CPNE1 promotes stemness and confers resistance to GPC3 CAR-T cell therapy in hepatocellular carcinoma via the STAT3–TGF-β signaling pathway Authors: Hao Zhang, Tong Xiang, Qiuzhong Pan, Mengjia Song, Lili Huang, Yao-Jun Zhang, Zili Hu, Xinyi Yang, Haoran Zhong, Yingzi Li, Jun Luo, Hao Chen, Song Gao, Chaopin Yang, Jian-Chuan Xia Published: 2026 ABSTRACT AbstractBackgroundHepatocellular carcinoma (HCC) remains a major global health burden with limited effective therapeutic strategies. Although chimeric antigen receptor (CAR) T-cell therapy has shown encouraging potential, its efficacy in HCC is profoundly constrained by the immunosuppressive tumor microenvironment.MethodsIn this study, we performed integrative analyses of HCC transcriptomic and clinical samples data to identify Copine-1 (CPNE1) as potential oncogenic drivers involved in tumor progression and immune suppression. Using functional and mechanistic assays in vitro with HCC cell line models and in vivo with mouse models, we examined the role of CPNE1 in regulating tumor proliferation, stemness, and its contribution to resistance against GPC3-targeted CAR-T cell therapy.ResultsWe found that CPNE1 is significantly overexpressed in HCC and correlates with poor patient prognosis. Functional assays revealed that CPNE1 enhances tumor proliferation and stemness by activating the STAT3–transforming growth factor beta (TGF-β) signaling pathway. Moreover, CPNE1-induced secretion of TGF-β promotes T-cell exhaustion, which impairs the efficacy of CAR-T cells. Knockdown of CPNE1 restored CAR-T cell function, enhanced T-cell infiltration into tumors, reduced exhaustion, and improved antitumor efficacy without causing systemic toxicity. Clinically, high CPNE1 expression was associated with lower T-cell infiltration and poor outcomes in HCC patients treated with immune checkpoint inhibitors.ConclusionsThese findings identify CPNE1 as a critical regulator of HCC progression and immune suppression, highlighting the CPNE1–STAT3–TGF-β axis as a promising therapeutic target in HCC. WHAT IS ALREADY KNOWN ON THIS TOPIC Hepatocellular carcinoma (HCC) currently lacks effective therapeutic options. During HCC development, aberrantly activated signaling pathways within tumor cells can trigger the excessive release of cytokines. These cytokines not only promote tumor cell proliferation and maintain cancer stemness but also suppress the antitumor activity of immune cells within the tumor microenvironment, which contribute substantially to malignant progression and the development of therapeutic resistance. Hepatocellular carcinoma (HCC) currently lacks effective therapeutic options. During HCC development, aberrantly activated signaling pathways within tumor cells can trigger the excessive release of cytokines. These cytokines not only promote tumor cell proliferation and maintain cancer stemness but also suppress the antitumor activity of immune cells within the tumor microenvironment, which contribute substantially to malignant progression and the development of therapeutic resistance. WHAT THIS STUDY ADDS CPNE1 is highly expressed in HCC and is associated with poor responses to immunotherapy. CPNE1 activates the STAT3 signaling pathway to drive excessive transforming growth factor beta secretion, which promotes tumor cell proliferation, maintains cancer stemness, and induces CAR-T cell exhaustion. Targeting CPNE1 suppresses tumor growth and enhances the therapeutic efficacy of CAR-T cells CPNE1 is highly expressed in HCC and is associated with poor responses to immunotherapy. CPNE1 activates the STAT3 signaling pathway to drive excessive transforming growth factor beta secretion, which promotes tumor cell proliferation, maintains cancer stemness, and induces CAR-T cell exhaustion. Targeting CPNE1 suppresses tumor growth and enhances the therapeutic efficacy of CAR-T cells HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY This study identifies a potential biomarker and therapeutic target for HCC. Targeting CPNE1 suppresses tumor progression while enhancing CAR-T cell antitumor immunity, providing a theoretical foundation for its future clinical application. This study identifies a potential biomarker and therapeutic target for HCC. Targeting CPNE1 suppresses tumor progression while enhancing CAR-T cell antitumor immunity, providing a theoretical foundation for its future clinical application. Introduction Hepatocellular carcinoma (HCC) is the most common primary liver malignancy, accounting for over 80% of all primary liver cancers. It ranks as the third-leading cause of cancer-related mortality and the sixth most frequently diagnosed cancer globally, with its incidence continuing to rise.1 2 The number of new cases is projected to reach 1.4 million by 2040, underscoring its growing impact as a major global health challenge.3 HCC is largely refractory to conventional chemotherapy and radiotherapy, making surgical resection the primary curative option. However, most patients are diagnosed at advanced stages, precluding surgical treatment. Furthermore, the lack of effective targeted therapies and immunotherapies, together with frequent development of therapeutic resistance, contributes to high rates of recurrence and distant metastasis.4 Therefore, novel therapeutic strategies capable of eliciting durable antitumor responses are urgently needed. Cancer stem cells (CSCs), a subpopulation of tumor cells with unlimited self-renewal properties, are major contributors to intratumoral heterogeneity and are strongly implicated in therapeutic resistance, recurrence, and metastasis.5 Furthermore, CSCs actively interact with the tumor microenvironment (TME), where they suppress T cell-mediated immunity to foster immune evasion.6 Consequently, mounting evidence suggests that targeting the core mechanisms underlying CSC maintenance offers a dual benefit: it directly restrains tumor progression and synergistically enhances the response to immunotherapies.7 Chimeric antigen receptor T cell (CAR-T) therapy, which involves engineering a patient’s T cells to express a receptor that specifically targets tumor antigens, has shown unprecedented success in hematologic malignancies.8 Glypican-3 (GPC3), a heparan sulfate proteoglycan, is specifically overexpressed on the surface of HCC cells, with little to no expression in normal adult tissues.9 In recent years, GPC3-targeted CAR-T cell therapy has entered early-phase clinical trials, demonstrating favorable safety profiles and encouraging preliminary antitumor activity.10 However, their therapeutic efficacy in HCC remains suboptimal, primarily hampered by the immunosuppressive TME, inadequate T cell persistence, and T cell exhaustion.11 Therefore, new strategies are needed to unlock the full potential of CAR-T therapy for HCC. CPNE1 is located on human chromosome 20 and encodes a 537-amino acid protein mainly localized in the cytoplasm.12 As a highly conserved calcium-dependent phospholipid-binding protein, CPNE1 contains two N-terminal C2 domains that mediate calcium-dependent membrane association and a C-terminal von Willebrand factor A domain that facilitates protein-protein interactions.13 These structural features enable calcium transport and suggest its involvement in intracellular signal transduction, membrane trafficking, and vesicle fusion.14 Previous studies have reported that CPNE1 is highly expressed in multiple types of cancers, including breast cancer, non-small cell lung cancer, renal cancer, and gastric cancer. Functionally, CPNE1 has been characterized as an oncogenic factor that promotes tumor progression and contributes to therapeutic resistance, thereby playing an important physiological role in cancer development.15–18 In this study, we sought to investigate the function of CPNE1 in regulating tumor stemness and CAR-T cell antitumor activity in HCC. By defining the CPNE1–STAT3–TGF-β axis, our work provides new mechanistic insights into strategies for suppressing cancer stemness and potentiating antitumor immunity, with the ultimate goal of improving therapeutic outcomes for HCC. Materials and methods Data sources and bioinformatics analysis RNA-seq expression data and survival information for HCC patients were obtained from The Cancer Genome Atlas (TCGA) and downloaded via the UCSC Xena platform. Additional transcriptomic datasets (GSE8763019 and GSE2509720) were retrieved from the Gene Expression Omnibus (GEO). Analyses of CPNE1 expression between tumor and normal tissues, correlations with clinicopathological features, survival outcomes, nomogram construction, and Cox regression modeling were performed using available online bioinformatics online analysis tool. Correlation analysis was performed based on CPNE1 expression levels in the TCGA-LIHC dataset. The 200 genes most strongly correlated with CPNE1 expression were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses to explore the biological processes and pathways associated with CPNE1. TCGA-LIHC dataset was stratified into high and low CPNE1 expression groups, and differential gene expression analysis was performed. Gene Set Enrichment Analysis (GSEA) was subsequently conducted, with |NES|>1, FDR<0.25, and adjusted p<0.05 considered statistically significant. KEGG pathway and GO enrichment analyses were applied to explore the potential biological functions of CPNE1. TCGA-LIHC dataset was conducted to analyze the association between CPNE1 expression and the infiltration levels of various immune cell subsets. In addition, the relationships between CPNE1 expression and TME parameters, including ESTIMATE score, stromal score, and immune score, were evaluated. All analyses were performed using available online bioinformatics online analysis website based on the R Studio V.3.6.3. Patient sample collection Fresh-frozen tumor and adjacent non-tumor tissue specimens were collected from eight HCC patients who underwent surgical resection at the Department of Liver Surgery, Sun Yat-sen University Cancer Center (Guangzhou, China). In addition, paraffin-embedded surgical resection tumor tissues from 51 HCC patients were obtained from the Department of Pathology, Sun Yat-sen University Cancer Center (Guangzhou, China). Furthermore, paraffin-embedded tumor samples from 28 HCC patients who had undergone immune checkpoint inhibitors (ICIs) prior to surgical resection were also collected from the Department of Liver Surgery, Sun Yat-sen University Cancer Center (Guangzhou, China). All 28 patients included in our analysis received anti-PD-1 therapy in combination with hepatic arterial infusion chemotherapy (HAIC) based on oxaliplatin, followed by surgical resection of HCC 4 weeks after treatment initiation. Cell culture The human HCC cell lines, including Huh7, Hep3B, MHCC97h, and SK-Hep1, as well as the human embryonic kidney cell line HEK293T, were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). All cell lines were authenticated and confirmed to be contamination-free. All cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Waltham, Massachusetts, USA), supplemented with 10% fetal bovine serum (FBS; Gibco, Waltham, Massachusetts, USA), and incubated in a 5% CO₂ atmosphere at 37°C. Fresh peripheral blood mononuclear cells (PBMCs) were isolated from the peripheral blood of healthy volunteers using density gradient separation with Lymphoprep (Stemcell, Vancouver, Canada). CD3+ T cells were subsequently isolated from PBMCs using the Human T Cell Isolation Kit (Stemcell, Vancouver, Canada). After activation with CD3 Dynabeads (Gibco, Waltham, Massachusetts, USA), the isolated T cells were cultured in X-VIVO 15 Medium (Lonza, Basel, Switzerland) supplemented with human-derived interleukin-2 (IL-2; Sihuan Biopharmaceutical, Beijing, China). Lentivirus packaging and cell infection The RNA interference and gene overexpression lentiviral vectors of the CPNE1 were constructed by the Obio Technology (Shanghai, China). The shRNA sequence is as follows: shRNA#1: CGCATGAATTTGCCTTGAATT, shRNA#2: CGCTTTGGAATCTATGACAA. The negative control (NC) sequence is TTCTCCGAACGTGTCACGTTTCAAGAGAACGTGACACGTTCG. The nucleotide sequence of GPC3 CAR, including GC33 scFv sequence, Myc-tag sequence, CD8 hinge domain, CD28 transmembrane, and intra domain, was synthesized and incorporated into the lentivirus vector by the Obio Technology (Shanghai, China). The constructed lentiviral plasmid, packaging plasmid psPAX2, and envelope plasmid pMD2.G were co-transfected into HEK293T cells by using Lipofectamine 3000 reagent (Thermo Fisher Scientific, Waltham, USA). At 24 and 48 hours after plasmid transfection, the culture supernatants containing lentiviral particles were collected from HEK293T cells. The collected supernatants were centrifuged to remove cell debris and subsequently filtered through a 0.45 µm filter. Lentivirus Concentration Solution (Yeasen, Shanghai, China) was then added to the filtered viral supernatant at a volume ratio of 1:4, followed by incubation at 4°C for 2 hours. After incubation, the mixture was centrifuged at 4000 g for 30 min. The supernatant was carefully discarded, and the viral pellet was resuspended in culture medium by gentle pipetting. The concentrated lentivirus was subsequently stored at −80°C until further use. For the cells lentiviral transfection, based on the titer of the concentrated lentivirus, the lentivirus was added to the cells, while the lentivirus volume was calculated for the requirements of the experiment. At the same time, the 1% of the total transfection volume of polybrene plus reagent (Obio Technology, Shanghai, China) was added to enhance the transfection efficiency of the lentivirus. Real-time quantitative PCR The total RNA of cells was extracted using the Universal RNA Extraction Kit (Takara Bio Inc, Kusatsu, Japan). The cDNA was synthesized from the extracted RNA through reverse transcription by Transcription Master Mix (ESscience, Shanghai, China), and then was performed to real-time quantitative PCR (RT-qPCR) by SYBR Green qPCR Master Mix (ESscience, Shanghai, China). The primer sequences are as follows: CPNE1 Forward: CCCTACCAACTTTGCACCCATC, CPNE1 Reverse: ATCCGTCACAGCACCATCAGTC. Cell viability assay The appropriate cells were inoculated into a 96-well plate, and then the corresponding drugs were carried out after the cells adhered to the 96-well plate. When the drug treatment was completed for 48 hours, CCK-8 reagents (NCM Biotech, Suzhou, China) were added to each well at a concentration of 10%. After incubation at 37 degrees for 2 hours, the absorbance at 450 nm (OD450) was recorded by an enzyme detector. Cell viability was calculated as a percentage of the untreated control group. Drug sensitivity was determined by three separate experiments, and the IC50 (the half maximal inhibitory concentration) was analyzed using a standard curve fitting procedure. Colony formation assay Approximately 1000–2000 cells were inoculated into the 6-well plate and incubated at 37°C with 5% CO2 cell incubator for 10 days. When clonal formation was observed under a microscope, 4% paraformaldehyde was added to fix the cells for 30 min, followed by 1% crystal violet staining for 30 min. The formed cell clones were photographed and counted. Spheroid formation assay Approximately 1000–2000 cells were inoculated into Ultralow Attachment 6-well plates (Corning, New York, USA). Then, they were incubated in DMEM/F12 medium (Gibco, Waltham, USA) supplemented with 2% B-27 supplement (Gibco, Waltham, USA), 20 ng/mL EGF (PeproTech, Waltham, USA), 10 ng/mL FGF (PeproTech, Waltham, USA), and 1% N-2 supplement (Gibco, Waltham, USA) at 37℃ with 5% CO2 in a cell incubator for 10 days. After observing the formation of tumor spheres under a microscope, the sphere count was conducted. Real-time live-cell analysis To assess the effect of CPNE1 on the proliferation of tumor cells, the mCherry-expressing tumor cells were seeded into a 96-well plate at a density of 2000 cells per well. Following cell attachment, the plate was loaded into the IncuCyte instrument (Sartorius, Göttingen, Germany), and the fluorescence signal was performed automatically every 3 hours for 72 hours. Tumor cell density within each well was quantified by measuring the mean fluorescence intensity (MFI). To assess the cytotoxic activity of CAR-T cells against tumor cells, the mCherry-expressing tumor cells were seeded into a 96-well plate at a density of 1.5×10⁴ cells per well. After the cells adhered, CAR-T cells were added at effector-to-target (E:T) ratios of 1:2 and 1:4. The plate was loaded into the IncuCyte instrument (Sartorius, Göttingen, Germany), and the fluorescence signal was performed automatically every 3 hours for 72 hours. Tumor cell density within each well was quantified by measuring the MFI. Western blot analysis After being washed twice with PBS, all cells were lysed using RIPA buffer (Solarbio, Beijing, China) supplemented with both protease and phosphatase inhibitor cocktails (NCM Biotech, Suzhou, China). Fresh tissue samples were homogenized in an appropriate volume of RIPA buffer using a frozen grinder until complete dissociation was achieved. The lysates were then centrifuged at 12 000 rpm for 15 min, and the supernatant was collected for protein concentration measurement using the BCA assay reagent (Thermo Fisher Scientific). The protein samples were separated by 10% SDS-polyacrylamide gel electrophoresis and electrophoretically transferred onto methanol-activated PVDF membranes (Millipore, Billerica, USA). The membranes were blocked with protein blocking buffer (NCM Biotech, Suzhou, China) for 20 min at room temperature, followed by incubation with primary antibodies overnight at 4°C with gentle shaking. Afterward, the membranes were washed three times with Tris-buffered saline containing 0.1% Tween-20 (TBST; Sigma-Aldrich, Billerica, USA) and then incubated with horseradish peroxidase (HRP)-conjugated secondary antibody for 1 hour at room temperature with shaking. Following additional TBST washes, protein bands were visualized using a chemiluminescence detection system (Bio-Rad, Hercules, USA). The information of all the antibodies used in the article is as follows: HRP-conjugated antirabbit IgG (Cell Signaling Technology, 7074); CPNE1 (Abcam, AB155675); SOX2 (Abcam, AB92494); OCT4 (Abcam, AB200834); Nanog (Abcam, AB109250); c-Myc (Abcam, AB32072); STAT3 (Abcam, AB68153); p-STAT3 (phospho Y705) (Abcam, AB76315); TGF-beta 1, 2, 3 (R&D Systems, MAB1835); GAPDH (Cell Signaling Technology, 2118); β-Tubulin (Cell Signaling Technology, 2146); Glypican 3 (Abcam, AB95363); CD3 (Abcam, AB135372); Granzyme B (Cell Signaling Technology, 46890). ELISA To quantify the secretion of TGF-β by tumor cells, 3×10⁶ cells were seeded per well of a 6-well plate. After 48 hours of culture, the supernatant was collected, centrifuged, and analyzed using a commercial ELISA kit (R&D Systems, Minneapolis, USA) according to the manufacturer’s instructions. GPC3 CAR-T cells and tumor cells were co-cultured in 6-well plates at a 1:1 ratio for 24 hours. The cell culture supernatant was then collected and centrifuged to remove cellular debris. Levels of IL-2, TNF-α, and IFN-γ secreted by GPC3 CAR-T cells were measured using commercially available ELISA kits (R&D Systems, Minneapolis, USA) according to the manufacturer’s protocols. Immunohistochemistry After tissue sections were baked at 65°C for 2 hours, they were deparaffinized with xylene, rehydrated through graded ethanol, and subjected to antigen retrieval using EDTA or sodium citrate buffer at 100°C for 10 min. Endogenous peroxidase activity was blocked by incubation with 3% hydrogen peroxide for 10 min at room temperature. The sections were then blocked with 10% goat serum for 1 hour at room temperature to prevent nonspecific binding, followed by incubation with primary antibodies at 37°C for 1 hour. After four washes with PBST (PBS containing 0.1% Tween-20), the sections were incubated with HRP-conjugated secondary antibodies (ZSGB-BIO, Beijing, China) for 30 min at room temperature and subsequently developed using a diaminobenzidine substrate kit (ZSGB-BIO, Beijing, China). Finally, the sections were counterstained with hematoxylin, dehydrated, mounted, and evaluated under a light microscope. Semiquantitative scoring of immunostaining was performed as follows: The proportion of positively stained tumor cells was scored as 0 (≤1%), 1 (2%–25%), 2 (26%–50%), 3 (51%–75%) or 4 (≥75%). Staining intensity was scored as 0 (negative), 1 (weak), 2 (moderate) or 3 (strong). The final IHC score was calculated by multiplying the proportion score by the intensity score for each specimen.21 Flow cytometry The PE anti-Myc-tag (71D10) antibody (Cell Signaling Technology 64025, Danvers, USA), BUV395 anti-human CD3 (SK7) antibody (BD Biosciences 564001, San Jose, USA), FITC anti-human CD4 (SK3) (BD Biosciences 566911), and APC anti-human CD8 (HIT8a) antibody (BD Biosciences 566852) were used to determine the positive rate of the CAR in the T cells isolated from PBMC. PE anti-Myc-tag (71D10) antibody (Cell Signaling Technology 64025, Danvers, United States), PE/Cyanine5 anti-human CD62L (DREG-56) antibody (BD Biosciences 561915), and BV421 anti-human CD45RO (UCHL1) antibody (BD Biosciences 562649) were used to determine memory T cells and effector T cells in CAR-T cells. BV605 anti-human PD-1 (EH12.1) (BD Biosciences 563245), BV421 anti-human LAG-3 (T47-530) (BD Biosciences 565720), and Alexa Fluor 647 anti-human TIM-3 (4S.B3) (BD Biosciences 565558) antibodies were used to determine T cell exhaustion. To assess the CD107a expression of CAR-T cell with tumor antigen stimulation, the CAR-T cells were incubated with tumor cells at a ratio of 2:1, along with the BV421 anti-human CD107a (H4A3) antibody (BD Biosciences 566261) and GolgiPlug protein transport inhibitor (BD Bioscience) for 6 hours. To determine the cytokines secretion levels of CAR-T cells with tumor antigen stimulation, the CAR-T cells were first incubated with tumor cells at a ratio of 2:1 with GolgiPlug protein transport inhibitor (BD Bioscience) for 6 hours. Then, these cells were subsequently fixed and permeabilized using the Fixation/Permeabilization Kit (BD Biosciences 554714) according to the manufacturer’s instructions, followed by intracellular staining with Alexa Fluor 647 anti-human IFN-γ (4S.B3) (BD Biosciences 563495, USA), PE anti-human TNF-α (MAb11) (BD Biosciences 562083, USA), and BV421 anti-human granzyme B (GB11) (BD Biosciences 563389, USA) antibodies. The spleens and xenografts of mice were prepared into single-cell suspensions. After erythrocyte lysis, the BV510 anti-human CD45 (HI30) (BD Biosciences 563204, USA) and PerCP-Cy 5.5 anti-human CD3 (SK7) (BioLegend 981008, USA) antibodies were used to detect human T cells. BV605 anti-human PD-1 (EH12.1) (BD Biosciences 563245, USA), BV421 anti-human LAG-3 (T47-530) (BD Biosciences 565720, USA), and Alexa Fluor 647 anti-human TIM-3 (4S.B3) (BD Biosciences 565558) antibodies were used to determine the exhaustion of tumor-infiltrating T cells. All results were collected using a CytoFLEX flow cytometer (Beckman Coulter, Brea, USA) and analyzed using FlowJo software V.10.8.1. Animal models All mice were purchased from GemPharmatech (Nanjing, China). Tumor volume was calculated using the formula: volume=(length×width²)/2. To evaluate the effect of CPNE1 knockdown on HCC growth, 5×10⁶ Huh7 cells stably transfected with CPNE1-shNC, CPNE1-sh1, or CPNE1-sh2 were suspended in 100 µL of PBS and mixed with 50 µL of Matrigel (Corning, NY, USA), and then subcutaneously injected into the flanks of 5-week-old NOD-Prkdcem26Cd52Il2rgem26Cd22/Gpt (NCG) mice (n=5 per group). Tumor growth was monitored weekly using caliper measurements. At the end of the experiment, tumors were excised, photographed, and weighed. To assess the effect of CPNE1 overexpression on HCC growth, 2×10⁶ SK-Hep1 cells stably transfected with CPNE1-Vector or CPNE1-OE were similarly prepared and injected subcutaneously into NCG mice (n=5 per group). Tumor growth was monitored weekly using caliper measurements. At the end of the experiment, tumors were excised, photographed, and weighed. To examine the impact of CPNE1 knockdown on the therapeutic efficacy of GPC3 CAR-T cells, 5×10⁶ Huh7 cells expressing CPNE1-shNC, CPNE1-sh1, or CPNE1-sh2 were suspended in 100 µL of PBS and mixed with 50 µL of Matrigel for subcutaneous implantation into NCG mice (n=5 per group). 10 days after tumor implantation, mice received intravenous injection of 4×10⁶ Mock T cells or GPC3 CAR-T cells via the tail vein. Tumor progression was monitored every 5 days using both caliper measurements and an IVIS Lumina Series imaging system (PerkinElmer, Waltham, USA) until the endpoint. To further investigate the effects of CPNE1 knockdown on CAR-T cell infiltration and function, the same xenograft model was established as described above. 14 days after tumor implantation, mice were intravenously infused with 4×10⁶ Mock T cells or GPC3 CAR-T cells. 10 days after T-cell infusion, mice were sacrificed, and tumors and spleens were collected for downstream analyses. Statistical analysis Bioinformatics analyses of TCGA-LIHC and GEO datasets were performed using R software (V.4.3.0) with appropriate statistical packages. Kaplan-Meier survival curves were generated to assess overall survival, and differences were evaluated using the log-rank test. Spearman’s correlation coefficient was used to assess associations between gene expression and immune parameters. All experimental data were analyzed using GraphPad Prism V.9.0 software. Statistical differences between two groups were determined using a two-tailed unpaired Student’s t-test, and comparisons among multiple groups were analyzed by one-way analysis of variance test. All data are presented as mean±SD unless otherwise specified. All experiments were independently repeated at least three times, and a p<0.05 was considered statistically significant (ns: p≥0.05, *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001). Results CPNE1 is upregulated in HCC and predicts poor clinical outcomes We conducted both unpaired and paired analyses of transcriptomic data from the TCGA LIHC cohort. CPNE1 was significantly upregulated in tumor tissues compared with normal tissues (figure 1A,B). This finding was further validated using independent GEO datasets, which consistently showed elevated CPNE1 expression in tumors (figure 1C,D). Analysis of clinical parameters revealed that higher CPNE1 expression was significantly associated with more advanced T stage and pathologic stage (figure 1E,F), but with no significant statistical difference in N stage and M stage (online supplemental appendix 1A). Moreover, elevated CPNE1 levels correlated with increased AFP expression (figure 1G). Kaplan-Meier survival analysis of the TCGA LIHC cohort demonstrated that higher CPNE1 expression was associated with poorer overall survival (OS) and progression-free survival (PFS) (figure 1H,I). To further validate the prognostic significance of CPNE1 in HCC, we performed immunohistochemical staining on tumor specimens from 51 patients who underwent surgical resection at Sun Yat-sen University Cancer Center. Based on staining intensity and distribution, patients were classified into CPNE1 high-expression and low-expression groups. Kaplan-Meier survival analysis revealed that patients with high CPNE1 expression had significantly shorter OS compared with those with low expression (figure 1J). Prognostic evaluation using a nomogram model indicated that advanced T, N, and M stage, higher pathologic grade, older age, and high CPNE1 expression were all associated with higher nomogram scores, which in turn predicted shorter survival probability and poorer prognosis (figure 1K). Moreover, univariate and multivariate Cox regression confirmed CPNE1 as an independent prognostic factor for OS in HCC patients (figure 1L). Western blotting and immunohistochemistry of paired clinical specimens validated that CPNE1 was highly expressed in tumor tissues compared with adjacent non-tumoural tissues (figure 1M–O). CPNE1 is highly expressed in hepatocellular carcinoma and serves as a predictor of poor prognosis. (A, B) Differential expression of CPNE1 in HCC and adjacent normal tissues from the TCGA-LIHC dataset by unpaired (A) and paired (B) analyses. (C, D) Differential expression of CPNE1 in HCC and adjacent normal tissues from GEO datasets GSE87630 (C) and GSE25097 (D). (E–G) Association of CPNE1 expression with clinicopathological features, including T stage (E), pathologic stage (F), and AFP level (G) in the TCGA-LIHC cohort. (H–I) Kaplan-Meier survival analyses of OS (H) and PFS (I) based on CPNE1 expression from the TCGA-LIHC dataset. (J) Kaplan-Meier survival analysis of OS in a cohort of 51 HCC patients who underwent surgical resection at SYSUCC. (K) Prognostic nomogram model integrating clinicopathological parameters and CPNE1 expression in HCC. (L) Forest plot showing univariate and multivariate Cox regression analyses evaluating CPNE1 as an independent prognostic factor for OS. (M) Western blotting validating CPNE1 protein expression in paired tumor and normal adjacent non-tumor tissues. T: tumor; N: adjacent non-tumor tissues. (N–O) Representative IHC images (4× and 20×) (N) and quantification of IHC scores (O) of CPNE1 protein expression in paired tumor and normal adjacent non-tumor tissues. Scale bars, 20 µm (4×) and 100 µm (20×). Mean±SD, n=8, two-tailed t-test. ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. AFP, alpha-fetoprotein; GEO, Gene Expression Omnibus; HCC, hepatocellular carcinoma; IHC, immunohistochemistry; OS, overall survival; PFS, progression-free survival; SYSUCC, Sun Yat-sen University Cancer Center; TCGA, The Cancer Genome Atlas. CPNE1 promotes HCC cell proliferation and stemness in vitro We first assessed CPNE1 protein expression in commonly used HCC cell lines (Hep3B, SK-Hep1, MHCC97h, and Huh7) by Western blotting. CPNE1 expression was high in MHCC97h and Huh7 cells but relatively low in Hep3B and SK-Hep1 cells (online supplemental appendix 1B). Based on these findings, we selected MHCC97h and Huh7 cells for CPNE1 knockdown experiments, and Hep3B and SK-Hep1 cells for CPNE1 overexpression, to investigate its functional role. The efficiency of CPNE1 knockdown and overexpression was confirmed at both protein and mRNA levels using Western blotting (online supplemental appendix 1C,D) and qPCR (online supplemental appendix 1E-F), respectively. After establishing CPNE1 stable knockdown and overexpression HCC cell lines, we examined the impact of CPNE1 on HCC cell proliferation using IncuCyte real-time monitoring and colony formation assays. The results indicated that CPNE1 knockdown significantly suppressed tumor cell proliferation, whereas its overexpression enhanced tumor cell proliferation (figure 2A–D). CPNE1 enhances proliferation, maintains stemness, and induces oxaliplatin resistance in HCC cells. (A, B) Incucyte live-cell imaging analysis evaluating the effects of CPNE1 overexpression (A) and knockdown (B) on HCC cell proliferation. Cells were pretransfected with mCherry fluorescence, and cell proliferation capacity was quantified by average fluorescence intensity. Mean±SD, n=3, two-tailed t-test. (C, D) Colony formation assay assessing the effects of CPNE1 overexpression (C) and knockdown (D) on clonogenic capacity. Mean±SD, n=3, two-tailed t-test. (E, F) GO and KEGG enrichment analyses of the top 200 genes most differentially expressed in association with CPNE1 in the TCGA-LIHC cohort to elucidate potential biological functions. (G) The TCGA-LIHC dataset was divided into high-CPNE1 and low-CPNE1 expression groups based on the median CPNE1 expression level. A heatmap illustrates the differential expression patterns of stemness-related marker genes between the two groups. (H) Correlation analysis between CPNE1 expression and stemness-related marker genes based on TCGA-LIHC dataset. (I, J) Western blot analysis showing protein expression levels of stemness markers SOX2, OCT4, NANOG and c-MYC in CPNE1-overexpressing (I) and CPNE1-knockdown (J) cells. (K, L) Sphere formation assay evaluating the effects of CPNE1 overexpression (K) and knockdown (L) on tumor cell stemness. Left: Representative images of spheroid formation under a 10× microscope. Right: Quantification of spheroid numbers recorded under a 4×microscope and statistically analyzed. Scale bars, 50 µm. Mean±SD, n=3, two-tailed t-test. (M) CCK-8 assay measuring IC₅₀ values of oxaliplatin in CPNE1-overexpressing and CPNE1-knockdown cells. ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. BC, biological process; GO, gene ontology; HCC, hepatocellular carcinoma; IC₅₀, half-maximal inhibitory concentration; KEGG, Kyoto Encyclopedia of Genes and Genomes; TCGA, The Cancer Genome Atlas. To explore the underlying mechanisms, we performed differential gene expression analysis using the TCGA-LIHC dataset. GO and KEGG enrichment analyses of differentially expressed genes suggested a potential role for CPNE1 in maintaining HCC stemness, in addition to its effect on tumor cell proliferation (figure 2E,F). Consistent with this, patients in the CPNE1 high-expression group (stratified by median expression) showed significantly elevated levels of multiple stemness-related genes compared with the low-expression group (figure 2G). A positive correlation between CPNE1 and these stemness-associated genes was further confirmed (figure 2H). To experimentally validate this prediction, Western blot analysis revealed that CPNE1 knockdown downregulated key stemness-associated markers (SOX2, OCT4, NANOG, and c-Myc), while its overexpression had the opposite effect (figure 2I,J). Consistently, sphere formation assays showed that CPNE1 knockdown markedly reduced the number and size of spheres, whereas its overexpression promoted sphere formation (figure 2K,L). Given the well-established association between tumor stemness and chemoresistance, and considering that oxaliplatin is a commonly used chemotherapeutic agent in HCC treatment, particularly in oxaliplatin-based HAIC regimens, we assessed whether CPNE1 modulates drug sensitivity by determining the IC50 of oxaliplatin in HCC cells. The results showed that overexpression of CPNE1 significantly increased the IC50 of oxaliplatin. Conversely, CPNE1 knockdown markedly sensitized HCC cells to the drug (figure 2M). CPNE1 enhances HCC proliferation and stemness through the STAT3-TGF-β pathway GSEA indicated that CPNE1 expression was positively associated with STAT3 and TGF-β signaling pathways in HCC (figure 3A,B). This was validated by Western blotting analysis, which showed that CPNE1 knockdown reduced, whereas its overexpression increased, the levels of phosphorylated STAT3 (p-STAT3) and TGF-β (figure 3C). Consistently, ELISA demonstrated that TGF-β secretion was decreased on CPNE1 silencing but elevated in CPNE1-overexpressing cells (figure 3D). CPNE1 promotes HCC proliferation and stemness by activating the STAT3–TGF-β signaling pathway. (A, B) Gene Set Enrichment Analysis (GSEA) of TCGA-LIHC data showing enrichment of STAT3 (A) and TGF-β (B) signaling pathways in CPNE1-high tumors. (C) Western blot analysis of phosphorylated STAT3 (p-STAT3) and TGF-β protein levels in CPNE1-overexpressing (left) and CPNE1-knockdown (right) cells. (D) ELISA quantification of TGF-β secretion in culture supernatants after 48 hours. Mean±SD, n=3, two-tailed t-test. (E, F) Colony formation (E) and sphere formation (F) assays evaluating proliferation and stemness of CPNE1-overexpressing cells treated with Stattic (2 µM) for 24 hours. Quantification of spheroid numbers recorded at 4×magnification. Mean±SD, n=3, one-way ANOVA. (G) Top: ELISA validation of TGF-β secretion in CPNE1-overexpressing cells after Stattic (2 µM) treatment. Bottom: Western blot analysis showing p-STAT3, TGF-β, and key stemness marker (SOX2, OCT4, NANOG, c-MYC) protein levels. Mean±SD, n=3, one-way ANOVA. (H, I) Colony formation (H) and sphere formation (I) assays in CPNE1-knockdown cells treated with TGF-β (10 ng/mL) for 24 hours. Mean±SD, n=3, one-way ANOVA. (J) Western blot showing restoration of stemness marker (NANOG, SOX2, c-MYC and OCT4) expression after TGF-β (10 ng/mL) supplementation. ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; GSEA, Gene Set Enrichment Analysis; OE, overexpression; STAT3, signal transducer and activator of transcription 3; TGF-β, transforming growth factor beta. To determine whether CPNE1 regulates the proliferation and stemness of HCC through the STAT3-TGF-β axis, we treated CPNE1-overexpressing cells with the STAT3 inhibitor Stattic. Stattic treatment abrogated the enhanced proliferation and sphere formation induced by CPNE1 overexpression (figure 3E,F and see online supplemental appendix 2A,B). At the molecular level, Stattic not only suppressed p-STAT3 but also reduced the upregulation of TGF-β and stemness-related markers caused by CPNE1 overexpression (figure 3G). Concordantly, ELISA further showed that Stattic markedly diminished TGF-β secretion in CPNE1-overexpressing cells (figure 3G). Conversely, supplementation with exogenous TGF-β restored the proliferation and sphere formation capabilities impaired by CPNE1-knockdown cells (figure 3H,I and see online supplemental appendix 2C,D). Western blot analysis confirmed that the addition of TGF-β rescued the expression of stemness-related markers in CPNE1-knockdown cells (figure 3J). CPNE1 drives tumor growth and stemness in vivo through STAT3/TGF-β signaling To evaluate the role of CPNE1 in tumor growth and stemness in vivo, we first established a subcutaneous xenograft model using Huh7 cells with CPNE1 knockdown (figure 4A). Compared with the shNC group, tumors derived from CPNE1-knockdown cells (sh1 and sh2) exhibited significantly slower growth, resulting in reduced tumor size, volume, and weight (figure 4B–D). Western blot analysis of xenograft tumor lysates further demonstrated that CPNE1 knockdown suppressed STAT3 phosphorylation, reduced TGF-β expression, and downregulated stemness-associated markers (figure 4E). Consistently, immunohistochemical staining confirmed that the expression levels of NANOG, SOX2, and c-MYC were markedly decreased in CPNE1-knockdown tumors (figure 4F,G). CPNE1 promotes tumor growth and maintains cancer stemness in vivo through activation of the STAT3/TGF-β signaling pathway. (A–D) Subcutaneous (s.c.) xenograft model using Huh7 cells with CPNE1 knockdown (sh1, sh2) and control (shNC). (A) Experimental schematic of xenograft model. (B) Representative tumor images. (C) Tumor growth curves (tumor volume=length×width²/2). (D) Xenograft weights. Mean±SD, n=5, two-tailed t-test. (E–G) Analysis of CPNE1-knockdown xenograft tissues. (E) Western blot analysis of p-STAT3, TGF-β, and stemness markers (NANOG, SOX2, c-MYC, OCT4) in CPNE1-knockdown xenografts. (F, G) Representative IHC images (4× and 20×) in stemness markers (NANOG, SOX2, c-MYC) (F) and quantification of IHC scores (G) in Huh7 shNC, sh1 and sh2 tumors. Mean±SD, n=3, two-tailed t-test. (H–K) Xenograft model using SK-Hep1 cells with CPNE1 overexpression (OE) or vector control. (H) Experimental schematic of xenograft model. (I) Representative tumor images. (J) Tumor growth curves (tumor volume=length×width²/2). (K) Xenograft weights. Mean±SD, n=5, two-tailed t-test. (L–N) Analysis of CPNE1-overexpressing xenograft tissues. (L) Western blot analysis of p-STAT3, TGF-β, and stemness markers (NANOG, SOX2, c-MYC, OCT4) in CPNE1-overexpressing xenografts. (M, N) Representative IHC images (4× and 20×) in stemness markers (NANOG, SOX2, c-MYC) (M) and quantification (N) in SK-Hep1 and SK-Hep1 OE tumors. Scale bars, 20 µm (4×) and 100 µm (20×). Mean±SD, n=3, two-tailed t-test. (O–S) STAT3 inhibition attenuates CPNE1-mediated tumor growth and TGF-β expression in vivo. (O) Experimental schematic of xenograft model using CPNE1-overexpressing SK-Hep1 cells treated with Stattic or vehicle control. (P) Representative tumor images from each group. (Q) Tumor growth curves (tumor volume=length × width²/2). (R) Xenograft weights. Mean±SD, n=5, one-way ANOVA. (S) ELISA analysis of TGF-β levels in xenograft tumor lysates from each group. Mean±SD, n=5, one-way ANOVA. ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; HCC, hepatocellular carcinoma; IHC, immunohistochemistry; STAT3, signal transducer and activator of transcription 3; TGF-β, transforming growth factor beta. We next established a xenograft model using SK-Hep1 cells with CPNE1 overexpression (figure 4H). In contrast to the knockdown model, CPNE1 overexpression significantly accelerated tumor growth and increased tumor size, volume, and weight compared with the control group (figure 4I–K). Western blot analysis revealed enhanced STAT3 phosphorylation, increased TGF-β expression, and elevated levels of stemness-related markers in CPNE1-overexpressing tumors (figure 4L). Immunohistochemical staining further confirmed increased expression of NANOG, SOX2, and c-MYC in the CPNE1-overexpression group (figure 4M,N). To further validate that CPNE1 regulates tumor growth and stemness through the STAT3–TGF-β axis in vivo, we established a xenograft model using CPNE1-overexpressing SK-Hep1 cells (figure 4O). The results showed that tumors derived from CPNE1-overexpressing cells exhibited significantly enhanced tumor growth compared with the control group. Importantly, treatment with the STAT3 inhibitor Stattic effectively attenuated the tumor-promoting effect induced by CPNE1 overexpression (figure 4P). Consistently, both tumor volume and tumor weight were markedly increased in the CPNE1-overexpression group, whereas inhibition of STAT3 activation by Stattic significantly suppressed the CPNE1-mediated increase in tumor burden (figure 4Q,R). Furthermore, tumor tissues were harvested for protein extraction, and TGF-β levels in tumor lysates were quantified by ELISA. We found that CPNE1 overexpression significantly increased TGF-β expression in tumor tissues, while Stattic treatment effectively reduced TGF-β levels induced by CPNE1 overexpression (figure 4S). Collectively, these in vivo results indicate that CPNE1 promotes HCC tumor growth and sustains cancer stemness through the STAT3-TGF-β signaling pathway. High CPNE1 expression in tumors may contribute to antitumor immune dysfunction in the TME In recent years, immunotherapy, particularly ICIs such as PD-1/PD-L1 blockade, has become a major strategy for HCC patients and has demonstrated clinical benefits.22 However, a substantial proportion of patients exhibit limited responsiveness to immunotherapy or acquire resistance to immunotherapy. Previous studies have shown that TGF-β derived from the immune microenvironment can impair T cell function and promote immunotherapy resistance.23 Similarly, previous studies have shown that aberrant activation of STAT3 can promote tumor stemness and induce immune evasion.24 Based on our findings that CPNE1 promotes STAT3-dependent TGF-β release, we hypothesized that elevated CPNE1 expression contributes to an immunosuppressive microenvironment in HCC. Consistent with this, bioinformatic analysis of HCC samples revealed that tumors with high CPNE1 expression exhibited significantly reduced infiltration of antitumor immune cells, including dendritic cells, macrophages, NK cells, and T cells (figure 5A,B). Furthermore, CPNE1 expression was strongly and positively correlated with key T cell exhaustion-related genes, such as PDCD1 (PD-1), CTLA4, LAG3, TIGIT, and HAVCR2 (TIM-3) (figure 5C). CPNE1 suppresses CD3+ T-cell infiltration and impairs immunotherapy efficacy in HCC. (A) Quantification of immune cell infiltration in the TME between CPNE1-high and CPNE1-low groups in TCGA-LIHC dataset. (B) Analysis of immune estimate scores in the TME between CPNE1-high and CPNE1-low groups. (C) Correlation analysis of CPNE1 expression with T-cell exhaustion-related markers. (D, E) Representative IHC staining (D) and quantification of IHC scores (E) for CPNE1 and CD3 in tumor tissues from 28 HCC patients who underwent surgical resection followed by immunotherapy. Patients were stratified into progressive disease (PD), stable disease (SD), partial response (PR) and complete response (CR) groups based on clinical outcomes. Images were captured at 4× and 20× magnifications. Scale bars, 20 µm (4×) and 100 µm (20×). Mean±SD, n=28, two-tailed t-test. (F) Correlation analysis of CPNE1 expression and CD3+ T cell infiltration levels in tumor tissues from HCC patients receiving immunotherapy. (G) Stratification of patients into high-expression and low-expression groups based on median IHC scores of CPNE1 (left) and CD3 (right), with comparison of PD/SD and PR/CR proportions across groups. (H, I) Kaplan-Meier survival analysis of OS in HCC patients receiving immunotherapy, stratified by CPNE1 (H) and CD3+T cells (I) IHC score levels. (J, K) Kaplan-Meier survival analysis of -PFS in HCC patients receiving immunotherapy, stratified by CPNE1 (J) and CD3+T cells (K) IHC score levels. (L) Validation of CPNE1 expression differences between immunotherapy responders and non-responders in independent GEO datasets (GSE279750 and GSE235863). ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. HCC, hepatocellular carcinoma; IHC, immunohistochemistry; OS, overall survival; PFS, progression-free survival. CPNE1 suppresses T cell infiltration and is associated with reduced immunotherapy responsiveness in HCC To further explore the potential clinical relevance of CPNE1 in tumor immunotherapy, we performed IHC analysis on tumor specimens from 28 HCC patients who received anti-PD-1/PD-L1 therapy prior to surgical resection. Representative images revealed higher CPNE1 expression and concomitantly lower CD3+ T cell infiltration in tumors from patients with progressive disease (PD) and stable disease (SD) groups compared with those with the partial response (PR) and complete response (CR) (figure 5D). Quantitative analysis confirmed that the CPNE1 IHC score was significantly elevated in the PD/SD group, while the CD3+T cell infiltration score was significantly higher in the PR/CR group (figure 5E). Moreover, correlation analysis across all patient samples demonstrated a significant negative association between CPNE1 expression and CD3+ T cell density (figure 5F). We further performed IHC analysis of PD-1 and TIM-3 expression in tumor tissues. The results showed no significant differences in PD-1 or TIM-3 expression levels between the PD/SD and PR/CR groups (online supplemental appendix 3A,B). In addition, correlation analyses between the IHC scores of CPNE1 and those of PD-1 or TIM-3 revealed no significant association between CPNE1 expression and PD-1 or TIM-3 levels in tumor tissues (online supplemental appendix 3C,D). Clinically, patients with low CPNE1 expression and high CD3+ T cell infiltration had a significantly greater probability of achieving overall response compared with those exhibiting high CPNE1 expression and low CD3+ T cell infiltration (figure 5G). Consistent with this, Kaplan-Meier survival analysis showed that HCC patients with high CPNE1 expression and low CD3+T cells infiltration had significantly shorter overall survival (OS) than those with low CPNE1 expression and high CD3+T cells infiltration following immunotherapy (figure 5H,I). Moreover, Kaplan-Meier survival analysis revealed that patients with high CPNE1 expression exhibited shorter PFS compared with those with low CPNE1 expression, whereas no significant difference in PFS was observed between the high and low CD3+ T-cell infiltration groups (figure 5J,K). Interrogation of independent public cohorts (GSE279750 and GSE235863) further confirmed that patients responding to immunotherapy exhibited markedly lower CPNE1 expression levels (figure 5L). Collectively, these findings suggest that elevated CPNE1 expression is associated with reduced immune cell infiltration and poorer clinical responses to immunotherapy in HCC, which supports a potential link between CPNE1 and an immune-cold TME. Development and validation of GPC3-targeted CAR-T cells for HCC GPC3 is a tumor-associated antigen highly expressed in HCC and has been widely explored as a therapeutic target in antibody–drug conjugates and CAR-T cell therapies. To functionally evaluate the impact of tumor cell-intrinsic CPNE1 on T cell function, we established a GPC3-targeted CAR-T cell model. We first analyzed TCGA-LIHC data and found that GPC3 was highly expressed in HCC (online supplemental appendix 4A,B). This finding was further validated using GEO datasets GSE54236 and GSE87630 (online supplemental appendix 4C,D). In addition, immunohistochemical staining of paired tumor and adjacent normal liver tissues from three surgical specimens revealed strong GPC3 expression in HCC tissues, whereas little to no expression was detected in adjacent non-tumoural tissues (online supplemental appendix 4E). These results indicate that GPC3 is a promising therapeutic target for HCC. Consistently, Western blot and flow cytometry analyses confirmed that GPC3 was highly expressed in Hep3B and Huh7 cells but nearly undetectable in MHCC97H and SK-Hep1 cells (online supplemental appendix 4F,G). Based on these findings, Hep3B and Huh7 cells were selected as target cell model for subsequent experiments. CAR-T cells were generated from healthy donor PBMCs using a lentiviral vector encoding a CAR construct containing the GC33 scFv derived from the GPC3 antibody, CD8 transmembrane domain, the CD28 costimulatory domain, and the CD3ζ signaling domain (online supplemental appendix 5A,B). The resulting CAR-T cell population exhibited a high transduction efficiency (65.2%) and phenotypic profiles (eg, CD4/CD8 ratios, memory phenotypes, and baseline exhaustion markers) comparable to control Mock T cells (online supplemental appendix 5C–G). These CAR-T cells expanded robustly in vitro while maintaining stable CAR expression (online supplemental appendix 5H,I). Subsequently, we further evaluated the tumor-killing activity of the constructed CAR-T cells. SK-Hep1 cells stably expressing GPC3 were established (online supplemental appendix 6A). In vitro cytotoxicity assays demonstrated that CAR-T cells exerted potent antitumor effects against SK-Hep1-GPC3 cells at both E:T ratios of 2:1 and 1:1 (online supplemental appendix 6B,C). In vivo studies also confirmed that CAR-T cells exhibited robust antitumor efficacy against tumors in the SK-Hep1-GPC3 cell model (online supplemental appendix 6D–F). CPNE1 induces T cell exhaustion and impairs CAR-T cell function Having established a robust CAR-T model, we investigated whether CPNE1-mediated signaling directly induces T cell exhaustion. Using both conditioned supernatant model (sup) and Transwell co-culture system (coc) (online supplemental appendix 7A), we found that exposure to CPNE1-overexpressing tumor cells significantly upregulated exhaustion markers (PD-1, TIM-3, LAG-3) on T cells, whereas exposure to CPNE1-knockdown cells reduced their expression (figure 6A,B). However, co-culture of T cells with either CPNE1-overexpressing or CPNE1-knockdown HCC cells did not significantly alter the proportion of Treg cells, suggesting that CPNE1-mediated immunosuppressive effects are unlikely to be primarily dependent on modulation of Treg cell abundance (online supplemental appendix 8A,B). Functionally, this translated to a direct impact on CAR-T cell efficacy. In in vitro cytotoxicity assays, GPC3 CAR-T cells exhibited enhanced killing of CPNE1-knockdown Huh7 (Huh7 sh1 and Huh7 sh2) cells but significantly reduced cytotoxicity against CPNE1-overexpressing Hep3B (Hep3B OE) cells at both E:T ratios of 1:4 and 1:2 (figure 6C and online supplemental appendix 7B,C). Mechanistically, when engaging CPNE1-knockdown tumors, CAR-T cells displayed enhanced degranulation (CD107a) and produced higher levels of the effector cytokines TNF-α and IFN-γ, and the cytolytic molecule granzyme B (figure 6D–G and online supplemental appendix 7D,E). Conversely, these effector functions were severely compromised when CAR-T cells targeted CPNE1-overexpressing tumors. ELISA confirmed that CAR-T cells secreted significantly higher levels of IL-2, IFN-γ, and TNF-α on co-culture with CPNE1-knockdown cells (figure 6H). CPNE1 induces T-cell exhaustion and impairs CAR-T cell cytotoxicity through modulation of the immune microenvironment. (A, B) Flow cytometry analysis of PD-1+TIM-3+ (A) and LAG-3+ (B) in T cells pretreatment with tumor cells. Left: Representative flow plots. Right: Statistical analysis. Mean±SD, n=3, two-tailed t-test. Sup: Supernatants from HCC cells (CPNE1 knockdown or overexpression (OE)) collected after 48 hours and mixed with fresh T-cell medium (E:T=1:1) for 48 hours before analysis. Coc: Transwell co-culture for 48 hours (E:T=1:1). (C) CAR-T cells co-incubated with mCherry-labeled HCC cells (CPNE1 knockdown or OE) at E:T=1:2 or 1:4 for 72 hours. Tumor growth quantified by fluorescence intensity. Mean±SD, n=3, one-way ANOVA. (D–G) CAR-T cells co-incubated with HCC cells (E:T=1:1) for 6 hours. Flow cytometric analysis of CD107a (D), TNF-α (E), Granzyme B (F), and IFN-γ (G). CD107a: Mean±SD, n=4; TNF-α, Granzyme B, IFN-γ: Mean±SD, n=3, two-tailed t-test. (H) ELISA of cytokines (IL-2, IFN-γ, TNF-α) in supernatants after 24 hours co-culture. Mean±SD, n=3, two-tailed t-test. ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; CAR-T, chimeric antigen receptor T cell; E:T, effector-to-target ratio; HCC, hepatocellular carcinoma; IFN-γ, interferon gamma; TNF-α, tumor necrosis factor alpha. CPNE1 knockdown enhances the antitumor efficacy of CAR-T cells in vivo To assess the role of tumor-intrinsic CPNE1 in modulating CAR-T cell efficacy in vivo, we established subcutaneous xenografts using Huh7 control (shNC) or CPNE1-knockdown (sh1, sh2) cells, followed by CAR-T cell treatment (figure 7A). The results showed that tumors derived from CPNE1-knockdown cells (sh1 and sh2) were efficiently eradicated after CAR-T treatment, whereas tumors derived from control group (shNC) exhibited only a limited response (figure 7B–E). Consequently, mice bearing CPNE1-knockdown tumors (sh1 and sh2) exhibited significantly prolonged survival following CAR-T cell therapy compared with the control group (shNC) (figure 7F). In contrast, in the SK-Hep1-GPC3 xenograft model (online supplemental appendix 9A), overexpression of CPNE1 significantly impaired the antitumor efficacy of CAR-T cells, resulting in enhanced tumor growth and reduced therapeutic response compared with the control group (online supplemental appendix 9B–D), further supporting the role of CPNE1 in promoting resistance to CAR-T cell therapy in HCC. CPNE1 knockdown enhances the in vivo efficacy of GPC3 CAR-T cells in HCC. (A) Experimental schematic illustrating the establishment of subcutaneous (s.c.) xenograft models using Huh7 cells with CPNE1 knockdown (sh1, sh2) and control (shNC), followed by treatment with GPC3-CAR-T cells. (B–E) Evaluation of the antitumor efficacy of CAR-T therapy in the Huh7 xenograft model. (B) Tumor growth curves (tumor volume=length×width²/2). Mean±SD, n=5, two-tailed t-test. (C) Statistical analysis of bioluminescence intensity. Mean±SD, n=5, two-tailed t-test. (D) Representative bioluminescence imaging of tumors. (E) Quantification of bioluminescence signals across groups. (F) Kaplan-Meier survival analysis comparing CPNE1-knockdown (sh1, sh2) and control (shNC) groups after CAR-T therapy (n=5). (G) Experimental schematic of the Huh7 xenograft model for subsequent immune profiling. (H) Flow cytometry analysis quantifying CD45+CD3+ T-cell infiltration in tumor tissues (left) and spleens (right) following CAR-T therapy. Mean±SD, n=3, two-tailed t-test. (I) Flow cytometric quantification of exhaustion marker expression (PD-1, TIM-3, and LAG-3) on tumor-infiltrating T cells. Mean±SD, n=3, two-tailed t-test. (J, K) Representative IHC images (J) and corresponding IHC score quantification (K) of CD3 and Granzyme B in tumor-infiltrating T cells after CAR-T therapy. Images were captured at 4×and 20×magnifications. Scale bars, 20 µm (4×) and 100 µm (20×). Mean±SD, n=3, two-tailed t-test. ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CAR-T, chimeric antigen receptor T cell; HCC, hepatocellular carcinoma; IHC, immunohistochemistry. We next investigated whether CPNE1 knockdown enhances CAR-T function in vivo. To this end, we established a separate cohort of tumor-bearing mice and harvested tumors and spleens prior to complete tumor clearance for analysis (figure 7G). Flow cytometry revealed a significant increase in T cell infiltration in both tumors and spleens from the CPNE1-knockdown groups compared with controls (figure 7H and online supplemental appendix 10A). Critically, these tumor-infiltrating T cells from the CPNE1-knockdown group also showed markedly reduced expression of the exhaustion markers PD-1, TIM-3, and LAG-3 (figure 7I and online supplemental appendix 10B), suggesting that CPNE1 knockdown alleviates tumor-infiltrating T cell exhaustion. Immunohistochemistry further confirmed these findings, showing enhanced T-cell infiltration and elevated granzyme B production within CPNE1-knockdown tumors (figure 7J,K). Histopathological examination of major organs (heart, lung, kidney, and liver) was performed by H&E staining to evaluate treatment-related toxicity. No significant pathological abnormalities were observed in mice bearing CPNE1-knockdown tumors following CAR-T therapy (online supplemental appendix 10C), indicating that the enhanced antitumor efficacy achieved by CPNE1 knockdown was not associated with systemic toxicity. CPNE1 promotes T cell exhaustion and impairs CAR-T cell function through TGF-β signaling Having established that CPNE1 knockdown enhances CAR-T function in vivo, we sought to determine whether this effect is mediated specifically through TGF-β signaling. We treated CPNE1-overexpressing Hep3B cells with a neutralizing anti-TGF-β antibody prior to co-cultured with CAR-T cells. In in vitro cytotoxicity assays, TGF-β blockade effectively rescued the impaired tumor-killing capacity of CAR-T cells against CPNE1-overexpressing tumors at both E:T ratios of 1:4 and 1:2 (figure 8A,B). CPNE1 promotes T cell exhaustion and impairs CAR-T cell cytotoxicity via TGF-β signaling. (A, B) CPNE1-overexpressing HCC cells labeled with mCherry fluorescence were co-incubated with CAR-T cells at effector-to-target (E:T) ratios of 1:2 and 1:4 for 48 hours in the presence or absence of 1 µg/mL anti-TGF-β antibody. Tumor cell growth was monitored using the Incucyte live-cell imaging system and average fluorescence intensity was quantified to assess CAR-T cytotoxicity. (A) Experimental design of treatment groups. (B) Quantification of relative fluorescence intensity. Mean±SD, n=4, one-way ANOVA. (C, D) Flow cytometry analysis of T-cell exhaustion markers PD-1 and TIM-3 (C) and LAG-3 (D) after anti-TGF-β treatment. Left: representative flow cytometry plots; right: statistical analysis of flow cytometry results. Mean±SD, n=4, one-way ANOVA. Supernatant (Sup): CPNE1-overexpressing HCC cells were treated with 1 µg/mL anti-TGF-β antibody for 48 hours, and the collected supernatant was mixed 1:1 with fresh T-cell medium before T-cell addition. After 48 hours, PD-1, TIM-3 and LAG-3 expression was measured by flow cytometry. Coculture (Coc): CPNE1-overexpressing HCC cells were treated with 1 µg/mL anti-TGF-β antibody and co-cultured with T cells in Transwell chambers for 48 hours, followed by flow cytometric analysis of PD-1, TIM-3, and LAG-3 expression. (E, F) CPNE1-overexpressing HCC cells were treated with 1 µg/mL anti-TGF-β antibody and co-incubated with CAR-T cells for 6 hours. Flow cytometry was performed to analyze the expression of CD107a (E), TNF-α, granzyme B and IFN-γ (F) in CAR-T cells. Left: representative flow cytometry plots; right: quantitative analysis. Mean±SD, n=4, one-way ANOVA. ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; CAR-T, chimeric antigen receptor T cell; E:T, effector-to-target ratio; IFN-γ, interferon gamma; TGF-β, transforming growth factor beta; TNF-α, tumor necrosis factor alpha. At the molecular level, anti-TGF-β antibody treatment reversed the upregulation of T cell exhaustion markers induced by CPNE1-overexpressing tumor cells (figure 8C,D). Furthermore, TGF-β blockade restored the effector functions of CAR-T cells, as evidenced by significantly increased levels of degranulation (CD107a) and production of granzyme B, IFN-γ, and TNF-α (figure 8E,F). Collectively, these findings demonstrated that CPNE1 promotes T-cell exhaustion and suppresses antitumor immunity primarily through TGF-β secretion, thereby identifying the CPNE1-STAT3-TGF-β axis as a key modulator of CAR-T cell efficacy in HCC. Discussion CPNE1 is recognized as a pivotal promoter of tumorigenesis and progression across multiple cancers. In clear cell renal cell carcinoma, CPNE1 promotes tumor cell proliferation and migration by activating the EGFR/STAT3 signaling pathway and serves as a reliable prognostic biomarker.17 Similarly, in non-small cell lung cancer, CPNE1 stabilization promotes abnormal proliferation and distant metastasis.25 These findings establish CPNE1 as a key regulator of oncogenic biological processes, including tumor cell proliferation, migration, and metastasis. While our previous study implicated CPNE1 in HCC cell proliferation,26 its precise mechanisms of action, particularly concerning cancer stemness and tumor immunity, remained elusive. Therefore, in this study, we further investigated how CPNE1 contributes to HCC pathogenesis, with a particular focus on these critical aspects. In this study, we first performed a comprehensive bioinformatics analysis to evaluate the expression pattern and clinical significance of CPNE1 in HCC. The results revealed that CPNE1 is markedly upregulated in HCC tissues, and its high expression is associated with unfavorable prognosis, suggesting an important role of CPNE1 in HCC progression. To further validate these findings, we analyzed tumor and adjacent non-tumor tissues from 8 HCC patients’ surgical resection. Consistently, CPNE1 was significantly overexpressed in tumor tissues compared with paired normal counterparts. Together, these results provide robust evidence supporting the oncogenic role of CPNE1 in HCC. Based on these findings, we further established HCC lines with either CPNE1 knockdown or overexpression to investigate its function. In vitro and in vivo assays demonstrated that CPNE1 is a critical driver of HCC proliferation, thereby validating its oncogenic role. CSCs are generally defined as a small subpopulation of tumor cells with the capacity for high tumorigenic potential, which are characterized by the expression of specific surface markers, such as CD133, CD44, CD90, EpCAM, and CD24, as well as the regulation of stemness-associated transcription factors, including Nanog, Oct4, Sox2, and c-Myc.27 CSCs are thought to represent the origin of the tumor, capable of initiating tumor formation and maintaining tumor heterogeneity.28 Moreover, CSCs exhibit enhanced migratory and invasive capacities, epithelial–mesenchymal transition (EMT) features, and resistance to chemotherapeutic agents such as sorafenib and oxaliplatin.29 30 These properties of CSCs collectively contribute to tumor recurrence and metastasis, highlighting the clinical significance of targeting CSCs in cancer treatment. In this study, our bioinformatic analyses initially suggested that, in addition to promoting HCC cell proliferation, CPNE1 may also be involved in the regulation of stemness. Consistent with this prediction, our experimental data demonstrated that CPNE1 overexpression markedly enhanced the stemness properties of HCC cells, whereas CPNE1 knockdown significantly suppressed these abilities. Furthermore, by evaluating the half-maximal inhibitory concentration (IC50) of oxaliplatin, we confirmed the role of CPNE1 in promoting chemoresistance. These findings further establish a direct mechanistic link between CPNE1-driven stemness and chemoresistance in HCC. Mechanistically, we identified the STAT3-TGF-β signaling axis as a key pathway activated by CPNE1. TGF-β is a well-established pro-tumorigenic factor that plays critical roles in sustaining cancer cell stemness, promoting EMT, and facilitating tumor metastasis.31 32 Moreover, TGF-β is widely recognized as a potent immunosuppressive cytokine that exerts multifaceted inhibitory effects on T-cell antitumor activity.33 TGF-β suppresses CD8+ T-cell proliferation by downregulating IL-2 expression, reduces the production of key effector molecules such as IFN-γ and granzyme B, and simultaneously promotes the expression of exhaustion-associated markers including PD-1 and TIM-3, ultimately impairing the antitumor function of T cells.34 35 In addition, TGF-β facilitates the recruitment and expansion of multiple immunosuppressive cell populations, including regulatory T cells (Tregs), M2-polarized macrophages, and myeloid-derived suppressor cells (MDSCs), within the TME. This process fosters a profoundly immunosuppressive TME that further compromises T-cell function.36 In our study, bioinformatics analysis suggested that CPNE1 may be involved in the STAT3-TGF-β signaling pathway. Subsequently, we conducted a series of experiments and further demonstrated that CPNE1-mediated aberrant activation of the STAT3–TGF-β signaling pathway is central to its effects on both HCC proliferation and stemness. This led us to a pivotal hypothesis: could CPNE1-driven TGF-β secretion also impair antitumor immunity? Supporting this notion, prior studies have shown that targeting stemness regulators (eg, HNRNPM) can enhance T-cell infiltration and improve immunotherapy responses.37 In the era where ICIs are cornerstone treatments for HCC,38 overcoming primary and acquired resistance remains a paramount challenge.39 We therefore investigated the impact of CPNE1 on CAR-T cell therapy, a highly personalized modality with proven success in hematological malignancies.40 41 Despite this promise, the immunosuppressive HCC TME significantly constrains the therapeutic efficacy of CAR-T cell therapy.42 Several mechanisms have been reported to contribute to the limited efficacy of CAR-T cell therapy in HCC, including antigen heterogeneity, hypoxia within the TME, and the accumulation of immunosuppressive myeloid cell populations.43 44 Heterogeneous or low tumor antigen expression may reduce the efficiency of CAR-T cell recognition and tumor targeting, while hypoxic stress and suppressive immune cells, such as MDSCs and tumor-associated macrophages, can promote T-cell dysfunction and exhaustion through multiple inhibitory signaling pathways.45–48 These factors collectively contribute to the establishment of a highly immunosuppressive TME that restricts the persistence and antitumor activity of CAR-T cells in solid tumors.49 Our data reveal that high CPNE1 expression creates a “cold” TME, characterized by reduced immune cell infiltration, and a signature of T-cell exhaustion. Subsequently, our in vitro and in vivo experiments demonstrated that CPNE1 promotes CAR-T cell exhaustion, thereby impairing their cytotoxic function. In contrast, CPNE1 knockdown markedly enhanced CAR-T cell activation, increased the production of antitumor cytokines, and improved CAR-T cell infiltration into the TME. Finally, treatment of CPNE1-overexpressing HCC cells with an anti-TGF-β antibody effectively reversed CPNE1-induced CAR-T cell exhaustion and restored their antitumor activity. These results further confirm that CPNE1 impairs CAR-T cell antitumor activity through TGF-β mediated mechanisms. Although our study identifies the CPNE1–STAT3–TGF-β axis as an important mechanism underlying CAR-T cell exhaustion and immune suppression in HCC, additional downstream immunosuppressive pathways or other CPNE1-regulated mechanisms may also contribute to tumor immune evasion. Further studies are needed to comprehensively elucidate the broader immunoregulatory network downstream of CPNE1. We acknowledge that a limitation of this study is that the CAR-T xenograft models were established in immunodeficient mice, which cannot fully recapitulate the immune microenvironment of HCC. Therefore, additional studies using immunocompetent, humanized, and PDX models will be required to further validate the immunomodulatory role and translational potential of CPNE1 in HCC immunotherapy. Furthermore, the clinical translation of targeting CPNE1 is still in its infancy and requires validation in advanced preclinical models and early-phase clinical trials. Future work could explore novel delivery strategies, such as nanoparticle-based systems, to overcome current limitations. Nevertheless, our findings robustly position CPNE1 inhibition as a promising strategy to concurrently suppress tumor growth and stemness while potentiating the efficacy of CAR-T cell therapy in HCC. Conclusions In summary, our study identifies CPNE1 as a critical regulator of HCC progression via activation of the STAT3–TGF-β axis. This activation drives a triad of pro-tumorigenic effects: enhanced tumor proliferation, maintenance of stemness properties, and induction of CAR-T cell exhaustion. Consequently, targeting CPNE1 presents a compelling dual-track therapeutic strategy, simultaneously inhibiting intrinsic tumorigenicity and reversing immunosuppression to enhance the therapeutic efficacy of CAR-T cells. Supplementary material