|
1.IntroductionA well-known, yet ever-evolving fluorescence technology that is used for measuring the mobility of molecules on a microscopic level is fluorescence recovery after photobleaching (FRAP).1, 2, 3 Basically, the fluorescent molecules are at first photobleached in a certain region of the sample and immediately after the bleaching phase, fluorescent molecules from the surrounding unbleached areas will start to diffuse into the bleached area. The resulting recovery of the fluorescence in the bleached area is subsequently measured by a highly attenuated light beam. The rate of fluorescence recovery reflects the mobility of the species in the system, expressed by the effective translational diffusion coefficient . FRAP has been used, for example, in cellular applications to assess the translational mobility of all kinds of solutes in cytoplasm, nuclei, and membranes3, 4, 5, 6, 7 and extracellular matrices such as cervical mucus, biofilms, cystic fibrosis sputum, and bovine vitreous.8, 9, 10, 11 The diffusion coefficient can be calculated from a FRAP experiment by fitting of an appropriate FRAP model to the experimental recovery curve. In many FRAP models—the so-called discontinuous photobleaching techniques—the bleaching phase is assumed to be very short compared to the recovery time. 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 Consequently, diffusion during bleaching can be neglected and the bleaching phase can be solely described by the photochemical bleaching process. In addition, the assumption is usually made that the bleaching process can be described by an irreversible first-order reaction: 11, 12, 13, 15, 16, 17, 18, 21, 22, 23 where is the spatial concentration distribution of fluorophores at a time , is the bleach rate, which is specific for a specific fluorophore in a particular medium, and is the bleaching illumination intensity distribution (BID). Equation 1 is based on the simple reaction scheme of product. If the bleaching beam is stationary, Eq. 1 immediately leads towhere is the fluorophore concentration distribution before bleaching and the fluorophore concentration distribution after bleaching during a time . In case of a geometry being bleached by a scanning beam, which is mostly the case for FRAP experiments on a CSLM, Eq. 1 leads to11 where is the line scanning speed and the distance between two adjacent scanning lines and where it is understood that is smaller than half the resolution of the BID. is the bleaching illumination intensity distribution that results from scanning the area with the BID and can be calculated from their convolution product:After the instantaneous photobleaching phase, the fluorescence inside the bleached region will gradually recover due to diffusion of the bleached fluorophores out of the bleached region and diffusion of unbleached fluorophores from the surroundings into the bleached region. The diffusion process is described by Fick’s second law:24 where is the diffusion coefficient and the spatial concentration distribution of the diffusing substance at a time after bleaching. To model the recovery phase, this differential equation has to be solved for the concentration distribution right after bleaching as calculated from Eq. 2 or 3. Hence, the correctness of the FRAP model depends directly on the validity of assuming first-order kinetics for the photobleaching process. In case first-order kinetics are not valid, the concentration distribution after bleaching will be different from the one calculated from Eq. 2 or 3. Consequently, the FRAP model will no longer be valid, leading to incorrectly calculated values for the diffusion coefficient.A very popular fluorophore for FRAP is fluorescein1, 3 because it photobleaches easily and therefore can comply with the assumption of a very short bleaching phase. It has been shown, however, that the bleaching process of fluorescein is not a simple first-order reaction and in microscopy, in general, it does not follow a single exponential decay.25 It is therefore to be expected that the concentration distribution after bleaching is generally no longer accurately predicted by Eqs. 2, 3 when using fluorescein, especially under the saturating conditions of the photobleaching phase. Here we study the effect of excitation saturation of fluorescein on the concentration distribution after photobleaching. First we will repeat in short the photochemical processes leading to the photobleaching of fluorescein. Then we will examine in detail, both experimentally and theoretically, the concentration distribution of fluorescein after photobleaching with the confocal microscope as a function of the bleaching light intensity. Finally, a discussion will be given on the implications of excitation saturation on various FRAP models. 2.Theoretical Background: Photobleaching of Fluorescein in Microscopy2.1.The Photobleaching Process of FluoresceinA detailed description of the photobleaching kinetics of fluorescein has been published by Song 25 As the photobleaching process in the computer simulations of this study is based on the work of Song, we will repeat the main conclusions here for convenience of the reader. A fluorophore can make an electronic transition from the ground state S to the excited singlet state (see Fig. 1 ) by absorbing a photon of wavelength (m) and corresponding energy , where is Planck’s constant and is the speed of light. If the excitation light has an irradiance and the fluorophore has an absorption cross-section , the rate of photon absorption is The absorption cross-section for fluorescein at and is .26 The number of fluorophores in the excited singlet state will be depopulated via fluorescence emission (rate ), radiationless internal conversion (rate ), and radiationless intersystem crossing to the excited triplet state (rate ). The spin-forbidden transition from to S will happen at a (usually very low) rate . While being in the excited singlet or triplet state, irreversible photochemical destruction of the fluorophores can occur, a process termed photobleaching. Those fluorophores that are photobleached can no longer take part in the fluorescence excitation-emission cycle.For fluorescein it has been demonstrated that the short-lived excited singlet state does ordinarily not contribute to the photobleaching mechanism. The long-lived triplet excited state, at the other hand, becomes depopulated via two major pathways: (1) the reaction between a triplet and another triplet or a ground state fluorescein molecule, and (2) the reaction between a triplet fluorescein molecule and an oxygen molecule. These two mechanisms are termed the dye-to-dye (D-D) and dye-to-oxygen (D-O) mechanisms and can lead to either reduced (R) or semi-oxidized (X) radical forms. These radicals can again react with triplet state molecules and revert either to stable nonfluorescent photoproducts or back to the ground state. Song 25 have incorporated all relevant photochemical reactions into a theoretical model comprising the following six coupled differential equations: The individual photochemical reactions and the corresponding rate constants as reported in Song 25 are shown in Table 1 . This set of differential equations has to be solved numerically for a given rate of photon absorption and initial values for the concentrations , , , , , and . The rate of photon absorption can be calculated from Eq. 6 for a given irradiance . Table 1Photochemical reactions of fluorescein.
Reproduced from Song
25 2.2.Photobleaching of Fluorescein by a Focused Laser BeamWhen a laser beam is focused down through an objective lens, a complex 3-D intensity distribution results around the focal point. For clarity and because most FRAP models are limited to diffusion in two dimensions, the study presented here will be restricted to two dimensions as well (focal plane). The conclusions from this study equally apply to the third dimension as well. The radial irradiance distribution in the focal plane of a (confocal) microscope’s objective lens is in good approximation given by the Fraunhofer–Airy formula, which is valid for small diffraction angles:27, 28, 29 where is the power (W) of the light emerging from the objective lens, is the first-order Bessel function of the first kind, and the resolution of the lens, which is defined as the position of the first minimum. For an objective lens of numerical aperture NA, the resolution can be calculated from the Rayleigh criterion:30 . Figure 2a shows the actual diffraction pattern in the focal plane of the objective lens that will be used in this study (see “Materials”), which is recorded by imaging a small fluorescent nanosphere with a completely opened confocal diaphragm. A logarithmic colormap was applied to make the first- and second-order diffraction rings visible. Since the real diffraction pattern is not perfectly radially symmetric, most likely due to slight imperfections of the microscope’s optical path, we will show throughout this study the intensity profiles along one direction only ( -direction). The intensity distribution along the -direction is shown in Fig. 2a. By analyzing the position of the first minimum, we find that the resolution is approximately . For comparison, the theoretical irradiance distribution according to Eq. 8 is shown in Fig. 2b for . As expected, Eq. 8 adequately represents the diffraction pattern of the focused laser beam at the focal plane.For most fluorescence applications it is sufficient to take the central part of the diffraction pattern into account. For example, the central peak of the Fraunhofer–Airy distribution contains 84% of the total amount of light in the diffraction pattern while the subsequent higher-order maxima have a relative intensity of only 1.7%, 0.4%, 0.1%, etc. Hence, also for FRAP models a more simple Gaussian distribution is ordinarily used to describe the BID: which accurately describes the central peak as is shown by the best fit of Eq. 9 in Figs. 2a and 2b. In that context, the parameter is usually referred to as the resolution of the BID as well, but differs slightly from because it has a different definition [ , as determined by a best fit of Eq. 9 to Eq. 8].FRAP models that explicitly take the BID into account never use the relatively complicated Eq. 8 but rather the simplified Gaussian intensity distribution Eq. 9 2, 16, 17, 18, 21 or possibly a 3-D extension with an additional axial (modified) Gaussian distribution.11, 15, 23 However, as we will show further on, the higher-order diffraction maxima cannot be neglected any longer in case of fluorescein being bleached by a high-intensity laser beam. Under the extreme illumination conditions of the bleaching phase, even the higher-order diffraction maxima will be able to populate the excited triplet state significantly, resulting in a complex expanded bleaching profile that can no longer be explained by simple first-order photobleaching with a Gaussian intensity distribution. 3.Materials and Methods3.1.The Bleaching and Imaging SystemThe FRAP experiments are performed on a CSLM (model MRC1024 UV, Bio-Rad, Hemel Hempstead, UK) equipped with the SCAMPER module for photobleaching experiments.31, 32 The module consists of a Ar-ion laser (model Stabilite 2017; Spectra-Physics, Darmstadt, Germany) and an acousto-optical modulator (AOM) (Brimrose, Campbell, Baltimore, MD, USA), controlled by a computer and dedicated software. The AOM controls the intensity of the laser beam that is sent to the CSLM. A high-power laser beam (“bleaching beam”) is used for photobleaching, while a low-power beam (“detection beam”) is used for the recording of the confocal images. Bleaching masks can be designed in the software that controls the AOM. While recording an image with the CSLM, low- and high-power laser light is sent to the sample according to the selected bleaching mask. All bleaching experiments have been performed with the line from the Ar-ion laser. The intensity of the laser beam at the sample can be controlled over a range of . A objective lens (CFI Plan Apochromat; Nikon, Badhoevedorp, The Netherlands) with a numerical aperture (NA) of 0.45 was used for bleaching. On the Bio-Rad confocal microscope, the back aperture of this lens is only partially filled, however, resulting in a lower effective NA of about 0.2 for the illumination beam and a corresponding resolution (and ). The resolution was determined by directly imaging fluorescent nanospheres (Molecular Probes, Leiden, The Netherlands) with a completely opened confocal diaphragm. Imaging of the bleached profiles was done with a high-resolution water immersion lens of 1.2 NA (CFI Plan Apochromat; Nikon, Badhoevedorp, The Netherlands) having a resolution of approximately . 3.2.Photobleaching ExperimentsFluorescein isothiocyanate dextrans (Sigma-Aldrich, Bornem, Belgium) of molecular weight (FD2000) were incorporated in an acrylamide gel. The acrylamide gel was prepared by radical polymerization of an aqueous acrylamide solution (30% w/w). The solution was prepared by dissolving acrylamide in deoxygenated phosphate buffer ( , 0.02% sodium-azide, adjusted with hydrochloric acid to pH 7.0). Prior to addition of the gelation reagents, gel of FD2000 was added. The polymerization reagents were -tetramethylene-ethylenediamine (TEMED; Fluka, Buchs, Switzerland) and ammonium persulphate (10% w/v) (APS; VWR International, Leuven, Belgium). The procedure involves adding TEMED solution to hydrogel. After mixing, APS solution was added to initiate gelation. The gelating mixture was transferred into a cuvette with a glass bottom for use with an inverted microscope (Nalge Nunc International, Naperville, IL, USA). The gelation required at room temperature. FRAP measurements using the CSLM11 revealed a very low diffusion coefficient of and a large fraction of immobile fluorescent FITC-dextrans of . Bleaching experiments were performed with a stationary laser beam (“beam parked”) or a line scanning beam. The laser power at the sample was between and for imaging and up to for bleaching. The time between the bleaching phase and imaging of the bleached profiles was approximately to allow diffusion of the (small) mobile fraction to complete. After that time, the remaining bleached profile is due to the immobile fraction and reflects directly the bleaching profile immediately after bleaching. Confocal images of the bleaching profiles were recorded with the NA objective lens to obtain the best possible resolution. Before normalizing the confocal images to the background fluorescence (i.e., unbleached fluorophores), they were convolved with a median smoothing mask to reduce the noise. 3.3.Photobleaching simulationsTo study the bleaching of fluorescein at the focal plane, a computer program was written to solve the six coupled differential Eqs. 7 numerically. For the simulation of the spot photobleaching experiments, the absorption rate is calculated from Eqs. 6, 8. At first a set of radial coordinates is defined (the spatial mesh) together with the initial concentrations of , , , , , and . In the simulations reported here, it is always assumed that all fluorescein molecules are initially in the ground state . Next, the differential equations are solved for each of the radial coordinates and the corresponding value of . When the illumination time has passed, in general there still will be a number of fluorescein molecules in the excited singlet and triplet state: & . Therefore, even when the illumination is switched off, the photobleaching reactions still can go on for some time until all the molecules are either bleached or back in the ground state. This has been taken into account in the simulation program by setting to zero after the bleaching time and continue to solve the differential equations until the number of bleached molecules together with the number of ground state molecules are equal to the initial number of molecules. In FRAP experiments on the CSLM, a scanning beam is often used for bleaching rather than a stationary beam. 11, 17, 23, 32, 33, 34, 35, 36 Therefore, the program for the simulation of the photobleaching of fluorescein was extended for a line-scanning beam as well. When bleaching a line segment, the bleaching will be uniform along the scanning direction (except at the beginning and the end of the line). Only in the direction perpendicular to the scanning direction a bleaching gradient will exist. It is therefore sufficient to calculate the bleaching profile perpendicular to the scanning direction for one position on the bleached line segment only. In the simulation algorithm, the line scanning is taken into account by first defining a start and end position of the laser beam ( and ) relative to the position on the line at which the bleaching profile is calculated : the laser beam starts at position and is moved at a scanning speed in discrete steps toward the position . At each step the laser beam is stationary for a time and the effect is calculated at position . The smaller the step size , the better the continuous movement of the laser scanning beam will be approximated (at the cost of an increased calculation time). All line scanning simulations reported here were calculated with a step size that is one fifth of the resolution of the bleaching beam since no appreciable differences in the calculated bleaching profiles were observed for smaller step sizes. The start and end positions of the bleaching beam were always taken symmetrically at 10 times the resolution of the bleaching beam: and . 4.Results4.1.Photobleaching of Immobilized FD2000To visualize the bleaching profiles at different bleaching conditions, FITC-dextrans of molecular weight were incorporated into an acrylamide gel. Using a stationary laser beam, spots were photobleached in the gel with a low NA objective lens for different bleaching times. Confocal images were subsequently recorded using a high NA objective lens to obtain the best possible resolution. Some examples are shown in Fig. 3 . To be able to relate the results of this study directly to the existing FRAP models (see Discussion), we will now define the “apparent BID” as the intensity distribution one would have to take into account to explain the experimentally observed bleaching profile if the bleaching process would follow first-order kinetics. In other words, the “apparent BID” is the intensity distribution that has to be used in Eq. 2 to correctly calculate the observed real bleaching profile. In case first-order photobleaching is valid, should be identical to the intensity distribution as shown in Fig. 2a. In the experiments below we will examine if this is true or not under a variety of photobleaching conditions. If denotes the fluorescence concentration distribution after photobleaching, the normalized “apparent BID” can be calculated from The first spot (Fig. 3a) was bleached at a low intensity of , which is typical for confocal imaging conditions, and a bleaching time of . The apparent BID calculated from Fig. 3a using Eq. 10 is shown in Fig. 4a . A best fit of Eq. 9 shows there is a very good correspondence of the apparent BID with a Gaussian intensity profile, in accordance with what is assumed theoretically in the FRAP models. We note that the bleached spot was not perfectly circular but was stretched along the -direction resulting in a slightly increased Gaussian radius. This is probably caused by a small difference in resolution in and because of the polarized laser light.37 To check the validity of a Gaussian BID for high-intensity laser beams, spots were bleached at and different bleaching times: (image not shown), (image not shown), (Fig. 3b), (Fig. 3c), (Fig. 3d), and (Fig. 3e). A Gaussian apparent BID is still found for very short bleaching pulses of (Fig. 4b) and (Fig. 4c), although a slight broadening of the profile is already observed compared to Fig. 4a. When bleaching for (Fig. 4d), the higher-order diffraction rings are already starting to show up as indicated by the arrows and the corresponding apparent BID starts to deviate from a pure Gaussian distribution. For bleaching times of (Fig. 4e) and (Fig. 4f) the higher-order diffraction rings become much more pronounced, expanding the apparent BID even more. Although the apparent BIDs do no longer correspond to a simple Gaussian distribution, it is worth noting that it is still possible to roughly approximate the apparent BIDs by an (expanded) Gaussian intensity distribution. Figure 3e, finally, shows the situation if the bleaching time is prolonged to . Besides the fourth-order diffraction ring, all kinds of diffraction spots take part in the bleaching process. Fitting of a Gaussian intensity distribution to the apparent BID in this case (data not shown) yielded a highly increased Gaussian resolution of . These experiments clearly show that the bleaching profiles obtained at high laser powers cannot be explained by first-order photobleaching alone. Assuming first-order kinetics leads to an apparently variable BID, which physically is nonsense. In reality, a higher laser power results in a higher amplitude of the BID but cannot change the shape of the distribution. Besides being dependent on the bleaching time, the apparent BID also depends on the bleaching intensity. Since many FRAP techniques make use of a scanning beam for bleaching, a line photobleaching experiment instead of spot photobleaching was chosen here to explicitly show that the results from this study equally apply to a scanning beam as well. Three line segments of are bleached at different bleaching powers, but with an equal total amount of light received. A first line is bleached by scanning two times at . A second one is bleached at by scanning 20 times. A third one is bleached at by scanning 267 times. A confocal image of the right part of each of the line segments is shown in Fig. 5 . The fluorescence intensity profile (along the -direction) and the corresponding apparent BID are shown for each line in Figs. 5a, 5b, 5c. Assuming first-order bleaching kinetics, the bleaching profile of a single line segment [a, b] can be calculated from Eq. 3 with and .11 For a Gaussian BID according to Eq. 9 and a line segment that is much longer than the resolution , the resulting bleaching intensity distribution finally becomes Equation 11 expresses that, when scanning a Gaussian intensity beam along the -direction, the Gaussian intensity distribution is preserved along the -direction. Hence, if the bleaching process follows first-order kinetics, the apparent BID calculated from Eq. 10 should still be Gaussian-distributed with a resolution equal to that of the scanning beam. At a low bleaching intensity, the apparent BID is indeed Gaussian distributed (Fig. 5c), while a deviation from a pure Gaussian distribution occurs for increasing bleaching intensities (Figs. 5b and 5a). Fitting of a Gaussian intensity distribution in Fig. 5 explicitly shows the expansion of the apparent BIDs with increasing bleaching intensity. From the spot photobleaching experiments we already know that the expansion is mainly due to the participation of the higher-order diffraction rings in the non-first-order bleaching process. The individual diffraction maxima are difficult to distinguish, however, because of the scanning process along the -direction. Additionally, it is clear from Fig. 5 that the amount of bleaching is different for each of the three situations as well, although all lines did receive an equal amount of light. Bleaching at high intensities for a short time appears to be less efficient than bleaching at lower intensities for a longer time.4.2.Simulation of Photobleaching Immobilized FluoresceinThe photochemical reactions according to Table 1 and the corresponding mathematical model Eq. 7 can be used to explain the observations of the previous section. We will show that the expansion of the bleaching profiles—which is unexpected from classical FRAP models, which use first-order photobleaching to calculate the bleaching profile—can indeed be explained by the complex photobleaching reactions of fluorescein in combination with the very high laser powers that are used for the bleaching phase of an FRAP experiment. We note that, as most of the FITC-dextrans are immobilized in the acrylamide gel, the contribution of the D-D interactions to the bleaching of fluorescein is expected to be very small and virtually all bleaching will be due to the D-O reactions. This can be taken into account in the simulations by setting the rate constants , , , , , and to zero. The initial concentration of fluorescein in the gel was approximately (which can be calculated from the concentration of FD2000 in the gel and the loading degree of fluorescein on the dextran chains). The initial oxygen concentration is taken to be as well.25 Assuming that all fluorescein molecules are initially in the ground state, the simulations therefore always start from the initial condition: , , and . We simulated the photobleaching of immobilized fluorescein as a function of time first to examine the evolution of the photobleaching process. The solid lines in Fig. 6 are calculated for a bleaching intensity of and a bleaching time of . From on, the ground state S is being depleted to the advantage of the excited singlet state that peaks at around . After , the excited singlet state starts to decay to both the ground state S and the long-lived excited triplet state . Between and , almost all fluorescein molecules are gradually getting trapped in the long-lived excited triplet state . Once the triplet state population starts to rise, the D-O reactions can start producing the photobleaching products X and R, which are formed from on. The formation of the photobleaching products X and R cause the triplet state to be depleted, together with the number of oxygen molecules. At , almost no molecules are left in the ground state because all are either photobleached or in the triplet state. At , the illumination is switched off and the reactions are allowed to continue until all molecules are either photobleached or back in the ground state. The first-order maximum of the BID has a relative intensity of 1.7% (see Fig. 2b). To see what happens at this intensity level, the same calculations were repeated at a bleaching power of (Fig. 6, dashed lines). Essentially the same processes are taking place, albeit somewhat delayed because of the reduced illumination intensity. It is important to note that the triplet state still becomes significantly populated to approximately 30%, despite the very much reduced illumination intensity. As a result, about the same level of bleaching is reached after , which explains why the higher-order diffraction rings are showing up in the bleaching experiments. To see this more clearly, calculations have been performed for a BID according to Eq. 8 (also see Fig. 2b) at a bleaching power of and for different bleaching times. Figure 7a shows the result for . The apparent BID is again calculated from the bleaching profile using Eq. 10. For comparison with the experimental data (Fig. 4b), it should be realized that the simulated bleaching profile is not what is observed directly through a microscope. When observing an object through a microscope, it is modulated by the microscope’s total point spread function (PSF), which depends directly on the objective lens being used.29, 38 The detection PSF of an objective lens in the focal plane of a confocal microscope can be approximated by a Gaussian intensity distribution corresponding to the central diffraction peak. This is allowed because the higher-order diffraction rings only have a small contribution to the collected fluorescence compared to the central peak and because the higher-order diffraction rings are largely blocked by the small confocal apperture (see, e.g., Fig. 2 of Jonkman and Keller29). The objective lens used to image the bleaching profiles has a Gaussian resolution of approximately . To obtain a more direct comparison between the simulated and experimental data, the simulated bleaching profile is therefore convolved with a corresponding Gaussian mask. The result is shown in Fig. 7b. The individual maxima seem less pronounced because of the limited resolving power of the objective lens. Together with the presence of noise, this explains why the individual maxima are not always easy to discern in the experimental apparent BIDs (compare with Fig. 4). The same calculations have been performed for (Fig. 7c), (Fig. 7d), (Fig. 7e), and (Fig. 7f). Only the convolved apparent BIDs together with a Gaussian fit are shown for clarity. Comparing Figs. 4b, 4c, 4d, 4e, 4f with Figs. 7b, 7c, 7d, 7e, 7f shows that the theory corresponds well with what is observed experimentally: an apparent BID that expands with increasing bleaching time due to the increasing contribution to the bleaching process of the higher-order diffraction rings. Again we find that each of the expanded apparent BIDs can roughly be approximated by a Gaussian distribution. It is worth noting that the simulation results not only correspond qualitatively to the experimental observations, but that there is also a reasonable quantitative correspondence between both. The simulation parameters used to obtain the results from Fig. 7 were chosen thus to directly reflect the experimental conditions leading to the results presented in Fig. 4. Figure 7b is in fact the simulation of the experiment in Figs. 4b and 7c is the simulation of 4c and so on (the experiment of Fig. 4a was not simulated). In all cases a Gaussian distribution was fitted to the apparent BID to quantify the expansion through the Gaussian resolution parameter . The values for are reported in each of the figures, from which it can be seen that there is a fairly good correspondence between the experimental and simulated values. The simulated results deviate from the experimental ones only by 19%, 9%, 7%, 12%, and 7% for the Figs. 7b, 7c, 7d, 7e, 7f, respectively. This further supports the validity of the model being used and the accuracy of the computations presented here. The line photobleaching experiments have been simulated as well for bleaching powers of , , and . The simulation program does not allow us to calculate repeated line scans because of the extensive computation times involved. Alternatively we have simulated line profiles using those three intensities but with different scanning speeds. The scanning speed at was of the scanning speed at , and at . Hence, the same amount of light will be received in all three cases. The resulting bleaching profiles are shown in Fig. 8a . In correspondence with the experimental observations, a better bleaching efficiency is obtained when using a low bleaching intensity with an increased bleaching time. This effect can be explained as follows. The high-intensity beam quickly depletes the ground state in favor of the triplet state, which will soon become saturated. The low-intensity beam, on the other hand, will only partly populate the triplet state at any one time, but integrated over the entire time span a larger number of triplet states will have existed, effectively leading to a larger number of photobleached molecules. Although in both cases an identical total amount of light dose was received, the lower light intensity applied for an extended period of time effectively causes more bleaching than the short saturating high-intensity bleach pulse. The apparent BIDs are shown in Fig. 8b. Again we find an expansion of the apparent BID with increasing bleaching intensities. Fitting of a Gaussian distribution to the simulated profiles (not shown in Fig. 8b) gives -values of , , and for the bleaching powers of , , and , respectively. As indicated in Fig. 5, the experimental -values are (in the same order) , , and . Thus we obtain a deviation of the simulated from the experimental results of 13%, 6%, and 0%. Again we find a good quantitative correspondence between the simulated and experimental results. 5.DiscussionBoth theory and experiments confirm that generally the photobleaching of fluorescein does not follow a simple exponential decay. This is especially the case for typical bleaching conditions of a FRAP experiment where high-intensity light levels are used for very fast photobleaching and cause saturation of fluorescein in the triplet excited state. We have demonstrated that under those conditions it is no longer sufficient to solely take the central diffraction spot into account as is assumed in the derivation of many FRAP models. Instead it has become clear that the higher-order diffraction rings can induce a significant amount of photobleaching, mainly due to the saturation of the fluorophore in the long-lived triplet state from which the bleaching actually proceeds. The net result is an apparent bleaching illumination intensity distribution that is expanded with regard to the actual bleaching illumination intensity distribution. In addition, not only does the apparent BID depend on the bleaching intensity, it also depends on the bleaching time or scanning speed in case of a stationary or scanning beam, respectively. We have shown by simulations and experiments that the actual situation is even more complicated. Besides being dependent on the different illumination parameters, the photobleaching of fluorescein is also dependent on multiple photochemical reactions, as summarized in Table 1. The individual reaction rates of the D-D and D-O reactions will, for example, depend on the mobility of the respective molecules in the system. The values reported in Table 1 are for free fluorescein in solution, but may be different when fluorescein is present in actual samples, such as cells or biopolymer gels. As a result, the various reaction rates will depend on the sample under study. It is also likely that in complex chemical systems, such as an intracellular environment, additional reactions will take part in the photobleaching process of fluorescein. Moreover, the bleaching process also depends on the molar concentrations of both fluorescein and oxygen in the system. Considering the variability of these parameters, whose exact values are unknown in most cases, it becomes clear that the apparent BID cannot be accurately predicted for actual experiments. FRAP models almost always assume first-order photobleaching kinetics. Hence the concentration distribution after bleaching can be calculated directly from a single exponential function [Eq. 2 or Eq. 3] and an invariant Gaussian BID [Eq. 9], which is solely determined by the microscope optics. For fluorescein we have demonstrated that, when assuming first-order kinetics, the BID can no longer be considered to be invariant. Instead an “apparent BID” had to be defined whose shape and intensity depends on many parameters. We have also shown that the apparent BID can be roughly approximated by a Gaussian intensity distribution whose resolution will depend on the specific experimental conditions. It is therefore to be expected that existing FRAP models are still able to calculate valid diffusion coefficients from experiments under saturating conditions, as long as the correct apparent Gaussian resolution can be determined for every single experiment. If not, serious errors will result. For example, in the FRAP equation for spot photobleaching with a Gaussian beam in the fundamental work of Axelrod,12 the diffusion coefficient is calculated from fitting the parameter to the experimental recovery curve, where is the Gaussian resolution of the bleaching beam. Hence the resolution has to be known with great precision to obtain an accurate diffusion coefficient. Our experiments have shown that an expansion of the apparent Gaussian radius with a factor 2 to 5 is very well possible. If one does not take this effect into account, the calculated diffusion coefficient would consequently be erroneous by a factor 4 to 25. We note that broadening of the bleach area can also occur when using bleach times that are too long, in which case diffusion will occur during bleaching. Most FRAP models assume an instantaneous photobleaching phase, i.e., no diffusion during bleaching. While long bleach times will lead to the similar result of bleach area broadening, the “anomalous” photobleaching discussed in this study is really of a different nature and is inherent to the fluorophore photobleaching mechanism. Diffusion during bleaching, on the other hand, can simply be avoided by understanding the FRAP model being used and carrying out the experiments accordingly. There are a few possibilities to deal with a variable apparent BID. For FRAP models that explicitly require the knowledge of the Gaussian resolution , the most direct solution in this case is to extract directly from the experiments, which can be done by acquiring an image of the bleaching profile immediately after the bleaching phase (which is independent of ) and using the FRAP formula to extract . In practice it is difficult, however, to obtain an image immediately after bleaching because of limited image acquisition rates. Even then most FRAP models would need an adjustment of the theory to take a different bleaching and illumination distribution into account. A workaround is to use FRAP models that do not require the precise knowledge of the BID or the concentration distribution after bleaching. Examples are the models that use the Fourier transform of the recovery images.39, 40, 41 They have shown that the diffusion coefficient can be calculated without determining the BID of the microscope by measuring the decay of the spatial Fourier transform of the recovery images. Another example is the model of Kubitscheck, 33 who have derived a moment’s analysis of the fluorescence recovery images that does not require the knowledge of the initial concentration distribution after bleaching. The results from this study suggest that methods using a relative determination of the diffusion coefficient3, 42 should be used with care in case of fluorescent dyes with a complex bleaching mechanism, such as fluorescein. Such a method relies on two series of FRAP experiments, one in the actual sample and one in a series of standard solutions of known viscosity. If both FRAP experiments are performed under exactly the same conditions, the diffusion coefficient in the sample can be calculated relative to the calibration reference solutions without knowing the actual bleaching profile. This approach could lead to invalid results because the reaction rates, and hence the apparent BID, depend directly on the mobility of the molecules that take part in the bleaching process. For example, in a gel system with small interstitial spaces, larger molecules such as FITC-dextrans can experience a strongly reduced diffusion coefficient due to sterical hindrance, while the smaller oxygen molecules can still diffuse as rapidly as in solution. Consequently, the dye-dye reaction constants (see Table 1) will have a lower value in the gel system as compared to in solution, leading to a change in the photobleaching process and hence in the apparent BID. Therefore, even under the exact same illumination conditions, the bleaching profile in the sample will, in general, be different from the one in solution, and consequently a general relation between the two recovery curves may not be possible any more. An alternative is to use a FRAP model where an area is uniformly bleached that is much larger than the characteristic dimensions of the BID. Recently, we derived such a model for use with the confocal microscope based on the bleaching of a uniform disk having a radius of at least 4 times the Gaussian resolution of the apparent BID.11 In that case, a change of the shape of the BID will only be “felt” at the boundaries of the bleached area. We have shown that neglecting the expansion of the BID usually will not cause errors larger than 10% in the determination of the diffusion coefficient. Hence this method can be used with great confidence and does not require special measures. The most straightforward solution is, of course, to avoid fluorescent dyes that easily saturate. From experience we know that one such fluorophore is NODD (N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)diethyl amine). NODD is incorporated in polystyrene microspheres that are encoded by photobleaching.31, 43 In this application it is essential to have the best possible resolution (i.e., a non-expanded BID) to write as much information as possible per unit area. It needs to be examined, though, which other fluorophores for FRAP meet this condition as well. This could be done, for example, by performing FRAP experiments on a series of samples with a known diffusion coefficient. If the diffusion coefficient is known, the apparent Gaussian resolution for each of the solutions can be calculated from the corresponding FRAP formula. If remains constant for all samples, the apparent BID is invariant for that particular fluorophore and is suitable for use with the existing FRAP models. We are currently developing a new FRAP model for use with the confocal microscope based on the bleaching of a line segment taking both an independent bleaching and imaging illumination intensity distribution into account. This model will be very well suited to characterize the expansion of the apparent BID—in terms of the Gaussian resolution —as a function of the bleaching intensity and duration for different kinds of fluorophores. It would especially be interesting to do this for GFP and its variants, which are popular fluorophores for intracellular FRAP experiments. Finally, we note that our bleaching experiments and simulations have shown that the change in shape of the BID is small when using a low-power bleaching beam (in combination with an extended bleaching time). Because most FRAP models assume a very short bleaching phase compared to the characteristic recovery time , this approach will only offer a solution in case of sufficiently slow diffusion. 6.ConclusionsIn this study we have examined the effect of excitation saturation on FRAP experiments. As a case study we have looked in detail at the photochemical bleaching reactions of fluorescein, one of the most frequently used fluorophores for FRAP. We have presented both experimental and theoretical evidence that, mainly due to saturation of the triplet excited state, the basic assumptions that are made in many FRAP models are generally no longer valid: an invariant bleaching illumination intensity distribution in combination with first-order photobleaching kinetics. Instead, saturation leads to a dramatically variable apparent bleaching illumination intensity distribution. This effect is not always immediately evident from the experiments, but neglecting it can induce errors of more than one order of magnitude in the calculation of the diffusion coefficient. Therefore, it seems advisable to look for photobleachable fluorophores that do not as easily saturate or to use/develop FRAP models that are insensitive to the actual shape of the bleaching illumination intensity distribution. AcknowledgmentsThe financial support of the IWT is acknowledged with gratitude. Ghent University (BOF) is acknowledged for its support through instrumentation credits. Kevin Braeckmans is a Postdoctoral Fellow of the Fund for Scientific Research–Flanders (Belgium). ReferencesT. K. L. Meyvis,
S. C. De Smedt,
P. Van Oostveldt, and
J. Demeester,
“Fluorescence recovery after photobleaching: A versatile tool for mobility and interaction measurements in pharmaceutical research,”
Pharm. Res., 16
(8), 1153
–1162
(1999). https://doi.org/10.1023/A:1011924909138 0724-8741 Google Scholar
J. G. McNally and
H. Smith,
“Photobleaching by confocal microscopy,”
Confocal and Two-Photon Microscopy, 525
–538 Wiley-Liss, New York
(2002). Google Scholar
A. S. Verkman,
“Diffusion in cells measured by fluorescence recovery after photobleaching,”
Biophotonics, Part A: Methods in Enzymology, 635
–648 Academic Press, New York
(2003). Google Scholar
M. Edidin,
“Translational diffusion of membrane proteins,”
The Structure of Biological Membranes, 539
–572 CRC Press, Boca Raton, Fla
(1992). Google Scholar
A. Ishihara and
K. Jacobson,
“A closer look at how membrane-proteins move,”
Biophys. J., 65
(5), 1754
–1755
(1993). 0006-3495 Google Scholar
O. Seksek,
J. Biwersi, and
A. S. Verkman,
“Translational diffusion of macromolecule-sized solutes in cytoplasm and nucleus,”
J. Cell Biol., 138
(1), 131
–142
(1997). https://doi.org/10.1083/jcb.138.1.131 0021-9525 Google Scholar
F. Umenishi,
J. M. Verbavatz, and
A. S. Verkman,
“cAMP regulated membrane diffusion of a green fluorescent protein-aquaporin 2 chimera,”
Biophys. J., 78
(2), 1024
–1035
(2000). 0006-3495 Google Scholar
W. M. Saltzman,
M. L. Radomsky,
K. J. Whaley, and
R. A. Cone,
“Antibody diffusion in human cervical-mucus,”
Biophys. J., 66
(2), 508
–515
(1994). 0006-3495 Google Scholar
S. S. Olmsted,
J. L. Padgett,
A. I. Yudin,
K. J. Whaley,
T. R. Moench, and
R. A. Cone,
“Diffusion of macromolecules and virus-like particles in human cervical mucus,”
Biophys. J., 81
(4), 1930
–1937
(2001). 0006-3495 Google Scholar
N. N. Sanders,
S. C. De Smedt, and
J. Demeester,
“The physical properties of biogels and their permeability for macromolecular drugs and colloidal drug carriers,”
J. Pharm. Sci., 89
(7), 835
–849
(2000). 0022-3549 Google Scholar
K. Braeckmans,
L. Peeters,
N. N. Sanders,
S. C. De Smedt, and
J. Demeester,
“Three-dimensional fluorescence recovery after photobleaching with the confocal scanning laser microscope,”
Biophys. J., 85
(4), 2240
–2252
(2003). 0006-3495 Google Scholar
D. Axelrod,
D. E. Koppel,
J. Schlessinger,
E. Elson, and
W. W. Webb,
“Mobility measurement by analysis of fluorescence photobleaching recovery kinetics,”
Biophys. J., 16
(9), 1055
–1069
(1976). 0006-3495 Google Scholar
D. M. Soumpasis,
“Theoretical-analysis of fluorescence photobleaching recovery experiments,”
Biophys. J., 41
(1), 95
–97
(1983). 0006-3495 Google Scholar
A. Lopez,
L. Dupou,
A. Altibelli,
J. Trotard, and
J. F. Tocanne,
“Fluorescence recovery after photobleaching (FRAP) experiments under conditions of uniform disk illumination—critical comparison of analytical solutions, and a new mathematical method for calculation of diffusion coefficient-D,”
Biophys. J., 53
(6), 963
–970
(1988). 0006-3495 Google Scholar
J. C. G. Blonk,
A. Don,
H. Van Aalst, and
J. J. Birmingham,
“Fluorescence photobleaching recovery in the confocal scanning light-microscope,”
J. Microsc., 169 363
–374
(1993). 0022-2720 Google Scholar
G. W. Gordon,
B. Chazotte,
X. F. Wang, and
B. Herman,
“Analysis of simulated and experimental fluorescence recovery after photobleaching—data for 2 diffusing components,”
Biophys. J., 68
(3), 766
–778
(1995). 0006-3495 Google Scholar
P. Wedekind,
U. Kubitscheck,
O. Heinrich, and
R. Peters,
“Line-scanning microphotolysis for diffraction-limited measurements of lateral diffusion,”
Biophys. J., 71
(3), 1621
–1632
(1996). 0006-3495 Google Scholar
T. J. Feder,
I. BrustMascher,
J. P. Slattery,
B. Baird, and
W. W. Webb,
“Constrained diffusion or immobile fraction on cell surfaces: A new interpretation,”
Biophys. J., 70
(6), 2767
–2773
(1996). 0006-3495 Google Scholar
Y. Cheng,
R. K. Prud’homme, and
J. L. Thomas,
“Diffusion of mesoscopic probes in aqueous polymer solutions measured by fluorescence recovery after photobleaching,”
Macromolecules, 35
(21), 8111
–8121
(2002). https://doi.org/10.1021/ma0107758 0024-9297 Google Scholar
G. Carrero,
D. McDonald,
E. Crawford,
G. de Vries, and
M. J. Hendzel,
“Using FRAP and mathematical modeling to determine the in vivo kinetics of nuclear proteins,”
Methods, 29
(1), 14
–28
(2003). 1046-2023 Google Scholar
L. S. Cutts,
P. A. Robberts,
J. Adler,
M. C. Davies, and
C. D. Melia,
“Determination of localized diffusion-coefficients in gels using confocal scanning laser microscopy,”
J. Microsc., 180 131
–139
(1995). 0022-2720 Google Scholar
C. Dietrich,
R. Merkel, and
R. Tampe,
“Diffusion measurement of fluorescence-labeled amphiphilic molecules with a standard fluorescence microscope,”
Biophys. J., 72
(4), 1701
–1710
(1997). 0006-3495 Google Scholar
U. Kubitscheck,
P. Wedekind, and
R. Peters,
“Three-dimensional diffusion measurements by scanning microphotolysis,”
J. Microsc., 192 126
–138
(1998). https://doi.org/10.1046/j.1365-2818.1998.00406.x 0022-2720 Google Scholar
J. Crank, The Mathematics of Diffusion, Clarendon Press, Oxford, UK
(1975). Google Scholar
L. L. Song,
E. J. Hennink,
I. T. Young, and
H. J. Tanke,
“Photobleaching kinetics of fluorescein in quantitative fluorescence microscopy,”
Biophys. J., 68
(6), 2588
–2600
(1995). 0006-3495 Google Scholar
R. Y. Tsien and
A. Waggoner,
“Fluorophores for confocal microscopy,”
Handbook of Biological Confocal Microscopy, 267
–279 Plenum Press, New York
(1995). Google Scholar
K. D. Mielenz,
“On the diffraction limit for lensless imaging,”
J. Res. Natl. Inst. Stand. Technol., 104
(5), 479
–485
(1999). 1044-677X Google Scholar
M. Born and
E. Wolf, Principles of Optics, Cambridge University Press, Cambridge, UK
(1999). Google Scholar
J. E. N. Jonkman and
E. H. K. Stelzer,
“Resolution and contrast in confocal and two-photon microscopy,”
Confocal and Two-Photon Microscopy, 101
–125 Wiley-Liss, New York
(2002). Google Scholar
S. Inoué,
“Foundations of confocal scanned imaging in light microscopy,”
Handbook of Biological Confocal Microscopy, 1
–17 Plenum Press, New York
(1995). Google Scholar
K. Braeckmans,
S. C. De Smedt,
C. Roelant,
M. Leblans,
R. Pauwels, and
J. Demeester,
“Encoding microcarriers by spatial selective photobleaching,”
Nat. Mater., 2
(3), 169
–173
(2003). 1476-1122 Google Scholar
P. Wedekind,
U. Kubitscheck, and
R. Peters,
“Scanning microphotolysis—A new photobleaching technique based on fast intensity modulation of a scanned laser-beam and confocal imaging,”
J. Microsc., 176 23
–33
(1994). 0022-2720 Google Scholar
U. Kubitscheck,
P. Wedekind, and
R. Peters,
“Lateral diffusion measurement at high-spatial-resolution by scanning microphotolysis in a confocal microscope,”
Biophys. J., 67
(3), 948
–956
(1994). 0006-3495 Google Scholar
P. Gribbon and
T. E. Hardingham,
“Macromolecular diffusion of biological polymers measured by confocal fluorescence recovery after photobleaching,”
Biophys. J., 75
(2), 1032
–1039
(1998). 0006-3495 Google Scholar
Y. Cheng,
R. K. Prud’homme, and
J. L. Thomas,
“Diffusion of mesoscopic probes in aqueous polymer solutions measured by fluorescence recovery after photobleaching,”
Macromolecules, 35
(21), 8111
–8121
(2002). https://doi.org/10.1021/ma0107758 0024-9297 Google Scholar
G. Carrero,
D. McDonald,
E. Crawford,
G. de Vries, and
M. J. Hendzel,
“Using FRAP and mathematical modeling to determine the in vivo kinetics of nuclear proteins,”
Methods, 29
(1), 14
–28
(2003). 1046-2023 Google Scholar
C. J. Cogswell and
K. G. Larkin,
“The specimen illumination path and its effect on image quality,”
Handbook of Biological Confocal Microscopy, 127
–137 Plenum Press, New York
(1995). Google Scholar
P. J. Shaw,
“Comparison of wide-field deconvolution and confocal microscopy for 3D imaging,”
Handbook of Biological Confocal Microscopy, 373
–387 Plenum Press, New York
(1995). Google Scholar
T. T. Tsay and
K. A. Jacobson,
“Spatial Fourier-analysis of video photobleaching measurements—principles and optimization,”
Biophys. J., 60
(2), 360
–368
(1991). 0006-3495 Google Scholar
D. A. Berk,
F. Yuan,
M. Leunig, and
R. K. Jain,
“Fluorescence photobleaching with spatial Fourier-analysis—measurement of diffusion in light-scattering media,”
Biophys. J., 65
(6), 2428
–2436
(1993). 0006-3495 Google Scholar
A. E. Cowan,
D. E. Koppel,
B. Setlow, and
P. Setlow,
“A soluble protein is immobile in dormant spores of Bacillus subtilis but is mobile in germinated spores: Implications for spore dormancy,”
Proc. Natl. Acad. Sci. U.S.A., 100
(7), 4209
–4214
(2003). https://doi.org/10.1073/pnas.0636762100 0027-8424 Google Scholar
H. P. Kao,
J. R. Abney, and
A. S. Verkman,
“Determinants of the translational mobility of a small solute in cell cytoplasm,”
J. Cell Biol., 120
(1), 175
–184
(1993). https://doi.org/10.1083/jcb.120.1.175 0021-9525 Google Scholar
K. Braeckmans,
S. C. De Smedt,
M. Leblans,
R. Pauwels, and
J. Demeester,
“Encoding microcarriers: Present and future technologies,”
Nat. Rev. Drug Discovery, 1
(6), 447
–456
(2002). 1474-1776 Google Scholar
|