Neural network fast-classifies biological images through features selecting to power automated microscopy.

Artificial intelligence is nowadays used for cell detection and classification in optical microscopy during post-acquisition analysis. The microscopes are now fully automated and next expected to be smart by making acquisition decisions based on the images. It calls for analysing them on the fly. Biology further imposes training on a reduced data set due to cost and time to prepare the samples and have the data sets annotated by experts. We propose a real-time image processing compliant with these specifications by balancing accurate detection and execution performance. We characterised the images using a generic, high-dimensional feature extractor. We then classified the images using machine learning to understand the contribution of each feature in decision and execution time. We found that the non-linear-classifier random forests outperformed Fisher's linear discriminant. More importantly, the most discriminant and time-consuming features could be excluded without significant accuracy loss, offering a substantial gain in execution time. It suggests a feature-group redundancy likely related to the biology of the observed cells. We offer a method to select fast and discriminant features. In our assay, a 79.6 ± 2.4% accurate classification of a cell took 68.7 ± 3.5 ms (mean ± SD, 5-fold cross-validation nested in 10 bootstrap repeats), corresponding to 14 cells per second, dispatched into eight phases of the cell cycle, using 12 feature groups and operating a consumer market ARM-based embedded system. A simple neural network offered similar performances paving the way to faster training and classification, using parallel execution on a general-purpose graphic processing unit. Finally, this strategy is also usable for deep neural networks paving the way to optimizing these algorithms for smart microscopy.

Automated screening of AURKA activity based on a genetically encoded FRET biosensor using fluorescence lifetime imaging microscopy.

Fluorescence Lifetime Imaging Microscopy (FLIM) is a robust tool to measure Förster Resonance Energy Transfer (FRET) between two fluorescent proteins, mainly when using genetically-encoded FRET biosensors. It is then possible to monitor biological processes such as kinase activity with a good spatiotemporal resolution and accuracy. Therefore, it is of interest to improve this methodology for future high content screening purposes. We here implement a time-gated FLIM microscope that can image and quantify fluorescence lifetime with a higher speed than conventional techniques such as Time-Correlated Single Photon Counting (TCSPC). We then improve our system to perform automatic screen analysis in a 96-well plate format. Moreover, we use a FRET biosensor of AURKA activity, a mitotic kinase involved in several epithelial cancers. Our results show that our system is suitable to measure FRET within our biosensor paving the way to the screening of novel compounds, potentially allowing to find new inhibitors of AURKA activity.

Optimized FRET Pairs and Quantification Approaches To Detect the Activation of Aurora Kinase A at Mitosis.

Genetically encoded Förster's Resonance Energy Transfer (FRET) biosensors are indispensable tools to sense the spatiotemporal dynamics of signal transduction pathways. Investigating the crosstalk between different signaling pathways is becoming increasingly important to follow cell development and fate programs. To this end, FRET biosensors must be optimized to monitor multiple biochemical activities simultaneously and in single cells. In addition, their sensitivity must be increased to follow their activation even when the abundance of the biosensor is low. We describe here the development of a second generation of Aurora kinase A/AURKA biosensors. First, we adapt the original AURKA biosensor-GFP-AURKA-mCherry-to multiplex FRET by using dark acceptors as ShadowG or ShadowY. Then, we use the novel superYFP acceptor protein to measure FRET by 2-color Fluorescence Cross-Correlation Spectroscopy, in cytosolic regions where the abundance of AURKA is extremely low and undetectable with the original AURKA biosensor. These results pave the way to the use of FRET biosensors to follow AURKA activation in conjunction with substrate-based activity biosensors. In addition, they open up the possibility of tracking the activation of small pools of AURKA and its interaction with novel substrates, which would otherwise remain undetectable with classical biochemical approaches.

A FRET biosensor reveals spatiotemporal activation and functions of aurora kinase A in living cells.

Overexpression of AURKA is a major hallmark of epithelial cancers. It encodes the multifunctional serine/threonine kinase aurora A, which is activated at metaphase and is required for cell cycle progression; assessing its activation in living cells is mandatory for next-generation drug design. We describe here a Förster's resonance energy transfer (FRET) biosensor detecting the conformational changes of aurora kinase A induced by its autophosphorylation on Thr288. The biosensor functionally replaces the endogenous kinase in cells and allows the activation of the kinase to be followed throughout the cell cycle. Inhibiting the catalytic activity of the kinase prevents the conformational changes of the biosensor. Using this approach, we discover that aurora kinase A activates during G1 to regulate the stability of microtubules in cooperation with TPX2 and CEP192. These results demonstrate that the aurora kinase A biosensor is a powerful tool to identify new regulatory pathways controlling aurora kinase A activation.