Harnessing the Potential of Carotenoids for Cancer Therapy: An Integrated Machine Learning and <scp>MST</scp> Based Approach

Receptor tyrosine kinases (RTKs) are high-affinity membrane-anchored receptors involved in cellular communication via various ligands and manage numerous biological processes such as cell growth, differentiation, and metabolism. However, dysregulation of RTKs is a key instigating factor in the development of a vast array of cancers. Carotenoids are a major family of secondary plant metabolites known for their anti-cancer activities in various cancer models by targeting several molecular intermediates. We aimed to decipher the potential carotenoids as RTK inhibitors through an integrated workflow of in silico approaches and in vitro microscale thermophoresis. The kinase domains of nine RTKs were subjected to molecular docking with potential carotenoids, and the best-scoring carotenoids were selected. The molecular interactions of the best-scoring carotenoids and respective RTKs were validated through dynamics simulation. The selected carotenoid candidates were further validated through comparative analysis with clinically established drugs using various machine learning algorithms to establish the drug likeliness. Microscale thermophoresis was performed to prove the interaction of the best-scoring carotenoid with recombinant PDGFRA and VEGFR2 in vitro. The following five receptors and respective carotenoids were recognized through docking, MDS, and ML analysis: EGFR-fucoxanthin, FGFR2-peridinin, VEGFR2-canthaxanthin, PDGFRA-canthaxanthin, and ALK-crocin. MST experiments further underlined the high binding affinity of canthaxanthin with the targeted RTKs, underlining the possibilities of plant-based chemotherapy. Interestingly, carotenoids were recognized as potential plant-based alternatives for conventional drugs in RTK-targeted cancer therapy via an innovative ML-assisted drug discovery approach, and they provide novel insights into the discovery of phytochemicals as cancer drugs.