Identification and validation of key miRNAs and a microRNA-mRNA regulatory network associated with liver cancer

ABSTRACT

MiRNAs play crucial regulatory roles in the growth and development of tumor cells by serving as carriers of post-transcriptional regulatory information derived from genes. Investigating the potential function and clinical significance of miRNA-mediated mRNA regulatory networks in liver cancer can offer novel insights and therapeutic strategies for the treatment of this disease. We identified 300 differentially expressed miRNAs, and five miRNAs were identified to be correlated with overall survival and could be used as an independent prognostic. GO enrichment analysis mainly included carboxylic acid biosynthesis, organic acid biosynthesis, peroxisomal membrane, microsomal membrane, DNA binding, C-acyltransferase activity, etc. KEGG enrichment analysis showed that the pathways of target genes related to liver cancer were mainly focused on butyric acid metabolism and partial amino acid metabolism. Eight of the top 10 HUB genes were associated with prognosis, and the expression of four genes was positively correlated with prognosis, of which ABAT, BHMT, and SHMT1 were target genes of hsa-miR-5003-3p. MiR-5003-3p inhibits ABAT/BHMT/SHMT1 expression, thereby promoting liver cancer development. Overall, our study provides new ideas for the treatment of liver cancer, and these five miRNAs may be independent prognostic biomarkers and therapeutic targets for liver cancer patients. And miR-5003-3p may be a critical factor in the mechanism of liver cancer development.

KEYWORDS:

• Liver cancer
• miRNA
• mRNA
• prognostic biomarker
• miR-5003-3p

1. Introduction

Globally, liver cancer is a common cause of cancer-related death, and its major risk factors include viral hepatitis and excessive alcohol consumption [1–3]. Due to the high degree of malignancy and poor prognosis of liver cancer, there are mostly no specific clinical symptoms in the early stages. Most of them are already in the middle or late stages when detected, at which time the effectiveness of treatment options is significantly reduced. The survival and prognosis of the patients become the focus of attention [4,5]. So far, the molecular mechanism of liver cancer is still unclear, and the incidence rate continues to rise [6,7], which requires researchers to continuously explore and find biomarkers for early diagnosis and prognosis with higher accuracy.

MicroRNAs (miRNAs) are short non-coding RNAs of approximately 22 nucleotides in length [8], which serve as carriers of post-transcriptional regulatory information encoded by genes and have important regulatory roles in the growth and development of numerous cancers [9–11]. For example, miR-155 is lowly expressed in liver cancer cells and inhibits the proliferation and metastasis of cancer cells [12]; MIR-373 inhibits autophagy and promotes apoptosis of cholangiocarcinoma cells by targeting ULK1 [13]; and miR-4733-5p is up-regulated in gallbladder cancer and promotes the proliferation of gallbladder cancer cells by directly binding to KLF [14]. This shows that miRNAs may be key factors in regulating oncogenes, which provides new ideas for us to study the pathogenesis of cancer.

In this study, we downloaded relevant data about liver cancer through the TCGA database, merged the obtained miRNAs, mRNAs, and clinical information, constructed models, screened key molecules in the miRNAs-mRNAs network, and provided new insights into future targeted therapies.

2. Materials and methods

2.1. Acquisition and processing of data

Liver cancer patients operated in the First Affiliated Hospital and the Second Affiliated Hospital of Bengbu Medical College from February 1, 2022 to February 1, 2023 were collected, all of them did not have any other treatments before the operation, and ten cases were randomly selected to extract the cancerous and para carcinomatous tissues for the subsequent experiments. Notify all patients and obtain their consent before surgery, and obtain approval from the Ethics Committee of Bengbu Medical College (Ethics Approval Letter (2023) No. 111).

MRNAs expression information: Case (371), Primary site (liver cancer, intrahepatic cholangiocarcinoma), Procedure (TCGA), Item (TCGA-LIHC), Disease Type (adenoma and adenocarcinoma); Files(424): Data category (Transcriptome Atlas), Data type (Quantitative gene expression)], Workflow Type (STAR-Counts). Among them, 50 are normal samples, and 375 are tumor samples.

MiRNAs Expression Information: Case (373), Primary Site (Liver cancer, Intrahepatic cholangiocarcinoma), Program (TCGA), Item (TCGA-LIHC), Disease Type (Adenomas and Adenocarcinomas), Files (425): Data Category (transcriptome profiling), Data Type (Isoform Expression Quantification), of which 50 were normal samples and 374 were tumor samples.

Clinical Information: Case (377), Primary Site (Liver cancer, Intrahepatic cholangiocarcinoma), Program (TCGA), Item (TCGA-LIHC), Disease Type (Adenomas and Adenocarcinomas); Files (377) Data Category: Clinical, Data Format (BCR XML), retrieved on February 14, 2023, from TCGA (https://cancergenome.nih.gov/) official website. The data were integrated through the Perl language.

2.2. Analysis of differentially expressed miRNAs and mRNAs and their integration with patient survival data

R Language version 4.2.1 was used to analyze and compare the expression of miRNAs and mRNAs in tumor samples and normal samples [15]. The expression profiles of miRNAs and mRNAs were analyzed, with the absolute value of log2FC set to be greater than 1 and fdr less than 0.05 as the filtering conditions. The top ten up-regulated, down-regulated differentially regulated genes were displayed in the heatmap.

2.3. Sample grouping and prediction model construction, validation and evaluation

Merge 377 clinical information samples with differentially expressed miRNAs using Perl for subsequent model construction. Randomly divide differentially expressed miRNAs with survival information into two groups using R Language version 4.2.1 (one group serves as the training group for constructing prognostic models, and the other group serves as the validation group for verifying model accuracy)., and conduct univariate Cox regression analysis on the training group [16]. Stepwise multifactorial Cox regression analysis was performed for miRNAs with p-value <0.05 to establish a prognostic model. Based on the product of each miRNA and the sum of its coefficients, risk scores of predicted miRNAs consisting of multiple miRNAs were established. Kaplan-Meier curves were used to assess the survival prognosis of patients in the train group, test group, and the all group; meanwhile, ROC curves were plotted to assess the accuracy of the risk-survival model [17,18].

2.4. MiRNA-independent prognostic analysis

By constructing a model and scoring three groups (train group) test group (all group), obtain the risk values for the all groups. Organize clinical variable data (age, gender, clinical stage, clinical grading, lymph nodes, and distant metastasis) and merge it with the risk values of the all group through Perl. Finally, the relationship between each clinical variable and survival time and survival status was analyzed through single factor and multivariate independent prognostic analysis using the survival package.

2.5. Target gene prediction of miRNAs and their potential functions

We downloaded data from three (mirtarbase, targetscan, miRDB) target gene prediction websites for miRNAs [19,20] and used the Perl language to find the target genes of miRNAs, setting the condition that at least 2 databases contain the target gene. Draw Venn diagram and Cytoscape 3.9.1 Draw miRNA and its target gene network diagram [21,22]. The intersection of the target genes of these miRNAs and the differentially expressed genes of liver cancer was further analyzed by obtaining the intersection of the target genes of these miRNAs and the differentially expressed genes of liver cancer to gain insight into whether the target genes of these miRNAs might be involved in the progression of liver cancer. Finally, set the filtering conditions pvalue < 0.05 and qvalue < 1 to obtain cross genes and perform the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analysis [23,24].

2.6. Screening of hub genes and survival-related genes

Construct a PPI network using a string database (https://string-db.org/) to reveal the relationships between target genes, and set the median confidence of the parameters to 0.400 to obtain the network relationship file [25]. Import the network relationship file into the cytoHubba plugin in Cytoscape 3.9.1 to identify HUB genes. Finally, the crossover genes were subjected to survival analysis using the Kaplan-Meier assay.

2.7. Western blotting

Human liver cancer cells (SMMC-7721) were purchased from MeisenCTCC (Zhejiang, China). Cells were grown in 1640 medium containing 10% fetal bovine serum. Protein lysis and extraction were performed on the cells after they reached 70 ~ 80% confluence. Protein quantification was carried out using the BCA kit (Biyuntian; Jiangsu, China), and the protein solution was added to 5* reduced protein uploading buffer at a ratio of 4:1 and denatured in a boiling water bath for 15 min. Electrophoresis was carried out on a PAGE gel and then transferred to a PVDF membrane by electroblotting. The PVDF membrane was closed with milk and then subjected to primary antibody incubation overnight at 4°C, followed by incubation with the corresponding secondary antibody for 2 h. Finally, ECL A and B were mixed in a 1:1 ratio and exposed to a gel imaging system, and the gray values of the bands were measured with Image J.

2.8. Lentiviral transfection

Human liver cancer cell lines SMMC-7721 and HuH7 were purchased from MeisenCCTCC (Zhejiang, China) and Chinese Academy of Sciences (Shanghai, China), respectively. Cells grown in 1640 or DMEM medium containing 10% fetal bovine serum. Inoculation: Prepare cell suspension (density 3–5 × 10⁴/ml), mix 1 ml of cell suspension and 1 ml of culture medium in a 6-well plate. Incubate in the incubator for 16–24 hours. Infection: Add an appropriate amount of virus and infection enhancing solution based on the cell’s MOI and viral titer (Purchased from Jikai Gene Co., Ltd., Shanghai, China). Continue cultivation for 12–16 hours, then replace the culture medium and continue cultivation (the medium can be changed in advance according to the cell morphology). Screening and experimentation: Add an appropriate amount of puromycin during the cultivation process to screen and observe stable transfected cell lines under a fluorescence microscope (miRNA-5003-3p knockdown group: SMMC-7721-KD) HuH-7-KD; miRNA-5003-3p empty transfection group: SMMC-7721-NC HuH-7-NC). The calculation formula used is based on virus volume= (MOI x cell count)/virus titer. Finally, the transfection efficiency was verified by RT-PCR.

2.9. Total RNA extraction from tissues or cells

Tissue: Cut and grind the tissue into powder, transfer it to a centrifuge tube, and add an appropriate amount of Trizol.Cells: Discard the cell culture medium, wash twice with PBS, then add Trizol and repeatedly blow and collect into a centrifuge tube.Add 200ul of chloroform to every 1.5 ml of enzyme free EP tube, let it stand for 10 minutes, and then centrifuge. Take the top layer (about 500 ul) and add isopropanol to precipitate RNA.Centrifuge after 10 minutes and discard the supernatant. Wash with 1 ml of 75% ethanol, centrifuge and air dry, dissolve in DEPC water, and measure concentration and purity.

2.10. Real time-PCR

Total tissue or cellular RNA was extracted by Trizol, reverse transcribed into cDNA, and then subjected to real-time fluorescence quantitative PCR by SYBR Green method under the reaction conditions: denaturation at 95°C for 40 s and annealing and extension at 60°C for 30 s, with a total of 40 cycles. The Stem loop synthetic miRNA-5003-3p primer sequence was: (5 ~ 3’): Reverse transcription primer: CTCAACTGGTGTCGTGGAGTCGG CAATTCAGTTGAGCCCCAACA, Forward primer: ACACTCCAGC TGGGTACTTTTCTAGGTTG, Universal primer – Reverse: TGGTG TCGTGGAGTCG. U6 (5 ~ 3’): Reverse primer: AACGCTTCACGAA TTTGCGT, Forward primer: CTCGCTTCGGCAGCACA.This experiment is used to detect the expression levels of miRNA-5003-3p in clinical specimens of cancer and paracancerous tissues, as well as to validate the efficiency of downregulating miRNA-5003-3p expression by lentivirus.

2.11. Immunohistochemistry experiment

Paraffin sections of liver cancer were deparaffinized and hydrated, and the sections were subjected to antigen repair. After 30 minutes of serum sealing, add the primary antibody and incubate the sections at 4°C in a humidor overnight. After washing, the sections were incubated with the corresponding secondary antibody for 50 min at room temperature. After washing, the color was developed with DAB color solution, and the color development time was controlled under the microscope, and the positive color was brownish yellow. After rinsing, hematoxylin was used to re-stain the nuclei. Finally, the slices were dehydrated and sealed with sealing gel, and the results were read under a white light microscope.

2.12. Colony formation experiment

Inoculate 500 cells per well in a 6-well plate, transfer to the incubator and incubate at 37°C, 5% CO₂ and saturated humidity for 2–3 weeks. Replace the culture once a week, and terminate the culture when clones visible to the naked eye appear in the culture dish. Discard the culture medium, carefully dip and wash with PBS for 2 times, add fixative (4% paraformaldehyde) to fix for 15 min, discard the fixative, wash with PBS for 2 times and then air-dry, air-dry and stain with crystal violet staining solution for 10 min, slow washing and air drying.

2.13. Transwell migration and invasion

For conventional trypsin digestion of cells, PBS washed 1–2 times to remove the effect of serum, resuspend the cells with serum-free medium, inoculate the suspended cells into the upper chamber of the Transwell nested 6-well plate chambers, divide the upper chamber into those containing and those not containing stromal gel, and in the lower chamber add the medium containing 10% FBS, incubate for 24 hours, remove the chambers. PBS drenching 2 times, wiping off the cells within the upper layer of the microporous membrane of the chambers, 4% paraformaldehyde fixation for 20 min, crystal violet solution staining for 15 min and then photographed.

2.14. Wound healing experiment

Cells were inoculated into 6-well plates, and the inoculation principle was that the fusion rate reached 100% after overnight culture. Scratch vertically with the tip of a gun, wash the cells with PBS 2–3 times, add serum-free medium, and take pictures at 0 and 24 hours, respectively.

2.15. Data analysis

The data has been presented as the mean±SEM from three independent experiments. All the statistical analyses were conducted using GraphPad Prism. If the P-value was less than 0.05, the result was considered as statistically significant.

3. Results

3.1. Identification of differentially expressed miRNAs and mRNAs

Differential genes were identified using the TCGA dataset, employing filtering conditions of a logFC absolute value greater than 1 and an FDR less than 0.05. This resulted in the identification of 882 differentially expressed mRNAs) Supplementary Figure S1), with 487 being up-regulated and 395 being down-regulated (Supplementary File 1 mRNAs). Additionally, 300 differentially expressed miRNAs were identified) Supplementary Figure S2), with 260 being up-regulated and 40 being down-regulated in expression (Supplementary File 2 miRNAs). The heatmaps in Figure 1(a,c)show the top 10 significantly differentially expressed mRNAs and miRNAs, respectively. Meanwhile, the volcano plot in Figure 1(b) displays all mRNAs, while the volcano plot in Figure 1(d) displays all miRNAs, where green represents downregulation, red represents upregulation, and black represents no difference.

Figure 1. Heatmap of the top 10 up- and down-regulated significantly differentially expressed mRNAs and miRNAs ((a) mRNAs, (c) miRNAs); the volcano plot shows all miRNAs and mRNAs, with green indicating downregulation, red indicating upregulation, and black indicating no difference)(b) mRNAs, (d) miRNAs).

3.2. Establishment and evaluation of prognostic modeling of miRNAs

The 300 differentially expressed miRNAs in liver cancer were randomly and equally divided into a train group (Supplementary File 3 Train) and a test group (Supplementary File 4 Test) using Caret package. Univariate Cox regression analysis showed that 10 miRNAs from the train group were associated with the overall survival of patients. Further multifactorial regression analysis was performed to screen 5 out of 10 candidate miRNAs (hsa-miR-103a-3p, hsa-miR-139-5p, hsa-miR-101-3p, hsa-miR-188-5p, hsa-miR-5003-3p) to establish a prognostic model according to the formula RISK SCORE=(−0.3601 × expression of hsa-miR-103a-3p) + (−0.2111 × expression of hsa-miR-139-5p) + (−0.2945 × expression of hsa-miR-101-3p) + (0.3485 × expression of hsa-miR-188-5p) + (0.2903 × expression of hsa-miR-5003-3p) to calculate risk values for the train and test groups, and divided into high and low groups according to the median risk value of the train group. The Kaplan Meier curve shows that the all group (p = 8.087e-10), test group (p = 4.935e-04), and train group (p = 2.493e-07), with p-values less than 0.05 (Figure 2(a-c)), suggesting that there is a difference in risky survival over time between the high- and low-expression groups, with the high-risk group having a significantly lower overall survival rate than the low-risk group.

Figure 2. Kaplan-Meier curves of survival risk and ROC curves of corresponding subgroups for the three groups: survival risk subgroups (a: all group, b: test group, c: train group); ROC curves subgroups (d: all group, e: test group, f: train group).

Next, we assessed the accuracy of the model by plotting the ROC curve, which showed that the area under the curve was 0.754, 0.773, 0.734 for the all, test and train groups, respectively (Figure 2(d-f)), which indicated that the model had a very high accuracy in predicting the survival risk of primary liver cancer patients.

Finally, the survival curves of these five miRNAs were plotted by Kaplan-Meier analysis and log-rank test, and the results showed that: the expression of hsa-miR-139-5p and hsa-miR-101-3p was positively correlated with the overall survival of the patients, and the expression of hsa-miR-188-5p and hsa-miR-5003-3p were overall survival were negatively correlated (Figure 3). Among them, hsa-miR-103a-3p expression was up-regulated in liver cancer. Still, the curve showed that high expression was positively correlated with overall survival, which was contradictory. Hence, the survival curves were no longer plotted, but this did not affect the construction of the model and the assessment of model accuracy.

Figure 3. Four miRNas associated with overall survival in liver cancer patients: (a) hsa-miR-101-3p.(b) hsa-miR-139-5p, (c) hsa-miR-188-5p, (d) hsa-miR-5003-3p.

3.3. Survival status and independent prognostic analysis

Using R Language version 4.2.1, an analysis was conducted on the risk values and survival status of the all group, test group, and train group. The findings were visually presented in graphs (Figure 4(a-c)), revealing a positive correlation between risk values and patient mortality rate. Ultimately, unifactorial and multifactorial independent prognostic analyses indicated that clinical stage and T stage hold potential as indicators for determining the prognosis of liver cancer (Table 1).

Figure 4. Survival status map: a: all group, b: test group, c: train group. Venn diagrams of 5 miRNAs target genes (d: hsa-miR-101-3p; e: hsa-miR-139-5p; f: hsa-miR-188-5p, g: hsa-miR-103a-3p, h: hsa-miR- 5003-3p).

Table 1. Univariate and multivariate independent prognostic analysis of clinical characteristics.

uniCox					multiCox
id	HR	HR.95 L	HR.95 H	pvalue	id	HR	HR.95 L	HR.95 H	pvalue
age	0.997	0.979	1.015	0.741	age	1.003	0.983	1.023	0.728
gender	0.767	0.468	1.258	0.294	gender	0.944	0.545	1.635	0.837
grade	0.994	0.720	1.371	0.971	grade	1.284	0.900	1.832	0.167
stage	2.078	1.597	2.704	<0.001	stage	0.795	0.286	2.205	0.658
T	1.991	1.562	2.538	<0.001	T	2.237	0.911	5.491	0.078
M	4.439	1.387	14.207	0.012	M	1.682	0.429	6.587	0.455
N	2.334	0.569	9.577	0.239	N	3.5853	0.534	24.068	0.1882
riskScore	1.409	1.237	1.606	<0.001	riskScore	1.349	1.155	1.574	<0.001

3.4. Target gene prediction of miRNAs and PPI network HUB gene screening

The target genes regulated by the five miRNAs (hsa-miR-101-3p, hsa-miR-139-5p, hsa-miR-188-5p, hsa-miR-103a-3p, and hsa-miR-5003-3p) were predicted using three miRNAs databases. Additionally, an analysis was conducted to identify the overlapping target genes among the three databases. The results indicated that the number of overlapping genes for hsa-miR-101-3p, hsa-miR-139-5p, hsa-miR-188-5p, hsa-miR-103a-3p, and hsa-miR-5003-3p were 166, 31, 15, 123, and 42 genes, respectively (Figure 4(d-h)). In order to investigate the potential involvement of the target genes of these miRNAs in the progression of liver cancer, the set of 882 mRNAs (487 up-regulated and 395 down-regulated) obtained previously was combined with the five miRNAs (down-regulated: hsa-miR-139-5p, hsa-miR-101-3p; up-regulated: hsa-miR-103a-3p, hsa-miR-188-5p, hsa-miR-5003-3p) for joint analysis. The resulting interaction network between these five miRNAs and their 34 target genes is depicted in Figure 5(a). Subsequently, we conducted additional screening on 32 genes out of the initial 34 target genes in the PPI network to reveal the relationships between target genes. Finally, we obtained the core 10 HUB genes (TOP2A, KPNA2, SMARCA4, ETS2, HMGB2, SHMT1, PSPH, BHMT, ABAT, SLC38A4) (Figure 5(b)).

Figure 5. a: MiRNAs-mRNAs regulatory network. Triangles represent miRNAs. Ovals represent mRNAs. The red table is up-regulated, green table is down-regulated, b: HUB genes of the PPI network. Blue, yellow, light orange, and deep orange indicate a gradual increase in correlation.

3.5. GO and KEGG enrichment analysis of related target genes

Furthermore, we performed Gene Ontology (GO) analysis on the target genes associated with liver cancer, wherein the top 5 GO outcomes were presented in biological process (BP), cellular component (CC), and molecular function (MF) plots (Figure 6(a-c)). Among them, BP analysis mainly included the carboxylic acid biosynthetic process, organic acid biosynthetic process and cellular amino acid biosynthetic process, and CC analysis mainly included peroxisomal membrane, microbody membrane, and MF analysis mainly included DNA binding, C-acyltransferase activity, and fatty acid ligase activity. KEGG enrichment analysis showed that the pathways of target genes associated with KEGG enrichment analysis showed that the pathways of target genes related to liver cancer mainly focused on Butanoate metabolism and Glycine, serine and threonine metabolism (Figure 6(d)).

Figure 6. Functional enrichment analysis of related target genes: (a) BP, (b) CC, (c) MF; (d) KEGG.

3.6. MiRNAs target gene survival correlation analysis

The expression levels of 32 genes were assessed using Kaplan-Meier curves and a log-rank test. The results indicated that the expression of 19 genes was significantly associated with prognosis (p < 0.05). Among these genes, 8 were identified as HUB genes (ABAT, BHMT, ETS2, HMGB2, KPNA2, PSPH, SHMT1, TOP2A) (Figure 7). Notably, the expression of four HUB genes (ETS2, ABAT, BHMT, SHMT1) exhibited a positive correlation with prognosis. Furthermore, ABAT, BHMT, and SHMT1 were identified as target genes of hsa-miR-5003-3p, suggesting the crucial regulatory role of hsa-miR-5003-3p in liver cancer.

Figure 7. Survival curves of 8 HUB genes (ABAT, BHMT, ETS2, HMGB2, KPNA2, PSPH, SHMT1, TOP2A).

3.7. MiR-5003-3p is highly expressed in liver cancer and inhibits the expression of ABAT/BHMT/SHMT1

To further substantiate the expression of miRNA-5003-3p in liver cancer, we conducted qRT-PCR analysis on cancerous and paracancerous tissues from 10 cohorts of liver cancer patients. Our findings revealed that miRNA-5003-3p exhibited a significantly higher expression level in liver cancer tissues compared to paracancerous tissues (Figure 8(a)). Additionally, utilizing the immunohistochemistry database, The Human Protein Atlas, we observed a diminished staining intensity of ABAT/BHMT/SHMT1 in liver cancer tissues compared to normal liver tissues. Moreover, the expression of ABAT/BHMT/SHMT1 demonstrated a positive correlation with patient survival. Subsequently, a random selection of paraffin specimens was obtained from 10 liver cancer patients for the purpose of conducting immunohistochemistry. The findings revealed a diminished staining intensity of ABAT/BHMT/SHMT1 in the liver cancer patients compared to the paraneoplastic individuals (Figure 8(b)).

Figure 8. a: The expression of miR-5003-3p was higher in cancer tissues than in paracancerous tissues in ten groups of liver cancer patients. b: Immunohistochemistry of ABAT/BHMT/SHMT1 in ten liver cancer patients showed lower staining intensity. c: The expression of miRNA-5003-3p was significantly reduced in SMMC-7721 and HuH7 liver cancer cell line after lentiviral transfection. d. Western blot results showed that the expression content of ABAT/BHMT/SHMT1 proteins in miRNA-5003-3p-KD cells was higher than that in miRNA-5003-3p-NC group and Control group. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Finally, miRNA-5003-3p low expression liver cancer cell lines (SMMC-7721-KD, Huh7-KD) were established by lentiviral transfection and the transfection efficiency was determined (Figure 8(c)). It was observed that the expression of ABAT/BHMT/SHMT1 was significantly increased after miRNA-5003-3p downregulation in liver cancer cell line (SMMC-7721-KD), as shown by Western blot analysis (Figure 8(d)). This finding suggests that the expression of ABAT/BHMT/SHMT1 is regulated by miRNA-5003-3p.

3.8. Down-regulation of miR-5003-3p inhibits liver cancer cell proliferation, migration and invasion

To further investigate the impact of miRNA-5003-3p, we assessed its effects on the migratory and invasive capabilities of SMMC-7721 and HuH7 liver cancer cells using wound healing and transwell migration and invasion assays. The findings demonstrated that the miRNA-5003-3p-KD group exhibited a reduced number of migrating and invading cells in comparison to both the miRNA-5003-3p-NC group and the Control group (Figure 9(a)). In contrast, the miRNA-5003-3p-KD group displayed a decreased rate of scratch healing after 24 hours when compared to both the miRNA-5003-3p-NC group and the Control group (Figure 9(b)). Furthermore, the results from the colony formation experiments indicated that the miRNA-5003-3p-KD group exhibited a significantly smaller number of colony cell clusters compared to both the miRNA-5003-3p-NC group and the Control group (Figure 9(c)). All observed differences were found to be statistically significant at a significance level of p < 0.05. The aforementioned findings suggest that the suppression of miRNA-5003-3p effectively hinders the proliferation, migration, and invasion capabilities of SMMC-7721 and HuH7 liver cancer cells.

Figure 9. a: Transwell assay showed that the number of migrating and invading cells through the compartment in miRNA-5003-3p-KD group was smaller than that in miRNA-5003-3p-NC and control groups. b: Wound healing assay showed that the wound healing rate in miRNA-5003-3p-KD group was lower than that in miRNA-5003-3p-NC and control groups. c: The cell colony experiment showed that the number of colonies in the miRNA-5003-3p-KD group was smaller than that in the miRNA-5003-3p-NC and control groups (**P < 0.01, ***P < 0.001, ****P < 0.0001).

4. Discussion

The pioneering work of Calin Dan Dumitru et al. in 2002 provided the initial compelling evidence of miRNA involvement in human diseases, specifically highlighting the crucial regulatory functions of miRNA-15 and miRNA-16 in B cell chronic lymphocytic leukemia (B-CLL) [26]. With the deepening of research, we are getting clearer and clearer about the network regulation of miRNA-mRNA interactions. A large number of studies have shown that miRNAs have important roles in the development of human diseases, especially cancer. For example, overexpression of miRNA148a inhibited the expression of WNT1 protein in non-small cell lung cancer [27]; MIR-802 targeted MYLIP to inhibit the growth of cervical cancer cells [28]; and lncRNA SNHG3 regulated osteosarcoma invasion and migration through miRNA-151a-3p/RAB22A axis [29]. However, miRNAs are involved in far more cancers than these, including pancreatic cancer [30], retinoblastoma [31], melanoma [32], nasopharyngeal carcinoma [33], prostate cancer [34], bladder cancer [35], and breast cancer [36].

Of course, the investigation of miRNAs mechanisms through experimental studies represents only a portion of the overall understanding. In recent years, computational modeling has emerged as a significant approach for examining the association between miRNAs and diseases [37–39]. In this particular study, we developed a novel miRNA prognostic prediction model by retrieving differentially expressed miRNAs from the TCGA database. The results of the model’s ROC curves demonstrated AUC values of 0.734, 0.773, and 0.754 for the train group, the test group, and the all group, respectively. This finding suggests that the model exhibits enhanced predictive efficacy, thereby enhancing our comprehension of the etiology and progression of primary liver cancer. Additionally, it offers novel perspectives for prospective targeted therapies in the field of primary liver cancer.

We performed functional enrichment analysis on the intersection of target genes. We differentially expressed genes of miRNAs, and the GO results of target genes were mainly focused on the carboxylic acid biosynthetic process, organic acid biosynthetic process, cellular amino acid biosynthetic process, peroxisomal membrane, microbody membrane, C-acyltransferase activity, etc. Among them, amino acid metabolism is a crucial biosynthetic process in the human body. Amino acid utilization is the basis of cellular activity and key to intracellular metabolic reactions and signal transduction [40]. Amino acid metabolism also plays a key role in targeted cancer therapy [41,42].

To further investigate the regulatory function of miRNAs in liver cancer, we conducted a screening of 10 HUB genes, with particular emphasis on three target genes (ABAT, BHMT, SHMT1) that exhibited a positive correlation with overall survival. Some studies have shown that ABAT, BHMT, and SHMT1 are lowly expressed in liver cancer, and the lower their expression, the worse the prognosis of liver cancer patients. And overexpression of ABAT and SHMT1 had an inhibitory effect on liver cancer [43–45]. Meanwhile, we analyzed and found that these three genes are target genes of hsa-miR-5003-3p, which suggests that hsa-miR-5003-3p may be a key node in the mechanism of liver cancer development. A comprehensive examination of hsa-miR-5003-3p holds significant value in elucidating the molecular mechanisms underlying liver cancer development.

Enzymes are a very important class of biocatalysts, which have high specificity and catalytic efficiency for the corresponding substrates. The life process of every organism is related to enzymes, the most important of which is the regulation of energy metabolism [46,47]. As important enzymes of amino acid metabolism, abat, BHMT and shmt1 have important regulatory roles in energy metabolism. At the same time, as the main site of amino acid metabolism, the liver must be affected by ABAT, BHMT and SHMT1. At present, studies have clarified their inhibitory effect on liver cancer, but the specific mechanism is still unclear. It is noteworthy that we found that amino acid metabolism may be an important factor in hepatocarcinogenesis through bioinformatics analysis of TCGA database. This provides a new idea for studying the pathogenesis of liver cancer. However, this is only an exploratory study, which needs a lot of experimental verification.

Author contributions statement

Jie Tang, Song Li and Shaobo Zhou provided the experimental design. Jie Tang, Song Li and Zixiao Zhou performed the experiments. Weicai Chang and Yongqiang Wang analyzed the data. Yongqiang Wang and Juan Mei prepared all figures. Jie Tang, Song Li and Shaobo Zhou wrote the draft of the manuscript. All authors read and approved the final manuscript.

Ethical Approval

All procedures in this study were conducted in accordance with the Ethics Committee of the Second Affiliated Hospital of Bengbu Medical College(Ethics Approval Letter (2023) No. 111).

Provenance and peer review

Not commissioned, externally peer-reviewed.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data used to support the findings of this study are available from the corresponding author upon request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15384101.2024.2335024.

Additional information

Funding

This work was supported by the Scientific research project of the Anhui Provincial Health Commission [AHWJ2021a012], Anhui University Natural Science Research Project [KJ2020A0564].