DPM1 expression as a potential prognostic tumor marker in hepatocellular carcinoma

Datasets

Datasets used for correlation analysis between DPM1/2/3 and chronic liver disease (CLD) were obtained from GEO database (http://www.ncbi.nlm.nih.gov/geo/) after searching for keywords related to CLD. We selected three separate gene expression profiles (GSE114783, GSE128726 and GSE89632) for our study and the detailed information of the datasets was shown in Table 1. The data used for ROC curve plotted were collected from TCGA LIHC datasets. The figures about the relationship between DPM1/2/3 expression and CLD were drawn using R package, ggplot2 v3.3.2. The significance of DPM1/2/3 expressions between normal and CLD samples were analyzed via unpaired Student’s t-test. The ROC curves were created by R package, pROC v1.16.2.

UALCAN

UALCAN (http://ualcan.path.uab.edu) is a comprehensive, user-friendly, and interactive web resource and provides data online analysis and mining based on cancer OMICS data (TCGA and MET500). It is designed to analyze relative transcriptional expression of potential genes of interest between tumor and normal samples and association of the transcriptional expression with relative clinicopathologic parameters. In addition, it is also used to evaluate epigenetic regulation of gene expression and pan-cancer gene expression (Chandrashekar et al., 2017). In our study, UALCAN was used to analyze the mRNA expressions of three subunits of DPMS in HCC samples and their relationship with clinicopathologic parameters. Difference of transcriptional expression or pathological stage analysis was compared by Student’s t-test and p <0.05 was considered as statically significant.

Human Protein Atlas

The Human Protein Atlas (https://www.proteinatlas.org) is a website that provides human proteins data in cells, tissues and organs, including immunohistochemistry-based expression data for near 20 common kinds of cancers (Asplund et al., 2012). The database can be conveniently used to compare the protein differential expressions of interest genes in tumors and normal tissues. In this study, direct comparison of protein expression of three subunits of DPMS between human normal and HCC tissues was performed by immunohistochemistry image.

GEPIA

Gene Expression Profiling Interactive Analysis (GEPIA) is a database developed and built by the team of professor Zhang of Peking University based on the data of the UCSC Xena project. It is an interactive web server that can dynamically analyze and visualize TCGA (The Cancer Genome Atlas) gene expression profile data. It can provide customizable and powerful functions, including differential expression analysis between tumor and normal samples, profiling plotting, survival analysis, similar gene detection, and so on Tang et al. (2017). In the current study, we operated correlative prognostic analysis and similar gene detection of DPM1, DPM2 and DPM3, respectively. p < 0.05 was considered as statically significant. The significance of expression analysis was completed using Student’s t-test. Kaplan–Meier curve was used to accomplish prognostic analysis.

cBioPortal

cBioPortal (http://www.cbioportal.org), an online open-access website resource, can display multidimensional cancer genomics data in a visual form. It can also help researchers explore the genetic changes between samples, genes and pathways, and combine them with clinical results (Gao et al., 2013). In this experiment, we studied the genomic profiles of DPMS three subunits, which included putative copy-number alterations (CNAs) from genomic identification of significant targets in cancer (GISTIC) and mRNA Expression z-Scores (RNASeq V2 RSEM) were gained with a z-score threshold ±1.8. Genetic alterations in DPMS and their association with overall survival (OS) and disease free survival (DFS) of HCC patients were exhibited as Kaplan–Meier plots and log-rank test was implemented to confirm the significance of the difference between the survival curves, and when a p value <0.05, the difference was statically significant.

Metascape

Metascape (http://metascape.org), a free and credible gene-list analysis device, can be used for gene annotation analysis and function analysis. It is a mechanized meta-analysis device that can realize habitual and different pathways in a set of orthogonal target-discovery studies (Zhou et al., 2019). In this work, Metascape was used to implement function and pathway enrichment analysis of DPMS members and their similar genes that acquired using GEPIA. Statistically significant difference was p < 0.05 and minimum enrichment number was 3. Databases containing OmniPath and BioGrid were used for protein-protein interactions enriched analysis. Futhermore, Molecular Complex Detection (MCODE) was supposed to recognize closely related protein components.

Cell culture

The human hepatoma cells SMMC-7721, HepG2 and immortal hepatic cell QSG-7701 involved in the experiment were gained from Institute of Cell Biology (Shanghai, China). All cell lines were cultured in RPMI-1640 or DMEM medium (Gibco/Invitrogen, Camarillo, CA, UNITED STATES) supplied with 10% fetal bovine serum (PAN-Biotech, Aidenbach, Germany), and then all cells were incubated at 37 °C in a 5% CO2 environment.

RT-qPCR

TRIeasy™ Total RNA Extraction Reagent (Yeasen, Shanghai, China) was used for total RNA extraction, and then the total RNA was reverse transcribed to cDNA with the Hifair^® 1st Strand cDNA Synthesis Kit (Yeasen, Shanghai, China) according to the product instruction. Hieff UNICON®Power qPCR SYBR Green Master Mix (Yeasen, Shanghai, China) was used to conduct RT-qPCR experiment on a Bio-Rad CFX96 System (Bio-Rad, Hercules, CA, USA). The reaction conditions were as follows: pre-denaturation at 95 °C for 30 s, followed by 40 cycles of amplification at 95 °C for 10 s and 60 °C for 30 s. Relative mRNA expression levels of DPM1/2/3 were measured based on the 2^−ΔΔCt method with 18S used for normalization. The significance of expression analysis was completed using Student’s t-test. Table 2 showed the primers we used in this study.

Transcriptional levels of DPMS in liver cancer

In order to explore the gene expressions of three subunits of DPMS in different types of cancer, mRNA expressions of DPM1, DPM2 and DPM3 were analyzed by UALCAN. As was shown in Fig. 1, we observed that DPM1, DPM2 and DPM3 had higher mRNA expressions for most kinds of tumor samples compared to normal samples, respectively. For example, mRNA expression levels of DPM1 and DPM2 were very highly expressed in colon adenocarcinoma (COAD) (DPM1, p = 1.62E−12; DPM2, p <1E−12), head and neck squamous cell carcinoma (HNSC) (DPM1, p <1E−12; DPM2, p = 1.62E−12), esophageal carcinoma (ESCA) (DPM1, p = 1.22E−07; DPM2, p = 2.30E−02), liver hepatocellular carcinoma (LIHC) (DPM1, p = 1.62E−12; DPM2, p <1E-12), rectum adenocarcinoma (READ) (DPM1, p = 4.07E−09; DPM2, p = 1.62E−12 ) and so on (Figs. 1A, 1B). Similarly, DPM3 gene was particularly highly expressed in breast invasive carcinoma (BRCA) (p = 1.62E−12), ESCA (p = 8.22E−10), LIHC (p = 1.11E−16) and glioblastoma multiforme (GBM) (p = 1.53E−05) (Fig. 1C). Thus, our results showed that transcriptional expressions of DPMS were significantly over–expressed in many different types of cancer. In particular, all three subunits of DPMS were expressed highly in LIHC and ESCA. Next, we examined the specific mRNA expressions of DPM1, DPM2 and DPM3 in liver tumor using UALCAN database. As was shown in Figs. 2A, 2B and 2C, mRNA expressions of three genes were all found significantly up-regulated in HCC tissues compared to normal samples (all p < 0.001). We next performed the protein expression levels of DPMS in HCC using Human Protein Atlas database. Results indicated that medium and low protein expressions of DPM1 and DPM3 were expressed in normal liver tissues, while high protein expressions of them were showed in HCC tissues (Figs. 2D, 2F). In addition, DPM2 protein were not detected in normal liver tissues, whereas medium expression of DPM2 were observed in HCC tissues (Fig. 2E). In general, the results indicated that transcriptional and proteinic expressions of DPMS were both over-expressed in patients with HCC.

Relationship between the mRNA levels of DPMS and the clinicopathological parameters in liver cancer patients

Because we observed mRNA and protein levels of DPMS were over-expressed in HCC patients, we subsequently investigated the connection between mRNA expressions of DPMS members with clinicopathological features of HCC patients with UALCAN, containing tumor grades and patients’ individual cancer stages. As presented in Fig. 3, mRNA expressions of DPMS members were significantly associated with tumor grades, and the mRNA expressions of DPMS headed to be higher with tumor grade elevated. The maximum mRNA expressions of DPM1/2 were showed in tumor grade 4 (Figs. 3A, 3B), whereas the supreme mRNA expression of DPM3 was found in tumor grade 3 (Fig. 3C). The reason why mRNA expression of DPM3 in grade 3 seemed to be higher than that in grade 4 may be attributed to the small sample size (only 12 HCC patients at grade 4). Similarly, the mRNA expressions of DPMS were noticeably related to the cancer stage of patients so, the patients with more advanced cancer, the higher in mRNA expressions of DPMS. The highest mRNA expressions of DPM1/2 were observed in tumor stage 3 (Figs. 3D, 3E), while the maximum DPM3 mRNA expression was noticed in stage 4 (Fig. 3F). Briefly, the results above indicated that mRNA expressions of DPMS were obviously associated with pathological parameters in HCC patients. Moreover, HCC usually developed from CLD caused by hepatitis B virus (HBV) and hepatitis C virus (HCV) infection, fatty liver and so on. The relationships between the expressions of DPM1/2/3 and CLD including HBV-related liver cirrhosis, HCV-related liver cirrhosis and non-alcoholic steatohepatitis were also analyzed. Three suitable datasets (GSE114783, GSE128726 and GSE89632) were chosen for verifying the expression of DPM1/2/3 in CLD. We found that the expressions of DPM1/2/3 in HBV and HCV-related liver cirrhosis and non-alcoholic steatohepatitis samples were more or less higher than normal samples (Fig. 4). Therefore, the expressions of DPM1/2/3 were also related to disease development of HCC.

Prognostic value of mRNA expression of DPMS in liver cancer patients

To assess the value of differentially expressed DPMS in the progression of HCC, we used GEPIA to evaluate the relationship between differentially expressed DPMS and clinical outcome. OS curves were presented in Fig. 5. We detected that liver cancer patients with low transcriptional levels of DPM1 (p = 0.007), DPM2 (p = 0.0032) and DPM3 (p = 0.029), were significantly connected with longer OS (Figs. 5A, 5B, and 5C). The worth of differentially expressed DPMS in the DFS of HCC patients was also estimated. Noteworthy, the longer DFS indicated to the HCC patients with lower DPM2 transcriptional levels (p = 0.049) (Fig. 5E). The receiver operating characteristic (ROC) curves were used to detect the prediction accuracy of DPM1/2/3 in distinguishing the HCC from the normal samples compared with existed liver tumor markers containing alpha-fetoprotein (AFP), glypican-3 (GPC-3) and transforming growth factor- β1 (TGFβ1). Our results indicated that DPM1/2/3 had a better performance than AFP and TGFβ1 for the diagnosis of HCC (Fig. 6). Area under the curve (AUC) of the DPM1/2/3 were 0.709, 0.860 and 0.746, respectively (Figs. 6A, 6B and 6C). AUC of the existed tumor markers including AFP, GPC-3 and TGFβ1 were 0.679, 0.879 and 0.577, respectively (Figs. 6D, 6E and 6F). Taken together, DPM1/2/3 may be also potential biomarkers for diagnosis or screening of HCC besides AFP, GPC-3 and TGFβ1.

DPMS genetic alteration and similar gene network in patients with HCC

Next, we implemented a universal analysis of the molecular characteristics of differentially expressed DPMS. Genetic variations of differentially expressed DPMS in HCC was analyzed utilizing cBioPortal. A total of 366 samples from TCGA pan cancer database were studied, and altered gene set or pathway was detected in 151queried samples (alteration rate was 41%). The alteration rates of DPM1, DPM2, and DPM3 were 19%, 6% and 24%, respectively (Figs. 7A, 7B). The most prevalent change in these samples was enhanced mRNA expression. The Kaplan–Meier plotter results and log-rank test presented a considerable difference in OS (p = 0.0264), but no remarkable difference in DFS (p = 0.0841) between the samples with changes in one of the target genes and those without variations in any target genes (Figs. 7C, 7D).

Functional enrichment analysis of DPMS in patients with HCC

Top 50 genes similar to DPM1, DPM2 and DPM3 respectively (a total of 150 genes) were searched by GEPIA (Table S1). Next, the functions of DPMS and their similar genes were predicted by analyzing GO and KEGG in Metascape. The top 20 GO enrichment items were classified into three functional groups: biological process group, molecular function group, and cellular component group (Figs. 8A, 8B and Table 3). The DPMS members and their similar genes were mainly enriched in biological processes such as ncRNA processing, DNA repair, viral gene expression, deoxyribonucleoside triphosphate metabolic process and so on. The molecular functions regulated by DPMS and their similar genes were snRNP binding, ubiquitin binding, nucleotidyltransferase activity and ubiquitin-like protein transferase activity. The cellular components affected by DPMS and their similar genes were involved in transferase complex, methyltransferase complex, chromosomal region and nucleolar part.

The 6 most significant KEGG pathways for the DPMS and their similar genes were displayed in Figs. 8C, 8D and Table 4. These pathways comprised pyrimidine metabolism, RNA transport, ubiquitin mediated proteolysis, mTOR signaling pathway and so on. Moreover, for more comprehending the relationship between DPMS and HCC, we performed enrichment analysis of protein–protein interaction with Metascape. Figures 8E and 8F exhibited the protein interaction correlation and important MCODE components. The top 3 essential MCODE components were achieved from the protein–protein interaction network. After function and pathway enrichment analysis for each MCODE constituents respectively, the results demonstrated that biological functions regulated by DPMS and their similar genes were mainly related to mRNA and RNA splicing, protein export form nucleus and nucleocytoplasmic transport.

The mRNA expression levels of DPM1/2/3 in vitro

We evaluated DPM1, DPM2 and DPM3 expression levels in a panel of three cell lines: two hepatoma cells (HepG2 and SMMC-7721) and one normal liver cell line (QSG-7701). The mRNA expression measured by RT-qPCR revealed that DPM1 transcription levels in cancerous cell lines were higher than that in normal liver cells (Fig. 9A) and the result was consistent with our prediction. Moreover, the expression of DPM2 and DPM3 in SMMC-7721 cell was significantly increased, while those expression did not change significantly in HepG2 cell (Figs. 9B, 9C). This discrepancy may be due to a number of differences between cell types and more cell and tissue samples are needed to validate the results. Therefore, DPM1 could be the most potential prognostic biomarker for survivals of HCC patients.