Application of a mouse model humanized for cytochrome P450-mediated drug metabolism to predict drug-drug interactions between a peptide and small molecule drugs.

Conventional preclinical in vitro approaches inaccurately predicted clinical trial outcomes of drug-drug interactions involving the peptide NN1177, a glucagon and glucagon-like peptide 1 receptor coagonist. To further study the mechanisms behind this discrepancy, we have exploited a mouse model (8HUM) humanized for the major cytochrome P450 (P450) enzymes involved in drug disposition in humans. We show that NN1177 administration to 8HUM mice suppressed hepatic in vivo expression of CYP3A4 (82% compared to vehicle) and CYP1A2 (58% compared to vehicle). This was consistent with in vitro sandwich culture hepatocyte data reported previously. However, reduction in CYP3A4 and CYP1A2 levels in vivo appeared to resolve over time, despite daily NN1177 administration. These findings suggest an adaptive response to the metabolic effects of NN1177. In vivo pharmacokinetic studies in 8HUM closely matched the findings observed in the clinical trial, because there was no relevant increase in the exposure of the CYP3A4 and CYP1A2 probe drugs. Furthermore, no suppression effects were observed when the mice had been pretreated with the inducing agents, St. John's wort or phenobarbital, respectively, suggesting that the mechanism of P450 reduction does not involve the transcription factors constitutive androgen receptor or pregnane X receptor. These data highlight the complexities associated with therapeutic peptide drug-drug interactions and the remaining challenges for accurate predictions of P450 suppression and potential clinical implications. The humanized 8HUM model provides a promising and informative preclinical tool that can add high value during drug development by providing further insights into the effects on P450 expression, together with the subsequent impact of coadministered probe drugs in an in vivo model. SIGNIFICANCE STATEMENT: The current work describes the application of a humanized cytochrome P450 mouse model that provides further insight into the potential mechanisms and outperforms conventional in vitro approaches for preclinical predictions of peptide drug-drug interaction risk. The results showed no significant effects on the Cooperstown 5 + 1 cocktail, in line with clinical findings, and thereby represent an exciting model to further explore future therapeutic peptide projects during drug development.