Acyclic retinoid (ACR) is a promising drug under clinical trials for preventing recurrence of hepatocellular carcinoma. The objective of this study was to gain insights into molecular basis of the antitumorigenic action of ACR from a metabolic point of view. To achieve this, comprehensive cationic and lipophilic liver metabolic profiling was performed in mouse diethylnitrosamine (DEN)-induced hepatic tumorigenesis model using both capillary electrophoresis time-of-flight mass spectrometry and liquid chromatography time-of-flight mass spectrometry. ACR significantly counteracted against acceleration of lipogenesis but not glucose metabolism in DEN-treated mice liver, suggesting an important role of lipid metabolic reprogramming in the initiation step of hepatic tumorigenesis. Knowledge-based pathway analysis suggested that inhibition of linoleic acid metabolites such as arachidonic acid, a proinflammatory precursor, played a crucial role in the prevention by ACR of DEN-induced chronic inflammation–mediated tumorigenesis of the liver. As a molecular mechanism of the ACR's effect to prevent the aberrant lipogenesis, microarray analysis identified that a key transcription regulator of both embryogenesis and tumorigenesis, COUP transcription factor 2, also known as NR2F2, was associated with the metabolic effect of ACR in human hepatocellular carcinoma cells. Our study provided potential therapeutic targets for the chemoprevention of hepatocellular carcinoma as well as new insights into the mechanisms underlying prevention of hepatic tumorigenesis. Cancer Prev Res; 9(3); 205–14. ©2016 AACR.

Hepatocellular carcinoma is the most common type of liver cancer and a leading cause of cancer-related death worldwide (nearly 600,000 deaths annually; ref. 1). Hepatocellular carcinoma is recorded with the worst prognosis, according to a population-based cancer registry data in China, in which the age-standardized 5-year relative survival rates for males and females are 10.2% and 10.3%, respectively (2). The high lethality of hepatocellular carcinoma is partly due to its high recurrence rate based on the concept of “field cancerization” (3). Acyclic retinoid (ACR), a synthetic vitamin A–like compound, was originally developed from a view of nutritional supplementation to improve the vitamin A content in hepatocellular carcinoma patients (4). Clinical trials revealed that ACR could significantly inhibit the recurrence of hepatocellular carcinoma after the removal of primary tumors (4, 5). It is hypothesized that this recurrence-preventive effect is associated with clonal deletion (6) through targeted elimination of cancer stem/progenitor cells, such as oval-like cells (7, 8). However, the detailed mechanisms underlying the prevention of hepatic tumorigenesis by ACR still need further investigation.

Metabolic alterations of cancer cells such as aerobic glycolysis are essential to generate energy and nutrients required for cancer cell processes (9). It is also believed that de novo lipogenesis contributes to the synthesis of membranes and signaling molecules in proliferating cancer cells (10). For example, serum metabolic profiling study demonstrated that the content of arachidonic acid (AA), a polyunsaturated fatty acid present in the phospholipids of cell membranes, was significantly elevated in hepatocellular carcinoma patients compared with the healthy controls (∼91.8-fold; ref. 11). In addition, metabolic reprogramming has been revealed to be regulated by proto-oncogenes and tumor suppressor genes, suggesting its primary function in tumorigenesis (12–14). Although viral hepatitis is the major causate of hepatocellular carcinoma, there is a growing recognition of the importance of metabolic syndrome such as obesity for the development of hepatocellular carcinoma (15). Dietary or genetic obesity can induce alteration of gut microbial metabolites such as increasing the levels of deoxycholic acid (DHA), which leads to chronic liver injury and facilitates hepatocellular carcinoma development (16). Interestingly, antibiotic treatments inhibiting DHA synthesis can prevent obesity-induced hepatocellular carcinoma development, suggesting abnormal metabolism of cancer cells is a potential therapeutic target for the systemic treatment and prevention of hepatocellular carcinoma (16, 17).

Our group and others have observed that ACR may suppress cell growth or inhibit hepatitis C virus replication by altering energy production and lipid metabolism in hepatocellular carcinoma cell lines (18, 19). However, in vitro study in malignant cells provides limited information to understand the mechanism of tumorigenesis. We reported that ACR prevented diethylnitrosamine (DEN)-induced hepatic tumorigenesis in obese and diabetic C57BKSL/J- +Leprdb/+Leprdb mice (db/db mice; ref. 20). ACR treatment reduced the prevalence of DEN-initiated liver precancerous lesions classified as foci of cellular alteration from 70% to 10%. Here, we performed metabolomics approaches to explore the molecular basis of the preventive effect of ACR on hepatic tumorigenesis from a metabolic point of view. To this end, comprehensive cationic and lipophilic metabolic profiles of liver tissues obtained from mouse DEN–induced hepatic tumorigenesis model were detected using capillary electrophoresis time-of-flight mass spectrometry (CE-TOFMS) and liquid chromatography time-of-flight mass spectrometry (LC-TOFMS), respectively (21). Our analyses revealed that ACR may suppress the enhanced lipogenesis during DEN-induced hepatic tumorigenesis.

Chemicals

ACR (NIK-333) was supplied by Kowa Co. Ltd. All-trans retinoic acid (atRA), DEN, and AA were purchased from Sigma-Aldrich.

Animal experiments

All experiments were performed in accordance with protocols approved by the Institutional Committee of Animal Experiment of Gifu University and RIKEN. Four-week-old male db/db mice were obtained from Japan SLC, Inc. and housed under constant temperature (22°C ± 1°C), with free access to food and water, 12-hour light/dark cycles, and were fed laboratory pellets. Detailed experimental procedure has been described previously in ref. (20). Briefly, mice were divided into four groups (n = 6, respectively). Three groups were given tap water containing 40 ppm of DEN for the first two weeks and then fed with basal diet alone (DEN group) and basal diet containing 0.03% ACR (DEN-0.03ACR group) or 0.06% ACR (DEN-0.06ACR group) till the end of experiment. The method for DEN treatment used in this study has been proved to be sufficient to develop liver neoplasms in db/db mice (20, 22, 23). The fourth group fed with basal diet containing 0.06% ACR alone without DEN treatment (0.06ACR group) was used as the negative control for changes during DEN-initiated liver tumorigenesis (Fig. 1A). After sacrifice by CO2 asphyxiation, liver tissues containing precancerous lesion but not cancer (approximately 60 mg/mouse) were isolated and stored at −80°C until further metabolic analyses. For histopathologic examination, 4-μm thick sections of formalin-fixed, paraffin-embedded livers were stained routinely with hematoxylin and eosin (H&E).

Figure 1.

Liver metabolic profiles in mouse DEN–induced hepatic tumorigenesis model. A, schematic overview of the experimental procedures. Mice were divided into four groups (n = 6, respectively). Three groups were given tap water containing 40 ppm of DEN for the first two weeks and fed with basal diet alone (DEN group) and basal diet containing 0.03% ACR (DEN-0.03ACR group) or 0.06% ACR (DEN-0.06ACR group) till the end of experiment. The fourth group fed with basal diet containing 0.06% ACR alone without DEN treatment (0.06ACR group) was used as the negative control for changes during DEN-initiated liver tumorigenesis. B, representative H&E staining of the liver precancerous lesions (top) and hepatocellular carcinoma (HCC; bottom) observed in DEN group. Scale bar 200 μm. To obtain a global view of the metabolic basis of the preventive effect of ACR on DEN-initiated tumorigenesis, unsupervised PCA analysis was applied on the quantification data measured by CE-TOFMS (C), LC-TOFMS (D), and the combined data (E). The score plot revealed clear discrimination in the metabolic profiles between DEN- and ACR-treated groups with the first component (PC1) representing 34.7%, 39.6%, and 30.1% of the total variance, respectively. FCA, foci of cellular alteration.

Figure 1.

Liver metabolic profiles in mouse DEN–induced hepatic tumorigenesis model. A, schematic overview of the experimental procedures. Mice were divided into four groups (n = 6, respectively). Three groups were given tap water containing 40 ppm of DEN for the first two weeks and fed with basal diet alone (DEN group) and basal diet containing 0.03% ACR (DEN-0.03ACR group) or 0.06% ACR (DEN-0.06ACR group) till the end of experiment. The fourth group fed with basal diet containing 0.06% ACR alone without DEN treatment (0.06ACR group) was used as the negative control for changes during DEN-initiated liver tumorigenesis. B, representative H&E staining of the liver precancerous lesions (top) and hepatocellular carcinoma (HCC; bottom) observed in DEN group. Scale bar 200 μm. To obtain a global view of the metabolic basis of the preventive effect of ACR on DEN-initiated tumorigenesis, unsupervised PCA analysis was applied on the quantification data measured by CE-TOFMS (C), LC-TOFMS (D), and the combined data (E). The score plot revealed clear discrimination in the metabolic profiles between DEN- and ACR-treated groups with the first component (PC1) representing 34.7%, 39.6%, and 30.1% of the total variance, respectively. FCA, foci of cellular alteration.

Close modal

CE-TOFMS measurement

Frozen liver tissues were immersed into 50% acetonitrile solution containing internal standards in crushing tubes and centrifuged (2,300 × g, 4°C, 5 minutes) using a desk-model crusher BMS-M10N21 (BMS Co., Ltd). After that, top layer was centrifugally filtered (9,100 g, 4°C, 120 minutes) using an UltrafreeMC-PLHCC 5-kDa cut-off filter (Human Metabolome Technologies Inc). The filtrates were dried and resuspended in 50 μL of Milli-Q water for CE-TOFMS analysis using an Agilent CE-TOFMS System (Agilent Technologies) as previously described (21, 24).

LC-TOFMS measurement

Frozen liver tissues were immersed into acetonitrile solution with 1% formic acid containing internal standards in crushing tubes and centrifuged (1,500 rpm, 4°C, 120 seconds) for three times using a desk-model crusher BMS-M10N21 (BMS Co., Ltd) following another centrifugation (1,500 rpm, 4°C, 120 seconds) after adding 167 μL of Milli-Q water. After that, supernatant was collected by centrifugation (5,000 × g, 4°C, 5 minutes). Meanwhile, the flow-through was mixed with 500 μL of acetonitrile solution with 1% formic acid and 167 μL of Milli-Q water, and supernatant was collected again by centrifugation. The supernatants were then mixed and centrifugally filtered (9,100 × g, 4°C, 120 minutes) using Pall Nanosep 3K Omega filters (Pall Corporation). The filtrate was dried and resuspended in 100 μL of isopropanol/Milli-Q water solution (1:1) for LC-TOFMS analysis using an Agilent 1200 series RRLC system SL equipped with an Agilent LC/MSD TOF system (Agilent Technologies).

Metabolites identification and quantification

Raw data obtained were analyzed with KEIO MasterHands software version 2.13 as previously described (25). Quantification of the major metabolites was performed as described previously (18, 25).

Cell culture

Hepatocellular carcinoma cell line, JHH7 cells were kindly supplied by Dr. Matsuura (Jikei University School of Medicine, Tokyo, Japan) in June 2012 (26). The cells were maintained in DMEM (Wako Industries) containing 10% FBS (Mediatech), 100 U/mL penicillin/streptomycin, and 2 mmol/L l-glutamine (Mediatech) and grown at 37°C in a 5% CO2 humidified incubator. No authentication for the cells was done by the authors.

Microarray analysis

Total RNA was isolated from JHH7 cells before (0 hour) and after treatment with 1 μmol/L atRA or 10 μmol/L ACR for 1 and 4 hours using a RNeasy Kit (Qiagen). The amount and purity of the isolated RNA were evaluated using a NanoDrop Spectrophotometer (NanoDrop Products). Then, oligonucleotide microarray experiment was performed using Affymetrix HG-U133 Plus 2.0 Array (Affymetrix). The arrays were scanned using a GenePix 4000B Microarray Scanner (Axon Instruments). Data normalization and analysis were performed with GeneSpring GX13.0 (Agilent Technologies) as previously described (27). All data are MIAME compliant, and the raw data were deposited in the Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo; accession no. GSE71856).

Real-time RT-PCR

JHH7 cells were treated with ethanol, 1 μmol/L atRA, or 10 μmol/L ACR for 4 hours, and then total RNA was isolated and quantified as described above. cDNA was synthesized using a PrimeScript RT Master Mix Kit (TaKaRa Bio). Oligonucleotide primers were designed using OligoPerfect Designer software (Invitrogen). The sequences of the primers (5′ to 3′) are as follows: GAPDH forward (CGACCACTTTGTCAAGCTCA) and reverse (AGGGGTCTACATGGCAACTG); CUOP transcription factor 2 (COUP-TFII, also known as NR2F2) forward (TGCCTGTGGTCTCTCTGATG) and reverse (ATATCCCGGATGAGGGTTTC). PCR reactions were performed on the Thermal Cycler Dice Real Time System (TP8000; TaKaRa Bio) with SsoAdvanced SYBR Green Supermix (Bio-Rad Laboratories).

Cell viability assay

Cells were seeded in 96-well plates 24 hours before treatment with AA in the presence and absence of 10 μmol/L ACR in DMEM containing 5% FBS. After 48 hours, cell viability was determined using the Cell Counting Kit-8 (Dojindo Molecular Technologies) in a plate reader (ARVO MX, PerkinElmer Inc).

ELISA

Cells were seeded in 10-cm dishes and allowed to grow to confluency. Cells were treated with 15 μmol/L ACR in FBS-free DMEM for 24 hours. The cell lysates were isolated and the contents of AA were measured using an ELISA kit (CEB098Ge, Cloud-Clone Corp.) according to the manufacturer's protocol.

Network generation and pathway analysis

MetaboAnalyst software (http://www.metaboanalyst.ca) was used to identify metabolites with significantly different levels across experimental conditions and apply the enrichment analysis of associated metabolic pathways (28). The Ingenuity Pathway Analysis (IPA) program (Ingenuity Systems) was used to identify networks and canonical pathways associated with differentially expressed genes following ACR treatment as previously described (27).

Statistical and multivariate analysis

The statistical significance of differences between values was assessed using ANOVA with post hoc Tukey HSD test or two-tailed Student t test. Values of P < 0.05 were considered to indicate statistical significance. Unsupervised principal component analysis (PCA) was run in SIMCA-P+ (version 12.0, Umetrics). Heatmap visualization of the metabolic data was generated using PeakStat version 3.18 (Human Metabolome Technologies Inc). Metabolic map was visualized in the network context using Visualization and Analysis of Networks containing Experimental Data (29). Hierarchical clustering analysis of the microarray data was applied using GeneSpring GX13.0 (Agilent Technologies).

Liver metabolic profiles in DEN-induced hepatic tumorigenesis mouse model

A total of 24 liver tissues (n = 6 per group) were obtained from mouse DEN–induced hepatic tumorigenesis model treated with or without ACR (Fig. 1A). Representative H&E staining images of the liver precancerous lesions and hepatocellular carcinoma observed in DEN group were presented in Fig. 1B. Peaks of 254 cationic metabolites (Supplementary Table S1) and 102 lipophilic metabolites (Supplementary Table S2) were detected using CE-TOFMS and LC-TOFMS, respectively. Metabolomics comparison using PCA analysis on CE-TOFMS data (Fig. 1C), LC-TOFMS data (Fig. 1D), and the combined data (Fig. 1E) revealed notable variations in the abundance of hepatic metabolites according differential ACR/DEN treatments. Notably, LC-TOFMS metabolomics data revealed the clearest distinction, suggesting an important role of lipid metabolism in the preventive effect of ACR on DEN-induced hepatic tumorigenesis. To provide a general view of the metabolic profiles, principle metabolic pathways of the detected metabolites were illustrated in Fig. 2 and Supplementary Fig. S1. Furthermore, a recent CE-TOFMS–based metabolomics study identified a novel biomarker pattern of ratio creatine/betaine of DEN-initiated hepatocellular carcinoma rat model, which could effectively predict the stage of human hepatocellular carcinoma (30). The ratio of creatine/betaine was calculated using the liver metabolomics data of our mouse DEN–initiated hepatocellular carcinoma model (Supplementary Fig. S2). Although no significance was observed, the ratio of creatine/betaine tended to increase in the DEN group.

Figure 2.

Principal metabolic maps illustrated using Visualization and Analysis of Networks containing Experimental Data. The relative quantities of the detected metabolites are represented as bar graphs (from left to right: DEN group, DEN-0.03ACR group, DEN-0.06 group, and 0.06ACR group). N.D., not detected.

Figure 2.

Principal metabolic maps illustrated using Visualization and Analysis of Networks containing Experimental Data. The relative quantities of the detected metabolites are represented as bar graphs (from left to right: DEN group, DEN-0.03ACR group, DEN-0.06 group, and 0.06ACR group). N.D., not detected.

Close modal

Statistical analysis of significantly changed metabolites

Heatmap analysis of all detected metabolites by CE/LC-TOFMS also revealed strong and reproducible ACR/DEN–induced changes of liver metabolic profiles in mouse DEN–induced hepatic tumorigenesis model under differential experimental conditions (Fig. 3A). ANOVA analysis with post hoc Tukey HSD test identified a total of 61 metabolites were significantly changed across differential chemical treatments with the threshold P values less than 0.05 (Fig. 3B). The top five of the most significantly changed metabolites were 1H-imidazole-4-propionic acid, cis-4,7,10,13,16,19-docosahexaenoic acid, arachidonic acid, phenylalanine, and cis-8,11,14-eicosatrienoic acid (Fig. 3C). With more details, 88 metabolites were significantly changed in the DEN alone–treated mice by comparing with the control mice treated with 0.06% ACR alone without DEN treatment (DEN vs. 0.06ACR). Meanwhile, 68 and 79 metabolites were significantly differential in mice treated with DEN combined with 0.03% and 0.06% ACR, respectively, by comparing with the mice treated with DEN alone (DEN-0.03ACR vs. DEN and DEN-0.06 ACR vs. DEN, respectively). Moreover, 50 common metabolites were shared among all the comparisons, indicating that ACR may specifically target the metabolic pathways dysregulated in DEN-induced tumorigenesis (Fig. 3D).

Figure 3.

Identification of metabolic targets involved in the antitumorigenic effect of ACR. A, heatmap visualization of ACR/DEN–induced changes based on all detected metabolite data by CE/LC-TOFMS. B, significantly changed metabolites across differential ACR/DEN treatment identified using ANOVA with post hoc Tukey HSD test. Red dot indicates significantly changed metabolites with P < 0.05. Top five metabolites were highlighted with arrows in B and ranked according to their −log10 of P values in C. Among the significantly changed metabolites observed in ANOVA analysis, Venn diagrams (D) show the number of metabolites that were significantly deregulated by comparing with the indicated chemical treatments. Eighty-eight metabolites were significantly differentiated between DEN group and 0.06ACR group. Sixty-eight metabolites were significantly differentiated between DEN-0.03ACR group and DEN group. Seventy-nine metabolites were significantly differentiated between DEN-0.06ACR group and DEN group. Fifty common metabolites were shared among all the three comparisons, which might be the candidate metabolic targets of ACR to prevent DEN-induced tumorigenesis.

Figure 3.

Identification of metabolic targets involved in the antitumorigenic effect of ACR. A, heatmap visualization of ACR/DEN–induced changes based on all detected metabolite data by CE/LC-TOFMS. B, significantly changed metabolites across differential ACR/DEN treatment identified using ANOVA with post hoc Tukey HSD test. Red dot indicates significantly changed metabolites with P < 0.05. Top five metabolites were highlighted with arrows in B and ranked according to their −log10 of P values in C. Among the significantly changed metabolites observed in ANOVA analysis, Venn diagrams (D) show the number of metabolites that were significantly deregulated by comparing with the indicated chemical treatments. Eighty-eight metabolites were significantly differentiated between DEN group and 0.06ACR group. Sixty-eight metabolites were significantly differentiated between DEN-0.03ACR group and DEN group. Seventy-nine metabolites were significantly differentiated between DEN-0.06ACR group and DEN group. Fifty common metabolites were shared among all the three comparisons, which might be the candidate metabolic targets of ACR to prevent DEN-induced tumorigenesis.

Close modal

Biologic process underlying preventive effect of ACR on hepatic tumorigenesis

Top metabolic pathways associated with metabolites significantly changed between DEN and 0.06ACR groups (Fig. 4A), DEN-0.03ACR and DEN groups (Fig. 4B), and DEN-0.06ACR and DEN groups (Fig. 4C) were identified using functional enrichment analyses in MetaboAnalyst software. Of interest, the top 2 metabolic pathways “Protein biosynthesis” and “Alpha linolenic acid and linoleic acid metabolism” were common in all the comparisons. Quantification results of the metabolites significantly inhibited by ACR treatment involved in the “Protein biosynthesis” and “α-linolenic acid and linoleic acid metabolism” were summarized in Fig. 5A and B, respectively. In contrast, no significant changes of metabolites involved in glucose metabolism such as glucose, glucose-6-phosphate, and lactic acid were observed (Fig. 5C). Furthermore, we investigated the effect of ACR on AA-induced cell growth of human hepatocellular carcinoma cells JHH7. ACR inhibited the contents of AA in JHH7 cells (Supplementary Fig. S3) and significantly suppressed the cell growth of JHH7 induced by a low dose of AA (1 μmol/L; Supplementary Fig. S4).

Figure 4.

Biologic process underlying the antitumorigenic effect of ACR. The list of significantly changed metabolites between DEN group and 0.06ACR group (A), DEN-0.03ACR and DEN group (B), and DEN-0.06ACR group and DEN group (C) were input into the MetaboAnalyst software. Top five associated metabolic pathways were presented and ranked according to their −log10 of P values. Numbers indicate the overlapping ratios of the number of enriched metabolites and the total number of metabolites involved in the pathways. The dashed lines indicate threshold of significance (P = 0.05).

Figure 4.

Biologic process underlying the antitumorigenic effect of ACR. The list of significantly changed metabolites between DEN group and 0.06ACR group (A), DEN-0.03ACR and DEN group (B), and DEN-0.06ACR group and DEN group (C) were input into the MetaboAnalyst software. Top five associated metabolic pathways were presented and ranked according to their −log10 of P values. Numbers indicate the overlapping ratios of the number of enriched metabolites and the total number of metabolites involved in the pathways. The dashed lines indicate threshold of significance (P = 0.05).

Close modal
Figure 5.

Quantification of the significantly changed metabolites. Relative quantitative data of enriched metabolites involved in the metabolic pathways “Protein synthesis” (A) and “Alpha linolenic acid and linoleic acid metabolism” (B) identified using MetaboAnalyst software and that of representative metabolites (glucose, glucose-6-phosphate, and lactic acid) involved in glucose metabolism (C). Data were presented as fold change as compared with the average of 0.06ACR group. Boxplot of quantitative data displays the full range of variation (from minimum to maximum). *, P < 0.05 compared with the DEN group identified using ANOVA with post hoc Tukey HSD test. n.s., not significant.

Figure 5.

Quantification of the significantly changed metabolites. Relative quantitative data of enriched metabolites involved in the metabolic pathways “Protein synthesis” (A) and “Alpha linolenic acid and linoleic acid metabolism” (B) identified using MetaboAnalyst software and that of representative metabolites (glucose, glucose-6-phosphate, and lactic acid) involved in glucose metabolism (C). Data were presented as fold change as compared with the average of 0.06ACR group. Boxplot of quantitative data displays the full range of variation (from minimum to maximum). *, P < 0.05 compared with the DEN group identified using ANOVA with post hoc Tukey HSD test. n.s., not significant.

Close modal

Gene expression profiles of hepatocellular carcinoma cells in response to ACR

Hierarchical clustering with Ward method of 44,907 genes detected by microarray analysis demonstrated diverse expression changes in hepatocellular carcinoma cells treated with ACR for 4 hours (Fig. 6A). A total of 973 differentially expressed genes in response to ACR by comparing with atRA for 4-hour treatments were identified with a fold change more than 2. Then, network analysis was performed on the altered gene expression profiles using IPA platform. The biologic functions of the top IPA-generated networks were summarized in Fig. 6B. The most highly populated networks were associated with the regulation of cell cycle and DNA replication, as ACR is well known to induce apoptosis and suppress cell proliferation in hepatocellular carcinoma cells (31). In addition, networks related to amino acid metabolism, protein synthesis, and lipid metabolism were also observed. For example, the biologic network entitled “Lipid Metabolism, Small Molecular Biochemistry, Vitamin and Mineral Metabolism” was highlighted in Fig. 6C. This network contains genes that play critical roles in controlling the development of tissues and organs such as the nuclear orphan receptor NR2F2. Then, the inhibitory effect of ACR on NR2F2 expressions was validated using real-time PCR (Fig. 6D).

Figure 6.

Gene expression profiles of hepatocellular carcinoma cells in response to ACR. JHH7 cells were treated with 1 μmol/L atRA or 10 μmol/L ACR for 0, 1, and 4 hours. Then, total RNA were isolated and applied to microarray analysis. A, hierarchical clustering with Ward method of 44,907 measured genes revealed that ACR treatment for 4 hours had outstanding effect on gene expression profiles of JHH7 cells. The list of differentially expressed genes between ACR treatment for 4 hours and atRA treatment for 4 hours were input into the IPA platform. B, the biologic functions of top populated networks generated in IPA were ranked by score, which is the likelihood of a set of genes being found in the network owing to random chance, identified by a Fisher exact test. A representative network related with cancer metabolism entitled "Lipid Metabolism, Small Molecular Biochemistry, Vitamin and Mineral Metabolism" is presented in C. Upregulated metabolites are indicated in red, downregulated metabolites indicated in green, and metabolites that were not annotated in this study but are part of this network are indicated in white. D, reduced levels of NR2F2 expression in JHH7 cells treated with ACR for 4 hours were verified using real-time PCR. Quantitative data were expressed as the means ± SD. *, P < 0.05 compared with the ethanol (/EtOH) control identified using two-tailed Student t test.

Figure 6.

Gene expression profiles of hepatocellular carcinoma cells in response to ACR. JHH7 cells were treated with 1 μmol/L atRA or 10 μmol/L ACR for 0, 1, and 4 hours. Then, total RNA were isolated and applied to microarray analysis. A, hierarchical clustering with Ward method of 44,907 measured genes revealed that ACR treatment for 4 hours had outstanding effect on gene expression profiles of JHH7 cells. The list of differentially expressed genes between ACR treatment for 4 hours and atRA treatment for 4 hours were input into the IPA platform. B, the biologic functions of top populated networks generated in IPA were ranked by score, which is the likelihood of a set of genes being found in the network owing to random chance, identified by a Fisher exact test. A representative network related with cancer metabolism entitled "Lipid Metabolism, Small Molecular Biochemistry, Vitamin and Mineral Metabolism" is presented in C. Upregulated metabolites are indicated in red, downregulated metabolites indicated in green, and metabolites that were not annotated in this study but are part of this network are indicated in white. D, reduced levels of NR2F2 expression in JHH7 cells treated with ACR for 4 hours were verified using real-time PCR. Quantitative data were expressed as the means ± SD. *, P < 0.05 compared with the ethanol (/EtOH) control identified using two-tailed Student t test.

Close modal

Finally, to mine the connection between the metabolic and genetic actions of ACR observed in this study, a NR2F2-dependent regulatory network was generated by knowledge-based pathway analysis in IPA platform (Fig. 7). A potential mechanism is that NR2F2 may bind to retinoid X receptor, which is known as a molecular target of ACR (32), and regulate lipid metabolism including linoleic acid metabolism through the downstream of peroxisome proliferator-activated receptor (PPAR) signaling pathways (33).

Figure 7.

A schematic model of NR2F2-dependent regulatory network underlying the antitumorigenic actions of ACR generated using IPA platform. PDK4, pyruvate dehydrogenase kinase 4; RXR, retinoid X receptor

Figure 7.

A schematic model of NR2F2-dependent regulatory network underlying the antitumorigenic actions of ACR generated using IPA platform. PDK4, pyruvate dehydrogenase kinase 4; RXR, retinoid X receptor

Close modal

There are raised efforts to target cancer metabolism as a potential anticancer strategy (17). However, these therapies have been partly elusive due to the poor understanding of metabolic phenotypes in the tissue-specific tumorigenic process. One major finding of our study was that alterations in lipid metabolism, but not glucose metabolism, were observed in DEN-induced hepatic tumorigenesis. ACR can significantly inhibit the DEN-induced acceleration of lipogenesis, suggesting lipogenic pathway as potential cancer targets (Fig. 4). Cancer metabolism has been long equated as aerobic glycolysis (known as "Warburg effect"; ref. 34), which is mainly based on the hypothesis of mitochondrial defects in cancer cells. However, this hypothesis is challenged by recent studies that most tumor mitochondria are not defective in their ability to carry out oxidative phosphorylation (12). Our previous in vitro study also raised questions regarding this issue that no significant difference in the content of lactic acid was observed between hepatocellular carcinoma cells, JHH7, and normal hepatic cells. In addition, the increased gene expression of pyruvate dehydrogenase kinase 4, which attenuates the flux of glycolytic carbon into mitochondrial oxidation, was found to be involved in the growth suppression action of ACR in JHH7 cells (18). As the JHH7 cells exhibited highly enhanced liver differentiation functions and stem cell–like features (35), it was possible that lipid metabolic reprogramming might be required in the initiation step of hepatic tumorigenesis. Consistently, increased biosynthesis of macromolecules, particularly lipids, has increasingly been recognized as an important component of cancer metabolic reprogramming (36). Limiting fatty acid synthesis can control cancer cell proliferation (10), whereas increased de novo fatty acid synthesis has been shown to correlate with breast cancer progression (37). This study further provided new evidences that lipid metabolic reprogramming might play a critical role in the development of hepatocellular carcinoma.

In this study, linoleic acid metabolism was identified as the mostly populated metabolic target of ACR (Fig. 3). Decreases in linoleic acid metabolites such as AA have been suggested as a potential mechanisms from mammary cancer prevention (38). An important issue is the difference between DEN model and human hepatocellular carcinoma. DEN is the most widely used chemical to induce hepatocellular carcinoma in mice, and DEN model is received as one of the best models to study the pathology of hepatocellular carcinoma, as it could mimic the inflammation–fibrosis process (39). The differences between the DEN model and human hepatocellular carcinoma observed by comparing their gene expression profiles have been reported (40). However, limited focus has been placed on their differences in metabolic profiles. A recent CE-TOFMS–based metabolomics study suggested that the metabolic profile of DEN model is effective to identify novel biomarkers for early diagnosis of human hepatocellular carcinoma (30). Indeed, inhibition of AA metabolic pathway has been recently reported to associate with the chemoprevention of high fat diet–enhanced colorectal carcinogenesis (41) and the apoptosis of hepatocellular carcinoma cells (42). Furthermore, significantly elevated content of AA was also reported in hepatocellular carcinoma patients compared with the healthy controls (11).

AA is known as a proinflammatory precursor playing etiologic roles in multiple cancers (43). It is possible that through inhibition of AA-regulated signaling pathways, ACR might prevent DEN-induced chronic liver inflammation, which finally contributes to hepatic tumorigenesis. This is reasonable as we previously showed enhanced inflammatory response and inhibited serum levels of TNFα and expression levels of TNFα, IL6, and IL1b mRNA in the liver of db/db mice treated with ACR compared with those treated with DEN alone (20). Furthermore, in a platelet-derived growth factor–overexpressed mouse model, genome-wide expression profile analysis revealed that the repressive effect of ACR on the development of hepatic fibrosis and tumors was related with inflammatory signaling pathways such as “MIF signaling” and “IL6 signaling” (44).

Further microarray analysis identified that inhibited NR2F2 expression in human hepatocellular carcinoma cells was associated with the metabolic effect of ACR through PPAR-dependent signaling pathways (Figs. 5 and 6). Deletion of NR2F2 led to embryonic lethality with defects in angiogenesis and heart development, suggesting an essential role of NR2F2 in embryogenesis and organization (45). It was known that aggressive tumor cells shared many characteristics with embryonic progenitors. The discovery of cancer stem cells further increased the interest in the interactions between cancer progression and embryologic development (46). Indeed, biologic function of NR2F2 has been proven in prostate carcinogenesis (47), suggesting NR2F2 as a potential drug target for preventing tumor progression.

In summary, to investigate the metabolic effect of ACR against DEN-induced hepatic tumorigenesis, comprehensive cationic and lipophilic metabolic analyses were performed using CE/LC-TOFMS. Significant preventive effect of ACR was observed on accelerated lipogenesis elicited by DEN, but not glucose metabolism. Pathway analysis suggested a crucial role of linoleic acid metabolism, such as the AA metabolic pathway, in the antitumorigenic action of ACR. Gene expression analysis identified NR2F2, a key transcription regulator of embryogenesis and tumorigenesis, was associated with the metabolic effect of ACR in human hepatocellular carcinoma cells. Our study provided potential therapeutic targets for the chemoprevention of hepatocellular carcinoma, as well as new insights into the mechanisms underlying hepatic tumorigenesis.

No potential conflicts of interest were disclosed.

Conception and design: X.-Y. Qin, S. Kojima

Development of methodology: X.-Y. Qin, S. Kojima

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X.-Y. Qin, H. Tatsukawa, Y. Shirakami, M. Shimizu, H. Moriwaki

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X.-Y. Qin, H. Moriwaki

Writing, review, and/or revision of the manuscript: X.-Y. Qin, K. Hitomi, N. Ishibashi, H. Moriwaki, S. Kojima

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N. Ishibashi, H. Moriwaki, S. Kojima

Study supervision: H. Moriwaki, S. Kojima

This study was partly supported by the HMT Research Grant for Young Leaders in Metabolomics 2012 from Human Metabolome Technologies Inc. (to X.-Y. Qin), a Grant-in-Aid for Scientific Research on Innovative Areas “Chemical Biology of Natural Products” from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to S. Kojima), and the Research on the Innovative Development and the Practical Application of New Drugs for Hepatitis B (H24-B Drug Discovery Hepatitis General 003) from the Ministry of Health, Labor and Welfare of Japan (to S. Kojima).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Venook
AP
,
Papandreou
C
,
Furuse
J
,
de Guevara
LL
. 
The incidence and epidemiology of hepatocellular carcinoma: a global and regional perspective
.
Oncologist
2010
;
15
Suppl 4
:
5
13
.
2.
Zeng
H
,
Zheng
R
,
Guo
Y
,
Zhang
S
,
Zou
X
,
Wang
N
, et al
Cancer survival in China, 2003-2005: a population-based study
.
Int J Cancer
2015
;
136
:
1921
30
.
3.
Lou
C
,
Du
Z
,
Yang
B
,
Gao
Y
,
Wang
Y
,
Fang
S
. 
Aberrant DNA methylation profile of hepatocellular carcinoma and surgically resected margin
.
Cancer Sci
2009
;
100
:
996
1004
.
4.
Muto
Y
,
Moriwaki
H
,
Ninomiya
M
,
Adachi
S
,
Saito
A
,
Takasaki
KT
, et al
Prevention of second primary tumors by an acyclic retinoid, polyprenoic acid, in patients with hepatocellular carcinoma. Hepatoma Prevention Study Group
.
N Engl J Med
1996
;
334
:
1561
7
.
5.
Okita
K
,
Izumi
N
,
Ikeda
K
,
Osaki
Y
,
Numata
K
,
Ikeda
M
, et al
Survey of survival among patients with hepatitis C virus-related hepatocellular carcinoma treated with peretinoin, an acyclic retinoid, after the completion of a randomized, placebo-controlled trial
.
J Gastroenterol
2015
;
50
:
667
74
.
6.
Shimizu
M
,
Takai
K
,
Moriwaki
H
. 
Strategy and mechanism for the prevention of hepatocellular carcinoma: phosphorylated retinoid X receptor alpha is a critical target for hepatocellular carcinoma chemoprevention
.
Cancer Sci
2009
;
100
:
369
74
.
7.
Sano
T
,
Kagawa
M
,
Okuno
M
,
Ishibashi
N
,
Hashimoto
M
,
Yamamoto
M
, et al
Prevention of rat hepatocarcinogenesis by acyclic retinoid is accompanied by reduction in emergence of both TGF-alpha-expressing oval-like cells and activated hepatic stellate cells
.
Nutr Cancer
2005
;
51
:
197
206
.
8.
Guan
HB
,
Nie
YZ
,
Zheng
YW
,
Takiguchi
K
,
Yu
HW
,
Zhang
RR
, et al
Acyclic retinoid induces differentiation and apoptosis of murine hepatic stem cells
.
Stem Cell Res Ther
2015
;
6
:
51
.
9.
Vander Heiden
MG
,
Cantley
LC
,
Thompson
CB
. 
Understanding the Warburg effect: the metabolic requirements of cell proliferation
.
Science
2009
;
324
:
1029
33
.
10.
Currie
E
,
Schulze
A
,
Zechner
R
,
Walther
TC
,
Farese
RV
 Jr
. 
Cellular fatty acid metabolism and cancer
.
Cell Metab
2013
;
18
:
153
61
.
11.
Zhou
L
,
Ding
L
,
Yin
P
,
Lu
X
,
Wang
X
,
Niu
J
, et al
Serum metabolic profiling study of hepatocellular carcinoma infected with hepatitis B or hepatitis C virus by using liquid chromatography-mass spectrometry
.
J Proteome Res
2012
;
11
:
5433
42
.
12.
Ward
PS
,
Thompson
CB
. 
Metabolic reprogramming: a cancer hallmark even warburg did not anticipate
.
Cancer Cell
2012
;
21
:
297
308
.
13.
Carroll
PA
,
Diolaiti
D
,
McFerrin
L
,
Gu
H
,
Djukovic
D
,
Du
J
, et al
Deregulated Myc requires MondoA/Mlx for metabolic reprogramming and tumorigenesis
.
Cancer Cell
2015
;
27
:
271
85
.
14.
Jeong
SM
,
Xiao
C
,
Finley
LW
,
Lahusen
T
,
Souza
AL
,
Pierce
K
, et al
SIRT4 has tumor-suppressive activity and regulates the cellular metabolic response to DNA damage by inhibiting mitochondrial glutamine metabolism
.
Cancer Cell
2013
;
23
:
450
63
.
15.
Welzel
TM
,
Graubard
BI
,
Zeuzem
S
,
El-Serag
HB
,
Davila
JA
,
McGlynn
KA
. 
Metabolic syndrome increases the risk of primary liver cancer in the United States: a study in the SEER-Medicare database
.
Hepatology
2011
;
54
:
463
71
.
16.
Yoshimoto
S
,
Loo
TM
,
Atarashi
K
,
Kanda
H
,
Sato
S
,
Oyadomari
S
, et al
Obesity-induced gut microbial metabolite promotes liver cancer through senescence secretome
.
Nature
2013
;
499
:
97
101
.
17.
Vander Heiden
MG
. 
Targeting cancer metabolism: a therapeutic window opens
.
Nat Rev Drug Discov
2011
;
10
:
671
84
.
18.
Qin
XY
,
Wei
F
,
Tanokura
M
,
Ishibashi
N
,
Shimizu
M
,
Moriwaki
H
, et al
The effect of acyclic retinoid on the metabolomic profiles of hepatocytes and hepatocellular carcinoma cells
.
PLoS One
2013
;
8
:
e82860
.
19.
Shimakami
T
,
Honda
M
,
Shirasaki
T
,
Takabatake
R
,
Liu
F
,
Murai
K
, et al
The acyclic retinoid Peretinoin inhibits hepatitis C virus replication and infectious virus release in vitro
.
Sci Rep
2014
;
4
:
4688
.
20.
Shimizu
M
,
Sakai
H
,
Shirakami
Y
,
Iwasa
J
,
Yasuda
Y
,
Kubota
M
, et al
Acyclic retinoid inhibits diethylnitrosamine-induced liver tumorigenesis in obese and diabetic C57BLKS/J- +(db)/+Lepr(db) mice
.
Cancer Prev Res
2011
;
4
:
128
36
.
21.
Soga
T
,
Ohashi
Y
,
Ueno
Y
,
Naraoka
H
,
Tomita
M
,
Nishioka
T
. 
Quantitative metabolome analysis using capillary electrophoresis mass spectrometry
.
J Proteome Res
2003
;
2
:
488
94
.
22.
Iwasa
J
,
Shimizu
M
,
Shiraki
M
,
Shirakami
Y
,
Sakai
H
,
Terakura
Y
, et al
Dietary supplementation with branched-chain amino acids suppresses diethylnitrosamine-induced liver tumorigenesis in obese and diabetic C57BL/KsJ-db/db mice
.
Cancer Sci
2010
;
101
:
460
7
.
23.
Shimizu
M
,
Sakai
H
,
Shirakami
Y
,
Yasuda
Y
,
Kubota
M
,
Terakura
D
, et al
Preventive effects of (-)-epigallocatechin gallate on diethylnitrosamine-induced liver tumorigenesis in obese and diabetic C57BL/KsJ-db/db Mice
.
Cancer Prev Res
2011
;
4
:
396
403
.
24.
Soga
T
,
Ueno
Y
,
Naraoka
H
,
Ohashi
Y
,
Tomita
M
,
Nishioka
T
. 
Simultaneous determination of anionic intermediates for Bacillus subtilis metabolic pathways by capillary electrophoresis electrospray ionization mass spectrometry
.
Anal Chem
2002
;
74
:
2233
9
.
25.
Sugimoto
M
,
Hirayama
A
,
Robert
M
,
Abe
S
,
Soga
T
,
Tomita
M
. 
Prediction of metabolite identity from accurate mass, migration time prediction and isotopic pattern information in CE-TOFMS data
.
Electrophoresis
2010
;
31
:
2311
8
.
26.
Fujise
K
,
Nagamori
S
,
Hasumura
S
,
Homma
S
,
Sujino
H
,
Matsuura
T
, et al
Integration of hepatitis B virus DNA into cells of six established human hepatocellular carcinoma cell lines
.
Hepatogastroenterology
1990
;
37
:
457
60
.
27.
Qin
XY
,
Kojima
Y
,
Mizuno
K
,
Ueoka
K
,
Muroya
K
,
Miyado
M
, et al
Identification of novel low-dose bisphenol a targets in human foreskin fibroblast cells derived from hypospadias patients
.
PLoS One
2012
;
7
:
e36711
.
28.
Xia
J
,
Psychogios
N
,
Young
N
,
Wishart
DS
. 
MetaboAnalyst: a web server for metabolomic data analysis and interpretation
.
Nucleic Acids Res
2009
;
37
:
W652
60
.
29.
Junker
BH
,
Klukas
C
,
Schreiber
F
. 
VANTED: a system for advanced data analysis and visualization in the context of biological networks
.
BMC Bioinformatics
2006
;
7
:
109
.
30.
Zeng
J
,
Huang
X
,
Zhou
L
,
Tan
Y
,
Hu
C
,
Wang
X
, et al
Metabolomics identifies biomarker pattern for early diagnosis of hepatocellular carcinoma: from diethylnitrosamine treated rats to patients
.
Sci Rep
2015
;
5
:
16101
.
31.
Tatsukawa
H
,
Sano
T
,
Fukaya
Y
,
Ishibashi
N
,
Watanabe
M
,
Okuno
M
, et al
Dual induction of caspase 3- and transglutaminase-dependent apoptosis by acyclic retinoid in hepatocellular carcinoma cells
.
Mol Cancer
2011
;
10
:
4
.
32.
Matsushima-Nishiwaki
R
,
Okuno
M
,
Takano
Y
,
Kojima
S
,
Friedman
SL
,
Moriwaki
H
. 
Molecular mechanism for growth suppression of human hepatocellular carcinoma cells by acyclic retinoid
.
Carcinogenesis
2003
;
24
:
1353
9
.
33.
Pozzi
A
,
Popescu
V
,
Yang
S
,
Mei
S
,
Shi
M
,
Puolitaival
SM
, et al
The anti-tumorigenic properties of peroxisomal proliferator-activated receptor alpha are arachidonic acid epoxygenase-mediated
.
J Biol Chem
2010
;
285
:
12840
50
.
34.
Levine
AJ
,
Puzio-Kuter
AM
. 
The control of the metabolic switch in cancers by oncogenes and tumor suppressor genes
.
Science
2010
;
330
:
1340
4
.
35.
Laurent
T
,
Murase
D
,
Tsukioka
S
,
Matsuura
T
,
Nagamori
S
,
Oda
H
. 
A novel human hepatoma cell line, FLC-4, exhibits highly enhanced liver differentiation functions through the three-dimensional cell shape
.
J Cell Physiol
2012
;
227
:
2898
906
.
36.
Schulze
A
,
Harris
AL
. 
How cancer metabolism is tuned for proliferation and vulnerable to disruption
.
Nature
2012
;
491
:
364
73
.
37.
Hilvo
M
,
Denkert
C
,
Lehtinen
L
,
Muller
B
,
Brockmoller
S
,
Seppanen-Laakso
T
, et al
Novel theranostic opportunities offered by characterization of altered membrane lipid metabolism in breast cancer progression
.
Cancer Res
2011
;
71
:
3236
45
.
38.
Banni
S
,
Angioni
E
,
Casu
V
,
Melis
MP
,
Carta
G
,
Corongiu
FP
, et al
Decrease in linoleic acid metabolites as a potential mechanism in cancer risk reduction by conjugated linoleic acid
.
Carcinogenesis
1999
;
20
:
1019
24
.
39.
Heindryckx
F
,
Colle
I
,
Van Vlierberghe
H
. 
Experimental mouse models for hepatocellular carcinoma research
.
Int J Exp Pathol
2009
;
90
:
367
86
.
40.
Wang
W
,
Iyer
NG
,
Tay
HT
,
Wu
Y
,
Lim
TK
,
Zheng
L
, et al
Microarray profiling shows distinct differences between primary tumors and commonly used preclinical models in hepatocellular carcinoma
.
BMC Cancer
2015
;
15
:
828
.
41.
Xie
G
,
Wang
CZ
,
Yu
C
,
Qiu
Y
,
Wen
XD
,
Zhang
CF
, et al
Metabonomic profiling reveals cancer chemopreventive effects of American ginseng on colon carcinogenesis in Apc Mice
.
J Proteome Res
2015
;
14
:
3336
47
.
42.
Li
J
,
Li
O
,
Kan
M
,
Zhang
M
,
Shao
D
,
Pan
Y
, et al
Berberine induces apoptosis by suppressing the arachidonic acid metabolic pathway in hepatocellular carcinoma
.
Mol Med Rep
2015
;
12
:
4572
7
.
43.
Meng
H
,
Liu
Y
,
Lai
L
. 
Diverse ways of perturbing the human arachidonic acid metabolic network to control inflammation
.
Acc Chem Res
2015
;
48
:
2242
50
.
44.
Okada
H
,
Honda
M
,
Campbell
JS
,
Sakai
Y
,
Yamashita
T
,
Takebuchi
Y
, et al
Acyclic retinoid targets platelet-derived growth factor signaling in the prevention of hepatic fibrosis and hepatocellular carcinoma development
.
Cancer Res
2012
;
72
:
4459
71
.
45.
Pereira
FA
,
Qiu
Y
,
Zhou
G
,
Tsai
MJ
,
Tsai
SY
. 
The orphan nuclear receptor COUP-TFII is required for angiogenesis and heart development
.
Genes Dev
1999
;
13
:
1037
49
.
46.
Topczewska
JM
,
Postovit
LM
,
Margaryan
NV
,
Sam
A
,
Hess
AR
,
Wheaton
WW
, et al
Embryonic and tumorigenic pathways converge via Nodal signaling: role in melanoma aggressiveness
.
Nat Med
2006
;
12
:
925
32
.
47.
Qin
J
,
Wu
SP
,
Creighton
CJ
,
Dai
F
,
Xie
X
,
Cheng
CM
, et al
COUP-TFII inhibits TGF-beta-induced growth barrier to promote prostate tumorigenesis
.
Nature
2013
;
493
:
236
40
.

Supplementary data