Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2025 Nov 25;15:45339. doi: 10.1038/s41598-025-29527-9

The prognostic value and functional role of CPSF3 in hepatocellular carcinoma

Weihao Kong 1,2,#, Yue Su 3,#, Long Teng 1,#, Kangjie Zhang 2, Yajun Zou 1, Xingyu Wang 1,✉, Jianlin Zhang 1,✉
PMCID: PMC12749310  PMID: 41286303

Abstract

The identification of reliable biomarkers is critical for improving the diagnosis and treatment of hepatocellular carcinoma (HCC). Cleavage and polyadenylation specific factor 3 (CPSF3) has been implicated in RNA processing and tumor progression, but its role, clinical significance, and functional mechanisms in HCC remain unclear. In this study, we systematically evaluated CPSF3 expression and its potential role in HCC through multi-omics analyses and experimental validation. We observed significant CPSF3 upregulation in HCC tissues in multiple cohorts, and its upregulation was associated with advanced AJCC stage and poor survival. A prognostic nomogram incorporating CPSF3 and AJCC stage showed excellent predictive performance for HCC. CPSF3 demonstrated high diagnostic accuracy in HCC in TCGA and ICGC datasets. Functional experiments revealed that CPSF3 knockdown suppressed HCC cell proliferation, migration, and invasion while promoting apoptosis. Single-cell sequencing analysis revealed predominant CPSF3 expression in malignant cells and proliferating T cells in HCC. CPSF3 expression in HCC correlated with immune cell infiltration and was enriched in cell cycle and MAPK signaling pathways. These findings establish CPSF3 as a promising diagnostic and prognostic biomarker and highlight its potential as a therapeutic target for HCC.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-29527-9.

Keywords: CPSF3, Diagnosis, Prognosis, Immune infiltrates, Hepatocellular carcinoma

Subject terms: Cancer, Computational biology and bioinformatics

Introduction

Hepatocellular carcinoma (HCC) is the most common primary liver cancer. It is among the most prevalent and deadly malignancies and the fourth most common cause of cancer-related death1–7. More than half of all new cases of HCC and deaths resulting from HCC occur in China8. Despite the advent of targeted drugs, the overall outcome for patients with HCC is poor. Therefore, early detection is crucial for patients with HCC. Alpha-fetoprotein (AFP) is a universally recognized biomarker for HCC, but it has limitations9–11. First, the result is distracting in pregnant patients and patients with hepatitis. Second, AFP is negative in a relatively small portion of patients with HCC, even those with a massive tumor burden12. Developing a strategy to precisely predict the treatment response and survival of patients with HCC is imperative to help improve patient outcome.

Cleavage polyadenylation specificity factor (CPSF) is the primary element of the 3′-end processing complex and controls how mRNAs are spliced and polyadenylated13. CPSF is comprised of six subunits: CPSF160 (also known as CPSF1), WD repeat domain 33, CPSF100 (also called CPSF2), CPSF3 (also called CPSF73), Fip1, and CPSF4 (also called CPSF30)13. Growing evidence has shown that CPSF is essential for the development and progression of tumors14. Compared with non-neoplastic squamous epithelium tissue, head and neck squamous cell carcinoma exhibits overexpression of CPSF1, and the abnormal expression of CPSF1 promotes disease progression through controlling alternative splicing15. Nilubol et al. showed that CPSF2 knockdown boosted thyroid cancer cell invasion and the number of cancer stem cells. A poor prognosis and lower CPSF2 expression were related to advanced T stage16. Several studies demonstrated the oncogenic activity of CPSF4 in colorectal cancer17. CPSF3 is a significant contributor to the development of several malignancies, such as esophageal squamous cell carcinoma, bladder cancer, glioblastoma, and pancreatic cancer18–25. Ma et al. demonstrated that the overexpression of CPSF3 was strongly associated with a decrease in overall survival (OS) and disease-free survival (DFS) in patients with bladder cancer. Experimental investigations revealed that the inhibition of CPSF3 substantially hindered colony formation, cellular proliferation, and tumor growth, which was linked to a G1 to S phase cell cycle arrest18. Abdulrahman et al. identified CPSF3 as a potential therapeutic target in pancreatic cancer and demonstrated that its increased expression correlated with adverse clinical outcomes. The genetic silencing of CPSF3 led to a reduction in cell proliferation and clonogenic potential in pancreatic ductal adenocarcinoma cells in vitro and diminished tumor growth in vivo23. Pagani et al. identified aberrant cleavage and polyadenylation as a defining characteristic of cancer, highlighting the importance of targeting this molecular mechanism in cancer research26.

We examined CPSF3 expression in HCC and its associations with clinicopathological traits and patient survival. The diagnostic and prognostic potential of CPSF3 was examined to determine its clinical role in HCC. Furthermore, to uncover the underlying mechanisms of CPSF3 in HCC, we examined the biological functions, genetic and epigenetic alternations, associations with immune filtration, and in vitro functions of CPSF3. These results provide clues for the clinical value of CPSF3 in HCCs.

Materials and methods

Resources for data and analysis of CPSF3 expression

We examined the expression of CPSF3 in a variety of cancers using the Cancer Genome Atlas (TCGA) database. CPSF3 expression was investigated in paired tumors and normal tissues using TCGA database and the Genotype-Tissue Expression Project (GTEx) database, considering the limited data of normal tissues in TCGA dataset. Gene Expression Omnibus (GEO) datasets including GSE22058, GSE25097, GSE36376, GSE46444, GSE54236, GSE63898, GSE64041, and GSE76427, TCGA-LIHC from TCGA database and ICGC-LIRI-JP from the International Cancer Genome Consortium (ICGC) database were used to assess CPSF3 expression in HCC27. We analyzed CPSF3 expression and its association with clinicopathological characteristics and OS in HCC cohorts from TCGA and ICGC. The phase-3 STORM trial used the GSE109211 dataset to examine the expression of CPSF3 in patients with sorafenib-sensitive and sorafenib-resistant HCC28. The scCancer platform (https://bianlab.cn/scCancerExplorer/) was used to investigate the expression patterns of CPSF3 in the single-cell sequencing dataset related to liver cancer29.

Receiver operating characteristic (ROC) curve analysis

The ability of CPSF3 expression to discriminate between HCC and non-tumor tissues was evaluated using ROC curve analysis. Time-dependent ROC curve analysis was used to assess CPSF3 and the constructed nomogram in predicting OS. To measure the diagnostic and prognostic capacity, we calculated the 95% confidence interval (CI) and the area under the ROC curve (AUC).

Kaplan–Meier survival analysis

Patients with HCC from HCC datasets from TCGA and ICGC were divided into two groups using the cutoff value of CPSF3 expression. The Youden index, computed using the formula (sensitivity + specificity) − 1, was used to identify the ideal cutoff value of CPSF3. Kaplan–Meier (K–M) survival analysis was used to evaluate the OS rates between the groups with high and low levels of CPSF3 expression. To determine the difference in survival rates between the two groups, the log-rank test was used.

Gene-immune infiltration analysis

The association between genes and immune infiltration was investigated using Tumor Immune Estimation Resource (TIMER) (https://cistrome.shinyapps.io/timer)30,31. We examined the correlation between CPSF3 expression and the percentage of tumor-infiltrating immune cells (TIICs) in HCC. TIICs include B cells, CD8 + T cells, CD4 + T cells, macrophages, neutrophils, and dendritic cells.

Functional enrichment analysis

Functional enrichment analysis was carried out using LinkedOmics (http://www.linkedomics.org/login.php) to identify co-expressed genes and the potential functional processes32. Co-expression genes were determined as significantly correlated with CPSF3 by Pearson correlation analysis. The co-expressed genes were subjected to enrichment analysis using KEGG and Gene Ontology (GO)33,34. Functionally enriched phrases were considered significant when the false discovery rate (FDR) value was less than 0.05.

Genetic alteration analysis

The correlation between genetic alteration of CPSF3 and OS in HCC was analyzed using the cBioPortal website tool (http://cbioportal.org)35–37, which is a web-based genomic database for genetic alteration analysis. The survival difference between patients with and without CPSF3 genetic alterations were compared using log-rank test.

DNA methylation analysis

Wanderer (http://maplab.imppc.org/wanderer/) is an online tool to investigate the relationship between DNA methylation and gene expression38. We used Wanderer to determine differences in DNA methylation of CPFS3 in tumor and normal tissues from TCGA HCC dataset. The individual probes with a significant difference in DNA methylation between tumor tissues and normal tissues were identified. Correlations between the DNA methylation of the identified probes and CPFS3 mRNA expression were also examined.

In vitro experiments

The THLE-2, Huh7, and LM3 cell lines were purchased from the Type Culture Collection Center at the Shanghai Cell Bank. Western blot was used to examine protein expression levels in normal liver cells and liver cancer cells and assess the effectiveness of RNA interference in liver cancer cells. Flow cytometry was performed to analyze apoptotic cells and the cell cycle dynamics of liver cancer cells after CPSF3 interference. The Edu assay was used to assess the changes in the proliferative capacity of HCC cells after CPSF3 interference. Wound healing and Transwell assays were performed to examine the migratory abilities of HCC cells after CPSF3 interference.

Statistical analysis

R program was used to perform statistical analysis. Depending on the kind of variable, the Student t-test or the Mann–Whitney U-test was used to analyze the variation in CPFS3 expression between the groups. The Chi-square test was used to assess the correlation between CPSF3 and clinicopathological features. Univariate and multivariate Cox regression analyses were performed to identify independent markers associated with OS. A P value < 0.05 was considered statistically significant.

Results

The expression of CPSF3 in malignancies and liver cancer

The expression of CPSF3 mRNA was explored in various malignancies using TCGA and GTEx datasets. CPSF3 expression was higher in tumor samples compared with the non-tumor samples in most cancer types from TCGA database (Fig. 1A). As data from normal samples in TCGA are limited, the normal samples from GTEx were incorporated for evaluation. Analysis of paired data patterns between tumor and nearby normal samples from 18 cancers confirmed higher expression level of CPFS3 mRNA in most tumors, including bladder urothelial carcinoma (BLCA), breast invasive cancer (BRCA), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), kidney renal papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), stomach adenocarcinoma (STAD), thyroid carcinoma (THCA), and uterine corpus endometrial carcinoma (UCEC) (Fig. 1B). CPSF3 mRNA in HCC was examined using data from 10 liver cancer datasets, namely GSE22058, GSE25097, GSE36376, GSE46444, GSE54236, GSE63898, GSE64041, GSE76427, TCGA-LIHC, and ICGC-LIRI-JP datasets. Overexpression of CPFS3 mRNA in liver tumor tissues compared with adjacent normal samples was observed in GSE22058, GSE36376, GSE46444, GSE54236, GSE63898, TCGA-LIHC, GSE64041, GSE76427, and ICGC-LIRI-JP. The expression level of CPSF3 mRNA in GSE25097 was significantly greater in HCC samples compared with adjacent, cirrhotic, and healthy samples (Fig. 1I).

Fig. 1.

Fig. 1

Open in a new tab

The CPSF3 mRNA expression in various malignancies. (A) The CPSF3 mRNA expression in tumor tissues and normal tissues in multiple malignancies; (B) The CPSF3 mRNA expression in tumor tissues and paired normal tissues in various malignancies; (C) ROC analysis of the diagnostic potential of CPSF3 in HCC from TCGA; (D) CPSF3 expression in paired normal and tumor tissues from TCGA; (E) ROC analysis of the diagnostic potential of CPSF3 in HCC from ICGC; (F) CPSF3 expression in paired normal and tumor tissues from ICGC; (G) Representative images of CPSF3 protein expression in normal tissues from HPA; (H) Representative images of CPSF3 protein expression in tumor tissues from HPA; (I) CPSF3 mRNA expression level in tumor tissue and non-tumor tissue in liver cancer datasets.

The diagnostic potential of CPSF3 in patients with HCC from TCGA and ICGC datasets was examined using ROC curve analysis. CPSF3 exhibited diagnostic potential in HCC with AUC values of 0.986 (CI: 0.976–0.996) and 0.938 (CI: 0.914–0.963) in TCGA and ICGC HCC cohorts, respectively (Fig. 1C, E). CPSF3 expression levels were considerably greater in tumor tissues than paired normal tissues in TCGA and ICGC databases (Fig. 1D, F).

Representative images of CPSF3 protein expression in normal tissues and tumor tissues were generated using The Human Protein Atlas database. CPSF3 protein was overexpressed in tumor tissues compared with healthy tissues (Fig. 1G, H). In the GSE140228 cohort, CPSF3 is mainly present in proliferating T cells, innate lymphoid cells, and dendritic cells. In the GSE149614 cohort, CPSF3 is found in cancer cells. In the GSE166635 cohort, CPSF3 is located in proliferating T cells, cancer cells, dendritic cells, and monocytes/macrophages (Fig. 2).

Fig. 2.

Fig. 2

Open in a new tab

The cellular subpopulation distribution of CPSF3 in single-cell cohorts.

(A)The cellular subpopulation distribution of CPSF3 in GSE140228; (B) The cellular subpopulation distribution of CPSF3 in GSE149614; (C) The cellular subpopulation distribution of CPSF3 in GSE166635.

Relationship between CPSF3 and clinicopathological characteristics

We next analyzed the clinicopathological traits and the expression of CPSF3 mRNA in patients with HCC in TCGA and ICGC. As presented in Fig. 3A–B, marked overexpression of CPSF3 was observed in the higher differentiation group (G3 + G4) in TCGA (P = 2.1e-06), advanced stage group (III + IV VS I + II) in TCGA (P = 0.018) and ICGC (P = 0.0058), and dead patient group in TCGA (P = 0.0078) and ICGC (P = 0.00073). CPSF3 was grouped as a low- and high-risk group by the optimal cutoff value. CPSF3 significantly correlated with AJCC stage (I + II vs. III + IV) in both TCGA (P = 0.012) and ICGC (P = 0.001) and survival status (alive vs. dead) in both TCGA (P < 0.001) and ICGC (P < 0.001). The results of the Chi-square test are shown in Table 1. Collectively, these results demonstrate that CPSF3 overexpression was associated with adverse clinicopathological characteristics and poor survival.

Fig. 3.

Fig. 3

Open in a new tab

CPSF3 correlated with clinicopathological characteristics in TCGA and ICGC HCC cohorts. (A) Differential expression of CPSF3 across distinct clinicopathological subgroups within the TCGA cohort; (B) Differential expression of CPSF3 across distinct clinicopathological subgroups within the ICGC cohort; (C) ROC analysis of the prognostic potential of CPSF3 in HCC from TCGA; (D) Survival analysis of HCC patients between high and low CPSF3 groups from TCGA; (E) Cox regression analysis identified independent risk factors affecting the prognosis of HCC patients in TCGA cohort; (F) ROC analysis of the prognostic potential of CPSF3 in HCC from ICGC; (G) Survival analysis of HCC patients between high and low CPSF3 groups from ICGC; (H) Cox regression analysis identified independent risk factors affecting the prognosis of HCC patients in ICGC cohort.

Table 1.

Correlation between CPSF3 expression and clinicopathological characteristics in TCGA and ICGC HCC cohorts.

Variables TCGA (n = 344) ICGC (n = 232)
Low risk
group		 High risk
group		 P-value Low risk
group		 High risk
group		 P-value
Age (years) 0.209 0.424
 < 60 149 22 34 11
 ≥ 60 158 15 130 57
Sex 0.494 0.773
 Male 207 27 120 51
 Female 100 10 44 17
Prior- Malignancy - 0.604
 No - - 144 58
 Yes - - 20 10
Histologic grade 0.279 -
 G1/2 194 20 - -
 G3/4 113 17 - -

TNM

stage

 0.012 0.001
 I/II 233 21 112 30
 III/IV 74 16 52 38
Survival status < 0.001 < 0.001
 Alive 216 14 144 45
 Dead 91 23 20 23
Open in a new tab

Bold values represent P-values less than 0.05.

Notes: TCGA: The Cancer Genome Atlas, HCC: hepatocellular carcinoma, AJCC: American Joint Committee on Cancer.

The prognostic potential of CPSF3 in patients with HCC from TCGA and ICGC datasets was also explored. As presented in Fig. 3C and F, the AUC values of 1-, 2-, and 3-year curves were 0.694, 0.634, and 0.635, respectively, in TCGA; the AUC values were 0.661, 0.668, and 0.680, respectively, in ICGC. CPSF3 was grouped into the low group and high group by the optimal cutoff value. Survival analysis revealed that high CPSF3 expression levels significantly correlated with dismal survival in TCGA (P < 0.0001) and ICGC (P < 0.0001) HCC cohorts (Fig. 3D, G).

Univariate and multivariate regression analyses were performed to identify the risk variables for survival in patients with HCC in TCGA and ICGC datasets. The results are illustrated in the forest plot in Fig. 3E and H. In univariate analysis, AJCC stage (HR: 2.50, 95% CI: 1.72–3.63, P < 0.001) and CPSF3 (HR: 3.57, 95% CI: 2.24–5.66, P < 0.001) were indicated as risk factors in TCGA dataset, and sex (HR: 1.93, 95% CI: 1.03–3.59, P = 0.039), AJCC stage (HR: 2.38, 95% CI: 1.30–4.36, P = 0.005), and CPSF3 (HR: 3.54, 95% CI: 1.93–6.48, P < 0.001) were revealed as risk factors in the ICGC dataset. In multivariate regression analysis, CPSF3 and AJCC stage were demonstrated as independent risk factors for survival in patients with HCC in both TCGA ((HR: 3.34, 95% CI: 2.10–5.32 P < 0.001) for CPSF3 and (HR: 2.39, 95% CI: 1.64–3.48 P < 0.001) for AJCC stage) and ICGC HCC cohorts ((HR: 3.17, 95% CI: 1.72–5.86 P < 0.001) for CPSF3 and (HR: 2.38, 95% CI: 1.27–4.48 P = 0.007) for AJCC stage).

Nomogram construction

In the multivariate Cox regression analysis, AJCC stage and CPSF3 were identified as independent risk factors for the OS of patients with HCC. Therefore, the AJCC stage and CPSF3 were used to construct a nomogram predictive of the OS of patients with HCC (Fig. 4A). The prognostic potential of the nomogram was assessed in TCGA training dataset and the ICGC validation dataset using a time-dependent ROC analysis. The AUC values at 1, 2, and 3 years were 0.711, 0.677, and 0.704 in TCGA and 0.788, 0.712, and 0.715, respectively, in ICGC (Fig. 4B–C). Agreement between the survival probability predicted by the nomogram and actual survival probability was evaluated by calibration plots. The results showed excellent agreements in predicting 1-, 2- and 3-year OS by the nomogram in both TCGA and ICGC datasets (Fig. 4D–I).

Fig. 4.

Fig. 4

Open in a new tab

Prognostic potential of the nomogram in predicting overall survival of HCC patients from TCGA and ICGC (A) Nomogram constructed using TNM stage and CPSF3 expression; (B) The time-independent ROC curve analysis of the nomogram in TCGA training dataset; (C) The time-independent ROC curve analysis of the nomogram in ICGC validation dataset; (D-I) The calibration curve of the nomogram for predicting overall survival at one year, two years, and three years in the TCGA training dataset and ICGC validation dataset.

Relationship between CPSF3 and tumor immune infiltration and Sorafenib response

Mounting evidence has demonstrated that the tumor immune environment plays a vital role in tumor progression39–43. We investigated the relationship between CPSF3 expression level and TIICs using the TIMER dataset. CPSF3 significantly correlated with purity (cor = 0.204, p = 1.3e-04), B cell (cor = 0.342, P = 7.51e-11), CD8 + T cells (cor = 0.355, P = 1.37e-11), CD4 + T cells (cor = 0.152, P = 4.71e-03), macrophage (cor = 0.333, P = 2.69e-10), neutrophil (cor = 0.202, P = 1.57e-04), and dendritic cell (cor = 0.377, P = 6.75e-13) (Fig. 5A–B).

Fig. 5.

Fig. 5

Open in a new tab

Relationship between CPSF3 expression level and tumor infiltration/ drug response. (A-B) Relationship between CPSF3 expression level and tumor infiltration (C) Differential expression of CPSF3 between sorafenib-resistant and sorafenib-responsive groups. (D) Predictive value of CPSF3 for sorafenib response.

Drug response plays a pivotal role in determining the treatment outcome and patient prognosis44–47. Sorafenib is the standard chemotherapeutic drug for advanced HCC and widely used in clinical practice48. We explored the relationship between CPSF3 and sorafenib response in GSE109211. The level of CPSF3 expression was significantly greater in the sorafenib-resistant group than in the sorafenib-sensitive group (Fig. 5C). ROC analysis showed the moderate predictive ability of CPSF3 in sorafenib response (AUC = 0.850) (Fig. 5D).

Gene set enrichment and biological functions

LinkedOmics was used to explore co-expression genes and the underlying biological mechanism of CPSF3 in HCC. The results revealed 5276 genes that were favorably associated with CPSF3 (FDR < 0.05) whereas 4949 genes were significantly inversely associated with CPSF3 (Fig. 6A). The top 50 genes that are negatively and positively associated with CPSF3 were identified using heatmaps, as shown in Fig. 6B–C. Next, GO and KEGG enrichment analysis were conducted to explore the biological function of CPSF3 co-expression genes. Ribonucleoprotein complex biogenesis, chromosome segregation, mRNA processing, DNA replication, positive regulation of cell motility, lipid localization, angiogenesis, extracellular structure organization, and regulation of small GTPase mediated signal transduction were the top 10 biological processes. Spliceosomal complex, cytosolic part, chromosomal region, mitochondrial inner membrane, membrane region, cell leading edge, cell-cell junction, apical part of cell, receptor complex, and extracellular matrix were the most related cellular components. The most strongly linked molecular functions were the structural constituent of the ribosome, catalytic activity acting on RNA and DNA, unfolded protein binding, oxidoreductase activity acting on paired donors, with incorporation or reduction of molecule, vitamin binding, guanyl-nucleotide exchange factor activity, lipid transporter activity, growth factor binding, and extracellular matrix structural constituent. KEGG enrichment analysis revealed that ribosome, spliceosome, RNA transport, cell cycle, Huntington disease, systemic lupus erythematosus, cGMP-PKG signaling pathway, MAPK signaling pathway, focal adhesion, and complement and coagulation cascades were the most involved pathways (Fig. 6D–G).

Fig. 6.

Fig. 6

Open in a new tab

Co-expression genes of CPSF3 in TCGA HCC cohort and the underlying biological mechanisms (A) Co-expressed genes related with CPSF3, (B-C) Heat maps presenting the top 50 genes negatively and positively correlated with CPSF3 in TCGA HCC cohort, (D-G) Functional enrichment analysis of CPSF3 in TCGA HCC cohort by GO and KEGG. Abbreviations: BP, biological process; CC, cellular component; MF, molecular function; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; TCGA, The Cancer Genome Atlas; HCC, hepatocellular carcinoma.

Genetic alterations and DNA methylation influence CPSF3 expression in patients with HCC

The cBioPortal tool was used to investigate CPSF3 genetic alterations in patients with HCC. There are 10% genetic alterations (Fig. 7A). The CPSF3 expression level differed among various alteration groups. The CPSF3 mRNA expression level was significantly higher in the gain group and prominently higher in the amplification group compared with the diploid alternation group (Fig. 7B). Survival analysis showed better survival in the patients with unaltered CPSF3 gene (P = 0.018) (Fig. 7C). The posttranscriptional modifications of CPSF3 include phosphorylation, acetylation, ubiquitination, glutathionylation, and sumoylation (Fig. 7D).

Fig. 7.

Fig. 7

Open in a new tab

The influence of genetic alternations/ DNA methylation on CPSF3 expression in HCC patients (A)Heat map of genetic alternations of CPSF3 in the TCGA HCC datasets; (B) The CPSF3 expression among different copy-number alternations groups; (C) Survival analysis between altered and unaltered CPSF3 groups; (D) posttranscriptional modification of CPSF3; (E) Mean DNA methylation of CPSF3 in tumor samples and non-tumor samples in the TCGA HCC cohort; (F) Correlation between CPSF3 expression and CPSF3 methylation in 6 methylation sites.

The mean methylation value of CPSF3 was explored in HCC tissues and normal tissues using the Wanderer web tool. The mean methylation value of CPSF3 differed significantly in tumor and normal tissues including at cg23889771, cg20549545, cg00024812, cg07974891, cg12057242, and cg18794882 (Fig. 7E). The correlation between CPSF3 methylation in the six probes and CPSF3 mRNA was further studied. We discovered that CPSF3 expression was positively correlated with methylation at cg23889771 (r = 0.150 P = 0.039) and negatively correlated with methylation at cg07974891 (r = 0.240 P = 0.001) and cg12057242 (r = 0.150 P = 0.035) (Fig. 7F).

Knocking down of CPSF3 inhibited Huh-7 and LM3 cell proliferation, invasion, and migration

Western blotting showed that the expression of CPSF3 in Huh7 and LM3 cells was higher than that in the THLE-2 cell line (Fig. 8A). To investigate the role of CPSF3 in HCC cells, siRNAs (siRNA-1 and siRNA-2) were constructed to inhibit the expression of CPSF3 in HCC cell lines. Western blotting confirmed effective downregulation of CPSF3 in the siRNA groups compared with the negative control (NC) group (Fig. 8B–C). Transwell assay revealed a significant reduction in the number of cells that traversed the membrane in the siRNA-treated group compared with the NC group (Fig. 8D). CPSF3 siRNA groups showed dramatically increased apoptosis in flow cytometry analysis (Fig. 8E). The impact of CPSF3 on cell proliferation was examined using cell cycle and Edu staining assays. The proportion of HCC cells in G0/G1 phase rose dramatically in the CPSF3 siRNA groups compared with the NC group, as shown by flow cytometry (Fig. 8G). A reduced percentage of Edu-positive cells was observed in the siRNA group compared with the NC group (Fig. 8H). Cell scratch assays showed that the CPSF3 siRNA groups exhibited dramatically reduced migration compared with the NC group (Fig. 8F). We further investigated the expression levels of proteins associated with epithelial–mesenchymal transition and cell proliferation. Western blot analysis indicated a decrease in the expression levels of N-cadherin, vimentin, and PCNA coupled with an upregulation of E-cadherin expression in the siRNA group relative to the NC group (Fig. 8I). These findings suggested that CPSF3 may facilitate liver cancer invasion, proliferation, and migration.

Fig. 8.

Fig. 8

Open in a new tab

Functional verification of CPSF3 in Huh-7 and LM3 cell lines. (A) Western blot analysis of CPSF3 expression in THLE-2, Huh7, and LM3 cell lines; (B) Evaluation of CPSF3 knockdown efficiency by Western blot in Huh7 cells; (C) Evaluation of CPSF3 knockdown efficiency by Western blot in LM3 cells; (D) Transwell assay quantifying the number of invasive cells in the NC group and CPSF3 knockdown group; (E) Flow cytometry analysis of apoptotic cell ratios in the NC group and CPSF3 knockdown group; (F) Wound healing assay evaluating migration ability in the NC group and CPSF3 knockdown group; (G) Flow cytometry analysis of cell cycle distribution in the NC group and CPSF3 knockdown group; (H) Edu assay measuring the proportion of Edu-positive cells in the NC group and CPSF3 knockdown group; (I) Western blot analysis of changes in EMT and proliferation-related proteins in the NC group and CPSF3 knockdown group.

Discussion

CPSF3 is an RNA endonuclease that belongs to the CPSF family. CPSF3 is implicated in mRNA cleavage at a location downstream of the poly(A) addition signal, which is where both polyadenylation occurs and histone messenger RNA precursors (pre-mRNAs) are processed49. Ning et al. reported that all members of the CPFS family were upregulated in the lung cancer sample from TCGA lung cancer cohort, and the overexpression of CPFS3 correlated with the prognosis and recurrence of lung adenocarcinoma. In addition, it can provide a determination by separating two histological subgroups of non-small cell lung cancer25. In the study of Luo et al., CPSF3 expression was shown to be considerably higher in colorectal cancer tissues than in non-tumor tissues. The authors revealed that CPSF3 interacts with lncRNA CASC9, exerting its oncogenic activity by regulating TGF-β signaling in colorectal cancer cells50. Ross et al. showed that in a fraction of acute myeloid leukemia and Ewing’s sarcoma cancer cell lines, CPSF3 indicates a synthetic lethal node. Inhibition of CPSF3 produces tumor-selective stasis and upregulates apoptosis in mouse xenografts24. Together, these studies indicate CPSF3 is a tumorigenic factor in cancer development and a potential therapeutic target.

To the best of our knowledge, the function and the underlying mechanisms of CPSF3 in HCC have been rarely studied. To determine the importance of CPSF3 in HCC, a bioinformatics study was carried out. Using TCGA and GTEx datasets, CPSF3 was found to be increased in tumor tissues relative to non-tumor tissues in a variety of malignancies, which is in line with the previous studies25. CPSF3 expression was examined in liver cancer, and prominent overexpression of CPSF3 was consistently demonstrated in 10 liver cancer datasets from GEO, TCGA, and ICGC. We examined the diagnostic and prognostic role of CPSF3 in HCC. CPSF3 exhibited great potential in discriminating between liver cancer and normal tissues and a predictive role in determining OS. Notably, high CPSF3 expression is significantly correlated with poor survival in HCC from TCGA and ICGC datasets. Further analysis revealed that CPSF3 mRNA was positively correlated with advanced stage and differentiation in HCC. CPSF3 was identified as an independent risk predictor for OS in HCC in both univariate and multivariate Cox regression analyses. Nomograms constructed using stage and CPSF3 expression exhibited great potential in predicting OS in HCC cohorts. The multi-kinase inhibitor sorafenib is the standard systemic therapy for advanced HCC, and VEGF receptors, KIT, and the mitogen-activated protein kinase/extracellular signal-regulated kinase pathway are the three targets of sorafenib. Here we found CPSF3 predicts HCC response to sorafenib, which is of great clinical significance.

CPSF3 is crucial for a variety of physiological activities. Our analysis revealed that CPSF3 is mainly involved in biological processes like mRNA processing, chromosomal segregation, and ribonucleoprotein complex biogenesis and pathways like ribosome, spliceosome, and RNA transport pathways. Thies information provides clues for future study directions. Further experimental research is warranted to elucidate the biological role of CPSF3 in liver cancer.

The tumor immune environment plays a crucial function in tumorigenesis and disease progression51. We explored the association between CPSF3 and TIICs. Our results suggest a significant association between CPSF3 and TIICs. Thus, the contact between CPSF3 and tumor immune cells may be the underlying mechanism for the development of liver cancer,.

We looked more deeply into the CPSF3 upregulation mechanisms. Genetic alteration analysis revealed shallow deletion, diploid, gain, and amplification alternations in CPSF3 in HCC. CPSF3 mRNA expression level was the highest in the amplification alteration group. A previous study examined DNA methylation, a frequent cause of changes in gene expression, in HCC52. The results indicated that CPSF3 expression was positively connected with methylation at cg23889771 and negatively correlated with methylation at cg07974891 and cg12057242. Therefore, we infer that genetic alteration and DNA methylation may be the underlying causes influencing CPSF3 expression.

This study investigated the role of CPSF3 in liver cancer, and our findings shed light on the clinical value of CPSF3 in liver cancer. However, this study has several limitations. Our study did not examine expression levels or prognostic relevance within our own collection of HCC specimens, instead depending solely on data sourced from public databases that display considerable variability in species origins. Therefore, future research should incorporate more tissue samples for confirmation. Additionally, while we performed functional phenotypic assays in two cell lines with CPSF3 knockdown, we did not carry out overexpression studies and we did not examine downstream targets. More studies are required using transcriptome sequencing or mass spectrometry analysis following CPSF3 knockdown to identify potential interacting molecules. Finally, we did not conduct animal experiments or design small-molecule inhibitors; therefore, future studies should include in vivo validation to further assess CPSF3 as a therapeutic target and develop small-molecule drugs to facilitate clinical translation.

Conclusion

This study demonstrates that CPSF3 is significantly overexpressed in HCC and its overexpression correlates with poor patient survival. CPSF3 shows strong diagnostic and prognostic value in HCC, and a nomogram combining CPSF3 with AJCC stage exhibited high predictive accuracy for HCC. Mechanistically, CPSF3 may promote HCC progression by modulating the tumor immune microenvironment, with its expression potentially regulated by genetic alterations and DNA methylation.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (229.2MB, pptx)

Abbreviations

HCC

Hepatocellular carcinoma

ROC

Receiver operator characteristic

TIMER

Tumor immune estimation resource

AFP

Alpha-fetoprotein

CPSF

Cleavage polyadenylation specificity factor

TCGA

The Cancer Genome Atlas

GTEx

Genotype-Tissue Expression Project

GEO

Gene Expression Omnibus

ICGC

International Cancer Genome Consortium

CI

Confidence interval

AUC

The area under the ROC curve

K-M

Kaplan-Meier

TIMER

Tumor Immune Estimation Resource

TIICs

Tumor-infiltrating immune cells

GO

Gene Ontology

BLCA

Bladder urothelial carcinoma

BRCA

Breast invasive cancer

CHOL

Cholangiocarcinoma

COAD

Colon adenocarcinoma

ESCA

Esophageal carcinoma

HNSC

Head and neck squamous cell carcinoma

KIRP

Kidney renal papillary cell carcinoma

LIHC

Liver hepatocellular carcinoma

LUAD

Lung adenocarcinoma

LUSC

Lung squamous cell carcinoma

STAD

Stomach adenocarcinoma

THCA

Thyroid carcinoma

UCEC

Uterine corpus endometrial carcinoma

BP

Biological processes

CC

Cellular components

MF

Molecular function

CRC

Colorectal cancer

Author contributions

W.K. Y.S and L.T conceived of the presented idea and drafted the manuscript. Y.Z and K.Z. modified figures and tables. X.W. and J.Z. conceived the presented idea and reviewed the draft. All authors agreed on the final version.

Funding

This research was supported by the Basic and Clinical Collaboration Enhancement Program Foundation of Anhui Medical University (2023xkjT038) and The University Natural Science Research Project of Anhui Province (2023AH053322).

Data availability

The data for this study were sourced from several public databases, including TCGA (https://www.cancer.gov/), GEO (https://www.ncbi.nlm.nih.gov/geo/), scCancer (https://bianlab.cn/scCancerExplorer/), ICGC (https://dcc.icgc.org/), TIMER (https://cistrome.shinyapps.io/timer), LinkedOmics (http://www.linkedomics.org/login.php), and Wanderer (http://maplab.imppc.org/wanderer/). For any further inquiries, please contact the corresponding author.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Weihao Kong, Yue Su, and Long Teng contributed equally to this work.

Contributor Information

Xingyu Wang, Email: wang_wxy@126.com.

Jianlin Zhang, Email: zhangjianlin@ahmu.edu.cn.

References

  • 1.Villanueva, A. Hepatocellular carcinoma. N. Engl. J. Med.380 (15), 1450–1462 (2019). [DOI] [PubMed] [Google Scholar]
  • 2.Li, H. et al. Peritumoral tertiary lymphoid structures correlate with protective immunity and improved prognosis in patients with hepatocellular carcinoma. Front. Immunol.12, 648812 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Li, H. et al. NCSTN promotes hepatocellular carcinoma cell growth and metastasis via β-catenin activation in a Notch1/AKT dependent manner. J. Exp. Clin. Cancer Res.39 (1), 128 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Cabibbo, G. et al. The present and the future of immunotherapy in hepatocellular carcinoma and biliary tract cancers. Cancer Treat. Rev.137, 102955 (2025). [DOI] [PubMed] [Google Scholar]
  • 5.An, L. & Li, Z. Molecular network of metabolic reprogramming and precision diagnosis and treatment of hepatocellular carcinoma. Biomark. Res.13 (1), 124 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Amin, N., Anwar, J., Sulaiman, A., Naumova, N. N. & Anwar, N. Hepatocellular carcinoma: a comprehensive review. Dis. (Basel Switzerland)13, 7 (2025). [DOI] [PMC free article] [PubMed]
  • 7.Li, J. et al. Epithelial-Mesenchymal transformation promotes the progression of hepatocellular carcinoma through NF-κB/MMP9 axis. Discov. Med.36 (183), 666–677 (2024). [DOI] [PubMed] [Google Scholar]
  • 8.Chen, W. et al. Cancer incidence and mortality in China, 2014. Chin. J. Cancer Res. = Chung-kuo Yen Cheng Yen Chiu. 30 (1), 1–12 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Malik, J. et al. Comprehensive evaluation of MicroRNA as a biomarker for the diagnosis of hepatocellular carcinoma. World J. Gastroenterol.28 (29), 3917–3933 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Lin, X., Luo, L., Zou, Y. & Chen, J. Cancer stemness-associated LINC02475 serves as a novel biomarker for diagnosis and prognosis prediction of hepatocellular carcinoma. Front. Genet.13, 991936 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Chen, T. Circulating Non-Coding RNAs as potential diagnostic biomarkers in hepatocellular carcinoma. J. Hepatocellular Carcinoma. 9, 1029–1040 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Luo, P. et al. A Large-scale, multicenter serum metabolite biomarker identification study for the early detection of hepatocellular carcinoma. Hepatol. (Baltimore Md). 67 (2), 662–675 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Misra, A. & Green, M. R. From polyadenylation to splicing: dual role for mRNA 3’ end formation factors. RNA Biol.13 (3), 259–264 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Liu, H. & Moore, C. L. On the Cutting Edge: Regulation and Therapeutic Potential of the mRNA 3’ End Nuclease. Trends Biochem. Sci. (2021). [DOI] [PMC free article] [PubMed]
  • 15.Sakai, A. et al. Aberrant expression of CPSF1 promotes head and neck squamous cell carcinoma via regulating alternative splicing. PloS One. 15 (5), e0233380 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Nilubol, N., Boufraqech, M., Zhang, L. & Kebebew, E. Loss of CPSF2 expression is associated with increased thyroid cancer cellular invasion and cancer stem cell population, and more aggressive disease. J. Clin. Endocrinol. Metab.99 (7), E1173–1182 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Yang, Q. et al. Cleavage and polyadenylation specific factor 4 promotes colon cancer progression by transcriptionally activating hTERT. Biochim. Biophys. Acta Mol. cell. Res.1866 (10), 1533–1543 (2019). [DOI] [PubMed] [Google Scholar]
  • 18.Ma, Z. et al. Comprehensive investigation identifies CPSF3 as a novel prognostic and oncogenic biomarker in bladder cancer. Discover Oncol.16 (1), 1847 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Lu, Y., Wang, T., Yan, X. & Zhang, H. Comprehensive assessment of cleavage and polyadenylation specificity factors in hepatocellular carcinoma: Expression, prognostic significance and immune infiltration analysis. Mol. Clin. Oncol.22 (6), 60 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Zhou, X., Zhao, M., Fan, Y. & Xu, Y. Identification of a necroptosis-related gene signature for making clinical predictions of the survival of patients with lung adenocarcinoma. PeerJ12, e16616 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Zhang, Y. et al. CPSF3 regulates alternative polyadenylation of CNIH2 to promote esophageal squamous cell carcinoma progression. Cancer Lett.593, 216925 (2024). [DOI] [PubMed] [Google Scholar]
  • 22.Lin, P. et al. RBBP6 maintains glioblastoma stem cells through CPSF3-dependent alternative polyadenylation. Cell. Discovery. 10 (1), 32 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Alahmari, A. A. et al. CPSF3 Inhibition blocks pancreatic cancer cell proliferation through disruption of core histone mRNA processing. RNA (New York NY). 30 (3), 281–297 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Ross, N. T. et al. CPSF3-dependent pre-mRNA processing as a druggable node in AML and ewing’s sarcoma. Nat. Chem. Biol.16 (1), 50–59 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Ning, Y., Liu, W., Guan, X., Xie, X. & Zhang, Y. CPSF3 is a promising prognostic biomarker and predicts recurrence of non-small cell lung cancer. Oncol. Lett.18 (3), 2835–2844 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Pagani, G. & Gandellini, P. Cleavage and polyadenylation machinery as a novel targetable vulnerability for human cancer. Cancer Gene Ther.31 (7), 957–960 (2024). [DOI] [PubMed] [Google Scholar]
  • 27.Lian, Q. et al. HCCDB: A database of hepatocellular carcinoma expression atlas. Genom. Proteom. Bioinform.16 (4), 269–275 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Pinyol, R. et al. Molecular predictors of prevention of recurrence in HCC with Sorafenib as adjuvant treatment and prognostic factors in the phase 3 STORM trial. Gut68 (6), 1065–1075 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Huang, C. et al. ScCancerExplorer: a comprehensive database for interactively exploring single-cell multi-omics data of human pan-cancer. Nucleic Acids Res.53 (D1), D1526–d1535 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Li, T. et al. TIMER: A web server for comprehensive analysis of Tumor-Infiltrating immune cells. Cancer Res.77 (21), e108–e110 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Li, B. et al. Comprehensive analyses of tumor immunity: implications for cancer immunotherapy. Genome Biol.17 (1), 174 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Vasaikar, S. V., Straub, P., Wang, J. & Zhang, B. LinkedOmics: analyzing multi-omics data within and across 32 cancer types. Nucleic Acids Res.46 (D1), D956–d963 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Kanehisa, M., Furumichi, M., Sato, Y., Matsuura, Y. & Ishiguro-Watanabe, M. KEGG: biological systems database as a model of the real world. Nucleic Acids Res.53 (D1), D672–d677 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Kanehisa, M. Toward Understanding the origin and evolution of cellular organisms. Protein Sci. Publ. Protein Soc.28 (11), 1947–1951 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Cerami, E. et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov.2 (5), 401–404 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Gao, J. et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal.6 (269), pl1 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.de Bruijn, I. et al. Analysis and visualization of longitudinal genomic and clinical data from the AACR project GENIE biopharma collaborative in cBioPortal. Cancer Res.83 (23), 3861–3867 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Díez-Villanueva, A., Mallona, I. & Peinado, M. A. Wanderer, an interactive viewer to explore DNA methylation and gene expression data in human cancer. Epigenetics Chromatin.8, 22 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Tian, H. et al. A novel FTO-targeting nanodrug induces Disulfidptosis and ameliorates the suppressive tumor immune environment to treat uveal melanoma. Biomaterials319, 123168 (2025). [DOI] [PubMed] [Google Scholar]
  • 40.Ishida, H. et al. Characterizing the tumor immune environment in thymic epithelial tumors using T-cell receptor repertoire analysis and gene expression profiling. JTCVS open.25, 435–465 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Hu, H. et al. Single-Cell sequence and machine learning identify a CD79A + B Cells-Related transcriptional signature for predicting clinical outcomes and immune microenvironment in breast cancer. Cancer Inform.24, 11769351251360675 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Xiong, B. et al. A Multi-Omics prognostic model capturing tumor stemness and the immune microenvironment in clear cell renal cell carcinoma. Biomedicines12, 10 (2024). [DOI] [PMC free article] [PubMed]
  • 43.Xiang, K. et al. TM-Score predicts immunotherapy efficacy and improves the performance of the machine learning prognostic model in gastric cancer. Int. Immunopharmacol.134, 112224 (2024). [DOI] [PubMed] [Google Scholar]
  • 44.Yousef, E. H., El Gayar, A. M. & El-Magd, N. F. A. Insights into Sorafenib resistance in hepatocellular carcinoma: Mechanisms and therapeutic aspects. Crit. Rev. Oncol./Hematol.212, 104765 (2025). [DOI] [PubMed]
  • 45.Sugimoto, R. et al. Advances in systemic therapy for hepatocellular carcinoma and future prospects. Curr. Oncol. (Toronto Ont)32, 9 (2025). [DOI] [PMC free article] [PubMed]
  • 46.Chen, R. et al. Regulated cell death in lenvatinib resistance of hepatocellular carcinoma: from molecular mechanisms to therapeutic strategies. Int. J. Biol. Sci.21 (5), 2012–2026 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Jiang, L. et al. Drug resistance mechanism of kinase inhibitors in the treatment of hepatocellular carcinoma. Front. Pharmacol.14, 1097277 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Llovet, J. M. et al. Sorafenib in advanced hepatocellular carcinoma. N. Engl. J. Med.359 (4), 378–390 (2008). [DOI] [PubMed] [Google Scholar]
  • 49.Mandel, C. R. et al. Polyadenylation factor CPSF-73 is the pre-mRNA 3’-end-processing endonuclease. Nature444 (7121), 953–956 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Luo, K. et al. LncRNA CASC9 interacts with CPSF3 to regulate TGF-β signaling in colorectal cancer. v. 38 (1), 249 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Cariani, E. et al. Immunological and molecular correlates of disease recurrence after liver resection for hepatocellular carcinoma. PloS One. 7 (3), e32493 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Esteller, M. Cancer epigenomics: DNA methylomes and histone-modification maps. Nat. Rev. Genet.8 (4), 286–298 (2007). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (229.2MB, pptx)

Data Availability Statement

The data for this study were sourced from several public databases, including TCGA (https://www.cancer.gov/), GEO (https://www.ncbi.nlm.nih.gov/geo/), scCancer (https://bianlab.cn/scCancerExplorer/), ICGC (https://dcc.icgc.org/), TIMER (https://cistrome.shinyapps.io/timer), LinkedOmics (http://www.linkedomics.org/login.php), and Wanderer (http://maplab.imppc.org/wanderer/). For any further inquiries, please contact the corresponding author.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES