FNDC4 Drives Metastasis and Immune Evasion in Pancreatic Cancer

FNDC4 is a key driver of pancreatic cancer invasiveness and immunosuppression that can be targeted to reprogram the pancreatic tumor microenvironment and suppress tumor metastasis, offering a promising therapeutic strategy.

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is highly metastatic and largely refractory to current therapies, underscoring the need to uncover the molecular drivers of progression to identify targetable vulnerabilities. In this study, we found that fibronectin type III domain–containing 4 (FNDC4), known for its role in macrophage polarization and metabolic regulation, was elevated in metastatic PDAC cells and correlated with poor patient outcomes. FNDC4 knockdown reduced tumor growth and metastasis in a diverse set of aggressive PDAC models. Mechanistically, FNDC4 enhanced cell cycle and apoptosis regulator 1 (CCAR1) stability, thereby sustaining CCAR1/β-catenin signaling. FNDC4 deficiency led to reduced CCAR1 and β-catenin expression and consequently impaired invasion and colony formation. Moreover, FNDC4 promoted immune evasion by driving macrophage polarization toward a protumorigenic M2 phenotype. FNDC4 loss shifted macrophage polarization toward an antitumor profile and increased CD4⁺ and CD8⁺ T-cell infiltration. Together, the effects of FNDC4 targeting resulted in reduced tumor burden, suppression of metastasis, and improved survival in immunocompetent murine PDAC models. Unexpectedly, FNDC4 localized to the nucleus, pointing to potential intranuclear activity. Transcriptomic and functional analyses further identified CCL5 as a critical downstream effector, required for recruiting CCR5⁺ T cells and mediating the immune effects of FNDC4 inhibition. Upstream, BHLHE40 directly activated FNDC4 transcription, which was stimulated by induction of epithelial–mesenchymal transition. Importantly, combining FNDC4 inhibition with claudin 18.2 chimeric antigen receptor T cells or chemotherapy resulted in enhanced tumor control compared with monotherapy. Together, these findings underscore the role of FNDC4 in promoting PDAC progression and the potential of FNDC4 as a target for innovative multimodal treatment strategies.

Significance:

FNDC4 is a key driver of pancreatic cancer invasiveness and immunosuppression that can be targeted to reprogram the pancreatic tumor microenvironment and suppress tumor metastasis, offering a promising therapeutic strategy.

Graphical Abstract

Introduction

Pancreatic cancer, particularly its most common subtype pancreatic ductal adenocarcinoma (PDAC), is among the most lethal malignancies, with a persistently dismal prognosis (1, 2). Despite accounting for approximately 2.6% of all cancer cases globally, PDAC is responsible for about 4.8% of all cancer-related deaths, reflecting its aggressive nature and poor survival rates. The 5-year survival rate remains below 5% across all stages, with only 0.9% for nonresectable patients and up to 17.4% for those undergoing surgical resection, primarily because most patients are diagnosed at an advanced stage, when curative treatment options are limited (3). Current therapeutic strategies, including surgical resection, chemotherapy, and radiotherapy, offer very limited efficacy, especially in the context of metastatic disease, which is present in more than 80% of patients at the time of diagnosis.

Pinpointing the upstream regulators that couple metastasis to immune evasion is now a central goal in PDAC research. Although several candidates have been proposed (4–6), the circuitry remains incompletely mapped. Metastatic spread hinges on epithelial–mesenchymal transition (EMT), a program that strips tumor cells of epithelial traits. This gives PDAC cells mobility while shielding them from immune surveillance, helping explain their marked resistance to immunotherapies (7–9). Understanding the underlying regulatory mechanisms could lead to the development of novel therapeutic strategies aimed at preventing metastasis and improving the survival of patients with PDAC.

In the present study, we identify and validate fibronectin type III domain–containing 4 (FNDC4) as a novel regulator of metastasis in PDAC. Although FNDC4 was originally identified for its involvement in metabolic processes and inflammation, more recent evidence has also implicated its potential role in the pathophysiology of some solid tumors (10–13). The full spectrum of its functions and mechanisms of action in cancer, particularly PDAC, had remained unknown. Our findings now demonstrate that FNDC4 plays a crucial role in PDAC progression by enhancing metastatic capacity and fostering immune evasion through cell cycle and apoptosis regulator 1 (CCAR1)/Wnt signaling. Targeting FNDC4 improved response to chemotherapy or claudin 18.2 (CLDN18.2) chimeric antigen receptor (CAR) T-cell therapy, thereby providing a novel multimodal therapeutic approach for combating this deadly disease.

Materials and Methods

Culture of primary human PDAC cells

Primary PDAC cells were derived from tumors (SiC020, 021, and 032) and circulating tumor cells (CTC; SiC002, 003, 005, and 007) following patients’ written informed consent. The studies were conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Shanghai Jiao Tong University School of Medicine (reference 2013-0905-70). Cells were cultured in RPMI medium supplemented with 10% FBS (#10270106, Gibco) and penicillin/streptomycin (#15140122, Invitrogen; ref. 14). Cells were authenticated annually by short tandem repeat DNA fingerprinting and screened for Mycoplasma contamination every 3 months by PCR, and only cultures testing negative were used in experiments. All experiments were performed using cells with passage numbers ≤10, and cell stocks were refreshed from low-passage frozen aliquots as needed.

Colony formation assay

A total of 2 × 10³ PDAC cells were seeded in 6-well plates to evaluate PDAC cell proliferation ability. After 14 days, 1% crystal violet stain solution was used to fix the PDAC cells, and the number of colonies was counted.

Sphere formation assay

Primary cell suspensions were cultured for 7 days in an ultralow attachment plate (#3471, Corning) at a density of 10,000 cells/mL in DMEM-F12 supplemented with 1× B-27, 20 ng/mL bFGF, penicillin/streptomycin, and amphotericin B (#15290026, Thermo Fisher Scientific).

In vitro transfection of siRNA or plasmids

Each well of a 24-well plate was seeded with 5 × 10⁴ cells overnight. Lipofectamine 3000 transfection reagent (#L3000015, Thermo Fisher Scientific) was used to deliver siRNA at a final concentration of 10 nmol/L. Cells were collected 48 hours following transfection for qRT-PCR measurement.

Cell fractionation

As previously described (15), cells were lysed in fractionation buffer (20 mmol/L HEPES, pH 7.4; 10 mmol/L KCl; 2 mmol/L MgCl₂; 1 mmol/L EDTA; 1 mmol/L EGTA; 1 mmol/L dithiothreitol; and protease inhibitors) and incubated on ice for 15 minutes. The cell suspension was passed through a 27-gauge needle 10 times, followed by centrifugation to pellet the nuclei. The supernatant was then centrifuged at 10,000 × g to pellet organelles including the endoplasmic reticulum, Golgi apparatus, and mitochondria, and the remaining fraction was further centrifuged at 100,000 × g to pellet the membrane fraction. All the pellets were washed once and lysed in TBS with 0.1% SDS before being processed for electrophoresis.

Western blot analysis

As previously described (15), tumor tissue or cultured cells were lysed in RIPA buffer (#89900, Thermo Fisher Scientific) supplemented with protease inhibitors (#11873580001, Sigma-Aldrich). The protein was resolved on 4% to 20% SurePAGE Bis-Tris protein gels (#M00655, GenScript) and transferred to polyvinylidene difluoride membranes (#ISEQ00010, Sigma-Aldrich). The membranes were blocked in Protein-Free Rapid Blocking Buffer (#PS108, Epizyme) for 10 minutes at room temperature, followed by an overnight incubation with primary antibodies at 4°C. After that, the membranes were incubated with secondary antibodies at room temperature for 1 hour, and the blots were developed using Clarity Western ECL Substrate (#1705061, Bio-Rad). ImageJ (RRID: SCR_003070) was used for band quantification.

Spheroid invasion assay

As previously described (15), appropriate numbers of spheres (diameter ≥40 μm) formed by primary PDAC cells and murine KPC cells were suspended in 300 μL RPMI (+ 20% FBS). The cell suspension was then mixed with 200 μL type I collagen (#354236, Corning), resulting in a final concentration of 1.5 mg/mL. The mixture was transferred into an ultralow attachment plate (#3473, Corning). Prior to this step, the collagen was neutralized with 3 μL of 1 mol/L NaOH. The mixture was allowed to solidify for 30 minutes at room temperature. Then 500 μL RPMI (+ 10% FBS) was added on top of the sphere–collagen gel. The plate was incubated for 24 hours, and the percentage of spheres with cells invading the surrounding matrix was determined.

Transwell migration and invasion assay

Transwell assays were performed using inserts (#3422, Corning) with an 8-μm pore size. For invasion assays, inserts were precoated with 100 μL Growth Factor Reduced Matrigel (#354230, Corning) and incubated for 2 hours at 37°C. For transmigration assays, uncoated inserts were used. After Matrigel incubation (in invasion assays), the excess Matrigel was aspirated, and cells were seeded into the upper chamber in serum-free medium. The lower chamber contained complete medium as a chemoattractant. Different primary human PDAC cell cultures and murine KPC–derived PDAC cells were seeded at the following densities and durations: SiC002 and SiC005 (2.5 × 10⁴ cells for 48 hours); SiC003 and CHX2000 (2.5 × 10⁴ cells for 24 hours). After incubation, nontransmigrated/invaded cells were removed from the upper surface using a cotton swab. Inserts were then fixed with 4% paraformaldehyde for 10 minutes at room temperature and stained with 0.1% crystal violet (#C6158, Sigma-Aldrich). Images were acquired using the EVOS imaging system (Thermo Fisher Scientific, RRID: SCR_023650), and the number of transmigrated or invaded cells was quantified.

qRT-PCR analysis

Total RNA was isolated using SteadyPure Quick RNA Extraction Kit (#AG21023, Agbio) according to the manufacturer’s instructions. cDNA synthesis was performed using the PrimeScript RT Reagent Kit with gDNA Eraser (#RR047A, Takara Bio). qRT-PCR analysis was conducted using the SuperReal PreMix Plus (SYBR Green #FP205, TIANGEN) and analyzed with the QuantStudio 6 Pro System (Thermo Fisher Scientific, RRID: SCR_020239). Relative mRNA expression was calculated using the ΔΔCt method. Primer sequences are provided in Supplementary Table S1.

In vivo tumorigenicity assay

PDAC cells were dissociated with trypsin, suspended in 20 μL Matrigel (#356234, Corning), and implanted subcutaneously into female 6 to 8-week-old BALB/c nude mice (RRID: IMSR_RJ:BALB-C-NUDE). Tumor growth was monitored for up to 6 weeks after implantation. Procedures were conducted in accordance with the Animals in Science regulations (Shanghai Jiao Tong University Project Approval A-2020-004).

In vivo PDAC models

Human PDAC cells (10⁵ in 20 μL Matrigel) were injected orthotopically into the pancreas of 6 to 8-week-old BALB/c nude mice, whereas murine KPC cells (500 or 5,000 in 20 μL Matrigel) were injected orthotopically into the pancreas or subcutaneously into the flanks of 6 to 8-week-old C57BL6/J mice (RRID: MGI:2159769), as described previously (16–18). For the metastasis model, cells were injected into the spleen and allowed to grow for 1 month before being harvested. All animal procedures were conducted in accordance with the 3Rs and the regulations for Animals in Science (Shanghai Jiao Tong University Project Approval A-2020-004).

In vivo delivery of siRNA

The siRNA was mixed with Entranster in vivo transfection reagent (#18668111, Engreen) in an appropriate volume of saline, following the manufacturer’s instructions. The mixture was injected intratumorally into subcutaneous tumors at the indicated dosage and frequency in certain experiments. Alternatively, in other experiments, the mixture was administered intraperitoneally.

Murine CLDN18.2 CAR T-cell generation and characterization

A second-generation CLDN18.2 CAR construct was engineered with murine CD8α as the signal peptide, the 8E5 anti-CLDN18.2 single-chain variable fragment as the binding domain (19), a FLAG tag for detection, murine CD8α as the hinge and transmembrane domain, 4-1BB (CD137) as the costimulatory domain, and CD3ζ for intracellular signaling. The CAR sequence was cloned into an murine stem cells virus retroviral vector (RRID: Addgene_24828) for stable transduction.

Flow cytometry analysis

Primary human PDAC cells or murine KPC cells were dissociated using TrypLE (#12604013, Thermo Fisher Scientific) and incubated with FcR blocking reagent (#130-092-575 for mouse or #130-059-901 for human, Miltenyi Biotec). The cells were then stained with the indicated antibodies and appropriate isotype-matched control antibodies, as listed in Supplementary Table S2 including RRID numbers, following the manufacturer’s instructions. DAPI (#G1012-10ML, ServiceBio) was used for exclusion of dead cells. Samples were processed using the Attune NxT flow cytometer (Invitrogen, RRID: SCR_019590) and analyzed using FlowJo V10 software (RRID: SCR_008520).

Immunofluorescence

Cells were fixed in 4% paraformaldehyde for 10 minutes at room temperature, followed by permeabilization with 0.1% Triton X-100 for 10 minutes at room temperature. Blocking was performed with 5% BSA for 1 hour at room temperature. Incubation with the indicated primary antibodies, as listed in Supplementary Table S2 including RRID numbers, was carried out at 4°C overnight, followed by incubation with fluorophore-conjugated secondary antibodies for 1 hour at room temperature. DAPI (#G1012-10ML, ServiceBio) was used as a counterstain for the nuclei. Samples were mounted in Fluoroshield mounting medium (#Ab104135, Abcam), and images were captured using a fluorescence microscope or confocal microscope (LSM 900, ZEISS, RRID: SCR_022263).

Cell counting kit 8 proliferation assay

Cell proliferation was assessed using the cell counting kit 8 assay (Beyotime, cat. #C0038) following the manufacturer’s instructions. PDAC cells were seeded in 96-well plates at a density of 2 × 10³ cells per well, depending on the cell culture, and allowed to adhere overnight. At the indicated time points, 10 μL of cell counting kit 8 reagent was added to each well, followed by incubation at 37°C for 2 hours. Absorbance was measured at 450 nm using a BioTek Synergy Neo2 Hybrid multimode microplate reader (BioTek, RRID: SCR_019765), with signal intensity proportional to the number of viable cells.

Peripheral blood mononuclear cell isolation and preparation of macrophage-conditioned medium

As previously described (15), human peripheral blood-derived mononuclear cells (PBMC) were obtained from healthy donors with informed consent. Monocytes were isolated by density gradient centrifugation using Ficoll–Paque Plus media (Cytiva, cat. #17144003) and cultured in Iscove’s Modified Dulbecco’s Medium (Invitrogen, cat. #12440061) supplemented with 10% human AB serum (Sigma-Aldrich, cat. #6914) in Nunc EasYFlask (Thermo Fisher Scientific, cat. #156499). M2 monocyte-derived macrophages were generated by treating cells with 0.5 ng/mL macrophage colony-stimulating factor (PeproTech, cat. #30025100) for 3 days, with the culture medium changed daily. From day 4 onward, when monocytes had differentiated into macrophages, the media was replaced with conditioned media (DMEM/F12 (BasalMedia, cat. #L320KJ) plus 20 ng/mL bFGF (PeproTech, cat. #10018C) plus 0.1× B-27 (Thermo Fisher Scientific, cat. #17504044) and cultured for two more days. The conditioned media was collected, centrifuged, and the supernatant used as macrophage-conditioned medium (MCM).

In vitro macrophage phagocytosis assay

Murine immortalized bone marrow–derived macrophages, kindly provided by Dr. Katherine Fitzgerald (UMass Medical School), or PBMC or THP-1 cells (Procell, cat. #CL-0233; RRID: CVCL_0006) activated with phorbol 12-myristate 13-acetate (Beyotime, cat. #S1819, 100 ng/mL for 3 days) were labeled with PKH26 (MaokangBio, cat. #MX4021), whereas cancer cells were labeled with PKH67 (MaokangBio, cat. #MX4023), following the manufacturer’s instructions. The macrophages were then mixed with the cancer cells at a ratio of 2:1 and seeded onto a 6-well plate. Images were taken at the indicated time points, and the phagocytic index was calculated as the number of phagocytosed cancer cells per 100 macrophages by fluorescent imaging and quantification.

Assessment of FNDC4 and CCL5 plasma concentration by ELISA

For FNDC4, plasma of patients with PDAC was collected from Ruijin Hospital, followed by ELISA (AdipoGen, cat. #AG-45B-0028-KI01) according to the manufacturer’s instructions for serum and plasma samples. For CCL5, CHX2000 cells were transfected with siRNA and seeded at 5 × 10⁵ cells per 6-well plate. The supernatants from each group were then harvested at 48 hours after transfection to measure mCCL5 concentrations by ELISA. The mCCL5 protein levels were measured using ELISA (Elabscience, cat. #E-EL-M0009) according to the manufacturer’s instructions for medium.

RNAscope ISH for FNDC4 mRNA detection

RNAscope ISH was used to detect FNDC4 mRNA expression in formalin-fixed, paraffin-embedded tissue sections. Tissue sections (5 μm) were mounted on positively charged slides, baked at 60°C for 1 hour, deparaffinized, and rehydrated. Hydrogen peroxide treatment, heat-induced antigen retrieval, and protease digestion were performed to optimize probe accessibility. Hybridization was conducted using a 1:1:1 mixture of three RNAscope probes targeting FNDC4 mRNA (probe sequences are provided in Supplementary Table S3), followed by signal amplification with the RNAscope Multiplex Fluorescent Reagent Kit (Bio-Techne, cat. #323100). For detection, the Opal 620 fluorophore (Akoya Biosciences, cat. #FP1495001KT) was used to generate a bright red fluorescence signal corresponding to FNDC4 mRNA expression. Nuclei were counterstained with DAPI, enabling precise spatial localization of fluorescent signals. High-resolution fluorescence microscopy was used for analysis, with FNDC4 mRNA expression visualized as distinct punctate red signals within individual cells. Signal intensity and distribution were evaluated across tissue regions, and positive and negative control probes ensured assay specificity and signal integrity.

T-cell isolation and activation

Murine CD3⁺ T cells were isolated from C57BL/6 splenocytes using the EasySep Mouse T Cell Isolation Kit (STEMCELL Technologies, cat. #19851). Purified T cells were stimulated with Dynabeads Mouse T-Activator CD3/CD28 (Thermo Fisher Scientific, cat. #11456D) in R10 medium [RPMI-1640 supplemented with 10% heat-inactivated FBS, 50 μmol/L β-mercaptoethanol (Sigma-Aldrich, cat. #M3148), 0.05 mg/mL gentamicin (MedChemExpress, cat. #HY-A0276A), and penicillin/streptomycin] and cultured at 1 × 10⁶ cells/mL with 80 U/mL IL2 for 24 hours before transduction.

Retroviral production and T-cell transduction

Retroviral supernatants were generated using Plat-E packaging cells transfected with the CAR construct. Murine T cells were transduced on RetroNectin-coated (Clontech, cat. #T100B) plates with viral supernatant, followed by centrifugation to enhance transduction efficiency. Transduced CAR-T cells were expanded in R10 medium supplemented with IL2 (R&D, cat. #202-IL-050, 50 U/mL), IL7 (Miltenyi Biotec, cat. #130-095-361, 10 ng/mL), and IL15 (Miltenyi Biotec, cat. #130-095-762, 10 ng/mL), with medium replenishment every 48 to 72 hours. CAR expression was validated by flow cytometry using a FLAG antibody (BioLegend, cat. #637309, RRID: AB_2563147).

The Cancer Genome Atlas dataset and gene set enrichment analysis

Transcriptomic data for patients with PDAC in The Cancer Genome Atlas (TCGA) database were downloaded from Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov/). Patients were grouped based on the expression of FNDC4. Gene set enrichment analysis (GSEA) tools (Java GSEA Desktop Application version 4.2, RRID: SCR_003199) were used to analyze key functional pathways enriched in different groups. An FDR < 0.05 was regarded as statistically significant. Gene Ontology and The Kyoto Encyclopedia of Genes and Genomes pathway analyses were performed using DAVID online tools (https://david.ncifcrf.gov/, RRID: SCR_001881). Differentially expressed genes (upregulated or downregulated) were analyzed separately, and P < 0.05 was considered statistically significant.

10× Genomics single-cell RNA sequencing

Primary PDAC cells were treated with MCM for 48 hours. Cells were captured, barcoded, and prepared for library construction according to the manufacturer’s instructions (10× Genomics, RRID: SCR_017344). The samples were sequenced on NovaSeq 6000 S4 flow cell (Illumina, RRID: SCR_010233). The Cell Ranger pipeline (v1.3, RRID: SCR_017344) and Seurat R package (RRID: SCR_016341) were used to analyze sequencing data in RStudio (RRID: SCR_000432; ref. 14). All raw FASTQ files and metadata have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession GSE184871.

Bulk RNA sequencing

Human PDAC cells were lysed in TRIzol (Thermo Fisher Scientific, cat. #15596026), and the lysates were processed by Genergy Biotech for library preparation and sequencing analysis. Differential expression analysis for genes across the different conditions was performed using DESeq2 (RRID: SCR_015687) with a threshold of |log₂FC| ≥1 and P ≤ 0.05. All raw FASTQ files and metadata have been deposited in the NCBI SRA under BioProject accession PRJNA1243292.

Tissue microarray and IHC

Tissue microarray slides from patients at different PDAC stages (Superchip, cat. #HPanA060CD02) were purchased from Superchip Company. Other tissue microarray samples were obtained from Ruijin Hospital with ethical approval and following informed consent (reference 2013‐0905‐70). FNDC4 IHC was performed on formalin-fixed, paraffin-embedded PDAC tumor sections using an FNDC4 antibody (ABclonal, cat. #A17758), followed by detection using the SABC‐HRP Kit (Beyotime, cat. #P0615) according to the manufacturer’s instructions. Staining results were processed using CaseViewer (3dhistech, RRID: SCR_017654) and the scores calculated using an IHC scoring system based on the product of the proportion of positive tumor cells (0–4) and staining intensity (0–3).

Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) was performed using SimpleChIP Plus Enzymatic Chromatin IP Kit (Cell Signaling Technology, cat. #9005) according to the manufacturer’s instructions. DNA fragments were sheared by sonication until they ranged from 200 to 500 bp. The nuclear lysate was immunoprecipitated with BHLHE40 (Cell Signaling Technology, cat. #45723; RRID: AB_3711224), PPARγ (Cell Signaling Technology, cat. #2435; RRID: AB_2166051), KLF6 (Merck, cat. #SAB2108745; RRID: AB_3711225), or IgG antibody (Cell Signaling Technology, cat. #2729; RRID: AB_1031062), see also Supplementary Table S2. The purified DNA fragments were analyzed by qRT-PCR with specific primers; the ChIP primers are provided in Supplementary Table S1.

Lentivirus production and transduction

The short hairpin RNA sequences targeting human FNDC4 and murine Fndc4 are provided in Supplementary Table S3. The vectors for short hairpin RNA and overexpression of FNDC4 [pLV(Exp)-mCherry:T2A:Puro-EF1A>{flag-FNDC4-myc] were purchased from VectorBuilder. Third-generation lentiviruses were generated in 293T cells (ATCC CRL-3216; RRID: CVCL_0063) using the respective lentiviral backbone, psPAX2 packaging plasmid (Creative Biogene, cat. #OVT2839), and pMD2.g (Creative Biogene, cat. #fgOVT2792) with polyethylenimine transfection reagent (Polysciences, cat. #23966-1). The viral particles were incubated with the target cells for 36 hours, then removed from the cell culture, and selective antibiotics added 24 hours later. The knockdown or overexpression efficiency was validated using qPCR or Western blotting.

Statistical analyses

The results are presented as mean ± SD, unless stated otherwise. Normality of the data was assessed with the Shapiro–Wilk and Kolmogorov–Smirnov tests. When a normal distribution was confirmed, parametric tests such as the t test or ANOVA were applied. Otherwise, nonparametric tests, including the Mann–Whitney U test for pairwise comparisons, were used. Survival analyses were assessed using the log-rank test. Statistical significance was defined as *, P < 0.05; **, P < 0.01; ***, P < 0.001, and ****, P < 0.0001. All statistical analyses were performed using GraphPad Prism 9.0 software (RRID: SCR_002798) or SPSS version 26 (RRID: SCR_002865).

Ethics and inclusion statement

Our research design and execution included the active involvement of local scientists, ensuring that diverse perspectives and expertise were appropriately considered throughout the study. The conducted research holds significant local relevance, addressing specific challenges within the local context and contributing to advancements in cancer research. We have developed plans to share the benefits of this research, including findings, data, and resources, with the scientific community, healthcare professionals, and the public. While adhering to local regulations, we ensured that ethical guidelines were followed throughout the study to uphold the welfare of both animals and human participants involved in the research process.

Key resource table

More detailed information about used materials can be found in the key resource table provided in the supplements.

Results

FNDC4 is overexpressed in invasive human PDAC and is associated with poor patient outcome

To identify genes critical for PDAC invasion, we analyzed single-cell RNA sequencing (scRNA-seq) data from primary human PDAC cultures derived from CTCs (SiC002, SiC003; ref. 20) and resected primary tumors (SiC021; ref. 21), with or without EMT-inducing MCM. Consistent with previous reports (22), MCM upregulated classical EMT-related markers and downregulated the epithelial marker CDH1 (E-cadherin) at the mRNA level in CTC-derived PDAC cultures (SiC002 and SiC005; Supplementary Fig. S1A and S1B). MCM treatment also significantly increased the invasive capacity of these PDAC cells (Supplementary Fig. S1C). Uniform Manifold Approximation and Projection analysis of the scRNA-seq data indicated increased FNDC4 expression in a variable fraction of MCM-treated PDAC cells compared with controls, suggesting that FNDC4 expression is upregulated during EMT in PDAC cells (Fig. 1A). The upregulation of FNDC4 expression was validated using qRT-PCR and Western blotting in five primary PDAC cultures following 48 hours of MCM treatment. Both mRNA and protein levels of FNDC4 were significantly increased, confirming transcriptional and translational upregulation of FNDC4 during EMT (Fig. 1B and C).

Figure 1.

FNDC4 expression in PDAC and association with outcome. A, Uniform Manifold Approximation and Projection (UMAP) of scRNA-seq data in primary PDAC cultures (SiC002, SiC003, and SiC021) after EMT-inducing MCM or control (Ctrl) treatment. B and C, FNDC4 expression by qRT-PCR (B) and FNDC4 protein levels by Western blot analysis (C) in five primary PDAC cultures following 48 hours of MCM treatment (n = 3 independent samples). D, Stage-specific FNDC4 expression in PDAC via GEPIA2 and survival analyses for OS and recurrence-free survival (RFS) in months after diagnosis based on high and low FNDC4 expression groups defined by the optimal cutoff in GEPIA2. E, FNDC4 expression in liver metastases compared with primary PDAC tumors analyzed across GEO datasets. F, FNDC4 IHC staining and scoring for healthy controls, pancreatitis, primary PDAC tumors (stage I to III), metastases (stage IV), and tumor-negative (Neg-LN) and tumor-positive (Pos-LN) lymph nodes. Representative images of each condition are shown. G, FNDC4 mRNA expression in paired PDAC tumors and adjacent normal tissues from the Ruijin Hospital cohort (n = 60); survival analysis of 58 patients using an optimal cutoff. FNDC4 protein expression by IHC (n = 56), with corresponding survival analysis based on median IHC score cutoff. ND, not determinable. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Analysis of the TCGA dataset demonstrates that FNDC4 mRNA expression is indeed markedly higher in PDAC tissues compared with the normal pancreas, with particularly elevated expression in stage IV tumors compared with lower stages (Fig. 1D; Supplementary Fig. S1D). Importantly, high FNDC4 mRNA expression correlated with worse overall survival (OS) and relapse-free survival in patients with PDAC, indicating its potential as a prognostic biomarker (Fig. 1D). Analysis of FNDC4 gene expression in metastatic settings revealed higher levels in liver metastases compared with primary tumors, as observed across different Gene Expression Omnibus (GEO) datasets (Fig. 1E).

Consistent with these findings, we examined FNDC4 expression in an independent patient cohort enrolled at Ruijin Hospital. IHC analysis revealed an increase in FNDC4 protein expression from normal and pancreatitis tissues to primary PDAC, lymph node metastases, and distant metastases (Fig. 1F). FNDC4 mRNA expression in tumor versus adjacent normal tissue in 60 paired samples was significantly upregulated in tumor tissues (Fig. 1G; Supplementary Fig. S1E), and high expression correlated with reduced OS (Fig. 1G). Similarly, elevated FNDC4 protein levels assessed by IHC also predicted poor patient outcomes (Fig. 1G; Supplementary Fig. S2A). In contrast, circulating FNDC4 levels did not show clear differences across disease stages or between patients with PDAC and healthy controls (Supplementary Fig. S2B), highlighting the tissue-specific role of FNDC4 in PDAC progression and patient outcomes. Together, data from TCGA, GEO, and the Ruijin cohort demonstrate that FNDC4 serves as a predictor of poor outcomes in PDAC.

FNDC4 promotes in vitro human PDAC invasion and aggressiveness

To investigate the functional role of FNDC4 in PDAC, we first assessed its expression following FNDC4 knockdown using two distinct shFNDC4 constructs. FNDC4 expression was significantly reduced in primary PDAC cultures at mRNA (Fig. 2A) and protein levels (Fig. 2B). This translated into a decrease in cell proliferation (Fig. 2C) and in vitro colony formation (Fig. 2D; Supplementary Fig. S3A), findings that were corroborated using two different siRNAs for FNDC4 (Supplementary Fig. S3B–S3D). Notably, FNDC4 knockdown did not affect early or late apoptosis (Supplementary Fig. S3E).

Figure 2.

In vitro effects on human PDAC cells following FNDC4 modulation. A and B, FNDC4 expression by qRT-PCR (A) and FNDC4 protein levels by Western blot analysis (B) after knockdown using two shFNDC4 constructs. C and D, Cell counting kit 8 (C) and colony formation (D) assays following FNDC4 knockdown in the primary PDAC culture SiC002. OD, optical density. E–G, Cell migration (E), transmigration (F), and invasion (G) assays after FNDC4 knockdown in the primary PDAC culture SiC002. H, FNDC4 expression by qRT-PCR following FNDC4 overexpression (OE). I–K, Proliferation (I), transmigration (J), and invasion (K) assays following FNDC4 overexpression in the primary PDAC culture SiC002. All functional assays were performed without MCM treatment (n = 3 independent samples). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FNDC4 knockdown also significantly impaired metastatic features of PDAC as demonstrated by reduced cell migration (Fig. 2E; Supplementary Fig. S3F), decreased transmigration (Fig. 2F; Supplementary Fig. S3G), and diminished invasive capacity (Fig. 2G; Supplementary Fig. S3H). Consistently, FNDC4 knockdown using siFNDC4 also significantly reduced the colony formation (Supplementary Fig. S3I) and invasive potential of PDAC cells (Supplementary Fig. S3J). These data suggest that FNDC4 promotes PDAC cell invasiveness, even without EMT induction. To explore the role of FNDC4 in other cellular contexts, we next examined its expression and function in 3D sphere cultures known to enrich for highly aggressive cancer stem cells (17). However, FNDC4 mRNA and FNDC4 protein levels were similar in spheres compared with adherent cells (Supplementary Fig. S4A). FNDC4 knockdown also did not significantly reduce sphere formation (Supplementary Fig. S4B) and had no effect on the content of CXCR4⁺CD133⁺ cancer stem cells (Supplementary Fig. S4C). These findings suggest that the role of FNDC4 in promoting PDAC cell invasiveness is independent of their self-renewal capacity.

To complement these loss-of-function studies, we overexpressed FNDC4 in our primary PDAC cultures. FNDC4 overexpression was confirmed by both qRT-PCR (Fig. 2H) and Western blotting (Supplementary Fig. S4D). Enhanced FNDC4 expression increased proliferation (Fig. 2I), colony formation (Supplementary Fig. S4E), cell migration (Supplementary Fig. S4F), transmigration (Fig. 2J; Supplementary Fig. S4G), and invasion (Fig. 2K; Supplementary Fig. S4H), further supporting its role in PDAC aggressiveness.

FNDC4 drives in vivo human PDAC progression and metastasis

To investigate the role of FNDC4 in PDAC progression in vivo, we examined subcutaneous tumors derived from primary PDAC cultures with FNDC4 knockdown, which displayed significantly reduced tumor weights compared with control groups (Fig. 3A; Supplementary Fig. S5A and S5B). IHC analysis revealed lower FNDC4 and Ki67 expression in FNDC4 knockdown xenograft tumors, suggesting reduced proliferative activity (Fig. 3B). In orthotopic PDAC models, FNDC4 knockdown also resulted in significantly smaller primary tumors and markedly fewer liver metastases compared with controls, whereas FNDC4 overexpression increased tumor burden and greatly enhanced metastatic spread (Fig. 3C). Consistently, FNDC4 knockdown reduced liver and lung metastases after intrasplenic and tail vein injections, respectively, whereas FNDC4 overexpression enhanced liver metastases following intrasplenic injection (Fig. 3D; Supplementary Fig. S5C and S5D). These findings demonstrate that FNDC4 strongly promotes human PDAC progression and metastasis in vivo.

Figure 3.

In vivo effects of FNDC4 on human PDAC progression and metastasis. A, Weights of subcutaneous tumors in shFNDC4 and control groups in two primary PDAC models. B, Representative images of FNDC4 and Ki67 IHC staining in xenograft tumors with FNDC4 knockdown or control. C, Primary tumors and liver metastases in an orthotopic PDAC model after intrapancreatic injection of primary PDAC cultures with FNDC4 knockdown, overexpression (OE), or control (Ctrl), with representative images with quantification of tumor weight and liver metastases. D, Liver metastases after intrasplenic injection of primary PDAC cultures with FNDC4 knockdown, overexpression, or control, with representative images and quantification of liver metastases. *, P < 0.05; **, P < 0.01.

FNDC4 promotes PDAC cell invasion by stabilizing CCAR1, thereby enhancing Wnt/β-catenin signaling

To elucidate the mechanism by which FNDC4 drives PDAC cell invasion, we first examined its localization and expression in primary PDAC cells. Immunofluorescence (IF) revealed that FNDC4 is predominantly localized in the nucleus of PDAC cells and, to a lesser extent, can also be detected on the cell membrane (Fig. 4A; Supplementary Figs. S6 and S7).

Figure 4.

Downstream effects of FNDC4. A, IF for FNDC4 in two primary PDAC cultures, shown alone or merged with DAPI. B, Subcellular fractionation and Western blot analysis of FNDC4 expression in SiC002 cells with shFNDC4 or shNC control. Full-length FNDC4 and C-terminal FNDC4 are shown alongside compartment-specific markers: E-cadherin (ECAD; membrane), calnexin [endoplasmic reticulum (ER)], histone H3 (nucleus), and GAPDH (cytosol). The same extracts were run on different gels, ensuring identical input across blots. Golgi, Golgi apparatus; Mito, mitochondria. C, Illustration of the FNDC4 construct featuring an N-terminal FLAG tag and a C-terminal Myc tag. Key domains include intracellular domain (C); fibronectin type III domain (FN III); hydrophobic domain (H); and signal peptide (SP). Immunoprecipitation for FNDC4 with FLAG and Myc tags in SiC002 cells with FNDC4 overexpression (OE), with coimmunoprecipitation for FLAG or Myc with target proteins and IgG as the control. D, Target protein expression in SiC002 PDAC cells with FNDC4 OE by Western blot analysis. β-Actin was used as the loading control. E, FNDC4, CCAR1, and β-catenin expression following FNDC4 knockdown in SiC002 and SiC005 PDAC cells by Western blot analysis. F, CCAR1 and β-catenin protein expression following CCAR1 knockdown in two primary PDAC models by Western blot analysis. G, Colony formation (left) and invasion assays (right) following CCAR1 knockdown in SiC005 cells. H, FNDC4, CCAR1, and β-catenin expression in SiC002 cells with or without FNDC4 OE and CCAR1 siRNA treatment. I, Colony formation (left) and invasion assays (right) with the cells used in H. All functional assays were performed without MCM treatment (n = 3 independent samples). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

This observation was corroborated by Western blotting using an antibody that distinguishes full-length FNDC4 from a smaller C-terminal fragment (CTF) in SiC002 primary PDAC cultures following knockdown of FNDC4. Full-length FNDC4 was specifically detected in the membrane fraction, whereas a smaller CTF was found in the nuclear fraction (Fig. 4B). The purity of each subcellular compartment was validated using established markers: E-cadherin (membrane), calnexin (endoplasmic reticulum), histone H3 (nucleus), and GAPDH (cytosol). Notably, both forms were reduced upon FNDC4 knockdown, but the nuclear signal of the CTF remained relatively high, with a nuclear-to-whole cell CTF-FNDC4 ratio of approximately 81% to 88%, comparable between FNDC4 knockdown and control cells.

These findings were further validated in FNDC4-overexpressing PDAC cells using a dual-tagged construct with Myc and FLAG epitopes to track the C- and N-terminal domains, respectively. Western blotting of subcellular fractions demonstrated distinct localizations: The C-terminal Myc-tagged fragment was enriched in the nucleus, whereas the N-terminal FLAG-tagged form localized to membrane-associated fractions (Supplementary Figs. S6 and S7). Quantitative analysis of the nuclear fraction revealed comparable nuclear enrichment of the CTF in FNDC4-overexpressing and control cells (∼83%), indicating that FNDC4 overexpression does not markedly alter the nuclear localization of the CTF (Supplementary Figs. S6 and S7). Confocal IF showed that FNDC4 is confined within the nucleus and clearly separates from lamin B1, confirming that FNDC4 resides intranuclearly rather than at the nuclear envelope (Supplementary Fig. S8A and S8B).

To identify potential FNDC4-interacting proteins, we performed immunoprecipitation using FLAG and Myc antibodies followed by mass spectrometry. Based on the resulting protein lists, we prioritized targets known to play roles in PDAC progression, including cell invasion, proliferation, and immune modulation. We then overexpressed a dual-tagged FNDC4 construct (FLAG at the N-terminal and Myc at the C-terminal) using a lentiviral vector. Coimmunoprecipitation confirmed specific interactions of the C-terminal Myc-tagged FNDC4 with AKAP8, RRBP1, IGF2BP2, and CCAR1, whereas the FLAG-tagged N-terminal FNDC4 did not show such interactions (Fig. 4C).

Among these, CCAR1 protein levels were notably increased upon FNDC4 overexpression, whereas RRBP1 and AKAP8 remained unchanged (Fig. 4D). FNDC4 knockdown reduced CCAR1 and β-catenin protein expression, supporting a regulatory link (Fig. 4E). Direct knockdown of CCAR1 also decreased β-catenin levels, confirming CCAR1 as a mediator of FNDC4-induced Wnt signaling (Fig. 4F; Supplementary Fig. S9A). These results demonstrate a pivotal role for CCAR1 in FNDC4-driven processes in PDAC and indicate that FNDC4 acts by stabilizing CCAR1, which functions as a coactivator of β-catenin in the Wnt/β-catenin pathway (23).

Notably, when PDAC cells were pretreated with cycloheximide to inhibit CCAR1 protein synthesis, FNDC4 overexpression still prolonged CCAR1 protein expression, whereas FNDC4 knockdown shortened its duration (Supplementary Fig. S9B). The effects of FNDC4 overexpression on CCAR1 and β-catenin were reversed by siCCAR1 treatment (Supplementary Fig. S9C). To further examine the role of FNDC4 in CCAR1 protein stability, the proteasome inhibitor MG132 was used to block protein degradation through the ubiquitin–proteasome system. MG132 increased FNDC4 protein levels approximately 2.5-fold compared with overexpression alone (Supplementary Fig. S9D). Western blotting revealed that MG132 treatment also elevated CCAR1 protein levels in FNDC4-overexpressing cells, suggesting that FNDC4 stabilizes CCAR1 by inhibiting its proteasomal degradation (Supplementary Fig. S9E). A time-course analysis confirmed this effect, showing that FNDC4 overexpression slowed CCAR1 degradation in the presence of MG132, reinforcing its role in extending CCAR1 protein stability (Supplementary Fig. S9E).

Functionally, CCAR1 knockdown significantly impaired colony formation, invasion, and proliferation (Fig. 4G; Supplementary Fig. S10A–S10D), thereby mimicking the effects of FNDC4 knockdown. Moreover, simultaneous FNDC4 overexpression and CCAR1 knockdown abolished the proinvasive and proliferative effects of FNDC4 overexpression (Fig. 4H and I; Supplementary Fig. S10B and S10D). Together, these results indicate that FNDC4 stabilizes CCAR1 protein, resulting in enhanced Wnt/β-catenin downstream signaling and promoting PDAC aggressiveness.

EMT-induced BHLHE40 directly activates FNDC4 transcription

To identify potential transcriptional regulators of FNDC4, we performed an integrative analysis using five independent RNA sequencing (RNA-seq) datasets, resulting in a common list of 26 candidate genes (Fig. 5A). After assessing their clinical significance using Kaplan–Meier survival analysis in PDAC patients, SIN3B, SP1, BHLHE40, HBP1, PPARG, and KLF6 were short-listed as potential upstream regulators (Fig. 5A; Supplementary Fig. S11A). Correlation analysis with FNDC4 expression narrowed the list to three candidates: BHLHE40, PPARG, and KLF6. Knockdown of KLF6 or PPARG by siRNA efficiently reduced their own transcripts and proteins, but neither FNDC4 mRNA nor FNDC4 protein levels were affected (Supplementary Fig. S11B and S11C). ChIP-qPCR analysis for these three candidates confirmed that only BHLHE40 binds to the FNDC4 upstream regulatory region (−1 kb from transcription start site) and acts as a primary transcriptional regulator of FNDC4 (Fig. 5B). Gel electrophoresis of immunoprecipitated DNA fragments verified BHLHE40 antibody binding to the FNDC4 upstream regulatory region (−1 kb; Supplementary Fig. S11D).

Figure 5.

Upstream transcriptional regulator of FNDC4. A, Venn diagram showing the overlap of five datasets and workflow identifying BHLHE40. DEG, differentially expressed genes. B, ChIP-qPCR for KLF6, PPARG (top), and BHLHE40 (bottom) binding to the FNDC4 upstream regulatory region [−1 kb from transcription start site (TSS)]. C, qRT-PCR for BHLHE40 and FNDC4 mRNA in SiC002 PDAC cells with or without EMT-inducing MCM treatment. D, Western blot analysis for BHLHE40 and FNDC4 proteins following MCM treatment in two primary PDAC models. E and F, qRT-PCR (E) and Western blot analysis (F) for BHLHE40 and FNDC4 expression after oncostatin M (OSM) treatment in two PDAC models. G and H, qRT-PCR in SiC002 cells (G) and Western blot analysis in SiC002 and SiC005 (H) for BHLHE40 and FNDC4 following BHLHE40 knockdown. I, Quantification of the invasion assay in SiC002 cells following BHLHE40 knockdown. J and K, qRT-PCR in SiC002 cells (J) and Western blot analysis in SiC002 and SiC005 cells (K) for BHLHE40 and FNDC4 after MCM or TGFβ treatment, alone or combined with TGFβ antibodies or siBHLHE40. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

We next evaluated BHLHE40 and FNDC4 expression following EMT induction. Treatment of SiC002 cells with EMT-inducing MCM led to a significant increase in both BHLHE40 and FNDC4 mRNA levels (Fig. 5C; Supplementary Fig. S11E). Western blot analysis corroborated these findings, showing elevated BHLHE40 and FNDC4 protein levels in two primary PDAC models following MCM treatment (Fig. 5D). Given the demonstrated role of the cytokine oncostatin M in promoting EMT in PDAC (15, 22), we also investigated its effect on BHLHE40 expression and subsequent FNDC4 upregulation. Oncostatin M treatment significantly upregulated BHLHE40 and FNDC4 at the mRNA level and the corresponding protein levels in two PDAC models (Fig. 5E and F; Supplementary Fig. S11F). Conversely, BHLHE40 knockdown using siBHLHE40 markedly prevented FNDC4 upregulation at both the transcript and protein levels, corroborating our finding that BHLHE40 is a key upstream inducer of FNDC4 (Fig. 5G and H; Supplementary Fig. S11G).

Functionally, reduced FNDC4 expression in response to BHLHE40 knockdown translated into significantly reduced invasiveness (Fig. 5I; Supplementary Fig. S12A). Treatment with either MCM or TGFβ alone resulted in increased FNDC4 expression, which was reversed by coadministration of TGFβ neutralizing antibodies or siBHLHE40 (Fig. 5J and K; Supplementary Fig. S12B–S12D). Finally, Kaplan–Meier analysis validated that increased BHLHE40 expression is associated with poorer OS (Supplementary Fig. S12E), mirroring our findings for FNDC4 (Fig. 1D). Together, these results establish BHLHE40 as a key upstream regulator of FNDC4 in PDAC.

FNDC4 drives progression of murine PDAC by reprogramming the tumor microenvironment

We next utilized murine PDAC models in fully immunocompetent mice to explore potential immunologic effects of Fndc4 modulation. Knockdown of Fndc4 reproduced the effects observed in human PDAC cells, including reduced proliferation, colony formation, and invasion (Supplementary Fig. S13A). Orthotopic murine PDAC models revealed a significant impact of Fndc4 knockdown on tumor progression with reduced tumor burden and virtually abrogated liver metastasis in the shFndc4 group compared with controls (Fig. 6A and B; Supplementary Fig. S13B). Histologic analysis confirmed the lack of metastatic nodules in the livers of mice injected with CHX2000 cells carrying shFndc4 knockdown.

Figure 6.

Immunologic effects of FNDC4. A, Explanted livers from mice injected with murine CHX2000 PDAC cells with shFndc4 knockdown or control (shNC), with arrows indicating metastatic nodules (left). Histologic images of explanted livers, with arrows indicating metastatic nodules (right). B, Quantification of primary tumor size and weight, and metastatic liver nodules. C, Flow cytometry analysis of the immune microenvironment in explanted PDAC tumors, showing percentages among live cells for CD4⁺, CD8⁺ T cells, macrophages, M1- and M2-type macrophages, and myeloid-derived suppressor cells (MDSC) in shFndc4 vs. control (shNC) tumors. TAM, tumor-associated macrophages. D, OS analysis of mice injected with CHX2000 cells carrying shFndc4 knockdown or shNC control. E, Flow cytometric quantification of M1- and M2-type macrophages following coculture of PBMC-derived macrophages with SiC002 PDAC cells carrying FNDC4 overexpression (OE) or control. F, Phagocytosis assay using PBMC-derived macrophages or THP-1–derived macrophages cocultured with SiC002 PDAC cells carrying FNDC4 overexpression, shFNDC4 knockdown, or control. The phagocytic index was measured as the percentage of macrophages that engulfed cancer cells. NC, negative control. G, qRT-PCR analysis of FNDC4-regulated target genes in SiC002 PDAC cells with FNDC4 knockdown using two different short hairpin RNA constructs or control (shNC), as well as in cells with FNDC4 overexpression. All functional assays were performed without MCM treatment unless indicated otherwise. Data shown in E–G represent n = 3 independent samples. ns, not significant; *, P < 0.05; ***, P < 0.001; ****, P < 0.0001.

Intriguingly, flow cytometry analysis of the tumor immune microenvironment revealed significant immune remodeling associated with Fndc4 knockdown, including increased infiltration of CD4⁺ and CD8⁺ T cells, a shift toward M1-type macrophages, and reduced M2-type macrophages compared with control tumors (Fig. 6C; Supplementary Fig. S13C). IHC also showed increased infiltration of CD4⁺ and CD8⁺ T cells, a higher presence of M1-type macrophages in Fndc4 knockdown tumors, a decrease in CD163⁺ M2-type macrophages, and fewer proliferative Ki67⁺ cells in Fndc4 knockdown tumors (Supplementary Fig. S14A). Fndc4 knockdown was confirmed at the histologic level using RNAscope (Supplementary Fig. S14A) and translated to improved OS (Fig. 6D).

As no pharmacologic compounds are currently available to target FNDC4, we used an in vivo siRNA approach to inhibit Fndc4 in fully established murine PDAC models and demonstrate its potential clinical utility. The efficiency of the siFndc4 treatment was demonstrated in vitro by qRT-PCR and in vivo by RNAscope (Supplementary Fig. S14B and S14C). Functionally, siFndc4 treatment also reduced proliferation, colony formation, and invasion (Supplementary Fig. S15A). Bioluminescence in vivo imaging (Supplementary Fig. S15B) revealed a significant reduction in tumor burden in mice treated with siFndc4 compared with controls, along with reduced tumor weight and diminished liver metastasis. This translated into significantly improved survival for mice treated with siFndc4 compared with controls (Supplementary Fig. S15C). Flow cytometry revealed remodeling of the immune microenvironment, characterized by increased CD8⁺ T cells and reduced M2 macrophages (Supplementary Fig. S15D).

To further extend these intriguing findings to human PDAC, we examined the effects of FNDC4 on immune cell modulation using human PDAC coculture models. Flow cytometry analysis demonstrated a significant decrease in the M1-type marker CD80 and a concurrent increase in the M2-type marker CD206 in PBMC-derived macrophages cocultured with FNDC4-overexpressing PDAC cells (Fig. 6E; Supplementary Fig. S16A), indicating a shift toward M2 macrophage differentiation. This was associated with a reduced phagocytic index in PBMC-derived and THP-1–derived macrophages (Fig. 6F; Supplementary Fig. S16B), whereas knockdown of FNDC4 enhanced phagocytic activity (Fig. 6F). Notably, these effects were not linked to relevant changes in major immune checkpoint molecule expression, e.g., CD24, CD47, IL34, or PD-1 (Fig. 6G). However, FNDC4 overexpression resulted in the robust upregulation of PD-L1 (Fig. 6G).

To better understand the paracrine mechanism by which FNDC4 remodels the tumor microenvironment (TME), we performed RNA-seq and found CCL5 to be consistently and significantly upregulated upon FNDC4 knockdown (Supplementary Fig. S16C). These findings were validated in human PDAC cells (SiC002 and SiC005; Supplementary Fig. S17A) and confirmed in murine CHX2000 cells at both the mRNA and protein levels (Supplementary Fig. S17B). Cosilencing of Ccl5 efficiently reduced the elevated Ccl5 mRNA levels induced by Fndc4 knockdown (Supplementary Fig. S17B), confirming a regulatory interaction between FNDC4 and CCL5.

Ccl5 knockdown in vivo did not significantly affect tumor burden but markedly diminished the antitumor effects observed upon Fndc4 knockdown (Supplementary Figs. S17C–S17E and S18A–S18C). The increased infiltration of CD4⁺ and CD8⁺ T cells induced by Fndc4 silencing was markedly diminished upon cosilencing of Ccl5, underscoring its critical role in mediating FNDC4-driven immune modulation (Supplementary Figs. S17D and S18B). Moreover, cosilencing of Ccl5 significantly reduced the intratumoral proportion of CCR5⁺CD4⁺ and CCR5⁺CD8⁺ T cells, indicating that recruitment of CCR5-expressing T cells is dependent on CCL5 and contributes to the antitumor immune response of Fndc4 knockdown (Supplementary Figs. S17E and S18C). These findings establish the FNDC4–CCL5–CCR5 signaling axis as a critical pathway governing T-cell infiltration and shaping the immune landscape of PDAC.

Fndc4 inhibition enhances response to both CAR T-cell therapy and chemotherapy

Although targeting Fndc4 alone already significantly extended the median survival of murine PDAC models, all mice eventually succumbed to disease progression. To further enhance the treatment effects, we first tested whether Fndc4 inhibition could improve the efficacy of CLDN18.2 CAR T-cell therapy (Fig. 7A; Supplementary Fig. S19A and S19B). The combination achieved greater tumor control compared with CAR T-cell therapy alone (Fig. 7B and C; Supplementary Fig. S19C), with IHC showing that in the CAR-T cohort, the combination of CAR-T cells with siFndc4 increased CD4⁺ T-cell infiltration, reduced CD163⁺ M2 macrophages, and lowered Ki67⁺ proliferation compared with CAR-T cells alone, whereas CD8⁺ T cells and total F4/80⁺ macrophages were not significantly altered (Supplementary Fig. S19D). We next combined Fndc4 inhibition with the standard-of-care chemotherapeutic agent gemcitabine (Fig. 7D), which led to significantly greater tumor reduction than gemcitabine treatment alone (Fig. 7E and F; Supplementary Fig. S20A), with IHC showing that in the gemcitabine cohort, the combination of gemcitabine with siFndc4 significantly increased CD4⁺ and CD8⁺ T-cell infiltration compared with gemcitabine alone, whereas total F4/80⁺ macrophages, CD163⁺ M2 macrophages, and Ki67⁺ proliferating cells showed no significant changes (Supplementary Fig. S20B). Together, these findings highlight Fndc4 inhibition as a promising approach to improve PDAC treatment efficacy.

Figure 7.

In vivo effects for treatments combined with FNDC4 inhibition. A, Schematic of the treatment regimen combining CLDN18.2 CAR T-cell therapy and siFndc4 in a murine PDAC model. B, Tumor growth curves (n = 5 mice per group; mean ± SEM). C, Tumor volume and weight on day 42 (n = 5; mean ± SD). D, Illustration of the treatment strategy integrating gemcitabine (GEM) and siFndc4 in a murine PDAC model. E, Tumor growth curves (n = 5 mice per group; mean ± SEM). F, Quantification of tumor volume and weight on day 42 (n = 5; mean ± SD). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Discussion

We demonstrate that FNDC4 is unexpectedly enriched in the nucleus of PDAC cells and drives CCAR1/β-catenin signaling, skews macrophages toward an M2 phenotype, suppresses CCL5-mediated T-cell recruitment, and—when silenced—sensitizes tumors to both CLDN18.2 CAR-T cells and gemcitabine.

Pathophysiologic context

FNDC4 belongs to a small fibronectin type III family better known for metabolic and anti-inflammatory roles; here, we connect it for the first time to PDAC aggressiveness. Upstream, the EMT transcription factor BHLHE40 binds the FNDC4 upstream regulatory region, coupling EMT to FNDC4 overexpression. Downstream, FNDC4 stabilizes CCAR1, which in turn cooperates with β-catenin to promote invasion, whereas its secreted fraction polarizes macrophages toward an immunosuppressive M2 state and dampens CCL5, blunting CCR5⁺CD4⁺/CD8⁺ T-cell infiltration. These intertwined tumor-cell and microenvironmental effects explain why siFndc4 alone restrains tumor growth and why combining it with CLDN18.2 CAR-T cells or gemcitabine produces markedly greater tumor control in vivo, positioning the BHLHE40–FNDC4 axis as a tractable therapeutic target in PDAC.

The family of fibronectin type III proteins includes FNDC1, FNDC3, FNDC4, and FNDC5 and is characterized by fibronectin type III domains that are integral to cell adhesion and signaling. Members of this family participate in diverse physiologic processes, such as metabolism, inflammation, and tissue repair, with emerging implications in cancer progression. For example, whereas FNDC5 is renowned for generating irisin to regulate energy metabolism, FNDC4 has attracted attention for its anti-inflammatory properties that influence macrophage polarization and metabolic regulation.

By comparison, FNDC4 is a relatively novel protein, and its metabolic functions have only recently begun to receive increasing interest. Bosma and colleagues (13) demonstrated that FNDC4 regulates the activity of inflammatory macrophages in the context of infections and tissue injury by suppressing the production of pro-inflammatory cytokines such as TNFα, IL6, and IL1β. In addition, FNDC4 promotes the polarization of macrophages toward the M2 phenotype, which is associated with tissue repair and anti-inflammatory functions, as opposed to the pro-inflammatory M1 phenotype. These observations suggest that FNDC4 could have therapeutic applications in inflammatory bowel diseases, including Crohn disease and ulcerative colitis, in which modulation of macrophage activity may reduce intestinal inflammation and facilitate tissue healing.

FNDC4’s contribution to oncogenesis is only now coming into view, and the protein’s well-documented anti-inflammatory and metabolic actions seem likely to shape tumor behavior as well. In some cancer types, FNDC4 has been implicated in promoting cell survival, proliferation, and migration—processes critical for metastasis (10, 24, 25). Our data now reveal that FNDC4 is markedly overexpressed in advanced PDAC and correlates with poor patient prognosis. Moreover, FNDC4 knockdown in primary PDAC models significantly reduced tumor growth and metastasis across subcutaneous, orthotopic, and metastatic xenograft systems, reinforcing its role in disease progression. Crucially, FNDC4 expression did not increase in cancer stem cell–enriched spheres, and its silencing left anchorage-independent 3D sphere growth unchanged, implying that FNDC4 amplifies PDAC aggressiveness in a broad sense rather than by acting on a specific stem-like subpopulation.

Immune reprogramming

Given FNDC4’s ability to modulate immune responses—specifically by driving macrophage differentiation toward an M2-like, protumorigenic phenotype—it may foster an immunosuppressive TME. Our findings show that in PDAC, FNDC4 not only promotes M2 macrophage polarization but also suppresses the differentiation of antitumor M1 macrophages. In a murine immunocompetent PDAC model, Fndc4 knockdown shifted macrophage polarization toward an antitumor profile, characterized by reduced M2 and increased M1 macrophages, thereby enhancing innate immune responses. This reprogramming was accompanied by increased intratumoral infiltration of adaptive immune cells, including CD4⁺ and CD8⁺ T cells, which further contributed to reduced tumor burden, diminished metastasis, and improved OS.

Mechanistic pathway

Our investigation reveals BHLHE40 as a key upstream and direct transcriptional regulator of FNDC4. Our data indicate that BHLHE40 binds to the FNDC4 promoter and drives its expression, thereby linking FNDC4 to the EMT process. Downstream, FNDC4 enhances PDAC cell invasion by stabilizing CCAR1 and thereby sustaining Wnt/β-catenin transcriptional output. FNDC4 knockdown significantly reduced the levels of CCAR1 and β-catenin, impairing both invasive behavior and colony formation. As a multifunctional protein, CCAR1 acts as a transcriptional coactivator, particularly within the Wnt/β-catenin and estrogen receptor signaling pathways. These observations align with previous reports implicating the CCAR1–β-catenin axis in the maintenance of cancer stemness and chemotherapy resistance (26–29).

Subcellular fractionation and IF unexpectedly showed nuclear localization of FNDC4, pointing to an additional layer of regulation and hinting at possible intranuclear functions for FNDC4 in cancer cells.

By establishing FNDC4 as a novel driver of metastasis downstream of BHLHE40, our findings suggest that targeting the BHLHE40–FNDC4 axis offers new therapeutic opportunities in PDAC, particularly for mitigating EMT-related tumor progression. Importantly, both BHLHE40 and FNDC4 have emerged as strong prognostic indicators in PDAC, as corroborated by previous studies and confirmed in our work (30, 31).

CCR5-mediated control of adaptive immunity

Our data uncover a previously unrecognized immunomodulatory circuit in which FNDC4 governs T-cell recruitment through CCL5. Bulk RNA-seq, confirmed by qRT-PCR and ELISA, showed that silencing FNDC4 markedly upregulates CCL5. Functional cosilencing revealed that CCL5 is indispensable for attracting CCR5⁺CD4⁺ and CCR5⁺CD8⁺ T cells into the TME: Simultaneous knockdown of Ccl5 markedly diminished the siFndc4-induced increase in CD4⁺ and CD8⁺ T-cell infiltration, particularly CCR5⁺ T-cell subsets, thereby attenuating the immune remodeling effects. Together, these findings establish the FNDC4–CCL5–CCR5 axis as a key regulator of adaptive immune cell recruitment and antitumor immunity in PDAC. Notably, alternative approaches that boost CCL5, e.g., treatment with the multikinase inhibitor anlotinib, likewise enhance CD8⁺ T-cell infiltration in lung cancer, underscoring the broader therapeutic potential of CCL5 upregulation (32).

Therapeutic implications

Given that FNDC4 not only promotes PDAC cell proliferation and invasion but also modulates the TME, we hypothesized that siFndc4 could enhance the efficacy of CLDN18.2 CAR T-cell immunotherapy. CLDN18.2, a tight junction protein normally restricted to gastric mucosal cells, is aberrantly overexpressed in several malignancies, including PDAC (33). Early-phase clinical studies have shown partial responses in only 16.7% of patients (33, 34), underscoring the need for combination strategies to fully exploit CAR T-cell immunotherapy in solid tumors. Our data now highlight the potential of FNDC4 inhibition to further enhance the therapeutic effects of CLDN18.2-targeted CAR-T cells by increasing CD4⁺ T-cell infiltration and reducing M2 macrophages and of gemcitabine by enhancing both CD4⁺ and CD8⁺ T-cell infiltration.

Importantly, combining siFndc4 with the standard-of-care chemotherapeutic agent gemcitabine yielded encouraging tumor control. Future studies should focus on optimizing the sequence and dosing of this combination to achieve synergistic effects, and long-term survival analyses will be essential to determine the durability of these responses. Together, our findings identify FNDC4 as a nuclear-localized, immunomodulatory oncogenic driver that promotes PDAC progression through macrophage polarization, Wnt pathway activation, and repression of CCL5-mediated recruitment of CCR5⁺ T cells. Our findings further indicate that FNDC4 may act both in a soluble form through receptor interactions and in the nucleus with functions yet to be defined.

Limitations

First, soluble FNDC4 has been shown to bind the metabolism-related orphan G protein-coupled receptor 116 (GPR116) in adipose tissue (11), raising the possibility of receptor-mediated tumor–metabolism interactions that were beyond the scope of our models. To date, the receptor responsible for FNDC4’s effects in immune cells has not been identified, leaving a critical signaling step unresolved. Second, our demonstration of nuclear FNDC4 in PDAC cells, described here for the first time, points to potential intranuclear functions that were not addressed in this study. An important open question is whether the observed repression of CCL5 reflects a direct nuclear effect of FNDC4 on transcription or an indirect consequence of altered secretory signaling, an issue that will require further dissection of nuclear versus soluble contributions. Defining these receptor pathways and clarifying the role of nuclear FNDC4 will be important goals for future studies in PDAC.

Taken together, our results place FNDC4 at the nexus of invasion and immune evasion in PDAC, making its blockade an attractive addition to future multimodal therapies.

Supplementary Material

Supplementary Figure S1

Figure S1 shows FNDC4 expression and functional assays in PDAC cells, including macrophage-conditioned medium, EMT markers, invasion assays, TCGA analysis, and IHC.

Supplementary Figure S2

Figure S2 shows FNDC4 expression in human PDAC, with IHC scoring examples and plasma FNDC4 levels measured by ELISA.

Supplementary Figure S3

Figure S3 shows proliferation, migration, invasion, apoptosis, and colony formation assays after FNDC4 knockdown in PDAC cells.

Supplementary Figure S4

Figure S4 shows sphere formation, CD133⁺CXCR4⁺ cell analysis, and functional assays after FNDC4 overexpression in PDAC cells.

Supplementary Figure S5

Figure S5 shows xenograft tumor growth and lung metastasis after FNDC4 knockdown or overexpression in PDAC cells.

Supplementary Figure S6

Figure S6 shows subcellular fractionation of FNDC4 in PDAC cells with marker proteins and schematic construct design.

Supplementary Figure S7

Figure S7 shows quantification of full-length and C-terminal FNDC4 bands from the fractionation blots.

Supplementary Figure S8

Figure S8 shows immunofluorescence staining of FNDC4, FLAG, Myc, and lamin B1 in PDAC cells.

Supplementary Figure S9

Figure S9 shows CCAR1 mRNA and protein regulation by FNDC4, with β-catenin analysis and proteasome inhibition assays.

Supplementary Figure S10

Figure S10 shows functional assays after CCAR1 knockdown or combined FNDC4 overexpression and CCAR1 knockdown.

Supplementary Figure S11

Figure S11 shows analysis of FNDC4 regulators, including survival, correlation, knockdown of KLF6/PPARG, ChIP-PCR for BHLHE40, and qRT-PCR after cytokine treatments.

Supplementary Figure S12

Figure S12 shows BHLHE40 regulation of FNDC4 and invasion assays, including responses to TGF-β and patient survival analysis.

Supplementary Figure S13

Figure S13 shows proliferation, invasion, and in vivo tumor growth after Fndc4 knockdown in murine CHX2000 cells.

Supplementary Figure S14

Figure S14 shows IHC, qRT-PCR, and RNAscope analysis of Fndc4 in orthotopic CHX2000 tumors with Fndc4 knockdown.

Supplementary Figure S15

Figure S15 shows proliferation, survival, tumor growth, and immune profiling in mice treated with siFndc4.

Supplementary Figure S16

Figure S16 shows macrophage co-culture, phagocytosis assays, and RNA-seq analysis after FNDC4 manipulation.

Supplementary Figure S17

Figure S17 shows CCL5 and FNDC4 regulation in human and murine PDAC cells, and tumor growth with siFndc4, siCcl5, or both.

Supplementary Figure S18

Figure S18 shows tumor progression and immune profiling in orthotopic CHX2000 PDAC treated with siFndc4, siCcl5, or both.

Supplementary Figure S19

Figure S19 shows CAR design, CAR expression in T cells, tumor growth, and IHC in siFndc4, CAR-T, and combination groups.

Supplementary Figure S20

Figure S20 shows tumor growth and IHC analysis of immune and proliferating cells after gemcitabine and siFndc4 treatment.

Key Resource Table

Table listing all relevant reagents including RRID where applicable.

Supplementary Table 1

Primers used for qPCR and ChIP-qPCR assays in this study.

Supplementary Table 2

Antibodies used in immunofluorescence (IF), flow cytometry (FC), immunohistochemistry (IHC), Western blot (WB), and chromatin immunoprecipitation (ChIP).

Supplementary Table 3

Sequences of shRNAs, siRNAs, and RNAscope probes used for knockdown and in situ hybridization experiments.

Data Availability

Expression data from human PDAC and normal tissues were obtained from TCGA, the Genotype-Tissue Expression Project database, and GEO (GSE71729, GSE42952, and GSE19279) and analyzed using the webserver Gene Expression Profiling Interactive Analysis 2 (TCGA and the Genotype-Tissue Expression Project databases; http://gepia2.cancer-pku.cn/, RRID: SCR_026154). Bulk RNA-seq data are publicly available in SRA under the accession number PRJNA1243292. scRNA-seq data are available in GEO under the accession numbers GSE184871, GSM5599107, and GSM5599108. All other data are available upon request to the corresponding author. This study did not generate new code.

Authors’ Disclosures

K. Jiang reports personal fees from government agency during the conduct of the study. A. Aicher reports grants from Ministry of Science and Technology, Taiwan, during the conduct of the study. No disclosures were reported by the other authors.

Authors’ Contributions

J. Li: Data curation, formal analysis, investigation, visualization, writing–original draft. R. Wu: Data curation, formal analysis, investigation. X. Jin: Data curation, formal analysis, investigation. Y. Yang: Data curation. K. Jiang: Data curation, formal analysis. Y. Wang: Conceptualization, supervision. S. Huang: Data curation, formal analysis. S. Tondi: Data curation, formal analysis. C.-H. Lai: Data curation, methodology. S. Dong: Data curation, formal analysis. B. Sabanovic: Data curation, formal analysis. M. Roberto: Data curation, formal analysis. C. Wen: Resources. Y. Jiang: Resources. D. Fu: Resources, supervision. A. Aicher: Formal analysis, visualization, methodology, writing–review and editing. B. Shen: Conceptualization, supervision. C. Heeschen: Conceptualization, resources, formal analysis, supervision, funding acquisition, visualization, writing–original draft, writing–review and editing.