INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

One of the most puzzling problems in biochemistry is how a polypetide chain, which is synthesized as a linear heterogeneous polymer, folds into a functionally unique three-dimensional structure. In this respect, the knowledge of the energetics of protein folding/unfolding is crucial to our understanding of the formation and functioning of protein molecules (Privalov, 1992). The stability of the native conformation of proteins is the result of all noncovalent interactions between backbone and side chain atoms, the covalent interchain disulfide bonds, and the coordinative bond between the metal ion and the protein side chains that act as ligands in all metallo-proteins. It has been shown that the integrity of the native three-dimensional structure of many proteins is promoted by the presence of disulfide bridges, because the removal of one or more of these bridges results in a reduction in the stability of the native relative to the denatured state (Creighton, 1974; White, 1982; Wetzel et al., 1988; Taniyama et al., 1988; Inaka et al., 1991; Cooper et al., 1992). Consequently, the investigation of the role of disulfide bonds for the stability of folded proteins has received a lot of attention. Protein engineering techniques have been recently employed in the attempts to increase both the overall stability of proteins by the introduction of non-native disulfide linkages into the structure and as specific probes of the folding pathway (Creighton, 1992; Clarke and Fersht, 1993). Although in some cases, the increase of the stability of a mutated protein relative to the wild type (wt) form has been observed, this is not generally the case (Wetzel, 1987; Matsumura et al., 1989). Many factors, including the location of the mutation site, the loop size connected by the -SS- linkage, and the strain energy introduced by the disulfide bridge, have to be considered to estimate the effect on the protein stability (Matsumura et al., 1989; Vogl et al., 1995). The mechanism of protein stabilization by disulfide bridge formation is difficult to resolve because the -SS- bond may influence enthalpy and entropy of both the native and unfolded state of the protein. In this respect, the two approaches commonly used are the introduction of artificial disulfide bridges (Clarke and Fersht, 1993; Gokhale et al., 1994; Tamura et al., 1994) and the removal of natural disulfide bonds either through Cys replacement by genetic engineering techniques or by opening existing disulfide bridges by chemical reduction (Schwarz et al., 1987; Pace et al., 1988; Cooper et al., 1992).

The generality of these concepts has been tested on azurin, a small blue copper protein belonging to the cupredoxin family, that acts as an electron transfer shuttle in the redox systems of certain bacteria. This protein has 128 amino acid residues and one copper(II) ion with Gly-45, His-46, Cys-112, His-117, Met-121 as metal ligands. The structure of wt azurin, that has been resolved by x-ray diffraction study (Nar et al., 1991), consists of eight -strands that form two sheets in a Greek-key motif (Fig. 1). This is the common fold in the cupredoxin family. In azurin, the two sheets are connected by an -helix at one side and by a kink in one of the -strands at the other side. A disulfide bridge connecting Cys-3 and Cys-26 is present at the southern end of the protein. Azurin has only one tryptophan residue, Trp-48, pointing with its side chain to the center of the hydrophobic core.

View larger version (76K): [in this window] [in a new window]   FIGURE 1   Model of wt azurin generated with Molscript program (Kraulis, 1991). The figure shows the copper ion with its ligands (top), the Trp-48 residue (middle), and the disulfide bridge (bottom).

It is generally recognized that the stability of the cupredoxins does not depend on the presence of the disulfide bridge. In fact, other members of the cupredoxin family, such as plastocyanin (Guss et al., 1992) and amicyanin (Kalverda et al., 1994) have not such a bond, and, in the case of stellacyanin (Strange et al., 1995), the disulfide bridge is in a region of the protein quite different from that of the disulfide bond in azurin. However, simple structural considerations can be misleading in estimating the stability of a protein. The knowledge of the energetics of the protein is also required. The thermal behavior of wt azurin has been previously investigated by some of us (La Rosa et al., 1995; Guzzi et al., 1996, 1998) from both a spectroscopic and a calorimetric point of view to relate the conformational changes occurring in the active site environment with the conformational changes of the whole protein. If the overall results are taking into account, a protein picture emerges that is consistent with a highly cooperative structure and a multistep unfolding pathway.

In this study, the structural and thermodynamic consequences of the removal of the -SS- bond connecting Cys-3 and Cys-26 in azurin are investigated. Site-directed mutagenesis has been used to replace both cysteines with alanine residues. The combined use of different techniques, such as optical absorption, steady-state fluorescence, circular dichroism (CD), and electron paramagnetic resonance (EPR), allowed us to monitor different regions of the protein and to gain more insight on the structural properties of the native state of the Cys-3Ala/Cys-26Ala (C3A/C26A) azurin mutant.

The results show that the spectroscopic properties of the C3A/C26A azurin mutant are very similar to those of wt azurin, indicating that the double mutation prevents neither the correct folding of the protein nor the formation of the metal-binding site.

The thermal unfolding of the C3A/C26A azurin mutant has been followed by differential scanning calorimetry, optical absorption at _max = 625 nm, and fluorescence emission using  = 295 nm as excitation wavelength. Moreover, the geometry of the copper site in the denaturated state has been resolved by EPR spectroscopy.

The analysis of the overall experimental data shows that the thermal transition from the native to the denaturated state of the modified azurin is irreversible and scan-rate dependent as in the wt azurin (La Rosa et al. 1995). However, the removal of the disulfide bridge results in a dramatic reduction of the protein stability. In fact, the midpoint transition temperatures registered by the different techniques for C3A/C26A azurin mutant are about 20°C lower than those previously obtained for wt azurin. Similarly, the Gibbs free energy calculated for the disulfide bond deficient protein is half that of the wt protein, suggesting that the presence of the disulfide bridge is an important contributor to the high stability of the wt azurin.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

Construction, culturing, and isolation of the C3A/C26A azurin mutant

The C3A and C26A azurin mutations were introduced separately in the azurin gene (Canters, 1987) using the oligonucleotide-directed polymerase chain-reaction mutagenesis (Picard et al., 1994). For the single C3A and C26A mutants, the mutagenesis primer, i.e., 5'CTGGCTGCCGAGGCCTCGGTGGACATC-, 5'ATCACCGTCGACAAGAGCGCTAAGCAGTTCACC- (Eurogentec) have been used, respectively. Subsequently, the DNA fragments containing the mutations have been substituted into the wt gene expression vector. Finally, to generate the double mutant, the two DNA fragments containing the single mutations have been ligated, and the intact reading frame encoding the double mutant has been obtained. The plasmid vector pUC 18-derived bearing the C3A/C26A mutant was used for Escherichia coli JM101 transformation and expression. Subsequently, the cells were grown in a fermentor containing 30 L Luria-Bertani medium supplemented with 100 µg/mL ampicillin and 100 µM of IPTG for gene induction. The cells were harvested at the end of the exponential growth phase. In principle, the protocol for C3A/C26A azurin purification could be kept similar to that of the wt azurin (Van der Kamp et al., 1990). The protein was judged pure based on standard I.E.F gel-electrophoresis.

Differential scanning calorimetry and spectroscopic measurements

Differential scanning calorimetry (DSC) scans were carried out with a SETARAM (Lyon, France) microdifferential scanning calorimeter (microDSC) with stainless steel 1-mL sample cells, interfaced with a BULL 200 Micral computer. The sampling rate was 1 point/s in all measuring ranges. The protein (1.25 mg/mL) was dissolved in 10 mM phosphate buffer at pH = 7.03. The ionic strength was adjusted at 0.1 by sodium chloride. Protein solution pH was adjusted by computer-controlled potentiometric titration. Titrations were performed using Metrohm digital pH meter (mod. 654) equipped with Metrohm 109 glass-saturated calomel microelectrode. The titration cell was thermostatted at 25.0 ± 0.2°C, and all solutions were kept under an atmosphere of nitrogen. The same solution without the protein was used in the reference cell. Both the sample and reference were scanned from 30 to 80°C with a precision of ±0.08°C at the scanning rates of 0.3, 0.5, 0.7, and 1°C/min. The calorimetric scans were carried out under an extra nitrogen pressure of 1.5 bar.

The average level of noise was about ±0.4 µW and the reproducibility at refilling was about 0.1 mJ/K/mL. Calibration in energy was obtained by dissipating a defined amount of energy, electrically generated by an EJ2 SETARAM Joule calibrator, within the sample cell.

To obtain the Cp curves, buffer-buffer base lines were recorded at the same scanning rate and then subtracted from sample curves (Sturtevant, 1987; Connelly et al., 1991). All the Cp_exc curves were obtained using a fourth-order polynomial fit.

Optical Density (OD) measurements were carried out with a JASCO 7850 spectrophotometer equipped with a Peltier-type thermostatted cell holder, model TPU-436 (precision ±0.2°C) and the EHC-441 temperature programmer. Quartz cuvettes with a 1-cm optical path were used throughout. The temperature of the samples was measured directly by a YSI precision thermistor dipped in the cuvette. The experiments were started 3 min after sample positioning in the thermostatted sample holder at the initial temperature of 30°C. The heating rates were 0.3, 0.5, 0.7, and 1°C/min. Protein concentration was 0.4 mg/mL.

Fluorescence emission curves were acquired with a Perkin-Elmer LS 50B spectrofluorimeter equipped with a Peltier Temperature Programmer PTP-1. The excitation wavelength was 295 nm, while the excitation and emission band-passes were of 6 and 4 nm, respectively. Temperature was scanned from 30 to 82°C at 1°C/min. The temperature of the samples was measured directly by a YSI precision thermistor dipped in the cuvette. The emission spectra were recorded at the scan speed of 400 nm/min. Protein concentration was 0.4 mg/mL.

CD measurements in the far UV region were performed with a JASCO 700 spectropolarimeter using quartz cuvettes of 1-cm optical path. Azurin concentration was 0.3 mg/mL.

The EPR measurements were carried out with a Bruker ER 200D-SRC X band spectrometer equipped with the ESP 1600 Data System. All the EPR spectra were recorded at 77 K by plunging the sample solutions in a finger dewar containing liquid nitrogen positioned into a TE₁₀₂ cavity. Protein concentration was 4 mg/mL.

All data presented in this paper are the average value of three measurements.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

Spectroscopic characterization of C3A/C26A azurin mutant

Spectroscopic techniques are essential tools in the investigation of the structure and dynamics of proteins. In Fig. 2 a the UV-Vis spectrum of C3A/C26A azurin mutant (dashed line) recorded at room temperature is compared to that of wt azurin (solid line). As can be seen, the spectral features of the two proteins are rather similar with absorption maxima at 280 and 625 nm. The intense absorption in the Vis region is usually assigned to a ligand-to-metal charge transfer transition between the S(Cys-112) and the d_x²-y² orbital of the Cu(II) ion (Solomon et al., 1992). The similarity of the two spectra suggests that the geometry of the copper environment is preserved in the mutated azurin, and the overlap between the two orbitals involved in the transition is maintained.

View larger version (22K): [in this window] [in a new window]   FIGURE 2   (a) UV-Vis spectra of wt (solid line) and C3A/C26A azurin mutant (dashed line) recorded at room temperature. Proteins concentration was 0.4 mg/mL in 10 mM PBS (pH 7.03). (b) Normalized fluorescence emission spectra of wt (solid line) and C3A/C26A azurin mutant (dashed line) in aqueous solution recorded at room temperature through excitation at 295 nm. (c) Far-UV CD spectra of wt (solid line) and C3A/C26A azurin mutant (dashed line) recorded at room temperature. (d) Comparison of the EPR spectra of wt (solid line) and C3A/C26A azurin mutant (dashed line) recorded at 77 K. Proteins concentration was 1 mM in 10 mM PBS (pH 7.03).

Figure 2 b shows the steady-state fluorescence emission spectra of wt (solid line) and C3A/C26A azurin mutant (dashed line) collected through excitation at 295 nm at room temperature. At this excitation wavelength the intensity of fluorescence emission is mainly due to the unique Trp residue of the protein. The spectra, perfectly overlapping, have emission maxima at 308 nm, which is consistent with a hydrophobic environment of the chromophore (Guptasarma, 1997). The similarity of the two spectra suggests that the double mutation does not affect the microenvironment where the emitting Trp residue is located. A similar result is obtained when the Tyr and Phe residues are also excited at 280 nm (data not shown).

In Fig. 2 c, the far-UV CD spectrum of C3A/C26A azurin mutant recorded at room temperature (dashed line) is compared to that of the wt protein (solid line). Both spectra show a strong absorption around 220 nm, which is characteristic of -structure (Mei et al., 1996). No significant deviation from the spectrum of the wt sample is observed in the mutated protein, indicating that both proteins have comparable secondary structure.

Finally, the EPR spectra at 77 K of wt (solid line) and C3A/C26A azurin mutant (dashed line) are shown in Fig. 2 d. The spectral features are typical of a type-1 copper ion with axial symmetry characterized by four hyperfine lines centered at g and separated by A and by a single, more intense resonance line centered at g at higher fields (Aqualino et al., 1991). The two spectra are very similar, although for the C3A/C26A azurin mutant sample the m_I = + resonance line in the parallel region of the spectrum is centered at a slightly higher magnetic field with respect to the corresponding line in the wt protein.

From the results presented in Fig. 2, it can be concluded that both the copper ion and the Trp environments as well as the secondary and the tertiary structure of the -SS- bond-deficient protein are very similar to those of wt azurin. In other words, the native state of the mutated azurin is not affected in a significant way from the replacement of the two cysteines with alanines. A different result has been previously found on azurin by substituting the two cysteines with serines (Bonander et al., 1995). This mutation gives rise to a significant alteration of the azurin folding and of the metal active site geometry. In fact, the type-1 Cu⁺⁺ is converted into a type-2 copper ion and the fluorescence emission change of the Trp residue shows that it becomes exposed to a polar environment. Moreover, the far-UV CD spectrum of such a mutated azurin indicates a loss of -structure. This result may be related to the polar character of the serine residues, which can induce a strong pertubation not only on the mutation site but also for the overall protein conformation. Such a perturbation does not occur when the two cysteines are replaced by alanines as in the present study.

Thermal behavior of the C3A/C26A azurin mutant. DSC

In Fig. 3 a, the calorimetric profiles of the C3A/C26A azurin mutant and wt azurin recorded in the same experimental conditions are shown. The comparison of the two curves evidences two main differences. The first one is that in the mutated azurin the DSC curve is broad and the temperature of maximum heat absorption, T_max, is about 62°C, whereas for the wt protein the same parameter is about 83°C. The broadening of the DSC profile and the reduction of T_max suggest a decrease of the cooperativity of the thermal transition and a strong reduction of C3A/C26A azurin mutant stability with respect to the wt protein, respectively. The second difference concerns the intense exothermic peak present at 87°C in the DSC profile of the wt protein, which disappears in the mutated one. This result suggests some differences in the final states reached by the two proteins. Notwithstanding these differences, the thermal unfolding of C3A/C26A azurin mutant is still irreversible like that of the wt protein, i.e., a second scan of a previously scanned sample did not show any endothermic peak. The irreversibility of the thermal transition requires establishment of whether the reaction is under kinetic control, and whether there is a change of molecularity during the heating of the solution (Sanchez-Ruiz, 1992; Galisteo and Sanchez-Ruiz, 1993; Tello-Solis and Hernandez-Harana, 1995). DSC scans at different scan rates allow us to elucidate the first point, whereas a change of molecularity is prevented because azurin is a monomeric protein.

View larger version (14K): [in this window] [in a new window]   FIGURE 3   (a) DSC thermograms of wt and C3A/C26A azurin mutant recorded under the same experimental condition. Protein concentration: 1.25 mg/mL, pH = 7.03, ionic strength 0.1 in NaCl; scan rate 0.5°C/min. (b) Normalized OD625 versus temperature of C3A/C26A azurin in aqueous solution recorded at a scan rate of 0.5°C/min.

In Table 1 are listed the T_max and the experimental H values as a function of the scan rates. As can be seen, T_max increases with the scan rate. Such an effect testifies to the irreversible character of the thermal denaturation of the protein. From the scan rate dependence of T_max, it is possible to calculate the apparent activation energy, E_app, of the denaturation process by using the equation (Sanchez-Ruiz, 1988),

(1)

where v is the scan rate (°C/min). From the slope of the linear plot of ln(v/T_max²) versus 1/T_max, we obtain E_app = 508 kJ/mol (correlation coefficient is 0.98).

On the other hand, some differences in the values of the denaturation enthalpy measured at varying scan rates are observed (Table 1) in agreement with other works (Sanchez-Ruiz et al., 1988; Grasso et al., 1995; Lyubarev et al. 1998). From a microscopic point of view, the reasons for these differences remain unclear. In principle, the denaturation enthalpy should not depend on the scanning rate, supposing the protein concentration and the initial and final states are identical in all the experiments performed with varying scanning rates. If we assume accurate sample preparation and the similarity of the initial states, it can be hypothesized that the differences registered for H can be attributed to differences in the final states reached by varying the scan rates. Such a situation is realized, for example, when denaturation is accompanied by the formation of aggregates whose characteristics depend on the rate of denaturation, i.e., on the scan rates. The H values obtained for the C3A/C26A azurin mutant (Table 1) are lower than those found for wt protein (La Rosa et al., 1995). This result is in agreement with the idea that the presence of a disulfide bond represents a constraint for the protein structure even in the unfolded state. As a consequence, the hydrogen bonding network in such a system is less favorable than in an unconstrained system. Moreover, the van der Waals' interactions between nonpolar groups, which are highly dependent on the distance, may also be disrupted by the introduction of an -SS- bond (Doig and Williams, 1991). Thus the H of unfolding is expected to increase in the presence of the disulfide bridge.

In general, the main problem concerning irreversible thermal transitions is that they cannot be analyzed in the light of classical thermodynamics. Then, the knowledge of the energetics of the system requires the use of theoretical models that describe the unfolding pathway. In previous papers authored by some of us (Milardi et al., 1994; La Rosa et al., 1995, 1998; Guzzi et al., 1998), we have developed a model to extract thermodynamic information from irreversible calorimetric data. According to this method, which has been already applied to wt azurin, the whole denaturation process of the C3A/C26A azurin mutant can be described as the sum of two steps: the first (unfolding) is reversible and contains thermodynamic details about the energetics of the protein; the second is irreversible and time dependent. The whole denaturation pathway can be summarized as
where N, U, and F are the native, unfolded, and final states, H_U and H_F are the enthalpy associated with the reversible and the irreversible steps, respectively, T_1/2 and T* are the temperatures at which the equilibrium constant, K, and the kinetic constant, k, approach the unity value, respectively. The reversible component can be separated from the irreversible one by using an extrapolation procedure of the Cp_exc curve at infinite scanning rate. According to this procedure, we have obtained H_U = 444 ± 18 kJmol¹ and T_1/2 = 64.72 ± 0.06°C. The details of the extrapolation procedure have been reported elsewhere (La Rosa et al., 1995).

To determine the thermodynamic and the kinetic parameters of the two steps of the denaturation process, the experimental DSC curves of C3A/C26A azurin, obtained at different scan rates, were simulated with the equation,

(2)

where the previously determined quantities H_U and T_1/2 have been used as starting values, R being the gas constant, T₀ is the onset temperature of denaturation, and

(3)

(4)

where E is the activation energy of the irreversible step.

This equation, which originates from the classical Lumry-Eyring model, has been previously improved (Milardi et al., 1994) by considering the enthalpy, H_F, of the U  F process not negligible. The unknown parameters H_F, T* and E have been obtained from the fitting procedure. In Fig. 4 the experimental heat capacity curve of C3A/C26A azurin mutant recorded at 0.5°C/min (solid line) is compared with the corresponding curve obtained by using Eq. 2 (dashed line). The two curves are in good agreement. This suggests that the proposed model provides a reliable description of the thermal denaturation of C3A/C26A azurin mutant.

View larger version (13K): [in this window] [in a new window]   FIGURE 4   Curve fitting with Eq. 2 of the experimental Cpexc profile recorded at 0.5°C/min. The starting values were HU = 444 kJ/mol and T1/2 = 64.7°C.

Optical density

The variation of the intense optical absorption at 625 nm as a function of temperature can be used to directly monitor the local conformational changes occurring in the copper environment when the temperature increases. Because the copper ion is located in a hydrophobic region of the protein, these changes reflect those occurring in the tertiary or secondary structure of the protein. Figure 3 b shows the normalized optical density at 625 nm, OD₆₂₅, in the temperature range 30-70°C of C3A/C26A azurin mutant in aqueous solution recorded at the scan rate of 0.5°C/min. The thermal profile shows a transition temperature, T_t, defined as the midpoint of the OD transition, of 59.5°C. This T_t value is remarkably lower than the corresponding one shown from the wt protein (T_t = 79.3°C) and reflects the strong decrease in stability of the mutated azurin. After the thermal denaturation, the characteristic blue color of the protein is not recovered by cooling back the solution to room temperature. Such an effect can be ascribed to conformational changes of the tertiary structure of the protein, which permanently alter the copper coordination environment (see also the section Electron Paramagnetic Resonance).

The occurrence of kinetic factors in the thermal disruption of the active site of C3A/C26A azurin mutant has been verified by performing OD₆₂₅/T measurements at different scan rates. The T_t values obtained at 0.3, 0.5, 0.7, and 1.0°C/min are listed in Table 1. As can be inferred from the data, T_t increases with the scan rate according to the trend observed in the DSC scans. The comparison of T_t and T_max shows that T_t has always a slight lower value with respect to T_max. The difference, T = T_max  T_t > 0 for each scan rate, suggests that the disruption of the active site precedes the whole protein denaturation.

The scan rate dependence of T_t can be used to calculate the activation energy, E_a, of the denaturation process using Eq. 1, where E_app and T_max are now substituted by E_a and T_t, respectively. The result of the fit of the experimental points with Eq. 1 gives an E_a value of 460 ± 8 kJmol¹, which is comparable, within the experimental error, with the value registered for the wt protein (La Rosa et al., 1995).

Electron paramagnetic resonance

Figure 5 shows the EPR spectra of C3A/C26A azurin mutant in aqueous solution recorded at 77 K after the protein solution has been heated at 75 (curve a) and 90°C (curve b) for 10 min. The first temperature has been chosen above the end of the DSC thermal transition. The EPR spectrum, when compared to the magnetic signal recorded on the native state of C3A/C26A azurin mutant (Fig. 2 d, dashed line), shows a strong temperature-induced effect both on the copper-ligands geometry and coordination atoms. In fact, the copper-ligand geometry undergoes a transition from trigonal bipyramidal in the native state to square planar in the final state as testified from the change of the A_// value from 55 to 175 Gauss. Moreover, in the high magnetic field region of the spectrum, traces of a superhyperfine structure are evident (Fig. 5, curve a). A better resolution of this superhyperfine structure can be observed in the spectrum recorded after heating the C3A/C26A azurin mutant solution at 90°C (Fig. 5, curve b). The superhyperfine structure arises from the interaction of the unpaired electron spin of the Cu⁺⁺ ion with the nuclear spin of the ligand atoms (McGarvey, 1967). This result is quite different from that previously obtained for wt protein (see Fig. 8 in La Rosa et al., 1995). In fact, although in both cases the final geometry can be assimilated to square planar, an increase in the hyperfine splitting from 160 Gauss in the denaturated wt azurin to 175 Gauss in denaturated C3A/C26A azurin mutant is registered. The differences observed in the EPR spectra of the denaturated state of wt and C3A/C26A azurin mutant suggest that the mutation affects the final conformational state reached by the protein in absence of the disulfide bridge. In particular, the presence of at least nine superhyperfine resonance lines suggests that there are four nitrogen atoms in the copper coordination sphere of the denaturated C3A/C26A azurin mutant. In contrast to this, two N and two O ligands had been proposed for the copper coordination of the denaturated state of wt protein (La Rosa et al., 1995).

View larger version (14K): [in this window] [in a new window]   FIGURE 5   EPR spectra of C3A/C26A azurin mutant recorded at 77 K after heating of the sample solution for 10 min at 75°C (curve a) and 90°C (curve b).

Fluorescence emission

The fluorescence emission signal of the Trp residue in a protein is closely related to its solvent exposure (Lakowicz, 1986), so that the Trp emission spectra can provide useful information about the changes of the protein conformational states. The fluorescence spectra of the wt and C3A/C26A azurin mutant collected through excitation at 295 nm have been recorded in steps of 1°C in the temperature range 30-82°C. The inset of Fig. 6 shows, as an example, the fluorescence spectra of C3A/C26A azurin mutant recorded at 30 and 62°C. At low temperature, the maximum emission is at 308 nm, whereas at high temperature it occurs at 357 nm. The same results are also observed for wt azurin at 30 and 80°C, respectively. The first wavelength is compatible with a Trp residue completely buried in the interior of the protein and surrounded by nonpolar amino acid residues (Lakowicz, 1986). The second one being compatible with the fluorescence emission of a Trp residue exposed to the aqueous solvent (Guptasarma, 1997).

View larger version (20K): [in this window] [in a new window]   FIGURE 6   Temperature-dependent fluorescence emission of wt and C3A/C26A azurin mutant at 308 and 357. exc = 295 nm. Protein concentration was 0.4 mg/mL in 10 mM PBS (pH 7.03). Inset: Fluorescence emission spectra of C3A/C26A azurin mutant recorded at 30 (curve a) and 62°C (curve b). The vertical axis of the fluorescence intensity on the left refers to sample (a) and that on the right to sample (b).

In Fig. 6 the variations of the fluorescence emission intensity at 308 and 357 nm as a function of temperature for wt and C3A/C26A azurin mutant are reported. As can be seen, the effect of the temperature on the fluorescence intensity is similar in both proteins. In fact, with temperature increase, a reduction of the fluorescence intensity at 308 nm is observed, then the maximum emission wavelength shifts at 357 nm, and an increase in the fluorescence intensity occurs, too. However, the transition temperature, defined as the midpoint of the fluorescence variation at 308 nm, is 78°C for wt protein and 58°C for the mutated azurin. These findings further support the results obtained using the other experimental techniques and confirm the strong reduction of stability of azurin upon removal of the -SS- bond.

Because the solvent accessibility is the major factor determining the fluorescence of the Trp residue, the data in Fig. 6 suggest that the fluorescence change may reach its limiting value before both proteins fully unfold. In fact, the temperatures of 78 and 58°C, determined by means of fluorescence emission for wt and C3A/C26A azurin mutant, respectively, are lower than those extracted from the DSC profiles (Fig. 3 a). In addition, the comparison between the fluorescence emission and optical absorption data reveals that the solvent exposure of the Trp residue occurs before the disruption of the copper site.

It is noteworthy that the Trp fluorescence intensity at 357 nm increases upon unfolding. Moreover, the fluorescence intensity of C3A/C26A azurin in the denaturated state is much higher than the corresponding one of the wt protein. This result supports the hypothesis that the conformational states of the two proteins after the thermal denaturation are quite different. In fact, because the Trp fluorescence in holo-azurin is extensively quenched compared to that of copper-depleted azurin (Petrich et al., 1987; Hansen et al., 1990; Sweeney et al., 1991), it turns out that the fluorescence emission intensity gain of the C3A/C26A azurin mutant may be related to a marked increase in the Cu-Trp-48 distance. On the other hand, this finding is consistent with EPR data that show a tight binding of the copper ion with ligand atoms in the denaturated-mutated azurin, which are different with respect to the wt protein.

Thermodynamic analysis of the unfolding process

The calculation of the Gibbs free energy relative to the unfolding process in all the temperature range considered requires the knowledge of three parameters: H_U, T_1/2, and Cp. Unfortunately, the irreversibility of the thermal transition of the C3A/C26A azurin mutant prevents the experimental determination of Cp = Cp_U  Cp_N, because the value of Cp at the offset temperature is ascribable to the final (F) state and not to the unfolded (U) one. To solve this problem, both alternative experimental methods and theoretical models can be used. In this paper, the calculation of Cp has been obtained by three different approaches. In the first one, the denaturational Cp is obtained from the temperature dependence of the unfolding enthalpy H_U using the Kirchhoff equation. Several sets of calorimetric experiments have been carried out at different pH values from 4.8 to 7.0. Then the H_U and the T_1/2 values, related to the thermally induced unfolding of C3A/C26A azurin mutant, have been calculated following the extrapolation procedure at infinite scanning rate (La Rosa et al., 1995). The temperature dependence of the extrapolated unfolding enthalpies H_U obtained at different pH values is linear with T_1/2. The slope of the linear fit gives 8.9 ± 0.9 kJK¹mol¹ for the heat capacity changes upon unfolding of C3A/C26A azurin mutant in H₂O.

The second approach consists in the use of the Murphy and Gill (1991) model. According to this model, the Cp value can be evaluated on the basis of the set of equations,

(5)

(6)

(7)

(8)

(9)

(10)

where Cp_ap is the denaturational heat capacity change ascribable to apolar groups, Cp_pol is the heat capacity change ascribable to polar groups, f_ap is the fraction of apolar buried surface area, N_res is the number of the residues in the protein, N_CH is the number of apolar hydrogen atoms (i.e., the hydrogen atoms directly bound to a carbon atom), (C⁰)p_-CH- is the specific contribution of one mole of apolar hydrogens to the overall denaturational heat capacity change and (C⁰)p_-CONH- is the specific contribution ascribable to one mole of polar residues (Murphy and Gill, 1991). Because there are 759 apolar hydrogen atoms and 128 aminoacid residues in C3A/C26A azurin mutant, Cp is 8.5 ± 1.0 kJK¹mol¹.

The third method is based on the average properties of globular proteins (Milardi et al., 1997). This method starts from the correlation of Cp with N_res and N_CH of well-known globular proteins at various temperatures. According to this method, we obtain a Cp value of 7.9 kJK¹mol¹ and in the temperature range 20-100°C the Cp variation is ±1.0 kJK¹mol¹.

To minimize the errors, ascribable either to experimental uncertainties or to the nonperfect efficiency of the models, the Cp used to calculate the thermodynamic functions has been taken as the average of the three values, which corresponds to 8.4 ± 1.0 kJK¹mol¹. This value is also in good agreement with the Cp value calculated according to the equation (Doig and Williams, 1991),

(11)

where N_ss is the number of disulfide bridges. The Cp obtained is 8.9 kJK¹mol¹. The denaturational heat capacity change, together with T_1/2 = 64.72°C and H_U = 444 kJmol¹ values, allow us to calculate H(T), S(T), and G(T), characterizing the thermal behavior of C3A/C26A azurin mutant by using the equations

(12)

(13)

(14)

The three functions are drawn in Fig. 7. In the same figure, the previously thermodynamic functions related to the thermal denaturation of wt azurin in H₂O (La Rosa et al., 1995) are also shown.

View larger version (16K): [in this window] [in a new window]   FIGURE 7   Temperature dependence of the unfolding (a) enthalpy, (b) entropy, and (c) Gibbs free energy for wt (solid line), C3A/C26A azurin mutant (dashed line). The thermodynamic functions expected using the Murphy and Gill model (see the text for details) are also shown (dotted line).

By comparing the two G functions (Fig. 7 c), it can be seen that the temperature of maximum stability is about 20°C in both cases. Moreover, G = G_wt  G_-SS-  27 kJmol¹ is constant over the whole temperature range (the subscript indexes wt and -SS- refer to wt and disulfide-deficient azurin). This result suggests that the presence of the disulfide bridge in the wt azurin contributes significantly to the stability of the protein and is not temperature dependent. If we consider the enthalpic contribution (Fig. 7 a) to G, it can be seen that H_wt and H_-SS- coincide in all the temperature range. The close similarity of the unfolding enthalpy value of the two proteins suggests that the reduced stability of C3A/C26A azurin mutant arises mainly from entropic effects. This result is in agreement with a DSC study of wt and three-disulfide lysozyme derivative (Cooper et al., 1992). An opposite conclusion has been proposed by Doig and Williams (1991) based on the correlations between the thermodynamic properties and the number of the disulfide bridges present in six proteins.

A different behavior is observed in the entropic component. S_wt is less than S_-SS- over the whole temperature range (Fig. 7 b). In particular, at 20°C, the entropy of unfolding of C3A/C26A azurin mutant is about 115 JK¹mol¹ higher than that of the wt protein (Table 2). This is consistent with the increase in the conformational entropy in the unfolded polypeptide chain resulting from removal of the -SS- cross-link (Vogl et al., 1995). Based on a series of studies on a large variety of proteins, it has been shown that this entropic contribution depends on the size of the loop connecting the -SS- bond according to the equation (Pace et al., 1988; Vogl et al., 1995),

(15)

where R is the gas constant and n is the number of residues in the loop. In azurin, the removal of the disulfide bridge should provide an increase in conformational entropy of 47.9 JK¹mol¹. Experimental data reported in Table 2 show that the removal of the disulfide bridge in azurin implies an entropy gain of 115 JK¹mol¹ on unfolding. By subtracting from this value the conformational entropy, we obtain the residual entropy, S_res, which is 67.1 JK¹mol¹. This residual entropy can be ascribed both to solvent-related effects that operate via a decrease in the water-exposed nonpolar surface area of the unfolded protein and to changes in the Cu coordination sphere in the unfolded mutant and wt protein (see Fig. 5). For some proteins, it has been shown that S_res can be calculated by means of the equation (Doig and Williams, 1991),

(16)

According to this point of view, the S_res for azurin should be 34 JK¹mol¹, which is lower than the expected value (67.1 JK¹mol¹). Hence, we believe that, in the case of azurin, other factors contribute positively to the entropic increase after removal of the disulfide bridge. We suggest that additional entropic factors can arise from a decrease of the entropy of the native state of the mutated azurin as a consequence of a modified hydrogen bonds network and nonpolar intramolecular interactions due to the removal of the disulfide bridge.

View this table: [in this window] [in a new window]   TABLE 2   Thermodynamic parameter values of wt (Milardi et al., 1996), C3A/C26A azurin mutant and calculated (Model) by using Murphy and Gill method at the temperature of maximum stability (20°C)

At a first approximation, the effect of the metal ion on unfolding can also be considered by comparing the G values calculated for the C3A/C26A azurin mutant and the one expected from applying the Murphy and Gill (1991) model, where the average thermodynamic features of proteins are considered only in terms of polar and apolar groups, i.e., neither the metal ion nor the -SS- bridge are considered. From the curves shown in Fig. 7 c it seems that the simultaneous absence of both the disulfide bridge and the Cu⁺⁺ ion makes the protein more stable by about 6 kJmol¹ with respect to the absence of the only -SS- cross-link. Structural details can be invoked to explain the different behavior of the thermodynamic functions, in particular of the entropic destabilization, in the mutated azurin with respect to the wt protein. The native structure of azurin is very compact, and as a consequence, the introduction of a disulfide bond in such a structure may induce only negligible effect on the entropy of the native state of the protein. In contrast, the number of the possible conformations that the protein can assume in the final state may be significantly lower if an -SS- bond is present as in the wt azurin. As a consequence, S_wt < S_-SS-. This hypothesis is also supported by the EPR and fluorescence emission measurements where differences in the final state of wt and C3A/C26A azurin mutant have been evidenced.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

The results presented in this study show that the double substitution of Cys-3 and Cys-26 with alanine residues does not affect the spectral properties and the most structural features of C3A/C26A azurin mutant with respect to the wt protein, whereas the thermodynamic consequences of the removal of the disulfide bond are considerable. The reduced stability of the C3A/C26A azurin mutant is evident from both the downward shift of the transition temperature from the native to the denaturated state of the protein as well as from the decrease of the G value. This reduction has mainly an entropic character. The presence of the disulfide bridge stabilizes the protein in two ways: 1) it decreases the conformational entropy of the unfolded state, and 2) it increases the surface area exposed to the solvent on unfolding of the nonpolar residues of the protein. In the case of azurin, these two entropic effects have almost the same weight, whereas the removal of the disulfide cross-link has little effect on the enthalpy of denaturation of the protein.

The different techniques used in this paper also allow us to hypothesize a sequence of events in the thermal denaturation process of azurin, which starts with the destabilization of the Trp environment, proceeds through the disruption of the copper site, and then the whole protein molecule collapses in the denaturated state.