1 INTRODUCTION

Sorafenib is the first approved oral multi-kinase molecular inhibitor for the treatment of advanced hepatocellular carcinoma (HCC). Two anticancer mechanisms of sorafenib to inhibit proliferation and angiogenesis of HCC include targeting the RAF/mitogen-activated protein kinase (MAPK) (also known as extra-cellular signal-regulated kinase, ERK) kinase (MEK)/ERK signaling pathways and tyrosine kinase receptors.1-4 Despite the successful approval and extensive application of sorafenib in the treatment of advanced HCC, only a limited number of patients have experienced a real and long-term benefit.5 Different characteristics of cancer cells have been proposed to explain the sorafenib resistance in HCC.6-10 However, there is still no conclusive marker to predict sorafenib efficacy in patients with advanced HCC.5, 11

Cancer stem cell or cancer cell with stem cell-like properties is considered as an important cause of tumor recurrence, metastasis, and failure of chemotherapy or radiation therapy.12, 13 NANOG is a DNA binding homeobox transcription factor and POU5F1 (also known as OCT3/4) is a transcription factor containing a POU homeodomain. Both genes play key roles in embryonic stem cell proliferation, renewal, and pluripotency and are considered as markers for cancer stem cells.12, 13 The purpose of this study was to investigate the associations of NANOG and POU5F1 expressions with the anti-cancer effects of sorafenib in HCC. Since the purchased HCC cell lines do not actually represent the clinical situations,14 primary cultured HCC cells obtained by the previously established method were applied.15 On the other hand, both transduction of lentiviral promoter to force expression of target gene and short hairpin RNA-mediated plasmid to knockdown gene expression are well established methods to understand the function of target gene. However, the present study did not apply these methods for investigation. The major concern was whether the magnitudes of these artificial expressions or repressions truly exist in real-world or not. Therefore, flow cytometric sorting from primary cultured cells to establish subgroups with different expressions of NANOG and/or POU5F1 rather than using artificial interference on gene expression was applied. All subgroups were originated from the existed cancer cells of the patients. Although analyzing the pharmacokinetics and pharmacodynamics of a drug can be influenced by different techniques applied and drug-drug or drug-protein interaction,16 the reported peak plasma concentrations after oral administration of clinical sorafenib doses in HCC patients with Child class A or B were around 5 to 10 μg/mL.3, 17-19 Therefore, 5 and 10 μg/mL two fixed sorafenib concentrations were applied for experiments. The results obtained from this unique study design could have the most possibility to mimic the clinical situation. All gene names described in this study were according to the official symbols from the HUGO Gene Nomenclature Committee.

2 MATERIALS AND METHODS 2.1 Parent HCC cell lines and sorafenib

Eight HCC parent cell lines (s003, s015, s020, s035, s058, s068, s071, s089) originated from primary culture of residual specimens of fine-needle aspiration of hepatic tumor in eight patients were applied for investigation (Table 1). The purpose for fine-needle aspiration was to diagnose the hepatic tumor. The aspirated specimens in all patients were diagnosed as HCC based on cytological and/or pathological findings. The detail methodology for the collection of residual specimens for primary culture was based on the previous study.15 Cultured HCC cancer cells were verified by positive stain for mouse anti-human hepatocyte antigen (Clone OCH1E5, Santa Cruz Biotechnology, Inc., Dallas, TX). All parent cell lines applied for the study were at the 4th to 6th passage in primary culture. All cultures were maintained at 37°C in a humidified atmosphere of 5% CO2 and 95% air. The serum-containing culture medium consisted of 10% fetal bovine serum, 90% DME/HIGH glucose, supplemented with 20 mM L-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin (HyClone, Logan, UT). Sorafenib was purchased from MedChem Express (Monmouth Junction, NJ) and dissolved in Dimethyl sulfoxide (DMSO), which was further diluted in culture medium to give 5 and 10 μg/mL concentrations resulting in a DMSO concentration equal to 1% (v/v) after addition to the cells. The same concentration of DMSO was applied to the control group. This study was approved by the Institutional Review Board of the hospital and all patients signed the informed consent.

TABLE 1. Characteristics of parent cell lines and the gene expression fold changes of NANOG and POU5F1 in subgroups Cell line Etiology HCC grading† AFP (ng/mL) TNM staging‡ NANOG/POU5F1 expressions s003 HCV 3 567.92 IIIB s003 NUPS 1.36/1.19 s015 HBV 3 4.5 II s015 NUPS 1.44/0.97 s015 NSPU 1.3/2.10 s015 NDPS 0.57/1.17 s020 unknown 2 3 IIIA s020 NSPU 0.85/5.21 s020 NDPU 0.28/3.17 s035 unknown 3 74.2 IB s035 NDPU 0.37/1.38 s058 HBV 2 49.36 IVB s058 NDPD 0.59/0.53 s068 HCV 2 3 II s068 NSPU 1.11/1.58 s071 unknown 3 151730 IVB s071 NUPD 2.37/0.25 s071 NUPS 1.82/0.84 s089 HCV 1 1149.33 IVB s089 NSPU 0.83/1.38 s089 NUPS 1.54/1.09 Note: The parent cell line was applied as control for comparison of gene expression in subgroup. Gene expression fold change was determined by quantitative reverse transcriptase-polymerase chain reaction. The value of gene expression fold change >1.3 was defined as up-regulation and <0.7 was defined as downregulation as compared with control. Abbreviations: AFP, alpha-fetoprotein (normal upper limit <20 ng/mL); D, downregulation; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; N, NANOG; P, POU5F1; S, same gene expression; U, upregulation. † Histological grading of HCC was based on Edmondson–Steiner grading system. ‡ The American Joint Committee on Cancer TNM staging system 8th edition was applied for tumor staging. 2.2 Establishment of HCC subgroups

Eight parent cell lines were undergone flow cytometric sorting using NANOG and POU5F1 (Thermo Fisher Scientific Inc., Waltham, MA) as targets. The collected cells were expanded and verified by quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR). A total of 13 subgroups were established (Table 1). The abbreviations for the names of these subgroup were defined as N (NANOG), P (POU5F1), U (upregulation), D (downregulation), and S (same expression).

2.3 Proliferation experiments

Cancer cells seeded in 96-well culture plate were incubated with serum-containing medium for 48 h. Then the medium was replaced with serum-free medium with various concentrations of sorafenib (0, 5, 10 μg/mL). The cells were incubated for further 72 h for proliferation assay. The premixed WST-1 cell proliferation reagent (Clontech Laboratories, Inc., A Takara Bio Company, Mountain View, CA) was applied. The experimental procedures were carried out following the manufacturer's protocols. The results were analyzed by an automated microplate reader (Synergy H1 Hybrid Multi-Mode Reader with Gen5 software, BioTek Instruments, Inc., Winooski, VT). The absorbance was measured at 450 nm wavelength (reference wavelength 630 nm). In all experiments, 16 replicated wells were used for statistical calculation.

2.4 qRT-PCR experiments

Cancer cells plated in 25-cm2 plastic culture flasks were incubated with serum-free medium with various concentrations of sorafenib (0, 5, 10 μg/mL) for 24 h for RNA extraction. The procedures to extract and purify RNA, validate specific oligonucleotide primer pairs, and calculate the fold expression or repression of the target gene were the same as previous studies.14, 20 The fold expression or repression of the target gene relative to the reference gene in each sample was calculated by the formula: gene expression fold change = 2−≥≥CP, where ≥CP = CPtarget gene–CPreference gene, and ≥≥CP = ≥CPtest sample–≥CPcontrol sample. The CP (crossing point, the turning point corresponds to the first maximum of the second derivative curve) value implies the beginning of exponential amplification. Each gene received triple qRT-PCR determinations to obtain the mean value for analysis. The value of gene expression fold change >1.3 was defined as up-regulation and <0.7 was defined as down-regulation. Ten genes including BRAF, KDR, MAPK1, MAPK3, MAP2K1, MAP2K2, NANOG, PDGFRB, POU5F1, and RAF1 were selected for qRT-PCR study. The housekeeping gene TBP (TATA box binding protein) was used as reference gene. The PCR primers used were 5ˊ-TCTCACCAGTCCGTCTCCTTC-3ˊ sense primer and 5ˊ-TCTCGTTGCCCAAATTGATTTCG-3ˊ anti-sense primer for BRAF, 5ˊ-TCTTGCCTCAGAAGAGCTGAA-3ˊ sense primer and 5ˊ-GCCTTCAGATGCCACAGACT-3ˊ anti-sense primer for KDR, 5ˊ-GAGCAGTATTACGACCCGAGTG-3ˊ sense primer and 5ˊ-TCCTTAGGCAAGTCATCCAATTCC-3ˊ anti-sense primer for MAPK1, 5ˊ-CAATCTCCCGCTGCTGCTG-3ˊ sense primer and 5ˊ-CCCACCCTCCACCTCTGC-3ˊ anti-sense primer for MAPK3, 5ˊ-CCCTAAGTGGATTGGCTTTGTG-3ˊ sense primer and 5ˊ-ATGGAGCAGGAACAGCAGTG-3ˊ anti-sense primer for MAP2K1, 5ˊ-ATGCCCCGCCCTTCCAG-3ˊ sense primer and 5ˊ-CGCCACGGTGCTCTCC-3ˊ anti-sense primer for MAP2K2, 5ˊ-TCTCGTATTTGCTGCATCGT-3ˊ sense primer and 5ˊ-CACTCGGTGAAATCAGGGTAA-3ˊ anti-sense primer for NANOG, 5ˊ- CCTTACCACATCCGCTCCATC-3ˊ sense primer and 5ˊ-TCACACTCTCCGTCACATTGC-3ˊ anti-sense primer for PDGFRB, 5ˊ- GGTGGAGAGCAACTCCGAT-3ˊ sense primer and 5ˊ- TGCAGAGCTTTGATGTCCTG-3ˊ anti-sense primer for POU5F1, 5ˊ- GGAGTGGTTCTCAGCAGGTTG-3ˊ sense primer and 5ˊ-AGTTCATACAATACGATGCCATAGG-3ˊ anti-sense primer for RAF1, and 5ˊ- CAATTTAGTAGTTATGAGCCAGAG-3ˊsense primer and 5ˊ- TTCTGCTCTGACTTTAGCAC-3ˊ anti-sense primer for TBP.

2.5 Statistical analysis

An unpaired two-tailed t-test was used to analyze the significance of any difference between the two means. The statistical significance was defined as p < 0.05.

4 DISCUSSION

The bioavailability of drug is determined by its character, and thus the in vitro experimental effects of the drug may be not consistent with the in vivo results.21, 22 To minimize this potential pitfall, the sorafenib concentrations applied in the study were based on the clinical plasma concentrations.3, 17-19 Our results were consistent with previous report20 that sorafenib could obviously downregulate RAF/MEK/ERK signaling pathways in one out of eight parent cell lines (12.5%). Consistency between these in vitro results and the real-world experience of anticancer effects of sorafenib on HCC indicates the reliability of our results. Dose-dependent antiproliferative effects of sorafenib on s003 parent cell line might be partially explained by that more genes in RAF/ERK signaling pathways were downregulated by10 μg/mL than by 5 μg/mL sorafenib. Two subgroups also demonstrated antiproliferative effects of sorafenib at 10 μg/mL but without obvious inhibition on RAF/MEK/ERK signaling pathways. This can be explained by that the antiproliferative effects of sorafenib on HCC can be caused by both MEK/ERK-dependent and MEK/ERK-independent signaling pathways.2

Both KDR and PDGFRB are upstream genes for several important signaling pathways such as the Ras, RAF/MEK/ERK, Rap1, and PI3K-Akt signaling pathways, and KDR is also the upstream gene for the vascular endothelial growth factor signaling pathway. Although they are considered as important targets for sorafenib, only s089 parent cell line demonstrated downregulation of both genes at 5 μg/mL sorafenib. However, no other tested gene related to RAF/MEK/ERK signaling pathways was concomitantly downregulated, and no inhibitory effects on proliferation was observed in s089 parent cell line at 5 μg/mL sorafenib. On the contrary, the expressions of KDR and PDGFRB cause by different concentrations of sorafenib were diverse which were also consistent with previous report.20 The responses of KDR and PDGFRB expressions induced by sorafenib in different HCC cells are hard to be predicted. On the other hand, the present study also demonstrated the unexpectedly promotive effects of sorafenib on cellular proliferation which was also consistent with previous report.20 However, analysis of the expressions of sorafenib target genes did not show consistent change. This suggests that the unexpected sorafenib effects were not related to RAF/MEK/ERK signaling pathways.

Overexpression of NANOG has been shown to be a hazard factor to predicate decreased overall survival and disease-free survival in HCC.23 NANOG induction can promote drug resistance in prostate and breast cancer cells.24 Knockdown of NANOG expression can reduce cell proliferation and differentiation potential, induce cell cycle arrest in cancer cells,25, 26 and suppress drug-resistance in esophageal squamous cancer cells.26 In the present study, dose-dependent antiproliferative effects of sorafenib were observed only in s003 parent cell line which showed the lowest expression of NANOG among all tested cell lines except s020 NDPU subgroup. However, the effects of sorafenib changed to promote proliferation in s003 NUPS subgroup. Sorafenib which have no influence on s015 and s071 parent cell lines also promoted proliferation in their upregulated NANOG alone subgroups. No influence of sorafenib on proliferation was observed in both s089 NUPS subgroup and its parent cell line. Different characteristics of s089 cancer cells causing no obvious response to sorafenib may be one of the explanations. On the contrary, both s015 NDPS and s020 NDPU subgroups demonstrated antiproliferative effects of sorafenib at 10 μg/mL which were not observed in their parent cell lines. Moreover, sorafenib which promoted s058 parent cell line proliferation changed to have no influence on s058 NDPD proliferation. These results indicate that the anticancer effects of sorafenib are limited in cancer cells with high NANOG expression. High NANOG expressed cancer cells will become the predominant cells after long-term sorafenib treatment due to the elimination of low NANOG expressed cancer cells by sorafenib. This can be one of the explanations for loss of anticancer effects of sorafenib on previously effectively treated HCC patients. Overexpression of POU5F1 has also been shown to be a hazard factor to predicate poor prognosis and/or recurrence in HCC.27 However, the present results showed that upregulation of POU5F1 alone had inconsistent influence on the antiproliferative effects of sorafenib among different HCC cell lines. Influence of NANOG and POU5F1 on differential expressions of sorafenib target genes among subgroups was also inconsistent. Sorafenib anticancer effects via MEK/ERK-independent signaling pathways may be one of the explanations.2 The above results indicate that NANOG rather than POU5F1 expression is a critical marker for the anticancer effects of sorafenib on HCC.

Not all parent cell lines could establish all kinds of subgroups with different NANOG and POU5F1 expressions. No such subgroup in parent cell line was the explanation. Nevertheless, successful establishment of subgroups by our method indicates the heterogenic characteristics of cancer cells within the tumor. Therefore, our results could be more accurate to reflex the real-word situation than results from artificial gene interference using purchased cell lines.

In conclusion, the anticancer effects of sorafenib on cancer cells with high NANOG expression were limited. In clinical practice, it is unable to determine NANOG expression in whole HCC tumors of all patients. Therefore, we suggest combination of sorafenib with other drug able to target cancer cells with high NANOG expression rather than sorafenib alone for the treatment of HCC.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

ETHICS APPROVAL, CONSENT TO PARTICIPATE, AND PATIENT CONSENT FOR PUBLICATION

This study was approved by the institutional review board of the hospital, and all patients signed the informed consent.