Dynamics of green fluorescent protein mutant2 in solution, on spin-coated glasses, and encapsulated in wet silica gels

Abstract

Single-molecule experiments are performed by investigating spectroscopic properties of molecules either diffusing in and out of the observation volume or fixed in space by different immobilization procedures. To evaluate the effect of immobilization methods on the structural and dynamic properties of proteins, a highly fluorescent mutant of the green fluorescent protein, GFPmut2, was spectroscopically characterized in bulk solutions, dispersed on etched glasses, and encapsulated in wet, nanoporous silica gels. The emission spectrum, the fluorescence lifetimes, the anisotropy, and the rotational correlation time of GFPmut2, encapsulated in silica gels, are very similar to those obtained in solution. This finding indicates that the gel matrix does not alter the protein conformation and dynamics. In contrast, the fluorescence lifetimes of GFPmut2 on glasses are two-to fourfold higher and the fluorescence anisotropy decays yield almost no phase shifts. This indicates that the interaction of the protein with the bare glass surface induces a significant structural perturbation and severely restricts the rotational motion. Single molecules of GFPmut2 on glasses or in silica gels, identified by confocal image analysis, show a significant stability to illumination with bleaching times of the order of 90 and 60 sec, respectively. Overall, these data indicate that silica gels represent an ideal matrix for following biologically relevant events at a single molecule level.

The green fluorescent protein (GFP) was discovered in the early 1960s (Shimomura et al. 1962), but only recently it has sparked a lot of interest as a biological tool to monitor complex cellular processes (Chalfie et al. 1994; Cubitt et al. 1995; Heim and Tsien 1996; Chalfie and Kain 1998). The chromophore that confers the typical green color and fluorescent properties to the protein is a p-hydroxybenzylideneimidazole, originated from an internal cyclization at residues Ser65, Tyr66, and Gly67, and 1,2 dehydrogenation of Tyr66 (Cubitt et al. 1995). The three-dimensional structure of the WT GFP and several mutants have been detemined (Ormo et al. 1996; Yang et al. 1996; Brejc et al. 1997; Palm et al. 1997; Wachter et al. 1998; Phillips 1997; Battistutta et al. 2000). The protein shows a β-can fold containing an α-helix to which the chromophoric moiety is linked. The color is completely but reversibly abolished on unfolding. The spectroscopic properties of GFP have been intensively investigated (Tsien 1998 and references therein; Volkmer et al. 2000). The WT protein shows a predominant absorption band centered at 397 nm, attributed to the neutral form of the chromophore, and a lower intensity band at 470 nm, attributed to the anionic form of the chromophore. The transition between the two species is controlled by a single ionizable residue with a pK_a of ∼4.5 for the WT GFP and between 5.8 and 7.9 for different mutants (Terry et al. 1995; Patterson et al. 1997; Haupts et al. 1998; Elsliger et al. 1999). The anionic form is a highly fluorescent species (Tsien 1998). Independent of the excitation wavelength, the emission band is observed at 504 nm, indicating that a proton transfer process takes place in the excited state. Several single or multiple mutations of GFP were obtained by random and site-directed mutagenesis to modify the spectral properties and increase the folding efficiency. In particular, mutations involving Ser65 lead to the selective stabilization of the anionic form (Tsien 1998).

The photophysics of the fluorescent emission of WT GFP and mutants were investigated by fluorescence up-conversion spectroscopy (Chattoraj et al. 1996), fluorescent correlation spectroscopy (Terry et al. 1995; Haupts et al., 1998), spectral hole-burning (Creemers et al. 1999, 2000), and one and two-photon time-resolved fluorescence (Volkmer et al. 2000). A peculiar property of GFP, revealed by single-molecule experiments on mutants immobilized in polyacrylamide gels (Dickson et al. 1997), was the blinking and switching between ionization states on light and dark cycles.

Single-molecule experiments are performed by investigating molecules either diffusing in and out of the observation volume or fixed in space by different immobilization procedures (Lu et al. 1998; Kelley et al. 2001; Edman and Rigler 2000; Weiss 2000; Zhuang et al. 2000a,b; Talaga et al. 2000). Examples of the latter case are the coating of glass surfaces with dilute chromophore solutions, the attachment to gold surfaces, and the entrapment in polymeric matrices, such as polyacrylamide, polymetacrylate, and agarose gels. A very promising strategy for protein encapsulation (to our knowledge not yet applied in single-molecule experiments) is the sol-gel technique (Brinker and Scherer 1990; Ellerby et al. 1992; Brennan 1999; Bruno et al. 2001; Mozzarelli and Bettati 2001, and references therein). A critical step of single-molecule experiments on immobilized proteins is the evaluation of the influence of immobilization on protein structure and dynamics to validate the biological relevance of these studies. To this purpose we have selected GFP as an ideal candidate because of its stability, highly fluorescent properties, and well-documented photophysics. In the present study, the emission properties of the triple mutant Ser65Ala, Val68Leu, and Ser72Ala, called GFPmut2 (Cormack et al. 1996), were characterized in bulk solutions, dispersed on spin-coated glasses, and encapsulated in wet, porous silica gels. Confocal imaging, steady-state and time-resolved one and two-photon fluorescence spectroscopy, and fluorescence correlation spectroscopy (FCS) were performed on concentrated protein solutions and at single molecule level. Results clearly indicate that encapsulation of GFP in silica gels does not perturb protein dynamics and, thus, is a powerful strategy for single-molecule experiments.

Results

GFPmut2 in solution

Absorbance and fluorescence spectra and fluorescence anisotropy

The absorbance spectrum of GFPmut2 shows a band centered at 485 nm at alkaline pH (Fig. 1a ▶) (Cormack et al. 1996). The fluorescence emission spectrum, on excitation at 485 nm, shows a band centered at 507 nm (Fig. 1b ▶) (Cormack et al. 1996). Similar to WT GFP (Haupts et al. 1998), absorbance and emission are pH dependent (Fig. 1a,b, insets). The pH dependence was monitored between 5.5 and 8.0, as lower pH values induced protein precipitation. The isosbestic point at 425 nm (Fig. 1a ▶) indicates that only the protonated and unprotonated forms, absorbing at 388 and 485 nm, respectively, are in equilibrium. A single ionizable group, showing a pK_a of 6.13 ± 0.13 (Fig. 1a ▶, inset) or 6.24 ± 0.06 (Fig. 1b ▶, inset), controls the distribution of chromophoric species. The fluorescence anisotropy r of GFPmut2, calculated from the band area, was found to be 0.31 ± 0.02, independent of pH between 6.2 and 7.5 (data not shown).

Fig. 1.

(a) Absorption spectra of a solution containing GFPmut2, 10 mM potassium citrate, 100 mM potassium phosphate buffer, pH 5.47 (———), 6.02 (— — —), 6.33 (– – – –), 6.59 (— • — • —), and 7.70 (— •• — •• —). (Inset) Best fit of absorbance at 485 nm to the titration of a single ionizable group with a pK_a of 6.13 ± 0.13. (b) Fluorescence emission spectra (λ_ex = 485 nm) of a solution containing GFPmut2, 10 mM potassium citrate, 100 mM potassium phosphate buffer, pH 5.42 (———), 6.04 (— — —), 6.34 (– – – –), 6.54 (— • — • —), and 7.60 (— •• — •• —). (Inset) Best fit of fluorescence emission intensity at 507 nm to the titration of a single ionizable group with a pK_a of 6.24 ± 0.06.

Time-resolved fluorescence

The fluorescence decays of GFPmut2 were measured for different excitation and emission wavelengths (Table 1). Decays are well described by double exponentials. The slower component (τ₁ = 3.5 ± 0.2 ns and fractional amplitude of 0.87 ± 0.03) dominates the emission and shows a slight but definite dependence on the emission wavelength. These results are in agreement with previous studies on enhanced green fluorescent protein (EGFP) (F64L/S65T) (Haupts et al. 1998) and GFP-S65T mutant (Volkmer et al. 2000). The fluorescence polarization anisotropy decays, measured on excitation at 488 nm and emission at 535 nm (Fig. 2 ▶), were best fitted with two components. The slower decay corresponds to a rotational correlation time φ₁ of 12.5 ± 0.3 ns and anisotropy r₀₁ of 0.34 ± 0.1. The faster relaxation shows a very low anisotropy, r₀₂ = 0.04 ± 0.02, and is characterized by a rotational time φ₂ of 0.6 ± 0.15 ns. Previous experiments indicated that the short rotational time is probably an artifact related to light scattering or instrument noise (Swaminathan et al. 1997).

Table 1.

GFPmut2 lifetimes in solution upon single photon excitation

λex	λem	τ1 (ns)	f1	τ2 (ns)	f2
496	520	3.39 ± 0.03	0.86 ± 0.02	0.92 ± 0.06	0.14 ± 0.02
	535	3.43 ± 0.05	0.87 ± 0.01	0.90 ± 0.04	0.13 ± 0.01
	560	3.51 ± 0.04	0.84 ± 0.02	0.91 ± 0.06	0.16 ± 0.02
488	520	3.33 ± 0.03	0.84 ± 0.02	1.01 ± 0.015	0.17 ± 0.02
	535	3.44 ± 0.04	0.86 ± 0.02	1.12 ± 0.01	0.13 ± 0.02
	560	3.85 ± 0.03	0.84 ± 0.01	1.26 ± 0.06	0.16 ± 0.02
457.9	520	3.21 ± 0.01	0.88 ± 0.008	1.28 ± 0.06	0.12 ± 0.008
	535	3.41 ± 0.04	0.90 ± 0.01	1.25 ± 0.04	0.10 ± 0.01
	560	3.76 ± 0.03	0.94 ± 0.005	0.66 ± 0.04	0.06 ± 0.002

Fig. 2.

Differential phase (squares) and modulated anisotropy (circles) for GFPmut2 in solution (closed symbols) and encapsulated in silica gels (open symbols), 50 mM potassium phosphate, pH 6.7, on laser excitation at λ = 488 nm. Lines represent the least-squares best fit with a two-component model.

Fluorescence correlation spectroscopy

Two-photon excitation (TPE) fluorescence correlation spectroscopy measurements were performed using TPE at λ_exc = 800 nm on very dilute protein solutions, with an excitation power of 6–20 mW. Representative auto-correlation functions (ACFs) are shown in Figure 3 ▶. For GFPmut2, the ACFs were analyzed according to a simple diffusion model for longer lag times plus an exponential relaxation that corresponds to protein flickering or triplet state interconversion (Song et al. 1996; Haupts et al. 1998; Zumbusch and Jung 2000):

Fig. 3.

Autocorrelation functions of the emission of a solution containing GFPmut2 (concentration of the order of 0.1 μM), 50 mM phosphate buffer, pH 7.6, on two-photon excitation (TPE) at λ_ex = 800 nm. Lines represent the fit with a diffusion component for long lag times and an exponential relaxation for short lag times. Symbols refer to different excitation powers: 24 mW (triangles), 17 mW (closed and open circles), and 6 mW (squares). Inset: triplet relaxation time obtained from the exponential fit of the short lag time component of the auto-correlation functions (ACFs) as a function of the excitation power.

(1)

From the value of g(0) of 0.011 ± 0.002 a protein concentration of 68 ± 20 nM was estimated, in agreement with the expected value. From the average diffusion time τ of 430 ± 30 μs and the measured beam waist, a diffusion coefficient D of 91 ± 6 μm²/s was calculated. This value is in excellent agreement with previous data obtained on GFP-S65T (Swaminathan et al. 1997). The exponential relaxation in equation 3 can be related to a combination of flickering modes (reversible transitions to dark states) and thermally activated delayed fluorescence from the triplet state of GFP (Zumbusch and Jung 2000), depending on the excitation power. Interestingly, the fraction of molecules in the dark states was approximately F ≅ 54 ± 10% for the two lowest excitation powers, and the relaxation time decreased from τ_T ≅ 40 μs at P ≅ 6 mW to τ_T ≅ 20 μs at P ≅ 17 mW (Fig. 3 ▶, inset). Similar results were interpreted as a light-driven flickering of GFP (Schwille et al. 2000). At higher excitation powers, the relaxation time drops to ≅5 μs, a typical value for the thermally activated delayed fluorescence (Zumbusch and Jung 2000; Schwille et al. 2000).

GFPmut2 on glasses

Imaging

The representative image of glasses spin coated with 1–2 μM GFPmut2 (Fig. 4a ▶), obtained using the confocal microscope, showed well-defined fluorescent spots over a scattering background. The distribution of spot intensity, evaluated for each pixel (Fig. 5a ▶), and the average pixel intensity (Fig. 5b ▶) indicate that the dimmer spots contain single GFPmut2 molecules, whereas more intense spots contain up to four molecules. For the single molecule spots the fluorescence intensity was constant with time, before a sudden increase and a successive drop to the background level (Fig. 6 ▶). The duration of the bright phase, T_bright, for GFP-mut2, as determined from 10 spots, is 94 ± 3 sec, and decreases linearly with the excitation power (data not shown). This finding indicates that the drop of fluorescence emission may be attributable to a thermally induced local rearrangement of the fluorophore pocket. We do not have an explanation for the sudden signal increase observed just before the bleaching. The background level was ≅0.03 a.u., which is about three orders of magnitude lower than the fluorescence output from single GFP molecules, ≅40 a.u. This high signal to background ratio is particularly important in single-molecule experiments. The different levels of fluorescence intensity, shown by individual molecules (Fig. 6 ▶), are not caused by different orientation of the molecular dipole moments because the exciting light is unpolarized. In some cases, during the bright phase, a switching on/off behavior is observed. This phenomenon, known as blinking (Dickson et al. 1997), is related to internal photodynamics of individual GFP molecules. The present acquisition time of our setup is 229 ms per image; the time interval between consecutive images is 458 ms, likely preventing a detailed observation of blinking that is expected (for excitation intensity ≅700 kW/cm²) to be much faster than our resolution time. In fact, the on/off time varies over a wide range depending on the mutant GFP and the light intensity. For example, for EGFP it was found that <T_ON> ≅ 100 ms and <T_OFF> ≅ 2 sec for an excitation intensity ≅14 kW/cm² (Garcia-Parajo et al. 2000), which is considerably lower than the present value (700 kW/cm²). The WT GFP evanescent wave fluorescence microscopy measurements at an excitation power of ≅10 mW (corresponding presumably to an excitation intensity much lower than 14 kW/cm²) indicated an even longer <T_ON> ≅ 8 sec (Pierce et al. 1997).

Fig. 4.

Typical fluorescence image of GFPmut2, spin coated on etched glasses (a), and encapsulated in silica gel (b). The view field is 10 × 10 μm and the residence time is 9 μ. Fluorescence imaging was performed by confocal microscopy (see Materials and Methods) on laser excitation at 488 nm. Protein concentration was of the order of 1 μM in 50 mM potassium phosphate, pH 6.7.

Fig. 5.

(a) Distribution of spot average fluorescence emission of GFPmut2 evaluated by averaging on a square of 441 pixels around the spots for 10 similar images of protein on glass and for six images of protein in silica gels. Open and closed squares represent the histogram of the number of fluorescent spots with a given average fluorescence for GFPmut2 in silica gels and on glass, respectively. The solid line is a multigaussian fit of the data. (b) The mean values of the gaussian fits reported in panel (a) are plotted as a function of the aggregation order to determine the average fluorescence intensity per spot. The open and closed circles refer to data for GFPmut2 in silica gels and on glass, respectively. The solid line is a linear fit of the data.

Fig. 6.

Kinetics of fluorescence emission (λ_ex = 488 nm) from 10 single GFPmut2 molecules spin-coated on etched glasses. Collection time: 458 ms. Excitation power: ≅1 mW.

Fluorescence lifetimes

For fluorescence lifetime measurements on etched glass slides, solutions containing higher protein concentrations with respect to imaging experiments were spread by spin coating. The lifetimes were measured at |gn = 80 MHz, using 1 μM rhodamine 6G in ethanol as a reference. The Ti:Sapph laser beam was passed through the epi-fluorescence port on the sample at a power ≅24 mW for TPE at λ = 800 nm. For 100 nM fluoresceine solutions spread on glasses, it was found that τ_PH was always equal to τ_MOD and the average value of lifetime was τ = 4.4 ± 0.3 ns, very close to the solution value at high pH, ≅4.0 ns. The lifetimes for ≅1 μM GPFmut2, dispersed on etched glasses, were τ_MOD = 6 ± 0.7 ns and τ_PH = 12 ± 2.5 ns (data not shown). We could not obtain measurements with sufficient accuracy for larger modulation frequencies because of the low fluorescence rate of GFPmut2.

GFPmut2 encapsulated in silica gels

Fluorescence spectra and fluorescence steady-state anisotropy

On excitation at 485 nm, the fluorescence emission spectrum of GFPmut2, encapsulated in wet porous silica gels, showed a peak centered at 507 nm (Fig. 7 ▶), as in solution (Fig. 1b ▶). The pH dependence is controlled by an ionizable residue with pK_a of 6.46 ± 0.03 (Fig. 7 ▶, inset), a value slightly higher than in solution (Fig. 1b ▶, inset). The fluorescence anisotropy was found to be 0.38 ± 0.01.

Fig. 7.

pH-dependence of fluorescence emission spectra (λ_ex = 485 nm) of GFPmut2 encapsulated in silica gels. Gels were soaked in a solution containing 10 mM potassium citrate, 100 mM potassium phosphate buffer, pH 5.57 (———), 6.08 (— — —), 6.44 (– – – –), 6.74 (— • — • —), and 7.71 (— •• — •• —). (Inset) Best fit of fluorescence emission intensity at 507 nm to the titration of a single ionizable group with pK_a of 6.46 ± 0.03.

Imaging

Gels for fluorescence imaging were prepared by spreading a drop of GFPmut2-sol mixture between two glass slides before gel formation and sealing to avoid drying. Imaging performed on GFPmut2 gels with the confocal microscope revealed bright spots on top of a background signal (Fig. 4b ▶). The distribution of the spot intensity, evaluated for each pixel (Fig. 5a ▶), and the average pixel intensity (Fig. 5b ▶) allow identification of spots originated from single molecules of GFPmut2 (Chirico et al. 2001). The less bright spots show a stable fluorescence emission with T_bright of 62.4 ± 3.9 sec (Fig. 8 ▶). It is worth noting that (1) the fluorescence level of different single GFPmut2 molecules within the silica gels is remarkably similar, in contrast with GFPmut2 on glass slides (Fig. 5 ▶); (2) no sudden increase of fluorescence was observed before the signal drop; and (3) the intensity of fluorescence arising from a single molecule is approximately twice the background signal in the gel. As in the case of GFPmut2 spin-coated glasses, in some cases, switching on-off behavior of the fluorescence signal was observed.

Fig. 8.

Kinetics of fluorescence emission (λ_ex = 488 nm) from single green fluorescent protein (GFP) molecules encapsulated in silica gels. Gels were soaked in a solution containing 10 mM potassium citrate and 100 mM potassium phosphate buffer, pH 7.6. Collection time: 458 ms. Excitation power: ≅4 mW.

Time-resolved fluorescence

Lifetime measurements were performed on GFPmut2 gels with single photon excitation (λ = 488 nm) at 12 modulation frequencies in the range of 20–150 MHz. From the analysis of the data, a lifetime τ₁ of 3.5 ± 0.2 ns with f₁ = 0.82 ± 0.04 and a lifetime τ₂ of 1.1 ± 0.2 ns with f₂ = 0.18 ± 0.03 were calculated, in very close agreement with solution data. The fluorescence polarization anisotropy was measured on silica gels at higher GFPmut2 concentration. The decays are reported in Figure 2 ▶. Although the uncertainties on the gel data are larger because of the lower concentration and the residual scattering from the gel, the fluorescence decays of GFPmut2 gels are remarkably similar to those observed in solution. The analysis of the data indicate two FPA rotational times, φ₁ = 15 ± 2 ns with anisotropy r₁ = 0.34 ± 0.06 and a fast relaxation time φ₂ = 0.3 ± 0.1 ns with anisotropy r₂ = 0.05 ± 0.002.

Discussion

The fluorescence properties of the mutant GFPmut2 in three different environments—solution, etched glasses, and wet silica gels—were investigated to evaluate the effect of different immobilization procedures on the dynamic properties of the protein. This information is instrumental to establish the biological and biotechnological relevance of spectroscopic studies on single immobilized molecules. Single-molecule experiments can be performed also in solution by on-the-fly experiments (Deschenes et al. 2001; Weiss 1999 Weiss 2000; Volkmer et al. 2000; Haupts et al. 1998) probing the conformational ensemble of molecules while diffusing in the excitation volume. However, experiments on immobilized molecules offer the possibility to monitor a wider range of dynamical processes and their kinetics. Experiments have been performed on proteins either fixed on etched glasses by chemi-or physi-adsorption (Talaga et al. 2000) or entrapped in polymeric matrices as polyacrylamide (Dickson et al. 1997) and agarose gels (Lu et al. 1998). We are not aware of any single-molecule study on proteins encapsulated in wet silica gels.

We first investigated the static and dynamic properties of GFPmut2 in solution. The absorption and emission spectra of GFPmut2 show bands slightly blue-shifted with respect to EGFP (Patterson et al. 1997), and their dependence on pH is controlled by a single ionizable residue with a pK_a similar to that observed for EGFP (Haupts et al. 1998).

The overall size of the protein was determined both from the translational diffusion coefficient measured by FCS and the rotational diffusion time provided by the analysis of fluorescent polarization anisotropy data. From FCS measurements we determined D = 92 ± 6 μm²/s, which corresponds to a hydration radius R_H = 2.4 ± 0.17 nm, if a simple spherical symmetry is assumed. The rotational time in solution, φ₁ = 12.5 ± 0.3 ns, corresponds to a rotational diffusion coefficient Θ =13.3 ± 0.3 MHz. Assuming a spherical symmetry for the protein, an average radius R_Θ = 2.3 ± 0.02 nm was obtained, in excellent agreement with the hydration radius and with the values reported for WTGFP (Terry et al. 1995), EGFP (Haupts et al. 1998), and S65T mutant (Volkmer et al. 2000).

The steady-state fluorescence properties of GFPmut2, encapsulated in silica gels, appear almost indistinguishable from those obtained in solution. This implies that the gel matrix does not alter the protein conformation. Regarding the dynamic properties, the fluorescence lifetimes of GFPmut2, encapsulated in silica gels, are very similar to the values obtained in solution. In contrast, the fluorescence lifetimes of GFPmut2 on glasses are two-to fourfold higher. This indicates that the interaction with the bare glass surface induces a significant perturbation of protein structure and dynamics that is not observed when the protein is entrapped in silica gels. Moreover, the rotational correlation time of the protein in silica gels, φ₁ = 15 ± 2 ns, is only slightly higher than the value observed in solution, φ₁ = 12.5 ± 0.3 ns, indicating that the constraints imposed by the gel matrix on protein rotation are limited. Similar experiments on glasses yield almost no phase shifts, although with large uncertainties (data not shown), indicating that the rotational motion of the protein physi-adsorbed on the glasses is severely restricted, as reported for other molecules embedded in various types of gels (Deschenes et al. 2001).

To investigate biologically relevant processes on single molecules, such as protein folding and unfolding and enzyme catalysis, a critical requirement is the photostability of the excited chromophore. It is well known that Trp residues are very rapidly bleached by the intense light pulse used in single-molecule experiments and cannot be used as a probe (Bent and Hayon 1975). The time course of the fluorescence intensity of single GFPmut2 molecules immobilized on glasses and in silica gels indicates that the fluorescence signal arising from individual GFPmut2 molecules is easily resolved from the background. The signal/background ratio is lower in silica gels than in the case of GFPmut2 spread on etched glasses. Another important feature is that T_bright, the time after which the fluorescence definitely drops to the background level, is remarkably long for both the protein on glasses and in silica gels and allows one to monitor biochemically relevant processes. Moreover, the constant fluorescence intensity of GFPmut2 in silica gels, compared with the variability observed on glasses (Figs. 6, 8 ▶ ▶), provides evidence for an unperturbed native conformation of the protein in silica gels. On the contrary, the different fluorescence intensities, observed for GFPmut2 on glasses (Fig. 6 ▶), might be explained by either different or fixed orientations of the molecules with respect to the excitation light or a distribution of folded and partially unfolded proteins, the unfolded GFP having completely lost the green fluorescence (Tsien 1998). Confocal and near-field scanning optical microscopy measurements of fluorescence emission of S65T GFP, immobilized in polyacrylamide gels, indicate that the on time dramatically depends on light intensity and polarization (Garcia-Parajo et al. 2000). Because the excitation light used in our experiments was unpolarized, the presence of partially unfolded molecules on the coated glasses seems to be a more likely explanation. This finding calls for a cautious interpretation of single-molecule experiments performed on biomolecules directly immobilized on glasses. In such studies, which were aimed to investigate catalysis and folding of RNA (Zhuang et al. 2000) and folding and unfolding of two-stranded coiled-coil peptides (Talaga et al. 2000), severe controls were performed.

In the case of immobilization of proteins in silica gels, functional properties are not significantly perturbed, as proven for several enzymes (Bettati and Mozzarelli 2001, and references therein) and hemoglobin (Bettati and Mozzarelli 1997; Bruno et al. 2001; Abbruzzetti et al. 2001b). The influence of immobilization on dynamic properties may vary among different proteins depending on constraints caused by specific interactions with the negatively charged silica matrix. In the case of albumin labeled with acrylodan, time-resolved anisotropy measurements indicated a small decrease in the global motion of the protein and an unrestricted local motion of the probe with respect to the protein in solution (Jordan et al. 1995). In the case of silica gel-encapsulated myoglobin, rotational diffusion was significantly impeded (Gottfried et al. 1999), and the amplitude of the carbon monoxide geminate rebinding was increased (Abbruzzetti et al. 2001a). Interestingly, myoglobin shows a positively charged surface, whereas GFP is negatively charged. This finding indicates that silica gel protein encapsulation is a well-suited method for single-molecule experiments, but some caution should still be exerted.

Materials and methods

GFPmut2 expression and purification

GFPmut2 gene, cloned in a pKEN1 vector (Ezaz-Nikpay et al. 1994), was kindly provided by Dr. Brendan P. Cormack (Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA). GFPmut2 is a GFP mutant containing a triple substitution, S65A, V68L, and S72A, conferring enhanced fluorescence emission and high yield of protein caused by a more efficient folding at 37°C with respect to WT (Cormack et al. 1996). For expression of GFPmut2, Escherichia coli strain XL1-Blue was used and bacteria were grown in 2xYT medium (16 g tryptone, 10 g yeast extract, 5 g NaCl per liter). A single E. coli colony containing the recombinant construct was inoculated in 10 mL of 2xYT medium supplemented with ampicillin (100 μg/mL) and tetracycline (50 μg/mL) and incubated overnight at 37°C on a rotating shaker. One liter of 2xYT medium containing the antibiotics was inoculated with the starter culture and shaken at 37°C until the cells reached the mid-log phase of growth (A₆₀₀ = 0.6). Expression of GFPmut2 was induced by adding isopropyl-β-D-galactopyranoside to a final concentration of 1 mM and incubating the culture at 37°C on a rotating shaker for 2.5 h. Cells were harvested by centrifugation and resuspended in 50 mM Tris/HCl buffer, pH 8.0, containing 1 mM EDTA, 1 mM β-mercaptoethanol, 0.1 M NaCl, 0.5 mM phenylmethylsulphonylfluoride, 0.5 mM benzamidine, 1μM leupeptin, and 1 μM pepstatin. Cells were sonicated until the fluorescence emission of the supernatant, on excitation at 485 nm, was constant. The crude extract was centrifuged at 14,000g for 45 min and streptomycin sulfate was added to the supernatant to a final concentration of about 3.3% (w/v). The solution was kept for 30 min at 4°C and then centrifuged for 45 min at 14,000g. The pellet was discarded and ammonium sulfate was added to the supernatant to 40% saturation. The solution was then centrifuged at 14,000g for 15 min and the pellet was again discarded. Ammonium sulfate was added to the supernatant to 70% saturation. The solution was centrifuged at 14,000g for 15 min. The pellet was redissolved in a minimum volume of 50 mM Tris/HCl buffer, pH 8.0, containing 1 mM EDTA and 1 mM dithiothreitol (buffer A). The solution containing GFPmut2 was dialyzed overnight against buffer A containing 0.5 M NaCl. Protein solution was concentrated to about 2 mL by ultrafiltration and loaded onto a size exclusion column (G-75 fine, Amersham Pharmacia Biotech; bed volume = 200 mL) equilibrated with buffer A containing 0.5 M NaCl. The column was eluted at a constant flow of 80 μL/min. Fractions containing GFPmut2 were collected and the purity of the sample was evaluated by SDS-PAGE. Fractions were pooled on the basis of the degree of purification of GFPmut2 and concentrated to about 100 μM. An 85% pure protein solution was stored at −80° C and used in the experiments.

Experiments in solution

GFPmut2 from stock solutions was diluted in phosphate-containing buffers. All experiments were performed at room temperature.

Spin coating on etched glasses

The glass slides were first soaked in a solution containing 1% sodium dodecyl-sulfate for 24 h, then in a methanol solution saturated with NaOH for 2 h. To remove residual traces of NaOH, the slides were first soaked in 0.1% HCl solutions for 2 h, then in a diluted chromic solution for 2 h, and, finally, rinsed extensively with Milli-Q water (Millipore, Inc.). After this procedure, glasses were stored for a few hours in ethanol. After rinsing thoroughly with Milli-Q water and drying under a filtered nitrogen flux, sample solutions were spread on the glass slides by spin coating.

Silica gels

Encapsulation of GFPmut2 in silica gels was performed according to Bettati and Mozzarelli (1997). The stock protein solution was diluted 50 to 100-fold in 10 mM citrate, 100 mM phosphate buffer, pH 7.5. Sixty-seven μL of the resulting solution were mixed with 100 μL sol, prepared from tetramethylorthosilicate, water, hydrochloric acid, and phosphate buffer. On gelation, silica gels were covered with the same buffer solution and stored at 4°C for at least 12 h before use. Gel pore size is less than ∼20 Å because GFPmut2 molecules, characterized by an average diameter of about 40 Å, did not leach. Transmission electron microscopy measurements of silica gels with and without hemoglobin (60 Å in diameter) showed a typical pore structure with a pore size of ∼30–40 and 20 Å, respectively (Abbruzzetti et al. 2001b).

Fluorescence imaging

The optical setup for the single-photon imaging experiments is based on an inverted microscope (TE300, Nikon), a Nikon PCM2000 scanning head, and an air-cooled Argon laser with excitation wavelength at 488 nm. The laser beam is sent to the entrance pupil of a Nikon objective (N.A. = 1.4, Plan Apochromat DICH 100X oil, working distance 0.19 mm, focal length 2 mm) by the scanning lens. The fluorescence signal, collected by the same objective and selected by a HQ535–50 filter (Chroma Inc.), is fed to a single-mode fiber connected to a R928 photomultiplier (Hamamatsu) in the PCM2000 controller.

Fluorescent molecules, either spin coated on etched glasses or encapsulated in silica gels, were imaged by the Nikon EZ-2000 software interfaced to the PCM2000 scanning head (Diaspro et al. 1999a). For the confocal setup the resolutions are 0.19 μm in the plane and 0.6 μm in the axial direction (Diaspro et al. 1999a). The acquisition of the images (512 × 512 pixels) with a residence time of about 9 μs per pixel takes 2.3 sec. The view field is in the range 35–140 μm and the excitation power is usually about 13 mW, which corresponds to a light intensity of 700 kW/cm².

Fluorescence kinetics of individual spots

A wide field (80 × 80 μm²) image was collected before the time course acquisition and compared with a 3D scanning performed right after the kinetics to ensure that the disappearance of the fluorescence was not caused by a shift of the focus plane. For this application, 160 × 160 pixels images were acquired on single spots (15 × 15 μm² field) with ≅9 μs residence time. The acquisition time is 229 ms per image and the time interval between consecutive images is 458 ms. The fluorescence intensity of each spot was computed by summing the pixel content in a circular area around each spot and by normalizing for the number of pixels in this area, which has a typical diameter of 10 pixels, corresponding to ≅0.9 μm. The computation of the spot intensity was performed with a home-coded MatLab (Mathworks, Inc.) program, which allows identification of a single spot on a time series of images and computation of the spot intensity as a function of the acquisition time.

Fluorescence spectra acquisition

Steady-state fluorescence spectra were acquired with a Perkin-Elmer LB-50 spectrofluorometer. The fluorescence anisotropy G values for GFP in solution and encapsulated in silica gels are 1.38 ± 0.05 and 1.33 ± 0.03, respectively.

Fluorescence lifetime and anisotropy

For time-resolved fluorescence measurements, the emitted light was detected through the front port of the microscope by a R928 photomultiplier tube (Hamamatsu). The gain was modulated by biasing the second dynode stage at a radio frequency. The cross-correlation frequency was 36 Hz and the modulation frequencies were provided by a master radio frequency synthesizer (Marconi Instruments, mod. 2023A) that biased a Pockels cell, which, coupled to a polarizer beamsplitter, modulated the amplitude of the laser light intensity (Argon Laser, 2025) at frequencies in the range of 20–150 MHz. The photomultiplier signal was fed to an ISS lock-in amplifier board (ISS) for the computation of the polarized modulation ratios and phase differences of the fluorescence light with respect to the excitation laser beam. The synchronization was performed by sending the signal of a second photomultiplier, measuring the intensity of a small fraction of the excitation laser beam, to the ISS board as a frequency reference.

The fluorescence lifetime measurements were performed with the polarizer at the magic angle (Θ = 54.7°). The lifetime reference was either an alkaline solution of fluorescein at p|gH ≅ 8, characterized by a lifetime of 4.05 ns (Tsien and Wagonner 1995) or rhodamine 6G in ethanol, characterized by a lifetime of 3.89 ns (Thompson and Gratton 1988). The polarization of the fluorescence was selected by a Glan-Thompson polarizer (extinction ratio <10^–6) and the fluorescence polarization anisotropy measurements were performed by rotating the polarizer between the directions parallel and perpendicular to that of the excitation light. The differential phase shifts and the polarized modulation ratios were provided directly by the ISS acquisition program. The G factor was measured directly on the TE300 microscope and found to be very close to unity, G = 1.000 ± 0.005.

TPE fluorescence correlation spectroscopy

The optical setup for the TPE experiments is based on the same inverted microscope used for fluorescence imaging and a mode-locked Ti:sapphire laser (Tsunami 3960, Spectra Physics; pulses width of about 100 fs, repetition frequency of 80 MHz). A portion of the laser beam is sent to the entrance pupil of the objective by the scanning lens. The laser power at the object plane can be adjusted by neutral filters. Only about 30% of the laser light entering the scanning head excites the sample because of losses and absorption in the confocal head and the objective. The excitation power was typically 5 mW on the sample, corresponding to 900 kW/cm². The fluorescence signal was collected by the same objective and selected by a filter (HQ535–50, Chroma Inc.). The point spread function of the TPE corresponds to a plane resolution of ≅0.22 μm and an axial resolution of ≅0.6 μm (Diaspro et al. 1999b). For measurements of the fluorescence fluctuations, the photon counts are detected through the bottom port of the microscope by an avalanche photodiode detector (SPCM-AQ-151, EG&G). The output signal is fed to a correlator board (ISS) with a sampling time of 50 μs. The ACF of the fluorescence signal F(t), defined as g_F(t) = <δF(t)δF(0)>/<F(0)², can be well approximated (Berland et al. 1995) for the TPE experiments using the equation:

(2)

The zero lag time term is related to the average number concentration ≤C> as:

(3)

where the term κ = 0.076 is determined by the laser beam shape at the objective focus, the volume of the excitation profile is related to the laser beam waist, w₀, by the relationship V_EXC = πw₀⁴/λ (Berland et al. 1995), and B indicates the contribution of an uncorrelated background to the detected signal. The relaxation time is related to the translational diffusion coefficient D_T of the fluorophore, as τ ( 1.15w₀²/(8D_T) (Chirico et al. 2000). The analysis of the ACFs was performed by a home-coded least squares fitting program based on the Marquardt algorithm (Bevington 1992) and using the routine MRQMIN (Press et al. 1993) for fitting. From the analysis of the ACFs, the relaxation time, τ, and the zero lag-time term, g_F(0), were determined. Measurements of the fluorescence fluctuations of fluorescein solutions at various concentrations (Chirico et al. 2000) provide an estimate of the laser beam waist, w₀, assuming a translational diffusion coefficient D_T of 280 μm²/s at 22°C (Rigler et al. 1993). The value of the excitation volume determined for the present setup is V_EXC = 0.17 ± 0.04 fL at λ = 770 nm (Chirico et al. 2000).

Abbreviations

• GFP, green fluorescent protein

• GFPmut2, GFP mutant containing the triple substitution S65A, V68L, S72A

• TPE, two-photon excitation

• ACF, auto-correlation function

• FCS, fluorescent correlation spectroscopy

• Tris, tris(hydroxymethyl)aminomethane