Excitation-multiplexed multicolor superresolution imaging with fm-STORM and fm-DNA-PAINT

Abstract

Recent advancements in single-molecule-based superresolution microscopy have made it possible to visualize biological structures with unprecedented spatial resolution. Determining the spatial coorganization of these structures within cells under physiological and pathological conditions is an important biological goal. This goal has been stymied by the current limitations of carrying out superresolution microscopy in multiple colors. Here, we develop an approach for simultaneous multicolor superresolution imaging which relies solely on fluorophore excitation, rather than fluorescence emission properties. By modulating the intensity of the excitation lasers at different frequencies, we show that the color channel can be determined based on the fluorophore’s response to the modulated excitation. We use this frequency multiplexing to reduce the image acquisition time of multicolor superresolution DNA-PAINT while maintaining all its advantages: minimal color cross-talk, minimal photobleaching, maximal signal throughput, ability to maintain the fluorophore density per imaged color, and ability to use the full camera field of view. We refer to this imaging modality as “frequency multiplexed DNA-PAINT,” or fm-DNA-PAINT for short. We also show that frequency multiplexing is fully compatible with STORM superresolution imaging, which we term fm-STORM. Unlike fm-DNA-PAINT, fm-STORM is prone to color cross-talk. To overcome this caveat, we further develop a machine-learning algorithm to correct for color cross-talk with more than 95% accuracy, without the need for prior information about the imaged structure.

DeepFRET, a software for rapid and automated single-molecule FRET data classification using deep learning

“…Recent advances in machine learning (ML) and specifically deep learning (DL) ( LeCun et al, 2015 ), have radically improved our capacity to access and extract information from abstract and noisy inputs independently of human interventions as we ( ATLAS collaboration, 2014 ) and others have shown ( Berg et al, 2019 ; Christiansen et al, 2018 ; Falk et al, 2019 ; Gómez-García et al, 2018 ; Jones, 2019 ; Ouyang et al, 2018 ; Smith et al, 2019 ; Zhang et al, 2018 ). DL implementations are providing high-level robust performances and have been successfully used to analyze and augment a wide range of the fluorescence microscopy analysis pipeline including assessing microscope image quality ( Yang et al, 2018 ), in-silico cell labeling ( Christiansen et al, 2018 ), single-cell morphology analysis ( Berg et al, 2019 ; Falk et al, 2019 ), detecting single molecules ( White et al, 2020 ; Wu and Rifkin, 2015 ) and linking smFRET experiments with molecular dynamics simulations ( Matsunaga and Sugita, 2018 ), amongst others ( Berg et al, 2019 ; Christiansen et al, 2018 ; Falk et al, 2019 ; Gómez-García et al, 2018 ; Jones, 2019 ; Ouyang et al, 2018 ; Smith et al, 2019 ; Zhang et al, 2018 ).…”

“…Deep learning-based analysis has several advantages over other approaches: It recognizes abstract patterns and learns useful features directly from the raw input data, which allows the implementation of analysis routines that do not require extensive data preprocessing or empirically defined rules, and thus offer reproducible and less opinionated evaluation of single-molecule data; It is significantly faster than human annotation for large single-molecule datasets; it comes close to, or outperforms human performance; and its performance is increased when increasing dataset size constituting an ideal case for evaluating the large datasets obtained from single-molecule data ( Berg et al, 2019 ; Christiansen et al, 2018 ; Falk et al, 2019 ; Gómez-García et al, 2018 ; Jones, 2019 ; Ouyang et al, 2018 ; Smith et al, 2019 ; Zhang et al, 2018 ). Especially important are convolutional deep neural networks (DNN), artificial neural networks that learn to approximate the underlying function that transforms input to associated output through multiple rounds of optimization.…”

DeepFRET: Rapid and automated single molecule FRET data classification using deep learning

“…Recent advances in machine learning (ML) and specifically deep learning (DL) (27), have radically improved our capacity to access and extract information from abstract and noisy inputs independently of human interventions as we (28) and others have shown (29)(30)(31)(32)(33)(34)(35)(36). DL implementations are providing high level robust performances and have been successfully used to analyze and augment a wide range of the fluorescence microscopy analysis pipeline including assessing microscope image quality (37), in-silico cell labeling (31), single cell morphology analysis (32,34), detecting single molecules (38) and linking smFRET experiments with molecular dynamics simulations (39), amongst others (29)(30)(31)(32)(33)(34)(35)(36).…”

“…Deep learning-based analysis has several advantages over other approaches: It recognizes abstract patterns and learn useful features directly from the raw input data which allows implementation of analysis routines that don't require extensive data preprocessing or empirically defined rules and thus offer reproducible and less opinionated evaluation of single molecule data; It is significant faster than human annotation for large single molecule data sets; it comes close to, or outperforms human performance; and its performance is increased when increasing data set size constituting an ideal case for evaluating the large data sets obtained from single molecule data (29)(30)(31)(32)(33)(34)(35)(36). Especially important are convolutional DNN which learn how to best recognize particular aspects of the given data through several rounds of optimization.…”

Confocal Laser-Scanning Fluorescence-Lifetime Single-Molecule Localisation Microscopy

“…(15) A very fascinating approach is multi-colour SMLM that combines PSF engineering with deep learning for identifying and sorting different molecular species without the need of spectrally resolved imaging. (16) In frequency-based multiplexing STORM/DNA-PAINT, (17) one uses frequency-encoded multiplexed excitation and colour-blind detection to circumvent chromatic-aberration problems. Another clever solution is Exchange-PAINT, (18) which sequentially images different targets with the same dye but uses different DNA-tags for directing the dye to different targets decorated with complementary DNA-strands.…”