You are here:
Quantitative Nuclease Protection Assays (qNPA) as Windows into Chemical-Induced Adaptive Response in Cultures of Primary Human Hepatocytes (Concentration and Time-Response)
Citation:
BEAM, A., D. ROTROFF, K. FREEMAN, A. FARMER, D. REIF, K. A. HOUCK, R. JUDSON, D. J. DIX, E. L. LECLUYSE, AND S. T. FERGUSON. Quantitative Nuclease Protection Assays (qNPA) as Windows into Chemical-Induced Adaptive Response in Cultures of Primary Human Hepatocytes (Concentration and Time-Response). Presented at ToxCast Data Analysis Summit, RTP, NC, May 14 - 15, 2009.
Impact/Purpose:
We have characterized the bioactivity of the 309 unique chemicals currently in the ToxCast library in cultures of primary human hepatocytes over ranges of concentration and time. Correlations were observed between activation of key receptor pathways and certain rodent in vivo toxicity endpoints. These correlations indicate the value of using this in vitro hepatocyte culture systems in predictive toxicity modeling, and identifies putative human toxicity pathways for specific disease endpoints.
Description:
Cultures of primary human hepatocytes have been shown to be dynamic in vitro model systems that retain liver-like functionality (e.g. metabolism, transport, induction). We have utilized these culture models to interrogate 309 ToxCast chemicals. The study design characterized both concentration- and time-response effects of these chemicals across two preparations of human hepatocytes by mRNA expression, CYP1A enzymatic activity (EROD), and cell morphology. mRNA expression was determined using quantitative nuclease protection assays (qNPA™) with the Omix™ Imaging System (HTG, Tucson, AZ). Fourteen liverrelated human gene targets ABCB1, ABCB11, ABCG2, SLCO1B1, CYP1A1, CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP3A4, UGT1A1, GSTA2, SULT2A1, HMGCS2 (Figure 1) were monitored based on their role in liver xenobiotic metabolism, hepatic transport, and sensitivity to receptor pathways (AhR, CAR, PXR, PPARα, FXR). These data were analyzed relative to negative and positive control receptor activators. These data were fit to sigmoidal concentration response model (Hill equation) to generate important potency and efficacy parameters (e.g EC50, Emax, Hillslope, R2 etc…). Concordance analysis was performed on the internal replicate ToxCast chemicals to assess the reproducibility of the assays. In addition, techniques from machine learning were leveraged to cluster compounds having similar gene response profiles. The concentration-response of a compound was abstracted as a vector (rather than classical scalar representations associated with standard microarray analysis) and used in algorithms such as K-means and algometric clustering, as well as creating representative phylogenies. Unique to this approach is the ability to assess if compounds behave similarly in a temporal sense. Using this methodology we were able to correlate how a chemical’s behavior compares with other compounds through time, as well as correlating gene targets with one another. These chemical signatures were further correlated with in vivo endpoints in relative risk assessments to define in vitro profiles that appear to be related to phenotypic outcomes.