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

Deep learning based local feature classification to automatically identify single molecule fluorescence events

Long-term single-molecule fluorescence measurements are widely used powerful tools to study the conformational dynamics of biomolecules in real time to further elucidate their conformational dynamics. Typically, thousands or even more single-molecule traces are analyzed to provide statistically meaningful information, which is labor-intensive and can introduce user bias. Recently, several deep-learning models have been developed to automatically classify single-molecule traces. In this study, we introduce DEBRIS (Deep lEarning Based fRagmentatIon approach for Single-molecule fluorescence event identification), a deep-learning model focusing on classifying local features and capable of automatically identifying steady fluorescence signals and dynamically emerging signals of different patterns. DEBRIS efficiently and accurately identifies both one-color and two-color single-molecule events, including their start and end points. By adjusting user-defined criteria, DEBRIS becomes the pioneer in using a deep learning model to accurately classify four different types of single-molecule fluorescence events using the same trained model, demonstrating its universality and ability to enrich the current toolbox.

Optical scattering methods for the label-free analysis of single biomolecules

Single-molecule techniques to analyze proteins and other biomolecules involving labels and tethers have allowed for new understanding of the underlying biophysics; however, the impact of perturbation from the labels and tethers has recently been shown to be significant in several cases. New approaches are emerging to measure single proteins through light scattering without the need for labels and ideally without tethers. Here, the approaches of interference scattering, plasmonic scattering, microcavity sensing, nanoaperture optical tweezing, and variants are described and compared. The application of these approaches to sizing, oligomerization, interactions, conformational dynamics, diffusion, and vibrational mode analysis is described. With early commercial successes, these approaches are poised to have an impact in the field of single-molecule biophysics.

Divergent molecular assembly and catalytic mechanisms between bacterial and archaeal RNase P in pre-tRNA cleavage

Ribonuclease P (RNase P) plays a vital role in the maturation of tRNA across bacteria, archaea, and eukaryotes. However, how RNase P assembles various components to achieve specific cleavage of precursor tRNA (pre-tRNA) in different organisms remains elusive. In this study, we employed single-molecule fluorescence resonance energy transfer to probe the dynamics of RNase P from E. coli (Escherichia coli) and Mja (Methanocaldococcus jannaschii) during pre-tRNA cleavage by incorporating five Cy3-Cy5 pairs into pre-tRNA and RNase P. Our results revealed significant differences in the assembly and catalytic mechanisms of RNase P between E. coli and Mja at both the RNA and protein levels. Specifically, the RNA of E. coli RNase P (EcoRPR) can adopt an active conformation that is capable of binding and cleaving pre-tRNA with high specificity independently. The addition of the protein component of E. coli RNase P (RnpA) enhances and accelerates pre-tRNA cleavage efficiency by increasing and stabilizing the active conformation. In contrast, Mja RPR is unable to form the catalytically active conformation on its own, and at least four proteins are required to induce the correct folding of Mja RPR. Mutation experiments suggest that the functional deficiency of Mja RPR arises from the absence of the second structural layer, and proper intermolecular assembly is essential for Mja RNase P to be functional over a broad temperature range. We propose models to illustrate the distinct catalytic patterns and RNA-protein interactions of RNase P in these two organisms.

Increasing the accuracy of single-molecule data analysis using tMAVEN

Time-dependent single-molecule experiments contain rich kinetic information about the functional dynamics of biomolecules. A key step in extracting this information is the application of kinetic models, such as hidden Markov models (HMMs), which characterize the molecular mechanism governing the experimental system. Unfortunately, researchers rarely know the physicochemical details of this molecular mechanism a priori, which raises questions about how to select the most appropriate kinetic model for a given single-molecule data set and what consequences arise if the wrong model is chosen. To address these questions, we have developed and used time-series modeling, analysis, and visualization environment (tMAVEN), a comprehensive, open-source, and extensible software platform. tMAVEN can perform each step of the single-molecule analysis pipeline, from preprocessing to kinetic modeling to plotting, and has been designed to enable the analysis of a single-molecule data set with multiple types of kinetic models. Using tMAVEN, we have systematically investigated mismatches between kinetic models and molecular mechanisms by analyzing simulated examples of prototypical single-molecule data sets exhibiting common experimental complications, such as molecular heterogeneity, with a series of different types of HMMs. Our results show that no single kinetic modeling strategy is mathematically appropriate for all experimental contexts. Indeed, HMMs only correctly capture the underlying molecular mechanism in the simplest of cases. As such, researchers must modify HMMs using physicochemical principles to avoid the risk of missing the significant biological and biophysical insights into molecular heterogeneity that their experiments provide. By enabling the facile, side-by-side application of multiple types of kinetic models to individual single-molecule data sets, tMAVEN allows researchers to carefully tailor their modeling approach to match the complexity of the underlying biomolecular dynamics and increase the accuracy of their single-molecule data analyses.

Deep-LASI, single-molecule data analysis software

By avoiding ensemble averaging, single-molecule methods provide novel means of extracting mechanistic insights into function of material and molecules at the nanoscale. However, one of the big limitations is the vast amount of data required for analyzing and extracting the desired information, which is time-consuming and user dependent. Here, we introduce Deep-LASI, a software suite for the manual and automatic analysis of single-molecule traces, interactions, and the underlying kinetics. The software can handle data from one-, two- and three-color fluorescence data, and was particularly designed for the analysis of two- and three-color single-molecule fluorescence resonance energy transfer experiments. The functionalities of the software include: the registration of multiple-channels, trace sorting and categorization, determination of the photobleaching steps, calculation of fluorescence resonance energy transfer correction factors, and kinetic analyses based on hidden Markov modeling or deep neural networks. After a kinetic analysis, the ensuing transition density plots are generated, which can be used for further quantification of the kinetic parameters of the system. Each step in the workflow can be performed manually or with the support of machine learning algorithms. Upon reading in the initial data set, it is also possible to perform the remaining analysis steps automatically without additional supervision. Hence, the time dedicated to the analysis of single-molecule experiments can be reduced from days/weeks to minutes. After a thorough description of the functionalities of the software, we also demonstrate the capabilities of the software via the analysis of a previously published dynamic three-color DNA origami structure fluctuating between three states. With the drastic time reduction in data analysis, new types of experiments become realistically possible that complement our currently available palette of methodologies for investigating the nanoworld.

Foundation model for efficient biological discovery in single-molecule data

Modern data-intensive techniques offer ever deeper insights into biology, but render the process of discovery increasingly complex. For example, exploiting the unique ability of single-molecule fluorescence microscopy (SMFM)1-5. to uncover rare but critical intermediates often demands manual inspection of time traces and iterative ad hoc approaches that are difficult to systematize. To facilitate systematic and efficient discovery from SMFM data, we introduce META-SiM, a transformer-based foundation model pre-trained on diverse SMFM analysis tasks. META-SiM achieves high performance-rivaling best-in-class algorithms-on a broad range of analysis tasks including trace selection, classification, segmentation, idealization, and stepwise photobleaching analysis. Additionally, the model produces high-dimensional embedding vectors that encapsulate detailed information about each trace, which the web-based META-SiM Projector (https://www.simol-projector.org) casts into lower-dimensional space for efficient whole-dataset visualization, labeling, comparison, and sharing. Combining this Projector with the objective metric of Local Shannon Entropy enables rapid identification of condition-specific behaviors, even if rare or subtle. As a result, by applying META-SiM to an existing single-molecule Förster resonance energy transfer (smFRET) dataset6, we discover a previously unobserved intermediate state in pre-mRNA splicing. META-SiM thus removes bottlenecks, improves objectivity, and both systematizes and accelerates biological discovery in complex single-molecule data.

Efficient, nonparametric removal of noise and recovery of probability distributions from time series using nonlinear-correlation functions: Photon and photon-counting noise

A preceding paper [M. Dhar, J. A. Dickinson, and M. A. Berg, J. Chem. Phys. 159, 054110 (2023)] shows how to remove additive noise from an experimental time series, allowing both the equilibrium distribution of the system and its Green's function to be recovered. The approach is based on nonlinear-correlation functions and is fully nonparametric: no initial model of the system or of the noise is needed. However, single-molecule spectroscopy often produces time series with either photon or photon-counting noise. Unlike additive noise, photon noise is signal-size correlated and quantized. Photon counting adds the potential for bias. This paper extends noise-corrected-correlation methods to these cases and tests them on synthetic datasets. Neither signal-size correlation nor quantization is a significant complication. Analysis of the sampling error yields guidelines for the data quality needed to recover the properties of a system with a given complexity. We show that bias in photon-counting data can be corrected, even at the high count rates needed to optimize the time resolution. Using all these results, we discuss the factors that limit the time resolution of single-molecule spectroscopy and the conditions that would be needed to push measurements into the submicrosecond region.

An optofluidic antenna for enhancing the sensitivity of single-emitter measurements

Many single-molecule investigations are performed in fluidic environments, for example, to avoid unwanted consequences of contact with surfaces. Diffusion of molecules in this arrangement limits the observation time and the number of collected photons, thus, compromising studies of processes with fast or slow dynamics. Here, we introduce a planar optofluidic antenna (OFA), which enhances the fluorescence signal from molecules by about 5 times per passage, leads to about 7-fold more frequent returns to the observation volume, and significantly lengthens the diffusion time within one passage. We use single-molecule multi-parameter fluorescence detection (sm-MFD), fluorescence correlation spectroscopy (FCS) and Förster resonance energy transfer (FRET) measurements to characterize our OFAs. The antenna advantages are showcased by examining both the slow (ms) and fast (50 μs) dynamics of DNA four-way (Holliday) junctions with real-time resolution. The FRET trajectories provide evidence for the absence of an intermediate conformational state and introduce an upper bound for its lifetime. The ease of implementation and compatibility with various microscopy modalities make OFAs broadly applicable to a diverse range of studies.

SEMORE: SEgmentation and MORphological fingErprinting by machine learning automates super-resolution data analysis

The morphology of protein assemblies impacts their behaviour and contributes to beneficial and aberrant cellular responses. While single-molecule localization microscopy provides the required spatial resolution to investigate these assemblies, the lack of universal robust analytical tools to extract and quantify underlying structures limits this powerful technique. Here we present SEMORE, a semi-automatic machine learning framework for universal, system- and input-dependent, analysis of super-resolution data. SEMORE implements a multi-layered density-based clustering module to dissect biological assemblies and a morphology fingerprinting module for quantification by multiple geometric and kinetics-based descriptors. We demonstrate SEMORE on simulations and diverse raw super-resolution data: time-resolved insulin aggregates, and published data of dSTORM imaging of nuclear pore complexes, fibroblast growth receptor 1, sptPALM of Syntaxin 1a and dynamic live-cell PALM of ryanodine receptors. SEMORE extracts and quantifies all protein assemblies, their temporal morphology evolution and provides quantitative insights, e.g. classification of heterogeneous insulin aggregation pathways and NPC geometry in minutes. SEMORE is a general analysis platform for super-resolution data, and being a time-aware framework can also support the rise of 4D super-resolution data.

Deep learning assisted single particle tracking for automated correlation between diffusion and function

Sub-cellular diffusion in living systems reflects cellular processes and interactions. Recent advances in optical microscopy allow the tracking of this nanoscale diffusion of individual objects with an unprecedented level of precision. However, the agnostic and automated extraction of functional information from the diffusion of molecules and organelles within the sub-cellular environment, is labor-intensive and poses a significant challenge. Here we introduce DeepSPT, a deep learning framework to interpret the diffusional 2D or 3D temporal behavior of objects in a rapid and efficient manner, agnostically. Demonstrating its versatility, we have applied DeepSPT to automated mapping of the early events of viral infections, identifying distinct types of endosomal organelles, and clathrin-coated pits and vesicles with up to 95% accuracy and within seconds instead of weeks. The fact that DeepSPT effectively extracts biological information from diffusion alone illustrates that besides structure, motion encodes function at the molecular and subcellular level.

Increasing the accuracy of single-molecule data analysis using tMAVEN

Time-dependent single-molecule experiments contain rich kinetic information about the functional dynamics of biomolecules. A key step in extracting this information is the application of kinetic models, such as hidden Markov models (HMMs), which characterize the molecular mechanism governing the experimental system. Unfortunately, researchers rarely know the physico-chemical details of this molecular mechanism a priori, which raises questions about how to select the most appropriate kinetic model for a given single-molecule dataset and what consequences arise if the wrong model is chosen. To address these questions, we have developed and used time-series Modeling, Analysis, and Visualization ENvironment (tMAVEN), a comprehensive, open-source, and extensible software platform. tMAVEN can perform each step of the single-molecule analysis pipeline, from pre-processing to kinetic modeling to plotting, and has been designed to enable the analysis of a single-molecule dataset with multiple types of kinetic models. Using tMAVEN, we have systematically investigated mismatches between kinetic models and molecular mechanisms by analyzing simulated examples of prototypical single-molecule datasets exhibiting common experimental complications, such as molecular heterogeneity, with a series of different types of HMMs. Our results show that no single kinetic modeling strategy is mathematically appropriate for all experimental contexts. Indeed, HMMs only correctly capture the underlying molecular mechanism in the simplest of cases. As such, researchers must modify HMMs using physico-chemical principles to avoid the risk of missing the significant biological and biophysical insights into molecular heterogeneity that their experiments provide. By enabling the facile, side-by-side application of multiple types of kinetic models to individual single-molecule datasets, tMAVEN allows researchers to carefully tailor their modeling approach to match the complexity of the underlying biomolecular dynamics and increase the accuracy of their single-molecule data analyses.

Deep learning assisted single particle tracking for automated correlation between diffusion and function

Sub-cellular diffusion in living systems reflects cellular processes and interactions. Recent advances in optical microscopy allow the tracking of this nanoscale diffusion of individual objects with an unprecedented level of precision. However, the agnostic and automated extraction of functional information from the diffusion of molecules and organelles within the sub-cellular environment, is labor-intensive and poses a significant challenge. Here we introduce DeepSPT, a deep learning framework to interpret the diffusional 2D or 3D temporal behavior of objects in a rapid and efficient manner, agnostically. Demonstrating its versatility, we have applied DeepSPT to automated mapping of the early events of viral infections, identifying distinct types of endosomal organelles, and clathrin-coated pits and vesicles with up to 95% accuracy and within seconds instead of weeks. The fact that DeepSPT effectively extracts biological information from diffusion alone indicates that besides structure, motion encodes function at the molecular and subcellular level.

Deep-LASI: deep-learning assisted, single-molecule imaging analysis of multi-color DNA origami structures

Single-molecule experiments have changed the way we explore the physical world, yet data analysis remains time-consuming and prone to human bias. Here, we introduce Deep-LASI (Deep-Learning Assisted Single-molecule Imaging analysis), a software suite powered by deep neural networks to rapidly analyze single-, two- and three-color single-molecule data, especially from single-molecule Förster Resonance Energy Transfer (smFRET) experiments. Deep-LASI automatically sorts recorded traces, determines FRET correction factors and classifies the state transitions of dynamic traces all in ~20-100 ms per trajectory. We benchmarked Deep-LASI using ground truth simulations as well as experimental data analyzed manually by an expert user and compared the results with a conventional Hidden Markov Model analysis. We illustrate the capabilities of the technique using a highly tunable L-shaped DNA origami structure and use Deep-LASI to perform titrations, analyze protein conformational dynamics and demonstrate its versatility for analyzing both total internal reflection fluorescence microscopy and confocal smFRET data.

Methods for automating the analysis of live-cell single-molecule FRET data

Single-molecule FRET (smFRET) is a powerful imaging platform capable of revealing dynamic changes in the conformation and proximity of biological molecules. The expansion of smFRET imaging into living cells creates both numerous new research opportunities and new challenges. Automating dataset curation processes is critical to providing consistent, repeatable analysis in an efficient manner, freeing experimentalists to advance the technical boundaries and throughput of what is possible in imaging living cells. Here, we devise an automated solution to the problem of multiple particles entering a region of interest, an otherwise labor-intensive and subjective process that had been performed manually in our previous work. The resolution of these two issues increases the quantity of FRET data and improves the accuracy with which FRET distributions are generated, increasing knowledge about the biological functions of the molecules under study. Our automated approach is straightforward, interpretable, and requires only localization and intensity values for donor and acceptor channel signals, which we compute through our previously published smCellFRET pipeline. The development of our automated approach is informed by the insights of expert experimentalists with extensive experience inspecting smFRET trajectories (displacement and intensity traces) from live cells. We test our automated approach against our recently published research on the metabotropic glutamate receptor 2 (mGluR2) and reveal substantial similarities, as well as potential shortcomings in the manual curation process that are addressable using the algorithms we developed here.

From Deep Mutational Mapping of Allosteric Protein Landscapes to Deep Learning of Allostery and Hidden Allosteric Sites: Zooming in on "Allosteric Intersection" of Biochemical and Big Data Approaches

The recent advances in artificial intelligence (AI) and machine learning have driven the design of new expert systems and automated workflows that are able to model complex chemical and biological phenomena. In recent years, machine learning approaches have been developed and actively deployed to facilitate computational and experimental studies of protein dynamics and allosteric mechanisms. In this review, we discuss in detail new developments along two major directions of allosteric research through the lens of data-intensive biochemical approaches and AI-based computational methods. Despite considerable progress in applications of AI methods for protein structure and dynamics studies, the intersection between allosteric regulation, the emerging structural biology technologies and AI approaches remains largely unexplored, calling for the development of AI-augmented integrative structural biology. In this review, we focus on the latest remarkable progress in deep high-throughput mining and comprehensive mapping of allosteric protein landscapes and allosteric regulatory mechanisms as well as on the new developments in AI methods for prediction and characterization of allosteric binding sites on the proteome level. We also discuss new AI-augmented structural biology approaches that expand our knowledge of the universe of protein dynamics and allostery. We conclude with an outlook and highlight the importance of developing an open science infrastructure for machine learning studies of allosteric regulation and validation of computational approaches using integrative studies of allosteric mechanisms. The development of community-accessible tools that uniquely leverage the existing experimental and simulation knowledgebase to enable interrogation of the allosteric functions can provide a much-needed boost to further innovation and integration of experimental and computational technologies empowered by booming AI field.

Observation of structural switch in nascent SAM-VI riboswitch during transcription at single-nucleotide and single-molecule resolution

Growing RNAs fold differently as they are transcribed, which modulates their finally adopted structures. Riboswitches regulate gene expression by structural change, which are sensitive to co-transcriptionally structural biology. Here we develop a strategy to track the structural change of RNAs during transcription at single-nucleotide and single-molecule resolution and use it to monitor individual transcripts of the SAM-VI riboswitch (riboSAM) as transcription proceeds, observing co-existence of five states in riboSAM. We report a bifurcated helix in one newly identified state from NMR and single-molecule FRET (smFRET) results, and its presence directs the translation inhibition in our cellular translation experiments. A model is proposed to illustrate the distinct switch patterns and gene-regulatory outcome of riboSAM when SAM is present or absent. Our strategy enables the precise mapping of RNAs' conformational landscape during transcription, and may combine with detection methods other than smFRET for structural studies of RNAs in general.

Machine learning and protein allostery

The fundamental biological importance and complexity of allosterically regulated proteins stem from their central role in signal transduction and cellular processes. Recently, machine-learning approaches have been developed and actively deployed to facilitate theoretical and experimental studies of protein dynamics and allosteric mechanisms. In this review, we survey recent developments in applications of machine-learning methods for studies of allosteric mechanisms, prediction of allosteric effects and allostery-related physicochemical properties, and allosteric protein engineering. We also review the applications of machine-learning strategies for characterization of allosteric mechanisms and drug design targeting SARS-CoV-2. Continuous development and task-specific adaptation of machine-learning methods for protein allosteric mechanisms will have an increasingly important role in bridging a wide spectrum of data-intensive experimental and theoretical technologies.

Reliability and accuracy of single-molecule FRET studies for characterization of structural dynamics and distances in proteins

Single-molecule Förster-resonance energy transfer (smFRET) experiments allow the study of biomolecular structure and dynamics in vitro and in vivo. We performed an international blind study involving 19 laboratories to assess the uncertainty of FRET experiments for proteins with respect to the measured FRET efficiency histograms, determination of distances, and the detection and quantification of structural dynamics. Using two protein systems with distinct conformational changes and dynamics, we obtained an uncertainty of the FRET efficiency ≤0.06, corresponding to an interdye distance precision of ≤2 Å and accuracy of ≤5 Å. We further discuss the limits for detecting fluctuations in this distance range and how to identify dye perturbations. Our work demonstrates the ability of smFRET experiments to simultaneously measure distances and avoid the averaging of conformational dynamics for realistic protein systems, highlighting its importance in the expanding toolbox of integrative structural biology.

Single-photon smFRET. I: Theory and conceptual basis

We present a unified conceptual framework and the associated software package for single-molecule Förster resonance energy transfer (smFRET) analysis from single-photon arrivals leveraging Bayesian nonparametrics, BNP-FRET. This unified framework addresses the following key physical complexities of a single-photon smFRET experiment, including: 1) fluorophore photophysics; 2) continuous time kinetics of the labeled system with large timescale separations between photophysical phenomena such as excited photophysical state lifetimes and events such as transition between system states; 3) unavoidable detector artefacts; 4) background emissions; 5) unknown number of system states; and 6) both continuous and pulsed illumination. These physical features necessarily demand a novel framework that extends beyond existing tools. In particular, the theory naturally brings us to a hidden Markov model with a second-order structure and Bayesian nonparametrics on account of items 1, 2, and 5 on the list. In the second and third companion articles, we discuss the direct effects of these key complexities on the inference of parameters for continuous and pulsed illumination, respectively.

Enhanced hexamerization of insulin via assembly pathway rerouting revealed by single particle studies

Insulin formulations with diverse oligomerization states are the hallmark of interventions for the treatment of diabetes. Here using single-molecule recordings we firstly reveal that insulin oligomerization can operate via monomeric additions and secondly quantify the existence, abundance and kinetic characterization of diverse insulin assembly and disassembly pathways involving addition of monomeric, dimeric or tetrameric insulin species. We propose and experimentally validate a model where the insulin self-assembly pathway is rerouted, favoring monomeric or oligomeric assembly, by solution concentration, additives and formulations. Combining our practically complete kinetic characterization with rate simulations, we calculate the abundance of each oligomeric species from nM to mM offering mechanistic insights and the relative abundance of all oligomeric forms at concentrations relevant both for secreted and administrated insulin. These reveal a high abundance of all oligomers and a significant fraction of hexamer resulting in practically halved bioavailable monomer concentration. In addition to providing fundamental new insights, the results and toolbox presented here can be universally applied, contributing to the development of optimal insulin formulations and the deciphering of oligomerization mechanisms for additional proteins.

Heterogeneous and Surface-Catalyzed Amyloid Aggregation Monitored by Spatially Resolved Fluorescence and Single Molecule Microscopy

Amyloid aggregation is associated with many diseases and may also occur in therapeutic protein formulations. Addition of co-solutes is a key strategy to modulate the stability of proteins in pharmaceutical formulations and select inhibitors for drug design in the context of diseases. However, the heterogeneous nature of this multicomponent system in terms of structures and mechanisms poses a number of challenges for the analysis of the chemical reaction. Using insulin as protein system and polysorbate 80 as co-solute, we combine a spatially resolved fluorescence approach with single molecule microscopy and machine learning methods to kinetically disentangle the different contributions from multiple species within a single aggregation experiment. We link the presence of interfaces to the degree of heterogeneity of the aggregation kinetics and retrieve the rate constants and underlying mechanisms for single aggregation events. Importantly, we report that the mechanism of inhibition of the self-assembly process depends on the details of the growth pathways of otherwise macroscopically identical species. This information can only be accessed by the analysis of single aggregate events, suggesting our method as a general tool for a comprehensive physicochemical characterization of self-assembly reactions.

Physical and barrier changes in gastrointestinal mucus induced by the permeation enhancer sodium 8-[(2-hydroxybenzoyl)amino]octanoate (SNAC)

Drug delivery systems (DDS) for oral delivery of peptide drugs contain excipients that facilitate and enhance absorption. However, little knowledge exists on how DDS excipients such as permeation enhancers interact with the gastrointestinal mucus barrier. This study aimed to investigate interactions of the permeation enhancer sodium 8-[(2-hydroxybenzoyl)amino]octanoate (SNAC) with ex vivo porcine intestinal mucus (PIM), ex vivo porcine gastric mucus (PGM), as well as with in vitro biosimilar mucus (BM) by profiling their physical and barrier properties upon exposure to SNAC. Bulk mucus permeability studies using the peptides cyclosporine A and vancomycin, ovalbumin as a model protein, as well as fluorescein-isothiocyanate dextrans (FDs) of different molecular weights and different surface charges were conducted in parallel to mucus retention force studies using a texture analyzer, rheological studies, cryo-scanning electron microscopy (cryo-SEM), and single particle tracking of fluorescence-labelled nanoparticles to investigate the effects of the SNAC-mucus interaction. The exposure of SNAC to PIM increased the mucus retention force, storage modulus, viscosity, increased nanoparticle confinement within PIM as well as decreased the permeation of cyclosporine A and ovalbumin through PIM. Surprisingly, the viscosity of PGM and the permeation of cyclosporine A and ovalbumin through PGM was unaffected by the presence of SNAC, thus the effect of SNAC depended on the regional site that mucus was collected from. In the absence of SNAC, the permeation of different molecular weight and differently charged FDs through PIM was comparable to that through BM. However, while bulk permeation of neither of the FDs through PIM was affected by SNAC, the presence of SNAC decreased the permeation of FD4 and increased the permeation of FD150 kDa through BM. Additionally, and in contrast to observations in PIM, nanoparticle confinement within BM remained unaffected by the presence of SNAC. In conclusion, the present study showed that SNAC altered the physical and barrier properties of PIM, but not of PGM. The effects of SNAC in PIM were not observed in the BM in vitro model. Altogether, the study highlights the need for further understanding how permeation enhancers influence the mucus barrier and illustrates that the selected mucus model for such studies should be chosen with care.

Deep learning in single-molecule imaging and analysis: recent advances and prospects

Single-molecule microscopy is advantageous in characterizing heterogeneous dynamics at the molecular level. However, there are several challenges that currently hinder the wide application of single molecule imaging in bio-chemical studies, including how to perform single-molecule measurements efficiently with minimal run-to-run variations, how to analyze weak single-molecule signals efficiently and accurately without the influence of human bias, and how to extract complete information about dynamics of interest from single-molecule data. As a new class of computer algorithms that simulate the human brain to extract data features, deep learning networks excel in task parallelism and model generalization, and are well-suited for handling nonlinear functions and extracting weak features, which provide a promising approach for single-molecule experiment automation and data processing. In this perspective, we will highlight recent advances in the application of deep learning to single-molecule studies, discuss how deep learning has been used to address the challenges in the field as well as the pitfalls of existing applications, and outline the directions for future development.

A blind benchmark of analysis tools to infer kinetic rate constants from single-molecule FRET trajectories

Single-molecule FRET (smFRET) is a versatile technique to study the dynamics and function of biomolecules since it makes nanoscale movements detectable as fluorescence signals. The powerful ability to infer quantitative kinetic information from smFRET data is, however, complicated by experimental limitations. Diverse analysis tools have been developed to overcome these hurdles but a systematic comparison is lacking. Here, we report the results of a blind benchmark study assessing eleven analysis tools used to infer kinetic rate constants from smFRET trajectories. We test them against simulated and experimental data containing the most prominent difficulties encountered in analyzing smFRET experiments: different noise levels, varied model complexity, non-equilibrium dynamics, and kinetic heterogeneity. Our results highlight the current strengths and limitations in inferring kinetic information from smFRET trajectories. In addition, we formulate concrete recommendations and identify key targets for future developments, aimed to advance our understanding of biomolecular dynamics through quantitative experiment-derived models.

A blind benchmark of analysis tools to infer kinetic rate constants from single-molecule FRET trajectories

Single-molecule FRET (smFRET) is a versatile technique to study the dynamics and function of biomolecules since it makes nanoscale movements detectable as fluorescence signals. The powerful ability to infer quantitative kinetic information from smFRET data is, however, complicated by experimental limitations. Diverse analysis tools have been developed to overcome these hurdles but a systematic comparison is lacking. Here, we report the results of a blind benchmark study assessing eleven analysis tools used to infer kinetic rate constants from smFRET trajectories. We test them against simulated and experimental data containing the most prominent difficulties encountered in analyzing smFRET experiments: different noise levels, varied model complexity, non-equilibrium dynamics, and kinetic heterogeneity. Our results highlight the current strengths and limitations in inferring kinetic information from smFRET trajectories. In addition, we formulate concrete recommendations and identify key targets for future developments, aimed to advance our understanding of biomolecular dynamics through quantitative experiment-derived models.

Mars, a molecule archive suite for reproducible analysis and reporting of single-molecule properties from bioimages

The rapid development of new imaging approaches is generating larger and more complex datasets, revealing the time evolution of individual cells and biomolecules. Single-molecule techniques, in particular, provide access to rare intermediates in complex, multistage molecular pathways. However, few standards exist for processing these information-rich datasets, posing challenges for wider dissemination. Here, we present Mars, an open-source platform for storing and processing image-derived properties of biomolecules. Mars provides Fiji/ImageJ2 commands written in Java for common single-molecule analysis tasks using a Molecule Archive architecture that is easily adapted to complex, multistep analysis workflows. Three diverse workflows involving molecule tracking, multichannel fluorescence imaging, and force spectroscopy, demonstrate the range of analysis applications. A comprehensive graphical user interface written in JavaFX enhances biomolecule feature exploration by providing charting, tagging, region highlighting, scriptable dashboards, and interactive image views. The interoperability of ImageJ2 ensures Molecule Archives can easily be opened in multiple environments, including those written in Python using PyImageJ, for interactive scripting and visualization. Mars provides a flexible solution for reproducible analysis of image-derived properties, facilitating the discovery and quantitative classification of new biological phenomena with an open data format accessible to everyone.

Heterogeneous migration routes of DNA triplet repeat slip-outs

It is unclear how the length of a repetitive DNA tract determines the onset and progression of repeat expansion diseases, but the dynamics of secondary DNA structures formed by repeat sequences are believed to play an important role. It was recently shown that three-way DNA junctions containing slip-out hairpins of CAG or CTG repeats and contiguous triplet repeats in the adjacent duplex displayed single-molecule FRET (smFRET) dynamics that were ascribed to both local conformational motions and longer-range branch migration. Here we explore these so-called "mobile" slip-out structures through a detailed kinetic analysis of smFRET trajectories and coarse-grained modeling. Despite the apparent structural simplicity, with six FRET states resolvable, most smFRET states displayed biexponential dwell-time distributions, attributed to structural heterogeneity and overlapping FRET states. Coarse-grained modeling for a (GAC)₁₀ repeat slip-out included trajectories that corresponded to a complete round of branch migration; the structured free energy landscape between slippage events supports the dynamical complexity observed by smFRET. A hairpin slip-out with 40 CAG repeats, which is above the repeat length required for disease in several triplet repeat disorders, displayed smFRET dwell times that were on average double those of 3WJs with 10 repeats. The rate of secondary-structure rearrangement via branch migration, relative to particular DNA processing pathways, may be an important factor in the expansion of triplet repeat expansion diseases.

Toward high-throughput oligomer detection and classification for early-stage aggregation of amyloidogenic protein

Aggregation kinetics of proteins and peptides have been studied extensively due to their significance in many human diseases, including neurodegenerative disorders, and the roles they play in some key physiological processes. However, most of these studies have been performed as bulk measurements using Thioflavin T or other fluorescence turn-on reagents as indicators of fibrillization. Such techniques are highly successful in making inferences about the nucleation and growth mechanism of fibrils, yet cannot directly measure assembly reactions at low protein concentrations which is the case for amyloid-β (Aβ) peptide under physiological conditions. In particular, the evolution from monomer to low-order oligomer in early stages of aggregation cannot be detected. Single-molecule methods allow direct access to such fundamental information. We developed a high-throughput protocol for single-molecule photobleaching experiments using an automated fluorescence microscope. Stepwise photobleaching analysis of the time profiles of individual foci allowed us to determine stoichiometry of protein oligomers and probe protein aggregation kinetics. Furthermore, we investigated the potential application of supervised machine learning with support vector machines (SVMs) as well as multilayer perceptron (MLP) artificial neural networks to classify bleaching traces into stoichiometric categories based on an ensemble of measurable quantities derivable from individual traces. Both SVM and MLP models achieved a comparable accuracy of more than 80% against simulated traces up to 19-mer, although MLP offered considerable speed advantages, thus making it suitable for application to high-throughput experimental data. We used our high-throughput method to study the aggregation of Aβ40 in the presence of metal ions and the aggregation of α-synuclein in the presence of gold nanoparticles.

Direct observation of heterogeneous formation of amyloid spherulites in real-time by super-resolution microscopy

Protein misfolding in the form of fibrils or spherulites is involved in a spectrum of pathological abnormalities. Our current understanding of protein aggregation mechanisms has primarily relied on the use of spectrometric methods to determine the average growth rates and diffraction-limited microscopes with low temporal resolution to observe the large-scale morphologies of intermediates. We developed a REal-time kinetics via binding and Photobleaching LOcalization Microscopy (REPLOM) super-resolution method to directly observe and quantify the existence and abundance of diverse aggregate morphologies of human insulin, below the diffraction limit and extract their heterogeneous growth kinetics. Our results revealed that even the growth of microscopically identical aggregates, e.g., amyloid spherulites, may follow distinct pathways. Specifically, spherulites do not exclusively grow isotropically but, surprisingly, may also grow anisotropically, following similar pathways as reported for minerals and polymers. Combining our technique with machine learning approaches, we associated growth rates to specific morphological transitions and provided energy barriers and the energy landscape at the level of single aggregate morphology. Our unifying framework for the detection and analysis of spherulite growth can be extended to other self-assembled systems characterized by a high degree of heterogeneity, disentangling the broad spectrum of diverse morphologies at the single-molecule level.

Real-time super-resolution microscopy analysis reveals the growth kinetics, morphology, and abundance of human insulin amyloid spherulites with different growth pathways.

Enhancing Inference on Physiological and Kinematic Periodic Signals via Phase-Based Interpretability and Multi-Task Learning

Physiological and kinematic signals from humans are often used for monitoring health. Several processes of interest (e.g., cardiac and respiratory processes, and locomotion) demonstrate periodicity. Training models for inference on these signals (e.g., detection of anomalies, and extraction of biomarkers) require large amounts of data to capture their variability, which are not readily available. This hinders the performance of complex inference models. In this work, we introduce a methodology for improving inference on such signals by incorporating phase-based interpretability and other inference tasks into a multi-task framework applied to a generative model. For this purpose, we utilize phase information as a regularization term and as an input to the model and introduce an interpretable unit in a neural network, which imposes an interpretable structure on the model. This imposition helps us in the smooth generation of periodic signals that can aid in data augmentation tasks. We demonstrate the impact of our framework on improving the overall inference performance on ECG signals and inertial signals from gait locomotion.

Determination of rate constants for conformational changes of RNA helicases by single-molecule FRET TIRF microscopy

RNA helicases couple nucleotide-driven conformational changes to the unwinding of RNA duplexes. Interaction partners can regulate helicase activity by altering the rate constants of these conformational changes. Single-molecule FRET experiments on donor/acceptor-labeled, immobilized molecules are ideally suited to monitor conformational changes in real time and to extract rate constants for these processes. This article provides guidance on how to design, perform, and analyze single-molecule FRET experiments by TIRF microscopy. It covers the theoretical background of FRET and single-molecule TIRF microscopy, the considerations to prepare proteins of interest for donor/acceptor labeling and surface immobilization, and the principles and procedures of data analysis, including image analysis and the determination of FRET time traces, the extraction of rate constants from FRET time traces, and the general conclusions that can be drawn from these data. A case study, using the DEAD-box protein eIF4A as an example, highlights how single-molecule FRET studies have been instrumental in understanding the role of conformational changes for duplex unwinding and for the regulation of helicase activities. Selected examples illustrate which conclusions can be drawn from the kinetic data obtained, highlight possible pitfalls in data analysis and interpretation, and outline how kinetic models can be related to functionally relevant states.

Real-Time Single-Molecule Studies of RNA Polymerase-Promoter Open Complex Formation Reveal Substantial Heterogeneity Along the Promoter-Opening Pathway

The expression of most bacterial genes commences with the binding of RNA polymerase (RNAP)-σ70 holoenzyme to the promoter DNA. This initial RNAP-promoter closed complex undergoes a series of conformational changes, including the formation of a transcription bubble on the promoter and the loading of template DNA strand into the RNAP active site; these changes lead to the catalytically active open complex (RPO) state. Recent cryo-electron microscopy studies have provided detailed structural insight on the RPO and putative intermediates on its formation pathway. Here, we employ single-molecule fluorescence microscopy to interrogate the conformational dynamics and reaction kinetics during real-time RPO formation on a consensus lac promoter. We find that the promoter opening may proceed rapidly from the closed to open conformation in a single apparent step, or may instead involve a significant intermediate between these states. The formed RPO complexes are also different with respect to their transcription bubble stability. The RNAP cleft loops, and especially the β' rudder, stabilise the transcription bubble. The RNAP interactions with the promoter upstream sequence (beyond -35) stimulate transcription bubble nucleation and tune the reaction path towards stable forms of the RPO.

Joint Detection of Change Points in Multichannel Single-Molecule Measurements

Recent developments in single-molecule measurement technology have expanded the capability to measure multiple parameters. These emergent modalities provide more holistic observations of complex biomolecular processes and call for new analysis methods to detect state changes in multichannel data. Here we develop an algorithm called MULLR (MUlti-channel Log-Likelihood Ratio test) to jointly identify change points in multichannel single-molecule measurements. MULLR is an extension of the popular single-channel implementation for change point detection based on a binary segmentation and log-likelihood ratio test framework. We validate the algorithm on simulated data and characterize the power of detection and false positive rate. We show that MULLR can identify change points in experimental multichannel data and naturally works with different noise statistics and time resolutions across channels. Further, we quantify the benefit of MULLR compared to single-channel analysis. We envision that the MULLR algorithm will be useful to a range of multiparameter single-molecule measurements.

Single-particle diffusional fingerprinting: A machine-learning framework for quantitative analysis of heterogeneous diffusion

Single-particle tracking (SPT) analysis of individual biomolecules is an indispensable tool for extracting quantitative information from dynamic biological processes, but often requires some a priori knowledge of the system. Here we present “single-particle diffusional fingerprinting,” a more general approach for extraction of diffusional patterns in SPT independently of the biological system. This method extracts a set of descriptive features for each SPT trajectory, which are ranked upon classification to yield mechanistic insights for the species under comparison. We demonstrate its capacity to yield a dictionary of diffusional traits across multiple systems (e.g., lipases hydrolyzing fat, transcription factors diffusing in cells, and nanoparticles in mucus), supporting its use on multiple biological phenomena (e.g., drug delivery, receptor dynamics, and virology).

Single-particle tracking (SPT) is a key tool for quantitative analysis of dynamic biological processes and has provided unprecedented insights into a wide range of systems such as receptor localization, enzyme propulsion, bacteria motility, and drug nanocarrier delivery. The inherently complex diffusion in such biological systems can vary drastically both in time and across systems, consequently imposing considerable analytical challenges, and currently requires an a priori knowledge of the system. Here we introduce a method for SPT data analysis, processing, and classification, which we term “diffusional fingerprinting.” This method allows for dissecting the features that underlie diffusional behavior and establishing molecular identity, regardless of the underlying diffusion type. The method operates by isolating 17 descriptive features for each observed motion trajectory and generating a diffusional map of all features for each type of particle. Precise classification of the diffusing particle identity is then obtained by training a simple logistic regression model. A linear discriminant analysis generates a feature ranking that outputs the main differences among diffusional features, providing key mechanistic insights. Fingerprinting operates by both training on and predicting experimental data, without the need for pretraining on simulated data. We found this approach to work across a wide range of simulated and experimentally diverse systems, such as tracked lipases on fat substrates, transcription factors diffusing in cells, and nanoparticles diffusing in mucus. This flexibility ultimately supports diffusional fingerprinting’s utility as a universal paradigm for SPT diffusional analysis and prediction.

FRETboard: Semisupervised classification of FRET traces

Förster resonance energy transfer (FRET) is a useful phenomenon in biomolecular investigations, as it can be leveraged for nanoscale measurements. The optical signals produced by such experiments can be analyzed by fitting a statistical model. Several software tools exist to fit such models in an unsupervised manner but lack the flexibility to adapt to different experimental setups and require local installations. Here, we propose to fit models to optical signals more intuitively by adopting a semisupervised approach, in which the user interactively guides the model to fit a given data set, and introduce FRETboard, a web tool that allows users to provide such guidance. We show that our approach is able to closely reproduce ground truth FRET statistics in a wide range of simulated single-molecule scenarios and correctly estimate parameters for up to 11 states. On in vitro data, we retrieve parameters identical to those obtained by laborious manual classification in a fraction of the required time. Moreover, we designed FRETboard to be easily extendable to other models, allowing it to adapt to future developments in FRET measurement and analysis.

Quantifying and visualising the nuances of cellular dynamics in vivo using intravital imaging

Intravital imaging is a powerful technology used to quantify and track dynamic changes in live cells and tissues within an intact environment. The ability to watch cell biology in real-time 'as it happens' has provided novel insight into tissue homeostasis, as well as disease initiation, progression and response to treatment. In this minireview, we highlight recent advances in the field of intravital microscopy, touching upon advances in awake versus anaesthesia-based approaches, as well as the integration of biosensors into intravital imaging. We also discuss current challenges that, in our opinion, need to be overcome to further advance the field of intravital imaging at the single-cell, subcellular and molecular resolution to reveal nuances of cell behaviour that can be targeted in complex disease settings.

AutoStepfinder: A fast and automated step detection method for single-molecule analysis

Single-molecule techniques allow the visualization of the molecular dynamics of nucleic acids and proteins with high spatiotemporal resolution. Valuable kinetic information of biomolecules can be obtained when the discrete states within single-molecule time trajectories are determined. Here, we present a fast, automated, and bias-free step detection method, AutoStepfinder, that determines steps in large datasets without requiring prior knowledge on the noise contributions and location of steps. The analysis is based on a series of partition events that minimize the difference between the data and the fit. A dual-pass strategy determines the optimal fit and allows AutoStepfinder to detect steps of a wide variety of sizes. We demonstrate step detection for a broad variety of experimental traces. The user-friendly interface and the automated detection of AutoStepfinder provides a robust analysis procedure that enables anyone without programming knowledge to generate step fits and informative plots in less than an hour.

•

Fast, automated, and bias-free detection of steps within single-molecule trajectories

•

Robust step detection without any prior knowledge on the data

•

A dual-pass strategy for the detection of steps over a wide variety of scales

•

A user-friendly interface for a simplified step fitting procedure

Single-molecule techniques have made it possible to track individual protein complexes in real time with a nanometer spatial resolution and a millisecond timescale. Accurate determination of the dynamic states within single-molecule time traces provides valuable kinetic information that underlie the function of biological macromolecules. Here, we present a new automated step detection method called AutoStepfinder, a versatile, robust, and easy-to-use algorithm that allows researchers to determine the kinetic states within single-molecule time trajectories without any bias.

Biased cytochrome P450-mediated metabolism via small-molecule ligands binding P450 oxidoreductase

Metabolic control is mediated by the dynamic assemblies and function of multiple redox enzymes. A key element in these assemblies, the P450 oxidoreductase (POR), donates electrons and selectively activates numerous (>50 in humans and >300 in plants) cytochromes P450 (CYPs) controlling metabolism of drugs, steroids and xenobiotics in humans and natural product biosynthesis in plants. The mechanisms underlying POR-mediated CYP metabolism remain poorly understood and to date no ligand binding has been described to regulate the specificity of POR. Here, using a combination of computational modeling and functional assays, we identify ligands that dock on POR and bias its specificity towards CYP redox partners, across mammal and plant kingdom. Single molecule FRET studies reveal ligand binding to alter POR conformational sampling, which results in biased activation of metabolic cascades in whole cell assays. We propose the model of biased metabolism, a mechanism akin to biased signaling of GPCRs, where ligand binding on POR stabilizes different conformational states that are linked to distinct metabolic outcomes. Biased metabolism may allow designing pathway-specific therapeutics or personalized food suppressing undesired, disease-related, metabolic pathways.

FRET-based dynamic structural biology: Challenges, perspectives and an appeal for open-science practices

Single-molecule FRET (smFRET) has become a mainstream technique for studying biomolecular structural dynamics. The rapid and wide adoption of smFRET experiments by an ever-increasing number of groups has generated significant progress in sample preparation, measurement procedures, data analysis, algorithms and documentation. Several labs that employ smFRET approaches have joined forces to inform the smFRET community about streamlining how to perform experiments and analyze results for obtaining quantitative information on biomolecular structure and dynamics. The recent efforts include blind tests to assess the accuracy and the precision of smFRET experiments among different labs using various procedures. These multi-lab studies have led to the development of smFRET procedures and documentation, which are important when submitting entries into the archiving system for integrative structure models, PDB-Dev. This position paper describes the current 'state of the art' from different perspectives, points to unresolved methodological issues for quantitative structural studies, provides a set of 'soft recommendations' about which an emerging consensus exists, and lists openly available resources for newcomers and seasoned practitioners. To make further progress, we strongly encourage 'open science' practices.

Single-Molecule FRET Imaging of Virus Spike-Host Interactions

As a major surface glycoprotein of enveloped viruses, the virus spike protein is a primary target for vaccines and anti-viral treatments. Current vaccines aiming at controlling the COVID-19 pandemic are mostly directed against the SARS-CoV-2 spike protein. To promote virus entry and facilitate immune evasion, spikes must be dynamic. Interactions with host receptors and coreceptors trigger a cascade of conformational changes/structural rearrangements in spikes, which bring virus and host membranes in proximity for membrane fusion required for virus entry. Spike-mediated viral membrane fusion is a dynamic, multi-step process, and understanding the structure–function-dynamics paradigm of virus spikes is essential to elucidate viral membrane fusion, with the ultimate goal of interventions. However, our understanding of this process primarily relies on individual structural snapshots of endpoints. How these endpoints are connected in a time-resolved manner, and the order and frequency of conformational events underlying virus entry, remain largely elusive. Single-molecule Förster resonance energy transfer (smFRET) has provided a powerful platform to connect structure–function in motion, revealing dynamic aspects of spikes for several viruses: SARS-CoV-2, HIV-1, influenza, and Ebola. This review focuses on how smFRET imaging has advanced our understanding of virus spikes’ dynamic nature, receptor-binding events, and mechanism of antibody neutralization, thereby informing therapeutic interventions.

Structural Dynamics of Glutamate Signaling Systems by smFRET

Single-molecule Förster resonance energy transfer (smFRET) is a powerful technique for investigating the structural dynamics of biological macromolecules. smFRET reveals the conformational landscape and dynamic changes of proteins by building on the static structures found using cryo-electron microscopy, x-ray crystallography, and other methods. Combining smFRET with static structures allows for a direct correlation between dynamic conformation and function. Here, we discuss the different experimental setups, fluorescence detection schemes, and data analysis strategies that enable the study of structural dynamics of glutamate signaling across various timescales. We illustrate the versatility of smFRET by highlighting studies of a wide range of questions, including the mechanism of activation and transport, the role of intrinsically disordered segments, and allostery and cooperativity between subunits in biological systems responsible for glutamate signaling.