Deep‐learning‐based removal of autofluorescence and fluorescence quantification in plant‐colonizing bacteria in vivo

Abstract

Summary Fluorescence microscopy is common in bacteria–plant interaction studies. However, strong autofluorescence from plant tissues impedes in vivo studies on endophytes tagged with fluorescent proteins. To solve this problem, we developed a deep‐learning‐based approach to eliminate plant autofluorescence from fluorescence microscopy images, tested for the model endophyte Azoarcus olearius BH72 colonizing Oryza sativa roots. Micrographs from three channels (tdTomato for gene expression, green fluorescent protein (GFP) and AutoFluorescence (AF)) were processed by a neural network based approach, generating images that simulate the background autofluorescence in the tdTomato channel. After subtracting the model‐generated signals from each pixel in the genuine channel, the autofluorescence in the tdTomato channel was greatly reduced or even removed. The deep‐learning‐based approach can be applied for fluorescence detection and quantification, exemplified by a weakly expressed, a cell‐density modulated and a nitrogen‐fixation gene in A. olearius. A transcriptional nifH::tdTomato fusion demonstrated stronger induction of nif genes inside roots than outside, suggesting extension of the rhizosphere effect for diazotrophs into the endorhizosphere. The pre‐trained convolutional neural network model is easily applied to process other images of the same plant tissues with the same settings. This study showed the high potential of deep‐learning‐based approaches in image processing. With proper training data and strategies, autofluorescence in other tissues or materials can be removed for broad applications.

Gold‐FISH enables targeted NanoSIMS analysis of plant‐associated bacteria

“…This would allow not only to eradicate doubts whether or not these xylem‐associated diazotrophs are truly active in situ , but also to assess their activity across the community to obtain an improved estimation of their contribution to plant nutrition. Similar considerations apply to other studies in the field of root‐associated diazotrophs, ranging from communities associated with roots of maize (Van Deynze et al ., 2018 ), rice (Jiang et al ., 2022 ; Yan et al ., 2022 ), silvergrass (Li et al ., 2022 ), and sugarcane (Boddey et al ., 1995 ), to marine seagrasses (Martin et al ., 2020 ). In addition, investigating plant‐associated diazotrophy through the lens of correlative microscopy and chemical imaging has the potential to inform conceptual models such as mucilage‐assisted N 2 fixation associated with cereal crops (Bennett et al ., 2020 ).…”

“…For rice plants, previous inoculation experiments with isolated strains (e.g. Azoarcus olearius ) demonstrated the expression of nif genes in diazotrophs, which were presumably fixing N 2 (Egener & Hurek, 1999 ; Faoro et al ., 2017 ; Sarkar et al ., 2017 ; Jiang et al ., 2022 ). Their actual in situ activity, however, has not been confirmed yet, corroborating the ongoing debate on the correlation between plant growth promotion and N 2 fixation activity of individual bacterial strains (Cassán & Diaz‐Zorita, 2016 ; Chalk, 2016 ; Dixon & Hartmann, 2017 ).…”

Volumetric imaging of fast cellular dynamics with deep learning enhanced bioluminescence microscopy

“…Thus, the neuronal network learned to recognize these autofluorescent signal even in bioluminescent images, where they were absent. In future, dedicated models 63 will be needed to assume perfect performance, that involve an automated discriminator or ‘forgetting’ layers.…”

Cutting through host autofluorescence: fluorescence lifetime imaging microscopy for visualising intracellular bacteria in Symbiodiniaceae

“…Various sample preparation approaches such as exposure of cells to sodium borohydride (Castillo-Medina et al, 2011), high light (Webster et al, 2001;Maire et al, 2021), combined hydrogen peroxide and light (Yakubovskaya et al, 2019), copper sulfate and ethanol (Cohen et al, 2021), fluorescence lifetime gating (Kodama, 2016), TrueVIEW quenching kit (Astafyeva et al, 2022b) and post image processing using deep learning tools (Jiang et al, 2022) with active ingestion of bacterial prey. Although a high ICB signal was observed under dark conditions, ~60% of B. minutum cells were devoid of endosymbionts in this study.…”