2.1 Cell lines

The human embryonic kidney cell line 293T were purchased from American Type Culture Collection (ATCC) (Manassas, VA, USA). The human hepatoma cell line HepG2 were purchased National Collection of Authenticated Cell Cultures (Shanghai, China). The HBV transgenic cell line HepG2-4D14 and human hepatoma cell line Huh7 were gifts from Dr. Dongping Xu (The Fifth Medical Center of Chinese PLA General Hospital, Beijing, China). The Huh7 cell was verified by short tandem repeat sequence in 2023. All the cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM, 12800-017, Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS, ST30-3302, PAN-Biotech, Aidenbach, Germany).

2.2 Datasets and analysis

The differentially expressed genes (DEGs) in HCC tissues compared to normal tissues were downloaded from Gene Expression Profiling Interactive Analysis (GEPIA) in which the RNA expression data from the Cancer Genome Atlas (TCGA, http://gepia.cancer-pku.cn/) were analyzed, and they were overlapped with the DEGs in HepG2-4D14 compared to HepG2 cells. The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA-LIHC) datasets were downloaded from TCGA, and the ETV4 mRNA level in tumors and normal tissues was analyzed. GSE64041, GSE83148, GSE94660 and GSE84005 were obtained from National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO) profiles, and the relative ETV4 mRNA level in these cohorts was analyzed. JASPAR (https://jaspar.genereg.net/) was used to predict the ETV4-binding sites on the promoter of TNF-α or MAPK11. HCC single-cell transcriptome data were from public database (https://db.cngb.org/PRHCCdb/), in which the expression profiles of tumor cells and different kinds of immune cells were obtained [25]. xCell (https://comphealth.ucsf.edu/app/xcell) was used to perform cell type enrichment analysis from gene expression data, and enumerate immune cell subsets from TCGA-LIHC transcriptomes [26].

2.3 Patient samples

Twenty-seven HCC tumor tissues and paired peritumoral tissues were defined as Cohort 1 in which the patients were underwent surgery between August 2012 and December 2014. Total RNA was extracted from all the tumor tissues and nontumor tissues and analyzed by quantitative real-time PCR (qRT-PCR) to measure ETV4 mRNA levels. A total of 128 formalin-fixed paraffin-embedded (FFPE) primary HCC liver specimens were defined as Cohort 2 and subjected to Immunohistochemistry (IHC) staining for ETV4, TNF-α, MAPK11 and CD68. Patients in Cohort 2 were hospitalized and underwent surgery between Jun 2012 and Dec 2013. The patients were all diagnosed with primary HCC, with no cholangiocarcinoma or other cancers. Patients in Cohort 2 were divided into HBV-low (≤40 DNA copies/mL) and HBV-high (>40 DNA copies/mL) groups according to the serum HBV DNA copy number before surgery. Sixty-six HCC patients who underwent surgery between July 2012 and November 2013 were defined as Cohort 3. The patients in Cohort 3 were divided into HBV-negative (both HBsAg and HBeAg were negative, and HBV DNA was undetectable in serum) and HBV-positive groups. The clinical characteristics of the enrolled patients are listed in Supplementary Table S1. All the patient samples were collected from the Fifth Medical Center of Chinese PLA General Hospital. All the participants provided written informed consent. The patient samples were assigned arbitrary identification numbers based on the order of enrollment.

For the single-marker overall survival (OS) analyses, the quartile was used to set the group cutoff according to the levels of ETV4, TNF-α or MAPK11. The top 25% was classified as high, the bottom 25% was classified as low, and the medium 50% was classified as intermediate.

For the double-marker OS analyses, firstly the medium was used to set the group cutoff according to the levels of ETV4, TNF-α or MAPK11, in which the top 50% was classified as high, the other 50% was classified as low. Then, patients with both ETV4-high and TNF-α-high were defined as ETV4+TNF-α-high, while patients with both ETV4-low and TNF-α-low were defined as ETV4+TNF-α-low. The other patients were classified as intermediate. The same group classification was applied for ETV4+MAPK11 groups.

2.4 Real-time quantitative PCR (qRT-PCR)

The human and mouse liver tissues were homogenized with the tissue grinder. Total RNA was extracted from tissues and cell lines with TRIzol (1559608, Invitrogen, Carlsbad, CA, USA). cDNA was synthesized with Hifair II 1^st strand cDNA Synthesis SuperMix (11123ES60, YEASEN, Shanghai, China). The random and oligo(dT)s primer mixtures were used for the cDNA synthesis. The reaction mixture was pre-denatured for 5 min at 95°C, then denatured for 10 s at 95°C, annealed and extended for 30 s at 60°C, for a total 40 cycles. qRT-PCR was performed with qPCR SYBR Green Master Mix (11202ES08, YEASEN, Shanghai, China). The primers specific for the indicated genes used in qRT-PCR are listed in Supplementary Table S2.

2.5 RNA sequencing and analysis

The human ETV4 gene was cloned into the lentiviral vector pLL-IRES-puro (provided by Dr. Wenjun Liu, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China) and named as pLL-ETV4. Short hairpin RNAs (shRNAs) targeting human ETV4 were cloned into the pSIH1-GFP lentiviral plasmid. The pLL-ETV4 vector or the pSIH1-GFP-shETV4 vector and the lentivirus packing plasmids were co-transfected into 293T cells. Lentivirus was harvested 72 h post transfection. HepG2 cells were infected with lentivirus carrying ETV4 to build the ETV4-overexpressing HepG2 cells (HepG2-ETV4-oe). Huh7 cells were infected with lentivirus carrying ETV4 shRNA to build the ETV4-knockdown Huh7 cells (Huh7-shETV4). Total RNA was extracted from HepG2-ETV4-oe and control cells or Huh7-shETV4 and control cells. The mRNAs were enriched and subjected to mRNA library construction and sequencing by BGI (BGISEQ-500 platform, Shenzhen, China). The raw reads were filtered and quantity controlled to obtain clean reads. Then, the clean reads were aligned with the Homo sapiens genome (GCF_000001405.37_GRCh38.p11). The gene abundance was measured in fragments per kilobase of exon per million fragments mapped (FPKM). The genes with fold change higher or lower than 2 in HepG2-ETV4-oe vs HepG2 cells, or Huh7-shETV4 vs. Huh7 cells, were defined as DEGs. The DEGs within the groups were further analyzed with Kyoto Encyclopedia of Genes and Genomes (KEGG) for signaling pathway enrichment.

2.6 Dual-luciferase reporter assay

Human TNF-α and MAPK11 promoters and their mutants were cloned into the pGL4 basic vector. HEK293T cells were seeded in 24-well plates (3×10⁵ cells per well) and transfected with the pLL-ETV4 or control vector together with the pGL4-TNF-α/MAPK11 promoter and pRL-TK Renilla reporter plasmids as an internal control for 24 h. Then, the cell lysates were harvested and subjected to luciferase assay with the dual-specific luciferase reporter assay system (E1980, Promega, Madison, WI, USA).

2.7 Chromatin immunoprecipitation (CHIP) assay

The CHIP assay was performed with the CHIP assay kit (P2078, Beyotime, Wuhan, Hubei, China). Briefly, Huh7 cells were cross-linked with 4% formaldehyde for 10 min, then treated with 125 mmol/L glycine for 5 min. Cells were collected and sonicated. The samples were centrifuged for 10 min at 8,000 g at 4°C. The supernatants were subjected to immunoprecipitation (IP) with IgG or anti-ETV4 antibodies for 4 h, followed by incubation with protein A Sepharose beads for 2 h. The IP samples were eluted and treated with proteinase K, and then the DNA were extracted with phenol-chloroform and subjected to qRT-PCR with indicated primers.

2.8 Mouse models

C57BL/6 ETV4 conditional knockout mice (ETV4^{fl/fl, alb-cre}) were generated with the CRISPR/Cas9-guided homologous recombination system (Biocytogen, Beijing, China). Briefly, an ETV4 targeting vector containing the ETV4 genome sequence from exon 6 to exon 13 with 5’loxp and 3’loxp flanking sites was constructed. Small guide RNA (sgRNA) sequences that effectively targeted intron 6 and the downstream sequence of exon 13 were selected with the UCA^TM assay, and Cas9/sgRNA plasmids were constructed. The ETV4 targeting vector and Cas9/sgRNA plasmids were microinjected into zygotes from the oviducts of 4-week-old female C57BL/6 mice. The genotype of 7-day-old male and female ETV4^fl/fl mice was confirmed by Southern blotting and DNA sequencing. For Southern blotting, the genomic DNA was digested with AseI, and then subjected to hybridization with digoxin-labeled probe of ETV4 exon 5. Hepatocyte-specific ETV4^{fl/fl, alb-cre} mice were generated by crossing ETV4^fl/fl mice with Alb-Cre mice.

To generate hepatocyte-specific ETV4 conditional transgenic mice (ETV4^{Rosa26-floxp-gfp-floxp-mETV4/+, alb-cre}, ETV4^Hep-TG), floxp-gfp-polyA-floxp mouse ETV4 (mETV4) was inserted into the Rosa26 locus of C57BL/6 wild type mice. Rosa26 is a non-coding gene on chromosome 6 of the mice. In general, the gene inserted into Rosa26 locus can be expressed normally while did not affect the expression of neighboring genes. The ETV4^Hep-TG mice were generated by crossing ETV4^{Rosa26-floxp-gfp-floxp-mETV4/+} mice with Alb-Cre mice.

HCC mouse model was generated as follows: 2-week-old male ETV4^fl/fl and ETV4^{fl/fl, alb-cre} or wild type and ETV4^Hep-TG mice were intraperitoneally injected with 25 mg/kg DEN (N0756, Sigma Aldrich, Saint Louis, MO, USA), followed by intraperitoneally injection of 0.5 mL/kg CCL₄ (C0731530924, Nanjing Reagent, Nanjing, China) every two weeks. The mice were euthanized via cervical dislocation at the 21-week-old. The livers were harvested and imaged, and the tumor number and size were measured. Hematoxylin and eosin (H&E) staining was performed and scanned by Leica digital pathology scanner Aperio CS2 (Wetzlar, Germany). The numbers of microscopic tumors were counted. The diameter of the tumor and the total tissue area were measured by Image Scope. Numbers of tumors per cm² were calculated.

For the acute inflammation model, 8-week-old male wild type and ETV4^Hep-TG mice were intraperitoneally injected with DEN (100 mg/kg) for 8 h and euthanized. Then the livers were fixed with formalin and subjected to IHC staining.

To establish chronic inflammation model, 2-week-old male ETV4^fl/fl and ETV4^{fl/fl, alb-cre} or wild type and ETV4^Hep-TG mice were intraperitoneally injected with a single dose of DEN (25 mg/kg), followed by CCL₄ (0.5 mL/kg) treatment every two weeks, and euthanized at 16 weeks. The livers were fixed with formalin and subjected to H&E staining or IHC staining with the indicated antibodies.

To deplete macrophages, 8-week-old male wild type and ETV4^Hep-TG mice were intravenously injected with 10 mg/kg GdCl₃ (203289, Sigma Aldrich, Saint Louis, MO, USA) every two days for 3 times [27, 28].

All the mice in this study were housed under specific pathogen-free (SPF) conditions at 25°C with 50% humidity. The mice were euthanized via cervical dislocation at the end of experiments. All the experiments were approved by the Ethical Review Committee of the Institute of Microbiology, Chinese Academy of Sciences (APIMCAS2019009).

2.9 IHC analysis

The human and mouse FFPE samples were sliced in 5 mm thickness and subjected to standard IHC procedures. In brief, the sections were treated with xylene firstly, and then hydrated in series of ethanol (99.9% ethanol, 95% ethanol, 75% ethanol, 50% ethanol, H₂O). The antigen retrieval was performed as follows: for ETV4 and TNF-α, the sections were treated with 1 mmol/L EDTA/10 mmol/L Tris HCl (pH 9.0); for MAPK11 and CD68, the sections were treated with 10 mmol/L sodium citrate (pH 6.0) for 30 min. The sections were blocked with 5% bovine serum albumin (BSA) (0032, Amresco, Solon, OH, USA) and incubated with primary antibodies (anti-ETV4, 1:50, overnight at 4°C; anti-MAPK11, 1:125, 2.5 h at room temperature; anti-TNF-α, 1:500, 2.5 h at room temperature; anti-CD68, 1:300, overnight at 4°C) and then incubated with horseradish peroxidase (HRP)-conjugated secondary antibody (PV-9000, ZSGB-BIO, Beijing, China) for 40 min at room temperature. The whole slide image was scanned by a NanoZoomer 2.0 HT (Hamamatsu, Japan) and analyzed with a HALO 2.3 digital pathology analysis system (Indica Labs, NM, USA). The tumor and adjacent nontumor areas were manually defined by using the HALO software (Albuquerque, NM, USA). The average density and percentage of positive cells for H-score was calculated with the equation: H-score = weak positive cell percentage × 1 + moderately positive cell percentage × 2 + strongly positive cell percentage × 3.

2.10 Primary mouse hepatocyte isolation

Mouse primary hepatocytes were isolated according to the procedures of Wen Zhang's Lab (http://www.mouselivercells.com/). Briefly, 8-week-old wild type or ETV4^Hep-TG male mice were anesthetized by intraperitoneally injection of 400 μg/g 2,2,2-tribromoethanol (T48402, Sigma Aldrich, Saint Louis, MO, USA), and then perfused with 100 units/mL type IV collagenase (17104, Gibco, Carlsbad, CA, USA) via the portal vein for 3-5 min until the livers were digested sufficiently. The livers were dissected out and teared apart to release the hepatocytes. The hepatocytes were collected after centrifugation (3 min, 50 g at 4°C) and plated. The ETV4, TNF-a and MAPK11 mRNA levels were measured by qRT-PCR.

2.11 Enzyme-linked immunosorbent assay (ELISA)

HepG2 cells overexpressing ETV4 or control cells were seeded in 24-well plates (2×10⁵ cells per well) for 36 h. Then, the supernatants were collected and subjected to ELISA with a human TNF-α ELISA kit (KIT10602, Sino Biological, Beijing, China).

2.12 Flow cytometry analysis

The mouse livers were minced into 1 mm³ pieces, followed by digestion with collagenase for 30 min. The cells were centrifuged at 350 g at 4°C for 3 min, and purified in 40% percoll. Subsequently, the cells were stained with anti-CD68 (137009, FA-11), anti-F4/80 (123110, BM8), anti-CD45 (103115, 30-F11), anti-CD11b (101212, M1/70), anti-CD3 (100205, 17A2), anti-CD4 (100515, RM4-5), anti-CD8 (100733, 53-6.7), anti-Ly6G (127617, 1A8) antibodies (BioLegend, San Diego, CA, USA), and sorted using a BD FACSAria II Cell Sorter (BD Bioscience, San Jose, CA, USA).

2.13 Statistical analysis

The OS of HCC patients after diagnosis was analyzed with the Kaplan-Meier estimator and calculated with the log-rank test. Student t test was used for two-group comparisons, and two-way analysis of variance (ANOVA) was used for multiple-group comparisons. P < 0.05 was considered significance.

3.1 The level of ETV4 was negatively associated with the survival of HCC patients

To investigate the HBV-related genes involved in the development of HCC, we overlapped the DEGs in the HBV transgenic cell line HepG2-4D14 compared to the parental cell line HepG2 and the DEGs in HCC tumor tissues compared to normal tissues in the TCGA-LIHC cohort (Figure 1A). Sixty-nine overlapped genes were identified (Supplementary Table S3), and the differential expression pattern is shown in Figure 1B. Among these 69 genes, ETV4 expression level was the most significantly associated with the OS of HCC patients (Supplementary Figure S1A), and the mRNA level of ETV4 was much higher in tumor tissues than in normal tissues according to the data from TCGA (Supplementary Figure S1B) and GEO HCC datasets (Supplementary Figure S1C). We next compared the mRNA level of ETV4 in 27 pairs of HCC tissues and their paired adjacent peritumoral tissues, and found that ETV4 was upregulated in HCC tumor tissues comparing to their paired adjacent peritumoral tissues (Figure 1C). The HCC datasets from GEO also showed that ETV4 was higher in tumors than in their paired peritumoral tissues (Supplementary Figure S1C-E). Then, we performed IHC staining of ETV4 in 128 HCC liver tissues which were collected from patients who underwent surgery between 2012 and 2013. The mean intensity of ETV4 staining was evaluated and representative images are shown in Figure 1D. Kaplan-Meier analysis demonstrated that patients with higher ETV4 levels had poorer survival rate (Figure 1E), which was consistent with the results shown in Supplementary Figure S1A, and both the percentage of ETV4-positive cells (Figure 1F) and the H-score (Supplementary Figure S1F) were higher in tumor tissues than in paired adjacent peritumoral tissues. All these data demonstrated that ETV4 was highly expressed in HCC tumor tissue, indicating that ETV4 level might predict poor survival rate of HCC patients.

3.2 ETV4 expression was upregulated in HBV-positive HCC and chronic hepatitis liver tissues

Both ETV4 mRNA and protein levels were higher in HepG2-4D14 cells than in parental HepG2 cells (Supplementary Figure S1G-H), which is consistent with the results in Figure 1B. We analyzed the relative ETV4 level in the GEO dataset (GSE83148), which included gene expression data from 122 liver samples from chronic hepatitis B (CHB) patients and 6 liver samples from healthy controls. The results indicated that the mRNA level of ETV4 was higher in the CHB samples than in the normal samples (Supplementary Figure S1I). Also, the mRNA levels of ETV4 were higher in tumor tissues from HBV-positive patients than in those from HBV-negative patients in Cohort 3 (Figure 1G). We then divided the HCC samples in Cohort 2 into two groups (HBV-low and HBV-high) according to the HBV DNA copy numbers in the serum of patients before surgery. The data of IHC staining demonstrated that the protein level of ETV4 was higher in tumor tissues from patients with a higher HBV load than in those from patients with a lower HBV load (Figure 1H). These data suggested that the ETV4 level was upregulated in HBV-related HCC.

In addition, we analyzed the TCGA dataset and found that the level of ETV4 mRNA was also higher in alcohol or non-alcoholic fatty liver disease-associated HCC tumors than in normal tissues (Supplementary Figure S1J).

3.3 ETV4 directly promoted the transcription of TNF-α and MAPK11

To identify downstream genes regulated by ETV4, we compared the transcription profiles of ETV4-overexpressing HepG2 cells (HepG2-ETV4-oe) and control cells (HepG2-Ctrl) and ETV4-knockdown Huh7 (Huh7-shETV4) cells and control cells (Huh7-shR) by RNA sequencing. We overlapped the DEGs in the 2 sets described above and identified 159 potential ETV4 downstream genes (Figure 2A-B). By using KEGG pathway analysis, we found that ETV4-upregulated genes were mainly involved in TNF, NF-κB and MAPK signaling pathways (Figure 2C). TNF-α is an inflammatory cytokine that abnormally expressed in many kinds of chronic or acute liver injury/inflammation [17, 29, 30]. MAPK11 (p38β) belongs to the family of the p38 MAPKs, which can induce the production of TNF-α and IL-1β [31-33]. Therefore, TNF-α and MAPK11 were evaluated in priority as they played critical roles in inflammation.

First, we confirmed that ETV4 knockdown in Huh7 cells caused a reduction in the mRNA levels of TNF-α and MAPK11 (Figure 2D) as well as the protein level of MAPK11 (Figure 2E). In contrast, overexpression of ETV4 in HepG2 cells enhanced both the mRNA and protein levels of TNF-α and MAPK11 (Figure 2F-H). We predicted the ETV4-binding sites on the promoters of TNF-α and MAPK11 on the JASPAR website. Then, we generated the luciferase reporters for the promoters of TNF-α and MAPK11 and performed luciferase assays. The data showed that ETV4 could significantly enhance the activities of the TNF-α and MAPK11 promoters (Figure 2I-J). Furthermore, we also performed luciferase assays with the promoters of TNF-α or MAPK11 in which the potential ETV4-binding sites were mutated, and identified the regions in the TNF-α and MAPK11 promoters that were critical for ETV4 binding (Figure 2I-J). In addition, we performed the CHIP assay, and the data showed that ETV4 could bind to the promoters of the TNF-α and MAPK11 directly in Huh7 cells (Figure 2K). In conclusion, ETV4 can directly promote the transcription of TNF-α and MAPK11.

3.4 ETV4 enabled macrophage accumulation and facilitated liver inflammation

To test whether ETV4 could regulate the expression of TNF-α and MAPK11 and enhance the liver inflammation in vivo, we generated hepatocyte-specific ETV4-transgenic mice (named as ETV4^Hep-TG) by crossing ETV4^{Rosa26-floxp-gfp-floxp-mETV4/+} mice with Alb-Cre mice (Figure 3A and Supplementary Figure S2A). The genotype of ETV4^Hep-TG was confirmed by PCR (Supplementary Figure S2B) and ETV4 protein was determined by immunoblotting (Figure 3B). We also generated hepatocyte-specific ETV4-knockout mice (ETV4^{fl/fl, alb-cre}) by crossing ETV4^fl/fl with Alb-Cre mice (Figure 3C-D and Supplementary Figure S2C-D). No significant changes of liver/body ratio were observed in 8-week-old ETV4 transgenic or ETV4 knockout mice (Supplementary Figure S2E-F).

The protein levels of hepatic TNF-α, MAPK11 and CD68 were examined. The data showed that levels of TNF-α, MAPK11 and CD68 were higher in ETV4^Hep-TG than that in wild type (WT) mice (Figure 3E-F). To further explore whether ETV4 enhanced the transcription of TNF-α and MAPK11 in hepatocytes, we isolated primary hepatocytes from the WT or ETV4^Hep-TG mice, and found that the mRNA levels of TNF-α and MAPK11 in hepatocytes were significantly higher in ETV4^Hep-TG than that in WT mice (Figure 3G). We further performed the flow cytometry analysis and found that both CD68⁺ and F4/80⁺ cells were significantly enriched in ETV4^Hep-TG than in WT mice (Figure 3H-I). The neutrophils were also higher in ETV4^Hep-TG mice than in WT mice, but there was no difference of CD4⁺ T cells and CD8⁺ T cells between WT and ETV4^Hep-TG mice (Supplementary Figure S3). Taken together, these results indicated that ETV4 promoted the expression of TNF-α and MAPK11 in hepatocytes and the accumulation of macrophages and neutrophils in the liver.

To investigate whether ETV4 promoted the expression of TNF-α and MAPK11 and the accumulation of macrophages during liver inflammation, we measured the protein levels of TNF-α, MAPK11 and CD68 in the livers of control mice and mice treated with DEN to induce acute inflammation and found that TNF-α, MAPK11 and CD68 were expressed higher in the ETV4^Hep-TG mice than in the WT mice (Figure 4A-B). We also performed macrophage depletion with GdCl₃ in mice treated with DEN and found that TNF-α expression was still higher in ETV4^Hep-TG than in WT mice (Supplementary Figure S4A-B). The above data indicated that ETV4 could upregulate TNF-α expression and enhance macrophage accumulation in the liver during acute inflammation. To uncover the effect of ETV4 during chronic liver injury and inflammation, we treated the mice with DEN combined with CCL₄ (Figure 4C). IHC staining showed that the levels of TNF-α, MAPK11 and CD68 in the livers of the ETV4^Hep-TG mice were higher than those in WT mice (Figure 4D-E), indicating that ETV4 could trigger liver inflammation and macrophage accumulation via upregulating TNF-α and MAPK11 expression when there was chronic liver injury. Otherwise, knockout of ETV4 caused a marked reduction in the TNF-α, MAPK11 and CD68 levels (Figure 4F-G). To completely exclude the possible influence of acute liver injury caused by CCl₄, we further measured the protein levels of TNF-α, MAPK11 and CD68 in the liver samples from the mice 7 days after the final injection of CCL₄ (Supplementary Figure S4C) and obtained consistent results in the ETV4^Hep-TG mice (Supplementary Figure S4D-E) and the ETV4^{fl/fl, alb-cre} mice (Supplementary Figure S4F-G). Overall, these results demonstrated that ETV4 enabled macrophage accumulation and facilitated liver inflammation via modulating the ETV4-TNF-α and ETV4-MAPK11 axis.

The data showed that ETV4 could promote the inflammation in the liver. In order to know whether inflammation could influence the level of ETV4, we treated the mice with DEN or CCL₄, and observed that the mRNA level of ETV4 in the liver was increased after treatment with DEN or CCL₄ (Supplementary Figure S5), indicating that inflammation may also trigger the transcription of ETV4.

3.5 ETV4 promoted liver tumorigenesis

Chronic inflammation plays an important role in HCC development. To identify whether ETV4 could promote liver tumorigenesis in vivo, we established the DEN-CCL₄-treated HCC model in transgenic mice (Figure 5A). After DEN-CCL₄ treatment, the number of total surface tumors and the number of tumors larger than 1 mm³ were lower in the livers of the ETV4^{fl/fl, alb-cre} mice than in those of the ETV4^fl/fl mice (Figure 5B-D), and the liver/body weight ratio was also lower in the ETV4^{fl/fl, alb-cre} mice than in the ETV4^fl/fl mice (Figure 5E). In addition, the number of microscopic tumors in the ETV4^{fl/fl, alb-cre} mice were lower than that in the ETV4^fl/fl mice (Figure 5F-H). These data indicated that knockout of ETV4 in hepatocytes suppressed liver tumor formation in DEN-CCL₄ HCC model.

A DEN-CCL₄ HCC model with ETV4^Hep-TG mice was also established (Figure 5I). Total number of surface tumors in the livers of the ETV4^Hep-TG mice was higher than those in the WT mice (Figure 5J-K). No significant difference was observed in the number of surface tumors larger than 1 mm³ and the liver/body ratio between the ETV4^Hep-TG and WT mice (Figure 5L-M), but total numbers of microscopic tumors and the numbers of large microscopic tumors were robust higher in the ETV4^Hep-TG mice than in the WT mice (Figure 5N-P). All these data suggested that ETV4 promoted HCC formation and could be a potential therapeutic target for HCC. In addition, the protein levels of TNF-α, MAPK11 and CD68 were lower in the liver tissues of the ETV4^{fl/fl, alb-cre} mice than in those of the ETV4^fl/fl mice (Figure 5Q-R) and significantly higher in the ETV4^Hep-TG mice than in the WT mice (Figure 5S-T), suggesting that ETV4 promoted the expression of TNF-α and MAPK11 and hepatic macrophage infiltration during the development of DEN-CCL₄-induced HCC.

3.6 ETV4 expression was positively correlated with TNF-α and MAPK11 expression and liver inflammation in human HCC samples

As shown in Figure 1C and Supplementary Figure S1C-F, the level of ETV4 was higher in HCC tumors than in adjacent peritumoral tissues. If ETV4 promoted liver tumorigenesis through TNF-α and MAPK11, there should be higher levels of TNF-α and MAPK11 in tumor tissues than in peritumoral tissues. As expected, there were more TNF-α- and MAPK11-positive cells and higher H-scores of TNF-α and MAPK11 in tumors than in peritumoral tissues according to the IHC staining results of Cohort 2 (Figure 6A-D, Supplementary Figure S6). TNF-α and MAPK11 expression were also higher in tissues with high ETV4 expression than in tissues with low ETV4 expression (Figure 6E). Then the mRNA levels of ETV4, TNF-α and MAPK11 in the HCC tumors in Cohort 3 were analyzed by qRT-PCR. The data showed that ETV4 expression was positively correlated with TNF-α and MAPK11 expression (Figure 6F-G). By analyzing the HCC single-cell transcriptome data from public database. we further found that ETV4 was mainly expressed in tumor cells (Figure 6H), and TNF-α and MAPK11 were co-expressed with ETV4 in tumor cells (Figure 6I-J). We divided the patients in Cohort 2 into different groups according to the levels of TNF-α, MAPK11 and ETV4 and performed Kaplan-Meier survival analysis. The results demonstrated that patients with higher TNF-α level had poorer survival than those with lower TNF-α level (Figure 6K), while the OS had no significant difference between patients with higher MAPK11 and those with lower MAPK11 (Figure 6L). Furthermore, patients with higher levels of ETV4+TNF-α had an even worse survival rate than those with lower levels of ETV4+TNF-α (Figure 6M), while patients with higher levels of ETV4+MAPK11 had poorer survival rate than those with lower levels of ETV4+MAPK11 (Figure 6N). These data suggested that ETV4+TNF-α might be used as prognostic markers for HCC patients. We analyzed the correlation of ETV4 level in tumor with the scores of immune cell types enumerated by xCell, and found that there were more monocytes (Figure 6O) and macrophages (Figure 6P) in ETV4-high tumors than that in ETV4-low tumors, indicating more monocyte and macrophage infiltration in ETV4-high tumors.

Taken together, our data suggested that ETV4 promoted liver inflammation and tumorigenesis via enhancing the TNF-α and MAPK11 signaling pathway (Figure 7).

AUTHOR CONTRIBUTIONS

Dandan Qi and Min Lu designed and performed the experiments, and wrote the draft. Pengfei Xu performed the IHC staining. Xinli Yao, Yongchen Chen, Yahua Cui and Lipeng Gan supported the mouse experiments. Yong Li performed the statistical analysis. Xiaomei Tong and Shuhong Liu contributed to the constructive discussions. Jingmin Zhao provided the clinical samples and contributed to the constructive discussions. Ningning Liu interpreted data and revised the manuscript. Xin Ye supervised the study and wrote the manuscript.