Hiroshi Arakawa, Shengyu Dai, Yukio Kato
Drug metabolism has traditionally been interpreted within enzyme-centric and extended clearance frameworks, in which intrinsic clearance is determined primarily by catalytic activity, enzyme abundance, and plasma membrane transport. Although these paradigms have been highly successful for cytochrome P450 (CYP)-mediated pathways, they do not fully explain persistent discrepancies in non-CYP metabolism, particularly glucuronidation. A key unresolved issue is that many drug-metabolizing enzymes, including UDP-glucuronosyltransferases (UGTs) and carboxylesterases, have catalytic domains oriented toward the endoplasmic reticulum (ER) lumen. Thus, substrates, cofactors, and metabolites must traverse intracellular membranes before and after catalysis, indicating that metabolic flux may be constrained by membrane-delimited transport processes. Here, we propose the multilayered clearance concept, a membrane-resolved flux framework in which effective metabolic clearance emerges from sequential transport and enzymatic processes across plasma and intracellular membranes. In this framework, clearance is governed by the rate-determining step among substrate delivery, ER access, cofactor supply, catalytic turnover, metabolite removal, and cellular export, rather than by enzyme activity alone. This perspective provides a mechanistic basis for interpreting microsome-hepatocyte discrepancies, underprediction of glucuronidation clearance in in vitro-in vivo extrapolation, limitations of current physiologically based pharmacokinetic models, and UGT-mediated drug-drug interactions (DDIs) that cannot be explained by direct enzyme inhibition alone. We further propose classification of UGT-mediated DDIs into enzyme-limited, system-level flux-limited, and intracellular transport-limited types. Although not yet based on a fully parameterized predictive model, the multilayered clearance concept identifies hidden flux-controlling processes that may improve mechanistic interpretation of non-CYP clearance and DDI prediction.