Integrated network findings reveal ubiquitin-specific protease 44 overexpression suppresses tumorigenicity of liver cancer

Patient sample collection

Retrospectively collected clinical specimens were used in this study. Consent was obtained from all patients. The patients’ tumors were histologically confirmed as hepatocellular carcinoma. The data were collected from 2007–2008. The Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster approved the use of the clinical specimens for research. The data were extracted from the electronic patient record system under the IRB reference number (UW 05-359 T/1022). Tissue specimens for RNA extraction were immediately preserved in RNAlater (Thermo Fisher Scientific, Waltham, MA, USA) and fixed in formalin for preparation of paraffin blocks. Clinical specimens were collected and processed by surgical tissue bank staff in the Department of Surgery, University of Hong Kong. Patients undergoing tumor resection at Queen Mary Hospital were included. The clinicopathologic parameters required for this study were available from the surgical tissue bank.

cDNA synthesis and quantitative real-time PCR

RNA was isolated from 40 pairs of HCC and adjacent nontumoral liver tissues using TRIzol reagent (Tiangen, Beijing, China) according to the manufacturer’s instructions. First-strand cDNA was synthesized from 1 μg total RNA using the FastKing gDNA Dispelling RT SuperMix Kit (Tiangen). Expression levels of USP44 were determined using the SuperReal PreMix Plus (SYBR Green) kit for quantitative real-time PCR (RT-PCR). Emission intensity was detected using the CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA) and following these steps: initial denaturation step at 95°C for 15 min, and 40 thermal cycling steps consisting of 10 s at 95°C, 20 s at 55°C, and 30 s at 72°C. Threshold cycles were averaged from triplicate reactions. To adjust for variations in the starting template, gene expression was normalized to glyceraldehyde 3-phosphate dehydrogenase. The mRNA expression of USP44 in HCC was compared with that in adjacent nontumoral liver tissue. The primer sequences were shown below (GAPDH_F: 5′-GAGAAGGCTGGGGCTCATTT-3′; GAPDH_R: 5′-AGTGATGGCATGGACTGTGG-3′; USP44_F: 5′-CCCTCAGACCAGAATGCTTTA-3′; USP44_R: 5′-CATTGCAGTGTACCCAGAACC-3′). For the transcriptome sequencing validation, the primer sequence of the select genes was shown below (CCND2_F: 5′-TTTGCCATGTACCCACCGTC-3′; CCND2_R: 5′-AGGGCATCACAAGTGAGCG-3′; CCNG2_F: 5′-CAGGATTGAGAAATGCCAAAGT-3′; CCNG2_R: 5′-TGACAGCCAGGACAAAAGTT-3′; SMC3_F: 5′-ATTGGTGCCAAAAAGGATCAGT-3′; SMC3_R: 5′-GATTGCTTCGAGAAAAACCAGC-3′; THBS1_F: 5′-GCTGGAAATGTGGTGCTTGTCC-3′; THBS1_R: 5′-CTCCATTGTGGTTGAAGCAGGC-3′; FZD2_F: 5′-GTGCCATCCTATCTCAGCTACA-3′; FZD2_R: 5′-CTGCATGTCTACCAAGTACGTG-3′; ROCK2_F: 5′-TGGTTTCTATGGGCGAGAATGT-3′; ROCK2_R: 5′-CAAGTCGTACCTCCCTATCTGTT-3′).

Cell culture

The HCC cell line HepG2 was cultured in minimal essential medium (Solarbio Science and Technology, Beijing, China) supplemented with 10% (v/v) fetal bovine serum (Gibco, Waltham, MA, USA) and 1% penicillin-streptomycin solution (Solarbio Science and Technology). Cells were grown in an air atmosphere with 5% (v/v) CO₂ at 37°C.

Lentiviral infection and establishment of USP44-stable cell lines

The full-length human USP44 gene was cloned into GV492 vector (GeneChem, Westmount, QC, Canada). Using Lipofectamine 2000 transfection reagent (Thermo Fisher Scientific), HEK293FT cells were transfected with packaging plasmids pCMV-VSV-G, pRSV-Rev, and pMDLg/pRRE along with the USP44–DEST vector or control vector. At 48 h post-transfection, the viral supernatant was precipitated with PEG-it Virus Precipitation Solution (System Biosciences, Palo Alto, CA, USA) in a 1:4 ratio to produce a concentrated viral stock. HepG2 cells were plated 1 day before transduction. At 72 h post transduction, cells were selected in medium containing 2 μg/mL puromycin for 10 days. Cell lines stably expressing USP44 or control vector were used for the functional study.

Cell proliferation assays

Stably transfected HCC cells were seeded into a 96-well plate at a cell density of 4 × 10³ cells per well, with four replicate wells. The cells were incubated for 1–6 days. After incubation, cell proliferation was measured using the CCK-8 assay (Data Inventory Biotechnology, Hong Kong). The colorimetric product formed was measured at an absorbance of 450 nm and 600 nm (optical density change: OD_{450 nm} − OD_{600 nm}).

Cell cycle analysis by propidium iodide

Stably transfected cells (8 × 10⁵) were harvested when they reached approximately 80% confluence. The cells were trypsinized, washed with phosphate-buffered saline, and fixed in 1 mL ice-cold 70% ethanol at 4°C overnight. The fixed cells were pelleted by centrifugation at 1000 g for 5 min. The cells were then washed with phosphate-buffered saline and stained with 0.5 mL staining solution containing 50 μg/mL propidium iodide and 0.5 μg/mL RNase A for 30 min. The cells then underwent flow cytometric analysis using the FACSCanto and CellQuest software applications (BD Biosciences, Franklin Lakes, NJ, USA). Average values of the G0/G1, S, and G2/M phases were obtained from at least three independent experiments.

Colony formation assay

Stably transfected HCC cells were seeded into 6-well plates (1000 cells/well). After incubation for 10 days, the cells were fixed in 4% paraformaldehyde for 30 min, stained with 0.1% crystal violet stain solution for 30 min, and then destained with water. The colonies (>20 cells/colony) were counted and recorded.

Transcriptome sequencing and bioinformatic analysis

RNA quality and quantity were assessed using a Bioanalyzer 2100 and the RNA 6000 Nano LabChip Kit (Agilent Technologies, Santa Clara, CA, USA), and high-quality samples with an RNA integrity number >7.0 were used to construct the sequencing library. The cDNA library was ligated with indexed adapters, and the ligated products were amplified by PCR. The average insert size of the final cDNA library was approximately 300 bp. Next, 2 × 150 bp paired-end sequencing was performed on a NovaSeq system (Illumina, San Diego, CA, USA). High-quality clean reads were filtered using Cutadapt and the quality-trimmed sequence reads were mapped to the human genome reference (Homo sapiens Ensembl v96) using the HISAT2 software application (version: hisat2-2.0.4) [17]. The mapped reads of each sample were assembled using StringTie (version: stringtie-1.3.4d.Linux_x86_64) with default parameters [18]. StringTie and ballgown were used to estimate the expression levels of all transcripts and all genes. Genes with a |log2 (fold change: USP44/vector)| > 1 and log10 (q value) > 1.3 were considered to be differentially expressed genes (DEGs). The DEGs were subjected to gene ontology and pathway enrichment analysis using the Database for Annotation, Visualization, and Integrated Discovery tool [19]. Ingenuity Pathway Analysis (IPA) was used to further understand the diseases, biologic functions, and gene networks controlled by USP44-mediated DEGs.

Statistical analysis

Statistical analyses of patient information, including age, sex, expression level of AFP, and the staging of HCC were performed using the Pearson chi-squared test. The positive of HBsAg in HCC, number of tumor nodules, degrees of differentiation, venous infiltration, and the presence of cirrhosis in HCC were performed using the Fisher's exact test. Size of largest tumor-length was assessed by the Student's t-test. The SPSS-PASW Statistics 18 software was used for statistical analyses. Correlations between clinicopathologic characteristics and USP44 expression levels were assessed by the Student’s t-test to establish differential gene expression in relation to specific tumor phenotypes. Quantitative data are reported as means with standard deviation or standard error of the mean. The Wilcoxon matched-pairs signed rank test was used to compare USP44 expression levels between groups. Patient survival was calculated using the Kaplan–Meier method from the date of surgery to the date of death or last follow-up. The log-rank test was used to compare survival between the groups. The Z-score is determined by using the Ingenuity Pathway Analysis. The DEGs of each treatment group were uploaded to the IPA system for functional analysis (Qiagen, Hilden, Germany). Canonical Pathway Analysis of IPA was used to identify significantly enriched canonical pathways in each treatment group, and Diseases and Functions Analysis of IPA was used to identify significantly enriched diseases and biological functions in each treatment group. The pathway or function with P-value < 0.05 was considered statistically significant. The activation z-score is to infer the activation states of predicted transcriptional regulators. The activation state of an upstream regulator is determined by the regulation direction associated with the relationship from the regulator to the gene.

Availability of data and materials

Transcriptome sequencing data from this study have been deposited in the National Center for Biotechnology Information BioProject database (https://www.ncbi.nlm.nih.gov/bioproject) with the accession code PRJNA791131.

Low expression of USP44 is prognostic for HCC

We began by examining USP44 expression in human HCC by analyzing 40 HCC tumors (T) and paired adjacent nontumoral liver tissue (NT). The quantitative RT-PCR results demonstrated significant downregulation of USP44 in HCC tumors (57.5%) compared with adjacent nonmalignant liver tissue (p < 0.0001, Figure 1A), indicating that USP44 is commonly reduced in HCC. We next conducted a correlative analysis of the USP44 expression in HCC tumors and the clinicopathologic features of the patients. The results showed that expression of USP44 was lower in advanced-stage (stages IV) than in early-stage (stages II and III) HCC tumors (p = 0.049, Table 1), suggesting that USP44 expression is related to tumor staging. In addition, overall survival (p = 0.0471) and disease-free survival (p = 0.0189) were significantly poorer in the low USP44 expression group (median overall survival: 27.31 months; median disease-free survival: 3.98 months) than in the normal and high USP44 expression groups (median overall survival: 52.68 months; median disease-free survival: 28.2 months; Figure 1B). Taken together, our results indicate that low USP44 expression is a prognostic factor in HCC.

Figure 1. Low expression of USP44 is a prognostic factor in hepatocellular carcinoma (HCC). (A) USP44 expression in 40 pairs of HCC nontumoral (NT) and tumoral (T) tissues. USP44 expression was lower in tumoral tissues than in nontumoral tissues (p < 0.0001). (B) Correlation between USP44 expression and survival rate of patients with HCC. Low USP44 expression was related to shorter overall survival (p = 0.047) and disease-free survival (p = 0.019) in patients with HCC. The x-axis indicates survival time in months. The y-axis indicates survival probability. The log-rank p < 0.05 indicates that USP44 expression is significantly associated with prognosis in HCC.

Overexpression of USP44 is associated with reduced proliferation in an HCC cell line

In a gain-of-function study, we used an HCC cell line model to determine the role of USP44 in the carcinogenicity of HCC. The HCC cell line HepG2 has relatively low USP44 mRNA expression as compared to other HCC cell lines including Hep3B, PLC-5, and MHCC97L (Figure 2A). Overexpression of USP44 in HepG2 cells was validated at the gene levels by quantitative RT-PCR (Figure 2B). We next investigated the functional role of USP44 in HCC. Our results showed that the overexpression of USP44 could significantly inhibit the growth of HCC cells (Figure 2C) and arrest the cell cycle at the G0/G1 phase (Figure 2D). In addition, USP44 overexpression reduced the ability of HCC cells to form colonies (Figure 2E). Collectively, these data suggest the importance of USP44 in HCC tumorigenicity.

Figure 2. Roles of USP44 in cell proliferation and the cell cycle in hepatocellular carcinoma (HCC). (A) USP44 expression in various HCC cell lines, including Hep3B, PLC5, MHCC97L, and HepG2. USP44 expression was relatively lower in the HepG2 cell line than in other HCC cell lines. (B) Quantitative real-time PCR analysis validated the overexpression of USP44 after lentiviral infection with either USP44 vector or control vector. (C) The CCK-8 assay showed inhibition of HepG2 cell proliferation after USP44 overexpression. Dashed line = control vector HepG2 cells; solid line = USP44-overexpressing HepG2 cells. (D) Flow cytometry analysis with propidium iodide staining shows cell cycle arrest at the G0/G1 phase in USP44-overexpressing HepG2 cells. (E) USP44 overexpression was associated with reduced colony formation by HepG2 cells. Colony formation was assessed 10 days after 1000 cells were seeded onto a 6-well plate.

USP44 controlled the processes and functions related to HCC carcinogenesis and metastasis

To delineate the molecular mechanisms underlying the role of USP44 in HCC carcinogenicity, a comparative transcriptomic analysis of USP44-overexpressing HepG2 cells and control vector–expressing HepG2 cells was conducted. We obtained at least 40 million qualified sequencing reads from each transcriptome sequencing sample, for a sequencing data total of 51.22 Gb (Table 2). The average mapping to the human genome exceeded 95%. When we compared the transcriptome profiles of USP44-overexpressing HepG2 cells and control cells, we identified 292 DEGs, including 115 upregulated and 177 downregulated genes (Figure 3A and Supplementary Table 1). Gene ontology enrichment analysis highlighted the importance of those genes in biologic processes related to cell proliferation and differentiation through the regulation of cell signaling such as Wnt signaling, TOR signaling, and protein kinase C signaling (Figure 3B). In addition, the USP44-mediated genes were found to control the DNA damage response, drug response, cell apoptosis via the regulation of TRAIL (Figure 3C). Furthermore, the targeted genes play important roles in metastasis and angiogenesis through the regulation of cell migration ability and angiotensin-activated signaling (Figure 3C). More importantly, USP44 could control the immune response via the regulation of T cell and production of interleukins, and mediate different metabolisms such as metabolisms of vitamin D, lipid, reactive oxygen species, and glutathione (Figure 3D). In the analysis of molecular function, we also observed many binding and kinase activities related to cell growth, metastasis, and immune response, including p53 binding, type 2 fibroblast growth factor receptor binding, insulin-like growth factor binding, and death receptor activity (Figure 3E). Taken together, our data suggest that USP44 controls a cluster of genes related to HCC carcinogenicity by modulating cell proliferation, apoptosis, metastasis, and immune functions.

Figure 3. Differential gene expression controlled by USP44 in hepatocellular carcinoma (HCC) cells. (A) A volcano plot shows differential gene expression associated with USP44 overexpression in HepG2 cell. Blue dots = downregulated genes; red dots = upregulated genes; and grey dots = unchanged genes. The x-axis represents differential gene expression in a log2 ratio for the USP44-overexpressing cells compared with the control vector cells. The y-axis represents the q value for differential gene expression, comparing USP44-overexpressing cells with control vector cells. (B) Gene ontology enrichment analysis of USP44-mediated differentially expressed genes highlights the biologic processes related to cell proliferation and differentiation. Bubble diagrams show the biologic processes related (C) to DNA damage response, drug response, cell apoptosis, metastasis, and angiogenesis; (D) to the immune response and metabolism; and (E) to the binding and kinase activities related to cell growth, metastasis, and immune response controlled by USP44 in HCC cells. The size of the bubbles reflects the number of genes being controlled. The color of the bubble reflects the significance of the processes being controlled.

Overexpression of USP44 induces biologic functions and pathways related to DNA damage response and cell apoptosis in HCC

In the IPA conducted to further understand the role of USP44 and its controlled gene networks, we focused on the diseases and biologic functions and the canonical pathways controlled by USP44 in HCC. The results showed that USP44 overexpression could induce biologic functions related to DNA damage, cell apoptosis, and necrosis in HCC (Figure 4A), which is reflected in a positive Z-score. In contrast, USP44 overexpression reduced the proliferation and migration ability of tumor cells in HCC (Figure 4A). In addition, induction of USP44 led to the inhibition of angiogenesis and vasculogenesis in HCC cells (Figure 4A).

Figure 4. USP44 overexpression reduced proliferation and induced apoptosis in hepatocellular carcinoma (HCC) cells. (A) Ingenuity Pathway Analysis (IPA) demonstrated induction of apoptosis-related and reduction of cell proliferation-related diseases and functions caused by USP44 overexpression in HepG2 HCC cells. A positive z score represents activation; a negative z score represents inhibition. (B) IPA gene network construction shows how the functional role of USP44 in HCC involves receptors, enzymes, cytokines, kinases, phosphatases, and transcription factors. Red diagrams = upregulated genes; green diagrams = downregulated genes; orange arrows = activation; and blue arrows = inhibition. (C) qPCR analysis was used to validate the differential gene expression obtained from transcriptome sequencing.

In the IPA, we further investigated the network of canonical pathways controlled by USP44 in HCC. Our results showed the involvement of cytokines such as DKK1, CCN2, EDN1, JAG1, and CCL20 in the functional roles of USP44 in HCC (Figure 4B). The cytokines targeted membrane proteins and receptors including KTN1, FZD2, NGFR, and SFRP5, leading to the control of enzymes such as PLA2G4B, RAC2, RASD1, GCLM, GCLC, ASNS, CYP1A1, and TAT (Figure 4B). In addition, USP44 mediated many transcription factors and cyclins involved in the control of cell proliferation, metastasis, and apoptosis in HCC—for example, NFAT5, EGR1, FOSL1, NFE2L2, NKX2-5, HIF1A, FOS, JUNB, MAFF, SIRT1, BRAC2, FOSB, ATF3, CEBPZ, MEF2B, CCND2, and CCNG2 (Figure 4B). The result of transcriptome was validated by using qPCR, and we found that the results of transcriptome sequencing and qPCR were well-matched (Figure 4C).