It is evident, therefore, that while animal studies may offer valuable
information regarding the potential of compounds for inducing those drug-
metabolizing enzymes whose mechanisms and molecular mediators are
qualitatively and quantitatively similar in preclinical species and in humans
(e.g., CYP1A family members), and they are expected to be poorly predictive
and marginally useful in cases where major species differences exist (e.g.,
CYP3A family members). However, recent technological developments offer a
third possibility. The generation of transgenic mouse models, which
incorporate genes that code for human-specific signaling molecules, can result
in humanized mice, which respond to potential inducers that interact with
those molecules in a manner very similar to that of humans. The main
determining factor for the species-specific pattern of CYP3A inducibility is the
interaction of the inducer with the species-specific PXR receptor, rather than in
the interaction of the activated PXR with the xenobiotic response element
(XRE) in the regulatory region of the CYP3A4 gene (Luo et al., 2004; Xie et
al., 2000). Xie et al. have generated a transgenic mouse strain that lacks the
mouse PXR gene, but expresses the human PXR gene. This animal model
appears to provide a reliable assessment of the CYP3A induction potential of
candidate molecules in an intact mouse whose response to xenobiotic PXR
ligands mirrors that of humans (Xie et al., 2000; Xie and Evans, 2002). These
humanized mice offer a potentially powerful tool for drug discovery and
development that will reliably predict the ability of candidate molecules to
induce CYP3A enzymes in humans. Furthermore, because this is a whole-
animal system, the effects of CYP3A induction (as well as the activation of
other PXR-mediated pathways) on the pharmacodynamics, pharmacokinetics,
and toxicological profiles of test compounds can be directly determined.
We see then, that while the results of testing for inducers of drug
metabolizing enzymes in preclinical species cannot be automatically assumed
to provide a reliable assessment of their potential for producing similar effects
in humans, in certain cases, animal models may serve as a valuable bridge
betweenin vitroand cell culture studies on one hand, and clinical studies on the
other. When the potential predictive value of animal studies warrant their
consideration, some general principles related to study design should be kept in
mind. Animals are normally treated from one to several days with multiple
bolus oral, IP, or IV doses of potential inducing agents, or exposed for longer
periods of time via the diet. The length of treatment will depend on the rate of
clearance of the inducing agent and the rate of turnover of the target protein.
Normally, 4–5 days of single dose treatment is required to achieve maximal
induction. The dose–response curve for induction is relatively steep, and
usually, only a maximal response is sought. Therefore, a complete dose–
response curve does not need to be generated, and animals can be treated at or
near the maximally tolerated dose, as determined by previous acute toxicity
studies. In the case of humanized transgenic mice, expense and availability may
be an issue. Numbers of animals may, therefore, be limited to three or four per
treatment group. The route of administration must also be considered. The
550 TESTING DRUG CANDIDATES FOR CYP3A4 INDUCTION