A Machine Learning Combined In Silico Approach for the Identification of Fungicide Combinations Targeting Plasmopara Viticola and Botrytis Cinerea Fungicide Resistance
Abstract
Downy mildew (caused by Plasmopara viticola) and Grey Mold (caused by Botrytis cinerea) are fungal diseases that significantly impact grape production globally. Cytochrome b plays a significant role in the mitochondrial respiratory chain of the two fungi that cause these diseases and is a key target for Quinone outside inhibitor (QoI) based fungicide development. QoI fungicides are common antifungal agents that are used to treat downy mildew or grey mold infections in fruits and vegetable crops by binding to cytochrome b and inhibiting respiratory function. Since the mode of action (MOA) of QoI fungicides is restricted to a single active site, the risk of developing resistance toward these fungicides is deemed high. Consequently, using a combination of fungicides in a rotational program is considered an effective way to reduce development of QoI resistance. In this study, a combination of in silico simulations that include Schrodinger Glide docking, molecular dynamics, MMGBSA and AutoQSAR modeling were used to screen the most potent QoI-based fungicide combinations to wild-type, G143A (Glycine to Alanine), F129L (Phenylalanine to Leucine) and double mutated versions that had both G143A and F129L mutations of fungal cytochrome b. The fungicides mandestrobin, fenaminstrobin and dimoxystrobin had high docking scores against multiple mutated versions of cytochrome b of Plasmopara viticola, which suggests their high affinity toward mutated variants of cytochrome b. Famoxadone, fenamidone, ametoctrodin and thiram also showed reasonable but relatively weaker binding affinity towards Plasmopara viticola cytochrome b. For the case of Botrytis cinerea, mandestrobin, pyribencarb and famoxadone showed strong binding affinity toward the four different variations of cytochrome b, which indicates that they are potential effective candidates against mutated cytochrome b. Four other fungicides, ametoctradin, fenamidone, metominostrobin and thiram were also effective against mutated cytochrome b. Based on both the docking simulations and QSAR/machine learning analysis ametoctradin emerged as a potential high-affinity QoI fungicide against the G143A mutation. The QoI-based fungicide combinations that include famoxadone, mandestrobin and ametoctradin preferentially are suggested to be considered in a fungicide management program in combination with fungicides that target other MOA as a potential treatment against Plasmopara viticola and Botrytis cinerea based fungal infections.
Subject
cytochrome bQoI fungicides
fungicide resistance
QSAR
machine learning
grapes
downy mildew
grey mold
Citation
Zhang, Junrui (2023). A Machine Learning Combined In Silico Approach for the Identification of Fungicide Combinations Targeting Plasmopara Viticola and Botrytis Cinerea Fungicide Resistance. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /198941.