A MALDI-MS biotyping-like method to address honey bee health status through computational modelling

Abstract

AbstractAmong pollinator insects, bees undoubtedly account for the most important species. They play a critical role in boosting reproduction of wild and commercial plants and therefore contribute to the maintenance of plant biodiversity and sustainability of food webs. In the last few decades, domesticated and wild bees have been subjected to biotic and abiotic threats, alone or in combination, causing various health disorders. Therefore, monitoring solutions to improve bee health are increasingly necessary. MALDI mass spectrometry has emerged within this decade as a powerful technology to biotype micro-organisms. This method is currently and routinely used in clinical diagnosis where molecular mass fingerprints corresponding to major protein signatures are matched against databases for real-time identification. Based on this strategy, we developed MALDI BeeTyping as a proof of concept to monitor significant hemolymph molecular changes in honey bees upon infection with a series of entomopathogenic Gram-positive and -negative bacteria. ASerratia marcescensstrain isolated from one “naturally” infected honey bee collected from the field was also considered. We performed a series of individually recorded hemolymph molecular mass fingerprints and built, to our knowledge, the first computational model made of nine molecular signatures with a predictive score of 97.92%. Hence, we challenged our model by classifying a training set of individual bees’ hemolymph and obtained overall recognition of 91.93%. Through this work, we aimed at introducing a novel, realistic, and time-saving high-throughput biotyping-like strategy that addresses honey bee health in infectious conditions and on an individual scale through direct “blood tests”.Significance StatementDomesticated and wild bees worldwide represent the most active and valuable pollinators that ensure plant biodiversity and the success of many crops. These pollinators and others are exposed to deleterious pathogens and environmental stressors. Despite efforts to better understand how these threats affect honey bee health status, solutions are still crucially needed to help beekeepers, scientists and stakeholders in obtaining either a prognosis, an early diagnosis or a diagnosis of the health status of the apiaries. In this study, we describe a new method to investigate honey bee health by a simple “blood test” using fingerprints of some peptides/proteins as health status signatures. By computer modelling, we automated the identification of infected bees with a predictive score of 97.92%.