1.

Introduction

The study of the intracellular mobility of proteins and other molecules is attracting increasing attention from biophysical and life science researchers, since molecular mobility is in many cases related to molecular functions and interactions and also provides a useful tool for assaying the physical properties of the intracellular environment.¹ The use of fluorescent proteins as intracellular reporters has dramatically advanced our ability to study molecular dynamics and interactions in live cells. A variety of fluorescence imaging and spectroscopy techniques are now routinely applied to measure molecular mobility and dynamics in live cells, such as photobleaching methods [fluorescence recovery after photobleaching (FRAP), inverse fluorescence recovery after photobleaching (iFRAP), fluorescence loss in photobleaching (FLIP)], particle tracking, and fluorescence correlation spectroscopy (FCS).^{2, 3} Fluorescent proteins are particularly useful for studying protein dynamics and interactions in live cells, because the gene for fluorescent proteins can be fused to DNA for the protein of interest and conveniently transfected into cells using standard protocols. The protein of interest can then be monitored by fluorescence imaging and spectroscopy techniques without exogenous labeling as the cells manufacture the fusion protein with the fluorescent reporter. Among many available fluorescent proteins, the enhanced green fluorescent protein (EGFP) is one of the most commonly used in live cell applications to study intracellular protein dynamics due to its high molecular brightness and good photostability. ^{4, 5, 6, 7}

Although expressing fluorescent fusion proteins is extremely useful and provides a versatile experimental tool, sometimes it is still advantageous or necessary to load external macromolecules into living cells. For example, some methods such as dual-color cross-correlation spectroscopy (FCCS), used to measure molecular interactions, require the use of multiple spectrally distinct fluorophores. Even though there are now many spectrally variant fluorescent proteins available, the choices are still limited for red emitters that are sufficiently bright and spectrally separated for dual-color fluctuation spectroscopy measurements. Therefore, red inorganic dyes are still often selected to pair with a green or blue fluorescent protein in dual-color fluctuation spectroscopy or even multicolor imaging applications.⁸ In addition, specific biological applications may require the introduction of other exogenous molecules such as purified peptides, oligonucleotides, or drugs, since such molecules cannot be transfected and expressed by cells. For some proteins, the relatively large size of the fluorescent proteins may also inhibit molecular function of the fusion protein, so other fluorescence labeling methods may be preferable. Finally, in the case of mobility studies where the complex disordered intracellular environment may have size-dependent diffusion behaviors, it can be important to measure molecular mobility without the $\sim 28\mathrm{kD}$ fluorescent fusion protein attached. So far it has been unclear to what extent the introduction of exogenous fluorescent reporters into living cells by membrane permeabilization will influence the observed molecular mobility through associated changes in the intracellular environment. A quantitative comparison of molecular mobility for proteins expressed in living cells versus those loaded by membrane permeabilization is the goal of these studies.

There are many ways to achieve cell permeabilization, and several methods are reviewed in reference.⁹ Common methods include physical approaches to create membrane wounds such as electroporation, laser-induced microporation, loading by scraping, syringe loading, and chemical approaches to form pores on the membrane such as using lipsomes, bacterial toxins, or signal peptides.⁹ Among the chemical methods, the use of streptolysin O (SLO), a bacterial toxin, has been demonstrated to achieve highly efficient and reversible permeabiliation as well as high cell viability.¹⁰ Cell viability is an important indication of the integrity of the cellular functions after the permeabilization treatment, yet viability itself may not be enough to detect subtle differences that the treatment induces. For example, when molecular mobility is of interest, it is not clear a priori whether or not the permeabilization will significantly alter the intracellular environment in a way that modifies molecular mobility in the cells, and if so, to what extent. To make quantitative comparisons between mobility measurements with different fluorescence labeling methods, i.e., expression versus loading, it is essential to investigate the effects of the labeling method on the observed mobility. Thus, we measured and compared the mobility of the fluorescent protein EGFP in live human embryonic kidney (HEK) cells using three different conditions: (1) EGFP expression, (2) loading purified EGFP following SLO treatment, or (3) EGFP expression but measured following SLO treatment. These three conditions allowed a thorough evaluation to determine whether quantitative comparisons between mobility measurements made using the various approaches are directly comparable or not.

All measurement were made using FCS, which has become an important tool for measuring the mobility of proteins and other macromolecules in live cells and other complex media. ^{11, 12, 13, 14, 15, 16} Some advantages of FCS include its high sensitivity, capability to measure dynamics at physiologically relevant low concentrations, and ability to make measurements with minimal cellular perturbation and low phototoxicity. FCS can be used to measure local concentration and molecular mobility as well as chemical or physical kinetic rates through the analysis of the autocorrelation function for the fluctuating fluorescence signal. ^{11, 12, 13, 14, 15, 16} The diffusion of macromolecules in live cells is commonly observed to be anomalous, presumably due to the complexity of the intracellular environment in which mobility can be hindered by various factors such as molecular crowding, transient nonspecific binding interactions between diffusing macromolecules and cellular components, and interactions with other molecules or obstacles. ^{17, 18, 19, 20} Therefore, the measured FCS data were fitted using an anomalous diffusion model,^{18, 21} and the results were used to evaluate how the SLO permeabilization treatment influenced the intracellular mobility of the cellular proteins.

2.

Methods

3.

Results and Discussion

We measured the intracellular mobility of EGFP expressed by HEK cells, the mobility of purified EGFP loaded into HEK cells using SLO permeabilization, and the mobility of the expressed EGFP after SLO treatment. Fluorescence imaging was performed before the FCS measurement to identify the cells of interest and to select the cellular regions for FCS measurements. Figure 1 shows sample images of cells expressing EGFP [1(a)], cells with successful EGFP loading by SLO permeabilization [1(b)], and cells expressing EGFP following SLO treatment [1(c) and 1(d)]. With the SLO treatment condition described in the methods section, about 50% of the cells were permeabilized, and over 90% of the permeabilized cells are viable after the treatment. A higher loading efficiency could be achieved if desired with a further optimized SLO dosage. For the experiment with loading of purified EGFP into HEK cells by SLO, it was straightforward to select the loaded cells because nonpermeabilized cells remain dark and were ignored for the FCS measurements [Fig. 1b]. For the case of cells expressing EGFP measured after SLO permeabilization treatment, it was not easy to distinguish successful and nonsuccessful permeabilization from the EGFP signal because there was no distinct difference between the two cases. To correctly identify permeabilized cells, we add Alexa 633-NLS, a red dye conjugated to a peptide coding a NLS, to the SLO treating medium. Since this peptide-dye conjugate was not cell permeable, its intake to the cells would indicate successful pore generation upon SLO treatment. The EGFP signal had some spectral bleedthrough into the red imaging channel, so images could show a red channel signal even if the cell was not permeabilized. However, the NLS peptide directed these red labeled molecules to the cell nucleus where they were concentrated, providing a straightforward and simple marker for SLO permeabilization. In Figs. 1c and 1d, cells show signal in both the green and red channels with a bright red cell nucleus, indicating successful permeabilization to the cells expressing EGFP. The two cells at the bottom of the image appearing in the red channel [Fig. 1d] but not appearing in the green channel [Fig. 1c] represent successful permeabilization of cells that were not expressing EGFP. One example of a cell expressing EGFP but with unsuccessful permeabilization is highlighted with a circle. Cells not expressing EGFP or not permeabilized were not included in the mobility measurement for the expressed EGFP with permeabilization treatment. We note that no Alexa 633 emission was detected in the EGFP detector channel, so this strategy for detecting successful SLO permeabilization did not distort FCS measurements of the intracellular EGFP mobility.

Fig. 1

Images of HEK cells (a) expressing EGFP or (b) loaded with EGFP by SLO treatment (b). (c) and (d) show cells expressing EGFP after SLO treatment and loading with Alexa 633-NLS. The green EGFP signal is shown in (c) and the red Alexa 633 signal is shown in (d). Most cells show Alexa 633-NLS intake, recognizable by the bright nuclear signal in (d), indicating successful SLO peameabilization, while the blue circles highlight an example of a cell with unsuccessful SLO permeabilization. (Color online only).

The imaging software allows computer mouse selection of points of interest in the imaged cells, and then a motorized stage moves the selected cellular points of interest to the focused laser beam for FCS measurements. For FCS experiments with each type of EGFP loading or expression, 10 to 15 cells are measured, and a total of 40 to 50 point measurements are taken. About half of the measurements are taken in the cytoplasm and half taken in the cell nucleus. For each point FCS measurement, four 30-second data runs were acquired. These four runs were then averaged and used to compute standard deviations.²⁴ Data were fit using an anomalous diffusion fitting function based on the published model,^{18, 21} with

We found that mobility measurements acquired from any single point within a cell were highly repeatable. For a collection of diffusion measurements from many points, we observe the expected biological variation in local mobility when analyzing measurements from different individual cells, but we found that the average diffusion coefficient and anomalous factor determined for a large number of point measurements were consistent from day to day. Within experimental error, the average diffusion parameters were consistently repeatable for the different labeling protocols, and the observed patterns and variations were statistically significant. Summaries of measured mobility for the cellular EGFP protein with all three different preparations are in Fig. 2 and Table 1 .

Fig. 2

(a) Anomalous factor $\alpha$ , and (b) apparent diffusion coefficient $D_{app}$ for EGFP expressed, purified EGFP loaded into cells by SLO, and EGFP expressed in cells after SLO treatment. Error bars indicate one standard error.

Table 1

Measured values for the anomalous factor α and the apparent diffusion coefficient Dapp with one standard error for expressed EGFP, purified EGFP loaded into cells by SLO, and expressed EGFP following SLO treatment.

3.2.

Mobility of Proteins upon SLO Permeabilization

After measuring the intracellular mobility of EGFP using in vitro expression of the EGFP, we then examined the mobility of EGFP proteins upon SLO permeabilization. The measured mobility data for the loaded EGFP and the expressed EGFP with SLO treatment are shown in Fig. 2 and Table 1. The measured mobility of the loaded EGFP is $15.3\pm 2.1\mathrm{\mu }{\mathrm{m}}^2∕\mathrm{s}$ and $14.2\pm 1.6\mathrm{\mu }{\mathrm{m}}^2∕\mathrm{s}$ in the cytoplasm and cell nucleus, respectively. These results show that the apparent diffusion coefficient of the loaded EGFP is indistinguishable from that of the expressed EGFP in both cellular compartments. For the loaded EGFP, the anomalous factor is slightly lower, i.e., slightly more anomalous, than for the expressed EGFP, perhaps suggesting some subtle variation in the intracellular environment induced by permeabilization. SLO treated cells also show a slightly larger standard error than expressed cells. However, on the whole the highly similar mobility measurements for EGFP expression and EGFP loading suggest that intracellular mobility measurements are not dramatically altered by exogenous loading, and that such results can indeed be compared quantitatively with measurements of expressed proteins. This finding will be useful in future investigations of intracellular dynamics that require some combination of expressed and loaded molecules.

When pores are created following membrane peameabilization, cellular contents can leak into the surrounding medium through the pores. We performed an additional control measurement to identify how the mobility changes when EGFP is expressed but cells are still treated with SLO (no exogenous EGFP protein loaded). We observed that after cells expressing EGFP underwent SLO treatment, the measured mobility of EGFP became significantly slower while the anomalous factor remained the same as was measured without SLO treatment. The apparent diffusion coefficient went down to $8.9\pm 1.5$ and $10.3\pm 1.3\mathrm{\mu }{\mathrm{m}}^2∕\mathrm{s}$ in the cytoplasm and cell nucleus, respectively. This somewhat surprising result, which at first appears to contradict the results shown above, can be explained after considering that the average fluorescence intensity of the expressed EGFP dropped significantly following the SLO treatment, with an average intensity decrease to around 40% of the original signal and individual cells with intensity decreases ranging from 10% to 80%. This variation presumably depends on the number of membrane pores induced in any given cell during the SLO treatment. The cells without successful permeabilization showed only minimal or no decrease in fluorescence intensity, and in all cases less than a 5% decrease was observed. The reduced intensity from SLO treated cells thus indicates that EGFP molecules are leaking from the cell into the surrounding media. The reduced diffusion coefficients following protein leaking seems to indicate that the most mobile EGFP molecules escaped from the cell, while a small subpopulation of EGFP molecules were for unknown reasons either trapped or confined in environments where they diffused more slowly. To further test this hypothesis, we examined how the measured diffusion coefficient of EGFP after SLO treatment correlated with the extent of membrane permeabilization, which was estimated by measuring the Alexa 633-NLS signal levels. We assumed brighter red signals indicated more permeabilized membranes and more protein leakage, which would thus predict slower mobility as well. As shown in Fig. 3 , this prediction was indeed observed in the cell cytoplasm, with higher Alexa 633 signals correlaed with reduced cytoplasmic mobility of the expressed EGFP. This is consistent with our hypothesis that the reduced mobility of EDFP after SLO treatment was indeed due to the leaking of more mobile EGFP through the SLO pores to the surrounding medium, resulting in an enhanced population of the less-mobile EGFP molecules.

Fig. 3

Apparent diffusion coefficient of the EGFP expressed in HEK cells after SLO treatment varied with the fluorescene intensity of loaded Alexa 633-NLS. The measured diffusion rate descreases with increasing Alexa 633-NLS intake. This observation is consistent with our explanation that the reduction in measured mobility is due to the loss of mobile EGFP from the cell to the surrounding media since increasing Alexa 633-NLS signal also indicates more permeabilization of the membrane.

By combining the intracellular mobility results for all three types of EGFP preparations, the scenario became clear. First, there existed both the main relatively fast-moving EGFP population and a small population of less-mobile EGFP molecules in the cells expressing EGFP. We do not detect the slower-moving population in cells expressing EGFP or loaded with exogenous EGFP, presumably because it was only a small fraction of the total protein in the cell and the diffusion coefficients were not sufficiently distinct to resolve two diffusing components. When pores are formed on a cell membrane through SLO permeabilization treatment, proteins up to $\sim 100\mathrm{kD}$ can leak through the pores, which can be as big as $35\mathrm{nm}$ in diameter.¹⁰ Cells treated with SLO without adding additional EGFP then undergo enhancement of the slow-moving population, since the faster-moving population leaks out of the cell more efficiently. This finding is also consistent with the dramatically reduced average EGFP signal in these cells. The observation that EGFP diffusion rates were the same for the expressed and loaded EGFP seems to suggest that the intracellular environment was not drastically different following SLO treatment, at least as seen by a $\sim 30\mathrm{kD}$ diffuser. However, we cannot rule out the possibility that some of the slow-diffusing population following SLO treatment was indeed related to some intracellular changes induced by the SLO permeabilization.

The somewhat surprising result that the mobility of the expressed EGFP without treatment had similar mobility to the loaded EGFP is relevent to the discussion of the role of molecular crowding in intracellular diffusion rates.^{1, 26, 27} In particular, as discussed above, the SLO treatment resulted in a significant loss of soluble protein from the cells to the surrounding media. While we do not have good quantitative indicators of crowding for these measurements, it seems reasonable to assume that with the loss of soluble proteins the intracellular space should become less crowded and molecules should diffuse more rapidly. That we did not observe an increase in diffusion rates following SLO treatment seems to indicate that molecular crowding was not the major determinant of the observed intracellular mobility. More systematic studies of this effect for different size proteins and with some direct measure of the intracellular crowding would be required to confirm this conclusion.

4.

Conclusion

We have shown that expressed and loaded EGFP have essentially the same measured intracellular mobility, and therefore, cell loading may be used as an experimental tool even when molecular mobility measurements are of interest. One caveat based on the above discussion is that if one is interested in the mobility of an expressed protein following SLO treatment, caution is needed in interpreting the data because the remaining molecular populations following treatment may be enhanced for less-mobile fractions. This need for caution is further confirmed by the observation that the measured apparent diffusion coefficient for the expressed EGFP with SLO treatment varies more from cell to cell as indicated by the larger standard error for this set of measurements (Fig. 2 and Table 1). Again, this is probably due to the variation in the number of membrane pores created by SLO treatment and the variation in the recovery process, with the degree of selection for less-mobile fractions influenced by the number of pores. Thus, if one is interested in the mobility of an expressed protein following SLO introduction of an additional species, careful control measurements will be required to find conditions where the mobility of the expressed protein is not dramatically altered.

In the context of fluorescence fluctuation spectroscopy, one of the main motivations for exogenous loading of fluorescent molecules is to use dual-color FCCS to measure molecular interactions within living cells. Our findings suggest that SLO loading may indeed provide an effective approach to introduce a red labeled molecule to complement a green EGFP fusion protein for FCCS measurements. Caution will be required in interpreting such measurements to account for possible changes in mobility and the possible selective enhancement to the population of less-mobile molecules, which could have some influence on the extent of molecular interactions. Thus, FCCS measurements with exogenous loading of red fluorescent interaction partners will require careful analysis of the variations in mobility and concentration of the EGFP fusion protein and selection of loading conditions that produce minimal variation in these parameters.