging | |||||
| T1/2 | 53 | 1 | 0.0089** | 1 | 0.025* |
| T3 | 22 | 2.12 (1.207-3.724) | 1.916 (1.086-3.381) | ||
| N staging | |||||
| N0 | 30 | 1 | 0.868 | ||
| N1/N2 | 45 | 1.048 (0.606-1.813) | |||
| TNM staging | |||||
| I/IIA | 28 | 1 | 0.637 | ||
| IIB/III | 47 | 1.144 (0.654-2.003) | |||
| EDC4 expression | |||||
| Low | 42 | 1 | 0.0096** | 1 | 0.038* |
| High | 33 | 0.468 (0.263-0.831) | 0.539 (0.301-0.966) | ||
MATN3 and SGCE were the potential downstream molecules of EDC4
EDC4 is an RNA uncapping protein that can degrade cancer-promoting mRNA [18]. First, we introduced three datasets: CCLE, TCGA, and CPTAC. The CCLE dataset contains RNA expression data from cell lines, the TCGA dataset provides RNA expression data from pancreatic cancer tissues, and the CPTAC dataset offers protein expression data from pancreatic cancer tissues. Since EDC4 degrades downstream mRNA, we utilized these three datasets to identify mRNA and proteins negatively correlated with EDC4 expression, spanning from cell to tissue level and from mRNA to protein level. By intersecting these three datasets, we identified 31 genes as potential downstream targets of EDC4, which negatively associated with EDC4 expression (Supplementary Table 4) (Fig. 5A). Among them, we found 7 proteins (GIPC2, PRSS3, MATN3, SGCE, PLD1, DEFB1, and TLR3) associated with poor prognosis of pancreatic cancer by performing univariate COX regression analysis in TCGA cohort (Fig. 5B). Then, we performed qRT-PCR experiment to validate these 7 mRNAs expression level after knockdown or overexpression of EDC4. As shown in Fig. 5C, MATN3, SGCE, GIPC2, and PRSS3 were highly expressed after knockdown. Meanwhile, MATN3, SGCE, PLD1, DEFB1, and TLR3 were down-regulated after overexpression (Fig. 5D). Hence, we presumed that MATN3 and SGCE were the potential downstream molecules of EDC4. We then performed Western blotting analyses to examine the protein expression levels of EDC4 and its downstream targets, MATN3 and SGCE. In MIA PaCa-2 cells, knockdown of EDC4 effectively reduced the protein levels of EDC4 but increased MATN3 and SGCE compared to the negative control (NC) (Fig. 5E). This indicated that EDC4 negatively regulated the expression of these proteins (Fig. 5F). Consistently, overexpression of EDC4 in AsPC-1 cells resulted in an increased expression of EDC4 and decreased expression of MATN3 and SGCE compared to cells transfected with the vector. These results collectively supported the hypothesis that EDC4 modulated the expression of MATN3 and SGCE as its downstream targets at both the mRNA and protein levels.
Fig. 5.
Potential Downstream Molecules Exploration of EDC4. (A) Venn diagram illustrating the overlap of genes from the CCLE, TCGA, and CPTAC datasets. (B) Table displaying the hazard ratio (HR), p-value, and confidence intervals for selected genes. (C, D) Relative expression levels of mRNA for selected genes in MIA PaCa-2 cells following EDC4 knockdown and in AsPC-1 cells with EDC4 overexpression. (E, F) Western blot analysis of EDC4, MATN3, and SGCE protein levels in MIA PaCa-2 cells following EDC4 knockdown, and in AsPC-1 cells with EDC4 overexpression. Significant results were observed at ***, P < 0.001; **, P < 0.01, and *, P < 0.05, respectively
EDC4 decreases the chemoresistance of pancreatic cancer in vivo
To validate the function of EDC4 in decreasing chemoresistance in vivo, we employed nude mice, stable EDC4-overexpressing AsPC-1 cells and EDC4-konckdown MIA PaCa-2 to construct subcutaneous xenografts. The impact of lentivirus-mediated EDC4 modulation on tumor growth was assessed in MIA PaCa-2 and AsPC-1 pancreatic cancer cells under different chemotherapy treatments. After all of the individual volume subcutaneous xenografts reached 80 mm3, the EDC4-overexpressing mice were injected with gemcitabine/nab-paclitaxel (GEM/AB) and EDC4-konckdown mice were administered FOLFIRINOX, and the tumor volumes were measured every three days. MIA PaCa-2 cells treated with GEM/AB and lentivirus siEDC4-1 exhibited significantly larger tumors and higher tumor weights compared to the control group (Figs. 6A-C), indicating that EDC4 knockdown promoted chemoresistance in this context. Consistently, AsPC-1 cells treated with mFOLFIRINOX and lentivirus EDC4 overexpression (OE) displayed significantly smaller tumors and reduced tumor weights compared to the control group (Figs. 6D-F), suggesting that EDC4 overexpression promoted chemosensitivity. These results highlight the role of EDC4 in modulating the response to chemotherapy in pancreatic cancer cell lines.
Fig. 6.
Analysis of Tumor Volume and Weight after chemotherapy. (A) Representative images of MIA PaCa-2 tumors treated with GEM/AB following lentivirus NC and siEDC4 administration. (B) Tumor volume measurements over time for MIA PaCa-2 cells post GEM/AB treatment. (C) Final tumor weights for MIA PaCa-2 cells treated with lentivirus NC and siEDC4-1. (D) Representative images of AsPC-1 tumors treated with mFOLFIRINOX following lentivirus NC and EDC4 OE administration. (E) Tumor volume measurements over time for AsPC-1 cells post mFOLFIRINOX treatment. (F) Final tumor weights for AsPC-1 cells treated with lentivirus NC and EDC4 OE. Statistical significance is indicated as *P < 0.05 and **P < 0.01
Discussion
Chemotherapy remains a cornerstone in the treatment of pancreatic cancer; however, its effectiveness is often limited by tumor cell resistance, which leads to suboptimal outcomes in many patients. GR50 (Growth Rate 50) represents the drug concentration required to reduce the growth rate of a cell population by 50%. Although IC50 (Inhibitory Concentration 50) is a widely used metric for evaluating drug potency, GR50 offers distinct advantages—particularly in accounting for differences in baseline cell proliferation and providing a more dynamic assessment of drug efficacy [19, 20]. By normalizing drug response to cell growth rates, GR50 minimizes variability arising from intrinsic proliferation differences and is less influenced by experimental inconsistencies, resulting in more reliable and reproducible measurements [8]. In this study, we used GR50 to assess the chemosensitivity of ten commonly used PDAC cell lines across five standard chemotherapeutic agents, thereby eliminating the confounding effects of differential cell proliferation.
Our analysis revealed that EDC4 plays a pivotal role in enhancing the chemosensitivity of PDAC cells. Differential gene expression analysis between chemosensitive and chemoresistant cell lines identified EDC4 as one of the most significant candidates. Functional experiments confirmed that EDC4 overexpression sensitized resistant cells to chemotherapy, while its knockdown conferred resistance in previously sensitive cells. These findings were further validated in vivo, where EDC4 expression modulated tumor growth in response to chemotherapy in xenograft models. Importantly, high EDC4 expression was also associated with improved patient survival in our clinical cohort, supporting its potential as both a predictive biomarker and a therapeutic target.
To stratify patients based on the chemosensitive score, we used the median value as the cutoff in both the TCGA and GEO datasets. This method is widely adopted in prognostic modeling due to its simplicity and its ability to evenly divide patient cohorts [19, 20], thereby reducing potential bias and enhancing statistical power. It also improves reproducibility in exploratory analyses by avoiding arbitrary thresholds. Nonetheless, we recognize that median-based cutoffs are inherently dataset-dependent and may not translate effectively across diverse clinical populations. For clinical implementation, it will be crucial to validate and refine this threshold in independent prospective cohorts, employing more robust statistical approaches such as maximally selected rank statistics or ROC curve-based optimization. Such strategies would enhance the consistency, interpretability, and practical utility of the chemosensitive score as a prognostic and therapeutic tool in pancreatic cancer.
EDC4 (Enhancer of Decapping 4) is a key component of the mRNA decapping complex, playing a critical role in mRNA decay and post-transcriptional gene regulation [18, 21]. Its involvement in cancer has been linked to processes such as DNA damage response and chemotherapy resistance. In cervical cancer, for example, EDC4 overexpression enhances cisplatin resistance by interacting with replication protein A (RPA) and attenuating DNA damage, thereby promoting tumor survival [22]. In contrast, our findings suggest a context-specific role for EDC4 in pancreatic cancer, where higher EDC4 expression correlates with improved survival and reduced chemoresistance. This may be due to EDC4-mediated degradation of oncogenic or chemoresistance-associated mRNAs. The divergent effects of EDC4 across tumor types likely reflect differences in the tumor microenvironment and genetic background.
MATN3 and SGCE were identified as potential downstream targets of EDC4. MATN3, a member of the matrilin protein family, participates in the formation of filamentous networks within the extracellular matrix (ECM) across various tissues [23, 24]. Previous studies have shown that MATN3 is predominantly expressed in cartilage, including tracheal, costal, and articular cartilage [25–27]. Klatt et al. reported that MATN3 can promote ECM degradation by inducing pro-inflammatory cytokines (TNFα, IL-1β, IL-6, IL-8) and matrix-degrading enzymes (ADAMTS4 and ADAMTS5) through IL-1β signaling. These enzymes, in turn, cleave MATN3, creating a feedback loop that exacerbates matrix breakdown and inflammation [26]. MATN3 has also been found to be overexpressed in gastric cancer and is associated with poor prognosis [28, 29]. Although its role in pancreatic cancer remains largely unexplored, our findings suggest that EDC4 may enhance chemosensitivity by downregulating MATN3, thereby reducing ECM stiffness, lowering interstitial fluid pressure, and improving drug penetration and response.
SGCE (Sarcoglycan Epsilon) is a single-pass transmembrane protein implicated in cancer progression through diverse mechanisms. In triple-negative breast cancer (TNBC), SGCE functions as an oncogene, promoting cancer stemness by stabilizing EGFR and enhancing FGF-BP1 transcription via Sp1 [30]. Bioinformatics analyses have also linked SGCE to poor prognosis in gastric [32, 33] and colorectal cancers [31, 32] and colorectal cancer [33]. Although its role in pancreatic cancer is not well defined, it is plausible that SGCE exerts similar oncogenic effects. In TNBC, SGCE impedes EGFR degradation through the ubiquitin-lysosomal pathway, contributing to drug resistance and metastasis [34]. If SGCE behaves similarly in PDAC, its overexpression may drive tumor progression, chemoresistance, and metastasis. Further studies are warranted to elucidate SGCE’s mechanistic role and therapeutic potential in pancreatic cancer.
This study is the first to comprehensively characterize the intrinsic chemoresistance of multiple pancreatic cancer cell lines across a range of chemotherapeutic agents. By generating GR50-based drug response profiles, we provide a valuable reference for future investigations into the mechanisms of chemoresistance and the development of targeted therapies. We identified EDC4 as a key modulator of chemosensitivity, validated its function both in vitro and in vivo, and revealed MATN3 and SGCE as potential downstream effectors. These findings offer new insights into resistance mechanisms and suggest that targeting the EDC4–MATN3 and EDC4–SGCE pathways may improve therapeutic outcomes in PDAC. Despite these findings, several limitations should be noted. First, although we used GR50 profiling to account for intrinsic proliferation differences, in vitro results may not fully recapitulate the complexity of the tumor microenvironment in vivo. Second, our classification of chemosensitive and chemoresistant cell lines was based on GR50 values across five drugs, but further validation in larger panels and clinical samples would strengthen generalizability. Third, while we identified EDC4 as a key regulator of multidrug sensitivity, we did not explore the potential influence of genetic mutations such as KRAS subtypes on drug response. Future studies integrating mutational status, tumor heterogeneity, and pharmacogenomic data will be essential to refine our findings and translate them into clinical application. Besides, our in vivo experiments confirm EDC4 regulates the chemosensitivity of pancreatic cancer xenografts, but the specific biological processes such as tumor cell proliferation, apoptosis and senescence involved in this regulation remain to be clarified. In follow-up studies, we will detect markers including Ki-67, Cleaved caspase 3 and P21, and further explore the molecular interaction between EDC4 and its downstream targets SGCE and MATN3.
Conclusion
Our study found the drug resistance profiles of eight cell lines based on the GR50 values and elucidated the differential expression and impact of genes in multi-drug chemosensitive and chemoresistant PDAC cell lines. Furthermore, we identified the pivotal role of EDC4 in modulating response to multiple chemotherapeutic agents. Its knockdown enhances resistance to chemotherapeutic agents, while overexpression confers sensitivity. In vivo experiments confirmed EDC4’s role in decreasing chemoresistance. EDC4 also regulates the expression of MATN3 and SGCE, highlighting its potential as a therapeutic target to improve chemotherapeutic efficacy in PDAC treatment.
Ethics statement
The Ethics Committee of Peking Union Medical College Hospital has approved this project (Approval number: I-23PJ1544). All methods were carried out in accordance with relevant guidelines and regulations (declaration of helsinki and the ICMJE guidelines on Protection of Research Participants). All patients involved in this study signed the informed consent. The animal experiment conducted observed the ARRIVE guidelines. The Ethics Committee of The Peking Union Medical College Hospital authorized the animal experiment.
Supplementary Information
Acknowledgements
None.
Author contributions
QC interpreted the data, performed experiments, and revised this manuscript. LTY performed data analysis, performed experiments, and drafted the original manuscript. ZBB performed experiments, and revised this manuscript. LZR, ZYT, XYP, YYT, HJS made important revisions to this manuscript. WWB and LC guided and supervised this study. All authors contributed to the article and approved the submitted version.
Funding
Wei-bin Wang received the support from the National Natural Science Foundation of China (No. 82173074), Beijing Natural Science Foundation (No. 7232127), the National High Level Hospital Clinical Research Funding (No. 2022-PUMCH-D-001, No. 2022-PUMCH-B-004), National Key R&D Program of China (2022YFF1202600, 2024YFB4708800), Peking Union Medical College Hospital Undergraduate Educational Reform Project (No. 2025bkjg005). Chen Lin received the support from the CAMS Innovation Fund for Medical Sciences (No. 2024-I2M-ZD-001). Cheng Qin received the support from the Postdoctoral Fellowship Program of CPSF under Grant Number GZC20240146.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
All patients signed an informed consent form.
Consent for publication
The participants gave their consent for the study to be published.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Cheng Qin, Tianyu Li and Bangbo Zhao contributed equally to this work.
Contributor Information
Chen Lin, Email: linchen0818@sina.com.
Weibin Wang, Email: wwb_xh@163.com.
References
- 1.Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–49. [DOI] [PubMed] [Google Scholar]
- 2.Mizrahi JD, Surana R, Valle JW, Shroff RT. Pancreatic cancer. Lancet. 2020;395(10242):2008–20. [DOI] [PubMed] [Google Scholar]
- 3.Ettrich TJ, Seufferlein T. Systemic therapy for metastatic pancreatic cancer. Curr Treat Options Oncol. 2021;22(11):106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Wu W, Miao Y, Yang Y, Lou W, Zhao Y, on behalf of Chinese Pancreatic Surgery Association CSoS, Chinese Medical Association. Real-world study of surgical treatment of pancreatic cancer in China: annual report of China Pancreas Data Center (2016–2020). Journal of Pancreatology. 2022;5(1):1–9.
- 5.Kunzmann V, Siveke JT, Algül H, Goekkurt E, Siegler G, Martens U, et al. Nab-paclitaxel plus gemcitabine versus nab-paclitaxel plus gemcitabine followed by FOLFIRINOX induction chemotherapy in locally advanced pancreatic cancer (NEOLAP-AIO-PAK-0113): a multicentre, randomised, phase 2 trial. Lancet Gastroenterol Hepatol. 2021;6(2):128–38. [DOI] [PubMed] [Google Scholar]
- 6.Conroy T, Castan F, Lopez A, Turpin A, Ben Abdelghani M, Wei AC, et al. Five-year outcomes of FOLFIRINOX vs gemcitabine as adjuvant therapy for pancreatic cancer: a randomized clinical trial. JAMA Oncol. 2022;8(11):1571–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Davodabadi F, Sajjadi SF, Sarhadi M, Mirghasemi S, Nadali Hezaveh M, Khosravi S, et al. Cancer chemotherapy resistance: mechanisms and recent breakthrough in targeted drug delivery. Eur J Pharmacol. 2023;958:176013. [DOI] [PubMed] [Google Scholar]
- 8.Hafner M, Niepel M, Chung M, Sorger PK. Growth rate inhibition metrics correct for confounders in measuring sensitivity to cancer drugs. Nat Methods. 2016;13(6):521–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Snoznik C, Medvedeva V, Mojsilovic-Petrovic J, Rudich P, Oosten J, Kalb RG, et al. The nuclear ubiquitin ligase adaptor SPOP is a conserved regulator of C9orf72 dipeptide toxicity. Proc Natl Acad Sci U S A. 2021;118:40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pramanik B, Debnath S, Rahimi M, Helal MMU, Hasan R. Morphometric frequency and spectrum of gamma-ray-induced chlorophyll mutants identified by phenotype and development of novel variants in lentil (Lens culinaris Medik). PLoS ONE. 2023;18(6):e0286975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Qin C, Wang Y, Zhao B, Li Z, Li T, Yang X, et al. Stoml2 restricts mitophagy and increases chemosensitivity in pancreatic cancer through stabilizing PARL-induced PINK1 degradation. Cell Death Dis. 2023;14(3):191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Vasaikar SV, Straub P, Wang J, Zhang B. Linkedomics: analyzing multi-omics data within and across 32 cancer types. Nucleic Acids Res. 2018;46(D1):D956-63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zhao B, Qin C, Li Z, Wang Y, Li T, Cao H, et al. Multidrug resistance genes screening of pancreatic ductal adenocarcinoma based on sensitivity profile to chemotherapeutic drugs. Cancer Cell Int. 2022;22(1):374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Chen Y, Xu R, Ruze R, Yang J, Wang H, Song J, et al. Construction of a prognostic model with histone modification-related genes and identification of potential drugs in pancreatic cancer. Cancer Cell Int. 2021;21(1):291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Delahoussaye AM, Abi Jaoude J, Green M, Fujimoto TN, Molkentine J, Garcia Garcia CJ, et al. Feasibility of administering human pancreatic cancer chemotherapy in a spontaneous pancreatic cancer mouse model. BMC Cancer. 2022;22(1):174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Rodriguez-Aznar E, Wiesmüller L, Sainz B, Hermann PC. EMT and stemness-key players in pancreatic cancer stem cells. Cancers (Basel). 2019;11:8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Rhim AD, Mirek ET, Aiello NM, Maitra A, Bailey JM, McAllister F, et al. EMT and dissemination precede pancreatic tumor formation. Cell. 2012;148(1–2):349–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hernández G, Ramírez MJ, Minguillón J, Quiles P, Ruiz de Garibay G, Aza-Carmona M, et al. Decapping protein EDC4 regulates DNA repair and phenocopies BRCA1. Nat Commun. 2018;9(1):967. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Clark NA, Hafner M, Kouril M, Williams EH, Muhlich JL, Pilarczyk M, et al. GRcalculator: an online tool for calculating and mining dose-response data. BMC Cancer. 2017;17(1):698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yu H, Kim D-J, Choi H-Y, Kim SM, Rahaman MI, Kim Y-H, et al. Prospective pharmacological methodology for establishing and evaluating anti-cancer drug resistant cell lines. BMC Cancer. 2021;21(1):1049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Brothers WR, Ali F, Kajjo S, Fabian MR. The EDC4-XRN1 interaction controls P-body dynamics to link mRNA decapping with decay. EMBO J. 2023;42(21):e113933. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wu X, Zhong Y, Chen Q, Zhang X, Zhang H. Enhancer of mRNA decapping protein 4 (EDC4) interacts with replication protein a (RPA) and contributes to cisplatin resistance in cervical cancer by alleviating DNA damage. Hereditas. 2020;157(1):41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Li D, Xu J, Dong X, Chen W, Pan L, Jiang H, et al. Diagnostic and prognostic value of MATN3 expression in gastric carcinoma: TCGA database mining. J Gastrointest Oncol. 2021;12(4):1374–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Shyamasundar LG, Loganathan L, Kumar A, Selina A, Madhuri V. MATN3 mutation causing spondyloepimetaphyseal dysplasia. Indian J Pediatr. 2020;87(3):227–8. [DOI] [PubMed] [Google Scholar]
- 25.Wagener R, Ehlen HWA, Ko Y-P, Kobbe B, Mann HH, Sengle G, et al. The matrilins–adaptor proteins in the extracellular matrix. FEBS Lett. 2005;579(15):3323–9. [DOI] [PubMed] [Google Scholar]
- 26.Klatt AR, Klinger G, Paul-Klausch B, Kühn G, Renno JH, Wagener R, et al. Matrilin-3 activates the expression of osteoarthritis-associated genes in primary human chondrocytes. FEBS Lett. 2009;583(22):3611–7. [DOI] [PubMed] [Google Scholar]
- 27.Otten C, Hansen U, Talke A, Wagener R, Paulsson M, Zaucke F. A matrilin-3 mutation associated with osteoarthritis does not affect collagen affinity but promotes the formation of wider cartilage collagen fibrils. Hum Mutat. 2010;31(3):254–63. [DOI] [PubMed] [Google Scholar]
- 28.Wu P-L, He Y-F, Yao H-H, Hu B. Martrilin-3 (MATN3) overexpression in gastric adenocarcinoma and its prognostic significance. Med Sci Monit. 2018;24:348–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lin Y, Huang K, Cai Z, Chen Y, Feng L, Gao Y, et al. A novel exosome-relevant molecular classification uncovers distinct immune escape mechanisms and genomic alterations in gastric cancer. Front Pharmacol. 2022;13:884090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Qiu T, Hou L, Zhao L, Wang X, Zhou Z, Yang C, et al. Sgce promotes breast cancer stemness by promoting the transcription of FGF-BP1 by Sp1. J Biol Chem. 2023;299(11):105351. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Zeng C, He R, Dai Y, Lu X, Deng L, Zhu Q, et al. Identification of TGF-β signaling-related molecular patterns, construction of a prognostic model, and prediction of immunotherapy response in gastric cancer. Front Pharmacol. 2022;13:1069204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zhou Y-Y, Kang Y-T, Chen C, Xu F-F, Wang H-N, Jin R. Combination of TNM staging and pathway based risk score models in patients with gastric cancer. J Cell Biochem. 2018;119(4):3608–17. [DOI] [PubMed] [Google Scholar]
- 33.Ortega P, Moran A, Fernandez-Marcelo T, De Juan C, Frias C, Lopez-Asenjo J-A, et al. MMP-7 and SGCE as distinctive molecular factors in sporadic colorectal cancers from the mutator phenotype pathway. Int J Oncol. 2010;36(5):1209–15. [DOI] [PubMed] [Google Scholar]
- 34.Zhao L, Qiu T, Jiang D, Xu H, Zou L, Yang Q, et al. SGCE promotes breast cancer stem cells by stabilizing EGFR. Adv Sci (Weinh). 2020;7(14):1903700. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.