Unfolding of Green Fluorescent Protein mut2 in wet nanoporous silica gels

Abstract

Many of the effects exerted on protein structure, stability, and dynamics by molecular crowding and confinement in the cellular environment can be mimicked by encapsulation in polymeric matrices. We have compared the stability and unfolding kinetics of a highly fluorescent mutant of Green Fluorescent Protein, GFPmut2, in solution and in wet, nanoporous silica gels. In the absence of denaturant, encapsulation does not induce any observable change in the circular dichroism and fluorescence emission spectra of GFPmut2. In solution, the unfolding induced by guanidinium chloride is well described by a thermodynamic and kinetic two-state process. In the gel, biphasic unfolding kinetics reveal that at least two alternative conformations of the native protein are significantly populated. The relative rates for the unfolding of each conformer differ by almost two orders of magnitude. The slower rate, once extrapolated to native solvent conditions, superimposes to that of the single unfolding phase observed in solution. Differences in the dependence on denaturant concentration are consistent with restrictions opposed by the gel to possibly expanded transition states and to the conformational entropy of the denatured ensemble. The observed behavior highlights the significance of investigating protein function and stability in different environments to uncover structural and dynamic properties that can escape detection in dilute solution, but might be relevant for proteins in vivo.

Biological macromolecules are usually studied in diluted solutions. However, the highly crowded cellular environment can have dramatic effects on protein stability, dynamics, and function by modifying solution viscosity, available volume, and solvent and solutes activity (Zimmerman and Minton 1993; Garner and Burg 1994; Minton 2000a,b, 2001; Ellis 2001a,b). This has recently boosted the interest for investigating biomolecules in artificially crowded and confined environments. Encapsulation in wet nanoporous silica gels, a widely used technique exploited to immobilize and confine protein molecules in a controlled environment (Ellerby et al. 1992; Avnir et al. 1994; Dave et al. 1996; Gill and Ballesteros 2000; Gill 2001; Livage et al. 2001; Mozzarelli and Bettati 2001; Jin and Brennan 2002; Bettati et al. 2004), was recently proved to be a valuable strategy to reproduce in vitro many of the effects of molecular crowding and confinement normally experienced by proteins in vivo (Eggers and Valentine 2001a,b; Klimov et al. 2002). The porous structure of silica gels allows a rapid exchange of solvent and solutes molecules between the gel matrix and the surrounding medium. Excluded volume and altered microviscosity and activity of solvent and solutes inside the gel pores are expected to influence the thermodynamics and kinetics of conformational equilibria. These effects have been exploited to gain insight into protein functional and regulatory properties either by selectively stabilizing distinct tertiary and/or quaternary states or by slowing down the kinetics of conformational transitions (Shibayama and Saigo 1995, 1999, 2001; Bettati and Mozzarelli 1997; Das et al. 1999; Abbruzzetti et al. 2001; Bruno et al. 2001; McIninch and Kantrowitz 2001; Samuni et al. 2002; Navati et al. 2003; West and Kantrowitz 2003; Bettati et al. 2004; Pioselli et al. 2004). Very few reports have addressed the effect of gel encapsulation on larger-scale conformational dynamics, such as those involved in protein folding and unfolding. The predicted stabilizing effect of caging and crowding on the folded state of globular proteins (Minton 1981, 2000a,b; Zhou and Dill 2001; Klimov et al. 2002; Takagi et al. 2003; Thirumalai et al. 2003) has been experimentally observed in the case of CO-myoglobin (Das et al. 1998; Samuni et al. 2000), cod III parvalbumin (Flora and Brennan 1998), cytochrome c (Lan et al. 1999), lysozyme, α-lactalbumin, apo- and met-myoglobin (Eggers and Valentine 2001b), ribonuclease A (Tokuriki et al. 2004), and a few other single domain proteins (Bolis et al. 2004). However, additive or contrasting effects on protein folding and unfolding can arise from the perturbation of solvent structure and of the entropy of denatured and transition states. This makes hard to draw general conclusions from a limited set of experimental results. Although, recently, the effect of alterations of protein hydration has been tackled by doping gels with several solutes (Eggers and Valentine 2001a; Brennan et al. 2003), many more systems should be investigated in order to achieve an adequate understanding of the type and scope of such effects.

In the present work, we have investigated the thermodynamics and kinetics of the guanidinium chloride (GdnHCl)–induced unfolding of the Green Fluorescent Protein (GFP) mut2 in solution and in silica gels. The triple substitution S65A, V68L, S72A confers to GFPmut2 a more efficient folding at 37°C, enhanced fluorescence emission upon excitation of the anionic form of the chromophore (Cormack et al. 1996), and an increased pK_a for the transition between the protonated and the highly fluorescent anionic form of GFP (Chirico et al. 2002). These properties render GFP-mut2 a good candidate for cell biology and biophysical applications. Hence, the investigation of GFPmut2 stability and unfolding mechanism in solution and in silica gel has the purpose to both add information on a valuable alternative to more widely used GFPs and gain insight into the effects of caging and crowding on protein stability and dynamics.

Results

Stability of GFPmut2 in solution and in silica gel

The far-UV circular dichroism (CD) and fluorescence emission spectra of GFPmut2 in solution and encapsulated in wet, nanoporous silica gels are reported in Figure 1 ▶. Encapsulation does not induce appreciable changes in the secondary structure of the protein (Fig. 1A ▶) and does not perturb or introduce any heterogeneity in the chromophore environment (Fig. 1B ▶). Full denaturation in the presence of 6.0 M GdnHCl, both in solution and in the gel, is indicated by the shape of CD spectra and by the complete abolishment of fluorescence emission (Fig. 1 ▶).

Figure 1.

(A) Far-UV CD spectra of a solution containing 6 μM GFPmut2, 20 mM Tris-HCl buffer, and 0.6 M NaCl (pH 7.4), in the absence (solid line) and presence of 6.0 M GdnHCl (dashed/dotted line), at 37°C, and gel-encapsulated GFPmut2 under the same buffer conditions, in the absence (dashed line) and presence of 6.0 M GdnHCl (dotted line). (B) Fluorescence emission spectra (λ_ex = 485 nm, slit_ex = 2 nm, slit_em = 2 nm) of a solution containing 100 nM GFPmut2, 20 mM Tris-HCl buffer, and 0.6 M NaCl (pH 7.4), at 37°C (solid line), and gel-encapsulated GFPmut2 under the same buffer conditions (dashed line). Spectra were normalized for the emission intensity at 507 nm. At 6.0 M GdnHCl, fluorescence is completely abolished both in solution and in the gel.

The effect of encapsulation on the thermodynamic stability of GFPmut2 was determined by measuring the dependence of molar ellipticity at 220 nm on GdnHCl concentration (Fig. 2 ▶). Control refolding experiments in solution indicated that <60% of the initial molar ellipticity was recovered after full denaturation in 6.0 M GdnHCl followed by rapid 15-fold dilution to 0.4 M GdnHCl (data not shown). The extent of aggregation appeared to depend on the residence time in the unfolded state. The refolding efficiency was not improved by the addition of DTT to the unfolding and refolding solutions, indicating that the formation of nonnative disulphide bridges is not the major determinant in the aggregation of denatured GFPmut2. In the gel, interprotein interactions are prevented and removal of denaturant after complete unfolding in 6.0 M GdnHCl results in the recovery of 90% of the initial molar ellipticity (data not shown). The recovery of fluorescence emission after dilution of the denaturant was consistently <50%, both in solution and in silica gel. This behavior cannot be ascribed to irreversible denaturation and/or aggregation. Unfolding–refolding experiments carried out by using two-photon fluorescence spectroscopy (F. Cannone, S. Bologna, B. Campanini, A. Diaspro, S. Bettati, A. Mozzarelli, and G. Chirico, in prep.) suggest that refolding GFPmut2 molecules can populate slowly equilibrating distinct electronic states, endowed with different fluorescent properties. This hampers a straightforward analysis of equilibrium curves and refolding kinetics but does not affect the observed unfolding kinetics at high denaturant concentration (see below).

Figure 2.

Equilibrium denaturation curves of GFPmut2 in solution (filled circles) and encapsulated in silica gel (open circles) monitored by CD at 220 nm. The fraction of unfolded molecules (f_u) is calculated according to Equation 5. Solution experiments were carried out in 20 mM Tris-HCl buffer and 0.6 M NaCl (pH 7.4) at 37°C. Experiments in the gel were carried out in the same buffer at pH 7.0. The fitting to a two-state transition gives the following parameters: ΔG⁰_0,U = 2.6 ± 0.2 kcal/mol and D₅₀ = 2.3 ± 0.1 M for solution (solid line), and ΔG⁰_0,U = 4.3 ± 0.7 kcal/ mol and D₅₀ = 2.1 ± 0.1 M for silica gels (dashed line).

Although the equilibrium data in solution must be taken with caution, due to the incomplete reversibility, the comparison of the denaturation curves indicates that the apparent unfolding midpoint, 2.3 ± 0.1 M in solution and 2.1 ± 0.1 M in silica gel, is scarcely affected by encapsulation (Fig. 2 ▶). The analysis of the equilibrium curves with the equation for a two-state transition (Equation 4) yields an unfolding free energy of 4.3 ± 0.7 kcal/mol in the gel and 2.6 ± 0.2 kcal/mol in solution. The apparent stabilizing effect of protein encapsulation is consistent with that previously observed for other proteins (Das et al. 1998; Flora and Brennan 1998; Lan et al. 1999; Samuni et al. 2000; Eggers and Valentine 2001b; Bolis et al. 2004; Tokuriki et al. 2004), and with the expectation that the reduced conformational entropy inside the gel pores raises the free energy of the denatured conformations.

Unfolding kinetics

The unfolding kinetics of soluble and silica gel–encapsulated GFPmut2 have been measured both by far-UV CD spectroscopy and fluorescence emission (Figs. 3 ▶, 4 ▶). Only denaturant concentrations > 3.0 M were used, since the equilibrium denaturation curves show that in this regime the unfolding transition is almost complete (Fig. 2 ▶). This implies that the microscopic unfolding rate constant k_u is much higher than is the refolding rate constant k_f, and is well approximated by the observed rate constant k_obs (Wallace and Matthews 2002).

Figure 3.

(A) Representative unfolding kinetics of GFPmut2 in a solution containing 4.5 M GdnHCl, 20 mM Tris-HCl buffer, and 0.6 M NaCl (pH 7.4) 37°C, monitored by CD at 220 nm (filled circles) and fluorescence emission at 507 nm (filled squares) (λ_ex = 485 nm, slit_ex = 2 nm, slit_em = 2 nm). Protein concentration was 6 μM in CD experiments and 100 nM in fluorescence experiments. Lines through data points represent fittings to a single exponential decay for CD data (r² = 0.996), and to a double exponential decay for fluorescence emission data (r² = 0.999). (B) Semilogarithmic plot of the denaturant dependence of the unfolding rates for GFPmut2 in solutions containing 20 mM Tris-HCl buffer and 0.6 M NaCl (pH 7.4) 37°C, with variable concentrations of GdnHCl, as monitored by CD at 220 nm (filled circles) and fluorescence emission at 507 nm (filled squares). Lines through data points represent the fitting to a linear equation. The values of lnk_obs at zero denaturant concentration are −11.7 ± 0.6 min⁻¹ and −11.8 ± 0.7 min⁻¹ as determined from CD and fluorescence measurements, respectively.

Figure 4.

(A) Representative unfolding kinetics of GFPmut2 in silica gel in the presence of 4.5 M GdnHCl, monitored by CD at 220 nm (open circles) and fluorescence emission at 507 nm (open squares) (λ_ex = 485 nm, slit_ex = 2 nm, slit_em = 2 nm), at 37°C, in 20 mM Tris-HCl buffer and 0.6 M NaCl (pH 7.0 and pH 7.4, respectively). Lines through data points represent fittings to a double exponential decay for CD data (r² = 0.997), and to a triple exponential decay for fluorescence emission data (r² = 0.999). (B) Semilogarithmic plot of the denaturant dependence of the unfolding rates for gel-encapsulated GFPmut2, in 20 mM Tris-HCl buffer and 0.6 M NaCl, 37°C, with variable concentrations of GdnHCl as monitored by CD at 220 nm (open and dotted circles) and fluorescence emission at 507 nm (open and dotted squares). Lines through data points represent the fittings to a linear equation. The values of lnk_obs at zero denaturant concentration are −7.5 ± 0.1 min⁻¹ and −11.6 ± 0.6 min⁻¹ as determined from CD measurements, −7.1 ± 0.2 min⁻¹ and −10.7 ± 0.5 min⁻¹ as determined from fluorescence measurements. The thick line is the denaturant dependence of the unfolding rate of GFPmut2 in solution, after averaging the rates measured by CD and fluorescence emission (data from Fig. 3 ▶).

A representative unfolding kinetics in solution, monitored by CD at 220 nm, is reported in Figure 3A ▶. At all denaturant concentrations between 3.0 and 5.5 M GdnHCl, CD unfolding kinetics are satisfactorily described by a single exponential decay (Table 1). The dependence of the unfolding rate on the denaturant concentration is shown in Figure 3B ▶. The unfolding kinetics monitored by fluorescence emission (Table 1; for a representative kinetic trace and fitting, see Fig. 3A ▶) exhibits a fast phase, partly lost within the manual mixing time, and a subsequent slower process dominating the total amplitude. Throughout the whole range of denaturant concentrations exploited, the amplitude of the fast phase is <20%. This phase is ascribed to the fluorophore response to the increased ionic strength upon the addition of GdnHCl, since comparable effects are induced by an equal amount of NaCl (data not shown). Thus, only the slow phase was considered in the comparison with the unfolding kinetics measured by CD. Figure 3B ▶ shows that rates and denaturant dependence determined by CD and fluorescence spectroscopy virtually overlap.

Table 1.

Rates and amplitudes of GFPmut2 unfolding kinetics in solution

	CD	Fluorescence
[GdnHCl] (M)	k × 10−3	k1 × 10−3	k2 × 10−3	A2
3.0	3.3	56.7	3.1	81
3.5	5.7	179.7	4.9	82
4.0	11.4	147.7	9.9	81
4.5	28.9	194.6	25.0	85
5.0	113.0	a	116.5	a
5.5	324.3	a	274.2	a

Rate constants are in min⁻¹, amplitudes are in relative percentage. Standard errors on the parameters are within 5% for both rate constants and amplitudes.

^a At 5.0 and 5.5 M GdnHCl, fitting to the equation for a biexponential process (Equation 2) does not allow to resolve two separate phases.

The unfolding kinetics of gel-encapsulated GFPmut2 are more complicated than are those observed in solution. The time course monitored by CD at 220 nm is well described by a double exponential decay, while three exponentials are needed to fit the unfolding kinetics monitored by the decrease of fluorescence emission at 507 nm (Fig. 4 ▶; Table 2). The slow and intermediate phases observed in the fluorescence experiments exhibit rates and denaturant concentration-dependence similar to those of the two processes measured in CD kinetics (Fig. 4B ▶). The additional, fast phase observed in fluorescence decays shows a negligible dependence on denaturant concentration and, although with a higher relative amplitude, can be ascribed to the same processes accounting for the fast phase of the experiments in solution (Table 1). Thus, the unfolding of GFPmut2 occurs with a single apparent rate in solution and two rates in silica gel. The thick line in Figure 4B ▶ represents the semilogarithmic plot of the unfolding rate in solution, as a function of denaturant concentration, obtained by averaging the rates observed by CD and fluorescence emission. Two features should be highlighted: The slope is more pronounced than those observed for the two unfolding phases in the gel, and noticeably, the unfolding rate extrapolated at zero denaturant concentration matches the extrapolated value for the slow phase in the gel.

Table 2.

Rates and amplitudes of GFPmut2 unfolding kinetics in silica gels

	CD			Fluorescence
[GdnHCl] (M)	k1 × 10−3	k2 × 10−3	A2	k1 × 10−3	A1	k2 × 10−3	A2	k3 × 10−3	A3
3.0	17.9	1.4	70	467.5	51	18.2	29	2.2	20
3.5	32.9	2.0	55	664.3	55	36.1	30	4.7	15
4.0	57.5	5.3	41	618.3	44	69.3	38	12.7	18
4.5	117.2	7.7	19	808.2	50	95.1	38	15.1	12
5.0	183.7	26.3	27	850.0	51	176.9	29	42.1	20
5.5	336.9	70.8	37	850.0	55	273.8	30	117.4	15

Rate constant are in min⁻¹, amplitudes are in relative percentage. Standard errors on the parameters are within 10% for both rate constants and amplitudes, except in the presence of 5.5 M GdnHCl (SE = 30%).

Discussion

After the early studies by Ward and Bokman (Bokman and Ward 1981; Ward and Bokman 1982), a limited number of investigations have been reported dealing with in vitro GFP stability toward chemical denaturation and folding/unfolding kinetics and mechanism (Reid and Flynn 1997; Fukuda et al. 2000; Iwai et al. 2001; Verkhusha et al. 2003). Among these, Kawajima and coworkers (Fukuda et al. 2000) characterized the unfolding and refolding kinetics of wt GFP and the Cycle3 mutant (Crameri et al. 1996). The investigators report that GFP has a strong tendency to aggregate, accounting for lack of complete reversibility and preventing the determination of meaningful thermodynamic parameters for the folding/unfolding equilibrium. A refolding yield not > 60%, upon denaturation induced by GdnHCl, was reported for Cycle 3 GFP and S65T GFP (Battistutta et al. 2000). In nanoporous silica gels, protein molecules are individually caged in the pores of the matrix, allowing rapid diffusion of small solutes but preventing interprotein interactions. Thus, encapsulation in silica gels avoids aggregation and permits determination of the thermodynamic stability under equilibrium conditions. Moreover, given the predicted and observed effects of caging on protein stability and dynamics (Minton 1981; Zimmerman and Minton 1993; Garner and Burg 1994; Das et al. 1998; van den Berg et al. 1999, 2000; Samuni et al. 2000; Eggers and Valentine 2001b; Ellis 2001b; Zhou and Dill 2001; Klimov et al. 2002; Takagi et al. 2003; Thirumalai et al. 2003; Bolis et al. 2004; Tokuriki et al. 2004), sol-gel encapsulation can be considered a good strategy to investigate how the altered microviscosity, solvent activity, and excluded volume effects influence thermodynamic and kinetic behavior. In this regard, entrapment in silica gels exposes proteins to a microenvironment with physico-chemical properties more similar to the crowded intracellular environment than the diluted solutions normally used for biochemical and biophysical investigations (Viappiani et al. 2004). GFP was shown to lose the capability to fold spontaneously in vitro in a crowded environment, becoming dependent on the presence of the complete GroEL/GroES chaperonin system (Martin 2002). On the other hand, it has been recently proposed that confinement itself could be a major determinant of accelerated protein folding within the chaperonin cavity (Betancourt and Thirumalai 1999; Brinker et al. 2001; Takagi et al. 2003; Bettati et al. 2004).

The chromophoric and fluorescent properties of native GFPs (Tsien 1998) are acquired through a slow oxidation process following protein folding and cyclization of the chromophore (Kolb et al. 1996; Makino et al. 1997; Reid and Flynn 1997). Since the chromophore remains intact upon GFP denaturation (Ward et al. 1980), the loss of fluorescence in the denatured protein upon excitation at 485 nm is due to the disruption of the network of interactions that stabilize the anionic form of the chromophore. Therefore, fluorescence has been assumed to be an indicator of the presence of a native tertiary structure in unfolding and re-folding kinetic experiments (Reid and Flynn 1997; Fukuda et al. 2000). The observation that GFPmut2 molecules can populate, upon refolding, a distinct electronic state (F. Cannone, S. Bologna, B. Campanini, A. Diaspro, S. Bettati, A. Mozzarelli, and G. Chirico, in prep.) indicates that under equilibrium conditions fluorescence emission is not a reliable probe for folding. Instead, the unfolding exhibited a good reversibility, in the gel, when monitored by far-UV CD. In solution, aggregation of denatured GFPmut2 occurs at a significant extent, preventing the possibility to exploit the calculated thermodynamic parameters to carry out a rigorous comparison between the thermodynamic stability in solution and in the gel. However, it should be observed that wt GFP and the Cycle 3 mutant exhibited much higher denaturation midpoints in the presence of GdnHCl (Fukuda et al. 2000). Therefore, it appears that the improved folding efficiency in vivo at 37°C conferred by the triple mutation S65A, V68L, S72A (Cormack et al. 1996) does not arise from an increased thermodynamic stability.

In order to determine the effect of encapsulation on the kinetics of large scale dynamics, we have measured the unfolding kinetics of GFPmut2 in solution and in the gel (Figs. 3 ▶, 4 ▶). The results reveal striking differences between solution and silica gel environment. In solution, the unfolding of GFPmut2 can be described by a single exponential process. The near perfect overlap of the apparent unfolding rates determined by CD and fluorescence emission (Fig. 3 ▶) indicates that the disruption of native secondary structure is concomitant to the quenching of the chromophore fluorescence. The comparison with the unfolding kinetics of wt GFP and the Cycle 3 mutant, at similar temperature and pH (35°C, pH 7.5) (Fukuda et al. 2000), indicates that the triple mutation S65A, V68L, S72A induces an increase of the unfolding rate extrapolated at zero GdnHCl concentration of more than two orders of magnitude. Still, the unfolding rate of GFPmut2 exhibits a very low value for a globular protein, suggesting a possible mechanism of kinetic stabilization (Sohl et al. 1998; Cunningham et al. 1999; Bettati et al. 2000; Manning and Colon 2004) compensating the relatively low thermodynamic stability.

In the gel, the unfolding kinetics measured by CD are clearly biphasic, while three exponential relaxations are required to fit the time course of fluorescence decrease upon exposure to GdnHCl (Fig. 4A ▶). The fast phase detected by fluorescence, virtually independent of denaturant concentration, accounts for ~50% of the total amplitude (Table 2) and is ascribed to nonspecific ionic strength effects on the fluorescence yield of the chromophore. The higher amplitude of this phase in the gel with respect to solution indicates an enhanced effect of solvent polarity changes on fluorescence emission. The remaining two phases of the unfolding kinetics monitored by fluorescence emission have the same apparent rates and denaturant dependence as those observed by CD spectroscopy (Table 2; Fig. 4B ▶).

The only phase detected in solution exhibits rates that lay between those measured in the gel, and is endowed with a more pronounced denaturant dependence. However, a meaningful comparison of the rates observed in solution and in silica gels has to be carried out considering the extrapolated value at zero denaturant concentration. It is evident that the unfolding rate in solution in native conditions is actually similar to the slowest phase detected in the gel, both by CD and fluorescence spectroscopy (Fig. 4B ▶). The depressed denaturant dependence in the gel (Fig. 4B ▶) is consistent with the fact that caging in the finite space of the pores poses constrains to possibly expanded transition states and limits the conformational entropy of the unfolded state ensemble. This is expected to alter the steepness of the semilogarithmic plot of the unfolding rate as a function of denaturant concentration, which is proportional to the change in solvent-exposed surface upon denaturation (Myers and Oas 2002).

By taking into account that gel encapsulation does not cause detectable changes in the secondary structure (Fig. 1A ▶) or in the shape of the fluorescence emission spectrum of GFPmut2 (Fig. 1B ▶), a simple scheme allows to rationalize the biphasic unfolding kinetics in the gel: On the time scale of unfolding experiments, encapsulation locks an equilibrium distribution between two native conformations of GFPmut2. One of these conformations, which is not significantly populated in solution, unfolds with a rate that is almost two orders of magnitude higher than that measured for the other conformer both in solution and in the gel. Although there are reports indicating that the encapsulation protocol does not perturb the equilibrium conformational distribution of entrapped proteins, as in the case of myoglobin (Samuni et al. 2002) and hemoglobin (Viappiani et al. 2004), the selective stabilization in the gel of specific tertiary conformations has been previously observed for pyridoxal 5′-phosphate–dependent enzymes (Pioselli et al. 2004) and lipases (Reetz 1997). This result confirms that sol-gel encapsulation of proteins can be exploited not only to slow down by orders of magnitude the relaxation kinetics between different protein conformations but also to bias under equilibrium conditions a pre-existing distribution of conformations. Recently, NMR experiments provided evidence for exchange processes, in solution, between different conformational states of mutant GFPs (Seifert et al. 2002, 2003). Finally, it should be noted that a significant fraction of encapsulated molecules unfolds with the same rate at zero denaturant concentration as measured for the protein in solution (Fig. 4B ▶), indicating lack of dramatic effects on the position of the transition state and the height of the energy barrier for unfolding. Alternative models, interpreting the observed double exponential kinetics in the gel in terms of a sequential unfolding mechanism, can be discarded provided the observed lack of heterogeneity in the secondary structure and fluorescent properties in the absence of denaturant (Fig. 1A,B ▶), the near-perfect overlapping of fluorescence and CD kinetics (Tables 1, 2; Fig. 4 ▶), and the parallel dependence on denaturant of the two unfolding rates (Fig. 4B ▶). Such behavior indicates that the two transitions imply comparable changes in the solvent accessible surface of the protein and are not related to the stabilization of a kinetic intermediate in the gel.

Conclusions

We have investigated the unfolding equilibrium and kinetics of GFPmut2 in solution and in silica gels in order to gain insight into the effects of caging and crowding on protein structure and large scale dynamics. The results indicate that a significant fraction of the encapsulated molecules have a different conformation from the dominant species in solution, despite undistinguishable fluorescent properties and secondary structure. This influences the overall equilibrium unfolding curve and the unfolding kinetics, clearly biphasic in the encapsulated samples. When the unfolding kinetics are analyzed in terms of separate contributions of the two alternative conformations, the interesting conclusion is that the unfolding rate of the conformation that dominates the kinetics in solution is virtually unaffected by encapsulation, except for a reduced dependence on denaturant concentration, that is expected based on the steric restrictions to the expansion of transition states and denatured conformations.

Immobilization in the optically transparent matrix and the high fluorescence yield of GFPmut2 allow to exploit entrapment in silica gel to undertake the investigation of protein folding and unfolding at a single molecule level (F. Cannone, S. Bologna, B. Campanini, A. Diaspro, S. Bettati, A. Mozzarelli, and G. Chirico, in prep.).

Materials and methods

Protein

The pKEN1 vector (Ezaznikpay et al. 1994) containing the GFP-mut2 gene was kindly provided by Dr. Brendan P. Cormack (Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA). Protein expression and purification were carried out as described elsewhere (Chirico et al. 2002). GFPmut2 was stored at −80°C as a concentrated solution (~100 μM) and diluted for the different applications.

Chemicals and buffers

All the chemicals were purchased from Sigma-Aldrich and were used without further purification.

Equilibrium and kinetic experiments were carried out in 20 mM TrisHCl and 600 mM NaCl (pH 7.0 or 7.4). Sodium chloride was added to the buffer to allow for free diffusion of the positively charged denaturant molecules in the pores of the gel, which at a pH around neutrality bear a net negative charge (Shen and Kostic 1997) that can lead to a partitioning of the solute. At an ionic strength of 600 mM, electrostatic interactions between the gel matrix and protein surface are negligible and the concentration of GdnHCl inside the pores of the gel is equal to the concentration in the surrounding solution (Badjic and Kostic 1999). GdnHCl was dissolved in 40 mM TrisHCl and 1.2 M NaCl to obtain a 7 M stock solution (Kawahara and Tanford 1966; Pace and Scholtz 1997). The stock solution was diluted with buffer to obtain the desired denaturant concentration.

Silica gels

GFPmut2-doped nanoporous silica gels were prepared according to the procedure of Bettati and Mozzarelli (1997) with some modifications. One volume of potassium phosphate buffer (pH 6.0), containing 1 mM EDTA was added to the silica sol prepared from tetramethyl orthosilicate (Ellerby et al. 1992). The solution was bubbled with humidified nitrogen for 40 min; 1.5 volumes of the resulting sol were mixed with 1 volume of GFPmut2 solution in potassium phosphate buffer (pH 7.5). The mixture was layered on quartz slides, and after gelation occurred, the silica gels were kept at 4°C in 100 mM potassium phosphate buffer (pH 7.0). Silica gels exhibited a thickness of ~0.5 mm. No protein leaking from the silica matrix was observed up to 4 d at 37°C.

Fluorescence measurements

Fluorescence emission spectra and single-wavelength kinetic traces were acquired with a FluoroMax-3 fluorometer (HORIBA Jobin Yvon), equipped with a thermostated cell-holder. The protein fluorescence was excitated at 485 nm, using 2-nm excitation slits. The emission of the chromophore was collected in the range 495–600 nm or at 507 nm (emission slits = 2 nm). The experiments were carried out in 20 mM Tris-NaCl buffer (pH 7.4) at a protein concentration of 100 nM in solution and 1 μM in silica gel, at 37 ± 0.5°C. The slides were fixed inside the optical cuvette at an angle of 45° with respect to the excitation light. This setup minimizes the amount of scattered light reaching the emission detector (Brennan 1999).

The estimated dead time due to manual mixing is ~10 sec for kinetic experiments in solution, and ~30 sec for those in silica gel.

CD measurements

CD measurements were carried out using a Jasco J-715 spectro-polarimeter equipped with a Peltier element for temperature control. All the measurements were carried out at 37 ± 0.5°C. The protein concentration used was 6 μM and 30 μM for experiments in solution and in silica gel, respectively. Experiments in solution were carried out in 20 mM Tris-NaCl buffer (pH 7.4) using a microcuvette with an optical path of 0.1 cm. Measurements on GFPmut2 encapsulated in silica gels were carried out in 20 mM Tris-NaCl buffer (pH 7.0). It was necessary to use a lower pH for experiments in silica gel because at high protein concentration some leaking occurs from the gel matrix (B. Campanini, unpubl.). At pH 7.0 the protein release from the silica gels is negligible within the experiment time-window. Lowering pH from 7.4 to 7.0 does not affect protein unfolding kinetics measured by fluorescence spectroscopy, allowing direct comparison of the data collected in the two conditions.

Spectra were collected in the range 210–260 nm, because TrisHCl buffer interferes with far-UV light at low wavelengths. Single-wavelength kinetics were collected at 220 nm.

Refolding

The reversibility of the unfolding reaction was assessed by measuring the degree of signal recovery upon removal of denaturant. The protein, in solution or encapsulated in silica gel, was fully denatured in 6.0 M GdnHCl. The solution of denatured protein was diluted 15-fold in Tris-NaCl buffer (pH 7.4) at 37°C to obtain a final GdnHCl concentration of 0.4 M. The denatured encapsulated protein was transferred to Tris-NaCl buffer (pH 7.0) kept at 37°C. In both cases the kinetics of the refolding reaction was followed until the signal reached a constant value.

Data analysis

Protein unfolding kinetic traces were fitted to a single, double, or triple exponential equation:

(1)

(2)

(3)

where I is the fluorescence or CD signal intensity, and I₀ is the signal intensity at time t = 0; a, b, and c are pre-exponential factors accounting for the amplitude of the corresponding kinetic phase; and k₁, k₂, and k₃ are the rate constants.

Equilibrium unfolding transitions where fitted to a two-state model according to the equation

(4)

where I and I_0,N are the observed signal intensity at a defined denaturant concentration and in the absence of denaturant, respectively; I_0,U is the signal intensity of the fully denatured species; ΔG⁰_0,U is the free energy change in the absence of denaturant, and m is the dependence of the unfolding free energy ΔG_U⁰ on denaturant concentration. The values I_0,N and I_0,U obtained by fitting experimental data with Equation 4 were used to calculate the fraction of unfolded protein (ƒ_U) according to the equation

(5)

The denaturant concentration at half transition (D₅₀) is calculated as

(6)

Article published online ahead of print. Article and publication date are at http://www.proteinscience.org/cgi/doi/10.1110/ps.041190805.