|Home | About | Journals | Submit | Contact Us | Français|
Hepatotoxicity is a major concern for both drug development and toxicological evaluation of environmental chemicals. The assessment of compound-induced hepatotoxicity has traditionally relied on in vivo testing; however, it is being replaced by human in vitro models due to an emphasis on the reduction of animal testing and species-specific differences. Since most cell lines and hybridomas lack the full complement of enzymes at physiological levels found in the liver, primary hepatocytes are the gold standard to study liver toxicities in vitro due to the retention of most of their in vivo activities. Here, we optimized a cell viability assay using plateable cryopreserved human hepatocytes in a 1536-well-plate format. The assay was validated by deriving inhibitory concentration at 50% values for 12 known compounds, including tamoxifen, staurosporine, and phenylmercuric acetate, with regard to hepatotoxicity and general cytotoxicity using multiple hepatocyte donors. The assay performed well, and the cytotoxicity of these compounds was confirmed in comparison to HepG2 cells. This is the first study to report the reliability of using plateable cryopreserved human hepatocytes for cytotoxicity studies in a 1536-well-plate format. These results suggest that plateable cryopreserved human hepatocytes can be scaled up for screening a large compound library and may be amenable to other hepatocytic assays such as metabolic or drug safety studies.
The liver is the key organ for the biotransformation of endogenous and exogenous compounds to aid in their clearance from the body. Paradoxically, these protective effects may increase susceptibility of the liver to toxicity from either the parental compound or its metabolites. Therefore, hepatotoxicity is a concern for the development and use of chemicals and drugs in industrial, environmental, and pharmaceutical applications.1,2 With examples such as potassium dichromate3,4 and troglitazone,5,6 the impact of hepatotoxicity can be significant, resulting in potential liver damage, liver transplantation, or death.
Human in vivo toxicities are usually detected in late-stage development or postmarket release and, therefore, cannot predict risks from new chemical entities due to the lack of human exposure during discovery and early development.7,8 Traditionally, animal models have been used to assess hepatotoxicity with success for overt hepatotoxins, but they fail to accurately predict species-specific toxicities and idiosyncratic drug-induced liver injury (DILI).1,9,10 Additionally, animal models are expensive, time-consuming, and not amenable to large chemical library screens. With increasing numbers of new chemical entities for environmental and pharmaceutical uses, it is necessary to find a rapid and efficient method to screen chemicals for their potential toxicities. Since most severe DILI is due to hepatocellular injury,1 alternative in vitro models are being used to estimate in vivo responses, to reduce and/or replace in vivo animal testing, and to increase the throughput of the evaluation of compounds screened and amount of data generated. Therefore, the ideal screening protocol would utilize human-derived cells in an in vitro assay to obtain human specific data while not endangering human volunteers.
As a result, there are several ongoing efforts dedicated to understanding chemical-induced toxicities in a variety of in vitro models. One key initiative is the National Toxicity Program (NTP), a U.S. federal government organization, started in 1978 to coordinate toxicological testing programs for the strengthening of toxicological sciences and development and validation of testing methods related to potentially toxic chemicals. Recently, the NTP, the NIH Chemical Genomics Center, and the U.S. Environmental Protection Agency initiated the Tox21 program for the development and validation of in vitro assays through the use of a high-throughput screening (HTS) platform.11–13
Various in vitro approaches have been described in the literature to screen for hepatotoxicity.14–18 Recently, the advent of quantitative high-throughput screening (qHTS) has enabled researchers to obtain inhibitory concentration at 50% (IC50) values directly from primary screening, such as viability assays, to assess the toxicity potential of compounds in cell lines.19,20 Cell lines of hepatic origin, such as HepG2 cells, have been previously adapted to HTS formats21,22 and utilized to assess hepatotoxicity.19,20,23 However, HepG2 cells lack the full expression of hepatocyte proteins, such as phase I and phase II metabolizing enzymes and transporters, and thus may not correlate to in vivo hepatotoxicity.24–26 As an alternative to HepG2 cells, primary human hepatocytes represent the best predictive in vitro model to determine liver function for metabolism,27,28 drug–drug interactions,29,30 and potential hepatotoxicity of compounds.30–32
Hepatocytes can be utilized in suspension for assays lasting a few hours or may be maintained in collagen-coated tissue culture plates for extended culturing. Traditionally, the use of hepatocytes has been limited to low density well formats such as 24-well for enzyme induction studies29 or culture tubes for drug metabolism assays.33 In addition, primary hepatocytes have been utilized as an in vitro model for determining hepatotoxicity and have shown strong correlation to in vivo hepatotoxicity.34–36
In spite of the acceptance of hepatocytes in pharmaceutical research, they have had minimal use for short-term suspension assays or multi-day culturing protocols in HTS studies. Human hepatocytes in suspension cultures lasting several hours have been utilized in 96-well and 384-well formats for determining the metabolic clearance of drugs.37,38 Further, Wolff et al. have established a method for plated rat hepatocytes cultured for multiple days in 384-well format for high content screening (HCS) to monitor cellular functions.39 However, no published reports have described employing cultured hepatocytes in 1536-well format. The low availability of freshly isolated human hepatocytes and plateable cryopreserved human hepatocytes had made them impractical for screening protocols and had limited assays to short-term incubations of six hours or less.32 However, recent improvements in availability and quality of plateable cryopreserved human hepatocytes have increased the opportunity for their use in HTS. Inclusion of primary hepatocytes in an HTS format would provide relevant data from the screening of large chemical libraries for the assessment of hepatotoxicity.
Herein, we describe the first reported multi-day culturing of plateable cryopreserved human hepatocytes from multiple donors in a 1536-well microtiter-plate format and its subsequent use in the determination of their hepatotoxicity potential of compounds by generating IC50 values. Intracellular adenosine triphosphate (ATP) levels were measured to assess viability and consistency of plating and retention of hepatocytic function was confirmed through inhibition of CYP3A4 activity. To determine hepatotoxicity in a miniaturized format, the assay was validated during a 40h exposure to a dozen of known toxic compounds, such as doxorubicin, tamoxifen, staurosporine, and phenylmercuric acetate.
All chemicals were purchased from Sigma (St. Louis, MO). CellTiter Glo® and P450-Glo™ CYP3A4 Assay with Luciferin-IPA kits were purchased from Promega (Madison, WI). Collagen I–coated 1536-well microtiter plates were purchased from Greiner Bio-One North America (Monroe, NC). Plateable cryopreserved human hepatocytes and medium were obtained from Celsis In Vitro Technologies (Baltimore, MD).
Plateable cryopreserved human hepatocytes were characterized for cytochrome P450 (CYP), UDP-glucuronosyltransferase (UGT), and sulfotransferase (ST) enzymatic activities, as well as for their ability to be cultured in vitro. For enzymatic characterization, the cryopreserved hepatocytes were prepared according to instructions for use provided by Celsis In Vitro Technologies. Briefly, cryopreserved human hepatocytes were thawed at 37°C for approximately 2 minutes and transferred into 48mL of InVitroGRO™ HT medium at 37°C. The cell suspension was centrifuged at 50×g for 5min. The supernatant was removed and the cell pellet was resuspended in InVitroGRO KHB medium. The hepatocyte suspension was counted for viability and cell concentration using Trypan blue exclusion. Cell density was diluted to a stock concentration of 2.0×106 viable cells per mL. The hepatocyte suspension was incubated for 60min at 37°C with equal volume of specific substrates for individual CYP enzymes as recommended by the FDA guidance40 at the following final concentrations: phenacetin [15μM] (CYP1A2), coumarin [8μM] (CYP2A6), tolbutamide [15μM] (CYP2C9), s-mephenytoin [20μM] (CYP2C19), dextromethorphan [5μM] (CYP2D6), chlorzoxazone [50μM] (CYP2E1), testosterone [50μM] (CYP3A4), and 7-hydroxycoumarin [30μM] (UGT and ST). The metabolites of the reactions were quantified using high performance liquid chromatography, liquid chromatography tandem mass spectrometry (LC/MS/MS), or ultra performance liquid chromatography tandem mass spectrometry (UPLC/MS/MS) methods. The data were converted to the final units of picomoles of metabolite per minute per million cells of hepatocytes (pmol/min/million cells). The hepatocyte functional data from 172 donors of cryopreserved human hepatocytes, characterized as described above, were provided by Celsis In Vitro Technologies.
The ability of cryopreserved hepatocytes to attach to cell culture plates was confirmed by culturing the plateable cryopreserved human hepatocytes for 5 days and by inducing the hepatocytes for protein synthesis of CYP1A2 and CYP3A4 as recommended in the FDA guidance.40 Briefly, cryopreserved human hepatocytes were thawed at 37°C for approximately 2 minutes and transferred into 5mL of InVitroGRO CP medium at 37°C. The hepatocyte suspension was counted for viability and cell concentration using Trypan blue exclusion. Cell density was diluted to a final concentration of 0.7×106 viable cells per mL. Cells were dispensed into 48-well collagen I-coated culture plate at a final density of 140,000 viable hepatocytes per well. The medium was changed during days 1 and 2 of culture with InVitroGRO CP medium. On days 3 and 4, wells were incubated with the positive controls omeprazole [50μM] and rifampicin [25μM] in InVitroGRO HI medium to induce CYP1A2 and CYP3A4 protein levels, respectively. Vehicle control wells were incubated with 1% acetonitrile in InVitroGRO HI medium. On day 5, the medium was removed and metabolism was determined from a 3-h incubation with phenacetin [100μM] and testosterone [125μM] in InVitroGRO KHB medium for CYP1A2 and CYP3A4 activity, respectively. The metabolites of the reactions were quantified using LC/MS/MS or UPLC/MS/MS methods. The data were converted to the final units of pmol/min/million cells. Fold induction was determined by the following equation: Positive Control Activity [pmol/min/million cells]/Vehicle Control Activity [pmol/min/million cells].
The preparation of the cryopreserved hepatocytes for culturing was modified from the instructions for use provided by Celsis In Vitro Technologies. Plateable cryopreserved human hepatocytes were thawed at 37°C and diluted into 48mL of InVitroGRO CP medium at 37°C. The cell suspension was centrifuged at 50×g for 5min. The supernatant was removed and the cell pellet was resuspended in InVitroGRO CP medium. The cell suspension was filtered through a 40μm nylon cell strainer (BD Biosciences, Durham, NC) to prevent obstruction of the dispensing tubes. The hepatocyte suspension was counted for viability and cell concentration using Trypan blue exclusion. Cell density was diluted to final concentrations of 0.4×106 viable cells per mL and 0.8×106 viable cells per mL to provide hepatocyte suspensions. The stepwise procedure from plating the cells to assay readout is described in Table 1. The hepatocyte suspensions were dispensed at 2,000 (2K) or 4,000 (4K) cells/5μL/well in collagen I-coated 1536-well microtiter plates using a Flying Reagent Dispenser (FRD) (Aurora Discovery, Carlsbad, CA). The assay plates were incubated in a 37°C, 5% CO2 humidified incubator for 4 or 24h to allow for cell attachment. Compound addition was performed via pin tool (Kalypsys, San Diego, CA), where 23nL of compound or dimethyl sulfoxide (DMSO) was added into the wells. The assay plates were incubated for an additional 24 or 40h, followed by the addition of 5μL per well of CellTiter-Glo reagent. This assay measures intracellular ATP content as a biomarker for cellular viability, where luminescent signal is proportional to the amount of ATP in metabolically active cells. Similarly, cell viability was also tested in HepG2cells after compound treatment. The cells were dispensed at 2K cells/5μL/well in 1536-well white/solid-bottom assay plates using an FRD. The cells were incubated a minimum of 5h at 37°C, followed by the addition of compounds using the pin tool. The assay plates were incubated for 40h, followed by the addition of 5μL/well of CellTiter-Glo reagent via FRD. After a 30-min incubation at room temperature, the luminescence of each well was determined with a ViewLux plate reader (PerkinElmer, Shelton, CT). Hepatotoxicity potential of 12 compounds was determined by deriving IC50 values from 16-point titrations, with each point in quadruplicate for a 40h exposure. A 40-h exposure timepoint was extrapolated from previous publications for assessing cellular toxicity of the 12 compounds as measured by ATP content.19,20
Retention of CYP3A4 activity was confirmed by inhibiting its activity with the specific inhibitor ketoconazole.41,42 Cryopreserved hepatocyte lots IZT, LHO, and LMP were plated at 2K cells/well as described above. At 4 and 24h after plating, 14 ketoconazole concentrations with each point in quadruplicate (9.2μM–1.1nM) were transferred to the wells using the pin tool and incubated for 10min. One microliter of the Luciferin-IPA substrate from the P450-Glo 3A4 assay was transferred to the wells using the FRD to achieve a 3μM final concentration. The plates were incubated for 1 hour at 37°C. A volume of 5μL of luciferin detection reagent with esterase was dispensed into each well using the FRD and the plates were incubated at room temperature for 30min. Luminescence values were detected using the ViewLux plate reader.
The IC50 values were determined and statistical analysis was performed using GraphPad Prism® 5 (La Jolla, CA). For viability, signal-to-background (S/B) ratio was calculated using luminescence values from DMSO-treated wells divided by tetraoctylammonium (92μM, positive control)-treated wells. For CYP3A4 activity, S/B ratio was calculated using luminescence values from uninhibited wells (DMSO 0.46%) divided by maximally inhibited wells (9.2μM, ketoconazole). Coefficient of variation (CV) was calculated by using the standard deviation (SD) divided by mean. Z′ factor, a measure of the quality of an HTS assay,43 was determined using the equation Z′ factor=1−[3×(SD of positive control+SD of basal)/(median of positive control−median of basal)].
To ensure appropriate metabolic activity associated with the cryopreserved hepatocytes, we first characterized hepatocytes from five different donors for metabolic enzyme level in traditional formats. Seven enzymes (CYP1A2, 2A6, 2C9, 2C19, 2D6, 2E1, and 3A4) representing phase I metabolism and two families of enzymes (UGTs and STs) representing phase II metabolism were evaluated for drug metabolizing activities (Table 2). Of the phase I enzymes, CYP3A4 is responsible for the metabolism of the majority of drugs on the market44 and thus is critical to be present at adequate levels. CYP3A4 activity of the five lots ranged between 50 to 108pmol/min/million cells. From a survey of 172 lots from Celsis In Vitro Technologies, the mean CYP3A4 activity was 100pmol/min/million cells with an SD of 98pmol/min/million cells, representing the diverse range of activity between individuals. CYP2E1 is of interest for occupational and environmental medicine45 given its metabolism of a broad range of chemical classes, including ketones, halogenated alkanes, and monocyclic compounds such as benzene and p-nitrophenol.46 The metabolic activity of CYP2E1 ranged between 13 and 48pmol/min/million cells of metabolite formation of 6-hydroxychlorzoxazone. This compared well against an average of 39pmol/min/million cells with an SD of 36pmol/min/million cells from the survey of 172 donors. All phase I enzyme activities were within one SD of the average as characterized by Celsis In Vitro Technologies (Table 2).
The phase II enzymes for UGT and ST family were measured using the substrate 7-hydroxycoumarin, which is not specific for individual isoforms. Therefore, the metabolite 7-hydroxycoumarin glucuronide reflects multiple UGT isoform activities, whereas the metabolite 7-hydroxycoumarin sulfate reflects multiple ST isoforms. The phase II enzymes were well represented with UGT metabolism ranging between 191 and 558, and ST between 11 and 58pmol/min/million cells. All lots were within one SD of the average as characterized by Celsis In Vitro Technologies (Table 2).
The ability for cryopreserved human hepatocytes to attach to collagen I-coated culture plates was further evaluated by maintaining the culture for 5 days and inducing CYP1A2 and CYP3A4 enzymatic activity in a traditional 48-well cell culturing format. The hepatocytes retained typical cuboidal morphology and a confluence of >70% on day 5 (data not shown). Induction of CYP1A2 and CYP3A4 by exposure to omeprazole and rifampicin was determined by measuring enzymatic activities using specific substrate phenacetin and testosterone, respectively. The fold induction was calculated by dividing the enzymatic rate of the induced hepatocytes by the uninduced rate. CYP1A2 was induced >6-fold and CYP3A4 >11-fold for the lots used (data not shown). According to the FDA guidance, a fold induction >2-fold should be obtained from potent inducers, and the results were within the expected range documented in the guidance.40 Therefore, the plateable cryopreserved human hepatocytes performed as expected compared to historical and documented activities and represent a suitable cellular system to mimic liver functions.
The retention of critical hepatocytic characteristics in miniaturized well format was investigated. At 48h, hepatocytes at 2K/well after 24h DMSO treatment (0.46% final concentration in the wells) showed a monolayer with typical morphology for a hepatocyte culture at approximately 40% confluence (Fig. 1) as expected from the cell density of 80,000 cells per cm2. Hepatocytes at 4K/well showed 80%–90% confluence at 48h incubation (data not shown). A cell density of ~200,000 hepatocytes per cm2 would provide near 100% confluence, which is concordant with literature values.32,47
Determining enzymatic activities using traditional methods as described above were not employed due to technical restraints of a 1536-well format and requirements of UPLC/MS/MS methods. The luminescent P450-Glo Assay with Luciferin-IPA is amenable to HTS formats and has been shown to be specific for CYP3A4 detection. Further, Luciferin-IPA has been used to determine inhibition and induction of CYP3A4 in human hepatocytes.42 Enzymatic activity of CYP3A4 was confirmed by calculating IC50 values of the known specific inhibitor ketoconazole at 4 and 24h after plating cryopreserved hepatocytes. The combined data from three of the donors provided average IC50 values of 0.197μM at 4h and 0.146μM at 24h. Further, the CYP3A4 activity was sufficient to produce sigmoidal curves for the calculation of IC50 values (Fig. 2) with S/B ratios of 11.9±6.8 at 4h and 5.4±1.5 at 24h. The reduction of the S/B over time was expected due to the known loss of CYP activity in cultured hepatocytes.48 The data provided evidence that the hepatocytes retained activity of CYP3A4 in the miniaturized well format.
To determine the potential of utilizing plateable cryopreserved hepatocyte culture in a miniaturized well format, hepatocytes were dispensed into collagen I–coated 1536-well plates at 2K/well and cultured for 4 or 24h before treatment with DMSO (0.46%). A cell density of 80,000 hepatocytes/cm2 (2K/well) was in line with HepG2 culturing conditions previously tested19,20 and therefore was the basis for testing compound cytotoxicity in the current study for IC50 determination as well as assessment of S/B ratio and well-to-well variation.
To confirm the ability to utilize hepatocytes in a cytotoxic assay, a CellTiter-Glo luminescent assay was used to evaluate hepatocyte viability. For control plate assessment, hepatocytes at 2K/well were treated with DMSO only in 1536-well-plate format. The cell viability assay performed well (Fig. 3) with CV values of 4.9% and 4.0% in the DMSO-treated wells for 24h after 4h and 24h plating, allowing hepatocytes attaching to the wells, respectively. To evaluate the assay performance, Z′ factor values of 0.78 and 0.87 were determined in hepatocytes treated with the DMSO (n=1,408) or tetraoctylammonium (92μM) (n=32) for 24h after the hepatocytes attached to the wells at 4 and 24h, respectively. The signal (DMSO wells)-to-background (tetraoctylammonium 92μM) ratios of 33 and 43 were observed after 24h DMSO treatment in the hepatocytes preculture for 4 and 24h incubations, respectively. The concentration response of tetraoctylammonium bromide and potassium dichromate was determined from 16-point curve performed in quadruplicate. The IC50 values of tetraoctylammonium bromide were 3.8 and 3.3μM at 24h treatment in the hepatocytes preculture for 4 and 24h, and 5.2 and 3.2μM for potassium dichromate at 24h treatment in the hepatocytes preculture for 4 and 24h, respectively.
The plateable cryopreserved hepatocytes performed well for attachment efficiency and viability in culture for 48h, and were responsive to concentration–response curves of two known cytotoxins. The assay performance was considered excellent as determined by Z′ factor and warranted further evaluation.
To further optimize the test system, we chose 12 compounds that have been cited in literature as cytotoxins in both in vitro and in vivo systems. The 12 compounds, a subset of the NTP 1408 library that has been used in our laboratory to validate HTS screens, have been shown to be toxic to human- and animal-derived cell lines, including liver-derived HepG2 cells.19,20,49 To evaluate the effects of the 12 compounds, two cell densities, 2K/well and 4K/well representing 40%–50% and 80%–90% confluence, respectively, were dispensed in a 1536-well plate to assess any confluence dependent differences. The 12 compounds were tested for their cytotoxic effects on these hepatocytes after 40h treatment and IC50 values were calculated from the concentration–response curve for 2K/well and 4K/well conditions (Fig. 4). There was good correlation (R2=0.95) between the IC50 values of these 12 compounds for 2K/well and 4K/well conditions and, as well, the curve fit was good with most R2 values above 0.9. At higher concentrations of compounds, decreased cell viability was observed with most compounds achieving viability near 0% with the exception of methylmercury chloride. The potencies of most compounds were slightly higher at 2K/well than 4K/well. The data demonstrate a robust and flexible range of responses, as well as consistency between the 2K/well and 4K/well cell densities. A cell density of 2K/well was used for subsequent testing.
Inter-donor variation was assessed between four of the five individual donors (Table 2). Lot SCT, though used for optimization of the test system, was not used for 12 compound toxicity validation due to lack of supply. We observed good correlation of the IC50 values from different donors with average R values of 0.92 (Table 3). Notably, lot LMP was more sensitive to the compounds tested with higher potencies for most compounds as compared to the other lots. However, the ranking order of the compound cytotoxicity values among the lots was very similar. Inter-assay variability was investigated by assessing the toxicities induced by the 12 compounds on lot IZT on different days. The IC50 values of these compounds are very similar between the experiments (data not shown) with an R value of 0.97. The averaged human hepatocyte responses were further compared to HepG2 cell line to confirm cytotoxicity equivalence (Table 3). The two cell types compared favorably across the compound toxicity panel. There was a good correlation of the IC50 values between HepG2 and human hepatocytes, with an R value of 0.78. Among the 12 tested compounds, doxorubicin was the most potent in both cell types with IC50 values of 1.8μM for averaged human hepatocytes and 1.6μM for HepG2 cells. The other compounds with IC50 values below 10μM were gentian violet, malachite green oxalate, and potassium dichromate. In contrast, tamoxifen was the least potent compound with IC50 values of 62μM for averaged human hepatocytes and 45.1μM for HepG2 cells.
The 12 compounds we selected for this study represent diversity in use and mechanism of toxicity, as well as having no known metabolic dependency for action. For example, malachite green oxalate, gentian violet, and pararosaniline are dyes and have been implicated as being a mitochondrial toxin.50–52 Doxorubicin and tamoxifen are cancer chemotherapeutics and have been cited for in vivo and in vitro hepatoxicities, presumably due to oxidative stress.53,54 Methylmercury chloride, phenylmercuric acetate, and mercury chloride are mercury-containing compounds that have been well established as an environmental toxins and public health risks.55,56 Despite the fact that no definitive primary mechanism of toxicity has been reported, a range of potential mechanisms have been proposed, such as mitochondrial dysfunction, lipid peroxidation, microtubule disruption, and oxidative stress.55
In addition, other researchers have demonstrated toxicity potential of these compounds in cell lines, such as mercuric compounds in human fetal hepatic cell line WRL-68,57 potassium dichromate in HepG2 cells,58 and malachite green oxalate in HepG2 and Caco-2 cell lines.59 Cytotoxicity has also been demonstrated with several of the compounds in primary cells. Staurosporine has been shown to induce apoptosis in murine hepatocytes60 and has been used as a positive control in human hepatocytes for HCS of hepatocellular toxicity markers.61 Several compounds have been documented to induce in vivo hepatotoxicities. For example, malachite green oxalate and gentian violet cause liver damage in fish and mammals.51,62 In addition, doxorubicin and tamoxifen have been implicated in human liver injury.53,54,63
We present here the first confirmed dispensing and culturing of plateable cryopreserved human hepatocytes in collagen I–coated 1536-well microtiter plates, and their survival up to 48h without medium changes. The assay system was consistent and robust as measured by ATP content providing an appropriate HTS platform for measuring viability of hepatocytes, while retaining metabolic function as confirmed by the inhibition of CYP3A4. We validated the system for determining in vitro hepatotoxicity by deriving IC50 values for 12 known cytotoxic compounds that represented diverse chemical properties. The combination of HTS and primary hepatocytes offers a new platform and potential use with qHTS to screen for hepatotoxicity of large chemical libraries. Further studies with compounds dependent upon hepatocytic function, such as metabolism and transport to modulate cytotoxicity, are necessary to delineate the benefits between human hepatocytes and commonly used cell lines such as HepG2. This system may also provide a suitable platform for other endpoints such as metabolic or drug safety studies.
We thank Srilatha Sakamuru and Jinghuan Zhao for technical support. This research was supported by the Intramural Research Programs of the National Toxicology Program, National Institute of Environmental Health Sciences Interagency Agreement Y2-ES-7020-01, and the National Human Genome Research Institute, National Institutes of Health, and the NIH Roadmap for Medical Research Molecular Libraries Program.
No competing financial interests exist.