Data obtainment and preprocessing

The mRNA expression data of FNDC3B and clinical characteristics of PC patients, including 178 tumoral tissues and 4 normal tissues (Data Type: HTSeq-TPM) were retrieved fromthe prestigious TCGA database (http://portal.gdc.cancer.gov) (http://tcga-data.nci.nih.gov/tcga/). Expression levels of FNDC3B in normal pancreatic tissues and other organs were obtained from the GTEx database and the National Center for Biotechnology Information (NCBI) Gene database, BioProject PRJEB4337 (http://gtexportal.org) (http://www.ncbi.nlm.nih.gov/gene/). TCGA and GTEx datasets were downloaded from the University of California, Santa Cruz Xena browser platform (http://xenabrowser.net/datapages). RNA expression profiles of FNDC3B were also collected from GEO database (http://www.ncbi.nlm.nih.gov/geo/) using accession numbers GSE16515, GSE28735, GSE62165, GSE62452, GSE15471 and GSE101448. Additional file 2: Table S1 provides comprehensive clinical data on PC patients included in this study based on TCGA dataset.

FNDC3B expression status in PC and normal pancreatic tissues

The expression of FNDC3B in PC and normal pancreatic tissues was compared using scatter plots and boxplots, along with the application of the Wilcoxon rank sum test and paired or unpaired student t-test. Visualization was achieved using the ggplot2 package (v3.3.6). To assess the diagnostic potential of FNDC3B in PC patients, receiver operating characteristic (ROC) curves were generated through pROC (v1.18.0) and ggplot2 (v3.3.6) package. Based on statistical ranking, FNDC3B expression above or below the median value was categorized as FNDC3B-high or FNDC3B-low groups.

Identification of differential expression genes (DEGs)

Using DESeq2 (v1.26.0), we leveraged TCGA database to discern DEGs between the FNDC3B-high and FNDC3B-low groups. The identified genes were considered significantly associated if the absolute log (fold change) exceeded 1.5 and the adjusted P < 0.05. Subsequently, we employed the ggplot2 package (v3.3.6) to visualize these findings through captivating volcano plots.

Enrichment of biological function and immune infiltration analysis

The enrichment of FNDC3B-associated DEGs was investigated using the Database for Annotation, Visualization and Integrated Discovery (DAVID) (http://david.ncifcrf.gov). Biological function evaluation of positive DEGs was performed through GO (geneontology.org) and KEGG (www.kegg.jp) enrichment analysis. To ensure significant differences, criteria were set as an enrichment factor > 1.5, a minimum count of 3, and a P < 0.01. GSEA was conducted using the FNDC3B differential expression matrix to identify potential pathways and biological processes associated with FNDC3B-related genes in patients stratified by high or low FNDC3B expression levels. Significance was determined when FDR < 0.25 and adjusted P < 0.05. ClusterProfile package (v4.4.4) was utilized for data visualization while gene sets were obtained from MSigDB database (https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp) [19]. The immune infiltration score of six immune cell types including B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages, and dendritic cells across pan-cancer was estimated using the TIMER method from IOBR (v0.99.9) package. Additionally, ssGSEA analysis based on gene expression profiles of 24 immune cell types [20], implemented via GSVA (v1.46.0) package, further explored immune infiltration patterns. Subsequently, the associations between FNDC3B mRNA expression levels and immune infiltration levels were assessed using Wilcoxon rank sum test and Spearman correlation analysis. Furthermore, ESTIMATE (v1.0.13) package was utilized to calculate Stromal Score, Immune Score, and ESTIMATE Score in order to explore the correlation between FNDC3B expression level sand stromal/immune scores. The quantification of Tumor immune dysfunction and exclusion (TIDE) can evaluate the sensitivity of immune checkpoint blockade by simulating tumor immune evasion mechanisms [21]. The TIDE score, along with T-cell dysfunction and exclusion scores for each patient, were determined using the online TIDE platform (http://tide.dfci.harvard.edu/).

The clinicopathological and prognostic value of FNDC3B, model construction and estimation

Initially, we evaluated the relationship between FNDC3B expression and clinicopathological variables using logistic regression and Wilcoxon rank sum test. Moreover, two-sided log-rank tests were employed to evaluate the prognostic significance of FNDC3B expression in PC patients and various subgroups based on overall survival (OS), disease-specific survival (DSS), and progression-free interval (PFI). To investigate the impact of FNDC3B expression on survival in conjunction with other clinicopathological factors (TNM stage, radiation therapy, primary therapy outcome, gender, race, age, residual tumor status, histologic grade, anatomic neoplasm subdivision), univariate and multivariate Cox analyses were conducted. A P < 0.05 was considered statistically significant while the median value of FNDC3B expression served as the cut-off for further analysis. Subsequently, a prognostic nomogram incorporating Cox analysis results along with FNDC3B expression was constructed to predict 1-, 2-, and 3-year survival outcomes using the survival (v3.3.1) package and RMS (v6.3.0) package (https://cran.r-project.org/web/packages/rms/index.html) and calibration plots were generated accordingly. Finally, time-dependent ROC curves were utilized to compare the prognostic efficacy of our novel prognostic model against pure FNDC3B expression using timeROC (v4.0) package in combination with ggplot2 (v3.6) package.

Cell culture and clinical specimens

The PC cell lines, including AsPC-1, BxPC-3, CFPAC, MIA PaCa-2, PANC-1, T3M4 and human pancreatic ductal epithelial (HPDE) cell line, were acquired from the American Type Culture Collection (ATCC) (Manassas, VA, USA). MIA PaCa-2, PANC-1 and T3M4 were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (Hyclone, Logan, UT, USA), BxPC-3 and AsPC-1 were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Hyclone, Logan, UT, USA) and CFPAC was cultured in Iscove’s Modified Dulbecco’s Medium (IMDM) (Hyclone, Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS) (Gibco, CA, USA) at 37 ℃ in a 5% CO2 cell culture incubator. Furthermore, seven paired samples consisting of tumor tissue and corresponding paratumor normal tissues collected from patients with PC who received surgical resection in the Second Affiliated Hospital of Zhejiang University (SAHZU) were procured for further study.

Immunohistochemistry (IHC)

Anti-FNDC3B polyclonal primary antibody (1:400 dilution; 22605-1-AP, Proteintech, Chicago, IL, USA) and a two-step staining kit (EnVision™ Detection System, Dako, Copenhagen, Denmark) were used to detect the expression of FNDC3B by IHC. The IHC procedure followed the methodology described in a previous study [22].

Cell transfection

After being seeded in 6-well plates at a density of 3 × 105 cells/well, the cells were cultured for 24 h. Once reaching 40% confluence, the cells were transfected with FNDC3B small interfering RNA (siRNA) and negative control siRNA (NC siRNA) following the manufacturer's instructions. The FNDC3B siRNAs, including sequences si1: GCATCCACGTGCAAAGGTA; si2: CGGAGAACGTGAACCAAGA; and si3: CCACTATTTTGT-AATGACA, as well as the retained sequence of NC siRNA, were purchased from RiboBio (Guangzhou, China). Subsequent assays were conducted on the respective cells after continuous culture for 24–48 h.

qRT-PCR

The total RNA was extracted from PC cell lines using TRIzol Reagent (15596026; Ambion, Life Technologies, Carlsbad, CA, USA), followed by first-strand synthesis utilizing a First-Strand Synthesis System for qRT-PCR (A6001, Promega, Madison, USA). The cDNA was then quantified by real-time PCR using a Veriti 96-well Thermal Cycler (4375786; Applied Biosystems, Foster City, CA, USA). Subsequently, PCR was conducted using a StepOnePlus™ system (Applied Biosystems) in accordance with the manufacurer’s instruction. The forward primer sequence of FNDC3B is 5′-GGCGGAATCCCCCATCAAA-3′, while the reverse primer sequence is 5′-ACCTCTCCGTTCAGCAATGG-3′. GAPDH served as the reference gene and its primer sequence was as follows: forward primer 5′-CGGAGTCAACGGATTTGGTCGTAT-3′, reverse primer 5′-AGCCTTCTCCATGGTGGTGAAGAC-3′. Fold changes relative to GAPDH were calculated using the − 2ΔΔCT method.

Western blot analyses

The detailed procedures of western blot analyses were analogous to a previous study [23]. The primary antibodies employed were as follows: rabbit anti-FNDC3B (1:1000 dilution; 22605-1-AP, Proteintech, Chicago, IL, USA), mouse anti-E-cadherin (1:1000 dilution; ab76055, Abcam, Cambridge, UK), rabbit anti-N-cadherin (1:5000 dilution; ab76011, Abcam, Cambridge, UK), rabbit anti-p21 (1:1000 dilution; ab109520, Abcam, Cambridge, UK), rabbit anti-cyclin B1 (1:50000 dilution; ab32053, Abcam, Cambridge, UK), rabbit anti-CDK1 (1:10000 dilution; ab172730, Abcam, Cambridge, UK) and rabbit anti-GAPDH (1:10000 dilution; ab181602, Abcam, Cambridge, UK).

Cell proliferation and colony formation assay

Cell Counting Kit-8 (CCK-8) assay was used to evaluate cell proliferation ability. Preprocessed cells were cultured at 3 × 103 cells/well in 96-well plates. And 10 μl/well CCK-8 reagent (Dojindo, Kumamoto, Japan) was added at 0, 24, 48, 72 and 96 h after cell adherence. Then the optical densities were measured at 450 nm (OD450) using microplate reader (Wellscan MK3; Thermo Labsystems, Helsinki, Finland). The values of OD630 were also measured as a reference. In the colony formation assay, 500 preprocessed cells were added to each well of 6-well plates and cultured in cell incubator for 14 days. Subsequently, the cells were first fixed with 4% paraformaldehyde for 20 min and then stained using crystal violet for 10 min before washed by phosphate buffered saline for three times. The colony were photographed and measured by Image J software (NIH, Bethesda, MD, USA).

Wound-healing and transwell migration and invasion assay

In wound-healing assay, the cells were cultured in 6-well plates at 3 × 105 cells/well in FBS-free medium. When cells reached 70–80% confluence, sterile pipette tips were used to scratch and form a “wound”. In transwell migration and invasion assay, we used non-coated and coated membranes in transwell chambers (24-well insert; 8-μm pore size; Corning Life Sciences, Corning, NY, USA) respectively. Preprocessed cells (3 × 104) were added in the upper chamber with FBS-free medium and medium containing 10% FBS was added to the lower chamber. After cultured in cell incubator for 48 h, a cotton swab was used to wipe the unpenetrated cells in the upper chamber, and the penetrated cells were fixed in methyl alcohol for 20 min and then subjected to hematoxylin and eosin staining 10 min and 5 min respectively for counting. The “wound” and the stained cells were observed and photographed via a DFC300FX microscope (Leica, Jena, Germany). The width of “wound” and cell number were further obtained using Image J software (NIH, Bethesda, MD, USA).

Statistical analysis

All the statistical analyses were conducted by RStudio software (v2022.12.0 + 353) (https://posit.co/download/rstudio-desktop/) and R software (v4.2.1). One-way analysis of variance (ANOVA) and two tailed Student t test were employed to analyze the data while other statistical methods were shown beforehand. All P < 0.05 were deemed as significantly difference.