Introduction
Hepatocellular carcinoma (HCC) is one of the most common malignant tumors and is the third leading cause of cancer-related deaths worldwide [
1]. The 5 year survival rate is less than 12%. Most HCC cases are at advanced stages when diagnosed [
2], thereby leading to poor prognosis and posing challenge to treatment. Traditionally, clinical staging and vascular tumor invasion are essential contributors to clinical outcomes and may help to predict survival [
3]. However, these clinicopathological risk factors are limited in terms of prognostic evaluation and are insufficient to distinguish between high-risk and low-risk patients. Sensitivity to adjuvant chemotherapy is even more unpredictable. Thus, there is an urgent need to explore novel prognosis-related genes, building a comprehensive model to predict clinical outcomes.
Epithelial-mesenchymal transition (EMT) is a favorable feature of malignant cells [
4]. Some cells lose their epithelial characteristics and obtain a mesenchymal phenotype during the transition, eventually leading to a loss of intercellular junctions [
4]. Thus, EMT not only promotes invasion and metastasis, but also leads to enhanced stemness of tumor cells, contributing to the development of chemoresistance [
5], immunosuppression [
6] and targeted therapy resistance [
7]. Therefore, developing new therapeutic strategies to control EMT is essential in oncogenesis, metastasis, and treatment. Unfortunately, reversing EMT in tumor cells has not yet been achieved [
8].
Metabolic reprogramming is generally recognized to be a new hallmark of cancer [
4], most notably the “Warburg effect.” In addition to dysregulated glucose metabolism, metabolic reprogramming in tumor cells is characterized by abnormal nucleotide metabolism, amino acid metabolism, mitochondrial biosynthesis, and the rest of pathways [
9]. The study of these metabolic reprogramming will shed light on the molecular events of malignancy and facilitate to identify preferable approaches for diagnosis and treatment. Recent findings suggested that metabolic demand is altered in EMT-activated cells to meet increased motility and aggressiveness [
10]. In some cases, metabolic reprogramming can also drive EMT, and the link between the two is reciprocal. In certain cancer types, tumors undergoing metabolic reprogramming are correlated with worse survival [
11]. Metabolic reprogramming of cancer cells has tremendous impact on immune microenvironment [
12], thereby influencing the efficacy of immunotherapy. Therefore, understanding the mechanisms of metabolic reprogramming in different EMT states is crucial for improving patient survival.
In this study, The Cancer Genome Atlas (TCGA)-LICH was used to analyze differential metabolic pathways according to different EMT status groups. Based on the unsupervised cluster analysis of the differential metabolic pathway genes, three clusters with significant differences in survival were obtained. Using differential expression analysis and LASSO-Cox regression, 11 genes were selected to establish a prognostic risk model. This prognostic model may help to optimize risk stratification and identify appropriate therapeutic strategies for HCC patients. Moreover, the correlation between Stard5 and EMT has been broadly verified in vitro and in patients, which provides a target for exploring the interaction between EMT and metabolic reprogramming.
Materials and methods
Data acquisition
The mRNA expression profiles of HCC patients and the corresponding clinical profiles, including age, gender, grade, stage, alcohol consumption, Hepatitis B, Hepatitis C and survival time, were downloaded from the TCGA-LIHC database (
https://gdc.nci.nih.gov/) and were detailed in Table.
1. Validation dataset GSE14520 was downloaded from Gene Expression Omnibus (GEO) database, and survival information for the samples was shown in Table.
2.
Table 1
The clinical characteristics of TCGA-LIHC samples
Age |
> 65 | 138 |
< = 65 | 229 |
Negative | 40 |
Gender |
MALE | 248 |
FEMALE | 119 |
Grade |
G1 | 55 |
G2 | 176 |
G3 | 119 |
G4 | 12 |
NA | 5 |
Stage |
Stage I | 171 |
Stage II | 85 |
Stage III | 83 |
Stage IV | 4 |
[Discrepancy] | 2 |
NA | 22 |
Alcohol consumption |
Yes | 115 |
No | 252 |
Hepatitis B |
Yes | 103 |
No | 264 |
Hepatitis C |
Yes | 56 |
No | 311 |
EMT group |
EMT-H | 19 |
EMT-L | 348 |
Cluster |
Cluster1 | 251 |
Cluster2 | 104 |
Cluster3 | 12 |
OS |
Alive | 237 |
Dead | 130 |
Table 2
Survival information for GSE14520 data set
Calculation of EMT enrichment scores
A total of 145 epithelium (EPI) genes and 170 mesenchyme (MES) genes were obtained from PMID: 25214461 [
13]. Based on the above gene sets, the samples’ EPI enrichment score and MES enrichment score were calculated by the R package GSVA (v1.34.0), and the EMT enrichment score was subtracted from the two. Surv_cutpoint of R package survminer (v0.4.8) was used to find the most appropriate node to differentiate EMT-H from EMT-L groups. Survival analysis of the two groups was performed by the R package surv (v3.2–7). Differences in clinical characteristics between the EMT subgroups were detected using Kruskal wallis test.
Enrichment scores for metabolic pathways were calculated for samples based on metabolic pathway-related genes provided by PMID: 33917859 [
14]. The differences of metabolic enrichment scores between EMT groups were analyzed using Kruskal wallis test.
Subtypes identification based on metabolic reprogramming
A subtype analysis related to metabolic reprogramming of HCC was performed based on genes of metabolic pathways, which had significant differences between the EMT-H and EMT-L groups. Unsupervised cluster analysis was applied to all samples in the TCGA-LIHC dataset through the R package ConsensusClusterPlus (v1.50.0) with the algorithm K-means. The clusters were then analyzed for survival using R package survival and survminer.
Functional enrichment analysis
Differentially expressed genes (DEGs) of the clusters were acquired by the R package limma [
15]. A threshold of |
\({\mathrm{log}}_{2}foldchange\)|> 1 and adjusted
p < 0.05, were considered for DEGs. Overlapping DEGs from the three clusters were used for subsequent analysis. The Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis and Gene Ontology term (GO) analysis, which consists of biological processes (BP), cellular component (CC), and molecular function (MF), were performed using DEGs shared by the three clusters [
16].
Construction of a prognostic model
Univariate Cox regression analysis was applied to the significant DEGs, using p < 0.01 as the threshold, in combination with the overall survival data. DEGs were then further filtered by LASSO-Cox regression analysis, and risk score models were constructed, a process that resorted to the R package glmnet (v4.0–2). Lambda screening was used for cross-validation. The model corresponding to lambda.min was used to collect the gene expression matrix. The risk score for each sample was calculated using the following equation: \({\mathbf{R}\mathbf{S}\mathbf{c}\mathbf{o}\mathbf{r}\mathbf{e}}_{{\varvec{i}}}= \sum_{{\varvec{j}}=1}^{{\varvec{n}}}{\mathbf{exp}}_{{\varvec{ji}}}\times {{\varvec{\upbeta}}}_{{\varvec{j}}}\). The median risk score was used to classify high-and low-risk groups. A p-value of Kaplan–Meier survival analysis < 0.05 was considered to indicate a significant difference between the two groups. The area under the curve (AUC) values for the model were calculated using the survival data and demonstrated by time-dependent receiver operating characteristic (ROC) curves, with AUC values greater than 0.6 indicating good predictive power of the prediction model.
Immune cell infiltration and chemotherapy resistance prediction analysis
To explore the response of patients to Erlotinib, Shikonin, Metformin, Bortezomib, Metformin, and Lapatinib, the predictive value of IC50 was obtained using the R package pRRophetic (v 0.5) analysis. The difference in IC50 between the high-and low-risk groups was tested using the Wilcoxon test. R package CIBERSORT (v1.03) was used to analyze the proportion of immune cells in all patients.
Clinical HCC patient samples and cell culture
To validate the expression levels of genes in the prognostic model, we collected 80 tumors and adjacent normal tissues from patients with HCC. All patients participating in this study signed an informed consent form. This project was approved by the Human Research Ethics Committee of Zhongshan Hospital, Fudan University (Y2021-242). Tumors and adjacent normal tissues were then collected from patients who underwent surgical resection of the liver. All tissues were obtained immediately after surgical resection and frozen at − 80 °C. Huh7 cells, derived from the Chinese Academy of Sciences, were cultured in DMEM medium (D5796, sigma) containing 10% fetal bovine serum (16140071, Gibco) at 37 °C and 5% CO2.
Quantitative real-time PCR analysis
Total RNA was extracted from the tumors and adjacent normal tissues and reverse transcribed to cDNA using the Kit (EZBioscience, MN, USA). Then, quantitative PCR amplification was operated by a CFX384 real-time PCR machine (Bio-Rad, USA) using SYBR Green (Vazyme, China). Gene abundances were normalized to GAPDH. The primer sequences were shown in Table
3.
Table 3
Primer sequences for Real-time PCR
STARD5-F | CCGGGAAGGCAATGGAGTTT |
STARD5-R | TCATCCCACTTCACTCGTAGG |
FTCD-F | TCCCGACTTATCGACATGAGC |
FTCD-R | GCCGTACAGGTAAACTGGC |
SCN4A-F | TTCACAGGGATCTACACCTTTGA |
SCN4A-R | CACAAACTCTGTCAGGTACGC |
ADH4-F | AGTTCGCATTCAGATCATTGCT |
ADH4-R | CTGGCCCAATACTTTCCACAA |
CFHR3-F | TACCAATGCCAGTCCTACTATGA |
CFHR3-R | CCGACCACTCTCCATTACTACA |
CYP2C9-F | GCCTGAAACCCATAGTGGTG |
CYP2C9-R | GGGGCTGCTCAAAATCTTGATG |
CCL14-F | CCAAGCCCGGAATTGTCTTCA |
CCL14-R | GGGTTGGTACAGACGGAATGG |
GADD45G-F | CAGATCCATTTTACGCTGATCCA |
GADD45G-R | TCCTCGCAAAACAGGCTGAG |
SOX11-F | AGCAAGAAATGCGGCAAGC |
SOX11-R | ATCCAGAAACACGCACTTGAC |
SCIN-F | ATGGCTTCGGGAAAGTTTATGT |
SCIN-R | CATCCACCATATTGTGCTGGG |
SLC2A1-F | ATTGGCTCCGGTATCGTCAAC |
SLC2A1-R | GCTCAGATAGGACATCCAGGGTA |
Immunohistochemistry (IHC)
HCC and adjacent normal tissues were deparaffinized, rehydrated, blocked for endogenous peroxidase activity, antigen repair, and blocking, before being incubated overnight at 4 °C with primary antibodies against Stard5(ab178688, Abcam), N-cadherin, vimentin, E-cadherin and zo-1(9782 T, CST). The sections were then incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature and stained with 3, 3-diaminobenzidine tetrahydrochloride (DAB). Finally, cells were observed under a microscope.
Construction of stable cell lines
The cDNA or shRNA (Genepharma, China) targeting Stard5 were recombined into lentiviral vectors to overexpress or knockdown Stard5, then transfected into 293 T cells. The mature infectious lentivirus was collected after 72 h. Stable Stard5-overexpressing and Stard5-knockdown Huh7 cell lines were constructed and verified by western blot.
Western blot
Cells were lysed to extract total protein and heated to 100 °C for 20 min. Protein was added onto 8–12% SDS-PAGE electrophoreses and transferred to the PVDF membrane, then the blocked PVDF membrane was incubated with 1:1000 diluted Stard5 (ab178688, Abcam), β-Actin (3700 T, CST), N-cadherin, vimentin, E-cadherin and zo-1 (9782 T, CST) antibody at 4 °C overnight. After washing with TBST, the PVDF membrane was incubated with a 1:10000 diluted secondary antibody for 1.5 h at room temperature. Finally, a chemiluminescence analysis was performed.
Wound healing assay
The cells were inoculated in six-well plates at 1 × 106 cells per well to form a dense monolayer after 12 h. Lines were drawn with a 200 μL tip to the cell layer to form straight cell wounds. After washing with PBS, the cells were incubated with serum-free medium at 37 °C for 48 h. The wound width was recorded at 0 and 48 h.
Transwell migration and invasion assays
80 μl BD Matrigel mixture (diluted 1:10 with DMEM) was pre-coated in a transwell chamber (3513, Corning) at 37 °C overnight. Cells were diluted with serum-free DMEM and 4 × 104 cells were added to the upper chamber. Then, 500 μl of DMEM containing 30% fetal bovine serum was added to the bottom chamber. After incubation at 37 °C for 48 h, the chambers were fixed in 4% paraformaldehyde for 2 h. Cells in the upper chamber were removed, then stained with crystal violet, washed with PBS and photographed under a microscope.
Statistical analysis
R software (version 4.0.1) and GraphPad Prism (version 9.0) were used for statistical analysis of the experimental data. Pair or unpaired Student t-tests were used for comparison of data between two groups. The Mann–Whitney test was used when the data did not conform to a normal distribution. One-way analysis of variance (ANOVA) was used to compare three or more groups. Differences between the groups were considered statistically significant at p < 0.05.
Discussion
In our study, we constructed a risk model selected from multiple metabolic pathways based on differences between the high and low EMT groups in HCC, which contained 11 metabolism-related genes. The high-risk group had poorer prognosis than the low-risk group. The high-risk group was positively associated with Tregs and negatively associated with CD4 + T cells, NK cells. In addition, the high-risk group was more likely to develop drug resistance. These revealed that our model might support the prediction of patients’ response to chemotherapy and immunotherapy and provide a reference for individualized therapy for patients with HCC. More importantly, we demonstrated for the first time that Stard5 expression was positively correlated with survival time of HCC patients and negatively correlated with EMT, providing a precise therapeutic target.
EMT is a key cellular process that transforms polarized epithelial cells into a mesenchymal phenotype with increased cell motility. In cancer, EMT allows malignant cells to separate from the primary tumor and spread into the circulation, a critical process for invasion and metastasis [
18]. Altogether, EMT is an inevitable state that occurs prior to invasion and metastasis and could be used for predicting cancer progression and prognosis. Therefore, it is logical to group patients according to their EMT enrichment scores, which may eliminate the impact of many confounders on prognosis.
We can reasonably speculate that the energy requirements of a cell switching between motion and resting states must be altered, ultimately leading to metabolic reprogramming. In fact, recent evidence suggests that the link between EMT and metabolism is reciprocal, and that altered metabolism can drive EMT in some cases. Multiple metabolic pathways, involving in glucose metabolism, lipid metabolism, amino acids metabolism, mitochondrial biosynthesis, and many other events, are simultaneously altered in tumor progression and metastasis [
14]. Currently prognostic models of HCC are mostly constructed from mono-metabolic part, rather than multiple metabolic parts. In contrast, we screened and constructed risk model by comparing metabolic pathways of carbohydrates, LIPID, NUCLEOTIDE, TCA, ENERG, VITAMIN, and AMINOACID between EMT subgroups. It will be more reliable in predicting prognosis, in view of its closer proximity to the molecular biological level of cancer cells prior to invasion. In our study, there were differences in metabolism between the EMT groups, most notably in terms of energy and lipids. We classified patients with HCC into three molecular subtypes based on genes of differential metabolic pathways. Prognostic differences existed among these three subtypes, and cluster 1 had the best prognosis. The KEGG analysis showed a major focus on cell polarity, extracellular matrix, and lipid metabolic pathways. Thus, targeting specific metabolic enzymes has the potential to reverse EMT and ultimately limit cancer metastasis.
In recent years, molecular prognostic markers have received increasing attention for predicting the survival of HCC [
19,
20]. Compared to single gene marker, multi-gene models have the advantage of higher predictive accuracy and more individualized results. Based on the DEGs among 3 clusters, a prognostic risk model including 11 genes (
STARD5, FTCD, SCN4A, ADH4, CFHR3, CYP2C9, CCL14, GADD45G, SOX11, SCIN, and
SLC2A1) was developed using univariate Cox and LASSO-Cox regression. External databases validated this risk model as valid and stable in predicting the prognosis of patients with HCC. SLC2A1, also named GLUT1, have been widely confirmed overexpressing in HCC and promoting metastasis [
21]. SOX11 was also significantly upregulated in HCC [
22]. FTCD, ADH4, CFHR3, CYP2C9, CCL14, GADD45G was down-regulated in HCC [
23‐
28]. These papers supported the high accuracy of our prognostic model. Among them, FTCD and CFHR3 have been reported to play a suppressive role in the invasion and migration of HCC [
23,
29]. No studies have shown a link between the remaining molecules and EMT in HCC. In addition, there are no reports on the expression of SCN4A or SCIN in HCC. We demonstrate for the first time that they were risk factors for HCC and were associated with EMT, which contributes to a better understanding of the molecular mechanisms of HCC progression. These molecules may be crucial triggers in controlling HCC metastasis.
Then, tumor microenvironment of patients was analyzed in the two groups. High risk group tended to be more immunosuppressed, with higher Tregs, and fewer CD8 + T, CD4 + T, and NK cells. In addition, patients in the high-risk group were more resistant to chemotherapy. These results imply that the risk model can help predict the effectiveness of immunotherapy and chemotherapy. Targeting these metabolic genes may improve the response of patients to treatment and provide new ideas for personalized medicine.
Stard5 became the focus of our attention, which had hardly been studied in cancers. Stard5, a lipid-binding protein, has a conserved steroidogenic acute regulatory protein-related lipid transfer domain [
30]. It is involved in the regulation of cholesterol homeostasis in vivo by binding and transporting cholesterol and other sterol-derived molecules to the liver [
31]. In hepatocytes, Stard5 reduces lipid accumulation, suggesting that Stard5 dysregulation may play an important role in fatty liver disease [
31]. Mutations in the
STARD gene may lead to autoimmune diseases or cancer [
32]. Additionally, Mulford et al. showed that knockdown of Stard5 expression resulted in reduced sensitivity of lung cancer cells to etoposide [
33]. In the present study, we demonstrated for the first time that Stard5 was down regulated in HCC tissue, and low Stard5 expression suggested poor prognosis. Stard5 deficiency contributed to the invasion and migration in HCC cell lines, while overexpression of Stard5 showed the opposite effect. The protein expression of EMT pathway was associated with Stard5 expression. These data suggest that Stard5 was a protective factor in patients with HCC.
Studies have found that endoplasmic reticulum (ER) stress increases Stard5 expression in mouse hepatocytes, and that Stard5 plays a key role in ER cholesterol homeostasis during ER stress [
31].
When tumor cells experience ER stress in response to intrinsic and extrinsic changes, a network of adaptive signals, known as the unfolded protein response (UPR), will be evoked to restore protein homeostasis. UPR hyperactivation has been demonstrated to regulate cell survival, angiogenesis, inflammation, invasion, and metastasis [
34]. Tumors exploit UPR signaling to promote EMT [
35]. Therefore, we speculate that when ER stress occurs, Stard5 may transport excess cholesterol from the ER to the Golgi and then to the efflux pathway during the UPR, preventing excessive cholesterol accumulation in the ER, restoring ER homeostasis, and promoting apoptosis. When stard5 deficiency, ER stress induces cholesterol imbalance, the UPR may be hyperactive and unfolded proteins activate ER-resident sensors, which in turn promotes the EMT. However, it remains to be experimentally verified. Targeting stard5 directly during EMT with concomitant metabolic reprogramming may offer a prospective direction for targeting therapy.
Some limitations remain in our study, the function of stard5 in inhibiting EMT still needs to be further explored. The role of the other 10 genes in HCC remains to be studied in vitro and in vivo. Furthermore, normal tissues require the same metabolic pathways for their survival and proliferation. This implies that targeting tumor metabolism faces a series of challenges.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.