This article presents a method for spatiotemporal analysis of mobile, single-molecule Förster resonance energy transfer (smFRET)-based probes using widefield fluorescence microscopy. The newly developed software toolkit allows the determination of smFRET time traces of moving probes, including the correct FRET efficiency and the molecular positions, as functions of time.
Single-molecule Förster resonance energy transfer (smFRET) is a versatile technique reporting on distances in the sub-nanometer to nanometer range. It has been used in a wide range of biophysical and molecular biological experiments, including the measurement of molecular forces, characterization of conformational dynamics of biomolecules, observation of intracellular colocalization of proteins, and determination of receptor-ligand interaction times. In a widefield microscopy configuration, experiments are typically performed using surface-immobilized probes. Here, a method combining single-molecule tracking with alternating excitation (ALEX) smFRET experiments is presented, permitting the acquisition of smFRET time traces of surface-bound, yet mobile probes in plasma membranes or glass-supported lipid bilayers. For the analysis of recorded data, an automated, open-source software collection was developed supporting (i) the localization of fluorescent signals, (ii) single-particle tracking, (iii) determination of FRET-related quantities including correction factors, (iv) stringent verification of smFRET traces, and (v) intuitive presentation of the results. The generated data can conveniently be used as input for further exploration via specialized software, e.g., for the assessment of the diffusional behavior of probes or the investigation of FRET transitions.
Förster resonance energy transfer (FRET) has been a major driver in molecular biological and biophysical research, as it allows the investigation of processes at sub-nanometer resolution. As the efficiency of energy transfer between donor and acceptor fluorophores strongly depends on the inter-dye distance in the sub-nanometer to nanometer range, it has been effectively used as a spectroscopic ruler to explore static and dynamic conformation of biomolecules1,2,3,4. Additionally, the FRET phenomenon has been widely used for colocalization studies of membrane-associated and intracellular proteins on a bulk level5,6. In the last two decades, the method was adapted for monitoring smFRET events7, which helped to substantially increase temporal and spatial resolution and resolved even rare subpopulations in heterogeneous samples. Equipped with these techniques, unique insights were gained into the dynamics of molecular machinery such as the transcript processing rate of RNA polymerase II8, replication speed of DNA polymerases9,10, nucleosome translocation rate11, transcript splicing and stalling rate of assembled spliceosomes12, the activity of ribosomal subpopulations13, and the walking speed of kinesin motors14, to name a few. Receptor-ligand interaction durations15 and molecular forces16 have been quantified.
Intensity-based smFRET studies typically rely on sensitized emission to measure FRET efficiency: a beam splitter in the emission path spatially separates light originating from donor and acceptor fluorophores upon donor excitation, allowing for the quantification of individual fluorescence intensities. The efficiency can subsequently be calculated as the fraction of photons emitted by the acceptor with respect to the total photon count17. In addition, acceptor excitation following donor excitation (ALEX) permits measurement of the stoichiometry of the FRET events, aiding in the discrimination between true low FRET signals from signals arising, e.g., from probes featuring a photobleached acceptor fluorophore18.
Single-molecule FRET experiments are commonly carried out in one of two ways. First, a small region in the sample volume is illuminated using a confocal microscope. Single probe molecules in solution are excited when they happen to diffuse within the focal volume. With this technique, fast photon-counting detectors can be used, enabling sub-microsecond time resolution. Second, probes are specifically immobilized on surfaces and monitored via widefield microscopy, often using total internal reflection (TIR) configuration to minimize background fluorescence. Probe immobilization allows for much longer recording times than using the first approach. In addition, the larger field of view permits the monitoring of multiple probes in parallel. The need for a camera makes this method slow compared to the one described above. Time resolution is limited to the millisecond to second range.
If long time traces are required, e.g., for studying dynamic processes on a millisecond to second time scale, the first method is not applicable, as the fluorescence bursts are typically too short. The second approach fails whenever immobilization is not feasible, e.g., in live-cell experiments featuring probes diffusing within the cell membrane. Furthermore, it has been observed that biological model systems can vary their response dramatically depending on the mobility of the contacted surface16.
While combined smFRET and single-particle tracking experiments recording mobile FRET probes have been performed in the past19, there is no publicly available software for the evaluation of the data. This prompted the development of a new analysis platform, which allows for the determination of multiple properties of mobile fluorescent probes, including smFRET efficiency and stoichiometry, positions with sub-pixel accuracy, and fluorescence intensities as functions of time. Methods for filtering the resulting traces by examining stepwise bleaching behavior, nearest-neighbor distances, emission intensities, and other traits were established to exclusively choose correctly synthesized and functional single-probe molecules. The software also supports experimental and analytical techniques recently agreed upon in a multilaboratory study to produce reliable, quantitative smFRET data17. In particular, the implementation adheres to the validated procedures for the calculation of FRET efficiency and stoichiometry. Fluorescence intensities upon donor excitation in the donor emission channel IDD and acceptor emission channel IDA are used for the calculation of the apparent FRET efficiency Eapp using Eq (1).
(1)
With the help of the fluorescence intensity in the acceptor emission channel upon acceptor excitation IAA, the apparent stoichiometry is calculated using Eq (2).
(2)
The FRET efficiency E and the stoichiometry S can be derived from Eapp and Sapp by considering four correction factors.
α describes the leakage of donor fluorescence into the acceptor emission channel and can be determined using a sample containing only donor fluorophores or by analyzing parts of trajectories where the acceptor has been bleached. δ corrects for the direct excitation of the acceptor by the donor excitation light source and can be measured using a sample with only acceptor fluorophores or by analyzing parts of trajectories where the donor has been bleached.
.
γ scales IDD to rectify diverging detection efficiencies in donor and acceptor emission channels and different quantum efficiencies of the fluorophores. The factor can be computed by analyzing the increase in donor intensity upon acceptor bleaching in trajectories with high FRET efficiencies20 or by studying a sample featuring multiple discrete FRET states.
β scales IAA to correct for disparate efficiencies of donor and acceptor excitation. If γ was determined via acceptor bleaching analysis, β could be calculated from a sample of known donor-to-acceptor ratio21. Otherwise, the multi-state FRET sample also yields β.
Together, the corrections allow the calculation of the corrected FRET efficiency using Eq (3).
(3)
and the corrected stoichiometry using Eq (4).
(4)
Ideally, the corrected stoichiometry for a 1:1 donor-to-acceptor ratio gives S = 0.5. In practice, a reduced signal-to-noise ratio produces a spread of the measured values of S, hampering the discrimination from donor-only signals (S = 1) and acceptor-only signals (S = 0). The resulting time traces can be used as input for a more detailed analysis of the single-molecule trajectories to obtain information such as spatiotemporal force profiles16, the mobility of the single-molecule events22, or transition kinetics between different states1.
The following protocol describes experimental parameters and procedures for smFRET tracking experiments, as well as the working principle behind data analysis using the newly developed software suite. For the acquisition of experimental data, it is recommended to use a microscopy setup meeting the following requirements: i) capability of detecting the emission of single dye molecules; ii) widefield illumination: in particular for live-cell experiments, total internal reflection (TIR23,24,25) configuration is recommended; iii) spatial separation of emission light according to wavelength such that donor and acceptor fluorescence is projected onto different regions of the same camera chip25 or different cameras; iv) modulation of light sources for donor and acceptor excitation with millisecond precision, e.g., using directly modulatable lasers or modulation via acousto-optic modulators. This permits stroboscopic illumination to minimize photobleaching of fluorophores as well as alternating excitation to determine stoichiometries; v) output of one file per recorded image sequence in a format that can be read by the PIMS Python package26. In particular, multipage TIFF files are supported.
1. Software prerequisites
2. Measurement of samples
Figure 1: Image Acquisition. (A) Excitation sequence. After recording an optional image of a dye-loaded cell using the 405 nm laser, donor and acceptor are excited alternately and repeatedly for illumination time till using 532 nm and 640 nm lasers, respectively. The time tr between donor and acceptor excitation must be long enough to allow for image readout by the camera. The delay time tdelay can be used to adjust the acquisition frame rate and, therefore, the observation time span before photobleaching. This panel is modified from 16. (B) Fiducial markers are used for the calculation of the coordinate transforms between the two emission channels. Matching fiducials are indicated by color. Several shifted images should be recorded to ensure that the whole field of view is covered. (C) Laser profiles for flatfield correction are recorded using a densely labeled sample. The acceptor profile is recorded and photobleached, followed by acquisition of the donor profile. Multiple images should be taken at different sample regions, averaged, and smoothed to mitigate the influence of sample imperfections (e.g., the bright spot in the center-top of the image). (D) Flatfield correction map p(x,y) calculated from 20 laser profile recorded as described in C. Abbreviations: FRET = Förster resonance energy transfer; ImDD = donor emission image upon donor excitation; ImDA = acceptor emission image upon donor excitation; ImAA = acceptor emission image upon donor excitation. Scale bars = 5 µm. Please click here to view a larger version of this figure.
3. Additional measurements for the determination of correction factors
4. Single-molecule localization algorithms
NOTE: Several analysis steps require single-molecule localization. Choose between a Gaussian fitting algorithm30 and center-of-mass computation31, depending on signal density, background, and signal-to-noise ratio.
5. Software initialization
Figure 2: Overview of a typical analysis pipeline. Note that filtering steps are subject to adaptation according to the experimental design. This figure is modified from 16. Abbreviation: FRET = Förster resonance energy transfer. Please click here to view a larger version of this figure.
NOTE: Sample data to try out the software can be downloaded from https://github.com/schuetzgroup/fret-analysis/releases/tag/example_files
6. Localization, tracking, and fluorescence intensity analysis of single molecules (01. Tracking.ipynb).
Figure 3: Single-molecule intensity measurement. (A) For a fluorophore located at the orange pixel, its uncorrected intensity Iuncorr is determined by summing up all pixels' intensities within a disk (yellow and orange pixels) large enough to cover all pixels affected by the signal: . The local background is computed as the mean of the pixels in a ring (blue pixels) around the disk: , where nring is the number of pixels in the ring. The fluorescence intensity I is the result of subtracting the background from the uncorrected intensity, I = Iuncorr - b × ndisk, where ndisk is the number of pixels in the disk. The circle radius is specified via the feat_radius parameter of the tracking method. The width of the ring is given by the bg_frame parameter. If the point spread function of one signal overlaps with the background ring of another (bottom panel), the affected pixels (red) are excluded from local background analysis. If two point spread functions overlap, fluorescence intensities cannot be calculated reliably and are therefore discarded. (B, C) Simulations show that applying a Gaussian blur with a standard deviation of 1 pixel improves the signal-to-noise ratio up to a factor of close to 2 at low fluorescence intensities (B) and introduces hardly any error (slight underestimation of less than 1%, (C))16. Moreover, the relative error (i.e., (Imeas - Itruth)/Itruth, where Itruth is the ground truth and Imeas is the outcome of the analysis) is constant over the whole intensity range and therefore cancels out for ratiometric quantities such as FRET efficiencies and stoichiometries. All plots are based on previously published work16. Abbreviations: SNR = signal-to-noise ratio; FRET = Förster resonance energy transfer. Please click here to view a larger version of this figure.
7. Visualization of FRET trajectories (optional)
8. Analysis and filtering of single-molecule data (02. Analysis.ipynb)
9. Plotting of results and further analysis (03. Plot.ipynb)
NOTE: Refer to Supplemental Information for screenshots of the Jupyter notebook and description of function call parameters.
A variety of low- and high-level information can be extracted from smFRET tracks depending on the scientific question of the experiment. Here, examples of analysis pipelines with analog and digital probes are presented: a peptide-based molecular force sensor16 and a DNA probe with stochastic switching of its conformation38, respectively. Refer to Figure 5 for the design and working principle of these probes.
After the localization and tracking algorithms have been executed as described in the protocol, the package offers multiple data visualization tools to optimize the initial parameters and subsequent filter steps: (i) visualization of individual smFRET events, (ii) optional image segmentation to analyze data in certain regions of interests, (iii) monitoring of filter steps via FRET efficiency vs. stoichiometry (E-S) plots. The visualization of the single-molecule data is presented in Figure 6.
Finally, the filtered FRET events are represented by an E-S plot and a FRET efficiency histogram (Figure 4). The E-S plot is a useful tool for optimizing the aforementioned filtering steps and investigating the final result. Partially bleached or incompletely labeled FRET sensors can be excluded by their stoichiometry value. Mobility parameters can be investigated by plotting an individual trajectory path in an x-y plot (Figure 6) or a mean square displacement (MSD) plot (Figure 4). The first method is especially useful for discriminating mobile from immobilized events, while the latter is used to calculate the diffusion coefficient.
Figure 4: Exemplary output. (A) The FRET efficiency is plotted versus stoichiometry (E-S plot) for a population of the molecular force sensor (left panel) decorating a glass-supported lipid bilayer and strained by a T cell. Only one population cloud is visible. The respective histogram of FRET efficiencies exemplifies the difference between a force sensor population in presence and absence of cells (middle panel). No shift to lower FRET efficiencies of the sensor population in presence of T cells can be observed, indicating little to no force-dependent stretching of the sensor module. The MSD plot of these experimental conditions confirms that the force sensor population beneath a T cell moves considerably slower than their unbound counterparts (right panel). (B) The same analysis was performed with Holliday junction DNA sensor decorating a glass-supported fluid lipid bilayer. The E-S plot clearly shows two populations, which are also apparent in the FRET efficiency histogram. The MSD plot indicates the presence of one fast-moving sensor population. Abbreviations: FRET = Förster resonance energy transfer; MSD = mean square displacement. Please click here to view a larger version of this figure.
Figure 5: Design and working principle of intramolecular FRET probes. (A) Analog peptide sensor for quantification of mechanical molecular forces. The donor and acceptor fluorophores are covalently attached to either end of the peptide backbone. The sensor module is site-specifically attached to a specific ligand, which in turn binds a cell-resident surface receptor of interest (here, an antibody fragment specifically recognizing the beta chain of the T cell receptor). Upon receptor-ligand binding, force is exerted, and the sensor module extends and eventually recoils after bond cleavage. This panel is modified from 16. (B) Digital DNA sensor for quantification of FRET transitions. The FRET sensor is composed of four DNA strands forming a Holliday junction. The donor and acceptor fluorophore are covalently attached to two strands. Holliday junctions frequently switch their conformation depending on the surrounding buffer conditions. The stochastic switching of these conformations can be monitored by quantifying the FRET efficiency of individual probes. Abbreviations: TCR = T cell receptor; FRET = Förster resonance energy transfer. Please click here to view a larger version of this figure.
Figure 6: Examples of localization and tracking of FRET probes. (A) The FRET efficiency and stoichiometry of individual events are calculated by quantifying the intensity of the donor fluorophore upon donor excitation (D → D), the acceptor fluorophore upon donor excitation (D → A), and the acceptor fluorophore upon acceptor excitation (A → A). Nearest neighbor filtering prevents bias by overlapping point spread functions of close emitters. Image segmentation allows the user to choose certain smFRET events localized within an area of interest (e.g., a cell or a micropattern). As an example of image segmentation, T cells were stained with Fura-2 (displayed on the left) and subjected to adaptive thresholding to identify the cell edges (orange dotted line). Scale bars = 5 µm. (B) smFRET trajectories using the molecular force sensor. Individual trajectories can be plotted in the x-y plane, visualizing their diffusion behavior and localization (left panel). Furthermore, each trajectory's intensities can be plotted over time to identify FRET transitions or bleaching steps (middle panel shows the red trajectory from the left panel). The resulting FRET efficiency and stoichiometry can be visualized similarly (right panel). (C) smFRET trajectories using the Holliday junction DNA sensor. HBSS + 12 mM MgCl2 was used as a buffer during the measurements. Apart from the apparent acceptor bleaching step near the sequence end of these examples, the frequency of FRET transitions for each sensor can be determined. The Holliday junctions switch their conformation with a high frequency, whereas the molecular force sensor does not exhibit FRET transitions. This information makes it possible to adjust the experimental conditions, such as the delay between the frames, to increase or reduce the number of observed transitions. Abbreviations: FRET = Förster resonance energy transfer; smFRET = single-molecule FRET; HBSS = Hank's balanced salt solution. Please click here to view a larger version of this figure.
Supplemental Information: Localization and tracking of single molecules (01. Tracking.ipynb). Please click here to download this File.
This article details a pipeline for the automated recordings and quantitative analysis of smFRET data originating from mobile yet surface-tethered probe molecules. It complements the two predominant approaches to smFRET experiments, involving either surface-immobilized probes or probes diffusing in solution into and out of a confocal excitation volume17. It provides the correct FRET efficiency and the molecular positions as a function of time. It can therefore be used as input for specialized analysis programs, e.g., to quantify transition kinetics1, FRET histograms39, or two-dimensional diffusion22.
The software is released under a free and open-source license approved by the Open Source Initiative that grants the user the perpetual right to free usage, modification, and redistribution. Github was chosen as a development and distribution platform to make it as easy as possible to obtain the software and participate in the development process by reporting bugs or contributing code40. Written in Python, the software does not depend on proprietary components. The choice of Jupyter notebooks as user interfaces facilitates the inspection of data at every analysis step and allows for tailoring and extending the pipeline specifically for the experimental system at hand. The sdt-python library32 serves as the foundation and implements functionality to evaluate fluorescence microscopy data, such as single-molecule localization, diffusion analysis, fluorescence intensity analysis, color channel registration, colocalization analysis, and ROI handling.
In principle, single-particle tracking can be performed in one-, two- or three-dimensional systems. Here, the single-molecule analysis pipeline was tailored to the study of 2D mobile systems. This choice mirrors the availability of simple systems, such as planar-supported lipid bilayers (SLBs), to present mobile fluorescent probes. Such lipid bilayer systems are typically composed of two or more phospholipids moieties, where the bulk fraction determines the key physicochemical parameters of the SLB (such as phase and viscosity), and the minor fraction provides attachment sites for biomolecules. These attachment sites can be biotinylated phospholipids for avidin- or streptavidin-based protein platforms or nickel-NTA conjugated phospholipids for protein platforms with histidine tags41. The choice of the appropriate platform for linking proteins to the SLB depends on the scientific question. Readers can refer to the literature16,38,42 for examples of successfully employed strategies. The density of probes in the sample should be sufficiently low to avoid overlapping point spread functions; typically, less than 0.1 molecules per µm2 are recommended. See the representative results section (in particular, Figure 6) for an example showing a suitable probe density. The analysis method is also applicable to single fluorescently labeled protein molecules diffusing in the plasma membrane of live cells.
One critical aspect of smFRET experiments is the production and characterization of the FRET probes themselves. When choosing fluorophores for a FRET pair, their Förster radius should match the expected inter-dye distances43. Dyes resistant to photobleaching are preferred as they yield long time traces. However, for elevated bleaching rates, one fluorophore species can be utilized to recognize multiemitter events originating from colocalized molecules via stepwise photobleaching analysis; see step 8.1.4 in the protocol section. Fluorophore pairs should be site-specifically and covalently attached to the molecules of interest, forming intra- or intermolecular FRET pairs.
Combining smFRET with other readily available techniques can increase its spatial resolution beyond the diffraction limit (via STED44). The smFRET tracking algorithm presented here widens the approach's applicability to new experimental settings and model systems. This includes studies of (i) kinetic changes in the stoichiometry of mobile biomolecules, (ii) dynamic association of mobile biomolecules, (iii) the rate of enzymatic reactions of freely diffusing reactants, and (iv) the kinetics of conformational changes of mobile biomolecules. The first two examples require model systems that show intermolecular FRET, i.e., donor and acceptor are conjugated to separate biomolecular entities of interest. The latter examples may make use of biosensors carrying donor and acceptor within the same molecular entity (intramolecular FRET).
Intramolecular FRET-based sensors can provide insight into intrinsic conformational changes of biomolecules1,2,3,4, conformational changes caused by endogenous or external force load (molecular force sensors16), or ion concentrations in the nano-environment such as calcium45 and pH46. Depending on the model system and the preferred anchoring platform, such smFRET events can either be tracked in 2D or 3D: (i) planar tracking of smFRET events can be employed for the quantification of receptor-ligand interaction times within a plasma membrane, the association of membrane-anchored signal amplification cascades, and the stoichiometry changes of surface receptors; (ii) volume tracking of smFRET events can be used for any intra- or intermolecular FRET probes in living cells or in in vitro reconstituted systems.
The smFRET tracking method was developed mainly with intramolecular FRET probes in mind. These probes feature a fixed and well-known number of fluorescent labels, a fact that was exploited to reject data from agglomerated and incorrectly synthesized (e.g., incompletely labeled) molecules, as well as from probes where one of the fluorophores has been photobleached. However, by adjusting the filtering steps, the method can also be applied to intermolecular FRET probes. For instance, instead of accepting only molecules featuring a single donor and a single acceptor fluorophore, one could examine the spatial trajectories of donor and acceptor dyes and select, for example, for co-diffusing donor-acceptor trajectories.
As the 3D-DAOSTORM algorithm has support for determining a signal's position along the optical axis via the astigmatism due to a cylindrical lens in the emission beam path, 3D experiments could be easily integrated into the analysis pipeline. In this case, the acceptor signal upon acceptor excitation would serve to determine the stoichiometry and the axial position. The analysis software can also be employed to evaluate data from experiments featuring immobilized probes by utilizing its large degree of automation and filtering schemes. In fact, smFRET efficiency datasets from Holliday junctions immobilized on gel-phase bilayers38 were analyzed using an early version of the software.
This work was supported by the Austrian Science Fund (FWF) projects P30214-N36, P32307-B, and by the Vienna Science and Technology Fund (WWTF) LS13-030.
Name | Company | Catalog Number | Comments |
1,2-dioleoyl-sn-glycero-3-[(N-(5-amino-1-carboxypentyl)iminodiacetic acid)succinyl] (nickel salt) (Ni-NTA-DOGS) | Avanti Polar Lipids | 790404P | |
1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) | Avanti Polar Lipids | 850375P | |
1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) | Avanti Polar Lipids | 850457P | |
α Plan-FLUAR 100x/1.45 oil objective | Zeiss | 000000-1084-514 | |
Axio Observer microscope body | Zeiss | ||
Bandpass filter | Chroma Technology Corp | ET570/60m | donor emission filter |
Bandpass filter | Chroma Technology Corp | ET675/50m | acceptor emission filter |
conda-forge | conda-forge community | community-maintaned Python package repository for Anaconda/miniconda | |
Coverslips 60 mm x 24 mm #1.5 | MENZEL | ||
Dichroic mirror | Semrock Inc | FF640-FDi01-25×36 | separation of donor and acceptor emission |
Dichroic mirror (quad band) | Semrock Inc | Di01-R405/488/532/635-25×36 | separation of excitation and emission light |
DPBS | Sigma-Aldrich | D8537 | |
FCS | Sigma-Aldrich | F7524 | for imaging buffer |
fret-analysis | Schütz group at TU Wien | Python package for smFRET data analysis; version 3 | |
Fura-2 AM | Thermo Fisher Scientific | 11524766 | |
HBSS | Sigma-Aldrich | H8264 | for imaging buffer |
iBeam Smart 405-S 405 nm laser | Toptica Photonics AG | ||
iXon Ultra 897 EMCCD camera | Andor Technology Ltd | ||
Lab-Tek chambers (8 wells) | Thermo Fisher Scientific | 177402PK | for sample preparation and imaging |
Millenia Prime 532 nm laser | Spectra Physics | ||
miniconda | Anaconda Inc. | Python 3 distribution. Min. version: 3.7 | |
Monovalent streptavidin (plasmids for bacterial expression) | Addgene | 20860 & 20859 | |
OBIS 640 nm laser | Coherent Inc | 1185055 | |
Optosplit II | Cairn Research | ||
Ovalbumin | Sigma-Aldrich | A5253 | for imaging buffer |
Plasma cleaner | Harrick Plasma | PDC-002 | |
sdt-python | Schütz group at TU Wien | Python library for data analysis; version 17 | |
TetraSpek bead size kit | Thermo Fisher Scientific | T14792 | Randomly distributed, immobilized fiducial markers for image registration |
USC500TH Ultrasound bath | VWR | for SUV formation |
This article has been published
Video Coming Soon
ABOUT JoVE
Copyright © 2024 MyJoVE Corporation. All rights reserved