Methods and approaches
The main aim of the research group is to decipher how protein clustering and membrane-mediated effects on protein interactions influence cell biological processes, e.g. transport and signaling, taking place at the plasma membrane. We usually use fluorescence labeling achieved by directly coupling fluorescent dyes to targets of interest (as an example see Fig. 1) or by antibodies (IgG, Fab, nanobody) or fluorescent proteins
Figure 1. Uptake of fluorescently-labeled penetratin. Penetratin is a cell-penetrating peptide, which was labeled either by AFDye532 (A) or by naphthofluorescein (B). Naphthofluorescein exhibits pH-dependent fluorescence that is quenched at acidic pH. Consequently, naphthofluorescein-labeled penetratin is not visible in acidic endo-lysosomes (arrows in A and B), while the pH-independent fluorescence of AFDye532-penetratin is visible in all compartments. Image C shows the transmission image of the cells.
The fluorescently-labeled cells are investigated using Förster resonance energy transfer (FRET), mobility measurements or correlation microscopic methods.
1. FRET is a non-radiative interaction between two fluorophores. An excited fluorophore, called donor, relaxes and transfers its energy without photon emission to a nearby molecule, called the acceptor. Since the acceptor is usually fluorescent as well, FRET leads to the emission of a photon from the acceptor. As a result, FRET leads to acceptor fluorescence after excitation of the donor. The probability of FRET is influenced by a couple of factors, out of which donor-acceptor distance is the most important. FRET only takes place if the acceptor is 2-10 nm away from the donor. Since this distance range corresponds to the dimension of molecular interactions, the occurrence of FRET implies that the donor-tagged and acceptor-tagged molecules interact with each other (Fig. 2).
Figure 2. The reaction scheme of FRET. A donor molecule is excited followed by its relation leading to non-radiative energy transfer to the acceptor. Since the acceptor is fluorescent in most cases, this process leads to the emission of an acceptor-derived photon.
Measurement of FRET can be accomplished by a couple of different approaches. We typically measure the intensities of the donor and the acceptor, or the fluorescence lifetime of the donor to calculate the efficiency of FRET. Fluorescence labeling of our targets of interest is achieved by external or internal labels (Fig. 3).
Figure 3. Measurement of FRET in cell biology. Targets of interests can be labeled by external labels (antibodies or their fragments) or they can be fused to fluorescence markers (fluorescent proteins or other genetically-encoded tags, e.g. SNAP, CLIP).
2. Measurement of mobility of different, fluorescently-labeled membrane components is carried out by fluorescence recovery after photobleaching (FRAP) or fluorescence correlation spectroscopy (FCS). As discussed below, correlation spectroscopy also provides information about clustering.
In FRAP, a certain area (region of interest, ROI) of the membrane is photobleached resulting in the sudden decrease in fluorescence measured in this ROI. Due to the diffusion of unbleached, fluorescently-labeled membrane components to the ROI, the fluorescence is gradually restored. The kinetics of this recovery is related to the diffusion constant, and the extent of the recovery is determined by the mobile fraction of the fluorescently-labeled membrane component (Fig. 4).
Figure 4. The principle of FRAP. A membrane component is labeled fluorescently, and the fluorescence intensity is measured in a region of interest (ROI) (1). The fluorescence in the ROI is bleached (2). Since bleached fluorophores are exchanged for unbleached ones in the rest of the membrane, the fluorescence is gradually restored (3). The bleached membrane region is shown by the red rectangle in the upper left image.
In FCS, the fluctuation of fluorescence intensity in single pixels is analyzed. A membrane component is fluorescently-labeled. The fluorescence intensity in this system fluctuates due to the change in the number of fluorescently-labeled molecules in a single pixel (Fig. 5).
Figure 5. The principle of fluorescence correlation spectroscopy (FCS). The intensity measured by a confocal microscope fluctuates due to the varying number of fluorophores in the confocal volume (orange oval). A small molecule (1) diffuses more slowly and consequently spends a shorter time in the confocal volume than a larger molecule (2). The longer time a fluorophore spends in the confocal volume, the longer the fluorescence intensity stays autocorrelated. This principle forms the bases of the determination of diffusion constant using FCS (see Fig. 6 for more detail).
The autocorrelation function, i.e. the correlation of pixel intensities with intensities measured in the same pixel with a certain delay, decays with the lag time. Fitting of a model equation to the data provides the diffusion constant and the number of diffusing units in a pixel (Fig. 6). If the average number of photons detected from a single pixel is divided by the number of mobile units in a pixel, the average molecular brightness (ε) of a single mobile unit can be calculated. Molecular brightness is a valuable piece of information regarding clustering of membrane proteins since the molecular brightness increases with oligomerization state. If the molecular brightness of a monomer is known, the oligomerization state of mobile units can be determined.
Figure 6. The autocorrelation function. The intensity of a single pixel is measured with high temporal resolution (A). The autocorrelation function, shown in C, is calculated by determining the correlation between pixel intensities with a certain lag time. The intensities of a pixel at time zero and at a short lag time (e.g. t1=10 μs) are correlated with each other (if this time lag is below the diffusional correlation time of the fluorophore, τD=ω^2 / D, where ω and D are the radius of the confocal detection volume and the diffusion coefficient, respectively). If this this time lag approaches the diffusional correlation time, the autocorrelation gradually declines (B and C). The diffusional correlation time is determined by fitting a model equation to the experimentally determined autocorrelation function. In this example, a diffusional correlation time of 42 μs was determined.
3. Since molecular brightness sheds light on oligomerization, clustering can be measured by FCS. However, molecular brightness can also be obtained by regular confocal microscopy by a technique called number and brightness (N&B) analysis (Fig. 7). N&B provides the molecular brightness according to the following equation:
where σ2 and µ are the variance and the mean of a single pixel, respectively, determined from a series of confocal microscopic images (Fig. 7).
Figure 7. The principle of number&brightness (N&B) analysis. The fluorescence intensity of a whole image (confocal section) is measured as a function of time. If the time between consecutive acquisitions of the same pixel (i.e. the frame time) is long enough, the fluctuation of the intensity of a pixel is determined by the average number of mobile units in a pixel. Few mobile units/pixel lead to large fluctuations, whereas the relative fluctuation is small if there are many mobile units/pixel. The relative magnitude of the fluctuation, i.e. SD2/mean, is proportional to the brightness of mobile units, forming the basis of number&brightness (N&B) analysis.
Experiments for measuring fluorescence intensities are mainly carried out by confocal microscopy, superresolution microscopy (STORM) and flow cytometry in our lab. The instruments are listed on the following page:
Post-experimental data processing
We excessively rely on quantitative analysis of the acquired data. These approaches usually involve identification of certain cellular components in microscopic images (image segmentation) by user-assisted or automatic methods, e.g. artificial neural networks. Image analysis is usually performed in Matlab (Fig. 8) or ImageJ. Several user-friendly GUI-controlled (GUI=graphical user interface) Matlab applications are available for the analyses (Fig. 9). For a list of such programs check the following page: https://peternagy.webs.com/what-do-i-do
Figure 8. Example for image segmentation implemented in Matlab. Cells were labeled with a fluorescent marker binding to the cell membrane (F) and with propidium iodide staining nuclei (A). The aim of the image segmentation was to identify membrane pixels so that image analysis could be restricted to the membrane. The nuclear stain served as a seed identifying each cell. The nuclear images were thresholded (B). In the thresholded image, every pixel above a certain threshold is binary 1 (red), other pixels are binary 0 (black). Holes in the identified nuclei were removed by morphological “closing” operation (C) followed by cleaning up the image by removing small speckles by morphological “opening” (D). Nuclei were shrunk to single pixels, which served as seeds for watershed segmentation (G). Membranes identified by the watershed segmentation algorithm are shown in red in G. Every cell and the corresponding membranes were individually identified (H) enabling analysis on a cell-by-cell basis.
Figure 9. The rFRET program used for analyzing microscopic FRET experiments (https://peternagy.webs.com/fret#rfret).