Samples from cancer tissues and their adjacent normal counterparts were obtained from 67 patients who had undergone surgical resection for HCC at the Fifth Hospital of Xiamen (Xiamen, China) between January 2015 and December 2016. The mean age of the patients was 52.6±11.3 years (range, 18–76 years). In total, 83.6% of the patients were male, consistent with a prior report that HCC is more prevalent in men compared with women (14). The patients were diagnosed with HCC via pathological examination (data not shown). All patients were informed of the aims of this study, and they provided written informed consent for the investigation in accordance with the Ethics Committee in the Fifth Hospital of Xiamen, following the clinical registration guidelines in China. This study was approved by the Ethics Committee of the Fifth Hospital of Xiamen. Tumor-node-metastasis (TNM) stage was determined according to the World Health Organization TNM staging 7th edition and the pathological analysis results. The samples were sectioned and some were stored at −80°C for western blot and high-performance liquid chromatography (HPLC)-mass spectrometry (MS) analysis, whereas others were flash-frozen in liquid nitrogen for PCR analysis.

Total RNA was extracted from tissue samples using TRIzol reagent (Thermo Fisher Scientific, Inc.), and RNA concentration was measured with a spectrophotometer (Beckman Coulter, Inc.). Total RNA (1 µg) was reverse-transcribed to cDNA with the ReverTra Ace^® qPCR RT kit (Toyobo Life Science), according to the manufacturer's protocols, and amplified with Ex Taq DNA polymerase, according to the manufacturer's protocols. qPCR was carried out on an Applied Biosystems 7500 Fast Real-Time PCR System (Thermo Fisher Scientific, Inc.) with SYBR Premix Ex Taq II (Takara Bio Inc., Otsu, Japan). The thermocycling conditions were as follows: Denaturation at 95°C for 5 sec, annealing/extension at 60°C for 31 sec (40 cycles). The quantitative values of mRNA were analyzed using the 2^–ΔΔCq method (15) and normalized relative to the levels of 18S. Each sample was set up in triplicate and the experiments were repeated three times.

The primers were as follows: N-acylphosphatidylethanolamine-hydrolysingphospholipase D (NAPE-PLD), forward primer (F): 5′-TGGCTGGGACACGCG-3′, reverse primer (R): 5′-GGGATCCGTGAGGAGGATG-3′; fatty acid amide hydrolase (FAAH), F: 5′-GCCTCAAGGAATGCTTCAGC-3′, R: 5′-TGCCCTCATTCAGGCTCAAG-3′; monoacylglycerol lipase (MAGL), F: 5′-CATGTGGATTCCATGCAGAAAG-3′, R: 5′-AGGATTGGCAAGAACCAGAGG-3′; diacylglycerol lipase (DGL)-α, F: 5′-AGAATGTCACCCTCGGAATGG-3′, R: 5′-GTGGCTCTCAGCTTGACAAAGG-3′; CB₁, F: 5′-AGCCTCTGGATAACAGCATGG-3′, R: 5′-AATCTTGACCGTGCTCTTGATG-3′; CB₂, F: 5′-CTCAGTGACCAGGTCAAGAAGG-3′, R: 5′-TTTTGCCTCTGACCCAAGG-3′; ceramide synthases (Cer)S1, F: 5′-TTTGGCTCCCGCACAATGT-3′, R: 5′-AAAAGCGAGATAGAGGTCCTCA-3′; CerS2, F: 5′-GCTCTTCCTCATCGTTCGATAC-3′, R: 5′-GTGTAGCCACGTACAGCTCA-3′; CerS3, F: 5′-CACCCAGCTGTCAAAGAGAAGG-3′, R: 5′-AGGACGATATCCGAAAGGTGG-3′; CerS4, F: 5′-CCGGATCCCGTCCAGTTTCAACGAG-3′, R: 5′-GGGAATTCGGCTATGTGGCTGTTGTG-3′; CerS5, F: 5′-GCTGCTCTTCGAGCGATTTAT-3′, R: 5′-CCTCCGATGGCGAAACCAG-3′; CerS6, F: 5′-TTTGGCTCCCGCACAATGT-3′, R: 5′-AAAAGCGAGATAGAGGTCCTCA-3′; 18S, F: 5′-CAGCCACCCGAGATTGAGCA-3′, R: 5′-TAGTAGCGACGGGCGGTGTG-3′.

Tumors and adjacent normal tissues were homogenized and sonicated (50/60Hz) in ice-cold RIPA buffer (Beyotime Biotechnology, Jiangsu, China) for 10 times, 5 sec each time. The suspension was centrifuged for 15 min at 4°C and 15,294 × g. The total protein content was determined using a bicinchoninic acid protein assay kit (Thermo Fisher Scientific, Inc.), according to the manufacturer's protocols, with bovine serum albumin (BSA; Solarbio Science & Technology Co. Ltd., Beijing, China) as the standard. Proteins (20 µg) were mixed with sample buffer, boiled, and then separated by 15% SDS-PAGE. After transference to polyvinylidene difluoride membranes (GE Healthcare Life Sciences), the membrane was blocked in 5% milk in Tris-buffered saline with 1% Tween-20 at room temperature for 1 h, and incubated at 4°C overnight with primary antibodies against CB₁ (Abcam; 1:1,000; cat. no. ab23703), CB₂ (Abcam; 1:1,000; cat. no., ab3561), SCD1 (Abcam; 1:500; cat. no. ab19862) and β-actin (Sigma-Aldrich, Merck KGaA; 1:500; cat. no. A5316). The membranes were subsequently incubated with an HRP-conjugated anti-rabbit IgG antibody (Sigma-Aldrich, Merck KGaA; 1:10,000; cat. no. SAB5600127) for 1 h at room temperature. Protein bands were visualized using an Enhanced Chemiluminescence Plus kit (GE Healthcare Life Sciences). Quantitative analysis was performed using the ImageJ software (version 2.1.4.7; National Institute of Health) with β-actin as the endogenous control.

In total, 50 µg frozen tissue samples were homogenized in 2 ml methanol/H₂O mixture (50:50, v/v) containing C17:1 FAE, and C17:0 ceramide as internal standards. Endogenous lipids were extracted with 3 ml chloroform, and then centrifuged for 10 min at 4°C and 3,000 × g. The organic phase was collected and dried under nitrogen by a nitrogen evaporator (Beijing TongTaiLian Technology Co., Ltd.). Lipids were re-dissolved in 1 ml chloroform and transferred to small Silica Gel G columns. Endogenous FAEs and ceramides were eluted with methanol/chloroform (10:90, v/v), dried under nitrogen, reconstituted in 100 µl methanol and detected using the 3200 Q Trap HPLC-MS system (Applied Biosystems; Thermo Fisher Scientific, Inc.) coupling with the 1100-HPLC system (Agilent Technologies).

The parameters of isolation and elution condition, ion monitor model and the molecular ion were all previously described in detail (16). The detailed information is as follows. The gradient elution of the mobile phase was as follows: 85% methanol (containing 15% H₂O; pH 7.5) was kept for the first 3 min, followed by a linear gradient from 85 to 100% methanol for 2 min, and then 100% methanol was continued for another 15 min. Finally, the elution condition went back to 85% methanol for 2 min at a flow rate of 0.7 ml/min. Column temperature was kept at 40°C. Ion detection was monitored by APCI⁺-MRM mode. The molecular ions were monitored at m/z 348.00/62.00 for AEA, m/z 379.10/287.10 for 2-AG, m/z 313.1/62.0 for C17:1 FAE, m/z 464.4/264.2 for C12:0 ceramide, m/z 520.4/264.2 for C16:0 ceramide, m/z 534.3/264.2 for C17:0 ceramide, m/z 548.4/264.2 for C18:0 ceramide, m/z 576.4/264.2 for C20:0 ceramide, m/z 604.5/264.2 for C22:0 ceramide, m/z 632.4/264.2 for C24:0 ceramide, and m/z 630.4/264.2 for C24:1 ceramide.

Tissue samples were cut into 200 µm thick sections and were homogenized in ice-cold Tris-HCl buffer (50 mM, pH 7.4) containing 0.32 M sucrose to obtain total proteins. To measure FAAH and MAGL activity, tissue proteins were incubated at 37°C for 30 min in 50 mM Tris-HCl buffer (pH 8.0, containing 0.05% fatty acid-free BSA), and 100 µg sample protein was incubated with 50 µM AEA or 2-oleoylglycerol as substrates. The reaction was stopped by adding 200 µl chloroform/methanol (1:1, v/v), containing C17:0 heptadecanoic acid as an internal standard. The reaction solution was centrifuged at 1,500 × g at 4°C for 5 min and the organic layers were subsequently collected and dried under nitrogen. The residues were re-dissolved in 100 µl methanol, and analyzed by 3200 Q Trap HPLC-MS system (Applied Biosystems; Thermo Fisher Scientific, Inc.) coupled with the 1100-HPLC system (Agilent Technologies, Inc.) in the negative-ion mode using 17:0 heptadecanoic acid as internal standard. A Hypersil Gold C18 column (dimensions, 250 × 4.6 mm; particle size, 5 µm; Thermo Fisher Scientific, Inc.) was used for analytical separation and the column temperature was kept at 40°C. Fatty acids were eluted using a linear gradient from 90% phase A (methanol containing 0.25% acetic acid and 5 mM ammonium acetate) to 100% phase B (water containing 0.25% acetic acid and 5 mM ammonium acetate) in 2.5 min at a flow rate of 1.0 ml/min. Capillary voltage was set at −4 kV and the fragmentor voltage was 120V. Nitrogen was used as drying gas at a flow rate of 13 liters/min and a temperature of 350°C. Nebulizer pressure is set at 60 psi. [M-H]⁻ ion was monitored in the selected-ion monitoring (SIM) mode (m/z=303 for arachidonic acid, m/z=281 for oleic acid, and m/z=269 for 17:0 heptadecanoic acid).

Data are expressed as the means ± standard error of the mean. Experiments were performed in triplicate. Student's t-test was performed using GraphPad Prism (version 5.01; GraphPad Software, Inc.), and P<0.05 was considered to indicate a statistically significant difference.

Tumor tissues and adjacent non-tumor tissues were collected from patients with HCC who underwent surgical resection between 2015 and 2016. The detailed clinical parameters of the 67 patients are listed in Table I. The mean age of the patients was 52.6±11.3 years (range, 18–76 years). In total, 83.6% of the patients were male, consistent with prior reports that HCC is more prevalent in males compared to females. It was reported that estrogen might repress HCC growth via inhibiting alternative activation of tumor-associated macrophages (17). Of the 67 samples collected, 65 were HBV infections, with a ratio of 92.5%. It was consistent with the fact that >80% of patients with HCC were HBV carriers in Fujian Province. As there was an insufficient number of HCC samples without HBV infection in the current study, the effect of HBV on the expression of these endogenous lipids and its role in promoting HCC was not investigated. Approximately 10% of the patients were recorded to have varying degrees of liver damage, and 23 patients had cancerous thrombi. In the cohort, ~55% was diagnosed with intermediate-grade cancer, while the percentage of high- grade was 25.4% and that of low-grade cases was 19.4% according to the Standardization of diagnosis and treatment for hepatocellular carcinoma (2017 edition, Bureau of Medical Administration, National Health and Family Planning Commission of the PRC) (18).

According to the current data, 38 patients (56%) have a recorded pathological status of cirrhosis, and 14 patients of them had Child-Pugh score. A total of 9 patients were recorded as HBV-related decompensated liver cirrhosis, 7 of them had a higher Child-Pugh score (10.29±0.68), while 2 of them had a lower Child-Pugh score (5 and 6, respectively). A total of 4 patients were recorded as decompensated alcoholic cirrhosis with Child-Pugh scores of 5, 9, 12, and 13, respectively. And only one patient was recorded as cardiogenic cirrhosis with a Child-Pugh score of 5.

The present study also included 7 patients who were HCV RNA positive and 6 patients were HCV RNA negative. There are no relevant records for the other patients. Some patients showed dyslipidemia, 5 of them were diagnosed with fatty liver and hypertriglyceridemia, 3 patients were diagnosed with hypercholesterolemia, and only 1 patient was suspected as nonalcoholic steatohepatitis.

The endogenous lipids in 67 tumor and matched non-cancerous tissue pairs were detected and quantified by HPLC-MS. Compared with the non-tumor counterparts, AEA levels were significantly decreased in tumor tissues (P=0.0329; 188.3±22.94 pmol/g in non-tumor tissues, and 131.7±12.04 pmol/g in HCC, Fig. 1A), while the level of 2-AG, another important endocannabinoid, was significantly increased in tumor tissues (P=0.0278; 68.55±10.64 nmol/g in non-cancerous controls, and 113.3±17.48 nmol/g in HCC, Fig. 1B). As congeners of AEA, palmitoylethanolamide (PEA) and oleoylethanolamide (OEA) belong to the superfamily of N-acylethnolamines (NAEs), however PEA and OEA interact with the PPAR-α receptor instead of cannabinoid receptors (19). The expression of PEA and OEA did not differ significantly between the two groups (Fig. 1C and D).

The most notable endocannabinoid changes in the present study were the decrease of AEA and increase of 2-AG. AEA is generated by NAPE-PLD and metabolized by FAAH (20). The checkpoint in 2-AG synthesis is considered to be DGL-α (21). Furthermore, MAGL is mainly involved in 2-AG degradation (21). To further investigate the molecular mechanisms underlying the aforementioned changes, the expression and activity profiles of these enzymes were examined in tissue samples.

There was no significant difference in the mRNA expression of NAPE-PLD between tumors and non-cancerous controls (Fig. 2A). The mRNA levels and activity of FAAH increased in HCC samples relative to their non-tumor counterparts (P=0.041; Fig. 2B and C), which suggested increased degradation of AEA in tumor tissues. A 5-fold higher level of DGL-α mRNA was detected in the tumor group (1.05±0.09 in non-tumor tissues and 5.27±0.81 in HCC; P<0.0001; Fig. 2D). The expression of DGL-α exhibited a greater increase compared with MAGL (Fig. 2E and F), indicating that 2-AG synthesis was markedly faster than its degradation. Whether the elevated expression of 2-AG and DGL-α may be used as markers for early HCC diagnosis warrants further investigation.

It has been reported that cannabinoid receptors 1 and 2 mediate functional responses to the AEA and 2-AG. As the levels of these two endocannabinoids changed significantly in tumor tissues, the present study investigated whether there were corresponding changes in the receptors. The mRNA expression of CB₁ and CB₂ was examined by quantitative PCR in both tumor and non-tumor samples. There was no significant difference between the two groups for CB₁ mRNA levels (Fig. 3A), while CB₂ expression was increased in tumor tissues (Fig. 3C). The protein levels of the two receptors were further investigated via western blot analysis. ImageJ software was used to calculate the optical density of the blots, using β-actin as the control. The results revealed that CB₁ protein levels in tumor tissues were decreased significantly (1.45±0.054 in non-cancerous tissues and 0.74±0.086 in tumor tissues; P=0.0022; Fig. 3B), and those of CB₂ were significantly increased (1.37±0.088 in non-tumor tissues and 2.06±0.2 in tumor tissues; P=0.0336, Fig. 3D). The trends in receptor protein expression were consistent with the changes in endogenous ligands.

Ceramides are a family of bioactive sphingolipids with tumor-suppressive properties (22). In the present study, HPLC-MS detection revealed that endogenous C12:0, C16:0, C18:1 and C24:1-ceramides were markedly increased in HCC tissues, whereas the levels of C18:0, C20:0 and C24:0-ceramides were significantly decreased in these samples, each compared with their adjacent normal counterparts (Fig. 4). The differences in C22:0-ceramide levels between tumor and non-tumor tissues were not significant. It was documented that C18:0-ceramide exerted anti-proliferative effects on head and neck squamous cell carcinomas (HNSCCs), while the roles of C18:0, C20:0 and C24:0-ceramides in the pathogenesis and progression of HCC require further investigation.

To examine the mechanisms underlying the changes observed with ceramides, the mRNA levels of CerS1-6 were examined in 67 pairs of HCC tumor and adjacent non-tumor tissues (Fig. 5A-F). The results revealed that the mRNA levels of CerS6 increased significantly in ~70% of the patients, which were associated with increased tumor levels of C16:0-ceramide. Furthermore, the decreased mRNA levels of CerS1 were in accordance with reduced C18:0-ceramide levels, which was highly relevant to the lower levels of C18:0-ceramide observed in HCC tissues. Decreases in C20:0 and C24:0-ceramides in tumor tissues may be attributed to the downregulated expression of CerS4, a synthase responsible for the synthesis of C18:0-24:0 ceramides (22). Notably, the levels of mono-unsaturated C18:1 and C24:1 ceramides were increased in tumor samples, consistently with the elevated levels of dihydroceramide desaturase 1 and 2 (DEGS-1/2) (Fig. 5G and H). However, the main role of DEGS and mono-unsaturated ceramides involved in HCC formation and progression warrants further investigation.

SCD1 is a transmembrane protein that converts SFAs to Δ-9 MUFAs to supply phospholipids for membrane biogenesis during cancer cell proliferation (23). The SCD1 mRNA and protein levels were significantly upregulated in HCC tissues compared with their non-tumor counterparts (Fig. 5I). High expression of SCD1 may be associated with the suppression of C18:0, 20:0 and 24:0-ceramide de novo synthesis, while the crosstalk between the SCD1 pathway and the ceramide metabolism network has, to the best of our knowledge, yet to be investigated.