Xenobiotic Metabolism Prediction by a novel methodology

Sumedha B S, Bangalore University

Xenobiotics:

Xenobiotics are chemical substances that are foreign (extrinsic to the normal metabolism of an organism). These include- fragrances, drugs, pesticides, cosmetics, food additives, environmental pollutants, and industrial chemicals. It was found that humans are exposed to about 2 million xenobiotics in a lifetime.

These chemicals gain access to the body through air, food, therapeutics, drinking water, and other lifestyle choices. Detoxification occurs in the body through many pathways. Partial detoxification in the animals consumed by humans determines the degree of exposure.  Xenobiotic metabolism could cause modification of the pharmacological properties of drugs. It can also activate inert chemicals into bioactive compounds.

Heterocyclic aromatic amines (HAAs):

These are a group of chemicals in the diet that are receiving increased attention as they were found to be carcinogenic. HAAs and heterocyclic amines are formed when meat (pork, fish, or poultry) is cooked at high temperatures. When amino acids, creatinine, and sugars react at high temperatures, HAAs are formed.

HAAs are compounds of concern because were found to be mutagenic in bacteria, and carcinogenic in animals. The IARC classifies them as possible and probable carcinogens. These compounds interact with DNA directly, by the formation of covalent adducts.  This process needs biological activation in the liver (by cytochrome P450 enzymes). DNA adducts could lead to mutations and may contribute to the development of cancer.

The metabolites and their DNA adduct formation capacity are examined using prediction tools.

Predicting Genotoxicity:

Genotoxin is a substance that causes DNA damage. To predict genotoxicity, many tools use analyze the capability of a metabolite to bind to DNA (potential DNA adduct). Methods used include:

• Searching for specific chemical structures which can bind to DNA.  As, a known compound with a similar structure, forms DNA adducts.
• Determining if a compound can form DNA adducts. This is based on a Quantitative Structure-Toxicity Relationship (QSAR) score. It models toxicity according to molecular descriptors of compounds.
• Other recent tools utilize deep learning to infer from the characters of each atom if it can bind to DNA. 

Previously, not much data was available about the formation of DNA adducts and the metabolic bioactivation of HAA. This information was obtained by the prediction tools. These tools also allowed the identification of probable biomarkers in humans. The Methods used for the prediction of metabolites and reactions use “biochemical transformation”. 

This links an input molecule (chemical structure) to an output chemical structure. For a compound, the prediction tool searches for the chemical structures identical to the input structure. When they are found, the biochemical transformation rule is applied and the resulting chemical structures are predicted as metabolites.

Several tools utilize such methods of prediction- UM-PPS, MetaSite, TIMES, METEOR, PROXIMAL, META, SyGMa. The predictions made are represented as a Metabolism Map. This is a map of the predicted metabolites and the reactions that link them.

The main drawback of this approach- use of a high number of transformation rules leads to many predictions including unknown metabolites. Also, areas for improvement were highlighted by predicting the map of caffeine and HAA metabolism. In the metabolism map of caffeine, a metabolite- AFMU, is not predicted by the SyGMa tool. This and other evidences suggest that the tool is inadequate in predicting reactions catalyzed by N-acetyl transferases.

Making the Predictions Efficient:

Researchers at the National Cancer Institute, NIH, introduced a new prediction model. It combines the concept of a filtered metabolic map and the ranked pathways. This uses a production probability score, instead of filtering metabolic maps according to individual reaction SOM scores.

This new method can be used to advance the ability in predicting the formation of DNA adducts by HAA. It consists of a three-step pipeline:

1. The prediction of metabolites of the compound of interest,

2. Annotation of the resulting metabolic map using SOM scores 

3. Computation of the production probability score for each metabolite using    Bayesian networks to rank and filter metabolite maps.

Conclusion:

Caffeine was used to validate the modeling approach. It is based on SOM predictions of xenobiotic metabolism enzymes. Caffeine metabolism is well described and it shares enzymes with HAA metabolism. After validation using caffeine, the model was applied to HAA. It was used to predict enzymes responsible for the formation of metabolites capable of binding to DNA.

In conclusion, the study introduced a new methodology for constructing and analyzing metabolism maps. It combines predictions of biotransformation rules, site of metabolisms (SOM), and reactivity to DNA. The method was validated for six xenobiotics. Further studies could apply this to the 24 other human HAAs. This approach opens the perspective to predict xenobiotic metabolites which can bind to DNA adducts in normal or physio-pathological conditions.

Also read: Alzheimer’s disease and the liver: Are they linked?

References:

1. Conan, M., Théret, N., Langouet, S., & Siegel, A. (2021). Constructing xenobiotic maps of metabolism to predict enzymes catalyzing metabolites capable of binding to DNA. BMC Bioinformatics, 22(1), 450. https://doi.org/10.1186/s12859-021-04363-6
2. Patterson, A. D., Gonzalez, F. J., & Idle, J. R. (2010). Xenobiotic metabolism: a view through the metabolometer. Chemical research in toxicology, 23(5), 851–860. https://doi.org/10.1021/tx100020p

• The Corrosion Prediction from the Corrosion Product Performance
• Nitrogen Resilience in Waterlogged Soybean plants
• Cell Senescence in Type II Diabetes: Therapeutic Potential
• Transgene-Free Canker-Resistant Citrus sinensis with Cas12/RNP
• AI Literacy in Early Childhood Education: Challenges and Opportunities