Free Energy Changes in Ribonuclease A Denaturation EFFECT OF UREA, GUANIDINE HYDROCHLORIDE, AND LITHIUM SALTS*

The unfolding of ribonuclease A by urea, guanidine hydrochloride, lithium perchlorate, lithium chloride, and lithium bromide has been followed by circular dichroic and difference spectral measurements. All three abnormal tyrosyl residues are normalized in urea and guanidine hydrochloride (At287 = -2700), only two are normalized in lithium bromide and lithium per-chlorate (At287 = -1700), and only one is exposed in lithium chloride solutions (&287 = -700). The Gibbs energies are 4.7 f 0.1 kcal mol” for urea- and guanidine hydrochloride-denaturation, 3.8 f 0.2 kcal mol-1 for lithium perchlorate-denaturation, and 12.7 2 0.2 kcal mol” for lithium chloride- and lithium bromide- denaturation of ribonuclease A. The latter results suggest that the mechanism of the unfolding process in urea and guanidine hydrochloride is quite different from that in lithium salts. hy-drodynamic, rotational,

Bigelow et al. (1)(2)(3)(4)(5)(6) have been interested for some time in the effect of various denaturants on the globular conformation of ribonuclease A, lysozyme, and a-lactalbumin. An examination of quantitative denaturation data obtained from hydrodynamic, difference spectral, optical rotational, and circular dichroic measurements show that urea and guanidine hydrochloride give the most unfolded state whereas, inorganic salts, LiC104, LiC1, and LiBr cause partial unfolding leading to unique intermediate states with some secondary structure. It should, however, not be inferred that the protein molecules which are converted from the folded to randomly coiled state (as obtained in concentrated solutions of urea and GdnHCl I ) necessarily go through either of the intermediate states.
One of the interesting results from a study of the denaturation of folded proteins is an estimation of the Gibbs energy (conformational free energy change) involved in converting the globular conformation to an unfolded state in water. This is done by measuring the free energy change AG,,, as a function of denaturant concentration in the transition region and extrapolating to zero concentration, AGz: (7). However, we do not know whether or not the values of AGZ: for all denatured states are internally consistent, in the sense that A G : ; ; is larger when more extensive unfolding takes place and is smaller when less extensive unfolding occurs. For this reason and others, we have been carrying out systematic studies of denaturation of proteins by urea and by various salt * This work was supported by a grant from the Natural Sciences and Engineering Council of Canada to Prof. Charles C. Bigelow. 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.
The abbreviation used is: GdnHC1, guanidine hydrochloride. denaturants, namely, GdnHCl, LiC104, LiBr, and LiC1. It is felt that it would be worthwhile to present in detail here some rather unexpected results of A G g measurements for the denaturation of one protein, namely, RNase A by salt denaturants giving intermediate states. It has been observed that although LiCl and LiBr cause partial unfolding, the value of AG&p is about three times larger than that obtained for a randomly coiled RNase A. A possible explanation for this discrepancy is proposed.

MATERIALS AND METHODS
RNase A (bovine pancreas) was obtained from Calbiochem. Ultrapure urea and GdnHCl were purchased from Schwarz/Mann. LiCl was from Baker, while LiClOl and LiBr were from Matheson, Coleman & Bell. These and other analytical grade chemicals were used without further purification.
Absorption spectra were measured in a Cary 118 or a Cary 219 spectrophotometer using tandem thermostated cells whose temperature was maintained within kO.01 "C. The protein concentration was determined using a value of 9800 for the molar extinction coefficient of native RNase A at 277.5 nm (1).
CD measurements were made in a Jasco 5-20 automatic recording spectropolarimeter. Thermostated cells of 0.1-and 1.0-cm path lengths were used. Base-line corrections for solvents were made routinely. CD data were reduced to the concentration-independent parameter [e], the mean residue ellipticity, defined as: where 8, is the observed ellipticity in millidegrees at wavelength X, M, the mean residue weight of RNase A (M, = 110), c the protein concentration (milligrams/cm3), and 1 the path length (centimeters).
All CD results are presented in units of degrees cm2 dmol".
RNase A solutions were prepared as follows. For unfolding experiments, known amounts of stock protein solution, buffer, and denaturant solutions, all in 0.1 M glycine-HC1 buffer (pH 3.0) were mixed and incubated overnight which was sufficient for completion of the reaction. A similar procedure was employed in preparing the protein solution for refolding experiments with the only exception that RNase A was first denatured in concentrated denaturant solution and then diluted with buffer. Solutions for all measurements were routinely filtered.
Measurements for pH were made with a Radiometer type TTTlC pH meter.

RESULTS
The unfolding transitions in urea and in salt denaturants, followed by observing changes in are shown in Fig. 1. A decrease in the magnitude of -[0]2z0 is taken to reflect the loss of secondary structure. All transitions were found reversible; the experimental points obtained from denaturation and renaturation experiments lie on the same transition curve.
In Fig. 2 the difference molar absorption changes at 287 nm, AtZe7, are plotted as a function of denaturant concentration. At this wavelength, it is expected that changes would reflect alteration in the environment of tyrosine residues on [DENATURANT], M

FIG. 2. Changes in AcZs7 of RNase A on denaturation by urea ( I ) , LiCl (2), LiC104 (3), and LiBr ( 4 ) .
The values of Ae287 for urea-, LiBr, LiC1-, and LiC101-denatured RNase A are -2700, -1700, -700, and -1700 cm" mol", respectively. Symbols and experimental conditions are the same as in Fig. 1. Lines in the pre-and posttransition regions are computed least squares best fit to the data. All data for the pretransition lines are not shown to maintain brevity. exposure to solvent. The intercepts at zero denaturant concentration, obtained from the linear extrapolation of results in the post-transition region are asumed to be the value of At287 which would be observed for the unfolded state, if it could somehow be studied in the absence of the denaturant (1). These effects of denaturation are also reversible.
The apparent equilibrium constant* and free energy change for a two-state process an be obtained from the results shown in Figs. 1 and 2 using the following relation: * The subscript app conveys the fact that the properties on which this thermodynamic analysis are based, usually by and large reflect the conformational change on denaturation. The other changes accompanying unfolding may be unjustifiably neglected or overlooked. Thus, K and AG values are apparent, that is, nonthermodynamic.
The subscripts N , U, and obs denote, respectively, the value of ([0],20 or Atzs7) for the native state, the value for unfolded state, and the value observed in the presence of denaturant at a temperature T"K. Unlike the CD results (Fig. l), since the difference spectral properties of the native and unfolded protein molecules showed dependence on the denaturant concentration (Fig. 2), allowance has been made for the dependence of both YN and Yr, on the denaturant concentration in calculating K,,.
The values of AG. , , were plotted against each denaturant concentration (Fig. 3). A linear least squares analysis was applied to the data according to empirical Equation 3 (8), which was later shown to be a more general thermodynamic model for analysing the solvent denaturation of proteins (9):   Table I. It has been found that the plots of AG,,, versus C are all good straight lines.

DISCUSSION
Everything we know about the effect of urea and GdnHCl on RNase A leads us to believe that these denaturants give rise to a molecule that is completely unfolded within the constraints imposed by four disulfide bonds. The denatured protein with its disulfide bonds reduced behaves as a linear random coil (11-13). On the other hand, lithium salts lead to an incomplete unfolding of RNase A (see Figs. 1 and 2), i.e. in our terminology a state "intermediate" between the native and randomly coiled state (1). Indeed such intermediate states have been shown to contain some residual structures that can be removed in a co-operative manner by the addition of urea to the partially denatured states (3, 4).
The observation that two independent physical methods used to follow a denaturation transition gave identical values of AG,"; and C,, within the experimental error, suggests that each transition shown in Figs. 1 and 2 is a two-state process. It should, however, be noted that this is a necessary but not a sufficient condition for an all-or-none process. But the high cooperativity of unfolding measured by equilibrium experiments inside the transition region suggests that intermediates are unstable relative to native and denatured protein (16) and therefore do not contribute significantly to K a w An attempt was made to study the kinetics of denaturation and renaturation of RNase A by lithium salts. It was concluded that meaningful kinetic data for the refolding of RNase A cannot be obtained due to a large heat of dilution of these salt solutions. For example, the rise in temperature of a 10 M LiBr solution at 25 "C was more than 10 "C on 50% dilution with water at 25 "C. Nevertheless, we assumed a two-state behavior of lithium salts denaturation in the light of the coincidence of transition curves measured by two independent physical properties (see Table I).
RNase A in concentrated solutions of urea and GdnHCl is devoid of all the elements of its native conformation and behaves as a cross-linked random coil (11-13). It is therefore possible to estimate the Gibbs energy change for the process, folded conformation $ random conformation in the absence of the denaturant. A value of 4.7 f 0.1 kcal mol" for AGE: was obtained (see Table I). This value for the two denaturants is the same, which indicates that the extent of unfolding in the two denaturants cannot differ appreciably.
In most cases the products of acid and thermal denaturation of several proteins are less completely unfolded than those of urea and GdnHCl denaturation (11). In his review Pace (8) compares the values of AGEB for GdnHCl and acid-thermal denaturation of several proteins. These results suggest that for a protein AGE* for unfolding to a random coil is greater than that for unfolding to a denatured state produced by acidthermal denaturation. All available data on the effect of lithium salts suggest that these denaturants unfold RNase A to a lesser extent than GdnHCl and urea (1-4). It is therefore expected that AG$? associated with the unfolding by a lithium salt is likely to be less than 4.7 +. 0.1 kcal mol". As can be seen in Table I, the analysis of equilibrium results of denaturation by LiCl and LiBr gave unexpected results; a value of 12.7 -e 0.2 kcal mol" for AGE? was estimated from these denaturation results in the absence of lithium salts.
The discrepancy between the observed and expected values of AGE? prompted us to study the denaturation by LiCl of two other proteins, namely lysozyme and a-lactalbumin; LiCI is known to give rise to a conformation intermediate between the native and denatured states of these proteins (5, 6). The results of the LiCl denaturation, along with those for denaturation by urea and GdnHCl (10) are listed in Table 11. As can be seen in Table 11, AGZ:," values from LiCl denaturation are larger than those associated with the process, native e random coil conformation. The discrepancy between the expected and observed values of A G : ; : is confusing unless it means that we have to reconsider the two-state behavior of protein denaturation by lithium salts, as assumed in the present analysis. The mechanism of denaturation is very important in analysing transition curves such as shown in Figs. 1 and 2. As we shall see below, one can explain the denaturation results by assuming a mechanism in which native protein ( N ) goes to a "definite" state A, obtained in a lithium salt, through another denatured state ( X ) which is unfolded more than A and is not on the direct path between N and A: On the other hand, as shown by Tanford (11) and Pace (8), if X is on the direct path between N and A , i.e. if X is less unfolded than A , then the estimate of AGZ:," derived from a two-state analysis will be lower than the true value of this parameter.
A mathematical treatment for the equilibrium constant for a process such as shown in Equation 4 is as follows. If all the states are characterized by a property Y and fractional concentration f , then Yobs at any point on the salt-induced denaturation curve between N and A is given by ycbs = fNyN + f X y X + f A y A (5) and The observed fraction of the denatured protein, fobs, for the process between N and A ( N 6 A ) is given by  It is obvious from Equation 7 that for 01 > 1, the square bracketed term of this equation is always greater than 1. This means that Kapp > KA, the true thermodynamic equilibrium constant for the process N + A . That is, AGE? > AGPO . We would therefore like to suggest that a cause for the apparent discrepancy among the values of AGz: for various denaturation processes seems to be due to a problem in analysing the transition curve obtained in lithium salt solutions.
It has recently been reported that the mechanism of denaturation of RNase A by LiC10, (17) is more complex than that proposed here (see Equation 4). In the folding of the denatured protein from state A to the state N , Denton et al.
(17) observed three intermediates. Thus, there is kinetic evidence that the lithium denaturation of RNase A is not a twostate process.