This protocol describes a method to determine protein half-lives in single living adherent cells, using pulse labeling and fluorescence time-lapse imaging of SNAP-tag fusion proteins.
Proteins are in a dynamic state of synthesis and degradation and their half-lives can be adjusted under various circumstances. However, most commonly used approaches to determine protein half-life are either limited to population averages from lysed cells or require the use of protein synthesis inhibitors. This protocol describes a method to measure protein half-lives in single living adherent cells, using SNAP-tag fusion proteins in combination with fluorescence time-lapse microscopy. Any protein of interest fused to a SNAP-tag can be covalently bound by a fluorescent, cell permeable dye that is coupled to a benzylguanine derivative, and the decay of the labeled protein population can be monitored after washout of the residual dye. Subsequent cell tracking and quantification of the integrated fluorescence intensity over time results in an exponential decay curve for each tracked cell, allowing for determining protein degradation rates in single cells by curve fitting. This method provides an estimate for the heterogeneity of half-lives in a population of cultured cells, which cannot easily be assessed by other methods. The approach presented here is applicable to any type of cultured adherent cells expressing a protein of interest fused to a SNAP-tag. Here we use mouse embryonic stem (ES) cells grown on E-cadherin-coated cell culture plates to illustrate how single cell degradation rates of proteins with a broad range of half-lives can be determined.
It is well known that cellular proteins undergo extensive turnover, with synthesis and degradation rates being specific for each protein and subject to physiological regulation. Traditionally, protein degradation rates have been measured using bulk methods, such as radioactive pulse chase analysis, or involving protein synthesis inhibitors such as cycloheximide1. More recently, stable isotope labeling with amino acids in cell culture (SILAC) in combination with mass spectrometry has been established to quantify protein turnover on a global scale2. However, these methods are limited by population averaging, and information about cell-to-cell variability is therefore lost. Furthermore, transient changes in protein degradation that are unsynchronized across the cell population cannot be identified.
Alternatively, protein half-lives can also be determined by fluorescence-based approaches, which often have the advantage of providing single-cell resolution. For example, a photoactivatable green fluorescent protein (paGFP) has been used to determine Oct4 half-life in the early mammalian embryo3. Another method to monitor protein decay in living cells is the use of a SNAP-tag in combination with fluorescence time-lapse imaging. The SNAP-tag is a mutant version of the DNA repair enzyme O6-alkylguanine DNA-alkyltransferase (AGT) that specifically reacts with benzylguanine (BG) derivatives, which can be coupled to molecular probes4,5,6. Therefore, any SNAP-tag fusion protein can be irreversibly labeled with a fluorescent, cell permeable dye. Pulse labeling of a protein of interest fused to the SNAP-tag, followed by washout of the residual dye, allows for monitoring the degradation of the labeled protein population and thus for determining protein half-life. SNAP-tags have been successfully used for pulse-chase labeling of proteins and for determining protein half-lives in adherent cell culture and in vivo5,7,8,9. A large variety of SNAP-tag substrates covering commonly used fluorescent spectra are commercially available, enabling the selection of the optimal dye for each specific application. Thus, SNAP-tags can also be used for multicolor imaging in combination with other fluorescent fusion proteins or dyes. Cell-impermeable dyes are suitable for labeling of membrane-tethered proteins, whereas cell-permeable dyes are applicable for monitoring both intracellular and membrane-bound proteins. Furthermore, some of these probes exhibit almost no basal fluorescence and only start emitting a strong fluorescent signal upon binding to a SNAP-tag10.
This protocol describes how to measure the degradation rates of different proteins of interest in single cells using a SNAP-tag. Here we apply this method to mouse embryonic stem (ES) cells cultured on E-cadherin, but it should be possible to use it with any adherent cultured cell type. We show that pulse labeling of SNAP-tag fusion proteins followed by fluorescence time-lapse imaging allows for determining the single cell half-lives of various proteins of interest and provides an estimate for the cell-to-cell variability of half-lives in a population of cultured cells.
Note: In this study, the E14 ES cell line was used. However, this protocol is directly applicable to any other mouse ES cell line expressing a protein of interest fused to a SNAP-tag, either by tagging the endogenous protein or by using overexpression. For the examples shown in the results section, doxycycline-inducible SNAP-tag fusion cell lines were used (SNAP-tag fused to the following proteins: Nanog, Oct4, Srsf11, or to the fluorescent proteins mOrange2 and sfGFP, and put under the control of a doxycycline-inducible promoter. See11 for further information on the plasmids used for the generation of the doxycycline-inducible SNAP-tag fusion cell lines). The doxycycline-inducible system can be particularly useful, as it allows for tightly controlling the timing and intensity of the expression of the protein of interest. C-terminal positioning of the SNAP-tag is recommended, as changing the N-terminal amino acid sequence is more likely to alter the half-life of the target protein (N-end rule12).
1. E-cadherin coating and cell seeding
2. Pulse labeling of the SNAP-tag
Note: For protein decay experiments it is crucial to use an adequate SNAP dye concentration. The concentration should be high enough to yield a bright signal in the beginning of the time-lapse, as the fluorescence will decrease over time. However, using too high dye concentrations might cause residual dye being left in the medium or in the cells even after washing. The free dye might subsequently bind to newly produced SNAP-tag molecules over the course of the movie, which will distort the decay curve. The observed fluorescence signal will depend on the properties of the dye, the cell line used, as well as the expression level of the corresponding protein. Therefore, it is crucial to optimize the dye concentration by testing different dilutions, starting from the dilution suggested for live cell imaging by the manufacturer. For this study, a far-red fluorescent substrate was used. An optimal concentration of 12 nM was determined for doxycycline-inducible overexpression cell lines.
3. Time-lapse microscopy
4. Image processing and analysis
The described protocol provides an estimate of the cell-to-cell variability in half-life for any given protein fused to a SNAP-tag. The use of recombinant E-cadherin-Fc for coating of the imaging plate allows for single cell resolution in ES cells, which otherwise grow in colonies. Single cells can be tracked separately throughout the course of the movie ( Figure 1A).
In order to determine the protein half-life for each single cell, it is crucial to measure the integrated, background-subtracted SNAP-tag fluorescence signal over time ( Figure 1B), with summing up the integrated intensities of both daughter cells in case of divisions. This results in an exponential decay curve for each cell, from which the decay rate and thus the half-life can be extracted by curve fitting ( Figure 1C). Importantly, if an average decay curve is calculated, the single cell traces should be normalized to the first frame to ensure that each cell has the same weight on the average, despite putative differences in initial fluorescence intensity between cells.
The adequate dye concentration depends on the type of dye and cell line used, and should thus be optimized for each experiment. To illustrate this, Figure 2 shows the decay curves of different concentrations of the far-red fluorescent substrate used for all experiments shown here, tested on a doxycycline inducible SNAP-tag fusion cell line. An initial dye concentration of 3 µM was chosen according to the manufacturer's instructions and a serial dilution of 1:3 was performed until reaching a concentration of 1.4 nM. For the two highest concentrations the signal does not decrease, whereas the observed decay is exponential for concentrations below 111 nM. An optimal concentration of 12 nM of dye was chosen for further experiments, since it ensured minimal amounts of residual dye left in the medium after washing, but the signal was still bright enough to observe clear decay curves for a variety of doxycycline inducible cell lines.
Figures 3A-3C show representative results, including the single cell decays and population averages, for 3 proteins with different average half-lives: the pluripotency associated transcription factors Nanog (average half-life 2.9 h) and Oct4 (average half-life: 4.8 h), and the much longer lived pre-mRNA processing factor Srsf11 (average half-life 25.3 h). The boxplots ( Figure 3D) represent the half-lives obtained from the single cell decay curves by curve fitting and illustrate the heterogeneity of half-lives for the three proteins.
Figure 1: Time-lapse acquisition and analysis
A: Series of representative images showing the decay of fluorescence in cells pulse labeled with SNAP dye. Cell line: doxycycline inducible TRE3G-Oct4-SNAPtag, induced with 500 ng/mL doxycycline 7 h before imaging.
B: Movie analysis and calculation of integrated intensity. Ic: Integrated intensity of the cell, IBG: Integrated intensity of the background, IC-BG: background-corrected integrated intensity.
C: Exponential curve fitting performed in MATLAB. Example of a single cell decay trace, the corresponding fit and the resulting estimates for the decay parameters a and b. The half-life is calculated as follows: t1/2 = ln(2)/b. Please click here to view a larger version of this figure.
Figure 2: Optimization of the dye concentration.
Fluorescence decay of cells pulse labeled with different concentrations of SNAP dye. A serial dilution of the dye starting from 3 µM was performed. The traces are averages of 5 cells each, normalized to the first frame and tracked for 12 h. Cell line: doxycycline inducible TRE3G-sfGFP-mOrange2-SNAPtag-PEST, induced with 500 ng/mL doxycycline 24 h before imaging. Please click here to view a larger version of this figure.
Figure 3: Single cell variability of half-lives for different SNAP-tagged candidate proteins.
A-C: Single-cell decays (light blue) and population averages (dark blue) for doxycycline inducible Nanog-, Oct4- and Srsf11-SNAP-tag fusion proteins.
D: Half-lives of individual cells, which were calculated by exponential curve fitting for each single cell. The boxplots are displayed in log2 scale. The boxes display the interquartile range (IQR), the black line indicates the median. The whiskers represent the minimum and maximum values, excluding the outliers (black dots), which are defined as values deviating more than 1.5x from the IQR. N = 20 cells each. Please click here to view a larger version of this figure.
The most crucial step when using a SNAP-tag to monitor protein decay is to ensure that no residual unbound dye is left in the medium or in the cells after washing, as otherwise it might bind to newly produced SNAP-tag molecules later in the course of the experiment and thereby compromise the decay curve. This is on one hand achieved by carefully performing all the described washing steps. On the other hand, the dye concentration should be kept as low as possible, while still being in a range that allows for obtaining a good signal to noise ratio. As shown in Figure 2, the dye concentration should be optimized for each experiment by testing different dilutions. In our case, the doxycycline inducible cell lines have relatively high expression levels, thus allowing to use a low dye concentration.
This protocol describes a general approach to quantify protein degradation in single cells by time-lapse fluorescence imaging and various aspects of the protocol can be modified, depending on the experimental conditions and the goal of the corresponding experiment. Protein degradation mostly follows a first order exponential decay. For the protein decays shown here, the R2 values of the fits were fairly high (0.95-0.99) and using multi-exponentials has not significantly improved the fits. However, if the obtained results from single exponential fits are not satisfying, fitting multi-exponentials might be considered, which would imply the existence of multiple labeled protein subpopulations decaying with different rates. Furthermore, our image analysis pipeline involves substantial manual work, which is only feasible if limited numbers of cells are analyzed. For higher numbers of cells, a more automated approach should be considered. Various automated tools have been developed in the past years, which allow for segmenting and tracking larger numbers of cells16,17. However, the movement of the cells and frequent cell divisions are challenging and prone to cause segmentation errors. In addition, when imaging protein decays, the fluorescence signal will decrease over time, making the selection of a threshold for segmentation even more difficult. Thus, tracking tools should be carefully tested before being implemented for this protocol, as segmentation errors can distort the data considerably. In addition, a different background subtraction method may be considered, depending on the quality of the acquired images. Here we use FIJI's rolling ball correction, which is suitable for evenly illuminated images with rather sparsely seeded cells. However, especially if images suffer from uneven illumination, more specialized background subtraction methods might be applied18,19.
One limitation to our approach is the duration of the time-lapse recordings that can be analyzed. For relatively short-lived proteins, an acquisition time of about 12 h is sufficient. For longer-lived proteins, as for example Srsf11, performing a curve fitting on a time range of 12 h is possible. However, it is important to keep in mind that the exponential fits might be less precise for longer-lived proteins, for which the acquisition time and the imaging intervals should be increased. If fast cycling cells are being imaged, longer acquisition times will complicate the analysis, due to the increasing number of daughter cells to be analyzed. Again, in this case, a more automated analysis pipeline might be useful. In addition, when using SNAP-tag fusion proteins to determine protein decay rates, the possibility of the SNAP-tag altering the half-life of the endogenous protein should be considered. In the case of the cell lines used here we see no evidence for such effects, as the determined half-lives of Oct4 and Nanog closely match published values3,20,21,22. To further rule out an effect of the SNAP-tag on protein degradation, especially when no data on half-life is available, one option would be to block protein synthesis by cycloheximide and to perform western blots at different time points using an antibody recognizing the endogenous protein, and directly compare the decay of the SNAP-tagged candidate protein with the untagged endogenous version.
Most previously published data on protein half-life is based on bulk biochemical experiments, mainly time-course western blot experiments upon inhibition of protein synthesis by cycloheximide. However, this approach does not provide single cell resolution and its temporal resolution is low due to the limited number of time-points that can be taken into account. In contrast, our single-cell approach does not require the use of protein synthesis inhibitors and can identify cell-to-cell variability in protein degradation rates. It also allows for identifying potential subpopulations of cells with different half-lives, which might be particularly interesting for the investigation of heterogeneously expressed proteins. A further advantage of our method is the possibility to in silico synchronize the cells and therefore monitor transient changes in half-life, for example throughout the cell cycle. However, the SNAP-tag technology requires the generation of a corresponding SNAP-tagged cell line, which might be its major drawback. Importantly, our approach can also be applied to endogenously tagged proteins9. As mentioned above we do not observe any evidence of the SNAP-tag altering the endogenous protein half-life for the proteins we have studied, suggesting that the accuracy of the SNAP-tag approach is in the same range as other methods.
The method presented here will be useful for various applications. Most importantly, it provides the option to study cell-to-cell variability in half-lives for any SNAP-tag fusion protein of interest. There are several other tags available (such as the CLIP-tag23 or the Halo-tag24) whose working principle is the same. Pulse labeling and washing of the residual dye will lead to a decay curve, which allows for determining protein half-life, whereas leaving the dye in the medium (in case of a ligand that is non-fluorescent unless bound to the tag) enables long-term imaging of a protein of interest. Due to the large selection of available fluorescent ligands, the SNAP-tag can thus be combined with other tags or with multiple other fluorescently labeled molecules, allowing for applying the technique in various contexts.
The authors have nothing to disclose.
Time-lapse microscopy experiments were performed at the Biomolecular Screening Facility (BSF), EPFL. We thank Marc Delachaux (Service Audiovisuel, EPFL) for the videography and editing of the movie.
Equipment | |||
ES cell line expressing a SNAP-tag fusion protein of interest | – | – | |
Falcon 100 mm TC-Treated Cell Culture Dish | Corning | 353003 | |
96 Well, Black/Clear, Tissue Culture Treated Plate | Corning | 353219 | |
Neubauer-improved counting chamber, 0.1 mm | Marienfeld-superior | 640030 | |
CO2 Incubator | Panasonic | MCO-170AICUV-PE | |
Centrifuge 5804 R | Eppendorf | 5804000528 | |
InCell Analyzer 2200 Cell Imaging System | GE Healthcare Life Sciences | 29027886 | |
Name | Company | Catalog Number | Comments |
Reagents | |||
Glasgow Minimum Essential Medium | Sigma-Aldrich | G5154 | |
Fetal Bovine Serum, embyonic stem cell-qualified | ThermoFisher | 16141-079 | |
Sodium pyruvate solution | Sigma-Aldrich | 113-24-6 | |
Minimum Essential Medium Non-Essential Amino Acids | ThermoFisher | 11140-035 | |
Penicillin-Streptomycin | BioConcept | 4-01F00H | |
L-Glutamine 200mM | ThermoFisher | 25030-024 | |
2-Mercaptoethanol | Sigma-Aldrich | 63689-25ML-F | |
Leukemia Inhibitory factor | – | – | Produced in the lab by transient transfection of HEK-293T cells, followed by collection and filtering of the supernatant. |
CHIR99021 (GSK-3 Inhibitor XVI) | Merck Millipore | 361559 | |
PD 0325901 | Sigma-Aldrich | 391210-10-9 | |
Gelatin from bovine skin | Sigma-Aldrich | 9000-70-8 | |
Dulbecco's PBS 10x concentrated | BioConcept | 3-05K00-I | |
Dulbecco's PBS Without Ca++/Mg++ | BioConcept | 3-05F29-I | |
Trypsin-EDTA-Solution 0.25% | Sigma-Aldrich | T4049 | |
Recombinant Mouse E-Cadherin Fc Chimera protein | R&D systems | 748-EC-050 | |
Doxycycline hyclate | Sigma-Aldrich | D9891 | |
SNAP-Cell 647-SiR | New England BioLabs | S9102S | |
FluoroBrite DMEM | ThermoFisher | A18967-01 | |
Name | Company | Catalog Number | Comments |
Software | |||
FIJI | – | – | Open-source image analysis software |
MATLAB R2014a | Mathworks | – | |
Microsoft Excel | Microsoft | – |