Quantitative structure-activity relationships for PPAR-gamma binding and gene transactivation of tyrosine-based agonists using multivariate statistics.

Peroxisome proliferator-activated receptor-gamma offers a molecular target for drugs aimed to treat type II diabetes mellitus, while its therapeutic potency against cancer disease is currently being explored in preclinical studies. Tyrosine derivatives constitute a major class of peroxisome proliferator-activated receptor-gamma agonists attracting considerable research interest in drug discovery. Thus, the establishment of adequate QSAR models would serve as a guide for further molecular design. In the present study, multivariate data analysis was applied on a large set of tyrosine-based peroxisome proliferator-activated receptor-gamma agonists for modelling binding affinity, expressed as pKi and gene transactivation, expressed as pEC(50). A pool of descriptors based on physicochemical and molecular properties as well as on specific structural characteristics was used and two PLS models with satisfactory statistics were produced for binding data. According to them, molecular weight, rotatable bonds and lipophilicity were found to exert a considerable positive influence, while excess negative and positive charge created by additional acidic or basic groups in the molecules was unfavourable. With gene transactivation data, an adequate model was obtained only for the highly active compounds if considered separately. The higher complexity incorporated in gene transactivation data was further investigated by establishing a PLS model, which improved the inter-relationship between pEC(50) and pKi.