We previously demonstrated that induction of hepatic cytochrome P4503A (CYP3A) immunoreactive protein is a response in rats, but not rabbits, treated with the antiglucocorticoid, pregnenolone 16acarbonltrile and in rabbits, but not rats, treated with rifampicin. These striking interspecies differences in response to CYP3A inducers prompted us to compare the effects of a variety of agents on CYP3A expression in primary cultures of hepatocytes from rats, rabbits, and humans, maintained under nearly identical conditions on Matngel. We used treatment with dexamethasone, the most effective inducer of CYP3A mRNA and CYP3A immunoreactive protein in cultures of rat hepatocytes, to define the 100% response. As expected from their effects in viva, incubations of cultures with medium containing pregnenolone 16a-carbonitrile or rifampicin induced CYP3A mRNA to high levels exclusively in rat hepatocytes or rabbit hepatocytes, respectively. Pregnenolone 16a-carbonitrile treatment also did not induce CYP3A immunoreactive protein in rabbit hepatocytes, although rifampicin treatment did increase CYP3A immunoreactive protein levels in rat hepatocyte cultures. Additions of phenobarbital to the cultures induced CYP3A mRNA and CYP3A immunoreactive protein to a greater extent in rabbit hepatocytes(94.-I 08% of the dexamethasone response) than in rat hepatocytes(38-57% of the dexamethasone response). In primary cultures of human hepatocytes, dexamethasone and phenobarbital treatments induced CYP3A mRNA (4.4-and I. 9-fold, respectively, over the amounts of CYP3A mRNA in incubated control cultures). In each preparation of cuftured human hepatocytes examined, rifampicin treatment increased the amount of CYP3A mRNA (1. 5-to 8.5-fold over untreated controls), whereas pregnenolone l6ci-carbonitrile treatment induced CYP3A mRNA in only 2 of 4 human cultures tested, raising the possibility of heteroge-neity among humans in this response. Incubation with RU-488 or trans-nonachlor(each at iO M) induced CYP3A gene products in cultured hepatocytes from all three species. Likewise, treatment of the cultures with lovastatin, a hypocholesterolemic agent reported to elevate CYP3A mRNA levels in cultured human hepatocytes, increased the amounts of CYP3A mRNA and CYP3A immunoreactive protein in rabbit (39-58% of the dexamethasone response) and in rat (11-35% of the dexamethasone response) hepatocytes. In contrast, incubation of cultures with spironolactone, cyproterone acetate, or clotrimazole resulted in species-specific effects, inducing CYP3A gene products to high levels in hepatocytes from rats and humans, but not rabbits. We conclude that there are dramatic interspecies differences in CYP3A induction, intrinsic to the hep-atocyte, that likely preclude the use of a single type of laboratory animal from reliably predicting the response in humans.

A major unsolved problem in modern toxicology is the uncertainty in the use of laboratory animals as test surrogates for humans in making safety assessments. It is clear that there can exist critical differences among species in expression of genes involved in the outcome of a toxic response. For example, the liver cytochromes P450 3A, an important locus of interaction between man and his chemical environment, have been detected in all species examined, including