The tumor tissues and paired paracancerous tissues used in this study were obtained from 16 HCC patients who underwent hepatectomy in the department of Hepatobiliary Surgery of the Affiliated Hospital of Southwest Medical University. The specimens were frozen in liquid nitrogen immediately after removal. The histopathological diagnosis of HCC specimens was diagnosed by two pathologists independently, and both were diagnosed as HCC. This study was approved by the Institutional Review Board of the Affiliated Hospital of Southwest Medical University (Luzhou, China). Further, the RNA-seq data and clinical data with survival information were downloaded from UCSC Xena (https://xenabrowser.net/datapages/) to evaluate the predictive power of AC093797.1 and analyze the correlation with clinical indicators.

HCC cell lines, including HCCLM3, MHCC97-H, Huh7, and HepG2, and human normal liver cells LO2 were purchased from Procell Life Science and Technology Co., Ltd. (Wuhan, China). The overexpression vector of AC093797.1 (OE-AC093797.1) and empty vector (pcDNA3.1) were purchased from Shanghai Generay Biotech Co, Ltd. (Shanghai, China). HCCLM3, MHCC97-H, and Huh7 cells were cultured in DMEM medium (Gibco), LO2 cells were cultured in RPMI 1640 (Gibco), and HepG2 cells were cultured in MEM medium (Gibco). Above media were supplemented with 10% fetal bovine serum (FBS). All the cells were incubated in an incubator with 5% CO₂ at 37°C.

For the transfection, HCC cells were incubated in a 24-well polypropylene plate and transfected with the empty vector pcDNA3.1 or OE-AC093797.1 according to the instructions of Lipofectamine™ 2000 (Thermo, USA) when the cell confluence reaches about 70–90%. To obtain the stable transfected cell lines, the transfected HCC cells were screened by G418 (1 mg/ml) for 2 weeks.

Total RNA of liver tissues or cells was isolated using Trizol reagent. RT first Strand cDNA Synthesis Kit was used to reverse transcribe 2 μg of total RNA into cDNA, and SYBR Green qPCR Master Mix (High ROX) was used for qRT-PCR. All the above reagents were purchased from Servicebio Biotechnology Co., Ltd. (Wuhan, China). The housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the reference gene, and the relative expression of AC093797.1 was calculated via the 2^-△△Ct method. The primers used in this study were as follows: GAPDH forward primer: 5′-GGA​CCT​GAC​CTG​CCG​TCT​AG-3′; GAPDH reverse primer: 5′-GTA​GCC​CAG​GAT​GCC​CTT​GA-3′; AC093797.1 forward primer: 5′-TGC​CGC​AAG​GAG​GAG​GCT​ATT​GTT-3′; AC093797.1 reverse primer: 5′-TGG​GAA​GGC​TTA​TTC​ATG​GAC​CTA-3′.

Cell-counting Kit-8 (CCK8) reagent was used to determine cell proliferation. Briefly, post-transfection for 24 h, the HCCLM3 cells were seeded into 96-well plates at 3 × 10³ cells/well. At 0, 24, 48, 72, and 96 h, 10 μl CCK8 reagent was added to the wells and incubated at 37°C for an extra 4 h, then to detect the absorbance value at 450 nm by a microplate reader.

After 24 h of transfection, the transfected cells were plated into a 24-well plate (2.5 × 10⁵ cells/well). A sterile 200 μl plastic pipette tip was used to make scratches, the image of scratches at 0 and 48 h was captured by an inverted microscope at 40× magnification, and Image J was used to calculate the healing area of scratches.

The 24-well transwell chambers with 8 μm pore size were employed to assess the motility potential of HCC cells. For the invasion experiment, the bottom membrane of the chamber was pre-coated with 50 μl of Matrigel (3 mg/ml). After 24 h of transfection, the transfected cells were harvested and suspended in the serum-free medium at a density of 4 × 10⁵ cells/ml, then 100 μl of cell suspension were dispensed into the upper chamber. Complete medium (600 μl) was added to the lower chamber. After 24 h, adherent cells on the membrane were fixed with paraformaldehyde fixative and stained with 0.1% crystal violet. A cotton swab was used to remove non-migrated cells, the images of 10 random fields at 200× magnification were captured by a microscope, and the number of cells was counted.

The 5-week-old female BALB/c nude mice weighing 15–20 g were maintained under specific-pathogen-free conditions and randomly divided into control group and OE-AC093797.1 group (n = 5 per group). In the nude mice tumor formation experiment, stably transfected HCCLM3 cells (5 × 10⁷ cells in 100 μl) were subcutaneously injected into each nude mice. The mice with tumors were observed, and the volume of the tumor was calculated by the following formula: V_tumor = length × width² × π/6. When the tumor grows to about 2000 mm³, the mice were sacrificed after anesthesia and tumor tissues were isolated. All procedures for the nude mice experiment were approved by the Animal Care Committee of Southwest Medical University (Luzhou, China).

Part of the tumor tissue isolated from the nude mice was fixed with 4% paraformaldehyde and used to detect the expression of Ki67 in the tumor tissue according to the instructions of the Ki-67 detection kit (Immunohistochemistry, Sangon Biotech Co., Ltd., Shanghai, China).

The total RNA was extracted from subcutaneous tumor tissue of nude mice in the OE-AC093797.1 and control group and used to prepare a common transcriptome library; then, the RNA sequencing was performed on the Illumina NovaSeq 6000 platform. The “DNSeq2” package in R 4.0 was used to filter the differentially expressed genes between two groups. |log2foldchange| > 1.5 and adjust p-value (FDR) < 0.05 were set as the criterion. Then, gene ontology (GO), KEGG signaling pathway enrichment analysis, and gene set enrichment analysis (GSEA, including c2.cp.kegg v7.4.symbols.gmt and c2.cp.reactome.v7.4.symbols.gmt) were performed. Adjusted p-value (FDR) < 0.05 indicates that the gene or the signaling pathway was significantly enriched in the corresponding category. Furthermore, the protein-protein interaction (PPI) network was constructed by the online tools STRING (https://string-db.org/) and visualized by Cytoscape 3.8. Then, the function modules and hub genes were identified by the plunge-in Molecular Complex Detection (MCODE) and cytoHubba using the default parameters and MCC algorithm, respectively.

All data were expressed as mean ± standard deviation (SD), and the standard two-tailed t-test or one-way analysis of variance (ANOVA) was used to compare the difference between two groups. The correlation between the expression of AC093797.1 and clinical variables was assessed by chi-square test. Univariate and multivariate Cox regression models were used to screen the independent prognostic factor of HCC patients in the software SPSS 26. The subgroup analysis between the expression of AC093797.1 and clinical variables was performed via the package “forestplot” in R, and the survival analysis between the expression of AC093797.1 and survival time was performed via the package “survival” and “survminer” in R. P < 0.05 indicates the differences with statistical significance.

The GEPIA database integrates the data from the TCGA and the GTEx database, which recomputed the RNA sequencing raw data based on a standard processing pipeline to minimize differences from distinct sources to make data from different sources more compatible. The GEPIA database contains RNA expression profiles of 369 HCC tissues and 160 normal tissues. Based on the expression profiles of GEPIA, AC093797.1 was significantly low-expressed in HCC tissues than normal liver tissues (p < 0.05, Figure 1A). In this study, the expression of AC093797.1 was also confirmed in paired pathological tissues of HCC patients and liver cancer cell lines. As shown in Figures 1B,C, the expression of AC093797.1 in the 16 HCC tissues and four liver cancer cell lines (including HCCLM3, Huh7, HepG2, and MHCC97-H) was significantly decreased than adjacent non-tumor specimens and normal liver cells LO2. Further, the HCC patients with lower expression of AC093797.1 have poorer 5-year survival, but the expression level of the lncRNA was not related to the 5-year disease-free survival (DFS) (Figures 1D,E), indicating that the expression of AC093797.1 may be a biomarker of 5-year survival of HCC patients.

To explore the correlation between the expression of AC093709.1 and clinical indicators, 365 HCC patients with survival information in the TCGA database were divided into low- and high-expression groups according to the median of AC093797.1. Combined with the 5-year overall survival information, the BMI (>23.9 vs. ≤23.9) and living status of HCC patients were significantly different between the two groups (Table 1), suggesting that the expression level of AC093797.1 may be related to the BMI and 5-year OS of HCC patients. While combined with the 5-year DFS information, the BMI (>23.9 vs. ≤23.9) and hepatic inflammation of HCC patients was significantly different between the two groups (Table 1).

Further, univariate Cox regression and multivariate Cox regression analyses were performed to identify the risk factors that affect the prognosis of HCC patients. As shown in Figure 2A, the BMI ≤23.9, TNM stage Ⅲ-Ⅳ, with tumor, and the low expression of AC093797.1 were the risk factors of 5-year survival. The low expression of AC093797.1 was not acting as a risk factor of 5-year DFS, and the risk factors of 5-year DFS of HCC patients were the TNM stage Ⅲ-Ⅳ, with tumor and vascular invasion (Supplementary Figures S1A, B). However, the expression level of AC093797.1 was not an independent risk factor that affects the 5-year OS (Figure 2B), implying that the expression of AC093797.1 may interact with other clinical indicators to affect the 5-year survival of HCC patients.

The results of subgroup analysis showed that the expression of AC093797.1 was an independent risk factor affecting 5-year survival in patients with age ≤60, BMI ≤23.9, related family history, TNM stage III-IV, or grade III-IV (Figure 2C). But the expression of AC093797.1 was not an independent risk factor affecting 5-year DFS in any subgroup (Supplementary Figure S2C).

Besides, the expression of AC093797.1 combined with age, BMI, TNM stage, grade, and related family history can better predict the 5-year survival in patients with age ≤ 60, BMI ≤ 23.9, related family history, TNM stage III-IV, or grade III-IV (Figures 2D–M).

It was speculated that the expression of AC093797.1 may have a significant clinical correlation with clinical indicators in specific HCC patients.

To further clarify the biological role of AC093797.1 in HCC, the overexpression vector OE-AC093797.1 was transfected into HCCLM3 cells. The expression of AC093797.1 in the transfected HCCLM3 was detected by qPCR. As shown in Figure 3A, the expression of AC093797.1 was significantly increased in the cells transfected with OE-AC093797.1. Through CCK8, transwell, and wound healing assays, it was found that the overexpressed AC093797.1 in the HCCLM3 cells could inhibit cell proliferation (Figure 3B), migration (Figures 3C,D, G–H), and invasion (Figures 3E,F).

Via the nude mice tumor formation experiment, it was found that the volume of the tumor isolated from the OE-AC093797.1 group was significantly smaller than the control group on the 45th day after subcutaneous injection (Figures 4A–D), which suggested that high expression of AC093797.1 can inhibit the growth of tumors formed by HCCLM3 in vivo. Ki67 protein is a nuclear protein strictly associated with cell proliferation, and it is generally believed that the higher the expression level of Ki67, the faster the tumor growth, the lower the degree of differentiation, and the worse the prognosis of tumor patients. The immunohistochemical results of ki67 showed that the positive staining area in the tumor tissue of the OE-AC097997.1 group was significantly lower compared with the control group (Figures 4E–G), indicating that AC093797.1 significantly weakened the cell division of HCC cells in vivo.

In this study, subcutaneous tumor tissues of three nude mice in the OE-AC093797.1 and control group were used to extract the total RNA and performed RNA sequencing independently. Then, the differentially expressed genes of the OE-AC097997.1 group and control group were filtered by the package “DESeq2” in R. The result showed that 710 differentially expressed genes were identified, including 243 upregulated genes and 467 downregulated genes (Figures 5A,B, Supplementary Table S1).

To explore the involved biological processes and functions of these differentially expressed genes, the GO, KEGG signal pathway enrichment analyses, and GSEA were conducted. The GO analysis contained three terms, including biological process (BP), cellular component (CC), and Molecular Function (MF). For the BP, the differentially expressed genes were significantly enriched in the extracellular matrix organization, extracellular structure organization, etc.; for the CC, the differentially expressed genes were significantly enriched in the collagen-containing extracellular matrix, endoplasmic reticulum lumen, nucleosome, etc.; for the MF, the differentially expressed genes were significantly enriched in peptidase regulator activity, extracellular matrix structure constituent, etc. (Figure 5C). Via the KEGG enrichment analysis, the differentially expressed genes were significantly enriched in retinal metabolism, alcoholism, drug metabolism-cytochrome P450, steroid hormone biosynthesis, metabolism of xenobiotics by cytochrome P450, etc. (Figure 5B). The result of GSEA showed that the KEGG pathway gene set of the metabolism of glycolipid metabolism, tyrosine metabolism, starch and sucrose metabolism, arginine and proline metabolism, valine leucine and isoleucine degradation, cell adhesion molecules cams, pentose and glucuronate interaction, amyotrophic lateral sclerosis (Figure 6A), and the Reactome pathway gene set of the NLRP3 inflammasome, signaling to Ras, and signaling to ERKs, and other three pathway was significantly enriched in the control group (Figure 6B).

Furthermore, a PPI network was constructed in the STRING database and visualized by the Cytoscape. By the MCODE plug-in in Cytoscape, three function modules of the PPI network were parsed out (Figures 5E–G), and the top 20 core genes were extracted from the PPI network via the plug-in CytoHubba using the MCC algorithms (Figure 5H). Combining the above two algorithms, 13 histone family members HIST1H2BJ, HIST1H2BK, HIST1H2AC, HIST2H2AA3, HIST1H2AI, HIST1H2BG, HIST2H4B, HIST2H4A, HIST1H2BC, HIST1H1C, HIST1H3H, HIST2H3D, HIST2H2BF, and TTR, A2M, PTPRC, and FGG were selected as hub genes finally.