Cell lines used in this study were obtained from the Cancer Cell Bank of the Chinese Medical Science Institute. BEL7402/5-FU is an MDR HCC cell line that was induced by 5-fluorouracil (5-Fu) from the human HCC cell line BEL7402. RPMI-1640 medium and fetal bovine serum (FBS) were purchased from Gibco (Grand Island, NY, USA). The complete medium for BEL7402 and BEL7402/5-FU consisted of 89% RPMI-1640 medium, 10% FBS and 1% penicillin-streptomycin solution. By adding 20 mg/l 5-Fu to the complete medium, the BEL7402/5-FU cell line maintained its biological characteristics of MDR. Cell lines were cultured in a water incubator (Thermo Fisher Scientific, Rockford, IL, USA) at 37°C and 5% CO₂.

The cytotoxicity of chemotherapeutic drugs to BEL7402 and its MDR cell line were measured by a WST-8 assay (5). BEL7402 and BEL7402/5-FU cells were counted and placed in 96-wells at a density of 5×10³ cells/well, and then cultured in an incubator at 37°C for 24 h before being treated with chemotherapeutic drugs. 5-Fu, cisplatin, adriamycin and vincristine (VCR) in graded concentrations were added to the two types of cells. After treatment for 48 h, the medium was replaced with fresh medium containing 10% WST-8 reagent, and the cells were incubated for an additional 1–4 h until their optical density was measured at 450 nm. The IC50 of BEL7402 and BEL7402/5-FU for each drug was analyzed using IBM SPSS Statistics 21 software (SPSS Inc., Chicago, IL, USA) by means of probit analysis (6).

The secretory proteins of BEL7402 and BEL7402/5-FU were extracted through ultrafiltration. When BEL7402 and BEL7402/5-FU were 90% confluent, the culture media were replaced with basic RPMI-1640 medium after the residual serum was washed away with PBS. Twenty-four hours after instituting this serum-free culture, the media of each cell line were collected and concentrated by centrifugation at 4°C for 1 h using centrifugal filter devices from Amicon (Carrigtwohill, Co., Cork, Ireland). The resulting concentrated solution was removed and placed into a tube and precipitated overnight with acetone at 20°C. Subsequently, the precipitated samples were separated by high-speed centrifugation, and then dissolved in a buffer containing 0.5 M triethylammonium bicarbonate and 8 M urea (urea TB buffer). Protein samples were stored at −80°C prior to isotope labeling.

Secretory protein samples were quantified using a 2-D Quant kit purchased from Amersham Biosciences, and diluted with the urea TB buffer to obtain a final concentration of 5 mg/ml. A 100 µg protein/sample of BEL7402 and BEL7402/5-FU was employed for isotope labeling, and a duplex was set for the HCC cell and its MDR phenotype, respectively. Samples in each tube were denatured and cysteine-blocked using buffers in the iTRAQ Buffer kit (AB Sciex, Foster City, CA, USA). Having been digested with 5 µg trypsin (Promega, Madison, WI, USA) at 37°C overnight and desiccated using a vacuum pump, the peptides were dissolved in isopropanol and then subjected to iTRAQ labeling. Isotope tags were allocated as follows: i) BEL7402 secretory protein −113 and −115 tags; and ii) BEL7402/5-FU secretory protein −114 and −116 tags. The labeled samples were mixed together for further mass spectrometric (MS) analysis.

Peptides labeled with isotope tags were fractionated as previously described (7). Briefly, the fractionation process is conducted through isoelectric focusing (IEF). Labeled samples were dissolved in Pharmalyte and urea solution, subjected to pH 3–10 IPG strips, and then focused on the IPGphor. The IPGphor IEF system was purchased from Amersham Biosciences (Uppsala, Sweden). Peptides were extracted from the gel using buffer A (2% acetonitrile and 0.1% formic acid) and lyophilized in a vacuum pump. The fractions were desalinated using a solid phase extraction (SPE) column (Supelco, Bellefonte, PA, USA) and lyophilized again. The fractions were then stored at −20°C prior to the MS analysis.

MS analysis was performed with the QStar^® Elite hybrid mass spectrometer (Applied Biosystems, Foster City, CA, USA) coupled to a liquid chromatography system (Amersham, The Netherlands). Lyophilized samples were re-suspended in buffer A, and 10 µl was loaded on a capillary column for peptide separation. Peptides were separated by eluting with buffer A and a series of buffer B (98% acetonitrile containing 0.1% formic) gradients at a 0.3 µl/min flow rate. The LC eluent was applied to the electrospray ionization (ESI) quadruple time-of-flight mass spectrometer (QTOF-MS) analysis. The mass spectrometer was operated in positive ion mode, with the mass range set at 300–2,000 m/z. The two most intensely ionized peptides >20 counts were selected for tandem mass spectrometry at a dynamic exclusion for 30 sec with a tolerance of ± 50 mDa.

MS analysis data were processed using the ProteinPilot version 2.0 software (Applied Biosystems) and searched against the International Protein Index (IPI) human protein database (v3.77). Cysteine modified by methane thiosulfate (MMTS) was specified as a fixed modification. Relative quantification of proteins was determined using ratios of the peak areas at 113, 114, 115 and 116 Da, respectively. These values corresponded to the abundance of the tags applied to label the samples.

To verify the differentially expressed secretory proteins identified through MS analysis, western blotting was subsequently carried out. Figures for the total proteins of BEL7402 and its MDR cell line were calculated by lysing cells with a lysis buffer as previously described (9). Concentrations of each protein sample were determined using the 2D-Quant kit. Equal amounts of intercellular and secreted proteins were separated by SDS-PAGE electrophoresis and then transferred onto polyvinylidene fluoride (PVDF) membranes. The membranes were subsequently blocked with blocking buffer [5% milk, 3% BSA or 0.04% gelatin in tris-buffered saline and Tween-20 (TBS-T)] at room temperature for 1 h. The blocked membranes were incubated with specific primary antibodies (1:1,000–1:10,000 dilution in blocking buffer) overnight at 4°C. Primary antibodies, including polyclonal and monoclonal against α-2-HS-glycoprotein (AHSG), angiopoietin 1 (ANGPT1), vitronectin (VTN), SerpinF1, NAD(P)H dehydrogenase (quinone) 1 (NQO1) and lipocalin-2 (LCN2), were purchased from Abcam (Cambridge, MA, USA). Species-specific horseradish peroxidase-conjugated secondary antibodies were applied to incubate the membranes, which were previously washed three times in TBS-T. Reactions between the target proteins and specific antibodies were then visualized by enhanced chemiluminescence (ECL) reagent (GE Healthcare, Buckinghamshire, UK) through the ChemiDOC^™ MP imaging system (Bio-Rad, Hercules, CA, USA).

Three AHSG siRNA sequences (5′-ACACCUUCAUGGGUGUGGUCUCAUU-3′, 5′-ACGCUCAGAACGGCUCCAAUUU-3′ and 5′-UCGUCCUGCUCCUUUGUCUUGCUCA-3′) and negative control siRNA purchased from Invitrogen-Life Technologies (Carlsbad, CA, USA) were each transfected into BEL7402/5-FU cells following the manufacturer's instructions. BEL7402/5-FU cells were evenly planted into 6-well plates and cultured until they were 60–80% confluent. AHSG siRNAs and negative control siRNA were incubated with the Lipofectamine^® 2000 Transfection reagent (Invitrogen-Life Technologies) for 20 min in Opti-MEM^® I Reduced-Serum Medium (Gibco) prior to being transfected into BEL7402/5-FU cells. The final concentration of the siRNAs was 40 nmol/l, which corresponded to the optimal concentration as measured in a pre-test. The medium in each well was replaced with RPMI-1640 complete medium 4–6 h after siRNA transfection and the cells were allowed to continue in culture for a further 48 h. The apoptotic rate was detected by Annexin V-FITC/PI using flow cytometry (FCM). The viabilities of RNA-interfered cells were examined using WST-8 following treatment with 5-Fu, adriamycin, VCR and cisplatin, respectively, in their IC50 concentrations for 48 h. Simultaneously, AHSG protein expression of knockdown cells was determined by western blotting.

Immunohistochemistry (IHC) was performed on the tissue microarray (TMA) LV801a obtained from US Biomax, Inc. (Rockville, MD, USA). The TMA contained 40 cases of HCC and an equal amount of adjacent normal tissue. The IHC was performed as previously described (7). Paraffin-embedded HCC tissue sections were deparaffinized and rehydrated, and antigen retrieval was performed with 0.01 M citrate for 5 min. Endogenous peroxidase activity was quenched in 3% H₂O₂ for 10 min. Tissue sections were then blocked with BSA, and incubated with specific monoclonal antibodies for AHSG (1:20), ANGPT1 (1:20), VTN (1:100), NQO1 (1:100), LCN2 (1:200) and SerpinF1 (1:400). The reactions were visualized by incubating with diaminobenzidine solution (DAB), counterstained with hematoxylin and evaluated under a light microscope. The protein expression levels were semi-quantitatively analyzed based on the percentage of positive cells, ranging from 0 to 100%, and the staining intensity was counted as 0, 1, 2, or 3. The arithmetic product of the two parameters was generated as the IHC score value. All evaluations were performed by the same pathologist to maintain consistency.

BEL7402 and HepG2 cells were transfected with negative control siRNA and AHSG siRNA sequences as previously mentioned. For the proliferation assay, transfected BEL7402 and HepG2 were added into 96-well plates at a density of 1×10³ cells/well, and the number of viable cells was monitored at regular intervals by WST-8. Cell migration and invasion assays were carried out in 24-well plates using polycarbonate membrane inserts (Cell Biolabs, Inc.) with 8 µm pores. For the cell invasion assay, an additional uniform basement membrane matrix layer was coated on the upper surface of the chambers to obstruct non-invasive cells. Transfected cells were re-suspended with serum-free media at a concentration of 1.0×10⁶ cells/ml and plated into the polycarbonate membrane chambers. Medium containing 10% FBS was added to the bottom of each well. After 24-h incubation at 37°C, any cells on the surface of the chambers that had not migrated or invaded were removed. The migrated and invaded cells on the polycarbonate membrane were stained with cell stain solution (Cell Biolabs, Inc.), dissociated with extraction solution, and the optical density was quantified at 560 nm.

The continuous variables in this study were presented as mean ± standard deviation (SD). Statistical analysis of data in this study was conducted by SPSS Statistics 21 software. The IC50 values of each chemotherapeutic drug were calculated by the probit analysis (8). Intergroup significance tests were conducted using a Student's t-test and the data were considered statistically significant when P<0.05.

The biological characteristics of BEL7402 and its MDR phenotype cell line BEL7402/5-FU have been previously described (9). The state of the cells was slightly different from that of the previous study, since they were of a different generation. Following the protocol mentioned above, dose-response relationships for 5-Fu, cisplatin and adriamycin were obtained and analyzed by SPSS software using the probit model, thereby determining the IC50 values. As is shown in Table I, the IC50 values of 5-Fu for BEL7402 and BEL7402/5-FU were 3.46±0.31 and 697.57±52.88, respectively. Thus, the resistance to 5-Fu of BEL7402/5-FU was 201.61-fold higher than that for BEL7402 (Fig. 1). Differences were also observed following treatment with cisplatin, adriamycin, and VCR, which amounted to 11.74-, 29.68- and 8.75-fold higher in BEL7402/5-FU cells, respectively (Table I and Fig. 1). The results indicated that BEL7402/5-FU was resistant to several types of chemotherapeutic drugs, and served as a typical model in the MDR study of HCC.

Secretory protein samples were enriched from a cell culture medium of the human BEL7402 HCC cell line and its MDR cell line BEL7402/5-FU, and subjected to iTRAQ-based MS analysis after being labeled with iTRAQ tags. A brief flowchart of the quantitative proteomic approach using isobaric labeling is presented in Fig. 2A. A duplicate of each sample was created to increase confidence in the result, and protein samples were labeled as follows: BEL7402, 113 and 115; BEL7402/5-FU, 114 and 116. Thus, the relative abundance of proteins from BEL7402/5-FU cells with respect to proteins from BEL7402 cells was revealed as the iTRAQ ratios 114:113 and 116:115. The isobaric-labeled peptides were analyzed by LC/MS/MS, searched for in the IPI database, and identified with ProteinPilot software. By searching the human IPI database under the ProteinPilot algorithm, 1,780 unique proteins were identified with 95% confidence. Considering the technical variations of the method and statistical analysis in the relative quantification analysis, and in order to reduce false positives and increase accuracy, stricter cut-off points were imposed when dividing the identified proteins. Proteins with a ratio >1.5 or <0.5 were classified as upregulated or downregulated (10). In total, 279 differentially expressed proteins were filtrated, of which 131 proteins were highly expressed in BEL7402/5-FU relative to its parental cell line, while 56 proteins were underexpressed. Table II shows the top 30 upregulated and downregulated proteins in BEL7402/5-FU cells compared to BEL7402 cells. Of the downregulated proteins, aldo-keto reductase (AKR) (peptide sequence: ALGVSNFSHFQIEK) ranks in the top 30, and its MS/MS spectrum and relative abundance by iTRAQ ratio 116:115 and 114:113 are shown in Fig. 2B.

To achieve a preview of the profile of the 279 differentially expressed proteins, the proteins were classified by the PANTHER classification system (http://www.pantherdb.org) according to protein class, cellular component, molecular function and biological process (Fig. 3). The differentially expressed proteins were various and divided into 28 classes, but mainly comprise nucleic acid binding (12.9%), hydrolase (11.7%), enzyme modulator (6.8%) and signaling molecule (6.1%) proteins (Fig. 3A). There were 10 molecular function categories, including catalytic activity (38.8%) and binding (31.1%) (Fig. 3B). The identified proteins participate in a range of biological processes, including metabolic processes (33.1%) and cell processes (17.9%) (Fig. 3C).

By studying protein classification and screening the relative abundance of proteins through the iTRAQ approach, the most significantly dysregulated proteins, AHSG, ANGPT1, VTN, SerpinF1, NQO1 and LCN2 were further validated. Western blotting was performed as described above to validate the identified differentially expressed proteins. The expression of AHSG, ANGPT1, VTN, SerpinF1, NQO1 and LCN2 in the conditioned medium and cell lysis product of BEL7402 and BEL7402/5-FU cells is shown in Fig. 4. The abundance of proteins AHSG, ANGPT1, VTN and SerpinF1 in secretory protein and total cytoplasmic protein were obviously higher in the BEL7402/5-FU than in BEL7402 cells, while NQO1 and LCN2 were downregulated. Thus, the results of immunoblotting were highly coincidental with the mass spectrum analysis.

To study the relevance of AHSG upregulation to the MDR phenomenon in the BEL7402/5-FU cell line, three specific siRNA sequences were employed to knock down AHSG expression in BEL7402/5-FU cells. By western blotting, AHSG expression in specific siRNA-transfected BEL7402/5-FU cells were proven to have been effectively silenced (Fig. 5A). To assess the relationship between AHSG and MDR, BEL7402/5-FU cells were treated with the aforementioned anticancer drugs for 2 days with their IC50 concentration after being transfected with specific siRNA sequences for 48 h. Fig. 5B shows that the inhibition rate of anticancer drugs 5-Fu, cisplatin, adriamycin and VCR on AHSG-knockdown BEL7402/5-FU cells was higher, compared with negative control. IC50 values of these drugs to AHSG-knockdown cells are detailed in Table III, showing that IC50 values of 5-Fu, cisplatin and adriamycin were markedly decreased in AHSG-knockdown BEL7402/5-FU cells than negative control. The result suggested that suppression of AHSG minimizes BEL7402/5-FU resistance to several chemotherapeutic drugs, which, in turn, indicated that AHSG may contribute to the MDR phenomenon in the cell line. Furthermore, through FCM detection, we found that AHSG-knockdown has no effect on the apoptosis of BEL7402/5-FU under the pressure of chemotherapeutic drugs (data not shown).

The selected differentially expressed proteins in the present study have been reported be linked to tumor development and progression in several types of cancer (11–14). However, their association with HCC is, by contrast, rarely reported. Moreover, chemoresistance phenotype is a common biological behavior in malignancy, while drug resistance may not be an isolated event in tumor development and progression. Thus, TMA containing 40 HCC and adjacent tissues were employed for the immunohistochemical assay. Fig. 6A shows the expression of AHSG, ANGPT1, VTN, SerpinF1, NQO1, and LCN2 in HCC, and adjacent tissues. The proteins AHSG, ANGPT1 and VTN were significantly upregulated in HCC tissues compared to adjacent tissues, and the semi-quantitative analysis described above shows the differential expression visually, as well as the statistical significance (Fig. 6B). The differential expression of SerpinF1 and NQO1 was observed in only a few locations, and there was no significant difference in the expression of SerpinF1 and NQO1 between HCC tissue and the adjacent normal tissue. The IHC results of AHSG, ANGPT1 and VTN were coincident with MS/MS analysis of the MDR cell line, whereas LCN2, which was a markedly underexpressed protein in BEL7402/5-FU cells, was overexpressed in HCC tissues compared to adjacent tissues.

The experimental data on TMA, coupled with the known information on AHSG, suggests that AHSG may not only be connected with the MDR of HCC, but also contributes to the progression of the cancer. In light of this, the expression of AHSG in human HepG2 and BEL7402 HCC cell lines were knocked down by its specific siRNAs to prepare the cell proliferation, migration and invasion assays. By measuring the OD values following incubation with media containing WST-8 reagent, the relative amounts of viable cells at each specified time interval are shown in Fig. 7, showing that AHSG-knockdown cells, HepG2 and BEL7402, exhibited fewer cells in contrast to the control cells over time. In the cell migration and invasion assays, however, silencing of AHSG had little or no effect on the migration or invasion process of HepG2 and BEL7402 cells (data not shown).