Folding Kinetics of the All-β-sheet Protein Human Basic Fibroblast Growth Factor, a Structural Homolog of Interleukin-1β*

The refolding and unfolding kinetics of the all-β-sheet protein human basic fibroblast growth factor (hFGF-2) were studied by fluorescence spectroscopy. The kinetics of the unfolding transition are monophasic. The refolding reaction at high and low guanidinium chloride (GdmCl) concentrations is best described by mono- and biphasic folding, respectively. Refolding and unfolding of hFGF-2 (155 amino acids) is very slow compared with other non-disulfide-bonded monomeric proteins of similar size. For example, the rate constant for unfolding at 4.5 mol·liter−1GdmCl is 0.006 s−1, and the refolding rate constants at 0.4 mol·liter−1 GdmCl are 0.01 s−1 and 0.0009 s−1 (15 °C, pH 7.0). A characterization of the thermodynamic nature of the folding process using transition state theory revealed that the slow refolding is almost exclusively controlled by entropic factors, namely the strong loss of conformational freedom during refolding. The rate of the slow unfolding kinetics is mainly (and at low denaturant concentrations exclusively) controlled by the large positive change in enthalpy. hFGF-2 shows similar slow folding kinetics to that of its structural homolog interleukin-1β. Since both proteins show very little sequence identity, it is suggested that their slow folding kinetics are determined by the complex β-sheet arrangement of the native molecules.

Human basic fibroblast growth factor (hFGF-2) 1 is a globular single chain heparin-binding polypeptide synthesized by different cell types and is involved in processes associated with proliferation and differentiation of cells (1). The wound healing activity of hFGF-2 renders it to a potential therapeutic agent of industrial importance (2).
The determination of the three-dimensional structure of hFGF-2 by x-ray crystallography revealed a complex all ␤-sheet protein without any disulfide bond (3)(4)(5). The overall structure shows strong homology with the fold of interleukin-1␤ and can be described as a trigonal pyramid with three topological units consisting each of four anti-parallel ␤-sheets. Both proteins share only 10 -13% sequence identity, but superposition of their carbon backbone involves 9 of the 12 ␤-sheet strands including the anti-parallel ␤-sheet incorporating the N-and C-terminal regions of the molecules (50 C ␣ of the two proteins could be superimposed with a root mean square discrepancy of 0.52 Å; see Ref. 3).
Electron density was not observed for the first 28 and the last 3 amino acids of hFGF-2 (155 amino acid form) suggesting that these parts of the molecule are disordered and/or are very flexible. hFGF-2 contains nine proline residues (6), a relatively high content compared with other molecules of similar size. Five of the Xaa-Pro peptide bonds are localized in the flexible N-terminal part of the protein, and they probably adopt conformations that are usually adopted by Xaa-Pro peptide bonds in unfolded proteins or short peptides (Ϸ70 -90% trans; see Ref. 7). The remaining four other Xaa-Pro peptide bonds are in trans conformations in native hFGF-2 (3)(4)(5).
Folding studies with the hFGF-2 structural homolog interleukin-1␤ revealed the very slow folding kinetics of this protein (8 -11). Slow folding kinetics result very often from peptidylprolyl trans/cis isomerization processes (7). Interleukin-1␤ (153 amino acids) contains eight prolines (3)(4)(5)12). One Xaa-Pro peptide bond is localized in the flexible N-terminal part of the protein, and six of the other Xaa-Pro peptide bonds are in trans conformations, and the Tyr 90 -Pro 91 peptide bond adopts the cis conformation (12). The molecular origin of the slow folding kinetics of interleukin-1␤ has not yet been identified and, despite the presence of a cis proline in the native protein, could not be attributed to prolyl isomerization processes (8 -10).
It was shown that the formation of unstable ␤-sheets in interleukin-1␤ occurred rapidly on the millisecond time scale (9). However, the stabilization of the three-dimensional structure of the protein with the establishment of native hydrogen bonds began only after 1 s with the protein passing through several folding intermediates containing an increasing content of stable hydrogen bonds within the ␤-sheets of the three topological units (9,10). Experimental results from folding studies using pulse-labeling hydrogen exchange and electrospray ionization mass spectrometry suggested that the folding of interleukin-1␤ proceeds through an obligatory, defined intermediate on the folding pathway (10,11). The final stabilization of the native structure of interleukin-1␤ was observed on a time scale of minutes and could be monitored by tryptophan fluorescence emission spectroscopy (8,9).
hFGF-2 and interleukin-1␤ both have a single tryptophan localized at a conserved position in an anti-parallel ␤-sheet segment whose ␣-carbon atoms are superimposable with a root mean square discrepancy of 0.52 Å in both proteins (3). The tryptophan is partly exposed to the solvent and localized at the end of a very tight loop connecting two anti-parallel ␤-strands (3)(4)(5)13). In contrast to interleukin-1␤, the fluorescence emission of the single tryptophan present in hFGF-2 is completely quenched in the native molecule (13). Unfolding of hFGF-2 leads to a tryptophan-dominated fluorescence spectrum with an increase in the intensity of the emission and an accompanying shift of the maximum wavelength from 306 to 355 nm. The unusual total quenching of the tryptophan fluorescence in hFGF-2 is also abolished in response to marginal perturbations within the native structure of hFGF-2. Therefore, the disappearance of the tryptophan emission during refolding of hFGF-2 can be used as a sensitive probe for the attainment of the correct interactions within the complex ␤-sheet arrangement of the native molecule.
In this study, we describe the refolding and unfolding kinetics of hFGF-2 at different temperatures and varying concentrations of guanidinium chloride (GdmCl) using fluorescence emission spectroscopy. In addition, we have characterized the thermodynamics of the unfolding and refolding process using transition state theory.

EXPERIMENTAL PROCEDURES
Unfolding and Refolding of hFGF-2-The preparation and storage of the hFGF-2 stock solution (10 mol⅐liter Ϫ1 hFGF-2 in 0.1 mol⅐liter Ϫ1 sodium phosphate buffer, pH 7.0) and fluorescence emission spectroscopy were performed as described previously (13). Unfolding of hFGF-2 was initiated by diluting hFGF-2 stock solutions into 0.1 mol⅐liter Ϫ1 sodium phosphate buffer, pH 7.0, in the presence of varying concentrations of GdmCl. Refolding of unfolded hFGF-2 (in 2 mol⅐liter Ϫ1 GdmCl) was initiated by dilution into 0.1 mol⅐liter Ϫ1 sodium phosphate buffer, pH 7.0, in the presence of varying concentrations of GdmCl. The reversibility of the unfolding and refolding processes were guaranteed by the presence of 0.1 mol⅐liter Ϫ1 ␤-mercaptoethanol in all solutions. Unfold-ing or refolding of hFGF-2 was monitored by measuring the time-dependent increase or decrease, respectively, of the protein fluorescence at 355 nm (excitation at 280 nm).
Kinetic and Thermodynamic Analysis-The kinetic analysis was performed as described previously (14). The number of kinetic phases and their corresponding time constants and amplitudes were determined from the time-dependent change of the fluorescence during protein (un)folding using the software package MATLAB.
The kinetic data of monomolecular (un)folding determined from fluorescence measurements can be described by Equation 1, where F(t) is the fluorescence at time t, F(ϱ) the fluorescence at the end of the reaction, k i the apparent rate constant of phase i, and F i the corresponding amplitude. The amplitude and the rate constant of the slowest phase were first determined from the linear region of the first-order plot of the reaction. In the case of biphasic folding, this component was then subtracted from the original time course and replotted against the time for determination of the amplitude and the rate constant of the major phase of folding ("pealing off method"). The free energies of activation for (un)folding were calculated from the temperature dependences of the kinetic constants according to transition state theory (15) (Equation 2), where k i is the experimentally determined rate constant of the kinetic phase i, the transmission factor, T the temperature, and k B , h, and R the Boltzmann, Planck, and universal gas constants, respectively. The apparent free energies of activation ⌬G ‡ of protein (un)folding were calculated assuming a transmission factor of 1, but may be smaller for protein folding reactions in solution. According to the equilibrium free energy ⌬G, the free energy of activation ⌬G ‡ can be decomposed into its enthalpic ⌬H ‡ and entropic components ⌬S ‡ (Equation 3).

FIG. 1. Kinetic analysis of unfolding and refolding of hFGF-2.
A, unfolding of hFGF-2 was monitored versus time in 3.0 mol⅐liter Ϫ1 GdmCl. B and C, refolding was initiated by dilution from 2.0 mol⅐liter Ϫ1 GdmCl into 1.0 and 0.5 mol⅐liter Ϫ1 GdmCl, respectively. The two curves in C represent the first-order plots of the folding reaction as determined by the pealing off method described under "Experimental Procedures" and correspond to the major (curve 1) and the very slow phase of refolding (curve 2). Experiments were performed at 15°C, and the protein concentration was 0.5 mol⅐liter Ϫ1 in 0.1 mol⅐liter Ϫ1 sodium phosphate buffer, pH 7.0, 0.1 mol⅐liter Ϫ1 BME.

FIG. 2. Determination of the GdmCl-dependent rate constants of refolding (triangles and diamonds correspond to the bi-and monophasic refolding region, respectively) and unfolding of hFGF-2 (circles).
Refolding and unfolding experiments were performed at 15°C at a protein concentration of 0.5 mol⅐liter Ϫ1 in 0.1 mol⅐liter Ϫ1 sodium phosphate buffer, pH 7.0, 0.1 mol⅐liter Ϫ1 BME, and varying concentrations of denaturant.
By combining Equations 2 and 3, the enthalpic ⌬H ‡ and the entropic components of the activated state ⌬S ‡ can be determined from the Eyring plot (ln(k⅐T -1 ) versus T -1 ) according to Equation 4.
If there is linearity for the temperature dependence of the folding rate constants, the enthalpy and entropy of activation can be determined from the slope and the y intercept of the Eyring plot, respectively. With the determination of the enthalpic ⌬H ‡ and entropic components of the activated state ⌬S ‡ , the activation free energy ⌬G ‡ is then calculated according to Equation 3.

RESULTS AND DISCUSSION
hFGF-2 is a very unstable protein and exhibits a strong tendency toward aggregation during refolding (13). However, it can be reversibly unfolded by urea or GdmCl at low protein concentrations (0.5 mol⅐liter Ϫ1 Х 9 g⅐ml Ϫ1 ) and in the presence of reducing agents.
Kinetics of Unfolding and Refolding-The results of typical unfolding and refolding kinetics of hFGF-2 as determined by fluorescence spectroscopy and using a manual mixing technique are shown in Fig. 1. The unfolding reaction of hFGF-2 is described by a simple single exponential decay at all GdmCl concentrations investigated (e.g. Fig. 1A). The refolding kinetics at intermediate GdmCl concentrations are best described by monophasic folding (Fig. 1B). At low GdmCl concentrations the refolding can be described by a biphasic reaction, and the rate constants for both phases can be determined with sufficient accuracy (Fig. 1C).
The GdmCl-dependent rate constants of refolding and un-folding of hFGF-2 display the typical V-shaped dependence on the denaturant concentration with a minimum in the transition region (Fig. 2) as has been reported for numerous other proteins (e.g. Refs. 7 and 16). At low denaturant concentrations, refolding displays a biphasic behavior. At higher GdmCl concentrations, both phases approach each other and refolding can be described by an apparent monophasic behavior. This monophasic folding and the faster phase in the case of biphasic folding are in the following referred to as the major phase of refolding. Refolding of hFGF-2 is an extremely slow process, e.g. the rate constants of the major and the very slow phase of refolding at 15°C and 0.4 mol⅐liter Ϫ1 GdmCl correspond to k f1 ϭ 0.01 s Ϫ1 and k f2 ϭ 0.0009 s Ϫ1 , respectively (Fig. 2). Unfolding of hFGF-2 is monophasic at all denaturant concentrations. Also, the slope of the GdmCl dependence of the lnk u is linear, indicating monophasic unfolding. The unfolding of the growth factor is also a very slow process, e.g. the rate constant of unfolding at 15°C and 4.5 mol⅐liter Ϫ1 GdmCl corresponds to k u ϭ 0.006 s Ϫ1 (Fig. 2).
The amplitude associated with the major phase of refolding represents 75-85% of the total amplitude observed at low GdmCl concentrations and decreases with increasing denaturant concentrations (data not shown). The amplitude or in the case of biphasic folding the sum of the two amplitudes associated with the refolding and unfolding transitions correspond to the entire amplitude expected from equilibrium unfolding experiments. This shows that the very fast and early phases of refolding that cannot be resolved by using manual mixing techniques are not associated with any or only with an insignificant change in tryptophan emission.
Temperature Dependence of Unfolding Kinetics and Kinetics of the Major Refolding Phase-The GdmCl-dependent refolding and unfolding kinetics of hFGF-2 were studied at temperatures ranging from 15 to 35°C (Fig. 3). The unfolding rate constant shows a strong dependence on the denaturant concentration and is also affected considerably by the temperature. Unfolding of hFGF-2 is slow at all the temperatures studied. The slopes of the denaturant dependence of lnk u at a given temperature are linear at all temperatures investigated indicating monophasic unfolding of hFGF-2. Linear denaturant dependences of lnk are in general taken as an indication of a two-state folding transition (17); however, it should be noted that they can also be associated with multiple state folding and cannot be taken as an unambiguous proof of a two-state folding process (18).
The refolding of hFGF-2 is also a slow process within the entire temperature range investigated. But in contrast to unfolding, there is not a pronounced effect of the temperature on the rate constants of refolding. Particularly, the rate constant of the major phase of refolding does not vary at a given temperature but is strongly affected by the denaturant concentration (Fig. 3). The small temperature effect on the rate constant of the very slow phase of refolding is discussed below. Thus, the kinetics of the major refolding phase are mainly determined by the residual denaturant concentration and not by the temperature. The denaturant dependence of lnk f of the major phase of refolding is linear suggesting that this phase is monophasic and not a composite of several folding transitions. However, other techniques such as circular dichroism spectroscopy or calorimetry which are not applicable, due to the low solubility of hFGF-2 folding intermediates, might reveal more intermediates on the folding pathway.
Transition State Analysis of Unfolding-The dependence of the unfolding rate constant on temperature and denaturant concentration was utilized to characterize the thermodynamic nature of the transition state during unfolding (15) (see "Experimental Procedures").
An Eyring plot (ln(k u ⅐T -1 ) versus T -1 ) of the temperature dependence of the unfolding rate constant revealed a linear dependence at all denaturant concentrations investigated (Fig.   4) indicating that there is no significant change in the heat capacity between the native and the transition state (19). Therefore, the free energy of activation and the corresponding enthalpic and entropic components of unfolding of hFGF-2 as function of the denaturant concentration were calculated assuming that ⌬H u ‡ and ⌬S u ‡ are independent of the temperature within the temperature range investigated.
These calculations revealed that the enthalpy, the entropy, and the free energy of activation are linearly dependent on the denaturant concentration (Fig. 5). The activation free energy of unfolding of hFGF-2 is positive at all denaturant concentrations and decreases with increasing concentrations of GdmCl. Decreasing free energies of activation of unfolding with increasing denaturant concentration have been observed before (19,20). They indicate that the transition state interacts more strongly with the denaturant compared with the native state of the protein. The enthalpy of activation is also positive at all denaturant concentrations and decreases with increasing concentrations of GdmCl. A positive activation enthalpy of unfolding reveals that internal noncovalent interactions that are stabilizing the native protein are disrupted during the transition from the folded to the activated state. The decrease of the activation enthalpy with increasing concentrations of GdmCl indicates stronger solvation of the transition state compared with the native state of the protein. The activation entropy also decreases with increasing denaturant concentrations. The entropic term is positive at low denaturant concentrations but becomes negative above 2.4 mol⅐liter Ϫ1 GdmCl. Negative entropic values for the activation entropy for unfolding appear unexpected, but they have been reported previously (20,21). They reflect the overcompensation of the loss of the conformational order by the concomitant ordering of solvent and cosolvent components caused by the disruption of hydrophobic Experimental data from refolding experiments were determined at 0.5 mol⅐liter Ϫ1 (squares), 0.7 mol⅐liter Ϫ1 (circles), and 0.9 mol⅐liter Ϫ1 GdmCl (triangles) at a protein concentration of 0.5 mol⅐liter Ϫ1 in 0.1 mol⅐liter Ϫ1 sodium phosphate buffer, pH 7.0, 0.1 mol⅐liter Ϫ1 BME, and varying temperatures. The experimentally determined rate constants near the transition region were corrected by subtracting the corresponding rate constants of unfolding to determine the true rate constant of the major refolding phase.
interactions during the unfolding of the protein molecule. The high enthalpic barrier for unfolding at low denaturant concentrations is reduced by the gain in (conformational) entropy during the transition from the folded to the activated state. At high denaturant concentrations both the positive change in enthalpy and the negative change in entropy caused by the ordering of solvent molecules during the transition from the folded to the activated state contribute to the energy barrier for unfolding. Altogether, the slow unfolding kinetics of hFGF-2 are mainly (and at low denaturant concentrations exclusively) determined by the large positive change in enthalpy.
Transition State Analysis of the Major Phase of Refolding-The Eyring plot of the rate constant of the major phase of refolding again emphasizes its temperature independence at a given denaturant concentration within the temperature range investigated (Fig. 6). Chemical reaction rates generally increase with increasing temperature, but the opposite has been observed for protein folding reactions (22,23). For example, it has been shown that the refolding rate of chymotrypsin inhibitor 2 and barnase initially increases and then goes through a broad maximum and finally decreases with increasing temperature. This affords into an Eyring plot of refolding with strong curvature resulting in positive activation enthalpies of refolding at low and negative values at high temperatures (22,23). The studies of the refolding of hFGF-2 have been in the temperature range where the rate constant of the major phase of refolding has reached its apparent highest value, thereby showing only a broad maximum in the Eyring plot (Fig. 6) where the enthalpy of activation for the major phase of refolding is approximately zero (Fig. 7).
The enthalpy, the entropy, and the free energy of activation of the major phase of refolding as a function of denaturant concentration were calculated from the data shown in Figs. 3 and 6 with the simplification that the enthalpy of activation of the major refolding phase is set to zero and the assumption that the enthalpic and entropic terms do not show a significant change at a given denaturant concentration within the temperature range investigated (Fig. 7). The calculations revealed that the free energy of activation of the major refolding phase is positive at all denaturant concentrations and increases linearly with increasing concentrations of GdmCl indicating that the unfolded state interacts more strongly with the denaturant than does the transition state. The entropy of activation for the major refolding phase has negative values at all denaturant concentrations which decrease with increasing concentrations of GdmCl. Thus, the kinetics of the major refolding phase are exclusively controlled by entropic factors. The slow refolding kinetics must result from the loss of conformational freedom during refolding causing a strong decrease in the entropy during the transition from the unfolded to the activated state. The gain in entropy during refolding through the release of solvent and co-solvent components bound to the unfolded protein is largely overcompensated by the massive loss in conformational entropy.
Transition State Analysis of the Very Slow Phase of Refolding-In contrast to the rate constant of the major refolding phase, the rate constant of the very slow phase of refolding (only observable at low denaturant concentration, e.g. below 0.8 to 0.9 mol⅐liter Ϫ1 GdmCl) does not show a pronounced dependence on the denaturant concentration (see also Fig. 2). The Eyring plot revealed a slight temperature effect reflecting the modest increase of the refolding rate of the very slow phase with increasing temperature (Fig. 8). The data shown in Fig. 8 were utilized to calculate the enthalpy, the entropy, and the free energy of activation for the very slow phase of refolding. These calculations revealed values for ⌬H f2 ‡ , ⌬S f2 ‡ , and ⌬G f2 ‡ of 33 kJ⅐mol Ϫ1 , Ϫ190 J⅐mol Ϫ1 ⅐K Ϫ1 , and 89 kJ⅐mol Ϫ1 , respectively. Therefore, neither the major nor the very slow phase of refolding of hFGF-2 has properties consistent with proline FIG. 7. Thermodynamic characterization of the transition state of the major refolding phase of hFGF-2. Determination of the free energy ⌬G ‡ (squares, the enthalpic ⌬H ‡ (circles), and the entropic components ⌬S ‡ (triangles) of the activated state of the major phase of refolding of hFGF-2 at 15°C as function of the denaturant concentration. Values were determined as described under "Results" and "Experimental Procedures" from data shown in Figs. 3 and 6. The maximum error associated with ⌬G ‡ , ⌬H ‡ , and ⌬S ‡ is 10%.
FIG. 8. Eyring plot of the rate constant of the very slow phase of refolding of hFGF-2. Experimental data from refolding experiments were determined at 0.4 mol⅐liter Ϫ1 (squares), 0.5 mol⅐liter Ϫ1 (circles), and 0.6 mol⅐liter Ϫ1 GdmCl (triangles) at a protein concentration of 0.5 mol⅐liter Ϫ1 in 0.1 mol⅐liter Ϫ1 sodium phosphate, pH 7.0, 0.1 mol⅐liter Ϫ1 BME, and varying temperatures. isomerization, particularly, the activation enthalpies of refolding are low (Ϸ0 and 33 kJ⅐mol Ϫ1 , respectively). The activation enthalpies for proline isomerization are in general high and of the order of 80 kJ⅐mol Ϫ1 (7,24,25).
Similar Slow Folding Kinetics of Structural Homologs hFGF-2 and Interleukin-1␤-Folding of hFGF-2 is very slow compared with other non-disulfide-bonded monomeric proteins of similar size. Folding studies with the hFGF-2 structural homolog interleukin-1␤ revealed that it also refolds in a slow and a very slow phase to its native structure at low denaturant concentrations (10). Since both proteins show very little sequence identity and there are no indications for rate-limiting prolyl isomerization processes, it is suggested that their slow folding kinetics are determined by their complex ␤-sheet arrangement. In both proteins, the slow folding kinetics can be monitored by fluorescence emission of the single conserved tryptophan suggesting that they may share a common folding pathway and that the final stabilization of the native structures occurs in the vicinity of this tryptophan.