Machine Learning Generated Synthetic Faces for Use in Facial Aesthetic Research.

Importance: A centralized repository of clinically applicable facial images with unrestricted use would facilitate facial aesthetic research. Objective: Using a machine learning neural network, we aim to (1) create a repository of synthetic faces that can be used for facial aesthetic research and (2) analyze synthetic faces according to contemporary aesthetic principles. Design, Setting, and Participants: Synthetic facial images were generated using an open source generative adversarial network. Images were refined and then analyzed using computer vision technology. Interventions: Not applicable. Main Outcomes and Measures: Synthetic facial images were created for use as a facial aesthetic research data set. Results: One thousand synthetic images were generated, and 60 images underwent analysis. Image attributes, including age, gender, image principle axis, facial emotion, and facial landmark points, were attained. Images demonstrated accordance with contemporary aesthetic principles of horizontal thirds and vertical fifths. Images demonstrated excellent correspondence when compared with real human facial photographs. Conclusions and Relevance: We have generated realistic synthetic facial images that have potential as a valuable research tool and demonstrate similarity to real human photographs while adhering to contemporary aesthetic principles.

A novel isotope dilution UHPLC-ESI-MS/MS method for the quantification of 3-monochloropropane-1,2-diol in Caco-2 cell transport and receiving buffers.

A routine, selective and sensitive ultra-high performance liquid chromatography-electrospray ionisation tandem triple quadrupole mass spectrometry (UHPLC-ESI-MS/MS) method was developed and validated for the quantification of 3-monochloropropane-1,2-diol (3-MCPD) in Caco-2 cell transport buffer (FaSSIF-V2, the second version of a fasted state simulated intestinal fluid) and receiving buffer (HBSS, Hank's balanced salt solution). The method involves measuring deuterated 3-MCPD (3-MCPD-d5) as internal standard (IS) during the entire analytical procedure to obtain precise and accurate results. The separation was performed on a Poroshell 120 HILIC column (2.7 µm, 3.0 × 50 mm) at a flow rate of 0.3 mL/min using water (containing 0.025% acetic acid) and acetonitrile (containing 0.025% acetic acid) as the mobile phases. Mass spectrometric detection was operated in dynamic multiple reaction monitoring (dMRM) in negative ion mode. The method exhibited high sensitivity. The limits of detection (LOD) for 3-MCPD in FaSSIF-V2 and HBSS were 0.012 and 0.014 µmol/L, and the limits of quantification (LOQ) were 0.039 and 0.045 µmol/L, respectively. Satisfactory results were observed for linearity (R2 > 0.999), intra-day precision (RSD% <7.7% in FaSSIF-V2 and <6.6% in HBSS), inter-day precision (RSD% <5.9% in FaSSIF-V2 and <5.6% in HBSS), accuracy (% error within ± 10%), and sample stability (RSD% <7.7% and % error within ± 10%). The method has been successfully applied to quantify 3-MCPD in Caco-2 cell transport and receiving buffers. The results were in good agreement with those obtained with gas chromatography-tandem mass spectrometry (GC-MS).

Assessment of intestinal absorption/metabolism of 3-chloro-1,2-propanediol (3-MCPD) and three 3-MCPD monoesters by Caco-2 cells.

3-chloro-1,2-propanediol (3-MCPD) and 3-MCPD esters are contaminants present in a variety of processed foods, including infant formulas. Toxicological data are unavailable in humans, but rodent studies have demonstrated renal and testicular toxicity from 3-MCPD and 3-MCPD esters. There is evidence that 3-MCPD esters are hydrolyzed in the digestive system, releasing 3-MCPD that would be absorbed and induce damage. We assessed absorption and metabolism of 3-MCPD and three 3-MCPD monoesters, 1-oleoyl (1-Ol), 1-linoleoyl (1-Li) and 1-palmitoyl (1-Pa) commonly found in U.S. infant formula using differentiated Caco-2 cells. After 1-hour incubation, all three monoesters released free 3-MCPD and free fatty acids (FFA) into Caco-2 cell supernatants. Free 3-MCPD had a high apparent permeability (Papp = 30.36 ± 1.31 cm/s × 10-6) suggesting that it is freely diffusible and highly absorbed by intestinal epithelium. 1-Li released 3-4-fold more 3-MCPD than 1-Ol and 1-Pa over 1 h, suggesting that this variable release rates might contribute to the overall in vivo exposure to 3-MCPD. None of the monoesters or FFA were detected in basolateral supernatants, suggesting that these compounds do not cross the intestinal wall without further transformation. In summary, this study provides relevant data to advance knowledge of in vivo intestinal absorption and metabolism of 3-MCPD monoesters.

Diglycolic acid induces HepG2/C3A liver cell toxicity in vitro.

Carboxymethyl starches are added to food products for thickening or tablet binding/filling purposes. Although they lack toxicity, their synthesis creates the chemical byproduct diglycolic acid (DGA), which is difficult to eliminate and whose toxicity is in question. A rare case of an accidental direct exposure to extremely high concentrations of DGA in a person revealed that DGA has the potential to be toxic to several organs, with the kidneys and liver being the most affected organs. Given that DGA is present in our food supply as a chemical byproduct of carboxymethyl starch food additives, we sought to perform in vitro testing of its potential hepatotoxicity to help complement a recent in vivo rat acute dose-response study that also tested for the potential hepatotoxic effects of daily DGA ingestion by oral gavage over a period of 28 days. Using the HepG2/C3A cellular in vitro model, we tested how escalating doses of DGA exposure over 24 h could induce hepatotoxicity. Both in vitro and in vivo testing systems revealed that DGA is indeed a hepatotoxin once a certain exposure threshold is reached. The concordance of these models highlights the utility of in vitro testing to support and help predict in vivo findings.

Simultaneous determination of intestinal permeability and potential drug interactions of complex mixtures using Caco-2 cells and high-resolution mass spectrometry: Studies with Rauwolfia serpentina extract.

Caco-2 cells are a commonly used model for estimating the intestinal bioavailability of single chemical entity pharmaceuticals. Caco-2 cells, when induced with calcitriol, also express other biological functions such as phase I (CYP) and phase II (glucuronosyltransferases) drug metabolizing enzymes which are relevant to drug-supplement interactions. Intestinal bioavailability is an important factor in the overall safety assessment of products consumed orally. Foods, including herbal dietary supplements, are complex substances with multiple chemical components. Because of potential interactions between components of complex mixtures, more reliable safety assessments can be obtained by studying the commercial products "as consumed" rather than by testing individual chemical components one at a time. The present study evaluated the apparent intestinal permeability (Papp) of a model herbal extract, Rauwolfia serpentina, using both whole plant extracts and the individual purified Rauwolfia alkaloids. All test compounds, endpoint substrates, and their metabolites were quantified using liquid chromatography and high-resolution mass spectrometry. The Papp values for individual Rauwolfia alkaloids were comparable whether measured individually or as components of the complete extract. Both Rauwolfia extract and all individual Rauwolfia alkaloids except yohimbine inhibited CYP3A4 activity (midazolam 1'-hydroxylation). Both Rauwolfia extract and all individual Rauwolfia alkaloids except corynanthine and reserpic acid significantly increased glucuronosyltransferase activity (glucuronidation of 4-methylumbelliferone). The positive control, ketoconazole, significantly inhibited both CYP3A4 and glucuronosyltransferase activities. These findings suggest that the Caco-2 assay is capable of simultaneously identifying both bioavailability and potentially hazardous intestinal drug-supplement interactions in complex mixtures.

Comparison of diglycolic acid exposure to human proximal tubule cells in vitro and rat kidneys in vivo.

Diglycolic acid (DGA) is present in trace amounts in our food supply and is classified as an indirect food additive linked with the primary GRAS food additive carboxymethyl cellulose (CMC). Carboxymethyl starches are used as a filler/binder excipient in dietary supplement tablets and a thickening ingredient in many other processed foods. We sought to utilize the human proximal tubule HK-2 cell line as an in vitro cellular model system to evaluate its acute nephrotoxicity of DGA. We found that DGA was indeed toxic to HK-2 cells in all in vitro assays in our study, including a highly sensitive Luminex assay that measures levels of an in vitro biomarker of kidney-specific toxicity, Kidney Injury Molecule 1 (KIM-1). Interestingly, in vitro KIM-1 levels also correlated with in vivo KIM-1 levels in urine collected from rats treated with DGA by daily oral gavage. The use of in vitro and in vivo models towards understanding the effectiveness of an established in vitro system to predict in vivo outcomes would be particularly useful in rapidly screening compounds that are suspected to be unsafe to consumers. The merit of the HK-2 cell model in predicting human toxicity and accelerating the process of food toxicant screening would be especially important for regulatory purposes. Overall, our study not only revealed the value of HK-2 in vitro cell model for nephrotoxicity evaluation, but also uncovered some of the mechanistic aspects of the human proximal tubule injury that DGA may cause.

Evaluation of "Dream Herb," Calea zacatechichi, for Nephrotoxicity Using Human Kidney Proximal Tubule Cells.

A recent surge in the use of dietary supplements, including herbal remedies, necessitates investigations into their safety profiles. "Dream herb," Calea zacatechichi, has long been used in traditional folk medicine for a variety of purposes and is currently being marketed in the US for medicinal purposes, including diabetes treatment. Despite the inherent vulnerability of the renal system to xenobiotic toxicity, there is a lack of safety studies on the nephrotoxic potential of this herb. Additionally, the high frequency of diabetes-associated kidney disease makes safety screening of C. zacatechichi for safety especially important. We exposed human proximal tubule HK-2 cells to increasing doses of this herb alongside known toxicant and protectant control compounds to examine potential toxicity effects of C. zacatechichi relative to control compounds. We evaluated both cellular and mitochondrial functional changes related to toxicity of this dietary supplement and found that even at low doses evidence of cellular toxicity was significant. Moreover, these findings correlated with significantly elevated levels of nephrotoxicity biomarkers, lending further support for the need to further scrutinize the safety of this herbal dietary supplement.

In vitro exposure of Adhatoda zeylanica to human renal cells lacks acute toxicity.

Adhatoda zeylanica is a dietary supplement ingredient present in several types of dietary supplements, including weight loss, respiratory relief, and immune regulating products. Due to its reported wide range of uses in folk medicine, it was hypothesized that it may have the potential to target multiple organs and lead to a range of toxicity features. As a preliminary evaluation of the safety of this herbal ingredient, an investigation into its effects on the kidney was sought. An in vitro study of its potential nephrotoxicity using the HK-2 human proximal tubule cell line in a variety of functional indicators was performed to capture both general forms of cellular toxicity as well as ones that are specific to proximal tubules. A. zeylanica was only capable of inducing detrimental short-term toxicity to HK-2 cells at relatively high treatment concentrations when exposed directly to the cells. The lack of acute and potent toxicity of A. zeylanica under our experimental conditions calls for further studies to better define its toxicant threshold and establish safe dosage levels.

Evaluation of CYP3A4 inhibition and hepatotoxicity using DMSO-treated human hepatoma HuH-7 cells.

A human hepatoma cell line (HuH-7) was evaluated as a metabolically competent cell model to investigate cytochrome P450 3A4 (CYP3A4) inhibition, induction, and hepatotoxicity. First, CYP3A4 gene expression and activity were determined in HuH-7 cells under three culture conditions: 1-week culture, 3-week culture, or 1 % dimethyl sulfoxide (DMSO) treatment. HuH-7 cells treated with DMSO for 2 weeks after confluence expressed the highest CYP3A4 gene expression and activity compared to the other two culture conditions. Furthermore, CYP3A4 activity in DMSO-treated HuH-7 cells was compared to that in a human hepatoma cell line (HepG2/C3A) and human bipotent progenitor cell line (HepaRG), which yielded the following ranking: HepaRG > DMSO-treated HuH-7 >> HepG2/C3A cells. The effects of three known CYP3A4 inhibitors were evaluated using DMSO-treated HuH-7 cells. CYP3A4 enzyme inhibition in HuH-7 cells was further compared to human recombinant CYP3A4, indicating similar potency for reversible inhibitors (IC 50 within 2.5-fold), but different potency for the irreversible inhibitor. Next, induction of CYP3A4 activity was compared between DMSO-treated HuH-7 and HepaRG cells using two known inducers. DMSO-treated HuH-7 cells yielded minimal CYP3A4 induction compared to that in the HepaRG cells after 48-h treatments. Finally, the cytotoxicity of five known hepatotoxicants was evaluated in DMSO-treated HuH-7, HepG2/C3A, and HepaRG cells, and significant differences in cytotoxic sensitivity were observed. Overall, DMSO-treated HuH-7 cells are a valuable model for medium- or high-throughput screening of chemicals for CYP3A4 inhibition and hepatotoxicity.

CYP3A4 inhibition by Psoralea corylifolia and its major components in human recombinant enzyme, differentiated human hepatoma HuH-7 and HepaRG cells.

Psoralea corylifolia (P. corylifolia) is a medicinal plant used primarily in herbal dietary supplements to treat skin diseases, such as vitiligo and psoriasis. Case reports of liver toxicity have recently emerged from its use, which often includes co-administration with other herbal products. In this study, CYP3A4 inhibition and hepatotoxicity of P. corylifolia and its major components were evaluated in human recombinant CYP3A4 enzyme, differentiated human hepatoma HuH-7 and HepaRG cells. LC/MS-TOF was used to identify the major components of P. corylifolia fruit methanol-water extract. P. corylifolia and its major bioactive components psoralen and isopsoralen were then incubated with human recombinant CYP3A4 (10 min) or differentiated HuH-7 and HepaRG cells (24 h) prior to CYP3A4 activity and cytotoxicity assays. P. corylifolia extract, psoralen, and isopsoralen concentration dependently inhibited CYP3A4 activity with different potency in the three in vitro systems. No cytotoxicity was observed at any concentration tested. In vitro CYP3A4 inhibition by P. corylifolia and its major components suggests potential drug-dietary supplement interactions that warrant further investigations in vivo.

Human kidney proximal tubule cells are vulnerable to the effects of Rauwolfia serpentina.

Rauwolfia serpentina (or Snake root plant) is a botanical dietary supplement marketed in the USA for maintaining blood pressure. Very few studies have addressed the safety of this herb, despite its wide availability to consumers. Its reported pleiotropic effects underscore the necessity for evaluating its safety. We used a human kidney cell line to investigate the possible negative effects of R. serpentina on the renal system in vitro, with a specific focus on the renal proximal tubules. We evaluated cellular and mitochondrial toxicity, along with a variety of other kidney-specific toxicology biomarkers. We found that R. serpentina was capable of producing highly detrimental effects in our in vitro renal cell system. These results suggest more studies are needed to investigate the safety of this dietary supplement in both kidney and other target organ systems.

Sex hormone modulation of both induction and inhibition of CYP1A by genistein in HepG2/C3A cells.

Genistein is a widely consumed phytoestrogen in dietary supplements and has been reported to play roles in both cancer prevention and promotion. These conflicting effects may be complicated by sex differences. Cytochrome P450 1A (CYP1A) participates in carcinogen activation and detoxification, and the enzyme may interact with genistein. Therefore, modulation of CYP1A by a combination of genistein and sex hormones could be responsible for sex differences related to cancer prevention and promotion. In the current study, a human liver cell line, HepG2/C3A, cultured in sex hormone-supplemented media was used to investigate the modulatory effect of genistein on CYP1A gene expression and activity. Genistein exerted both long-term (72 h) induction and short-term (immediate) inhibition of CYP1A activity in HepG2/C3A cells. In the long-term study, CYP1A gene expression and enzyme activity were induced to a greater extent in male hormone-supplemented cells than female ones. In the short-term study, CYP1A activity was inhibited more strongly by genistein in the male hormone-supplemented cells than in the female hormone-supplemented cells. These significant differences suggest that male hormones can modulate the effects of genistein on CYP1A gene expression and activity.

Correlating in vitro data to in vivo findings for risk assessment.

A special session at the Toxicology and Risk Assessment Conference in Cincinnati, OH, USA in May, 2012 presented approaches expanding upon current uses of in vitro toxicity data for risk assessment. Evaluation of xenobiotics through use of in vitro study methods is increasing exponentially and these methodologies offer a relatively fast and considerably cheaper way to determine toxicities in comparison to traditional approaches. One of the challenges with in vitro data is to effectively use this information for risk assessment purposes. Currently, in vitro studies are used as supportive for hazard characterization and identifying mechanisms associated with toxicity. Being able to effectively correlate in vitro effects to in vivo observations represents a major challenge for risk assessors. The presentations in this special session provided innovative approaches toward effectively using in vitro data for the human health risk assessment process.

Use of the Combination Index to determine interactions between plant-derived phenolic acids on hepatotoxicity endpoints in human and rat hepatoma cells.

The beneficial or adverse effects of isolated phytochemicals are not always concordant with effects of the botanical dietary supplements from which they were derived. This disparity could be due to interactions between the various phytochemicals present in the whole plant. The phenolic acids, rosmarinic acid (RA), caffeic acid (CA) and ferulic acid (FA) are widely present in foods and dietary supplements, and they are assumed to exert various beneficial biological effects. However, there is little data on the potential biological interactions of these three phenolic acids which commonly occur together and are linked metabolically. In the present study, liver toxicity of the three phenolic acids was assessed on the three compounds singly and in various binary and one ternary combinations. A series of in vitro endpoints relevant to liver toxicity were evaluated in both a human (HepG2/C3A) and rat (MH1C1) hepatocyte cell line. The Combination Index (CI) was calculated for each endpoint from both the concentration responses of the single compounds and the responses of the various binary and ternary mixtures. Both synergistic and antagonistic interactions were observed for some endpoints and some combinations of test agents. Interactions were most prevalent in measures of oxidative stress and cytochrome P450 activities in both cell types. There was only a 53% concordance between the rat and human cells which may be suggestive of species differences. The data suggest an approach for better characterizing the beneficial or adverse effects of complex botanical products through evaluation of interactions between individual phytochemical components.

Cellular glutathione in fatty liver in vitro models.

The range of non-alcoholic fatty liver disease (NAFLD) includes simple hepatic steatosis, the inflammatory non-alcoholic steatohepatitis (NASH), fibrosis and cirrhosis. The accumulation of specific lipids in hepatocytes has been associated with oxidative stress and progression of the disease. Elevated serum free fatty acids and hepatocyte lipotoxicity can be studied in an in vitro cellular model. For this purpose, we cultured the human liver cell line, HepG2/C3A, in medium supplemented with increasing amounts of oleic acid (C18:1) and evaluated oxidative stress by measuring the content of the cellular antioxidant, glutathione (GSH). We observed a dose-dependent steatosis, as determined by Nile Red staining, with concurrent increases of GSH; similar findings were also observed in cultured human hepatocytes. Cells cultured with palmitic acid (C16:0) or the combination oleic/palmitic acids (2:1 ratio) also exhibited a dose-dependent increase of GSH; however palmitic-supplemented cultures did not sustain the GSH increase after 24h. We also detected an increase in the formation of lipid peroxides (LPO) indicating that the increase of GSH was a cellular mechanism that may be related to the high exposure of fatty acids. The results of this in vitro study suggest an antioxidant response against fat overloading and indicate potential differences in response to specific fatty acid-induced hepatic steatosis and associated lipotoxicity.

Effects of dietary phenolics and botanical extracts on hepatotoxicity-related endpoints in human and rat hepatoma cells and statistical models for prediction of hepatotoxicity.

Toxicity assessment of botanical materials is difficult because they are typically complex mixtures of phytochemicals. In the present study, 16 phenolics were tested in both human (HepG2/C3A) and rat (MH1C1) hepatoma cells using a battery of eight toxicity endpoints. Cluster analysis was used to group the phenolics into four clusters for each cell type. Comparison of overall and individual liver activity of phenolics on both human and rat hepatoma cell lines showed significant differences for some endpoints. However, the cluster membership was similar across both cell types with the majority of phenolics clustering with the solvent control group (cluster 1). Each cell type produced a cluster of compounds with reported in vivo liver toxicity (cluster 2). Five herbal extracts were prepared and then tested as above. Using the cluster model developed with the phenolics, in the HepG2/C3A cells green tea was assigned to cluster 2 and the remaining four extracts to cluster 1. In the MH1C1 cells, green tea and thyme were assigned to cluster 2, cinnamon to cluster 4, and juniper berry and peppermint to cluster 1. The data suggest that this in vitro model may be useful for identifying hepatotoxic phenolics and botanical preparations rich in phenolics.

An in vitro system for studying potential biological mechanisms of human sex differences in susceptibility to acute liver injury.

Women are more susceptible than men to acute liver injury from drugs and other xenobiotics. The biological mechanisms for this sex difference are unknown, but known sex differences in steroid hormone levels and immune response could play a role. A human hepatocyte cell line, HepG2, was cultured for 8 days in either a male hormone, female hormone, or sex hormone-free medium. The cells were then exposed to a mixture of pro-inflammatory cytokines (interleukin (IL)-1beta, IL-6, TNFalpha) for 72h to simulate acute inflammation. Cell viability (total DNA) and various metabolic functions (reactive oxygen species (ROS), neutral and polar lipid (PL) accumulation, mitochondrial membrane potential, cytochrome P450 (CYP) activities) were measured fluorometrically. Acute phase proteins (albumin, IL-1ra) were measured in the culture medium by ELISA. This model gave both significant hormone only effects (ROS, PL accumulation) and cytokine only effects (total DNA, CYP1A, neutral and PL accumulation, albumin, IL-1ra) consistent with known biological responses. Significant hormone-cytokine interactions were observed for several endpoints (total DNA, ROS, neutral and PL accumulation, albumin). These findings suggest that sex hormones and pro-inflammatory cytokines can interact to alter liver metabolism in ways that may contribute to the marked sex difference in susceptibility to chemical-induced acute liver injury.

Multiendpoint mechanistic profiling of hepatotoxicants in HepG2/C3A human hepatoma cells and novel statistical approaches for development of a prediction model for acute hepatotoxicity.

HepG2/C3A human hepatoma cells were exposed to serial concentrations of seven known hepatotoxicants for 48h. Six endpoint assays were selected to model different mechanisms of acute hepatotoxicity. Each compound produced a unique concentration-response pattern across all endpoints. The endpoints did not correlate strongly, suggesting that each endpoint monitored an independent cellular process. Prediction models were developed using five statistical methods. The models used only known hepatotoxicants for the training set. The zero concentration (control) and all concentrations not significantly different from control were programmed as non-toxic levels and concentrations significantly different from control as toxic levels. So, rather than a binary classification of each compound (i.e., toxic or non-toxic), the models gave a prediction of the concentration, if any, at which a compound showed behavior similar to liver toxicants at their toxic concentrations. The discriminant analysis model gave the best overall performance with positive and negative predictive values of 1.00 and 0.83, respectively. Ten additional compounds were tested using this prediction model. The model predicted liver active concentrations for each compound that were consistent with their known biologically active concentrations. This model system may be useful for predicting concentration levels at which unknown compounds would display undesirable liver activity.

Rat liver clone-9 cells in culture as a model for screening hepatotoxic potential of food-related products: hepatotoxicity of deoxynivalenol.

Deoxynivalenol (DON) is a mycotoxin food contaminant found in several cereal grains. The literature on the liver toxicity of DON in vivo is conflicting and does not clearly characterize its hepatotoxic effects. Cultured rat liver clone-9 cells were used as a model to assess the hepatotoxic potential of DON. The cell cultures, seeded onto 96-well plates, were treated at confluence with varying concentrations of DON (0-100 microg ml(-1)) for 48 h at 37 degrees C in 5% CO2. After the treatment period, the cells were assayed for a number of hepatotoxic endpoints that included cytotoxicity, double-stranded DNA (ds-DNA) content, oxidative stress and mitochondrial function. The concentration-dependent toxicity of DON, as measured by cytotoxicity and ds-DNA content, was observed over the entire concentration range studied beginning at 0.5 microg ml(-1). DON also induced a significant concentration-dependent increase in oxidative stress at DON concentrations starting at 10 microg ml(-1). The mitochondrial function of the treated cells decreased with the increasing concentration of DON exposure, but it was not statistically different from that of the control value. Liver histopathology observed at 3, 24 and 72 h following a single intraperitoneal administration dose of DON (10 mg kg(-1) BW) to adult male rats is consistent with early mild hepatotoxicity. The overall results of this study suggest that acute DON exposure has early mild cytotoxic effects on hepatocytes in vivo that are expressed as severe effects in rat liver clone-9 cells in vitro.

Validation of an in vitro model for assessment of androstenedione hepatotoxicity using the rat liver cell line clone-9.

Androstenedione, a naturally occurring steroid hormone, has been used to enhance athletic performance. Little is known, however, about its hepatotoxicity. Clone-9 cells, a non-transformed epithelial cell line that was originally isolated from normal liver of a 4-week old Sprague-Dawley rat, were used as an in vitro model to assess the hepatotoxic potential of androstenedione. The cultures were treated with androstenedione for 24 h at 37 degrees C in 5% CO(2) at concentrations of 0-100 microg ml(-1). After the treatment period, the cells and the culture supernatants were assayed for markers of cytotoxicity which included: release of liver enzymes, cell viability, cellular double-stranded DNA content, oxidative stress, steatosis, cellular ATP content, caspase-3 activity, the mitochondrial permeability transition and induction of cytochrome P450 activity. Significant concentration-dependent differences from control were observed in some endpoints at medium concentrations of 10 microg ml(-1) and above. These in vitro findings were compared with comparable endpoints obtained from an in vivo study of androstenedione toxicity in female Sprague-Dawley rats. Of the eight endpoints that could be compared between the two studies, only three (lipid accumulation, ATP depletion and P450 activity) appeared to be concordant. This suggests that, under the experimental conditions used, the clone-9 cells were not a good model for androstenedione hepatotoxicity.