The current study was approved by the Ethics Committee of Zhongshan Hospital. Human blood samples for the control (CS) group were collected from healthy volunteers (n=50) who had no obvious signs of disease, as determined by clinical hematology. In the CS group there were 38 males and 12 females, between 45 and 75 years of age. Blood samples for the HCS group were collected from patients with HCC (n=50) who were diagnosed by pathology. In the HCS group there were also 38 males and 12 females, the age range was the same as in the CS group (Table SI). Samples of HCC (HCT group) and adjacent tissue (AT group) (n=10), and histological sections, were supplied by the Hepatocellular Carcinoma Clinical Sample Library (Zhongshan Hospital, Xiamen University). The samples were collected between June 2015 and June 2016. Individuals provided informed consent to participate in the study, according to the requirements of the Ethics Committee of Zhongshan Hospital. Clinical data for the HCC patients involved in the present study are shown in Tables SI and SII.

The current study used carrier ampholyte (pH 5-8; GE Healthcare), dithiothretol (DTT), 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS) and α-cyano-4-hydroxy cinnamic acid (HCCA) were purchased from Sigma-Aldrich; Merck KGaA. Trypsin was obtained from Promega Corporation. The new lysis buffer (lysis buffer N) was composed of 7 M urea, 4% (w/v) CHAPS, 2 M thiourea, 60 mM DTT, 10 mM Tris, 1 mM EDTA, 0.5% (v/v) carrier ampholyte (pH 5–8) and 1% protease inhibitor. The lack of SDS was the principal difference from traditional lysis buffer.

A total of 0.2 ml serum from the CS and HCS groups was collected from volunteers and mixed to identify differential proteins among individuals for proteome separation. An ice-cold mixture serum sample (20 µl serum from each volunteer mixed to 1 ml for detection) was mixed with 9 ml lysis buffer N, stored overnight at −70°C, then thawed and centrifuged at 6,000 × g for 30 min at 4°C. The protein concentration of the supernatants was determined using the Bradford method and the samples were then stored at −70°C prior to 2-DE.

Tissue samples for the HCT and AT groups, containing a mixture of 50 mg from each sample for the different groups, were homogenized in lysis buffer N and placed on ice for 2 h prior to centrifugation at 10,000 × g for 30 min at 4°C. The supernatants were collected and stored at −70°C following determination of the protein concentration.

A high-resolution 2-DE method for separating serum and tissue proteins was performed using the Investigator 2-D Gel system (PerkinElmer, Inc.). The protein sample (150 µg) for each group was loaded on a 14-cm isoelectric focusing strip for the first-dimension electrophoresis, which was performed at a total of 11,000 V via the following stepwise increases: 200 V for 0.25 h; 300 V for 0.5 h; 400 V for 1 h; and 600 V for 16 h. Prior to PAGE, the gel strips were equilibrated for 15 min in equilibration buffer (6 M urea, 30% glycerol, 50 mM Tris-HCl, 2% SDS and 1% DTT) and then for 15 min in the same buffer containing 2.5% iodoacetamide instead of DTT. SDS-PAGE (T=12%) for the second-dimension electrophoresis was performed at 15 mA per gel for ~5 h. For each group, three experimental replicates were subjected to 2-DE.

The silver-stained gels were scanned using the Image Scanner II apparatus (GE Healthcare). ImageMaster 2D Platinum software (version 7.0; GE Healthcare) was employed for digital analysis of the protein spots. Protein spots were detected and compared between the CS and HCS groups, and the HCT and AT groups, and individual spot volume values were obtained, according to the manufacturer's protocol.

The differentially expressed proteins were manually excised from the 2-DE gels. Each protein was washed at room temperature with 50 µl Milli-Q water, twice for 20 min each time. The protein was then immersed in destaining solution [30 mM K₃Fe(CN)₆ and 100 mM Na₂S₂O₃ at a 1:1 ratio] for 2 min and washed at room temperature with 100 µl NH₄HCO₃ (25 mM) containing 50% acetonitrile (ACN), twice for 20 min each time. Subsequently, the gel spots were dehydrated using 100% ACN for 10 min, dried in a vacuum centrifuge for 10 min, and rehydrated with 10–20 µl trypsin solution in an ice-bath for 30 min, followed by incubation at 37°C for 16 h. The supernatant was collected in new tubes, and the extracts were pooled and dried in a vacuum centrifuge (Eppendorf). Finally, 2.5 µl 0.5% trifluoroacetic acid (TFA) dissolved in 60% ACN was added to each protein sample to dissolve the protein powder for MALDI-TOF/TOF MS analysis.

Each protein solution was mixed with matrix solution (HCCA saturated in 60% ACN and 0.1% TFA) at a 1:1 ratio. The mixture (1 µl) was then spotted onto the MALDI target and analyzed using a REFLEX III MALDI-TOF/TOF MS system (Bruker Corporation). The peptide spectra were recorded in the reflector mode and calibrated with acidic peptides [residues 26; M (H⁺)=2,961.5] and angiotensin I (1,296.5 Da). The accuracy of the molecular ion mass was >±0.1 Da up to a mass of 6,000 Da. All other settings were as previously published (9). Peptide mass fingerprinting (PMF) of selected proteins was performed using the MASCOT database (http://www.matrixscience.com/search_form_select.html). The maximum tolerance for masses was adjusted to 40 ppm by internal calibration. Mr and pI values of the analyzed spots were obtained from the 2-DE gel. The SWISS-PROT (https://www.expasy.org/) databases were used for the identification of significant proteins.

TRIzol^® reagent (Thermo Fisher Scientific, Inc.) was used to extract RNA from the HCC tissue samples and adjacent tissue samples. Complementary DNA was obtained using the Takara PrimeScript™ RT-qPCR kit (Takara Bio, Inc.) as follows: 65°C for 5 min, rapid cooling on ice, 30°C for 10 min, 42°C for 45 min, 95°C for 5 min and cooling on ice. RT-qPCR was performed with the SYBR^® Premix Ex Taq™ II (Perfect Real Time) kit (Takara Bio, Inc.), according to the manufacturer's protocol. The housekeeping gene β-actin was used as the internal control. The primers used in the current study are listed in Table I. RT-qPCR was performed using an Applied Biosystems FAST 7500 real-time PCR system (Thermo Fisher Scientific, Inc.) with the following steps: Amplification was initiated with pre-denaturing at 95°C for 30 sec, denaturing was performed at 95°C for 5 sec, annealing and extension were performed at 62°C for 35 sec over 40 cycles during the second stage, and there was a final stage of 60–95°C to determine the dissociation curves of the amplified products. The 2^−ΔΔCq method (10) was employed to perform the analysis of differential gene expression. A Student's t-test was used for statistical analysis.

IHC staining was performed on 4 µm-thick HCC tissue sections and adjacent tissue sections. The slides were rehydrated in dimethylbenzene and a descending alcohol series. After washing three times with double distilled H₂O, citrate buffer (0.01 mol/liter, pH 6.0) was used for antigen retrieval by boiling (≥95°C) the slides in a microwave oven for 10 min, after natural cooling to room temperature, slides were washed twice using double distilled H₂O and twice using PBS buffer at room temperature. Then 10% BSA (Sigma-Aldrich; Merck KGaA) blocking solution was used and incubated for 30 min at 37°C. Apolipoprotein E (APOE) rabbit monoclonal antibody (cat. no. CY5573; Abways Technology, Inc.) was applied at 1:100 in PBS for 2 h at room temperature. A universal secondary antibody (EnVision™+/HRP rabbit/mouse; DAKO; Agilent Technologies, Inc.) was then applied for 30 min at 37°C. Diaminobenzidine was used for chromogen detection. Slides were incubated with diaminobenzidine for 5 min at room temperature and after washing for 2 min with water, slides were counterstained with hematoxylin for 40 sec and hydrochloric acid alcohol solution [70% alcohol:HCl at a ratio of 100:1 (v:v)] for 2 sec at room temperature. Dehydration was achieved using an increasing alcohol series and dimethylbenzene. Slides were sealed using neutral balsam and a coverslip. A light microscope (Diaphot 300, Nikon Corporation) was used to observe the sections after staining.

Survival analysis of differentially expressed proteins in liver cancer was performed using the Kaplan-Meier Plotter mRNA/RNA-seq database of liver cancer (http://kmplot.com/analysis/index.php?p=background). The survival analysis was carried out using the Kaplan-Meier method on 364 patients, and hazard ratios with 95% confidence intervals and log-rank P-values were calculated. Pathway analysis of identified proteins was performed using the Search Tool for Interactions of Chemicals (STITCH; http://stitch.embl.de). The functional properties and subcellular localization of the proteins were determined using the UniProt (www.uniprot.org/) and PSORT (https://db.psort.org) databases, respectively.

The data are expressed as the mean ± SD of triplicate experiments. Significant differences among groups were determined using one-way ANOVA with SPSS version 19 (IBM Corp.) followed by the LSD post-hoc test. P<0.05 was considered to indicate a statistically significant difference.

The serum sample in the lysis buffer was separated using 2-DE gel at a pH of 5–8. A total of 150±5 µg serum protein was considered optimal for separation using 2-DE. The new lysis buffer was conducive to obtaining more protein spots (Fig. S1). As demonstrated in Fig. 1A and B, ~800 proteins were visualized. Fig. 1C and D demonstrates the 2-DE results for the HCT and AT groups, in which almost 1,200 proteins were revealed by sliver staining in each tissue gel.

Quantitative image analysis based on the percentage of the spots revealed that certain proteins were either significantly upregulated or downregulated between the two groups (Fig. 2). Compared with the CS group, the expression levels of 24 proteins were significantly different in the HC group (Fig. 2A). Furthermore, 12 proteins were revealed to be differentially expressed in the AT and HCT groups (Fig. 2B).

MALDI-TOF/TOF MS, PMF and a database search were used to evaluate the proteins identified in Fig. 2. MALDI-TOF/TOF MS can identify small amounts of protein in 2-DE gels. The differentially expressed proteins were digested by trypsin in the gel to further assist with the identification of the proteins (Table II). Of the 24 differentially expressed proteins between the CS and HCS groups, seven were identified. Additionally, all 12 differentially expressed proteins in the AT and HCT groups were identified. Biochemical parameters based on the results of the database search are presented in Table II, including accession number, protein name, score, Mw and PI.

The ratio of protein spots compared with the whole gel and the spot grey value may demonstrate the expression level of the corresponding protein. Taking the CS or AT groups as the control, the relative intensity ratio of the HCS and HCT groups revealed significant changes in the expression levels of certain proteins, which are presented in Table II. A total of three upregulated proteins were identified (spot nos. 4, 7 and 9) and four downregulated proteins were revealed (spot nos. 4, 12, 15 and 16) in the HCS group compared with the CS group. For the tissue groups, seven upregulated proteins were identified (spot nos. 4, 5, 6, 8, 10, 11 and 12) and five downregulated proteins were revealed (spot nos. 1, 2, 3, 7 and 9) in the HCT group compared with the AT group.

RT-qPCR analysis was performed to assess the extent of the changes in protein expression levels in the 2-DE results. A total of seven differentially expressed genes from the CS and HCS groups, and twelve genes from the AT and HCT groups (Table II) were selected for this analysis. The gene expression levels of the proteins identified in the CS/HCS groups were analyzed in HCC tissues and corresponding adjacent tissues by RT-qPCR, and the results are demonstrated in Fig. 3A and Table SIII. The mRNA expression levels determined by RT-qPCR for ten samples of each group were relatively consistent with the protein expression levels revealed by 2-DE (Figs. 1A, B and 2A; Table II). The expression level of one gene, oxysterol binding protein-like 11 (OSBPL11), was revealed to be lower in HCC tissues compared with adjacent tissues in all 10 samples; this difference was identified to be significant in nine of the samples, and the difference between the HCT and AT groups reached a highly significant level (P<0.001; Table SIII). This result was consistent with the results of 2-DE. In addition, the expression levels of six other genes, including heparan sulfate (glucosamine) 3-O-sulfotransferase 3A1 (HS3ST3A1), tubulin βD, the RAS oncogene family, apolipoprotein A-D, 60S acidic ribosomal protein P0 and cyclin H, were partly consistent with the results of 2-DE. These results further confirmed the validity of the proteomic approach for the identification of serum biomarkers in HCC.

The expression levels of 12 differentially expressed genes identified in the AT and HCT groups (Figs. 1A, B and 2; Table II), were also evaluated using RT-qPCR in a total of ten tissue samples (Fig. 3B; Table SIV). Among the 12 genes investigated, the expression levels of 11 genes were revealed to be significantly higher in the HCT group compared with the AT group. By contrast, the expression level of the APOE gene was identified to be lower in the HCT group compared with the AT group. The gene expression levels in the HCT group identified by RT-qPCR were not completely consistent with the protein changes identified by 2-DE. A total of six spots (nos. 4, 5, 6, 8, 11 and 12) demonstrated similar expression levels in 2-DE compared with RT-qPCR for the AT and HCT groups.

As presented in Fig. 4A and Table III, the identified proteins were predominantly localized in the cytoplasm, nucleus, Golgi, endosomes, endoplasmic reticulum, lysosomes and extracellular space. The statistical analysis was not performed separately for the proteins identified in the CS/HCS groups and the AT/HCT groups. In total, 41.67% of the proteins were located the in cytoplasm, 16.67% were located in the nucleus, 12.5% in the Golgi, 8.33% in endosomes, 4.17% in lysosomes or endoplasmic reticulum, and 12.5% were identified to be located in the extracellular space as secretory proteins.

In total, the proteins were revealed to be enriched in 19 molecular function terms (Fig. 4B), including ‘sulfotransferase activity’, ‘GTPase activity’, ‘ferroxidase activity’, ‘kinase activity’, ‘oxidoreductase activity’, ‘inorgnesium diphosphatase activity’, ‘ribosome binding’, ‘ion binding’, ‘GTP binding’, ‘RNA binding’, ‘U6 snRNA binding’, ‘lipid binding’, ‘protein binding’, ‘structural constituent of ribosome’, ‘structural molecule activity’ and ‘antioxidant’. Comprehensive analysis revealed that 63.64% of the functions were associated with protein binding, 27.27% were associated with enzyme activity and 6.06% were associated with structure. The enriched biological processes varied markedly and included the following: Biosynthetic process, cell cycle, signal transduction, transport, metabolic and immune response (Table III).

Pathway analysis of the differentially expressed proteins in the CS/HCS and AT/HCT groups was performed using the STITCH database, and the protein-protein network is presented in Figs. 5 and S2; the network was not generated separately for the proteins identified in the CS/HCS groups and the AT/HCT groups. The network indicates a strong interaction among members of the RAS oncogene family and the APO family, the expression of which had been detected in HCC tissue (Fig. S2). Additionally, an association was revealed between these two families via the proteins ferritin heavy polypeptide 1 (FTH1) and RAB35.

The differentially expressed proteins OSBPL11 and APOE, detected by 2-DE, were selected to evaluate their association with survival in liver cancer (Fig. 6). Compared with a low expression level, it was revealed that high expression of OSBPL11 was associated with a higher survival probability in liver cancer. A significant difference (P=0.036) was identified in the 5-year survival rate for liver cancer cases with high OSBPL11 expression compared with low expression (Fig. 6A). However, no significant difference was identified for the 10-year survival rate (Fig. 6B). Furthermore, the survival rate for patients with liver cancer with a high APOE expression level was higher compared with patients with a low APOE expression level. A significant difference was identified in the 5-year survival rate (P=0.0047; Fig. 6C) and the 10-year survival rate (P=0.0026; Fig. 6D) for patients with liver cancer with a high APOE expression level compared with patients with a low APOE expression level.

The results of IHC staining to detect APOE expression in HCC and adjacent tissues of different liver cancer grades are presented in Fig. 6E. The expression of APOE was higher in HCC tissues compared with corresponding adjacent tissues in samples obatined from patients with grade I–III liver cancer. Additonally, the expression level of APOE increased with increasing tumor grade, both in HCC tissues and adjacent tissue samples.