|Home | About | Journals | Submit | Contact Us | Français|
Justin B. Greer, Division of Marine Biology and Fisheries, Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, 4600 Rickenbacker Cswy, Miami, FL, USA 33149, Phone: 305-421-4906, FAX: 305-421-4600, jgreer/at/rsmas.miami.edu (J. B. G.)
Fujiang Guo, Current address for Fujiang Guo is School of Pharmacy, Shanghai University of Traditional Chinese Medicine, 1200 Cailun Road, Shanghai, 201203, China, gfj/at/shutcm.edu.cn
Douglas C. Crawford, Division of Marine Biology and Fisheries, Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, 4600 Rickenbacker Cswy, Miami, FL, USA 33149, Phone: 305-421-4121, FAX: 305-421-4600, dcrawford/at/rsmas.miami.edu
Kathleen S. Rein, Department of Chemistry and Biochemistry, 11200 SW 8th St, Florida International University, Miami, FL, USA 33199. Phone: 305 348-6682, FAX: 305 348-3772, reink/at/fiu.edu
The marine toxin, okadaic acid (OA) is produced by dinoflagellates of the genera Prorocentrum and Dinophysis and is the causative agent of the syndrome known as diarrheic shellfish poisoning (DSP). In addition, OA acts as both a tumor promoter, attributed to OA-induced inhibition of protein phosphatases as well as an inducer of apoptosis. To better understand the potentially divergent toxicological profile of OA, the concentration dependent cytotoxicity and alterations in gene expression on the human liver tumor cell line HepG2 upon OA exposure were determined using RNA microarrays, DNA fragmentation, and cell proliferation assays as well as determinations of cell detachment and cell death in different concentrations of OA. mRNA expression was quantified for approximately 15,000 genes. Cell attachment and proliferation were both negatively correlated with OA concentration. Detached cells displayed necrotic DNA signatures but apoptosis also was broadly observed. Data suggest that OA has a concentration dependent effect on cell cycle, which might explain the divergent effects that at low concentration OA stimulates genes involved in the cell cycle and at high concentrations it stimulates apoptosis.
Okadaic acid (OA) and the related dinophysistoxins (DTX) are phycotoxins produced by several marine dinoflagellates belonging to the genera Prorocentrum and Dinophysis. Dinoflagellates are a highly diverse group of flagellated, unicellular protists which are responsible for the majority of toxic harmful algal blooms (HAB). Filter feeding shellfish can accumulate the toxin and the consumption of OA-contaminated shellfish results in a syndrome known as diarrheic shellfish poisoning (DSP) which is characterized by severe gastrointestinal symptoms (Reguera and Pizarro 2008). DSP has been associated with the consumption of mussels, scallops or clams tainted with OA and its derivatives. Outbreaks of DSP have been reported in the Americas, Asia and Europe (Reguera and Pizarro 2008; Swanson et al. 2010; Deeds et al. 2010). The effects of chronic exposure to OA may be of greater concern, as it has been demonstrated to be a potent tumor promoter that differs from the phorbol ester class (Fujiki and Suganuma 1993; 2009).
The adverse effects of OA exposure are varied and cell-type and concentration-dependent (Gehringer 2004). OA is a potent inhibitor of the serine/threonine protein phosphatases PP-1, PP-2A (Fernandez et al. 2002; Dawson and Holmes 1999), PP-4 and -5 (Zhang et al 1994) and PP-6 (Prickett and Brautigan 2006). It has been suggested that most or all of the toxic effects of OA may be attributed to the resulting hyperphosphorylation of numerous proteins leading to a loss of regulation of multiple intracellular processes, including cell cycle regulation, maintenance of cell shape, cystoskeletal disruption, cell motility, transport of vesicles, and caveolae (Sheppeck et al. 1997; Lontay et al. 2005; Honkanen and Golden 2002; Kuroda et al. 2007; Botos et al. 2007; Kiss and Botos 2009; Vale and Botana 2008; Blankson et al. 1995; Jayaraj et al. 2009). The commercial availability of OA isolated from cultures of Prorocentrum concavum has led to its prominent role in cell biological research, where it is often employed to enhance the activity of protein serine/threonine kinases such as the microtubule associated protein (MAP) kinases (Gómez et al. 1990), mitogen-activated protein kinase kinase (MEK) and extracellular-signal-regulated protein kinase (ERK) ( Barford 1996) in studies of cell cycle, apoptosis, nitric oxide metabolism and calcium signaling.
OA may act both as a tumor promoter and inducer of apoptosis in its role as a phosphatase inhibitor, via activation or inactivation of intracellular signaling cascades that are dependent on the phosphorylation state of proteins. In primary cell cultures OA exposure resulted in hyperphosphorylation of both p53 and the retinoblastoma protein (Yatsunami et al. 1993). After acute OA exposure, hyperphosphorylation of p53 led to cell cycle arrest and apoptosis (Yan et al. 1997), while hyperphosphorylation of the retinoblastoma protein led to upregulation of genes involved in DNA replication and cell proliferation (Gehringer 2004). Which of these opposite effects predominates in different human cell types has begun to be addressed (Valdiglesias et al. 2010; 2011a, 2011b, 2011c).
A growing body of evidence suggests that OA and its metabolites produced in the liver exhibit genotoxic effects (Fessard et al. 1996; Tohda et al. 1993; Pinto-Silva et al. 2005; Carvalho et al. 2003; Le Hegarat et al. 2003, 2004; Valdiglesias et al. 2010). The dependence of these effects on specific phosphorylation events has not been studied. OA exerts toxic effects independent of phosphatase inhibition, producing oxygen free radicals that impair protein synthesis (Matias et al. 1999).
In mice, OA was distributed equally to all organs but appeared to have the longest residence time in liver (Matias et al. 1996, 1999; Ito et al. 2002). The aim of this investigation was (1) to expand these studies and those of Valdiglesias et al. (2010, 2011a, 2011b, 2011c) in the human liver tumor model, HepG2, (2) examine the gene expression profile of exposure to OA and (3) correlate effects on cell cycle and apoptosis with concentration.
Okadaic acid used in this study was isolated from cultures of Prorocentrum hoffmanianum according to published methods (Hu et al. 1992). The isolated OA was dried to a constant weight, quantitated gravimetrically and compared by HPLC-MS to OA that was purchased from Calbiochem (San Diego, CA). Isolated OA was found to be >96% pure by HPLC-MS and identical to commercial OA.
HepG2 cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA) and maintained in normal media consisting of Eagle’s Minimal Essential Medium (ATCC) with 10% fetal bovine serum (Atlanta Biologicals, Atlanta, GA) and 100 U/ml penicillin plus 100 μg/ml streptomycin (Gibco) in a 5% CO2 humidified atmosphere at 37°C. Cells were passaged via trypsinization (0.05%; CellGro, Mediatech, Inc., Manassas, VA) into vented T25 flasks (BD Biosciences, Bedford, MA). Cells for an experiment were plated from the same mixture of donor cells at 1 × 106 cells in 5 ml media for bromodeoxyuridine (BrdU), RNA or DNA extraction experiments. Time zero counts for each flask (T0), representing the number of cells just before toxin addition, were calculated approximately18 hr after plating by averaging the cells counted in 3 × 1.08 mm2 grids along the midline of the flask’s long dimension and scaling up to the 25 cm2 area of the flask. Following T0 cell counts OA was added via substitution of 1 ml of media with 1 ml of a 5 fold working stock in media from a serially diluted stock of 620 μM OA in media. Calculated initial OA concentrations in triplicate flasks were 6.2, 41, 62, 206 or 620 nM. Media in control flasks was substituted with 1 ml of normal media without OA. Numbering of control and OA-exposed flasks was randomized by a colleague not involved in the study and assessments of cell morphology and cell numbers were made without knowing the OA concentration before breaking the code at the conclusion of experiments.
Cell morphological changes such as process retraction and cell rounding, and granular appearance suggesting impending cell death were assessed by inspection of cultures at 200× and scored as either present, if noted in >10% of attached cells, or absent, if present in <10% of attached cells. Cell detachment was graded by focusing the microscope objective on the surface of the media where these cells accumulated. Detachment was scored as absent or present at either a low or high extent based on this observation, as well as on cell counts (see Tfinal, below) and cell morphological assessments.
Tfinal (Tf) cell counts, representing the end of toxin exposure, were made on live cells after 48 hr exposure to OA according to the method described above. Cell-free spent media was frozen at −20°C for quantitation of OA and OA metabolites. The cells were harvested for either RNA or DNA extraction, or the cells adherent to the flask were fixed for BrdU assays (see below).
After 24 hr incubation with OA, HepG2 cells were exposed to BrdU (10 μl/ml diluted from a concentrated stock solution of 10:1 5-‘bromo-2′-deocyuridine:5-fluoro-2′-deoxyuridine according to the manufacturer’s recommendations; Invitrogen). After an additional 24 hr, cells were fixed in the flasks with ethanol-glycine and Tf cell counts were made. BrdU positive cells were identified using a biotinylated antibody, mouse anti-BrdU (Calbiochem, San Diego, CA) and visualized using streptavidin-horseradish peroxidase and diaminobendizine (DAB; Invitrogen, Camarillo, CA). Total cell numbers and % BrdU positive cells were determined by averaging cell numbers in 3 × 0.27mm2 grids along the flask midline, then scaling to flask Tf cell numbers. Flasks with no BrdU and flasks with no primary antibody served as negative controls.
After OA exposure, the media was removed, and HepG2 cells scraped from the substrate using a sterile cell scraper in 2 ml of digestion solution (50 mM Tris pH 8, 200 mM NaCl, 100 mM EDTA pH 8, 1% SDS, 0.2% DTT, 2 mg/ml Proteinase K). For exposures resulting in substantial cell detachment, cells were recovered from the culture media via centrifugation (250 × g for 8–10 min) and the resulting cell pellet added to scraped cells from the same flask. Samples were digested overnight at 55°C in a water bath. DNA was extracted from cells using standard phenol-choroform extraction followed by precipitation in isopropanol. Purified DNA was resuspended in Tris EDTA and subsequently treated with 1–2 μl RNaseA (1mg/ml; Sigma-Aldrich), depending on volume of sample, for 20 min on ice to remove degraded RNA, then spectrophotometrically analyzed for concentration. Electrophoresis was performed on a 1.2% agarose gel at 80 V for 90 min. DNA gel-loading volumes were 2 μg of sample DNA per well for the 620 μM OA time course (Figure 2A) and 3 μg of sample DNA for cells exposed OA concentrations of 200–607 nM and analyzed as adherent vs detached cell fractions, to enhance ladder visualization (Figure 2B). DNA fragments were stained with GelRed (Biotium, Hayward, CA) and imaged using a Gel Doc imaging system (Bio-Rad) with Q capture imaging software.
One ml of each sample of spent media was extracted with methylene chloride (4 × 1 ml). The combined extracts weredried under a stream of N2 and reconstituted in either 1 ml (highest 3 concentration samples) or 200 μl (control and lowest 2 concentration samples) of methanol. HPLC-MS analysis was carried out on a Finnigan Surveyor series HPLC system (Thermo Fisher Scientific Inc., Waltham, MA) using a 5 μm C18, 250 mm × 4.6 mm Discovery column (Supelco, Bellefonte, PA) coupled to a Finnigan LCQ Deca XP MAX mass ion-trap spectrometer (Thermo Fisher Scientific Inc., Waltham, MA) as described previously (Guo et al. 2010). Ten μl of extract was injected. OA was quantitated against a single point calibration curve generated by injecting 10 μl of a 62 nM solution of OA. The instrument’s limit of detection (LOD, defined as S/N = 3) was 0.1 ng for OA, and was presumed the same for OA metabolites. This detection limit was approached only in the case of the lowest concentration sample. OA concentrations presented in the Results and Discussion sections reflect those determined for spent media: 0, 2.4, 37, 51, 200 and 607 nM. Recovery of OA was determined by extraction from 0.5 ml of OA unused spiked media (125 nM and 6.2 nM OA concentrations) exactly as described above. Recoveries were calculated based on peak area compared to an OA standard. At 125 nM OA recovery was calculated to be 90%. However at 6.2 nM recovery was only 30%.
CYP (cytochrome P 450) 3A4 activity was assayed according to methods of Yoshitomi et al. (2001) and Omasa et al. (2005). Four T75 flasks were plated with 4 × 106 cells each in 15 ml of culture media. A 100 mM stock solution of testosterone (Sigma-Aldrich Chemical) in methanol was diluted in media to give 4 15× stock solutions (150 μM, 375 μM, 750 μM or 1.5 mM). One ml of culture media from each flask was replaced with 1 ml of testosterone containing media for final concentrations of 10, 25, 50 or 100 μM. Two ml aliquots of culture media were removed after 0.5, 1, 1.5 or 2 hr incubation for testosterone and 6β-hydroxytestosterone analysis. Each 2 ml aliquot was extracted with ethyl acetate (3 × 2 ml). The combined extracts were evaporated in vacuuo and reconstituted in methanol (1 ml). HPLC analysis was carried out on a Finnigan Surveyor series HPLC system (Thermo Fisher Scientific Inc., Waltham, MA) using a 5μm C18, 250 mm × 4.6 mm Discovery column (Supelco, Bellefonte, PA) and PDA (UV6000LP) detector monitoring at 244 nm. Twenty μl of sample was injected. The chromatographic conditions were as follows: flow rate of 1 ml/min; solvent A, CH3CN; solvent B, H2O; 0 – 5 min, 30% A; 5 – 10 min, 30–60% A; 10 – 20 min, 60% A; 20 – 25 min 60–90 % A. For additional confirmation, 10μl of the 100 μM, 2 hr incubation sample was also analyzed using the more sensitive HPLC-MS on a Finnigan Surveyor series HPLC system (Thermo Fisher Scientific Inc., Waltham, MA) using a 5 μm C18, 250 mm × 4.6 mm Discovery column (Supelco, Bellefonte, PA) coupled to a Finnigan LCQ Deca XP MAX mass ion-trap spectrometer (Thermo Fisher Scientific Inc., Waltham, MA) monitoring for m/z 288±1 (testosterone) and m/z 304±1 (6β-hydroxytestosterone). The chromatographic conditions were identical to those described above. Control reactions (2 ml) were conducted by incubating testosterone (100 μM) with human recombinant CYP 3A4 (50 pmol, Codexis Corp.) or human liver microsomes (1 ml, 1 mg/ml, BD Biosciences). The reaction mixtures were extracted with ethyl acetate (3 × 2 ml). The combined extracts were evaporated in vacuuo and reconstituted in methanol (1 ml) and analyzed by HPLC-PDA as described above. The instrument’s limit of detection (LOD, defined as S/N = 3) was 2 ng for testosterone.
After 48 hr exposure of HepG2 cells to OA or control solutions, culture media was removed from flasks and 2 ml chaotropic buffer (4.5M guanidinium thiocyanate, 2% N-lauroylsarcosine, 50 mM EDTA, 25 mM Tris-HCl, 0.1 M β-mercaptoethanol, and 0.2% antifoam A) was added. The cells were scraped off the bottom of the flask using a cell scraper, and removed by pipetting. At the two highest concentrations of OA some or all of the cells were detached. Spent media from these flasks was centrifuged at 250 × g for 5 min and the cell pellet was added to the scraped cells from the same flask. RNA was isolated using standard protocols (Chomczynski and Sacchi 1987). Isolated RNA was purified using a Qiagen RNeasy Mini Kit (Valencia, CA) and quantitated using a Nanodrop spectrophotometer (ThermoFisher, Waltham, MA). RNA quality was assessed with an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA).
Agilent G4112F Whole Human Genome 44K, Oligo microarrays were used to quantify mRNA expression. Hybridization used a loop design which is statistically superior to common reference design (Kerr and Churchill 2001; Churchill 2002) such that Control 37 nM 2.5 nM Control 37 nM 2.5 nM 51 nM 37 nM 200 nM 51 nM 607 nM 200 nM control 607 nM 200 nM control 37 nM 200 nM 607 nM 37 nM 2.5 nM 607 nM 2.5 nM 37 nM 51 nM 2.5 nM control 51 nM 51 nM control 2.5 nM 51 nM first sample (control); where each arrow represents an array with Cy3 labeled sample at its base and Cy5 labeled sample at the arrow head. Purified RNA was labeled using T7 linear amplification using a modified Eberwine protocol as described previously (Oleksiak et al. 2005; Whitehead and Crawford 2005, 2006a, 2006b; Crawford 2007; Scott et al. 2009). Two labeled reactions (Cy3 & Cy5) are re-suspended at 1.5 pmol/μl and 2μl /cm2 (or 100 μl for each 50 cm2 Agilent Array) of each reaction is used. Forty-eight hr hybridizations using Agilent reagents and protocols were employed. Following hybridization, slides were scanned using the Packard Bioscience ScanArray Express microarray scanner (PerkinElmer Life Sciences Inc., Boston, MA, USA) using a fixed power that maximizes signal without causing saturation. The resulting tiff images were imported into spot grids built in ImaGene (Biodiscovery Inc., Marina Del Ray, CA, USA), and spot signals were collected as fluorescence intensities for each dye channel.
Genes whose mRNA expression were below background or too highly expressed were excluded. Specfically, the fluorescent signals were considered to be too low if they were within 2 standard errors of background or too high if they saturated the photo-multiplier tube (i.e., equal to 216-1 = 65,535). Of the 43,376 experimental probes, 15,037 were greater than background or not saturated (i.e., signal intensities were less than 65,535). Statistical analyses of the mRNA expression data were carried out using JMP genomics (SAS JMP Genomics v.7.0.2). All analyses used fluorescent measures that were log2 transformed and loess normalized.
An ANOVA of the 15,037 log2-loess measures (yijk) for the significant differences in gene expression among individuals within each group used the linear mixed model (Kerr and Churchill 2001; Wolfinger et al. 2001; Patterson et al. 2006; Yu et al. 2004)
where μ is the sample mean, Ai is the effect of the ith array (i=1–17), Dj is the effect of the jth dye (Cy3 or Cy5), and Tm is one of 6 exposure to OA, and εijkm is the stochastic error. Each concentration was treated as a fixed effect, even though OA concentrations varied among replicates. With 6 treatments and 3 replicates per treatment the ANOVA used an F-table with 5 and 10 degree of freedom. The number of genes whose expression is significantly different across treatments ( p-value of 0.01) without multiple correction is provided.
Dunnett’s post-hoc test (JMP 7) was used to determine which concentrations differed from the control. “DAVID” at NCBI (Huang et al. 2009a, 2009b) was used to determine if functional categories (GO terms, metabolic pathways, protein structure function) were statistically over-represented. Over-represented is a statistical analysis and using a Chi-square analysis asks if the frequency of genes in a functional category is greater than expected given the data. Hierarchical clustering of gene expression uses Macintosh’s version (de Hoon et al. 2004) of Eisen’s Cluster and Treeview (Eisen et al. 1998). The criterion for significance was set at p<0.05.
Twenty-four hr exposure to OA resulted in marked morphological changes and detachment of HepG2 cells. Cells exposed to the highest concentrations of OA (200 or 607 nM based on concentrations in Tf medium) showed high cell detachment (>50%) from the substrate while cell detachment was less severe at 51 nM OA. However, at 51 nM OA, even cells remaining attached displayed rounded morphology suggesting cytoskeletal changes leading to impending detachment (data not shown). Fewer than 5% of cells were lost from the substrate in 37 nM OA at 24 hr but all cells were rounded, while in 2.5 nM no marked difference from control was noted in either attached cell numbers or in normal cell morphology. Deleterious changes were magnified after 48 hr exposure to OA, with both cell attachment and proliferation, as evidenced by BrdU staining, negatively correlated with OA (Figure 1A and B, respectively). Concentration response analyses resulted in slopes of 0.98 and 1.0 for attachment and proliferation, respectively, and half-inhibitory concentrations (EC50) of 25.4 nM and 5.03 nM OA.
The DNA fragmentation assay for OA exposures to ≥200 nM OA demonstrated apoptosis and possible necrosis in adherent and detached cell fractions (Figure 2A and B). Apoptosis was indicated by the low molecular weight ladder (Figure 2A lanes 4–5; Figure 2B lanes 7–12) while necrosis was suggested by high molecular weight smear (lanes 8, 10). At each concentration in Figure 2B, detached fractions (lanes 8, 10, 12) appeared to have increased ladder formation relative to their adherent counterparts (lanes 7, 9, 11, respectively).
Guo et al. (2010) demonstrated that among 9 human CYP tested CYP3A4 and 3A5 metabolize OA to 4 metabolites. In order to ensure that all observed effects on HepG2 cells might be attributed to OA and not its metabolites, CYP 3A4 or 3A5 activity was assayed by incubation with testosterone. Incubation of HepG2 cells with testosterone produced no detectable 6β-hydroxytestosterone, while control experiments using human recombinant CYP3A4 or human liver microsomes resulted in 70 and 60% conversion of testosterone respectively. HPLC-MS analysis of spent culture media indicated recoveries of OA ranging from 82 to 98% for the 4 most concentrated samples and as low as 39% for the least concentrated sample (Table 1). These recoveries are consistent with those observed from spiked media at similar concentrations. The poor recovery for the lowest concentration sample (HPLC-MS analysis of the original 6.2 nM sample was determined to be 2.4 nM) sample is more likely due to losses of OA at the lower concentration during the extraction process rather than metabolism of OA as none of the 4 previously characterized metabolites (Guo et al. 2010) were detected in any of the samples.
Of the approximately 44,000 probes, only 15,037 were either less than saturating or displayed fluorescent data significantly greater than controls. Among these, 403 (2.7%) were significantly different across all 6 treatments (Table 2; Figure 3). Although 403 genes with significant differences in mRNA expression is 2.7 fold higher than the expected random frequency, only one of these is significant with Bonferroni’s correction or 26 with FDR correction. Dunnett’s test (a post-hoc test) indicates that 389 genes were significantly different between one of the treatments and the control (Figure 3), with the greatest number of significant genes at intermediate concentrations (37 and 51 nM, Table 2). Three-hundred and eighty-nine genes (2.6%) that were significant using the post-hoc Dunnett’s test were 2.6 fold greater than the 1% random expectation. Our studies show that the expression of CYP RNAs was low and not significantly induced.
Regression analysis was performed on the 389 genes for which there was a significant effect on mRNA expression in the presence of OA. One hundred sixty nine genes had notable regressions. Specifically, 65 genes (17% of 389 genes with significant concentration effect) had a significant linear relationship using log10 (Figure 4). One hundred fifty (39%) other genes had a non-linear response best fitted to a quadratic equation that significantly predicted a maximal response at intermediate concentrations (Figure 4). The more complex non-linear regression for these 150 patterns of mRNA expression was used because there was no significant linear regression. Data suggested that OA exerted complex effects on mRNA expression in this cell line. This complexity includes both log-linear and non-linear response to OA and thus the response to OA as measured by mRNA expression is not a simple dose response.
“DAVID”, a program that uses NCBI GO terms, pathways and enzyme designations, was applied to determine if there were specific pathways or gene ontologies associated with either the significant difference relative to the control, or if there were concentration dependent effects. For both sets of significant genes (Dunett’s test comparing concentration to control and regression) many more genes than expected by chance were involved in two pathways: cell cycle/cell replication or division and extracellular (secreted protein and extracellular matrix). The p-value indicated that genes in these pathways were greatly over-represented (Figure 4). Among these genes are several identified cell cycle related genes: CDKN2B (cyclin-dependent kinase inhibitor 2B), CCNA2 (cyclin 2A), ERBB3 (receptor tyrosine-protein kinase erbB-3), as well as apoptosis-related genes: CASP9 (caspase 9), CIDEB (cell death-inducing DFF45-like effector; DNA fragmentation factor) and DFFB (a caspase-activated DNase), and PERP (TP53 apoptosis effector). Several apoptosis-related genes, CASP9-related and ADAM17 (disintegrin and metalloprotease domain 17), were upregulated at intermediate concentrations of OA. Expression of AIF (apoptosis inducing factor), however, was not significantly altered at any concentration. CCNA2 was downregulated at low OA concentration while CDKN2 (cyclin-dependent kinase inhibitor 2) and CCNB2 (cyclin B2) were upregulated at low OA concentrations. ERBB3 was downregulated at high and intermediate OA concentrations. RBBP6 (retinoblastoma protein binding protein 6) was upregulated only at 200 μM OA, whereas the RBR1 gene was not contained in this array.
Flanagan et al. (2001) reported a total loss of viability of HepG2 at 48 hr exposure to 1.24 μM OA, producing changes prior to cell death that reflect cellular responses to toxic challenge such as alterations of the actin cytoskeleton. The studies presented here demonstrated toxic effects of OA at 50 nM. At OA concentrations below 25 nM for 48 hr, HepG2 cells proliferated at 50–80% the rate of controls, but at higher concentrations more negative effects of OA such as pronounced cell rounding and detachment were observed. These effects culminated in nearly complete cell detachment from the substrate at concentrations above approximately 50 nM OA in a process that included some apoptosis, as previously shown for OA (Valdiglesias et al 2011c; Fujiki and Suganuma 1993; Rossini et al. 2001). The similarity in slopes of the concentration-response relationships for detachment and proliferation suggest that cessation of proliferation and cell detachment are linked in a continuum of damage with increasing OA concentration. In addition, greater DNA laddering in detached fractions suggested greater apoptosis in detached cells.
The effects of OA on HepG2 cells appear to be independent of its metabolism via induction of phase 1 enzymes such as the CYP which is the route by which many carcinogens act (Perkinson 1996). CYP3A transforms several procarcinogens to their carcinogenic form (Medina-Diaz and Elizondo 2005; Perera 2010). Although Hashizume et al. (2009) demonstrated an increased tolerance to OA in HepG2 cells transformed to express CYP1A2. Guo et al. (2010) noted that among 9 human CYP tested including CYP1A2, only CYP3A4 and 3A5 metabolize OA. Therefore it is likely that OA did not induce CYP expression in HepG2 cells and OA toxicity in HepG2 cells may be attributed entirely to OA itself and not to its metabolites. Several lines of evidence lead us to this conclusion.
HepG2 cells exhibited low or undetectable levels of CYP3A4 or 3A5 activity in 2 studies (Yoshitomi et al. 2001; Omasa et al. 2005), with no CYP mRNA transcript levels detected in the former study and the latter study demonstrating only 0.6 pmol/min/mg testosterone 6β hydroxylation by HepG2 cells. With the single exception of o, p-DDT (Medina-Diaz and Elizondo 2005), there is little evidence that most CYPs are more than minimally induced in HepG2 cells by the introduction of xenobiotics (Yoshitomi et al. 2001; Wilkening et al. 2003). OA was shown to be converted to 4 metabolites by human cytochrome P450 3A4 and 3A5 (Guo et al. 2010). However, incubation of OA with HepG2 cells failed to produce any of these metabolites. Testosterone metabolism, which is dependent on CYP3A4, was also not observed. Finally, mRNA expression of all members of the (CYP3A) family was low.
Valdiglesias and colleagues (2011a; 2011b; 2011c) studied OA toxicity in the range 0.005–1 μM in different preparations including HepG2 cells to demonstrate differences in OA effects in different human cell types. Caspase 3-dependent apoptosis occurred in HepG2 cells exposed to OA for 3 hr without evidence of necrosis (Valdiglesias et al 2011c). A marker of γH2AX phosphorylation denoting DNA damage in chromatin domains near DNA double strand breaks (Andrievski and Wilkins 2009) was highest in HepG2 of the 3 cell types tested (Valdiglesias et al 2011b). In the comet assay, OA showed no marked oxidative damage to DNA in HepG2 cells (Valdiglesias et al 2011a). OA-induced DNA damage in HepG2 cells over a 48 hr exposure period was suggested by both apoptosis and necrosis in the detached cell fraction. DNA damage induced by CYP-dependent conversion of OA to its metabolites may not have been detected in these experiments due to absence of CYP expression in HepG2 cells. The lack of CYP expression is considered a shortcoming of the HepG2 cell line for genotoxicological testing studies. Guillouzo et al. (2007) developed a human liver-derived cell line, HepaRG, which is inducible for many CYP. It is possible that HepaRG might distinguish DNA damage induced by OA metabolites from the cell death induced by OA as documented here.
Several studies on gene expression in humans have demonstrated diverse actions of OA on cell processes. Chin et al (2000) used differential display in human glioma model T89G cells to show OA-induced upregulation of transcription factors, oxidative metabolism proteins, phosphorylation substrates, and stress response genes after 2.5 hr exposures. Valdiglesias et al (2012) used subtractive hybridization in the human neuroblastoma cell line SHSY5Y to show OA induced up and down gene regulation depending on the time of exposure from 2–48 hr with the affected genes playing roles in translation, transcription, signal transduction, metabolism, cell cycle control and apoptosis, transport, and cytoskeletal processes. The study of OA-mediated concentration effects adds to these insights. More genes were significantly different from the controls at OA concentrations of 37–200 nM than at low or high concentrations, with twice as many genes having curvilinear rather than linear responses. Expression of many genes peaked at intermediate concentrations. Most of these gene expression responses involved cell cycle and secretion or extracellular matrix genes. The array data showed specific cell cycle genes were downregulated at moderate OA concentrations consistent with the significant decrease in cell proliferation observed, while dozens more were upregulated at these same OA concentrations. Cell cycle inhibition at G0/G1 in HepG2 cells during short exposures to OA, with concomitant decreased expression of cyclins A2 and B1 and increased expression of cyclin D has been observed (Rubiolo et al. 2011; Valdiglesias et al. 2011c).
The upregulation of many array cell cycle genes may reflect the phosphorylation state of retinoblastoma protein, pRb, in the presence of OA, rather than a physiological response. pRb is a gatekeeper of cell cycle gene transcription that binds to the promoter region of cell cycle genes and whose activity is controlled by its phosphorylation state. pRB is active and in the promoter-bound form when dephosphorylated. When phosphorylated, it detaches from the promoter, allowing gene transcription. Since pRb is normally dephosphorylated by a serine/theronine phosphatase such as PP-1 or PP-2A (Magenta et al. 2008; Krucher et al. 2006; Grana 2008), in the presence of the PP-1 and PP-2A inhibitor OA, pRb would remain phosphorylated and inactivated, and cell cycle gene transcription would proceed. RBBP6 interacts with both pRb and p53 (Chibi et al. 2008), and promotes degradation of the latter, a sequence hypothesized to increase cell proliferation (Motadi et al. 2011). Thus OA may have induced inappropriate inactivation of both pRb and RBBP6, although it was possible to track only RBBP6 in our array, stimulating cell cycle gene transcription even though the cells were unable to divide due to damage and activation of apoptosis gene pathways. The observed gene expression patterns allude to both changes in morphology and the loss of viability in HepG2 cells.
In conclusion, the deleterious effects of OA on HepG2 cells reflect OA-induced toxicity and not that of its metabolites. While, as in past studies, many genes were affected by OA, cell cycle genes were significantly elevated at low OA while some apoptotic gene pathways were upregulated at moderate and high OA concentrations. Overall, the concentration dependent effects of OA on gene expression may explain the divergent effects of OA at low concentrations stimulating genes involved in the cell cycle while and at high concentrations stimulating apoptosis.
This work was supported by the National Institute of Environmental Health Sciences (NIEHS) Grant S11 ES11181. Microarray data obtained in this study has been uploaded to NCBI Geo.