HYDRATION AND PROTEIN SUBSTATES: FLUORESCENCE OF PROTEINS IN REVERSE MICELLES

SUMMARY The fluorescence properties of indole derivatives, lysozyme and azurin were investigated in reverse micelles of detergent sodium bis[2-ethyl-hexllsulfosuccinate (Aerosol OT)* in n-hexane. L-tryptophan, l-methyl-trypto- phan and n-acetyl-1-tryptophanamide exhibited complex fluorescence decays in reverse micelles. Fluorescence decays were best described using Gaussian bimodal distributions of lifetimes. Increasing hydration levels in the micelles resulted in a decrease in decay heterogeneity, as indicated by a large decrease in lifetime distribution widths. Steady-state polarization and fluorescence emission measurements indicated both an increase in average polarity of the environment around the indole derivatives and an increase in the mobility of the probes with increasing hydration levels. The fluorescence decays of lysozyme and azurin in reverse micelles were also found to be very complex and were described with Gaussian lifetime distributions. Increasing water content in the micelles caused marked decreases in both center and width of the lifetime distributions for these two proteins. Steady-state polarization measurements as a function of the extent of hydration revealed an increase in the average rotational rates of the tryptophan residues in lysozyme upon increasing water content. Thus, static polarization and lifetime measurements indicate that the amount of water present in the micelle may influence the amount of structural flexibility of the polypeptide chains and the rates of interconversion between conformational substates. *Abbreviations


INTRODUCTION
The first comprehensive review on protein motions was by Careri,Fasella and Gratton (ref. 1). The focus was on the characterization of the time scale of different protein time events. The underlying idea of that work was that hydration of the protein matrix is required for most protein motions and that the network of hydration would serve as a common medium for the connection between solvent and protein dynamics. Since that original work, the amount of experimental evidence for protein motions has enormously increased. Protein dynamics is now a well accepted reality of protein systems. Theoretical methods based on molecular dynamics have been used to describe, although for few hundred picoseconds, the motion of each atom of a protein (ref. Despite the amount of research and the advances on protein dynamics, there are still some basic aspects that need to be understood. Among those aspects is the role of hydration. Essentially, the problem can be explained using the following highly simplified description (ref. 8).
(1) At room temperature, in the absence of the hydration shell, the protein is essentially frozen. There is very little motion occurring in the time range of 1 ns and longer.
(2) As water is gradually added to the protein, nothing major occurs with respect to the protein dynamics until about 0.2 mg of water per mg of protein. At this value there is a sudden transition and all dynamics reach the so-called "solution value" at about 0.25 mg water per mg protein. The effect of water is minimal on the protein structure, but is paramount on the dynamics (ref. 9). The effect of hydration on protein dynamics has been addressed very little in published literature and its physical origin is virtually unknown. This investigation presents a study of protein hydration using reverse micelles, which can contain a single protein molecule in a microscopic pool. The total amount of water in the pool can be easily controlled.
In the following, we present the relevant information on the use of reverse micelles for hydration studies.
Dissolution of amphiphiles in organic solvents may give rise to formation of reverse micelles (for reviews, see refs. 10, 11). These inverted micellar solutions are capable of taking up relatively large amounts of water, forming "water pools" in the core of the micelles (Fig. 1). The thermodynamics of micelle formation (refs. 12, 13) as well as the characterization of phase-diagrams and Fig. 1 a) interfacial, in contact with the AOT side-chains in the interface between the organic solvent and water; b) in the bound-water layer; and c) in the water pool.
For hydrophilic molecules solubilized in reverse micelles the partition of the molecule in the interfacial region presented by the hydrophobic AOT side chains is probably negligible. Thus, for hydrophilic molecules, the reverse micellar systems essentially reduce to a two-state system: that of the bound water layer and that of the water pool (bulk-like water).
Over the past few years, a large number of small molecules and biopolymers have been solubilized in reverse micelles of AOT in hydrocarbon solvents.
The study of proteins in reverse micelles has received a great deal of attention We present steady-state and lifetime measurements of the single tryptophan protein azurin and of the six-tryptophan protein lysozyme in reverse micelles of AOT in hexane. Our results show that at low water content, the fluorescence decays of these two proteins are very complex and broad distributions of lifetimes are needed to describe the decay. As the water content is increased the center of the distributions moves to shorter lifetimes toward the value observed for the proteins in solution. The width of the distribution is also sensitive to hydration, and decreases as w0 is increased. However, even at the highest hydration values attainable in reverse micelles (w0 = 70) the recovered parameters for the lifetime distributions are still far from the solution values.
Comparison of steady-state polarization and lifetime data for lysozyme in reverse micelles indicates increased average mobility of the tryptophan residues with increasing hydration levels. In

MATERIALS AND METHODS
Excitation and emission spectra were recorded on an ISS, Inc.
(Champaign, IL> GREG PC photon counting spectrofluorometer (4 nm slits in both excitation and emission). Spectral center of mass was calculated as I hI(hldh xav = I I(h)dh ' and the integrated emission intensity was calculated as

Itotal = 1 IWdk
where h and I(h) are the emission wavelength and the intensity at a given wavelength, respectively. Static anisotropy measurements were performed in the same instrument using the L format and were corrected by subtracting the background signal from blanks which did not contain the fluorophore.

Fluorescence lifetime wurements
The fluorescence decays of indole derivatives (tryptophan, l-methyltryptophan, and N-acetyl-L-tryptophanamide), as well as of azurin and lysozyme, were measured in a phase fluorometer using the harmonic content of a Spectra-Physics series 3000 Nd-YAG laser or, in some cases, a Coherent Antares model laser. The light fmm the mode-locked laser was used to pump a Rhodamine 6G dye laser which was cavity-dumped and externally frequency-doubled, providing 5 to 15 mW light at the excitation wavelength of 295 nm. Fluorescence emission was observed through a WG-335 filter. Color errors due to photomultiplier response were minimized by use of a reference solution of p-terphenyl in cyclohexane (lifetime 1.0 ns> in the measurements. Phase and modulation data were acquired at 8 to 10 frequencies in the 2 to 200 MHz range, with uncertainties of ti.2" and rtO.004 for phase angles and modulation ratios, respectively. Data were analyzed with a sum of exponential decays or, alternatively, with continuous lifetime distributions. In both cases, the analysis software provided by ISS, Inc. included the above errors in the calculation of x2, which was used to judge the goodness of the fits. The relevant equations concerning the use of lifetime distributions and a discussion of the resolvability of the distributions with respect to exponent&& were presented by Alcala et al. (ref. 26).
Hvdr&.n studies Reverse micelles of Aerosol OT were prepared by dissolution of the detergent in n-hexane to a final concentration of 0.11 M. Concentrated stock solutions of the indole derivatives or proteins were prepared in deionized water at pH 7-6.
A small aliquot (2 to 8 ~1) of sample was injected into the fluorescence cuvette containing 3 ml of reverse micellar solution. Water was added to give the desired concentration and the cuvette was gently hand-shaken for 30-60 set until complete clarification. Hydration levels are expressed as overall concentration of water in the cuvette or as w0, the ratio between molar concentrations of water and AOT. Temperature was kept at 30°C using a thermostated sample holder.

Chemicalg
The n-hexane used was of spectroscopic grade from Merck (Darmstadt, Germany   suggesting that the average polarity in the water pool at high w. was similar to that of aqueous solution. Similar results were found for I-tryptophan in reverse micelles (not shown).
Steady-state anisotropy measurements of NATA as a function of w. are shown in Figure 2b. At low hydration, the anisotropy of the probe was high (about 0.09 in the w0 range 0.34 to 5.0). This high anisotropy indicated that the rotational motion of the probe was severely restricted in this situation, possibly suggesting that the probe was attached to the micelle wall. As w. was increased above 5.0, the anisotropy decreased to a value of about 0.06 at the highest water content investigated. This could mean that a solubilization process occurred, with the probe going from the micelle wall to the inner water pool. However, it should be noted that even at a w0 of 60 (close to the maximum amount of water that can be taken up by the micelles) the anisotropy value was still far above the value of 0.015 found for NATA in solution (Fig. 2b, dashed- Tables I, JI, and III. shows the recovered bimodal Gaussian lifetime distributions for the three indole derivatives investigated. Figure 3 shows that for the three compounds the fluorescence decay could be almost entirely described by a major lifetime distribution component centered at 1.7-2.0 ns, depending on hydration level. In the case of NATA (Fig. 3a)  throughout hydration levels and allowing the fractional amplitudes to vary to provide the best fit of the data. Tables I, II, and III show the results obtained with the global analysis for NATA, I-methyl-tryptophan and L-tryptophan, respectively. For the three compounds, the decay presented a major component of 3.1-3.4 ns and a smaller component of 0.9-1.3 ns. In all cases, the fractional intensity of the major long-lived component appeared to decrease up to a w0 of 10-15, and then displayed a tendency to increase again at higher hydration levels (which was more pronounced in the measurements with NATA, Table I). Thus, our results do not support a simple exchange of the fluorescent probes between two different environments (namely, bound and free probes) with increasing hydration levels. In order to more easily assess the effects of hydration on the fluorescence decay of NATA in reverse micelles, we analyzed our lifetime data with a unimodal Gaussian distribution of lifetimes.
The results thus obtained are shown in Figure 4. Increasing hydration produced interesting effects on the center of the lifetime distributions of NATA in reverse micelles (Fig. 4a) solution value. .

Fluorescence emission snectra of uroteins m reverse r&&g
Fluorescence emission spectra were acquired for azurin and lysozyme in reverse micelles as a function of w6. Figure 5 illustrates the result obtained for the center of gravity of the emission for lysozyme as a function of wo. Increasing w. from 1 to 60 produced a red shift in the spectral center of mass from 342 nm to 351 nm, thus approaching the solution value of 352 nm (dashed line). This effect is qualitatively similar to that obtained for NATA or tryptophan (Fig. 2) and probably reflects increased polarity of the tryptophan environment upon hydration of the protein.  In order to assess the influence of hydration on the dynamics of the protein matrix, steady-state anisotropy measurements were performed as a function of water content for lysozyme in reverse micelles. Figure 7a shows that the static anisotropy of lysozyme was mostly constant up to w0 = 25 and then increased slightly up to w. = 50. Since the behavior of the average lifetime of lysozyme as a function of hydration was known (Fig. 6a), it was possible to calculate an average rotational correlation time for lysozyme in reverse micelles. The average rotational correlation time was calculated from the Perrin equation as where o is the average lifetime, r. is the limiting anisotropy for indole at &zc = 300 mn (r. = 0.3 according to Valeur and Weber,ref. 321, and r is the measured anisotropy. Figure   7b shows that the average rota- The fluorescence properties of model indole compounds in reverse micelles were found to be markedly dependent on the amount of water present. Steadystate measurements indicated that the polarity of the environment around the indole moiety increased with increasing hydration (Fig. 2a). On the other hand, polarization measurements suggested that at low hydration the indole derivatives were partially attached to the micelle wall, and increasing hydration resulted in increased probe mobility (Fig. 2b). This feature could, in principle, be attributed to a simple solubilization of the probe in the water pool, but this interpretation, as discussed below, is not compatible with our time-resolved measurements.
The fluorescence decay of indole derivatives in reverse micelles was complex. Global analysis of the datasets obtained under various hydration conditions in terms of a two-state model gave incompatible results with a single solubilization process (Tables I, II, and III). For the three probes tested, the fits yielded two lifetime components, a major component at 3.  Table III, which cover a larger w0 range). This implies that above w0 = 15 the solubility of the probe in the water pool would be decreased. Given the inadequacy of the exponential analysis to describe the system, we have analyzed the data with continuous lifetime distributions. For the three probes tested, the decay could be well described by bimodal Gaussian lifetime distributions (Tables I, II, and III). is interesting that very similar lifetimes were found both at the lowest possible hydration regime in reverse micelles and in aqueous solution. One possible explanation for this is that the structure of water in the hydration shell around the indole may limit the efficiency of quenching of indole fluorescence. Thus, at very low hydration there is simply not enough water to quench the fluorescence;.
as hydration increases in reverse micelles, a structured water layer is formed covering the polar heads of the detergents and also the fluorescent probe, which at low water levels remains attached to the micelle wall (Fig. 2b). Several methods have been previously employed to characterize the structure of water in reverse micelles (refs.19-231, revealing that indeed a highly structured water is present up to an overall concentration of about 1 M (i.e., wo e 10). This structured water layer could be optimally organized to quench the fluorescence of the indole derivatives. Further additions of water result in a progressive disorganization of the structure of water, which could lead to less effective quenching.
The limit situation would be the aqueous solution, where the structure of water could be too loose to effectively quench the indole derivatives.
Previous studies on the fluorescence of perylene derivatives and of rose bengal in AOT reverse micelles (refs. 29,30) indicated that a simple solubilization process occurred upon increasing water levels. As discussed above, this conclusion is not supported by our results on indole derivatives, which may indicate that the structure, rather than just the average polarity of the surrounding water, may affect the fluorescence decay of the indole moiety.
An alternate explanation for our findings could arise from the compart- We extended our studiesto investigate the fluorescence decay of proteins in reverse micelles. The decays were found to be very complex and were described by unimodal Gaussian lifetime distributions.  used lifetime distributions to describe complex fluorescence decays in proteins. Broad lifetime distributions were found for even single-tryptophan proteins, and the width and center of the distribution appeared to be related to the interconversion rates between different conformational substates in proteins (refs. 28,35).
Our results for azurin and lysozyme in reverse micelles showed that increasing water resulted in a decrease in both the center and the width of the lifetime distributions (Fig. 6). These results likely indicate faster interconversion rates between conformational substates (ref. 35) as hydration increased in the micelles. It should be noted that the lifetime parameters for the two proteins tested decreased continuously with increasing hydration (Fig. 6), and did not display the bell-shaped water-dependence of the indole derivatives (Fig. 4a). Both the average lifetime and distribution width for the two proteins decreased toward their solution values with increasing hydration but even at the highest attainable hydration level in the micelles the lifetime parameters were still far from reaching their solution values.
Further evidence of increased protein dynamics as hydration proceeded in reverse micelles was obtained from anisotropy measurements which enabled us to calculate average rotational correlation times for the tryptophan residues in lysozyme at various hydration levels (Fig. 7). Our results are, thus, in qualitative agreement with previous reports on the effects of hydration on several dynamic properties of lysozyme (ref. 9), which showed that, at low hydration, the protein presented a rigid structure and that increasing hydration sharply increased the dynamics of the protein matrix.
The activities of several enzymes have been studied in reverse micelles as a function of w6 (for reviews, see refs. 11 and 24). At very low wo, the measured activities were low; increasing w. results in an overshoot of activity at w. around 10-15 and the activity decreases again towards its aqueous solution value with further additions of water. This feature was interpreted as resulting from hyperactivity of the enzymes entrapped in reverse micelles, and several models have been suggested to account for this behavior (ref. 36). These models take into account, for example, the partition of substate molecules between the organic solvent and the water pool inside the micelles, and it has also been suggested that the shape of the micelles, as well as the presence of electric fields arising from the negatively charged detergent polar heads may influence activity (ref. 11). The overshoot of activity at intermediate w. is apparently not compatible with previous studies (ref. 9) which showed that all measured physico-chemical properties of lysozyme, including enzyme activity, gradually approached the aqueous solution value as hydration was increased. In addition, our fluorescence measurements also indicate a progressive and monotonic increase in the flexibility of the protein matrix with increasing hydration. The correlation between enzyme activity and flexibility of the polypeptide chain in reverse micelles should provide information on the role of hydration in catalysis. We are currently developing a model which takes into account the compartmental nature of the micellar solutions, and which may help explain some of the apparent incompatibilities between our results on the dynamics of proteins and their activities in reverse micelles