Cryopreserved human hepatocytes help HTS take the next step forwardThe advantages of high-throughput screening have been expounded upon by many for its efficiencies and confirmed by its acceptance in the pharmaceutical and biotechnology industries and government agencies. This has developed from advances in liquid handling systems, diagnostic equipment, and reagents.
The science behind and understanding of the processes involved in deriving useful and accurate data have brought us success with HTS. Growth in HTS will continue as it moves away from simple systems and binary information to more complex models and interpretable data. Two recent publications illustrate these improvements.
HTS started with simple systems such as bacterial cultures and receptor binding. It quickly incorporated cell lines to add to the complexity of biological systems while fulfilling the need for a scalable and renewable source of material. However, cell lines may not fully mimic the in vivo physiology of the cell type they are intended to represent.
A case in point is the HepG2 cell line derived from a human hepatocellular carcinoma. HepG2 cells have been used as a surrogate for hepatocytes, the primary functional cell type of the liver, to study various liver cell functions like metabolism, transport, drug-drug interactions, and cytotoxicity in traditional cell culture and HTS models. However, significant differences, such as lower expression levels of drug-metabolizing enzymes compared to primary human hepatocytes, have limited the ability of HepG2 to predict in vivo situations.
It is well accepted that primary human hepatocytes provide the most apposite in vitro model to predict in vivo function and are, therefore, the gold standard for pharmaceutical and toxicological studies. However, the use of primary human hepatocytes in HTS has been limited by reliance on unscheduled fresh isolations and inability to retest from the same donor. Recent advances in the quality and availability of cryopreserved human hepatocytes have provided the opportunity to exploit them in HTS models for metabolism, drug-drug interaction, and toxicity studies.
The Flexibility of HTSIn a 2011 manuscript, Moeller and colleagues presented the first reported use of cryoplateable human hepatocytes in 1536-well format.1 The validation included confirmation of metabolic activity of cytochrome P450 (CYP) 3A4 in culture and the derivation of toxicity IC50 values from a 16-point concentration response curve after a 40-hour exposure to known cytotoxic compounds. The format provided a balance between a high number of compounds tested and the high number of data points observed. The former was in line with the original purpose of HTS to screen large chemical libraries, while the latter has followed the current trend toward quantitative HTS.
Traditional HTS relied on a single concentration to provide a qualitative assessment. This approach may be acceptable with criteria that are well established to provide actionable and interpretable data. However, qHTS has provided a wider range of concentrations in order to assess the effect when significant concentration thresholds are not known. In this study, the concentration response curve was between 92 µM and 3 nM. The range of IC50 values was between 2 µM for doxorubicin and 90 µM for tamoxifen.
Primary human hepatocytes provide the most apposite in vitro model to predict in vivo function and are the gold standard for pharmaceutical studies.If a single concentration of 10 µM had been employed, about half of the compounds tested would have been negative for cytotoxicity. The significance of the in vitro data from a single concentration or concentration response curve must be vetted with observed toxicity. Although this example doesn’t determine the more significant approach, it does illustrate the flexibility of HTS in meeting the needs of research.
Match Data Quantity, QualityThe quantity of data provided by HTS must be matched by quality. That is, if designed appropriately, the data generated should be able to be interpreted with confidence as to the mechanism of action for the observed event. To achieve this, multiple markers have been used to capture sufficient data in order to discern among multiple pathways in a cell-based screen.
One method is multiplexing, employing two or more signals obtained from a single well. Larson and colleagues presented a model for measuring the induction of metabolic enzymes with drugs using three distinct markers.2 The model was based on the gold standard of measuring enzyme activity for key metabolic enzymes—CYP1A2 and 3A4. The novelty was to perform these measurements, along with a viability marker, from the same well of a 384-well microtiter plate.
Traditional methods relied upon separate incubations between wells to measure the individual event. The data would then be combined to assess the capacity of a test compound for its induction potential. Inherent error, including seeding densities between wells, position of the wells on the plate, microenvironments between wells, and dosing differences, may affect the fidelity of the data when combining separately acquired data into one profile. By measuring these markers from the same well, many of the potential errors may be mitigated, increasing confidence in the data. EC50 values from concentration response curves may be generated due to the capacity of the 384-well format, a process made more difficult in the commonly used 24-well formats.
Two examples highlight the benefits of multiplexing CYP1A2 and 3A4 signals with a viability marker.
Rifampicin is a well-known potent inducer of CYP3A4. The data generated from a concentration response curve provided an EC50 of approximately 2 µM with slight induction of CYP1A2 at the highest concentration (see Study 1A). The viability as measured by ATP content remained stable at 100% compared to vehicle control.
A lower fold induction than the Emax was observed at the highest concentration of rifampicin. If this had been observed with only the CYP3A4 activity as the marker, possible interpretations might be cytotoxicity, inhibition, or suppression, and separate investigations for each possibility would be required to determine its cause. With viability included as a marker, cytotoxicity can be ruled out, leaving inhibition and suppression as possible mechanisms. From the literature, rifampicin has been shown to be an inhibitor of CYP3A4 at high concentration, providing a plausible explanation for the reduced induction observed at the highest concentration.
Lansoprazole presented a different profile and interpretation (see Study 1B). Lansoprazole was confirmed to be an inducer of CYP1A2 from the concentration response curve. It, too, showed lower induction values than the Emax at higher concentrations; however, the viability marker was lower at higher concentrations, indicating lansoprazole-induced cytotoxicity. In this case, cytotoxicity was the cause of the drop in induction at higher concentrations. This was collaborated by the observed reduction of CYP3A4 signal at similar concentrations. Interpretation of the data can be made with confidence, given the multiple markers observed across the concentration response curve.