Methods we use
We develop and use an array of innovative methods and assays that are tailored to resolve specific features of biological systems, in-vitro, in cells and in tissue. These methods are extremely sensitive as they rely on single-molecule detection capabilities, and therefore provide unique critical information that cannot be obtained using conventional methods.
By combining the information obtained from the different methods we can generate a detailed description of a biological system.
Some of the single-molecule methods commonly used in our lab are detailed below ::
Single Molecule Fluorescence Localization Microscopy (Super-resolution STORM)
Fluorescent microscopy of biological samples is one the most commonly used methods in biomedical research since it offers unique specificity for marking molecules of interest in side cells. However, the most significant constraint of standard fluorescent microscopy is the diffraction limit of light that was formalized by Lord Rayleigh (Rayleigh criterion) and Ernst Abbe (Abbe limit), and states that the ability to resolve two adjacent objects in the microscope image plane is limited to about half of the wavelength of the fluorescent moiety. Assuming a fluorescent molecule emitting at a wavelength of 600 nm, this mean that the resolution limit is about 300 nm (in X-Y plane) and twice that in Z. When considering the size of proteins and cellular complexes (few nanometers to tens of nanometers) this limitation means that all the specific features under 300 nm will just be blurred, and therefore one cannot resolve the “real” arrangement of molecular complexes.
In the past decade several methods have emerged to address this limitation, generally known as super-resolution microscopy methods. While some of these new methods (such as structured illumination microscopy) offer improved resolution, but are still subjected to a resolution limit essential, methods such as single-molecule localization microscopy offer, in principle, unlimited resolution.
The single-molecule localization microscopy method utilizes fluorescence intermittency (or blinking) and photobleaching event of fluorophores to localize individual molecules and generate super-resolved images. The blinking behavior is a stochastic process that arises from the recurring transition of the emitter between a non-emitting state (off), and an emitting state (on).
The rate of transition and duration in each state depends on a number of parameters, such as irradiation intensity and buffer conditions (oxidizing agents, reducing agents, and free oxygen), and can be tuned accordingly.
In the movie, we show the principle of STORM imaging, where stochastically emitting fluorophores are localized below the diffraction limit. New localization events are mapped as old fluorophores deactivate and new ones activate, allowed for a fully super resolution image to be reconstructed. We also provide an example of the enhanced imaging capabilities of STORM in the Figure below, comparing a regular (diffraction limited) fluorescence image and reconstructed super-resolved image of immunofluorescently labeled DNA damage response markers. In contrast to the diffraction limited blurred foci the super-resolve foci show significantly enhanced spatial features.
Single-molecule Fluorescence (or Förster) Resonance Energy Transfer (smFRET) and single-molecule co-localization (smCL)
The underlining principle of FRET is a distance dependent interaction between a pair of different dye molecules known as the FRET pair. This interaction involves the direct excitation of one donor molecule (D) and the emission from the other acceptor molecule (A), and arises from a dipole-dipole interaction between the two fluorophores (D-A). Being strongly distance dependent, FRET is sometimes referred to as a “spectroscopic ruler”. The effective distance for FRET interaction of a standard FRET pair is between 2 to 8 nm, providing an excellent tool for probing distance changes in bimolecular systems.
To measure changes in the FRET efficiency, the emission intensities of the donor and acceptor are spectrally resolved, and recorded. The FRET efficiency can then be found through the relationship:
Where IAcceptor (Donor) is the emission intensity of the Acceptor (Donor). At large D-A separation, a strong Donor emission is observed, along with a weak (or no) emission from the Acceptor. At smaller D-A separation, the Donor emission will decrease while the Acceptor will increase, as more energy is transferred from the donor to the Acceptor, until at very close proximity the Donor emission will dim while the Acceptor strongly emits.
This is illustrated in the cartoon, where a four-way DNA junction (Holliday junction) is shown tethered to the surface. These DNA structures undergo rapid structural rearrangements, switching between parallel and antiparallel conformations as a function of salt concentration in the buffer, as shown in the movie. We show a representative trajectory of donor-acceptor intensity and its corresponding FRET trajectory of an individual Holliday junction as it undergoes rapid transitions in FRET corresponding to conformational changes.
Read more about how we used this approach to study DNA repair proteins here, here, here, here, and here
In cases when the distance between the two FRET pair is too large for FRET interaction we can utilize a different measurement that relies on the localization of individual molecules (sometimes refer to as single-molecule co-localization smCL). This method uses a surface bound molecule as reference points, and monitors the binding and dissociation kinetics of other molecules to those points, thereby providing single-molecule kinetics. Moreover, by analyzing the intensity of the binding molecules we can derive metrics regarding molecular complex formation and stoichiometry. See the details on how we used this approach to study V(D)J recombination.
Live cell imaging and single-molecule tracking
This method is used for tracking single-molecules/particles in real-time, and was initially applied to resolve the step sizes of individual molecular motors. Briefly, the target of interest is labeled with a fluorophore (fluorescent protein, dye molecule or Quantum Dot) and imaged, resulting in the detection of diffraction-limited spots on the EMCCD or next generation sCMOS camera. Each diffraction-limited spot is fitted with a 2D Gaussian curve, which localizes the fluorophore with nanometer accuracy when there is a sufficiently high signal to noise (S/N), enable to localize a single Cy3 molecule to within 2.3 nm, and temporal resolution of 30 milliseconds. Nanoscale localization in the z (axial) axis was also demonstrated, and can be achieved using defocusing or astigmatism approaches.
Single-molecule trajectories are analyzed to yield precise quantification of various motion related factors, such as diffusion coefficients and displacement characteristics. These quantifications contain information on the nature of observed motion and its immediate environment, such as directed motion mediated by molecular motors, and free and confined diffusion in the membrane or inside cells. The main advantage of this approach is the ability to observe the motions of proteins and other molecules at their length scales, well below the physical limitations of diffraction limited microscopy.
Below we provide two samples movies from our single-molecule tracking experiments, where in the first we monitor recruitment of DNA repair proteins (red dots) to sites of DNA damage (green globular foci) in a human cancer cell, and in the second we visualize bacterial proteins in the phage shock response pathway. We also utilize other live cell imaging methods to monitor various cellular processes. In the example movie below we track the dynamics of DSB induction and repair over long time in human cells where the damage site is marked by EGFP.
Single-molecule pulldown and photobleaching analysis
The ability to detect the emission of individual fluorophores enables the counting of the number of molecules contained in a diffraction-limited spot. This is primarily accomplished through analyzing discrete photobleaching events in the intensity trajectories of individual fluorescently labeled complexes; the numbers of photobleaching events are obtained. In this way, if each subunit in a complex is labeled with a single fluorophore, the number of subunits in the complex can be extracted (proportional to the number of photobleaching events observed in the trajectory.
In this approach, the pulldown target is tethered specifically to the biofunctionalized surface of micro-fluidics chamber and imaged under the microscope using a tagged target protein that is captured with the specific anti-tag antibody. The pull-down complex can be labeled either endogenously (by fusion to a fluorescent protein) or by using fluorescently labeled primary antibody with a dye/antibody ratio of 1:1. The fluorescent signal from individual surface tethered pull-down complexes is analyzed for both co-localization and number of photobleaching steps to yield the composition and stoichiometry of each target.
In the Figure we show an example of photobleaching analysis of DNA damage repair complexes composed of replication protein A (RPA) and DNA repair protein PALB2 that were bound to single stranded DNA. Here the fluorescently labeled oligonucleotide was immobilized on a coverslip, followed by addition of RPA. Nuclear extracts containing a second protein, PALB2 are then allowed to bind and the proteins are labeled with antibodies. Using this approach, we see that PALB2 associates with RPA, and are able to gage relative stoichiometries based on the known immunoreactivity of the antibodies used.
Read more about it here
Computational approaches for analysis of multi-color STORM images
This new type of data as well as the incredible resolution, enhanced sensitivity and detection capabilities of SMLM provide an abundance of previously undetectable information, which presents new challenges for big-data image analysis. Conventional microcopy such as confocal and epifluorescence microscopy generate images that contain information on the detected light intensity for each pixel in the image. However, SMLM–STORM imaging provides an entirely different type of data: the specific coordinates for each single-molecule localization event as well as the number of localizations events. Analysis of this sort of data is especially challenging when dealing with multi-color images that display dense molecular populations and amorphous structures, which is the majority of the biological matter in the cell. To address these challenges, we develop a number of new image analysis computational approaches. In collaboration with the lab of David Fenyo we utilized new Monte Carlo simulation-based analysis to determine the interaction landscape of cellular complexes. We also developed cluster methods and various correlation approaches for SMLM data analyses. The latter are especially useful for unbiased data mining in dense three-color SMLM data sets via the development of new triple-pair correlation algorithms.