5.3.3 Induction Models
The number of models to evaluate induction of drug-metabolizing enzymes has
increased tremendously over the past decade. As mentioned previously, animal
models are not amenable to assessing the induction potential of humans due to
differences in the LBD of most nuclear receptors. As a result, nearly all of the
current induction models are based on human reagents. One of several high
throughput models (cell free) involves a measurement of ligand binding to the
expressed NR in scintillation proximity type assays (Zhu et al., 2004). A simple
cell-based model involves the coexpression of a nuclear receptor along with an
appropriate gene promoter region coupled to a reporter gene. An example of
this type of model utilizes HepG2 cells where human PXR is expressed along
with a second vector expressing a partial CYP3A4 promoter fused to the
luciferase gene. Activation of PXR proceeds as previously described except
that the luciferase gene is induced and a direct luminescent readout can be
measured (Luo et al., 2002, 2004). In the case of CAR activation, measurement
of CAR translocation into the nucleus of primary hepatocytes has been used
successfully to identify activators of this receptor (Wang et al., 2004). Another
cell-based system that has the ability to measure multiple drug-metabolizing
enzymes and transporters within a single cell system is the immortalized
hepatocyte cell line, Fa2N-4 (Mills et al., 2004). Ultimately, the present ‘‘gold
standard’’ for assessing human induction of drug-metabolizing enzymes are
primary cultures of fresh or cryopreserved human hepatocytes (LeCluyse,
2001a). In this situation, all of the NR, corepressors/activators, dimerization
partners, response elements, and target genes are present in their natural
environment. Although there are issues with the timely acquisition of human
hepatocytes from multiple donors, this model remains the most representative
of thein vivosituation in humans.
Transgenic animal models of human induction have received greater attention
as a means of moving away from the staticin vitromodels to a more dynamicin
vivosituation where concentration–time effects, and how other absorption–
distribution–metabolism–excretion (ADME) properties affect the overall level of
enzyme induction (Xie and Evans, 2002). As an example, there are mouse strains
where the mouse PXR or CAR has been knocked out and the human NR
inserted. It has been shown in the case of the humanized PXR mouse, that
rifampicin (a human PXR ligand) activates the hPXRin vivoand initiates
transcription of mousecyp3a11(Ma et al., 2007; Xie et al., 2000). At similar
doses of rifampicin in wild type mice, no induction ofcyp3a11is found because
rifampicin is not an effective ligand for mouse PXR. One caveat exists for these
models, only one gene has been humanized (i.e., the NR) and all other aspects of
the mouse remain normal. Drugs evaluated in thesein vivosystems will be
absorbed, distributed, metabolized, and eliminated according to mouse
physiology and not human physiology, albeit they are sometimes similar.
Additional studies and validation will be necessary to fully support the use of
these humanized animal models to predict human induction.
ENZYME INDUCTION 125