Folding intermediates of a model three-helix bundle protein. Pressure and cold denaturation studies.

The stability and equilibrium unfolding of a model three-helix bundle protein, alpha(3)-1, by guanidine hydrochloride (GdnHCl), hydrostatic pressure, and temperature have been investigated. The combined use of these denaturing agents allowed detection of two partially folded states of alpha(3)-1, as monitored by circular dichroism, intrinsic fluorescence emission, and fluorescence of the hydrophobic probe bis-ANS (4,4'-dianilino-1,1'-binaphthyl-5,5'-disulfonic acid). The overall free-energy change for complete unfolding of alpha(3)-1, determined from GdnHCl unfolding data, is +4.6 kcal/mol. The native state is stabilized by -1.4 kcal/mol relative to a partially folded pressure-denatured intermediate (I(1)). Cold denaturation at high pressure gives rise to a second partially (un)folded conformation (I(2)), suggesting a significant contribution of hydrophobic interactions to the stability of alpha(3)-1. The free energy of stabilization of the native-like state relative to I(2) is evaluated to be -2.5 kcal/mol. Bis-ANS binding to the pressure- and cold-denatured states indicates the existence of significant residual hydrophobic structure in the partially (un)folded states of alpha(3)-1. The demonstration of folding intermediates of alpha(3)-1 lends experimental support to a number of recent protein folding simulation studies of other three-helix bundle proteins that predicted the existence of such intermediates. The results are discussed in terms of the significance of de novo designed proteins for protein folding studies.

Understanding the mechanisms by which a polypeptide adopts a stable and functional three-dimensional structure still represents a challenging problem (1). The folding of small proteins usually takes place on timescales close to a millisecond or less, and is believed to occur in a highly cooperative fashion without the presence of populated folding intermediates (2)(3)(4)(5)(6)(7)(8). However, recent simulation studies have suggested the existence of intermediate states during the folding of a small model three-helix bundle protein (9 -12). Three-helix bundles represent a simple folding motif found in a variety of soluble and membrane proteins, including spectrin (13) and the extramembranous portion of Staphylococcus aureus protein A (14). Using sequence patterns discovered in coiled coils, the synthesis of amphiphilic ␣-helices that self-assemble into three-or fourhelix bundles stabilized by a hydrophobic core has been successfully achieved (15)(16)(17). The de novo design of proteins represents a versatile tool to gain insight into the interplay of forces resulting in conformational stability. Artificial proteins are generally less complex than their native counterparts but at the same time retain the features responsible for the folding process.
Recently, Johansson and co-workers (18) reported the synthesis and initial characterization of a native-like three-helix bundle protein, designated ␣ 3 -1. The three different helices of this 65-amino acid polypeptide are joined by (glycine) 4 linkers. NMR solution studies revealed a well structured conformation with ␣-helical secondary structure. GdnHCl 1 -induced unfolding of ␣ 3 -1 followed by CD revealed a Gibbs free-energy of unfolding of ϩ4.6 kcal/mol (18), comparable to that observed for small monomeric natural proteins of similar size, such as myoglobin (7.6 kcal/mol) (19) or the 43-amino acid residue peripheral subunit-binding domain of the pyruvate dehydrogenase complex (3.1 kcal/mol) (20).
In the last two decades, hydrostatic pressure has been extensively used as a reversible thermodynamic variable to characterize subunit association in oligomeric proteins (21,22). In general, unfolding of monomeric proteins requires significantly higher pressures (i.e., 5-7 kilobars (kbar)) than those required for subunit dissociation of oligomers (typically up to 3.5 kbar) (21). Few examples to date demonstrate denaturation of monomeric proteins at pressures below 3 kbar. In the present study, we have used a combination of hydrostatic pressure (up to 3.5 kbar) and low temperatures to investigate the folding stability of ␣ 3 -1. Interestingly, our results revealed the existence of partially (un)folded intermediate states of ␣ 3 -1, giving support to the predictions from the above mentioned simulation studies. Bis-ANS binding studies revealed the existence of significant residual hydrophobic structure in the pressure-denatured and especially in the cold-denatured state of ␣ 3 -1, suggesting molten globule-like conformations for these intermediates.

EXPERIMENTAL PROCEDURES
Chemicals-All reagents were of the highest analytical grade available. Distilled water was filtered and deionized through a Millipore water purification system. Bis-ANS was from Molecular Probes (Eugene, OR).
Peptide-The design, synthesis, and purification of ␣ 3 -1 have been previously described (18). The amino acid sequence of ␣ 3 -1 is given below in single-letter codes. The sequence is arranged in heptads, with the coiled-coil heptad positions labeled a through g (Table I). The N terminus of the peptide is acetylated, and the C terminus is amidated.
Fluorescence Measurements-Unless otherwise indicated, fluores-* This work was supported by National Institutes of Health Grant GM55876 (to J. S. J.) and by a Howard Hughes Medical Institute International Research Scholar Award (to S. T. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § A visiting professor at the Universidade Federal do Rio de Janeiro. ʈ To whom correspondence should be addressed. Tel.: 5521-270-5988; Fax: 5521-270-8647; E-mail: ferreira@bioqmed.ufrj.br. cence emission spectra were measured at 25°C on a spectrofluorometer (PC 1, ISS Inc., Champaign, IL). For intrinsic fluorescence measurements, excitation was at 280 nm and emission spectra were recorded from 300 to 420 nm. Bis-ANS fluorescence was measured with excitation at 375 nm and emission from 420 to 600 nm. Fluorescence measurements under pressure were performed using a pressure cell similar to that originally described by Paladini and Weber (23), equipped with sapphire optical windows. The temperature of the pressure cell was controlled by means of a jacket connected to a circulating bath and was monitored by a telethermometer. All experiments were carried out in 20 mM Tris-HCl, pH 7.4, containing 130 mM NaCl. Protein concentration in all experiments was 2 M (determined with a UV-visible Ultraspec 2000 spectrometer (Amersham Pharmacia Biotech) using ⑀ 280 ϭ 5700 M Ϫ1 cm Ϫ1 ), and bis-ANS concentration, when used, was 1.0 M. All samples were deoxygenated by bubbling with a stream of nitrogen for 5 min prior to the experiments. In low temperature experiments, the windows of the pressure bomb were flushed with nitrogen to prevent condensation.
Fluorescence spectral centers of mass (intensity-weighted average emission wavelengths, av ) were calculated with software provided by ISS Inc. as follows, where is the emission wavelength and I() represents the fluorescence intensity at wavelength . Shifts in the spectral center of mass were converted into extent of denaturation (␣ p ) at each pressure according to the following phenomenological relationship (24): where N and U are the spectral centers of mass of native-like and fully unfolded protein obtained in the absence of denaturant and in the presence of a high concentration of GdnHCl, respectively, p is the spectral center of mass at pressure p, and Q is the ratio of fluorescence quantum yields of unfolded and native-like ␣ 3 -1.

Unfolding of ␣ 3 -1 by GdnHCl and Hydrostatic Pressure-
The equilibrium unfolding of ␣ 3 -1 by GdnHCl was initially investigated. Control measurements showed that unfolding was very rapid and was complete within a few minutes after addition of GdnHCl. Samples were incubated at increasing GdnHCl concentrations for 2 h at room temperature, and intrinsic fluorescence emission spectra were recorded. Unfolding was accompanied by a significant red shift of the fluorescence emission of ␣ 3 -1 ( Fig. 1), indicating increased exposure of the single tryptophan residue (Trp-32) to the aqueous medium. Fig. 1 (inset) shows the degree of denaturation (␣) of ␣ 3 -1 as a function of GdnHCl concentration. For comparison, data on the unfolding of ␣ 3 -1 monitored by far-UV CD measurements (18) have also been included. A single, cooperative transition from the native to the unfolded state was observed in both fluorescence and CD measurements. The almost exact superimposition of the unfolding profiles revealed by fluorescence and CD suggests that, in the presence of GdnHCl, the transition between native and unfolded ␣ 3 -1, at atmospheric pressure and room temperature, is essentially a two-state transition with no evidence for the existence of populated folding intermediates. Fig. 1 also shows that unfolding of ␣ 3 -1 takes place between 1 and 3 M GdnHCl, with a transition mid-point at 2.4 M GdnHCl. Fig. 2 shows the effect of pressure on the fluorescence spectral center of mass of ␣ 3 -1 in the absence or in the presence of GdnHCl. In the absence of GdnHCl (triangles) the center of mass did not reach a plateau even at the highest pressure used (3.5 kbar), indicating that complete unfolding was not achieved by the pressurization of ␣ 3 -1. From the equilibrium GdnHCl unfolding experiments ( Fig. 1) it is apparent that GdnHCl concentrations up to 1 M are subdenaturing for ␣ 3 -1. Pressure unfolding experiments were then repeated in the presence of different subdenaturing concentrations of GdnHCl (0.4 M and 1 M, Fig. 2) to poise the system toward unfolding. Although addition of 0.4 M GdnHCl had little effect on the pressure sensitivity of ␣ 3 -1, pressure denaturation in the presence of 1 M GdnHCl exhibited a clearly defined plateau of the spectral centers of mass at about 342 nm (Fig. 2). It is important to note that fully unfolded ␣ 3 -1 (i.e. in the presence of 6 M GdnHCl) exhibited a very red-shifted fluorescence emission, with a spectral center of mass of 355 nm. Thus, the plateau observed for the spectral center of mass of the pressure-denatured state at 342 nm seems to correspond to a stable partially unfolded intermediate. Upon stepwise release of pressure, the fluorescence spectra underwent a blue shift and reached complete recovery of the spectral center of mass at atmospheric pressure (data not shown), indicating reversible refolding of ␣ 3 -1 to a state qualitatively similar to the native-like protein. The fluorescence changes thus indicate that application of pressure in the presence of a subdenaturing concentration of GdnHCl (1 M) induced a transition to a stable conformation different from the fully denatured state induced by 6 M GdnHCl.
The pressure unfolding data for ␣ 3 -1 were analyzed using a two-state model for monomer unfolding. The dimensionless equilibrium denaturation constant at atmospheric pressure (K 0 ) and the molar volume change of folding (⌬V) can be calculated from the following thermodynamic relation, where K p is the denaturation constant at pressure p, and R and T have their usual meanings. The equation can be rewritten by introducing the degree of unfolding, ␣ p , at pressure p: where ln[␣ p /(1 Ϫ ␣ p )] equals lnK p for the denaturation of a monomer. Thus, a plot of ln[␣ p /(1 Ϫ ␣ p )] versus pressure (Fig. 2, lower panel) yields the molar volume change of folding (⌬V) from the slope and lnK 0 from the intercept on the ordinate. The parameters obtained for pressure unfolding of ␣ 3 -1 are shown in Table II. Cold Denaturation of ␣ 3 -1-To further characterize the existence of folding intermediates of ␣ 3 -1, we carried out low temperature unfolding experiments under pressure. The freezing point of water is significantly decreased under pressure (25), allowing aqueous samples to be analyzed at sub-zero temperatures without the need for addition of cryosolvent additives. Fig. 3 shows the fluorescence spectral centers of mass of ␣ 3 -1 as a function of decreasing temperature at 3.5 kbar in the absence and in the presence of 1.0 M GdnHCl. The starting points of the two curves (at 25°C) are similar to the spectral centers of mass obtained in the pressure denaturation experiments at the corresponding GdnHCl concentrations. In the absence of GdnHCl (circles) the spectra became progressively red-shifted but did not reach a plateau at low temperatures (down to Ϫ12°C), indicating that a stable partially (un)folded intermediate had not been reached. By contrast, in the presence of 1.0 M GdnHCl, a further red shift of the fluorescence emission occurred, with a low temperature plateau observed at about Ϫ10°C. Interestingly, the fluorescence spectra of the cold-denatured state and the completely unfolded protein (i.e., Folding of a Model Three-Helix Bundle Protein in 6 M GdnHCl) differ by about 8 nm in spectral center of mass (Fig. 4). After return of the sample to room temperature the fluorescence spectral center of mass returned to the original value, reflecting the reversibility of the process (Fig. 3, open  symbols). At constant pressure the temperature dependence of the equilibrium constant for a two-state unfolding transition is described by the van't Hoff equation, where K T is the equilibrium constant for denaturation at temperature T and ⌬G is the corresponding Gibbs free-energy change. From a plot of ⌬G/T versus the inverse temperature, the changes in enthalpy (⌬H) and entropy (⌬S) of unfolding can be extracted (Fig. 3, lower panel). The thermodynamic parameters obtained from such analysis are summarized in Table II.
Bis-ANS Binding Studies-The environment-sensitive fluorescent dye bis-ANS was used to characterize different partially folded conformations of ␣ 3 -1. Bis-ANS tends to bind exposed hydrophobic surfaces in partially folded intermediates more tightly than both the native and random coil states of proteins (26). Bis-ANS binding is accompanied by an increase in its fluorescence quantum yield, as well as by a blue shift of the fluorescence emission. On the basis of the increase in bis-ANS fluorescence, the pressure-denatured state of ␣ 3 -1 bound more bis-ANS than the native-like state, and the cold-denatured state exhibited significantly stronger binding (Fig. 5). In addition, bis-ANS binding to the cold-denatured state was also accompanied by a 14-nm blue shift of the fluorescence emission.

DISCUSSION
Structural transitions of a single-chain 65-amino acid threehelix bundle polypeptide, ␣ 3 -1, induced by hydrostatic pressure and by a combination of low temperature and high pressure revealed the existence of partially folded intermediate states, which are not observable in GdnHCl unfolding experiments of this model protein. Bis-ANS binding studies support the idea that organized hydrophobic surfaces persist, or can form, at both high pressures and low temperatures. Taken into account the "new view" of protein (un)folding, which models the chain collapse of a polypeptide by a multiple pathways "funnel," our results suggest that one possible unfolding transition of ␣ 3 -1 can be summarized by the following scheme, N^I 1^I2^U SCHEME 1 where N is the native-like and U is the unfolded state, and I 1 and I 2 represent the two partially (un)folded intermediates revealed in high pressure and low temperature experiments, respectively (Fig. 6).
Pressure-induced changes in the intrinsic fluorescence emission spectrum of ␣ 3 -1 took place between atmospheric pressure and 2.5 kbar in the presence of a subdenaturing concentration (1 M) of GdnHCl. The shift in spectral center of mass from 335 to 342 nm indicates partial exposure of the previously solventshielded tryptophan at the central heptad a position of helix II to the aqueous environment. The single-chain polypeptide nature of ␣ 3 -1 renders this native-like three-helix bundle of particular interest for pressure unfolding studies. It is generally assumed that pressures below 5 kbar do not significantly disturb the secondary or tertiary structures of proteins (27). Hydrogen bonds, the stabilizing elements of helices and ␤-sheets, are permanent dipoles and relatively insensitive to pressure

Folding of a Model Three-Helix Bundle Protein
changes. Moreover, the volume change attendant on replacement of protein-protein hydrogen bonds by protein-water hydrogen bonds is rather small. Therefore, pressure-induced unfolding of small monomeric proteins is generally only observed at high temperature (28), at low pH (29,30), or with mutant proteins (31,32). The finding that ␣ 3 -1 can be unfolded by pressure in the presence of a subdenaturing concentration of GdnHCl opens interesting possibilities for further studies of the stability of helical bundles and, in particular, of the structure of the pressure-stabilized partially folded state.
It is interesting to compare the volume change measured for the unfolding of ␣ 3 -1 with the volume changes reported for pressure denaturation of other proteins. The specific volume changes observed upon dissociation of dimeric proteins are dependent on the molecular weight. Arc repressor (M r 13,000), for example, shows a specific volume change of Ϫ7.7 l/g (33), whereas Enolase (M r 80,000) is reported to have a change in volume of Ϫ0.7 l/g (23). This can be explained by a larger proportion of buried amino acid residues, which becomes exposed to the solvent upon dissociation in smaller dimers, because in these cases the subunit interfaces involve a larger fraction of the entire structure. Alternatively, volume changes can also be interpreted in terms of the balance of forces responsible for protein stability. Disruption of electrostatic interactions leads to a large decrease in volume caused by electrostriction of water around the unpaired charged residues (34). By contrast, breaking of hydrophobic interactions is accompanied by much smaller volume changes. The denaturation of monomeric proteins is accompanied by similar effects, resulting in stronger hydration and the replacement of longer dispersion bonds by shorter dipolar interactions. Therefore, the relatively large specific volume change of 2.3 l/g observed for the folding transition of the intermediate I 1 to the native-like state N of ␣ 3 -1 (Table II) occurs most likely with the burial of polar side-chain groups.
Cold denaturation experiments at high pressures take advantage of the depression of the freezing point of water (25). Such an experimental setup and the presence of a subdenaturing concentration of GdnHCl allowed characterization of another folding intermediate, which showed strong bis-ANS binding. Destabilization of proteins at low temperatures indicates a Calculated assuming a two-state transition from native to the pressure-denatured state, using the relation ⌬G ϭ ϪRT lnK 0 , T ϭ 298 K. c Calculated assuming a two-state transition from pressure to cold-denatured states, as described in the text.  Fig. 1). Cold denaturation was obtained by applying pressure up to 3.5 kbar in the presence of 1 M GdnHCl and decreasing the temperature to Ϫ10°C (see Fig. 2). Spectra are normalized for maximal emission intensity. significant contribution of hydrophobic interactions to the folding process. Studies of the small dimeric protein Arc repressor showed that folding and association are accompanied by the displacement of solvent molecules, suggesting the burial of previously solvent-exposed nonpolar side chains (35). According to Privalov (36), hydration of polar residues decreases the entropy of the folding process. On the other hand, Weber (37) described the entropy-driven condensation of proteins as a consequence of the conversion of stronger solvent-protein interactions into weaker (entropy-rich) protein-protein interactions (London dispersion forces). In several cases, protein-pro-tein interactions involved in folding and subunit association have indeed been found to be predominantly entropy-driven. For example, the subunit association of hexokinase is characterized by a strong entropic contribution (T⌬S ϭ ϩ38 kcal/mol), which outweighs the unfavorable enthalpy of ϩ17 kcal/mol (38). The folding of ␣ 3 -1 reveals an entropy-driven transition from I 2 to I 1 , with T⌬S ϭ ϩ8.2 kcal/mol at 25°C and a van't Hoff enthalpy of ⌬H ϭ ϩ7.1 kcal/mol, resulting in Ϫ1.1 kcal/ mol of conformational stability (⌬G). The entropy-driven nature of the I 2 3 I 1 transition suggests that a hydrophobic collapse may be involved at this stage of folding of ␣ 3 -1.
The design of ␣ 3 -1 contains six distinct hydrophobic core layers, each consisting of three amino acids of either two a and one d or one a and two d heptad positions of the corresponding helices I, II, and III. Very likely, these areas are involved in formation of the organized hydrophobic domains revealed by bis-ANS binding at low temperatures. Of note is that the changes in intrinsic fluorescence emission of ␣ 3 -1 induced by pressure and low temperature were fully reversible, indicating that the protein refolds to a state that is qualitatively similar to the native-like state upon return to atmospheric conditions.
The folding of small (Ͻ100 amino acid residues), singledomain proteins is assumed to occur in a concerted fashion well accounted for by a two-state transition without well populated intermediates (39,40). On the other hand, there is strong evidence, especially from hydrogen exchange experiments, that partially folded conformations can be present (41,42). For a number of proteins, residual structure has been detected and linked to partially folded states, which are believed to be important as nucleation sites for condensation (43,44). For example, small patterns of stable residual structure were found in barnase during acid denaturation and were assumed to be formed during the early stage of folding (45). In addition, urea denaturation of the chaperonin GroEL also revealed persistent hydrophobic surfaces at high urea concentrations as probed by bis-ANS binding (46). Low temperature unfolding studies of ␤-lactamase provided evidence for the existence of two equilibrium intermediates between native and unfolded states (47).
Small synthetic helical proteins that undergo metal-directed transitions from molten globule-like to native-like states via folding intermediates have also been reported (48,49). Thermodynamic calculations of a model three-helix bundle using either a simple off-lattice or an all-atom approach revealed the existence of a metastable minimum (9 -11). In addition, Zhou and Karplus (12) have very recently used the same model protein to calculate different folding trajectories (12). The phase diagram could be varied by changing a single parameter related to the relative stability of native and non-native contacts. The simulation revealed that the folding mechanism for helical proteins changes from a cooperative (diffusion-collision) transition to one that involves on-pathway intermediates depending on the difference between the strength of native and non-native interactions (12). Our results on ␣ 3 -1 give direct support to the idea that even small helical bundle proteins may indeed present metastable folding intermediates, which can be detected under appropriate experimental conditions designed to stabilize them.
In conclusion, the present results, together with the small size of this three-helix bundle protein, make ␣ 3 -1 an ideally suited system for detailed protein folding studies. In this regard, an interesting possibility could be the use of high pressure NMR studies to characterize the structure of the folding intermediates.
Acknowledgments-A. C. thanks Fundaçã o de Amparo à Pesquisa do Estado do Rio de Janeiro for a previous fellowship. Pressure denaturation was carried out in the presence of 1 M GdnHCl at 25°C (see Fig. 1). Cold denaturation was carried out at 3.5 kbar in the presence of 1 M GdnHCl and decreasing the temperature to Ϫ10°C (see Fig. 2).