DUSP12 regulates the tumorigenesis and prognosis of hepatocellular carcinoma

Public databases

Cell lines and culture conditions

The human liver-cancer cell line Huh7 was acquired from American Type Culture Collection (Manassas, VA, USA) and cultivated in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Grand Island, NY, USA) with 10% fetal bovine serum (Gibco) and 1% penicillin–streptomycin (Gibco) at 37 °C in an atmosphere of 5% CO₂.

Western blotting

Cell proteins were extracted by denaturing buffer and then quantified by a bicinchoninic acid protein assay (Thermo Scientific, Waltham, MA, USA). Protein lysates from the HCC cell line were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes (Millipore, Bedford, MA, USA), blocked, and then detected by primary antibody DUSP12 (1:2000 dilution; catalog number: ab237008; Abcam, Cambridge, UK) and horseradish peroxidase-conjugated secondary antibody (Sigma–Aldrich, Saint Louis, MO, USA). These actions were followed by exposure to enhanced chemiluminescence. The housekeep gene β-tubulin (1:500; ab6046; Abcam) was used as a loading control.

Plasmids and lentivirus production

Annealing and connection of short hairpin (sh)RNA were undertaken followed by construction into the modified plasmid pLKO.1. Well-constructed vectors were transinfected into HEK293T cells by lentivirus packaging plasmids psPAX and pMD2.0G. The shRNA sequences and DUSP12 primer sequence (forward and reverse, respectively) were: CCGGGTTGAGTGGCAACTGAAATTATCTCGAGATAATTTCAGTTGCCACTCAAGTTTTTG, and AATTCAAAAAGTTGAGTGGCAACTGAAATTATCTCGAGATAATTTCAGTTGCCACTCAAG for shDUSP12-1; CCGGGTGGATACCTCTAGTGCAATTCTCGAGAATTGCACTAGAGGTATCCACTTTTTG, and AATTCAAAAAGTGGATACCTCTAGTGCAATTCTCGAGAATTGCACTAGAGGTATCCAC for shDUSP12-2.

Cell-growth assay

Lentivirus-infected stable cells were seeded into 96-well plates and cultured in DMEM containing 10% fetal bovine serum (2000 cells per well, five parallel wells). Then, cells were collected at different timepoints. The cell number in each well was counted by Cell Counting Kit 8 (CCK8). Absorbance at 450 nm was measured to determine of the number of viable cells.

Transwell™ assay

Lentivirus-infected stable cells were seeded in the upper chamber of a Transwell chamber (24-well (8-µm pore; Corning, Corning, NY, USA) in 200 µL of serum-free DMEM (1 ×10⁵ cells per well, five parallel wells). Then, 800 µL of DMEM containing 10% fetal bovine serum was added to the lower chamber and incubation allowed for 36 h at 37 °C. After removing the cells at the upper surface of the membrane, cells were passed through a filter and fixed with 4% paraformaldehyde, stained with 0.1% Crystal Violet solution and photographed using an inverted fluorescence microscope.

Pattern of DUSP12 transcriptional expression using Ualcan and HCCDB databases

Analyses of the TCGA-LIHC dataset in the Ualcan database and analyses of ICGC-LIRI-JP and GSE14520 cohorts revealed that transcriptional expression of DUSP12 was higher in LIHC tissues compared with that in normal liver tissues (Fig. 1A, Fig. S1). Analyses of clinical subgroups demonstrated that DUSP12 expression was higher in an Asian, tumor–node–metastasis (TNM) stage-III, grade-3, P53-mutant group than that in a Caucasian, TNM stage-I/II, grade-1/2, P53-nonmutant group. The non-significant change in DUSP12 expression between the stage-IV/grade-4/N1 group compared with that in other groups may have been caused by the number of samples in the stage-IV/grade-4/N1 group being significantly less compared with that in other groups. The top positively and negatively correlated genes in TCGA dataset for DUSP12 were also downloaded from UALCAN database (Figs. S1 and S2). We suspected that DUSP12 could be a pathological and prognostic marker of LIHC (Figs. 1B–1F). Integrative analyses of HCCDB revealed that mRNA expression of DUSP12 was higher in liver-cancer tissue compared with that in normal liver tissue in 11 cohorts (Fig. 2A). Analyses of functional enrichment of genes co-expressed with DUSP12 in liver-cancer tissues using Metascape showed that these genes were engaged mainly in “histone methylation”, “cullin RING ubiquitin ligase complexes”, “nuclear specks” and “ubiquitin-like protein transferase activity” (Figs. 2B–2E).

Validation of DUSP12 expression in tissues at the protein level and liver-cancer lines

DUSP12 expression was higher in liver-cancer tissues than that in normal liver tissues at the protein level, and this observation was validated by results from the HPA database (Antibody: HPA008840; cancer-patient ID: 2399; normal-patient ID: 3222) (Figs. 3A, 3B) and HUH1 had the highest transcriptional expression of DUSP12 in liver-cancer cell lines (Fig. 3C).

Survival analyses of liver-cancer patients with different DUSP12 expression

LIHC patients with higher expression of DUSP12 had shorter OS and DFS in the TCGA-LIHC dataset (Figs. 4A, 4B). Next, we validated the result by survival analyses of liver-cancer patients in the Kaplan–Meier Plotter database. We showed that patients with higher expression of DUSP12 had shorter OS, DFS, PFS and DSS than that of patients with lower expression of DUSP12 (Figs. 4C–4F).

Analyses of DUSP12 expression using the cBioPortal database

An online tool in the cBioPortal database was utilized to analyze the mutant status of DUSP12 in the Liver Hepatocellular Carcinoma (TCGA, Firehose Legacy) dataset. The mutant frequency of DUSP12 in HCC was 33.0% (Fig. 5A), which was composed mainly of amplification and high expression of mRNA (Fig. 5B). mRNA expression of DUSP12 in HCC with amplification was higher than that in those without alteration (Fig. 5C). HCC patients with altered DUSP12 had shorter OS than HCC patients with nonaltered DUSP12 (Fig. 5D). CNV analyses revealed that mRNA expression of DUSP12 was higher in HCC patients with DUSP12-amplification and DUSP12-gain patients than in those with DUSP12-shallow deletion and DUSP12-diploid (Fig. 6A). In general, mRNA expression of DUSP12 was correlated negatively with methylation of the promoter region of DUSP12 in 359 HCC samples from the TCGA-LIHC (Firehose Legacy) dataset (Fig. 6B). DUSP12-altered patients had a higher serum level of alpha fetoprotein (AFP) at procurement (Fig. 6C), fraction of genome altered (Fig. 6D) and worse histology grade in neoplasms (Fig. 6E). In addition, we screened 392 DEGs between DUSP12-altered patients and DUSP12-nonaltered patients with false discovery rate <0.05 and —log(ratio)—>1 (Fig. 7A). Analyses of functional enrichment and disease-related enrichment of these 392 DEGs revealed that these genes mainly took part in: “M61392: CHIANG LIVER CANCER SUBCLASS PROLIFERATION DN”; “M16496: CHIANG LIVER CANCER SUBCLASS CTNNB1 UP”; “M3268: CHIANG LIVER CANCER SUBCLASS PROLIFERATION UP” (Figs. 7B, 7C). Minimal Common Oncology Data Elements (MCODE) analyses revealed clustering of seven MCODEs (Fig. 7D). MCODE 1 mainly included proteins that took part in the cell cycle (e.g., cluster of differentiation (CD)K1, CDC20, PLK1), cell division or mitosis (e.g., AURKB, BIRC5, KIF2C, CENPA, CENPF). MCODE 2 mainly included cytochrome P450 (CYP) monooxygenase proteins that took part in various types of metabolism. MCODE 3 mainly included UDP-glucuronosyltransferase (UDPGT) proteins. MCODE 4 included GCK, OTC, SDS, POLE2 and ATP6V1B1. MCODE 5 included SULT4A1, SULT1B1, HS6ST2, and GAL3ST1. MCODE 6 mainly included C-C motif chemokine proteins. MCODE 7 included FMO3, ACSL6 and HSD11B1. MCODE 1 and MCODE 2 may play a critical part in the proliferation and biological activity of HCC cells. MCODE 6 may have a correlation with the infiltrations of immune cells in HCC samples.

Analyses of DUSP12 expression using the TIMER database

We explored the correlation of DUSP12 expression with TIICs by TIMER database. DUSP12 expression was moderately (partial correlation >0.2) positively correlated with the abundance of infiltrating B cells, CD4+ T cells, macrophages, neutrophils, and dendritic cells (Fig. 8A). Then, we explored the correlation between DUSP12 and immune cells with other immune-infiltration analyzing methods (xCELL, EPIC, CIBERSORT). The result suggested that DUSP12 expression was moderately (partial correlation >0.2) positively correlated with the abundance of infiltrating B cells and CD4+ T cells regardless of the method employed (Table 1). DUSP12 expression had a moderately positive correlation with the immune-checkpoint moieties HAVCR2, TIGIT, CTLA4 and PD-1 (Fig. 8B). The infiltration level of CD8+ T cells, macrophages, neutrophils, and dendritic cells was significantly different in DUSP12 with different CNV (Fig. 8C). Furthermore, we explored mRNA expression of DUSP12 in patients with different immune-subgroup liver cancer using the TISIDB. We revealed that the C1 (wound healing) subgroup had the highest DUSP12 expression, whereas the C3 (inflammatory) and C6 (TGF-β dominant) subgroups had lower expression of DUSP12 (Fig. 8D). Furthermore, groups with high expression of DUSP12 had a higher ratio of T-regulatory cells (T_regs) and activated natural-killer cells compared with those with low expression of DUSP12 (Fig. 9).

Knockdown of DUSP12 expression decreases the proliferation and migration of Huh-7 cells

We evaluated the probabilities of dependency of DUSP12 in 22 HCC cell line types with data from CRISPR (DepMap 21Q2 Public + Score, CERES) cohort. The CERES scores of cell lines ranged from −0.04 to −0.47 while mean value was equal to −0.21 (Fig. S2). In general, a lower score meant that a gene is more likely to be essential in a given cell line and a score <0 meant that down-regulation of a gene may inhibit the proliferation of a given cell line. Human liver-cancer cells (Huh-7) were transfected with a specific shRNA for DUSP12 (shDUSP12) and a nonspecific shRNA (NC) (Fig. 10A). CCK8 and Transwell assays were utilized to evaluate the ability of cells to proliferate and migrate. Knockdown of DUSP12 expression led to the reduced proliferation and motility of Huh7 cells (Figs. 10B, 10C).