In silico modeling of the AHAS inhibition of an augmented series of pyrimidine herbicides and design of novel derivatives

Abstract

Pyrimidine compounds comprise a class of acetohydroxyacid synthase (AHAS) inhibitors, thus possessing herbicidal activity. Due to the ongoing development of resistance by weeds to current herbicides, the design of new agrochemical candidates is often required. This work reports the proposition of unprecedented pyrimidines as herbicides guided by quantitative structure-activity relationship (QSAR) modeling. Multivariate image analysis (MIA) descriptors for 66 pyrimidine derivatives obtained from different sources were regressed against inhibitory activity data, and the resulting QSAR models were found to be reliable and predictive (r2 = 0.88 ± 0.07, q2 = 0.53 ± 0.06, and r2pred = 0.51 ± 0.10 in a bootstrapping experiment using electronegativity-based descriptors). The chemical features responsible for the herbicidal activities were analyzed through MIA contour maps that describe the substituent effects on the response variables, whereas the interaction between the proposed compounds and AHAS was analyzed through docking studies. From the proposed compounds, at least five pyrimidine derivatives exhibited promising performance as AHAS inhibitors compared to the known analogs.