Japanese researchers have built a computer model that reproduces how the human enzyme CYP1B1 handles the chemicals it metabolizes and blocks, and have paired it with an existing model of its close relative CYP1A1. The work, published in the journal Genes and Environment, reports that more than 260 reactions of ligands for these two enzymes could be verified on the paired template systems, with placements on the templates and a set of interaction rules faithfully reproducing which compounds are good or poor substrates, where on each molecule oxidation occurs, and how inhibitors interfere with the enzymes. Because cytochrome P450-mediated metabolism is central to judging the safety of chemicals, a system that reliably anticipates these reactions could matter for toxicology and drug development.
Cytochrome P450 enzymes are a family of proteins that metabolize hydrophobic molecules the body encounters, both endogenous compounds and foreign chemicals. CYP1A1 and CYP1B1, the two enzymes modeled here, are expressed polymorphically in hepatic and extrahepatic human tissues and mediate the metabolism of endobiotics and xenobiotics, including the metabolic activation of promutagens and procarcinogens. Assessing metabolism at the level of individual CYP enzymes is necessary for efficacy and safety evaluations. Yet despite decades of crystal-derived three-dimensional protein models, predictions of the regio- and stereo-selective metabolism of small ligands remain far from established, which is the gap the template approach is designed to close.
The new CYP1B1 system was constructed by reverse-engineering the ligand-accessible space of the enzyme from assemblies of its substrates. The authors first gathered experimental information on substrate specificities and metabolism from the literature, preferring published data from recombinant human CYP1A1 and CYP1B1 systems because these directly reflect the properties of each enzyme. CYP1B1 substrates were picked up at random, and their three-dimensional structures, flattened into conformations that fit the template and adjusted with molecular mechanics calculations, were assembled with a defined width between two virtual walls. Focusing on the positions metabolized at the bottom of this assembly produced a ligand-accessible space, which was then overlaid on the previously established CYP1A1 template to extract the CYP1B1 template and assign its rings and position numbers.
The resulting templates consist of hexagonal grids and sticks. A substrate atom sitting at the corner of a hexagonal grid, termed a Ring, counts as occupancy, and placements are recorded in a hyphen-linked form tracing which rings a molecule occupies. For a ligand to function, the model requires five things: simultaneous plural contact with the Rear-wall, contact with the Facial-wall, contact with a feature called the Bay-2 residue, interaction with the Site of oxidation, and interaction with a Trigger-residue. Most ligands satisfy these requirements as a single molecule; some, like phenanthrene in its 9,10-oxidation, cannot interact simultaneously with both the Site of oxidation and the Trigger-site and are modeled as two-molecule bindings.
The comparison revealed close similarities between the CYP1B1 and CYP1A1 templates in distribution area, Site of oxidation, and allowable width, but also two telling differences. CYP1B1 substrates use a feature called Position 52′, described as Pier-sitting, that CYP1A1 substrates do not, and CYP1B1 substrates do not enter the bottom middle region of the template, Rings F and Ea, which CYP1A1 substrates can occupy. Both enzymes oxidize substrates mainly at the region connecting Positions 4 and 52. The authors report that these distinct uses of template regions, together with differences in Bay-2 residue contacts, account for the different catalytic behaviors of the two enzymes across examples including benzo[c]chrysene, febuxostat, imatinib, and estrone.
A second distinction concerned the Bay-2 residues themselves, regions of avoidance for ligand sitting whose locations were set from the ligand assembly results. Comparing simulation placements with experimental data for compounds such as dibenzo[a,l]pyrene, Sudan I, benzo[c]phenanthrene, and 1-nitropyrene suggested that the CYP1A1 Bay-2 residue sits on the facial side of the template’s width gauge, while the CYP1B1 Bay-2 residue lies in the middle region. The CYP1A1 residue was drawn as a square of 0.6 ring-size diameter and the CYP1B1 residue as an oval of 0.5 ring-size diameter, sizes chosen arbitrarily from the agreement between simulations and measured substrate behavior. For 1-nitropyrene, the model predicted that the CYP1B1 residue would contact the nitro group in the 2,3-oxidation placement while the CYP1A1 residue would not, matching experimental data showing substantial lack of CYP1A1 activity for that reaction.
The quantitative performance of the system is a headline claim. In total, 267 reactions of 214 distinct chemicals reported as CYP1B1 ligands were examined on the CYP1B1 template: 137 good oxidations or reductions, 42 poor catalyses, and 105 inhibitions, with molecular masses ranging from 131 for 3-methylindole to 532 for ponatinib. Placements were available for all reactions of good catalysis and inhibition, and the authors report verifications of good and poor substrates and of regio- and stereoselectivity with more than 99 percent accuracy.
The templates were also used to probe how inhibitors work. Applying various intense inhibitors to both templates, the authors found two common structural features: placements hanging at Position 32 among CYP1A1 ligands, and placements adhering to the Positions 24–52 line among CYP1B1 ligands. Both features, they suggest, retard the dissociation of the Bay-2 residue from the ligand molecule, making interference with Bay-2 residue dissociation a possible mechanism of ligand-mediated inhibition on both enzymes. The pattern fit chemicals including 7,8-benzoflavone, quinazolinone derivatives with CYP1B1 inhibitory concentrations in the low nanomolar range, rutaecarpine derivatives, methoxy-stilbenes, and several polycyclic aromatic hydrocarbons such as dibenzo[a,h]anthracene and 3-methylcholanthrene, which inhibit CYP1B1-mediated activity far more strongly than CYP1A1.
The authors are careful to frame the inhibitory mechanism as a proposal rather than a demonstrated result, stating that further study is necessary to prove it. The template system itself is a ligand-based reconstruction, not a crystal structure, and its Bay-2 residue sizes and positions were set by fitting simulation output to experimental observations rather than measured directly. The underlying substrate data come from published recombinant enzyme experiments, so the model inherits the scope and limitations of that literature. No datasets were generated or analyzed during the study, and the work was partially supported by a grant from Japan’s Ministry of Health, Labour, and Welfare. Within those limits, the paired CYP1A1 and CYP1B1 templates offer toxicologists and medicinal chemists a transparent, rule-based way to anticipate how these promutagen-activating enzymes will handle new chemicals.
Subject of Research: An in silico fused-grid template system for predicting CYP1A1- and CYP1B1-mediated metabolism and inhibition of chemicals
Article Title: CYP1B1- and CYP1A1-Template systems and their application to metabolism and inhibition
Article References: Yamazoe, Y., Ambe, K., Tohkin, M., Yamada, T., & Masumura, K. (2025). CYP1B1- and CYP1A1-Template systems and their application to metabolism and inhibition. Genes and Environment, 48(1), Article 1. https://doi.org/10.1186/s41021-025-00351-x
Image Credits: AI Generated
DOI: 10.1186/s41021-025-00351-x
Keywords: CYP1B1, CYP1A1, cytochrome P450, template system, in silico prediction, xenobiotic metabolism, regioselectivity, enzyme inhibition, Bay-2 residue, toxicology, procarcinogen activation, computational model
News Source: Drew Townsend. (October 29, 2026). New Template System Predicts How Two Key Enzymes Metabolize and Inhibit Chemicals. Scienmag.
