Construction of a predictive model for the design of triptamin analogues with potential activity in Parkinson's and Alzheimer's diseases

Abstract

A growing public health issue has been caused by the diagnosis of neurodegenerative pathologies in millions of people. These conditions are caused by the progressive degeneration of brain cells, which results in the loss of synaptic connections between neural networks and causes issues with a variety of cognitive, motor, and memory functions. Although the symptoms of these disorders can be reduced with current medications, patients often have side effects and develop drug dependence. Tryptamines have gained attention in recent years due to their varied agonist activity at 5-hydroxytryptamine (5 HT) receptors and their structural resemblance to serotonin. This is based on current knowledge and numerous advancements in neuroscience. The INQA Artificial Neural Network was used to train a QSAR model with 31 samples of tryptamines with IC50 values. Pearson Correlation (descriptor vs. descriptor) was used to choose the molecular descriptors that fed the model looking for low or no categorization between them. The descriptors and derived descriptors were then correlated again with the IC50 value in order to achieve high precision; among the resulting ones, we chose 8 from constitutional and topological categories for their chemical relevance. With the built model, we designed more than 30 tryptamine analogues, predicted their IC50, calculated their affinity in kcal/mol, and evaluated their toxicity in silico, yielding two candidates with high affinity, low IC50, and good toxicology profiles that could be potential candidates for treating parkinson (5p) and Alzheimer's (6A) disease.