Integration of genome-scale metabolic networks into whole-body PBPK models shows phenotype-specific cases of drug-induced metabolic perturbation

Henrik Cordes,Vanessa Baier,Christoph Thiel,Lars Kuepfer,Lars M Blank

doi:10.1038/s41540-018-0048-1

Abstract

Drug-induced perturbations of the endogenous metabolic network are a potential root cause of cellular toxicity. A mechanistic understanding of such unwanted side effects during drug therapy is therefore vital for patient safety. The comprehensive assessment of such drug-induced injuries requires the simultaneous consideration of both drug exposure at the whole-body and resulting biochemical responses at the cellular level. We here present a computational multi-scale workflow that combines whole-body physiologically based pharmacokinetic (PBPK) models and organ-specific genome-scale metabolic network (GSMN) models through shared reactions of the xenobiotic metabolism. The applicability of the proposed workflow is illustrated for isoniazid, a first-line antibacterial agent against Mycobacterium tuberculosis, which is known to cause idiosyncratic drug-induced liver injuries (DILI). We combined GSMN models of a human liver with N-acetyl transferase 2 (NAT2)-phenotype-specific PBPK models of isoniazid. The combined PBPK-GSMN models quantitatively describe isoniazid pharmacokinetics, as well as intracellular responses, and changes in the exometabolome in a human liver following isoniazid administration. Notably, intracellular and extracellular responses identified with the PBPK-GSMN models are in line with experimental and clinical findings. Moreover, the drug-induced metabolic perturbations are distributed and attenuated in the metabolic network in a phenotype-dependent manner. Our simulation results show that a simultaneous consideration of both drug pharmacokinetics at the whole-body and metabolism at the cellular level is mandatory to explain drug-induced injuries at the patient level. The proposed workflow extends our mechanistic understanding of the biochemistry underlying adverse events and may be used to prevent drug-induced injuries in the future.

Highlights

Drug-induced adverse events are a common clinical, and an increasing public health problem.[1]
Context-specific extraction algorithms make use of the genome-scale metabolic network (GSMN) model structure to account for RESULTS The multi-scale physiologically based pharmacokinetic (PBPK)-GSMN modeling workflow Cellular toxicity is a key manifestation of drug-induced adverse metabolic flux activity that is not reflected in gene expression data.[20]
The resulting multi-scale PBPK-GSMN models allow the quantification of organ-specific endogenous cellular responses and changes in exometabolome pools

Summary

Introduction

Drug-induced adverse events are a common clinical, and an increasing public health problem.[1]. A computational workflow that combines both scales of scale flux distributions are iteratively calculated with the resulting PBPK-GSMN models that allow a quantitative assessment of biological organization would be a viable tool for the under- cellular responses during drug-induced metabolic perturbations.

Results

Conclusion