Potential conflict of interest: Nothing to report.
Supported by a grant from Buergermeisterfonds, Vienna, Austria (to W.S.), by the Ludwig Boltzmann Gesellschaft LBG, the Austrian Science Fund FWF grant SFB F28 and DK-plus grant IAI (to R.E.), and the Austrian Federal Ministry of Science and Research GENAU grant “Austromouse” to (R.E.).
Activation of the activator protein 1 (AP-1) transcription factor as well as increased serum levels of vascular endothelial growth factor (VEGF) and interleukin (IL)-8 predict poor prognosis of patients with hepatocellular carcinomas (HCCs). Moreover, HCC patients display reduced selenium levels, which may cause lipid peroxidation and oxidative stress because selenium is an essential component of antioxidative glutathione peroxidases (GPx). We hypothesized that selenium-lipid peroxide antagonism controls the above prognostic markers and tumor growth. (1) In human HCC cell lines (HCC-1.2, HCC-3, and SNU398) linoleic acid peroxide (LOOH) and other prooxidants enhanced the expression of VEGF and IL-8. LOOH up-regulated AP-1 activation. Selenium inhibited these effects. This inhibition was mediated by glutathione peroxidase 4 (GPx4), which preferentially degrades lipid peroxides. Selenium enhanced GPx4 expression and total GPx activity, while knock-down of GPx4 by small interfering RNA (siRNA) increased VEGF, and IL-8 expression. (2) These results were confirmed in a rat hepatocarcinogenesis model. Selenium treatment during tumor promotion increased hepatic GPx4 expression and reduced the expression of VEGF and of the AP-1 component c-fos as well as nodule growth. (3) In HCC patients, increased levels of LOOH-related antibodies (LOOH-Ab) were found, suggesting enhanced LOOH formation. LOOH-Ab correlated with serum VEGF and IL-8 and with AP-1 activation in HCC tissue. In contrast, selenium inversely correlated with VEGF, IL-8, and HCC size (the latter only for tumors smaller than 3 cm). Conclusion: Reduced selenium levels result in accumulation of lipid peroxides. This leads to enhanced AP-1 activation and consequently to elevated expression of VEGF and IL-8, which accelerate the growth of HCC. Selenium supplementation could be considered for investigation as a strategy for chemoprevention or additional therapy of early HCC in patients with low selenium levels. (HEPATOLOGY 2012)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide. Solid tumor growth depends on neoangiogenesis. Microvessel density within the tumor is an independent prognostic marker and predictor of HCC recurrence.1 A variety of cytokines is involved in neoangiogenesis, but clinical relevance in HCC was shown only for interleukin (IL)-8 and vascular endothelial growth factor (VEGF). An enhanced concentration of serum IL-8 is associated with poor prognosis.2 Similarly, increased VEGF is linked to earlier HCC recurrence and shorter overall survival.3-5
Both VEGF and IL-8 gene expression are regulated by the transcription factor activator protein 1 (AP-1). AP-1 binding sites were found in the promoter regions of VEGF and IL-8 genes.6, 7 AP-1 is supposed to be involved in tumor formation and angiogenesis8 and can be activated by oxidative stress.9 Oxidative stress is defined as an imbalance between production and antioxidative elimination of reactive oxygen species (ROS). Moreover, oxidative stress is a common feature of inflammatory liver diseases that predispose to cancer10, 11 and is associated with a higher incidence of HCC recurrence in hepatitis C patients.10
Selenium counteracts oxidative stress because selenoproteins such as glutathione peroxidases (GPx) eliminate ROS.12, 13 Low selenium levels are associated with an increased cancer risk including HCC.14-18 The liver is particularly affected under selenium deficiency because other organs such as brain, testis, and endocrine tissues are supplied preferentially with selenium.19
GPx2 and GPx4 are supposed to attenuate cancer development.20 GPx4 is the only known enzyme that is efficiently able to reduce lipid peroxides13 formed through ROS-mediated oxidation of unsaturated lipids. Lipid peroxides elevate AP-1 activity and VEGF formation in colorectal cancer cells.21 Likewise, in cultured HCC cells we found an increase of AP-1 components c-jun and c-fos by lipid peroxides.22 We hypothesized that AP-1 activation as well as expression of its target genes VEGF and IL-8 in HCC are controlled by the selenium/lipid peroxide antagonism. The results of the present study support this hypothesis by evidence gathered from cell lines, an HCC animal model, and HCC patients.
Selenium was quantified by an inductively coupled plasma mass spectrometer (ICP-MS). Details are given in Supporting Methods.
Twenty-nine adult patients with histologically confirmed HCC (Supporting Table 1) underwent orthotopic liver transplantation at the General Hospital of Vienna, Austria. HCC tissue arrays were constructed.23 Healthy persons from the local population without known liver disorders were used as controls. Data analysis was performed with the permission of the local Ethics Committee. Correlations were also calculated from published microarray human HCC data.24
HCC in Fisher-344 rats (Charles River) was initiated by 200 mg/kg intraperitoneal diethylnitrosamin (DEN) as described.25 Promotion was performed by 0.02% dietary 2-acetylaminofluoren (2-AAF) for 4 days, followed by a 2/3 partial hepatectomy (PH) and two intragastric injections of 2-AAF (20 mg/mL in agar) 2 and 4 days after PH. Fifty mg selenite / 100 g body weight was administered by way of drinking water. In the promotion study, selenium exposure started 1 week before 2-AAF feeding until sacrifice at days 7 and 21 post-PH. In the progression study, selenium exposure was for 3 months starting 3 weeks after PH.
Primary Cells and Cell Lines.
Primary human hepatocytes were obtained from LONZA (Basel, Switzerland). Primary rat hepatocytes were isolated.26 HCC-1.2 and HCC-3 cell lines were established in our laboratory27; SNU398 cell line was purchased from ATCC (LGC Standards, Wesel, Germany). The cell lines were kept under standard tissue culture conditions. Fifty nM of sodium selenite (Sigma-Aldrich) was added 24 hours before treatment. Synthesized linoleic acid hydroperoxides (LOOH)28 was dispersed by sonication into serum-free medium containing 1 mg/ml fatty acid-free bovine serum albumin (BSA). ROS was quantified by the 2′,7′-dichlorofluorescin diacetate (DHFC) method.21
Detection of Linoleic Acid Hydroperoxide-Related Antibodies (LOOH-Ab).
LOOH-Ab were detected in plasma according to the modified method of Rolla et al.29; 1 mM DTPA (Sigma-Aldrich) was added to washing phosphate-buffered saline (PBS) (Invitrogen, Carlsbad, CA).
Nuclear c-jun Localization.
HCC tissue arrays were stained for c-jun by immunohistochemistry, counterstained with hematoxylin, and scanned using TissueFaxs software (TissueGnostics, Vienna, Austria). Nuclear localization of c-jun was evaluated using HistoQuest software (TissueGnostics). Proper recognition of nuclei by the hematoxylin nuclear mask was confirmed prior to quantification of c-jun nuclear intensity.
RNA was isolated according to a standard Trizol-extraction protocol (Invitrogen, Austria). Complementary DNA (cDNA) was synthesized using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) and assessed for gene expression with the real-time RT-PCR TaqMan System using the following primers: Hs00173626_m1 for VEGF, Hs00174103_m1 for IL-8, Hs01591589_m1 for GPx2, and Hs00157812_m1 for Gpx4 (Applied Biosystems). The ΔΔCt method was applied for quantification.
GPx4 Activity and Western Blotting.
Total GPx activity in cell lysates was measured as described.30 Western blotting was performed as described.31 More details are given in the Supporting Materials.
VEGF and IL-8 Protein Measurements.
Human serum VEGF and IL-8 were determined by Quantikine enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, Abingdon, UK) and IL-8 Human ELISA Kit (BenderMedSystems, Vienna, Austria), respectively, according to the manufacturers' instructions.
AP-1 and Hypoxia Inducible Factor 1 (HIF-1) Binding Assay.
AP-1 and HIF-1 DNA binding was measured in nuclear extracts by TransAM transcription factor ELISA (Active Motif Europe, Rixensart, Belgium) according to the Instruction Manual.
All cellular experiments were performed at least three times. If not indicated otherwise, data are expressed as mean ± standard error of the mean (SEM), and statistical differences were determined using analysis of variance (ANOVA) with significance accepted at P < 0.05.
Increase of ROS, VEGF, and IL-8 by LOOH and Inhibition by Selenium in HCC Cells.
Previous studies in cultured colon cancer cells showed that treatment with LOOH caused intracellular ROS formation and VEGF synthesis.21 We tested whether cultured HCC cells respond to LOOH in a similar way, and whether selenium can inhibit ROS formation and VEGF synthesis. Treatment of HCC-1.2 cells with LOOH increased intracellular ROS formation, which was inhibited by selenite. Selenite alone had no effect on ROS formation (Fig. 1A). These data suggest that selenium protects from increased ROS formation induced by LOOH.
VEGF and IL-8 were released by HCC cells (Supporting Table 2). The expression was induced by LOOH in HCC cells (Fig. 1B,C) and in primary human hepatocytes (Supporting Fig. 1A,B). Selenium decreased LOOH-induced VEGF and IL-8 expression in HCC-1.2 and SNU398 cells (Fig. 1B,C), but only marginally in HCC-3 cells (data not shown). No VEGF or IL-8 induction was observed with nonoxidized linoleic acid (Fig. 1B,C), supporting the importance of peroxidized linoleic acid in this activation.
In order to test if intracellular ROS accumulation is responsible for increased IL-8 and VEGF expression, we evaluated the ability of other known ROS sources to induce these cytokines.
Consistently, VEGF and IL-8 expression was induced in HCC cells by the ROS sources copper, hydrogen peroxide, and sodium hypochlorite (Fig. 2A,B). The same effect was observed in primary rat hepatocytes except for hypochlorite (Supporting Fig. 1C). The ROS scavenger N-acetylcystein reduced the induction of VEGF and IL-8 expression by LOOH (Fig. 2C). These data indicate that the increase of intracellular ROS is responsible for up-regulation of VEGF and IL-8 in HCC cells and primary hepatocytes.
Effects of LOOH and Selenium on DNA Binding of Transcription Factors HIF-1α and AP-1.
Hypoxia inducible factor 1α (HIF-1α) and AP-1 are transcription factors that regulate VEGF expression in response to oxidative stress.32 We investigated whether DNA binding activities of HIF-1α or AP-1 are enhanced by LOOH. LOOH treatment did not increase HIF-1α DNA binding in HCC-1.2 and HCC-3 cells (Fig. 3A,B). Thus, LOOH-induced VEGF expression in HCC cells is not due to HIF-1α activation.
In contrast, DNA binding of the transcription factor AP-1 was significantly enhanced after exposure to LOOH but not to nonoxidized linoleic acid (Fig. 3C,D). Selenium reduced LOOH-mediated AP-1 activation substantially in HCC-1.2 and moderately in HCC-3 cells.
Selenium Treatment Blocks VEGF and IL-8 Increases by Way of Induction of GPx4.
Accumulation of ROS and particularly of lipid peroxides is antagonized by GPx2 and GPx4. In HCC cells, expression of GPx4 was higher than of GPx2 (Fig. 4A,B). Selenium further enhanced GPx4 RNA and protein (Fig. 4A-D), whereas GPx2 expression remained unchanged (Fig. 4A,B). Raw values of GPx expression are listed in Supporting Table 3. Total GPx activity was also increased by selenium treatment (Fig. 4).
Knockdown of GPx4 expression by small interfering RNA (siRNA) increased VEGF and IL-8 at the messenger RNA (mRNA) and protein level (Supporting Table 4). These data indicate that VEGF and IL-8 production in HCC cells is controlled by selenium through a GPx4-dependent mechanism.
Effect of Selenium on VEGF, Chemokine (C-X-C motif) Ligand 1 (CXCL1) (IL-8 Analog), and AP-1 Activity in a Rat Model of Hepatocarcinogenesis.
Next we tested whether selenium levels modulate AP-1 activity, VEGF, and IL-8 also in the animal organism and affect growth of early tumor stages. Because the IL-8 gene is not conserved in rats, its analog CXCL1 was investigated. The Solt-Farber model of rat hepatocarcinogenesis was used with and without selenium supplementation.
Selenium supplementation increased serum selenium levels (Table 1). In the promotion phase, cell proliferation as well as volume of preneoplastic liver nodules were reduced from 38% to 14% volume fraction in the selenium-supplemented rats.25 Hepatic mRNA expression of VEGF and c-fos was reduced in the promotion but not in the progression phase (Table 1). Nuclear translocation of c-jun and expression of CXCL1 were not influenced by selenium (Table 1). Serum VEGF and CXCL1 proteins were below the detection limit of commercially available ELISAs. Thus, in this rat model selenium supplementation decreases VEGF and c-fos expression as shown above in vitro; this effect is associated with a dramatic reduction of nodule growth.
Table 1. Impact of Selenium Supplementation on Gene Expression and AP-1 Activation in DEN-Induced Model of Rat Liver Carcinogenesis
Percent of Control Without Selenium Supplementation
Serum selenium was measured in ng/ml; GPx4 protein is given as a quantified density of western blot bands adjusted for housekeeping gene GAPDH; percentage proportion of positive cells is given for nuclear c-jun; original expression values adjusted to house keeping gene GAPDH are given for VEGF, c-fos and CXCL1.
HCC Patients: LOOH-Ab and Selenium Levels as Related to VEGF, IL-8, AP-1, and Tumor Size.
Finally, we analyzed the effects of selenium and LOOH on growth factors and tumor size in patients with HCC. LOOH-Ab in blood plasma were determined similar to work published previously.33-35 Interestingly, LOOH-Ab levels were significantly higher in HCC patients than in healthy controls (Fig. 5A), suggesting higher amounts of circulating LOOH. Selenium levels inversely correlated with VEGF and IL-8 and also with tumor size in HCC patients, the latter only in those with tumors diameters up to 3 cm (Table 2; Fig. 5B). LOOH-Ab levels correlated positively with VEGF (only in patients with HCC <3 cm) and IL-8 and also with nuclear localization of c-jun indicating AP-1 activation (Table 2; Fig. 5C,D). These correlations in HCC patients are consistent with the above finding that LOOH enhances VEGF and IL-8 expression and AP-1 activation in cultured HCC cells, and that selenium antagonizes these effects.
Table 2. Correlation Coefficients Between Systemic Oxidative Stress Parameters, Tumor Diameter, and AP-1 Activation in HCC Patients
Finally, we reevaluated published gene expression data from HCC tissue of 60 patients.24 GPx4 but not GPx2 inversely correlated with VEGF and c-fos expression. GPx correlations with IL-8 and c-jun expression were not statistically significant, but VEGF positively correlated with IL-8 (Supporting Table 5). These data agree with the inhibitory role of the selenium-inducible GPx4 on VEGF expression in HCC cells found in vitro (see above).
Inflammation and associated formation of ROS are widely accepted risk factors in hepatocarcinogenesis but important mechanistic details are still unknown. Here we report that peroxides of linoleic acid (LOOH) can activate the transcription factor AP-1, a sensor of oxidative stress36-38 and important promoter of hepatocarcinogenesis.39-42 Moreover, LOOH and other inducers of oxidative stress enhanced expression of AP-1 targets VEGF and IL-8 which are considered as key factors for angiogenesis and growth of HCCs (present study).2-5, 43
Oxidized lipids are immunogenic44, 45 and antibodies against lipid peroxidation products are supposed to reflect systemic oxidative stress. Antibodies against oxidized lipids have been shown to correlate with the amount of lipid peroxidation products.46 Therefore, patients with oxidative stress-associated liver diseases have higher titers of lipid peroxidation-related antibodies.33-35 We observed higher levels of LOOH-Ab in HCC patients when compared to controls, which is consistent with the expected ROS-mediated increase in lipid peroxidation under inflammatory conditions. The increase in LOOH levels could be partly due to elevated levels of free fatty acids resulting from obesity and metabolic syndrome, which are increasing risk factors of hepatocarcinogenesis.47, 48 Free fatty acids and ROS might act synergistically to increase lipid peroxides, thereby leading to the observed AP-1 activation and increased expression of VEGF and IL-8. However, lipid peroxidation products from LOOH decomposition or from LOOH-initiated membrane lipid peroxidation22 could also be involved. Interestingly, a positive correlation between LOOH-Ab and VEGF levels was only seen in patients with small HCCs, suggesting that VEGF production is regulated by alternative mechanisms in more advanced liver tumors.
In addition to HCCs, oxidative stress might provoke similar molecular effects in other tumor cells. VEGF was induced by oxidative stress in hepatitis C-infected HUH7 cells49 and in immortalized 3T3 fibroblasts.50 Oxidative stress induced VEGF and IL-8 in an AP-1-dependent manner in breast carcinoma cells.38
The LOOH-mediated HCC-promoting molecular effects were antagonized by the antioxidant selenium. Selenium decreased the LOOH-induced AP-1 binding to DNA in cultured HCC cells and the subsequent induction of VEGF and IL-8 expression. These selenium effects were shown to be mediated, at least in part, by the selenoenzyme GPx4, which is specifically implicated in the decay of lipid peroxides. We demonstrated that GPx4 expression in HCC is induced by selenium treatment, which is consistent with data in normal rat liver.51 Increased GPx4 levels were associated with reduced VEGF and AP-1/c-fos expression and with a decline in tumor growth. Importantly, selenium levels inversely correlated with VEGF and IL-8 serum levels and tumor size in HCC patients. Moreover, expression of GPx4 inversely correlated with expression of VEGF and AP-1/c-fos, supporting the significance of our findings for human patients.
Selenium is also an inhibitor of VEGF and IL-8 expression in other tumor types.52 In rat mammary tumors, selenium treatment impaired angiogenesis by way of a VEGF-dependent mechanism.52-54 In leukocytes,55 epithelial cells,57 and hepatoblastomas,56 selenium has been reported to inhibit IL-8 expression. Moreover, selenium has been described to inhibit AP-1.55, 58, 59 These results suggest that the proposed molecular interaction between LOOH, AP-1 activity, and expression of VEGF/IL-8 as well as the antagonistic effect of selenium/GPx4 (Fig. 5E) could be important not only for development of HCCs but also for other tumor types.
The inverse correlation between selenium levels and tumor size described here in HCC patients is consistent with several epidemiologic studies. An inverse relation between plasma selenium levels and HCC risk was observed in Taiwan.18 Based on previous animal experiments60 an intervention trial was performed in Quidong/China, a region with low selenium intake. Daily doses of 200 μg selenium decreased HCC rates by 35% and cessation of selenium supplementation brought tumor rates back to initial values.17, 60-62 Consistently, an intervention study in the USA demonstrated protection by selenium against prostate cancer.63 In contrast, the more recent SELECT study did not show any benefit of selenium supplementation.64 This might, however, be due to the high baseline plasma levels of selenium observed in this study that could conceal potentially beneficial effects of selenium supplementation.
Although comparison of selenium levels between different studies is difficult because of inconsistent methodologies, conclusions can be drawn from environmental parameters. In particular, low selenium concentrations in the serum have been documented for the Austrian population that are due to low selenium in the soil.65
In conclusion, the mechanistic data in the present study support the notion that the inverse correlation between selenium levels and the risk to develop HCC may have a causal basis. Therefore, selenium supplementation could be considered a strategy for chemoprevention or additional therapy for HCC patients with low selenium levels.
We thank M. Seif, E. Hangelmann, M. Eisenbauer, and N. Kandler for excellent technical assistance, M. Vidali for help in optimization of LOOH-Ab detection, M. Jakupec for help in selenium quantification, B. Marian for critical reading of the article, and A. Kaider for statistical evaluation of the data.