Integrating machine learning, deep learning, and docking to predict aristolochic acid A carcinogenesis
Longzhu Li1, Jiacheng Liao1, Xintian Chen2
1Guangdong Provincial Key Laboratory of Autophagy and Major Chronic Non-Communicable Diseases, Key Laboratory of Prevention and Management of Chronic Kidney Disease of Zhanjiang City, Affiliated Hospital of Guangdong Medical University, Zhanjiang, Guangdong, China.
Objective:
This study investigates the molecular mechanisms of renal clear cell carcinoma (RCC) induced by Aristolochic acid A (AAA) using machine learning, deep learning, and molecular docking approaches.
Methods:
To identify AAA target genes associated with RCC, differential expression analysis was performed on multiple datasets. Network toxicology, machine learning, deep learning, and molecular docking were used to explore the binding interactions between AAA and target proteins. The top candidate gene was validated using molecular dynamics simulation and in vitro Western blot assays.
Results:
A total of 74 genes were identified as potential targets in AAA-induced RCC. Subsequent machine learning analysis identified seven core genes as key regulators of RCC. Deep learning classification further highlighted five of these seven genes, including PYGL, ADH1B, PTGS1, EDNRA, and AURKA. Additionally, molecular docking simulations revealed strong binding affinities between AAA and these target proteins. Molecular dynamics simulation demonstrated the binding stability of the AAA-PYGL complex, and in vitro studies highlighted PYGL as a potential target of AAA. Elevated expression of PYGL was observed in both 786-O and AAA-induced HK-2 cells. Moreover, treatment with CP-91149 (a PYGL inhibitor) or PYGL knockdown restored the expression of E-cadherin, an epithelial-mesenchymal transition (EMT) marker, in HK-2 cells.
Conclusion:
By combining advanced computational methods with in vitro studies, this work elucidates a key toxicity mechanism of AAA in RCC. Our approach provides a feasible and efficient framework for toxicological studies, offering significant value for toxicologists with limited access to clinical specimens.
Insights
Aristolochic acid A (AAA) causes kidney cancer. This study used AI and lab tests to find AAA targets, identifying PYGL as a key gene involved in AAA-induced renal clear cell carcinoma (RCC) development.
Area of Science:
- Toxicology
- Computational Biology
- Oncology
Background:
- Renal clear cell carcinoma (RCC) is a significant health concern.
- Aristolochic acid A (AAA) is a known nephrotoxic and carcinogenic agent implicated in RCC.
- Understanding the molecular mechanisms of AAA-induced RCC is crucial for developing effective prevention and treatment strategies.
Purpose of the Study:
- To elucidate the molecular mechanisms underlying Aristolochic acid A (AAA)-induced renal clear cell carcinoma (RCC).
- To identify key target genes and pathways affected by AAA exposure using a multi-omics and computational approach.
- To validate the role of identified targets in RCC pathogenesis through in vitro experiments.
Main Methods:
- Differential gene expression analysis across multiple datasets to identify AAA-responsive genes.
- Network toxicology, machine learning, and deep learning for identifying core regulatory genes.
- Molecular docking and dynamics simulations to assess binding interactions between AAA and target proteins.
- In vitro validation using Western blot assays and cell culture models.
Main Results:
- Identified 74 potential AAA target genes in RCC, with seven identified as key regulators by machine learning.
- Deep learning highlighted PYGL, ADH1B, PTGS1, EDNRA, and AURKA as significant drivers.
- Molecular docking confirmed strong binding affinities for AAA with these targets.
- In vitro studies validated PYGL as a direct target, showing elevated expression in RCC cells and its involvement in epithelial-mesenchymal transition (EMT).
Conclusions:
- This study elucidates a key toxicity mechanism of AAA in RCC through integrated computational and experimental approaches.
- PYGL is identified as a critical molecular target in AAA-induced RCC.
- The developed framework offers an efficient approach for toxicological studies, particularly valuable when clinical samples are scarce.