Fragment-Based Design of Pesticide-Like Molecules: From Method Development to Agrochemical Discovery

Abstract

Driven by rapid global population growth, an unprecedented demand for food has placed severe pressure on agricultural production. While synthetic pesticides have served as a cost-effective means of safeguarding crop yields, the effectiveness of many existing agents is increasingly undermined by rising pest resistance and stricter regulatory safety margins, necessitating continuous innovation in more effective and safer agrochemicals. Conventional agrochemical discovery relies on phenotypic screening directly against whole organisms, which requires substantial labor, funding, and time. Alternatively, target-based approaches using in vitro assays have emerged as a powerful paradigm for the rational design of agrochemicals, increasingly supported by computational strategies for ligand identification and optimization. Thus, advancing computational methods for target-based pesticide design is essential for accelerating the discovery of next-generation agrochemicals and supporting sustainable agriculture. Fragment-based design, a promising target-based approach developed primarily in the pharmaceutical industry, can be extended to agrochemicals discovery. By leveraging the additive binding contributions of molecular fragments to discrete protein subpockets, this technique enables small molecules to be optimized fragment by fragment, thereby significantly expanding chemical space for exploring novel bioactive scaffolds. The growing number of experimentally determined structures of agrochemical targets, together with rapid progress in protein structure modeling technologies, has provided valuable foundations for fragment-based pesticide design. However, adapting the fragment-based strategy from pharmaceutical research to agrochemical discovery requires careful consideration of unique agrochemical molecular properties such as photostability, ecotoxicity, and bioavailability. Therefore, the implementation of fragment-based agrochemical discovery still poses critical questions, such as how to acquire structurally diverse fragments that can be readily assembled into pesticide molecules and how to assess the suitability of a fragment or ligand for subsequent optimization into pesticide leads. In this Account, we describe our efforts to address these issues by incorporating the consideration of pesticide-likeness and ecotoxicity into the fragment-based design pipeline and establishing an integrated computational framework for fragment-based agrochemical discovery. This framework supports key stages from fragment library design to fragment-to-lead optimization through a comprehensive suite of molecular tools, including (i) the qualitative and quantitative assessment methods for pesticide-likeness (HerbiPAD, InsectiPAD, FungiPAD, and Pesti-DGI-Net); (ii) the ecotoxicity predictive tools (beetox, AquaticTox, and PEAR) for rapid screening of eco-friendly candidates; (iii) the PADFrag database of pesticide- and drug-derived bioactive fragments, and the DigFrag method for de novo fragment generation; (iv) the pharmacophore-linked fragment virtual screening (PFVS) method and the online ACFIS server for fragment-to-lead optimization; and (v) the Auto In Silico Ligand Directing Evolution (AILDE) server for iterative hit-to-lead optimization. Furthermore, we present representative applications of this framework in the identification of novel pesticide leads targeting 4-hydroxyphenylpyruvate dioxygenase, acetohydroxyacid synthase, the cytochrome bc1 complex, succinate dehydrogenase, and histone deacetylase. Our successful discovery of novel herbicides or fungicides through fragment-based design for pesticide-like molecules may also stimulate greater interest in computer-driven agrochemical discovery.