- Email: [email protected]
Ph-dependent interactions and the stability and folding kinetics of the N-terminal domain of L9. electrostatic interactions are only weakly formed in the transition state for folding1
doi:10.1006/jmbi.2000.3752 available online at http://www.idealibrary.com on
J. Mol. Biol. (2000) 299, 1091±1100
pH-Dependent Interactions and the Stability and Folding Kinetics of the N-terminal Domain of L9. Electrostatic Interactions are Only Weakly Formed in the Transition State for Folding Donna L. Luisi1 and Daniel P. Raleigh1,2* 1
Department of Chemistry State University of New York at Stony Brook, Stony Brook NY, 11794-3400, USA 2
Graduate Program in Biophysics and Graduate Program in Molecular and Cellular Biology
The role of electrostatic interactions in the stability and the folding of the N-terminal domain of the ribosomal protein L9 (NTL9) was investigated by determining the effects of varying the pH conditions. Urea denaturations and thermal unfolding experiments were used to measure the free energy of folding, G , at 18 different pH values, ranging from pH 1.1 to pH 10.5. Folding rates were measured at 19 pH values between pH 2.1 and pH 9.5, and unfolding rates were determined at 15 pH values in this range using stopped-¯ow ¯uorescence experiments. The protein is maximally stable between pH 5.5 and 7.5 with a value of G 4.45 kcal molÿ1. The folding rate reaches a maximum at pH 5.5, however the change in folding rates with pH is relatively modest. Over the pH range of 2.1 to 5.5 there is a small increase in folding rates, ln (kf) changes from 5.1 to 6.8. However, the change in stability is more dramatic, with a difference of 2.6 kcal molÿ1 between pH 2.0 and pH 5.4. The change in stability is largely due to the smaller barrier for unfolding at low pH values. The natural log of the unfolding rates varies by approximately four units between pH 2.1 and pH 5.5. The stability of the protein decreases above pH 7.5 and again the change is largely due to changes in the unfolding rate. ln (kf) varies by less than one unit between pH 5.5 and pH 9.5 while G decreases by 2.4 kcal molÿ1 over the range of pH 5.4 to pH 10.0, which corresponds to a change in ln Keq of 4.0. These studies show that pH-dependent interactions contribute signi®cantly to the overall stability of the protein but have only a small effect upon the folding kinetics, indicating that electrostatic interactions are weakly formed in the transition state for folding. # 2000 Academic Press
*Corresponding author
Keywords: protein stability; electrostatics; protein folding; L9
Introduction Interactions involving charged residues are thought to play an important role in protein folding and in stabilizing the native structure of proteins (Barlow & Thornton, 1983; Hendsch & Tidor, Abbreviations used: CD, circular dichroism; GdHCl, guanidine hydrochloride; Fmoc, N-9¯uorenylmethyloxycarbonyl; HBTU, O-benzotriazoleN,N,N0 ,N0 -tetramethyl-uronium-hexa¯uorophosphate; MALDI-TOF, matrix assisted laser desorption and ionization time of ¯ight; PAL-PS, polystyrene Fmoc support for peptide amides; TFA, tri¯uoroacetic acid. E-mail address of the corresponding author: [email protected] 0022-2836/00/041091±10 $35.00/0
1994; Honig & Nicholls, 1995; Kumar & Nussivov, 1999; Perutz, 1978). They depend on the geometry and separation of the charged residues, the degree of exposure to solvent, and on neighboring residues. They include classic salt-bridges between oppositely charged residues which are in close proximity, as well as interactions involving more distant charged residues. They also include interactions between multiple residues and interactions involving charged residues with uncharged residues. While individual interactions may make minor contributions to the overall stability of a protein, numerous interactions can have a substantial effect. The role played by pH-dependent interactions in protein stability has been extensively investigated, resulting in a large volume of exper# 2000 Academic Press
1092 imental and theoretical literature (Bashford & Karplus, 1990; Blasie & Berg, 1997; Gilson et al., 1985; Hollecker & Creighton, 1982; Horovitz et al., 1990; Kohn et al., 1997; Lumb & Kim, 1995; Lyu et al., 1992; Pace et al., 1990; Sali et al., 1991; Serrano, 1994; Shaller & Robertson, 1995; Spector et al., 2000; Strop & Mayo, 2000; Waldburger et al., 1995; Warshel & Aqvist, 1991; Yang & Honig, 1993). In contrast, there have been far fewer investigations of the pH dependence of protein folding kinetics. The pH dependence of the irreversible unfolding of papain-related proteinases has been examined by the HernaÂndez-Arana group (LoÂpezArenas et al., 1999; SolõÂs-Mendiola et al., 1998), while Chan and co-workers have compared the pH dependence of the unfolding of mesophilic and thermophilic rubredoxins (Cavagnero et al., 1998). Neither the proteinases nor the rubredoxin studies examined the pH dependence of the folding rates. Fersht and co-workers analyzed the pH dependence of the folding and unfolding kinetics of CI-2, which folds via a two-state mechanism, and of barnase, which folds in a more complicated fashion (Oliveberg & Fersht, 1996; Tan et al., 1996; Tissot et al., 1996). In principle, comparison of the pHdependent stability pro®le and the pH dependence of folding rates provides information about the development of electrostatic interactions in the transition state for folding, information which is dif®cult to obtain by other means. Here, we present an analysis of the effect of pH on the stability and folding kinetics of a small mixed a/b, protein, the N-terminal domain of L9 (NTL9) (Figure 1). The N-terminal domain contains 56 residues and has been shown to fold rapidly by a two-state mechanism (Kuhlman et al., 1998b; Luisi et al., 1999). The N-terminal domain is a simple example of a common class of mixed helix sheet structures which are found in a number of other proteins. The domain lacks disul®de groups and does not require the binding of metal ions or cofactors to fold, thus it is well suited for kinetic studies. It contains a single proline residue, which is trans in the folded state. The protein is folded over the pH range of at least 1.0 to 12, making NTL9 particularly well suited for pH-dependent studies.
Results Description of the structure of NTL9 and the local environment of the charged residues The structure of NTL9 is displayed in Figure 1. The acidic residues and the N terminus are shown. There are four glutamic acid residues and two aspartic acid residues distributed throughout the structure. NTL9 also contains 12 basic residues, consisting of 11 lysine residues and one arginine residue, as well as a free N terminus. The basic residues were omitted from Figure 1 for clarity. All of the charged groups are located at the protein surface and there are no completely buried ion-
Importance of Electrostatic Interactions in NTL9
Figure 1. MOLSCRIPT (Kraulis, 1991) diagram of NTL9. The positions of the aspartic and glutamic acid residues are shown as is the location of the N terminus.
pairs, nor are there any isolated, buried charged groups. There is only one obvious salt-bridge in the structure. It involves the protonated N terminus and the side-chain of Asp23. These groups are on the surface of the protein but are not completely solvent-exposed. We have previously measured the pKa values of the acidic residues in NTL9 by two-dimensional NMR. The individual titration curves could all be ®t well by using the Henderson-Hasselbalch equation. Fitting the data with a modi®ed Hill equation did not noticeably improve the ®t, and the measured Hill coef®cients were all very close to 1. These observations strongly suggest that there are no strong cooperative interactions between the carboxylate groups (Kuhlman et al., 1999). The pKa values of Glu38, Glu48, and Glu54 are not signi®cantly perturbed from those of model compounds. The unperturbed pKa values are consistent with the structure of L9. Glu48 and Glu54 are in the C-terminal helix and both are solvent-exposed. Lysine residues at positions 45 and 51 could in principle form favorable ion-pairs but the lack of any signi®cant pKa perturbation suggests that these interactions, if present, are very weak. The side-chain of Glu38 is solvent-exposed Ê away from the nearest charged group. and is 7 A In contrast, the pKa values of Asp8, Asp23 and Glu17 are all lower than model compound values. The perturbed pKa values can be rationalized by considering the structure of NTL9. Asp8 is located at the start of the lysine-rich loop that connects the ®rst and second b-strands. The carboxylate group Ê from the posioxygen atoms of Asp23 are 1.8 A tively charged amino group at the N terminus, and the carboxylate group oxygen atoms of Glu17 form a hydrogen bond with the backbone amide of Lys14 (Kuhlman et al., 1999).
Importance of Electrostatic Interactions in NTL9
1093
Equilibrium unfolding experiments In order to probe the contribution of electrostatic interactions to the stability of NTL9, the free energy of unfolding was measured over a pH range of 1.1 to 10.45. NMR and CD experiments demonstrate that the protein remains folded over the entire pH range. We have previously measured the stability at six pH values between pH 2 and pH 7 (Kuhlman et al., 1999). Here, we have collected data at an additional 12 pH values in order to de®ne more precisely the pH stability pro®le, and we have extended our measurements to pH values above neutrality. At low pH values, Gu was determined by CD-monitored temperaturedenaturation experiments (Figure 2). Due to the loss of post-transition in the temperature-denaturation experiments at higher pH values (pH > 5.5), chemical-denaturation experiments using urea were used to determine Gu. Urea was chosen as the denaturant, since it is non-ionic. The ureadependent unfolding curves could all be ®t well using standard methods. Adequate pre and posttransition baselines were present. The results of the urea and thermal unfolding experiments are selfconsistent. At pH 5.5, the stability determined via extrapolation of the urea unfolding to zero denaturation was 4.45 (1.4, ÿ1.1) kcal molÿ1. The numbers in parentheses represent the estimated 95 % con®dence limits. Extrapolation of the thermal unfolding experiments to 25 C using the measured H(tm), tm and the known value of Cp yields a value of G (25 C) of 4.71 (0.3, ÿ0.4) kcal molÿ1. Although it was impossible to perform urea-denaturation experiments at low pH conditions at 25 C, it was possible to generate complete unfolding curves at low pH and low temperature (4 C). Again the data are entirely consistent with the extrapolation of the thermal unfolding data.
Figure 2. Thermal unfolding curves of NTL9 shown as plots of fraction unfolded versus temperature. The open circles represent data collected at pH 1.5 and the ®lled circles represent data collected at pH 5.4. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water.
Figure 3. Protein stability as a function of pH. The values of G were obtained from thermal and chemical denaturations followed by CD. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water at 25 C.
As shown in Figure 3 the Gu values display a bell-shaped dependence on pH, with maximum stability occurring near neutral pH values. The shape of the pH-dependent stability pro®le can be rationalized at least on a qualitative basis. At very low pH values the carboxylate groups are all protonated and there should be unfavorable interactions amongst the positively charged groups. As the pH is increased the carboxylate groups become deprotonated, giving rise to the possibility of favorable Coulombic interactions with the basic residues. At still higher pH values the N-terminal amino group will deprotonate, which is expected to weaken the interaction with the side-chain of Asp23. Finally, deprotonation of the lysine sidechains will diminish the interactions involving Lys14 and Glu17, and interactions involving Asp8 and the lysine-rich loop. NTL9 is a basic protein with an estimated isoelectric point of 10. 5. NTL9 is positively charged over the entire pH range studied, with the total charge varying from approximately 13 at pH 1 to 1 at pH 10. It is interesting to note that the pH of maximum stability does not correspond to the isoelectric point. The denaturant m value, which is thought to be directly related to the change in solvent-exposed surface area upon unfolding (Myers et al., 1995; Tanford, 1970), is independent of pH, with an average value of 0.72 kcal molÿ1 Mÿ1. This value is in good agreement with a previously determined m value for NTL9 (Kuhlman & Raleigh, 1998). The lack of any pH dependence on the m value strongly suggests that the change in solventexposed surface area upon unfolding is independent of the pH of the measurement. The pH-dependent thermal unfolding curves can be analyzed to determine the midpoint of the thermal denaturation curves, tm, and the enthalpy change at tm, H (tm). These values of H (tm) and tm can in turn be used to determine Cp. Since
1094
Importance of Electrostatic Interactions in NTL9
the heat capacity de®nes the temperature dependence of the enthalpy, one can accurately determine Cp from the slope of a plot of H (tm) versus tm: dH 1 Cp dT p We have previously estimated Cp to be 0.57 kcal molÿ1 based upon global analysis of the temperature and guanidine HCl denaturations at pH 5.4 (Kuhlman & Raleigh, 1998). This analysis required a multiparameter ®t involving a large number of independent variables. In principle, analysis of the pH dependence of stability should give a more precise estimate of Cp. Using this method, Cp was determined to be 0.52 kcal molÿ1 Kÿ1 (Figure 4). This value is in good agreement with the previously reported value and also agrees with estimates based upon changes in solvent-accessible surface area upon unfolding (Kuhlman & Raleigh, 1998). Kinetic measurements The pH dependence of both the folding rate kf and the unfolding rate ku were measured between pH 2.1 and pH 9.5. The folding and the unfolding rate constants in the absence of denaturant were determined by extrapolation from the plot of ln (kobs) versus [Urea], where kobs is the apparent ®rstorder rate constant and is given by the sum of kf and ku. The data were ®t assuming the standard linear dependence of ln (kf) and ln (ku) on [Urea]: Urea ln kobs ln kf 0M Urea exp mf RT Urea 2 ku 0M Urea exp mu RT where mu is the slope of the unfolding branch and mf is the slope of the folding branch. kf(0 M Urea) and ku(0 M Urea) are the folding and unfolding rate constants at 0 M urea. Representative plots of the observed rate versus concentration of urea, so-called chevron plots, are shown in Figure 5, for pH 2.1 and pH 5.5. The plots display the expected V-shape for two-state folding. We have shown that NTL9 folds via a two-state mechanism at pH 5.4 (Kuhlman et al., 1998b). All of the pH-dependent data indicate that folding remains two-state over the entire range of pH values. There is no evidence for any rollover in any of the individual chevron plots at any pH value considered. Evidence to support further the two-state nature of the folding is provided by the agreement between m values calculated from the kinetic measurements (mu ÿ mf) and those measured in equilibrium experiments. The m values for the kinetic experiments do not vary with pH, and the average m value of 0.72 kcal molÿ1 Mÿ1 is in excel-
Figure 4. A plot of H versus tm. The data were obtained from temperature-denaturation experiments at various pH values. The slope of the linear ®t yields a Cp value of 0.52 kcal molÿ1 Kÿ1. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water.
lent agreement with the value obtained by denaturation experiments. There is a slight discrepancy at the lowest pH values (below 2.5). However, the kinetic and experimental measurements still agree within the limits of the experimental uncertainty. The relative size of mu and mf provide an estimate of how much surface area is being buried in the transition state ensemble for folding (Myers et al., 1995; Tanford, 1970). The ratio of mf/(muÿmf), referred to as ym, provides one measurement of the position of the transition state. The average value of ym is 0.64, which corresponds to a relatively open transition state (Figure 6). The scatter in the data is due to the dif®culty in determining mf at low pH conditions and in determining mu at high pH conditions. At low pH values the folding branch is small, making an accurate determination of mf dif®cult. The converse is true at high pH values and it is dif®cult to measure mu accurately above pH 6.5. There is no discernible systematic variation of ym with pH or with G . This is an interesting observation, since the values of G display a signi®cant pH dependence. There is only a small increase in the folding rates between low and neutral pH values with ln kf varying from 5.06 at pH 2.1 to 6.78 at pH 5.5 (Figure 7). The change in stability, as measured by equilibrium denaturation experiments, is more dramatic, increasing from 1.90 kcal molÿ1 at pH 2.0 to 4.45 kcal molÿ1 at pH 5.4. This corresponds to a change in the natural log of the equilibrium constant, lnKeq, from 3.21 at pH 2.0 to 7.51 at pH 5.4. The change in G is primarily due to the smaller barrier for unfolding at low pH values. The natural log of unfolding rates varies by more than four units between pH 2.1 and pH 5.5. Note that because of the dif®culties in measuring ln (kf) and G at low pH, there is a small discrepancy in the variation of ln Keq calculated from the change in
1095
Importance of Electrostatic Interactions in NTL9
Figure 5. A plot of ln(kobs) versus [Urea] at pH 2.1 (open circles) and pH 5.5 (®lled circles). The continuous line represents a ®t with equation (2). The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water at 25 C.
rate constants and the change in stability. The effect is relatively minor and does not alter the conclusion that the pH-dependent stability is mainly due to changes in ku. These observations strongly suggest that interactions involving the deprotonated carboxylate side-chains of Asp and Glu are only weakly formed in the transition state for folding. As the pH is raised to even higher values, the stability of the protein decreases, but again the effect upon the folding rate is much less than the effect on the overall stability. The natural log of kf changes by less than one unit between pH 5.5 and pH 9.5. In contrast, the natural log of the equilibrium constant, ln Keq, changes by slightly more than three units over the same range. It is likely that it is the N terminus that is titrating over this pH range. This suggests that changing the ionization of the N terminus has little effect upon the folding rate but does have a signi®cant effect upon stability. This inference, if true, is very interesting, since the protonated N terminus is involved in the salt-bridge with the side-chain of Asp23. This interaction is the one clear pair-wise interaction present in the native state that involves a carboxylate group. Support for this interpretation is provided by experiments conducted with a variant which has an acetylated N terminus. This protein still folds in a two-state fashion. The value of G is 1.9 kcal molÿ1, the value of y m is 0.72 and the folding rate kf is 315 sÿ1 at pH 5.4. The chevron plot for this variant shows no rollover, and the values of G and m obtained from analysis of the kinetic data are in good agreement with the equilibrium values. Comparison to the wild-type protein allows the calculation of the -value for this ``mutation'', which is 0.23. A -value of 1 is interpreted to indicate that interactions involving the mutated group are strongly formed in the transition state, while a -value near zero is evidence that interactions
Figure 6. Plot of ym versus (a) pH and (b) G . The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water at 25 C.
involving that group are not formed. The small value for the acetylated variant suggests that interaction involving the protonated N terminus are not signi®cantly formed in the transition state and is consistent with suggestions that the speci®c saltbridge is not formed. A more rigorous test of the development of this interaction can be obtained from double mutant cycles. These experiments are in progress.
Discussion It is clear from these studies that interactions involving charged residues contribute signi®cantly to the overall stability of NTL9. NTL9 is most stable near neutral pH and the stability decreases at both higher and lower pH values. The folding rate is a maximum near neutral pH and the unfolding rate is a minimum in this region. It is interesting to note that the pH of maximum stability does not correspond to the isoelectric point. In the classical LinderstroÈm-Lang model a protein is treated as a charged sphere and is predicted to be most stable near its isoelectric point (LinderstroÈm-Lang, 1924). Several other proteins with basic or acidic isoelec-
1096
Importance of Electrostatic Interactions in NTL9
the transition state for folding of the N-terminal domain of L9. This is consistent with the openness of the transition state, (average ym, 0.64), and with the fact that the charged residues lie on the surface of the protein. A more quantitative measure of the development of electrostatic interactions in the transition state can be derived by considering the classic Wyman-Tanford linkage relationship (Wyman, 1964; Tanford, 1970): @ ln K Q @aH
3A
@G 2:303 RTQ pH @pH
3B
K is the apparent equilibrium constant for folding, aH is the activity of the protons, G is the apparent free energy of folding and Q(pH) is the difference in the number of protons bound to the folded and unfolded states, respectively. This expression has been used to treat kinetic data by assuming that a similar functional form describes the folding kinetics. G0 is replaced by G{, ln K by ln (kf) and Q{ represents the difference in the number of protons bound to the transition state and the unfolded state. The equation is often recast by integrating to obtain: Figure 7. The pH-dependence of the folding and unfolding rates of NTL9. (a) ln(kf) versus pH (b) ln(ku) versus pH. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in H2O at 25 C.
tric points have been observed to be maximally stable near neutral pH, indicating that effects other than just the overall charge on the protein are important for determining stability. If the electrostatic interactions were fully formed in the transition state for the folding of NTL9, then varying the pH would affect the stability of both the transition state and the folded state in a similar fashion. Since ln kf is related to the free energy difference between the unfolded protein and the transition state, a noticeable affect on the folding rate would be observed in this case, but only a minor change in the unfolding rate would be expected. On the other hand, if the interactions are weakly developed in the transition state, then changing the pH will effect the stability of the folded state more than the stability of the transition state. In this case, the unfolding rate would change as a function of pH, while only minor variations in the folding rate would be expected. Our experimental data are consistent with the second scenario, which leads to the conclusion that electrostatic interactions are only weakly formed in
Gz pH2 ; pH1 Gz pH2 ÿ Gz pH1 pH2 Qz pH dpH 4 2:303 RT pH1
Note that the integral Q{ can be evaluated from the difference in G{. The expression is formally equal to the difference in free energy required to charge the transition state relative to the denatured state evaluated at pH2 minus the same quantity evaluated at pH1. Note that it is not simply equal to the free energy required to charge the transition state minus the free energy required to charge the unfolded state as is sometimes stated, rather it represents the difference in those quantities evaluated at pH2 and pH1. This interpretation follows directly from the treatment of the pH dependence of protein stability (Yang & Honig, 1993). Comparison of the ratio of: pH2 pH2 pH z Q pHdpH Q pHdpH 5A b pH1
pH1
which is equivalent to the ratio: ln kf pH2 ÿ ln kf pH1 =ln K pH2 ÿ ln K pH1 5B provides a useful quantitative measure of the development of electrostatic interactions in the transition state for folding (Tan et al., 1996). The
1097
Importance of Electrostatic Interactions in NTL9
value of this ratio evaluated using pH 2 and pH 7 is a measure of the extent to which interactions involving deprotonated Asp and Glu side-chains have developed in the transition state. The experimentally measured value of the quantity: ln kf pH 2 ÿ ln kf pH 7 =ln K pH 2 ÿ ln K pH 7
6
is 0.33. The small value of this ratio indicates that interactions involving charged Asp and Glu side-chains are only partially formed in the transition state for folding. In principle, evaluation of the expression at neutral pH and at high pH would allow an estimate of the development of interactions involving positively charged residues in the transition state, however it was not possible to obtain data at a pH value high enough to ensure that the positively charge residues were fully deprotonated, and there is also the effect of deprotonating the side-chain of Tyr25 to consider. Thus the interpretation of the ratio is, in this case, slightly less straightforward. Nonetheless, the very modest change in the natural log of the folding rates between neutral and high pH compared with the signi®cantly larger change in the natural log of the equilibrium constant for folding indicates that interactions involving the positively charge side-chains are also weakly formed. The experiments described here give a broad global picture of the extent to which electrostatic interactions involving the charged acidic residues have developed in the transition state, but they do not provide information about the development of speci®c interactions. The most prominent pair-wise interaction in the native state appears to be the salt-bridge formed by the side-chain of Asp23 and the protonated N terminus. Removing the charge on the N terminus by acetylation leads to a noticeable decrease in stability, but has very little effect on the folding rate. The low -value for this ``mutant'' strongly suggests that this speci®c interaction is at best only weakly formed in the transition state. More detailed experiments involving double mutant cycles are in progress to characterize further this interaction. Although interactions involving the acidic residues are not strongly developed in the transition state, they may still play a role in the early stages of folding. Burial of charged groups in a hydrophobic environment is energetically unfavorable and interactions in the unfold state or interactions which develop early in folding may help to keep the charged residues separated from the developing hydrophobic core. It is interesting to compare the ratio determined from equation (6) to global measures of the position of the transition state. The urea or guanidine HCl (GdHCl)-dependence of folding and unfolding rates is commonly used to calcu-
late ym, which is taken as a measure of the fraction of solvent-accessible surface area buried in the transition state. For the case of NTL9, the values of ym calculated from urea, 0.64, or GdHCl, 0.60 are in good agreement. This corresponds to a relatively open transition state compared with some other proteins which fold in a two-state manner. These values are also in good agreement with estimates of the position of the transition state, yCp, obtained by analyzing the difference in heat capacity between the unfolded state and the transition state and between the transition state and the folded state. The value of yCp for NTL9 is 0.59. The values of yCp and of ym are related to the amount of surface area buried in the transition state ensemble, while the ratio de®ned by equation (6) reports on the formation of electrostatic interactions involving the acidic residues. In NTL9 these interactions involve residues which are located on the surface of the protein. The observation that they are less well developed than the burial of solventaccessible surface area likely indicates that the core of the protein is relatively compact in the transition state, while the surface is less well consolidated and is ¯uxional. Again, it is important to remember that these experiments provide a global view and it is possible, although we believe unlikely in this case, that some electrostatic interactions are strongly formed in the transition state for folding, while others are unformed. Our data obtained with the acetylated variant strongly argue that the one well-de®ned pair-wise interaction which is present in the native state is weakly formed in the transition state. The development of electrostatic interactions during the folding process has been analyzed for two other proteins. Fersht and co-workers have conducted pH-dependent folding studies of CI-2, which folds in a two-state fashion and of barnase, which folds via a more complicated process. The behavior of CI-2 and NTL9 appear to be very similar. In CI-2, the acidic residues are located at or near the surface and the ratio de®ned by equation (5A) is 0.28, while ym is 0.6 (Jackson & Fersht, 1991; Tan et al., 1996). For barnase, the electrostatic interactions appear to be better developed and the ratio de®ned by equation (5A) is close to 0.5. The Fersht group has shown that much of the difference may be due to a buried salt-bridge involving residues 69 and 93, which is almost fully formed in the transition state (Tan et al., 1996). When the effects of this interaction are factored out, the ratio decreases to a value very similar to that found for CI-2. Although it is clearly hazardous to extrapolate from such a small data set, weakly formed electrostatic interactions involving charged surface residues may be a general feature of proteins which fold through a relatively open transition state.
1098
Importance of Electrostatic Interactions in NTL9 GDÿN Urea GH2 O ÿ m Urea
Materials and Methods
9
Peptide synthesis and characterization NTL9 was prepared by solid-phase peptide synthesis using a Millipore 9050 Plus peptide synthesizer and standard Fmoc chemistry (Kuhlman et al., 1998a). Acetylated NTL9 was prepared by treating a portion of the resin with acetic anhydride prior to the deprotection and cleavage step. PAL-PS resin was purchased from Perseptive Biosystems (Framingham, MA). HBTU was purchased from Advanced Chemtech (Louisville, KT). Fmoc-protected amino acid residues were purchased from Advanced Chemtech and Perseptive Biosystems. The proteins were puri®ed using reverse phase HPLC with a C4 column. The purity was checked using analytical HPLC and was greater than 95 %. An A-B gradient was used where buffer A was 0.1 % (v/v) aqueous TFA and buffer B was 0.1 % (v/v) TFA, 90 % (v/v) isopropanol, 9.9 % (v/v) water. Matrix-assisted laser desorption time of ¯ight mass spectroscopy (MALDI-TOF) con®rmed the identity of the pure product. Circular dichroism spectroscopy Equilibrium data were recorded using an Aviv model 62A DS circular dichroism spectrometer equipped with a Peltier temperature-control system. All of the CD experiments were performed using a buffer of 100 mM sodium chloride and 10 mM sodium phosphate with protein concentrations in the range of 450 mM to 16 mM. Temperature-denaturation experiments were performed by monitoring the signal at 280 nm over the range of 2 to 98 C, with 2 deg. C steps and a 45 second equilibration time. Urea-denaturation experiments were followed at a wavelength of 222 nm. The concentration of urea was determined by refractometry (Pace et al., 1989). The experimental data, plots of ellipticity versus temperature, were ®tted with the program Kaleidagraph (Abelbeck Software) to equation (7):
f T
aN bN T aD bD Teÿ GDÿN T=RT 1 eÿ GDÿN T=RT
Stopped-¯ow ¯uorescence measurements were performed using an Applied Photophysics SX.18MV stopped-¯ow reaction analyzer equipped for asymmetric mixing at a ratio of 10:1. Fluorescence measurements were made with an excitation wavelength of 279(4) nm using a 305 nm cut-off ®lter. NTL9 was denatured in 10 M urea and folding was initiated by 11-fold dilution into 0 to 6 M urea. The unfolding of NTL9 was initiated by 11-fold dilution into 4 to 10 M urea. For experiments below pH 3.5 very small urea jumps were made because the folding branch is small. NTL9 was denatured in 3 M urea and folding was initiated by 11-fold dilution into 0 to 1.5 M urea. All solutions contained 100 mM sodium chloride and 10 mM sodium phosphate. Final protein concentrations were approximately 75 mM. The resulting curves were ®tted with a single exponential in order to determine rate constants for each reaction. Each curve was the average of three individual stopped-¯ow experiments. The dead time was 3.5 ms. The temperature of the syringes and the ¯ow-cell was maintained at 25(0.2) C with a circulating water-bath. Error analysis Upper limits of the errors were estimated by perturbing a parameter from its best ®t value and then observing how this change affects its chi-squared value for a new ®t. The parameters which are not being analyzed are allowed to change. An F-test was performed to determine what chi-squared value corresponds with the 95 % con®dence limit. The error for the analysis is determined by the range of parameter values that give chi-squared values lower than maximum allowed chi-squared determined from the F-test. This analysis does not include cross-correlation effects and, if anything, overestimates the uncertainties. This statistical procedure is described in more detail by Shoemaker et al. (1989).
7
where:
Acknowledgments
GDÿN T H tm 1 ÿ T=tm ÿ Cp tm ÿ T T ln T=tm
Stopped-flow fluorescence
8
f is the ellipticity and T is temperature. aN, bN, aD and bD are parameters that de®ne the ellipticity of the native (N) and denatured states (D). a and b describe a line with a slope equal to b and a y-intercept equal to a, tm is the transition mid-point for the temperature denaturation. H (tm) is the change in enthalpy at tm. The fraction folded at a given temperature was determined by using f(T), aN, bN, aD and bD. Thermal denaturations were monitored using near-UV CD, since measurements at these wavelengths gave ¯atter baselines. Urea, unfolding was monitored with far-UV CD, since excellent baselines were obtained at all wavelengths and far-UV CD requires less material. Thermal denaturations were 8595 % reversible. Urea-denaturations were ®tted using the analogous form of equation (7) appropriate for chemical denaturation, with:
We thank Dr Daniel F. Moriarty for his assistance with the MALDI-TOF mass spectroscopy analysis and Professor Bruce Tidor for helpful discussions. This work was supported by NSF grant MCB 9600866 to D.P.R. D.P.R. is a Pew Scholar in the Biomedical Sciences. D.L.L. was supported in part by an GAANN fellowship from the Department of Education and an NSF teaching fellowship. The NMR facility at SUNY Stony Brook is supported by a grant from the NSF, Che9413510.
References Barlow, D. J. & Thornton, J. M. (1983). Ion-pairs in proteins. J. Mol. Biol. 168, 867-885. Bashford, D. & Karplus, M. (1990). pKa's of ionizable groups in proteins: atomic detail from continuum electrostatic model. Biochemistry, 29, 10219-10225. Blasie, C. A. & Berg, J. M. (1997). Electrostatic interactions across a b-sheet. Biochemistry, 36, 62186222.
Importance of Electrostatic Interactions in NTL9 Cavagnero, S., Debe, D. A., Zhou, Z. H., Adams, M. W. W. & Chan, S. I. (1998). Kinetic role of electrostatic interactions in the unfolding of hyperthermophilic and mesophilic rubredoxin. Biochemistry, 37, 3369-3376. Gilson, M. K., Rashin, A., Fine, R. & Honig, B. H. (1985). On the calculation of electrostatic interactions in proteins. J. Mol. Biol. 183, 503-516. Hendsch, Z. S. & Tidor, B. (1994). Do salt bridges stabilize proteins? A continuum electrostatic analysis. Protein Sci. 3, 211-226. Hollecker, M. & Creighton, T. E. (1982). Effect on protein stability of reversing the charge on aminogroups Biochim. Biophys. Acta, 701, 395-404. Honig, B. & Nicholls, A. (1995). Classical electrostatics in biology and chemistry. Science, 268, 1144-1149. Horovitz, A., Serrano, L., Avron, B., Bycroft, M. & Fersht, A. R. (1990). Strength and co-operativity of contributions of surface salt bridges to protein stability. J. Mol. Biol. 216, 1031-1044. Jackson, S. E. & Fersht, A. R. (1991). Folding of chymotrypsin inhibitor 2. Evidence for a two state folding transition. Biochemistry, 30, 10436-10443. Kohn, W. D., Kay, C. M. & Hodges, R. S. (1997). Salt effects on protein stability: Two-stranded a-helical coiled-coils containing inter- or intrahelical ion pairs. J. Mol. Biol. 267, 1039-1052. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structure. J. Appl. Crystallog. 24, 946-950. Kuhlman, B., Boice, J. A., Fairman, R. & Raleigh, D. P. (1998a). Structure and stability of the N-terminal domain of the ribosomal protein L9: evidence for the rapid two-state folding. Biochemistry, 37, 10251032. Kuhlman, B., Luisi, D. L., Evans, P. A. & Raleigh, D. P. (1998b). Global analysis of the effects of temperature and denaturant on the folding and unfolding kinetics of the N-terminal domain of the protein L9. J. Mol. Biol. 284, 1661-1670. Kuhlman, B., Luisi, D. L., Young, P. & Raleigh, D. P. (1999). pKa values and the pH dependent stability of the N-terminal domain of L9 as probes of electrostatic interactions in the denatured state. Differentiation between local and nonlocal interactions. Biochemistry, 38, 4896-4903. Kuhlman, B. & Raleigh, D. P. (1998). Global analysis of the thermal and chemical denaturation of the N-terminal domain of the ribosomal protein L9 in H2O and D2O. Determination of the thermodynamic parameters, Ho, So, and Cop, and evaluation of solvent isotope effects. Protein Sci. 7, 24052412. Kumar, S. & Nussivov, R. (1999). Salt bridge stability in monomeric proteins. J. Mol. Biol. 293, 1241-1255. LinderstroÈm-Lang, C. R. (1924). The ionization of proteins. Compt. Rend. Trav. Lab. Carlsberg, Ser. Chim. 15, 29. LoÂpez-Arenas, L., SolõÂs-Mendiola, S. & HernaÂndezArana, A. (1999). Estimating the degree of expansion in the transition state for protein unfolding: analysis of the pH dependence of the rate constant for caricain denaturation. Biochemistry, 38, 1593615943. Luisi, D. L., Kuhlman, B., Sideras, K., Evans, P. A. & Raleigh, D. P. (1999). Effects of varying the local propensity to form secondary structure on the stability and the folding kinetics of a rapid folding mixed a/b protein: characterization of a truncation
1099 mutant of the N-terminal domain of the ribosomal protein L9. J. Mol. Biol. 289, 167-174. Lumb, K. J. & Kim, P. S. (1995). Measurement of interhelical electrostatic interactions in the GCN4 leucine zipper. Science, 268, 436-439. Lyu, P. C., Gans, P. J. & Kallenbach, N. R. (1992). Energetic contribution of solvent-exposed ion pairs to alpha-helix structure. J. Mol. Biol. 223, 343350. Myers, J. K., Pace, C. N. & Scholtz, J. M. (1995). Denaturant m values and heat capacity changes: relation to changes in accessible surface areas of protein unfolding. Protein Sci. 4, 2138-2148. Oliveberg, M. & Fersht, A. R. (1996). Formation of electrostatic interactions on the protein-folding pathway. Biochemistry, 35, 2726-2737. Pace, C. N., Laurents, D. V. & Thomson, J. A. (1990). pH dependence of urea and guanidine hydrochloride denaturation of ribonuclease A and ribonuclease T1. Biochemistry, 29, 2564-2572. Pace, C. N., Shirely, B. A. & Thomson, J. A. (1989). Measuring the conformational stability of a protein. In Protein Structure: A Practical Approach (Creighton, T. E., ed.), IRL press, Oxford. Perutz, M. F. (1978). Electrostatic effects in proteins. Science, 201, 1187-1191. Sali, D., Bycroft, M. & Fersht, A. R. (1991). Surface electrostatic interactions contribute little to the stability of barnase. J. Mol. Biol. 220, 779-788. Serrano, L. (1994). Protein engineering and the dissection of protein folding pathways. Curr. Opin. Struct. Biol. 4, 107-111. Shaller, W. & Robertson, A. D. (1995). pH, ionic strength, and temperature dependences of ionization equilibria for the carboxyl groups in turkey ovomucoid third domain. Biochemistry, 34, 47144723. Shoemaker, D. P., Garland, C. W. & Nibler, J. W. (1989). Experiments in Physical Chemistry, 5th edit., pp. 819820, McGraw-Hill Inc., New York. SolõÂs-Mendiola, S., GutieÂrrez-GonzaÂlez, L. H., ArroyoReyna, A., Padilla-ZuÂnÄiga, J., Rojo-DomõÂnguez, A. & HernaÂndez-Arana, A. (1998). pH dependence of the activation parameters for chymopapain unfolding: in¯uence of ion pairs on the kinetic stability of proteins. Biochim. Biophys. Acta, 1388, 363-372. Spector, S., Wang, M., Carp, S. A., Robblee, J., Hendsch, Z. S., Fairman, R., Tidor, B. & Raleigh, D. P. (2000). Rational modi®cation of protein stability by the mutation of charged surface residues. Biochemistry, 39, 872-879. Strop, P. & Mayo, S. L. (2000). Contribution of surface salt bridges to protein stability. Biochemistry, 39, 1251-1255. Tan, Y.-J., Oliveberg, M. & Fersht, A. (1996). Titration properties and thermodynamics of the transition state for folding: comparison of two-state and multi-state folding pathways. J. Mol. Biol. 264, 377389. Tanford, C. (1970). Protein denaturation. Part C. Theoretical models for the mechanism of denaturation. Advan. Protein Chem. 24, 1-95. Tissot, A. C., Vuilleumier, S. & Fersht, A. R. (1996). Importance of two buried salt bridges in the stability and folding pathway of Barnase. Biochemistry, 35, 6786-6794. Waldburger, C. D., Schildbach, J. F. & Sauer, R. T. (1995). Are buried salt bridges important for protein
1100
Importance of Electrostatic Interactions in NTL9
stability and conformational speci®city? Nature Struct. Biol. 2, 122-128. Warshel, A. & Aqvist, J. (1991). Electrostatic energy and macromolecular function. Annu. Rev. Biophys. Biophys. Chem. 20, 267-298.
Wyman, J., Jr. (1964). Linked functions and reciprocal effects in hemoglobin: a second look. Advan. Protein Chem. 16, 223-245. Yang, A.-S. & Honig, B. (1993). On the pH dependence of protein stability. J. Mol. Biol. 231, 459-474.
Edited by C. R. Matthews (Received 21 December 1999; received in revised form 29 March 2000; accepted 31 March 2000)
J. Mol. Biol. (2000) 299, 1091±1100
pH-Dependent Interactions and the Stability and Folding Kinetics of the N-terminal Domain of L9. Electrostatic Interactions are Only Weakly Formed in the Transition State for Folding Donna L. Luisi1 and Daniel P. Raleigh1,2* 1
Department of Chemistry State University of New York at Stony Brook, Stony Brook NY, 11794-3400, USA 2
Graduate Program in Biophysics and Graduate Program in Molecular and Cellular Biology
The role of electrostatic interactions in the stability and the folding of the N-terminal domain of the ribosomal protein L9 (NTL9) was investigated by determining the effects of varying the pH conditions. Urea denaturations and thermal unfolding experiments were used to measure the free energy of folding, G , at 18 different pH values, ranging from pH 1.1 to pH 10.5. Folding rates were measured at 19 pH values between pH 2.1 and pH 9.5, and unfolding rates were determined at 15 pH values in this range using stopped-¯ow ¯uorescence experiments. The protein is maximally stable between pH 5.5 and 7.5 with a value of G 4.45 kcal molÿ1. The folding rate reaches a maximum at pH 5.5, however the change in folding rates with pH is relatively modest. Over the pH range of 2.1 to 5.5 there is a small increase in folding rates, ln (kf) changes from 5.1 to 6.8. However, the change in stability is more dramatic, with a difference of 2.6 kcal molÿ1 between pH 2.0 and pH 5.4. The change in stability is largely due to the smaller barrier for unfolding at low pH values. The natural log of the unfolding rates varies by approximately four units between pH 2.1 and pH 5.5. The stability of the protein decreases above pH 7.5 and again the change is largely due to changes in the unfolding rate. ln (kf) varies by less than one unit between pH 5.5 and pH 9.5 while G decreases by 2.4 kcal molÿ1 over the range of pH 5.4 to pH 10.0, which corresponds to a change in ln Keq of 4.0. These studies show that pH-dependent interactions contribute signi®cantly to the overall stability of the protein but have only a small effect upon the folding kinetics, indicating that electrostatic interactions are weakly formed in the transition state for folding. # 2000 Academic Press
*Corresponding author
Keywords: protein stability; electrostatics; protein folding; L9
Introduction Interactions involving charged residues are thought to play an important role in protein folding and in stabilizing the native structure of proteins (Barlow & Thornton, 1983; Hendsch & Tidor, Abbreviations used: CD, circular dichroism; GdHCl, guanidine hydrochloride; Fmoc, N-9¯uorenylmethyloxycarbonyl; HBTU, O-benzotriazoleN,N,N0 ,N0 -tetramethyl-uronium-hexa¯uorophosphate; MALDI-TOF, matrix assisted laser desorption and ionization time of ¯ight; PAL-PS, polystyrene Fmoc support for peptide amides; TFA, tri¯uoroacetic acid. E-mail address of the corresponding author: [email protected] 0022-2836/00/041091±10 $35.00/0
1994; Honig & Nicholls, 1995; Kumar & Nussivov, 1999; Perutz, 1978). They depend on the geometry and separation of the charged residues, the degree of exposure to solvent, and on neighboring residues. They include classic salt-bridges between oppositely charged residues which are in close proximity, as well as interactions involving more distant charged residues. They also include interactions between multiple residues and interactions involving charged residues with uncharged residues. While individual interactions may make minor contributions to the overall stability of a protein, numerous interactions can have a substantial effect. The role played by pH-dependent interactions in protein stability has been extensively investigated, resulting in a large volume of exper# 2000 Academic Press
1092 imental and theoretical literature (Bashford & Karplus, 1990; Blasie & Berg, 1997; Gilson et al., 1985; Hollecker & Creighton, 1982; Horovitz et al., 1990; Kohn et al., 1997; Lumb & Kim, 1995; Lyu et al., 1992; Pace et al., 1990; Sali et al., 1991; Serrano, 1994; Shaller & Robertson, 1995; Spector et al., 2000; Strop & Mayo, 2000; Waldburger et al., 1995; Warshel & Aqvist, 1991; Yang & Honig, 1993). In contrast, there have been far fewer investigations of the pH dependence of protein folding kinetics. The pH dependence of the irreversible unfolding of papain-related proteinases has been examined by the HernaÂndez-Arana group (LoÂpezArenas et al., 1999; SolõÂs-Mendiola et al., 1998), while Chan and co-workers have compared the pH dependence of the unfolding of mesophilic and thermophilic rubredoxins (Cavagnero et al., 1998). Neither the proteinases nor the rubredoxin studies examined the pH dependence of the folding rates. Fersht and co-workers analyzed the pH dependence of the folding and unfolding kinetics of CI-2, which folds via a two-state mechanism, and of barnase, which folds in a more complicated fashion (Oliveberg & Fersht, 1996; Tan et al., 1996; Tissot et al., 1996). In principle, comparison of the pHdependent stability pro®le and the pH dependence of folding rates provides information about the development of electrostatic interactions in the transition state for folding, information which is dif®cult to obtain by other means. Here, we present an analysis of the effect of pH on the stability and folding kinetics of a small mixed a/b, protein, the N-terminal domain of L9 (NTL9) (Figure 1). The N-terminal domain contains 56 residues and has been shown to fold rapidly by a two-state mechanism (Kuhlman et al., 1998b; Luisi et al., 1999). The N-terminal domain is a simple example of a common class of mixed helix sheet structures which are found in a number of other proteins. The domain lacks disul®de groups and does not require the binding of metal ions or cofactors to fold, thus it is well suited for kinetic studies. It contains a single proline residue, which is trans in the folded state. The protein is folded over the pH range of at least 1.0 to 12, making NTL9 particularly well suited for pH-dependent studies.
Results Description of the structure of NTL9 and the local environment of the charged residues The structure of NTL9 is displayed in Figure 1. The acidic residues and the N terminus are shown. There are four glutamic acid residues and two aspartic acid residues distributed throughout the structure. NTL9 also contains 12 basic residues, consisting of 11 lysine residues and one arginine residue, as well as a free N terminus. The basic residues were omitted from Figure 1 for clarity. All of the charged groups are located at the protein surface and there are no completely buried ion-
Importance of Electrostatic Interactions in NTL9
Figure 1. MOLSCRIPT (Kraulis, 1991) diagram of NTL9. The positions of the aspartic and glutamic acid residues are shown as is the location of the N terminus.
pairs, nor are there any isolated, buried charged groups. There is only one obvious salt-bridge in the structure. It involves the protonated N terminus and the side-chain of Asp23. These groups are on the surface of the protein but are not completely solvent-exposed. We have previously measured the pKa values of the acidic residues in NTL9 by two-dimensional NMR. The individual titration curves could all be ®t well by using the Henderson-Hasselbalch equation. Fitting the data with a modi®ed Hill equation did not noticeably improve the ®t, and the measured Hill coef®cients were all very close to 1. These observations strongly suggest that there are no strong cooperative interactions between the carboxylate groups (Kuhlman et al., 1999). The pKa values of Glu38, Glu48, and Glu54 are not signi®cantly perturbed from those of model compounds. The unperturbed pKa values are consistent with the structure of L9. Glu48 and Glu54 are in the C-terminal helix and both are solvent-exposed. Lysine residues at positions 45 and 51 could in principle form favorable ion-pairs but the lack of any signi®cant pKa perturbation suggests that these interactions, if present, are very weak. The side-chain of Glu38 is solvent-exposed Ê away from the nearest charged group. and is 7 A In contrast, the pKa values of Asp8, Asp23 and Glu17 are all lower than model compound values. The perturbed pKa values can be rationalized by considering the structure of NTL9. Asp8 is located at the start of the lysine-rich loop that connects the ®rst and second b-strands. The carboxylate group Ê from the posioxygen atoms of Asp23 are 1.8 A tively charged amino group at the N terminus, and the carboxylate group oxygen atoms of Glu17 form a hydrogen bond with the backbone amide of Lys14 (Kuhlman et al., 1999).
Importance of Electrostatic Interactions in NTL9
1093
Equilibrium unfolding experiments In order to probe the contribution of electrostatic interactions to the stability of NTL9, the free energy of unfolding was measured over a pH range of 1.1 to 10.45. NMR and CD experiments demonstrate that the protein remains folded over the entire pH range. We have previously measured the stability at six pH values between pH 2 and pH 7 (Kuhlman et al., 1999). Here, we have collected data at an additional 12 pH values in order to de®ne more precisely the pH stability pro®le, and we have extended our measurements to pH values above neutrality. At low pH values, Gu was determined by CD-monitored temperaturedenaturation experiments (Figure 2). Due to the loss of post-transition in the temperature-denaturation experiments at higher pH values (pH > 5.5), chemical-denaturation experiments using urea were used to determine Gu. Urea was chosen as the denaturant, since it is non-ionic. The ureadependent unfolding curves could all be ®t well using standard methods. Adequate pre and posttransition baselines were present. The results of the urea and thermal unfolding experiments are selfconsistent. At pH 5.5, the stability determined via extrapolation of the urea unfolding to zero denaturation was 4.45 (1.4, ÿ1.1) kcal molÿ1. The numbers in parentheses represent the estimated 95 % con®dence limits. Extrapolation of the thermal unfolding experiments to 25 C using the measured H(tm), tm and the known value of Cp yields a value of G (25 C) of 4.71 (0.3, ÿ0.4) kcal molÿ1. Although it was impossible to perform urea-denaturation experiments at low pH conditions at 25 C, it was possible to generate complete unfolding curves at low pH and low temperature (4 C). Again the data are entirely consistent with the extrapolation of the thermal unfolding data.
Figure 2. Thermal unfolding curves of NTL9 shown as plots of fraction unfolded versus temperature. The open circles represent data collected at pH 1.5 and the ®lled circles represent data collected at pH 5.4. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water.
Figure 3. Protein stability as a function of pH. The values of G were obtained from thermal and chemical denaturations followed by CD. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water at 25 C.
As shown in Figure 3 the Gu values display a bell-shaped dependence on pH, with maximum stability occurring near neutral pH values. The shape of the pH-dependent stability pro®le can be rationalized at least on a qualitative basis. At very low pH values the carboxylate groups are all protonated and there should be unfavorable interactions amongst the positively charged groups. As the pH is increased the carboxylate groups become deprotonated, giving rise to the possibility of favorable Coulombic interactions with the basic residues. At still higher pH values the N-terminal amino group will deprotonate, which is expected to weaken the interaction with the side-chain of Asp23. Finally, deprotonation of the lysine sidechains will diminish the interactions involving Lys14 and Glu17, and interactions involving Asp8 and the lysine-rich loop. NTL9 is a basic protein with an estimated isoelectric point of 10. 5. NTL9 is positively charged over the entire pH range studied, with the total charge varying from approximately 13 at pH 1 to 1 at pH 10. It is interesting to note that the pH of maximum stability does not correspond to the isoelectric point. The denaturant m value, which is thought to be directly related to the change in solvent-exposed surface area upon unfolding (Myers et al., 1995; Tanford, 1970), is independent of pH, with an average value of 0.72 kcal molÿ1 Mÿ1. This value is in good agreement with a previously determined m value for NTL9 (Kuhlman & Raleigh, 1998). The lack of any pH dependence on the m value strongly suggests that the change in solventexposed surface area upon unfolding is independent of the pH of the measurement. The pH-dependent thermal unfolding curves can be analyzed to determine the midpoint of the thermal denaturation curves, tm, and the enthalpy change at tm, H (tm). These values of H (tm) and tm can in turn be used to determine Cp. Since
1094
Importance of Electrostatic Interactions in NTL9
the heat capacity de®nes the temperature dependence of the enthalpy, one can accurately determine Cp from the slope of a plot of H (tm) versus tm: dH 1 Cp dT p We have previously estimated Cp to be 0.57 kcal molÿ1 based upon global analysis of the temperature and guanidine HCl denaturations at pH 5.4 (Kuhlman & Raleigh, 1998). This analysis required a multiparameter ®t involving a large number of independent variables. In principle, analysis of the pH dependence of stability should give a more precise estimate of Cp. Using this method, Cp was determined to be 0.52 kcal molÿ1 Kÿ1 (Figure 4). This value is in good agreement with the previously reported value and also agrees with estimates based upon changes in solvent-accessible surface area upon unfolding (Kuhlman & Raleigh, 1998). Kinetic measurements The pH dependence of both the folding rate kf and the unfolding rate ku were measured between pH 2.1 and pH 9.5. The folding and the unfolding rate constants in the absence of denaturant were determined by extrapolation from the plot of ln (kobs) versus [Urea], where kobs is the apparent ®rstorder rate constant and is given by the sum of kf and ku. The data were ®t assuming the standard linear dependence of ln (kf) and ln (ku) on [Urea]: Urea ln kobs ln kf 0M Urea exp mf RT Urea 2 ku 0M Urea exp mu RT where mu is the slope of the unfolding branch and mf is the slope of the folding branch. kf(0 M Urea) and ku(0 M Urea) are the folding and unfolding rate constants at 0 M urea. Representative plots of the observed rate versus concentration of urea, so-called chevron plots, are shown in Figure 5, for pH 2.1 and pH 5.5. The plots display the expected V-shape for two-state folding. We have shown that NTL9 folds via a two-state mechanism at pH 5.4 (Kuhlman et al., 1998b). All of the pH-dependent data indicate that folding remains two-state over the entire range of pH values. There is no evidence for any rollover in any of the individual chevron plots at any pH value considered. Evidence to support further the two-state nature of the folding is provided by the agreement between m values calculated from the kinetic measurements (mu ÿ mf) and those measured in equilibrium experiments. The m values for the kinetic experiments do not vary with pH, and the average m value of 0.72 kcal molÿ1 Mÿ1 is in excel-
Figure 4. A plot of H versus tm. The data were obtained from temperature-denaturation experiments at various pH values. The slope of the linear ®t yields a Cp value of 0.52 kcal molÿ1 Kÿ1. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water.
lent agreement with the value obtained by denaturation experiments. There is a slight discrepancy at the lowest pH values (below 2.5). However, the kinetic and experimental measurements still agree within the limits of the experimental uncertainty. The relative size of mu and mf provide an estimate of how much surface area is being buried in the transition state ensemble for folding (Myers et al., 1995; Tanford, 1970). The ratio of mf/(muÿmf), referred to as ym, provides one measurement of the position of the transition state. The average value of ym is 0.64, which corresponds to a relatively open transition state (Figure 6). The scatter in the data is due to the dif®culty in determining mf at low pH conditions and in determining mu at high pH conditions. At low pH values the folding branch is small, making an accurate determination of mf dif®cult. The converse is true at high pH values and it is dif®cult to measure mu accurately above pH 6.5. There is no discernible systematic variation of ym with pH or with G . This is an interesting observation, since the values of G display a signi®cant pH dependence. There is only a small increase in the folding rates between low and neutral pH values with ln kf varying from 5.06 at pH 2.1 to 6.78 at pH 5.5 (Figure 7). The change in stability, as measured by equilibrium denaturation experiments, is more dramatic, increasing from 1.90 kcal molÿ1 at pH 2.0 to 4.45 kcal molÿ1 at pH 5.4. This corresponds to a change in the natural log of the equilibrium constant, lnKeq, from 3.21 at pH 2.0 to 7.51 at pH 5.4. The change in G is primarily due to the smaller barrier for unfolding at low pH values. The natural log of unfolding rates varies by more than four units between pH 2.1 and pH 5.5. Note that because of the dif®culties in measuring ln (kf) and G at low pH, there is a small discrepancy in the variation of ln Keq calculated from the change in
1095
Importance of Electrostatic Interactions in NTL9
Figure 5. A plot of ln(kobs) versus [Urea] at pH 2.1 (open circles) and pH 5.5 (®lled circles). The continuous line represents a ®t with equation (2). The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water at 25 C.
rate constants and the change in stability. The effect is relatively minor and does not alter the conclusion that the pH-dependent stability is mainly due to changes in ku. These observations strongly suggest that interactions involving the deprotonated carboxylate side-chains of Asp and Glu are only weakly formed in the transition state for folding. As the pH is raised to even higher values, the stability of the protein decreases, but again the effect upon the folding rate is much less than the effect on the overall stability. The natural log of kf changes by less than one unit between pH 5.5 and pH 9.5. In contrast, the natural log of the equilibrium constant, ln Keq, changes by slightly more than three units over the same range. It is likely that it is the N terminus that is titrating over this pH range. This suggests that changing the ionization of the N terminus has little effect upon the folding rate but does have a signi®cant effect upon stability. This inference, if true, is very interesting, since the protonated N terminus is involved in the salt-bridge with the side-chain of Asp23. This interaction is the one clear pair-wise interaction present in the native state that involves a carboxylate group. Support for this interpretation is provided by experiments conducted with a variant which has an acetylated N terminus. This protein still folds in a two-state fashion. The value of G is 1.9 kcal molÿ1, the value of y m is 0.72 and the folding rate kf is 315 sÿ1 at pH 5.4. The chevron plot for this variant shows no rollover, and the values of G and m obtained from analysis of the kinetic data are in good agreement with the equilibrium values. Comparison to the wild-type protein allows the calculation of the -value for this ``mutation'', which is 0.23. A -value of 1 is interpreted to indicate that interactions involving the mutated group are strongly formed in the transition state, while a -value near zero is evidence that interactions
Figure 6. Plot of ym versus (a) pH and (b) G . The solutions contained 100 mM NaCl, 10 mM sodium phosphate in water at 25 C.
involving that group are not formed. The small value for the acetylated variant suggests that interaction involving the protonated N terminus are not signi®cantly formed in the transition state and is consistent with suggestions that the speci®c saltbridge is not formed. A more rigorous test of the development of this interaction can be obtained from double mutant cycles. These experiments are in progress.
Discussion It is clear from these studies that interactions involving charged residues contribute signi®cantly to the overall stability of NTL9. NTL9 is most stable near neutral pH and the stability decreases at both higher and lower pH values. The folding rate is a maximum near neutral pH and the unfolding rate is a minimum in this region. It is interesting to note that the pH of maximum stability does not correspond to the isoelectric point. In the classical LinderstroÈm-Lang model a protein is treated as a charged sphere and is predicted to be most stable near its isoelectric point (LinderstroÈm-Lang, 1924). Several other proteins with basic or acidic isoelec-
1096
Importance of Electrostatic Interactions in NTL9
the transition state for folding of the N-terminal domain of L9. This is consistent with the openness of the transition state, (average ym, 0.64), and with the fact that the charged residues lie on the surface of the protein. A more quantitative measure of the development of electrostatic interactions in the transition state can be derived by considering the classic Wyman-Tanford linkage relationship (Wyman, 1964; Tanford, 1970): @ ln K Q @aH
3A
@G 2:303 RTQ pH @pH
3B
K is the apparent equilibrium constant for folding, aH is the activity of the protons, G is the apparent free energy of folding and Q(pH) is the difference in the number of protons bound to the folded and unfolded states, respectively. This expression has been used to treat kinetic data by assuming that a similar functional form describes the folding kinetics. G0 is replaced by G{, ln K by ln (kf) and Q{ represents the difference in the number of protons bound to the transition state and the unfolded state. The equation is often recast by integrating to obtain: Figure 7. The pH-dependence of the folding and unfolding rates of NTL9. (a) ln(kf) versus pH (b) ln(ku) versus pH. The solutions contained 100 mM NaCl, 10 mM sodium phosphate in H2O at 25 C.
tric points have been observed to be maximally stable near neutral pH, indicating that effects other than just the overall charge on the protein are important for determining stability. If the electrostatic interactions were fully formed in the transition state for the folding of NTL9, then varying the pH would affect the stability of both the transition state and the folded state in a similar fashion. Since ln kf is related to the free energy difference between the unfolded protein and the transition state, a noticeable affect on the folding rate would be observed in this case, but only a minor change in the unfolding rate would be expected. On the other hand, if the interactions are weakly developed in the transition state, then changing the pH will effect the stability of the folded state more than the stability of the transition state. In this case, the unfolding rate would change as a function of pH, while only minor variations in the folding rate would be expected. Our experimental data are consistent with the second scenario, which leads to the conclusion that electrostatic interactions are only weakly formed in
Gz pH2 ; pH1 Gz pH2 ÿ Gz pH1 pH2 Qz pH dpH 4 2:303 RT pH1
Note that the integral Q{ can be evaluated from the difference in G{. The expression is formally equal to the difference in free energy required to charge the transition state relative to the denatured state evaluated at pH2 minus the same quantity evaluated at pH1. Note that it is not simply equal to the free energy required to charge the transition state minus the free energy required to charge the unfolded state as is sometimes stated, rather it represents the difference in those quantities evaluated at pH2 and pH1. This interpretation follows directly from the treatment of the pH dependence of protein stability (Yang & Honig, 1993). Comparison of the ratio of: pH2 pH2 pH z Q pHdpH Q pHdpH 5A b pH1
pH1
which is equivalent to the ratio: ln kf pH2 ÿ ln kf pH1 =ln K pH2 ÿ ln K pH1 5B provides a useful quantitative measure of the development of electrostatic interactions in the transition state for folding (Tan et al., 1996). The
1097
Importance of Electrostatic Interactions in NTL9
value of this ratio evaluated using pH 2 and pH 7 is a measure of the extent to which interactions involving deprotonated Asp and Glu side-chains have developed in the transition state. The experimentally measured value of the quantity: ln kf pH 2 ÿ ln kf pH 7 =ln K pH 2 ÿ ln K pH 7
6
is 0.33. The small value of this ratio indicates that interactions involving charged Asp and Glu side-chains are only partially formed in the transition state for folding. In principle, evaluation of the expression at neutral pH and at high pH would allow an estimate of the development of interactions involving positively charged residues in the transition state, however it was not possible to obtain data at a pH value high enough to ensure that the positively charge residues were fully deprotonated, and there is also the effect of deprotonating the side-chain of Tyr25 to consider. Thus the interpretation of the ratio is, in this case, slightly less straightforward. Nonetheless, the very modest change in the natural log of the folding rates between neutral and high pH compared with the signi®cantly larger change in the natural log of the equilibrium constant for folding indicates that interactions involving the positively charge side-chains are also weakly formed. The experiments described here give a broad global picture of the extent to which electrostatic interactions involving the charged acidic residues have developed in the transition state, but they do not provide information about the development of speci®c interactions. The most prominent pair-wise interaction in the native state appears to be the salt-bridge formed by the side-chain of Asp23 and the protonated N terminus. Removing the charge on the N terminus by acetylation leads to a noticeable decrease in stability, but has very little effect on the folding rate. The low -value for this ``mutant'' strongly suggests that this speci®c interaction is at best only weakly formed in the transition state. More detailed experiments involving double mutant cycles are in progress to characterize further this interaction. Although interactions involving the acidic residues are not strongly developed in the transition state, they may still play a role in the early stages of folding. Burial of charged groups in a hydrophobic environment is energetically unfavorable and interactions in the unfold state or interactions which develop early in folding may help to keep the charged residues separated from the developing hydrophobic core. It is interesting to compare the ratio determined from equation (6) to global measures of the position of the transition state. The urea or guanidine HCl (GdHCl)-dependence of folding and unfolding rates is commonly used to calcu-
late ym, which is taken as a measure of the fraction of solvent-accessible surface area buried in the transition state. For the case of NTL9, the values of ym calculated from urea, 0.64, or GdHCl, 0.60 are in good agreement. This corresponds to a relatively open transition state compared with some other proteins which fold in a two-state manner. These values are also in good agreement with estimates of the position of the transition state, yCp, obtained by analyzing the difference in heat capacity between the unfolded state and the transition state and between the transition state and the folded state. The value of yCp for NTL9 is 0.59. The values of yCp and of ym are related to the amount of surface area buried in the transition state ensemble, while the ratio de®ned by equation (6) reports on the formation of electrostatic interactions involving the acidic residues. In NTL9 these interactions involve residues which are located on the surface of the protein. The observation that they are less well developed than the burial of solventaccessible surface area likely indicates that the core of the protein is relatively compact in the transition state, while the surface is less well consolidated and is ¯uxional. Again, it is important to remember that these experiments provide a global view and it is possible, although we believe unlikely in this case, that some electrostatic interactions are strongly formed in the transition state for folding, while others are unformed. Our data obtained with the acetylated variant strongly argue that the one well-de®ned pair-wise interaction which is present in the native state is weakly formed in the transition state. The development of electrostatic interactions during the folding process has been analyzed for two other proteins. Fersht and co-workers have conducted pH-dependent folding studies of CI-2, which folds in a two-state fashion and of barnase, which folds via a more complicated process. The behavior of CI-2 and NTL9 appear to be very similar. In CI-2, the acidic residues are located at or near the surface and the ratio de®ned by equation (5A) is 0.28, while ym is 0.6 (Jackson & Fersht, 1991; Tan et al., 1996). For barnase, the electrostatic interactions appear to be better developed and the ratio de®ned by equation (5A) is close to 0.5. The Fersht group has shown that much of the difference may be due to a buried salt-bridge involving residues 69 and 93, which is almost fully formed in the transition state (Tan et al., 1996). When the effects of this interaction are factored out, the ratio decreases to a value very similar to that found for CI-2. Although it is clearly hazardous to extrapolate from such a small data set, weakly formed electrostatic interactions involving charged surface residues may be a general feature of proteins which fold through a relatively open transition state.
1098
Importance of Electrostatic Interactions in NTL9 GDÿN Urea GH2 O ÿ m Urea
Materials and Methods
9
Peptide synthesis and characterization NTL9 was prepared by solid-phase peptide synthesis using a Millipore 9050 Plus peptide synthesizer and standard Fmoc chemistry (Kuhlman et al., 1998a). Acetylated NTL9 was prepared by treating a portion of the resin with acetic anhydride prior to the deprotection and cleavage step. PAL-PS resin was purchased from Perseptive Biosystems (Framingham, MA). HBTU was purchased from Advanced Chemtech (Louisville, KT). Fmoc-protected amino acid residues were purchased from Advanced Chemtech and Perseptive Biosystems. The proteins were puri®ed using reverse phase HPLC with a C4 column. The purity was checked using analytical HPLC and was greater than 95 %. An A-B gradient was used where buffer A was 0.1 % (v/v) aqueous TFA and buffer B was 0.1 % (v/v) TFA, 90 % (v/v) isopropanol, 9.9 % (v/v) water. Matrix-assisted laser desorption time of ¯ight mass spectroscopy (MALDI-TOF) con®rmed the identity of the pure product. Circular dichroism spectroscopy Equilibrium data were recorded using an Aviv model 62A DS circular dichroism spectrometer equipped with a Peltier temperature-control system. All of the CD experiments were performed using a buffer of 100 mM sodium chloride and 10 mM sodium phosphate with protein concentrations in the range of 450 mM to 16 mM. Temperature-denaturation experiments were performed by monitoring the signal at 280 nm over the range of 2 to 98 C, with 2 deg. C steps and a 45 second equilibration time. Urea-denaturation experiments were followed at a wavelength of 222 nm. The concentration of urea was determined by refractometry (Pace et al., 1989). The experimental data, plots of ellipticity versus temperature, were ®tted with the program Kaleidagraph (Abelbeck Software) to equation (7):
f T
aN bN T aD bD Teÿ GDÿN T=RT 1 eÿ GDÿN T=RT
Stopped-¯ow ¯uorescence measurements were performed using an Applied Photophysics SX.18MV stopped-¯ow reaction analyzer equipped for asymmetric mixing at a ratio of 10:1. Fluorescence measurements were made with an excitation wavelength of 279(4) nm using a 305 nm cut-off ®lter. NTL9 was denatured in 10 M urea and folding was initiated by 11-fold dilution into 0 to 6 M urea. The unfolding of NTL9 was initiated by 11-fold dilution into 4 to 10 M urea. For experiments below pH 3.5 very small urea jumps were made because the folding branch is small. NTL9 was denatured in 3 M urea and folding was initiated by 11-fold dilution into 0 to 1.5 M urea. All solutions contained 100 mM sodium chloride and 10 mM sodium phosphate. Final protein concentrations were approximately 75 mM. The resulting curves were ®tted with a single exponential in order to determine rate constants for each reaction. Each curve was the average of three individual stopped-¯ow experiments. The dead time was 3.5 ms. The temperature of the syringes and the ¯ow-cell was maintained at 25(0.2) C with a circulating water-bath. Error analysis Upper limits of the errors were estimated by perturbing a parameter from its best ®t value and then observing how this change affects its chi-squared value for a new ®t. The parameters which are not being analyzed are allowed to change. An F-test was performed to determine what chi-squared value corresponds with the 95 % con®dence limit. The error for the analysis is determined by the range of parameter values that give chi-squared values lower than maximum allowed chi-squared determined from the F-test. This analysis does not include cross-correlation effects and, if anything, overestimates the uncertainties. This statistical procedure is described in more detail by Shoemaker et al. (1989).
7
where:
Acknowledgments
GDÿN T H tm 1 ÿ T=tm ÿ Cp tm ÿ T T ln T=tm
Stopped-flow fluorescence
8
f is the ellipticity and T is temperature. aN, bN, aD and bD are parameters that de®ne the ellipticity of the native (N) and denatured states (D). a and b describe a line with a slope equal to b and a y-intercept equal to a, tm is the transition mid-point for the temperature denaturation. H (tm) is the change in enthalpy at tm. The fraction folded at a given temperature was determined by using f(T), aN, bN, aD and bD. Thermal denaturations were monitored using near-UV CD, since measurements at these wavelengths gave ¯atter baselines. Urea, unfolding was monitored with far-UV CD, since excellent baselines were obtained at all wavelengths and far-UV CD requires less material. Thermal denaturations were 8595 % reversible. Urea-denaturations were ®tted using the analogous form of equation (7) appropriate for chemical denaturation, with:
We thank Dr Daniel F. Moriarty for his assistance with the MALDI-TOF mass spectroscopy analysis and Professor Bruce Tidor for helpful discussions. This work was supported by NSF grant MCB 9600866 to D.P.R. D.P.R. is a Pew Scholar in the Biomedical Sciences. D.L.L. was supported in part by an GAANN fellowship from the Department of Education and an NSF teaching fellowship. The NMR facility at SUNY Stony Brook is supported by a grant from the NSF, Che9413510.
References Barlow, D. J. & Thornton, J. M. (1983). Ion-pairs in proteins. J. Mol. Biol. 168, 867-885. Bashford, D. & Karplus, M. (1990). pKa's of ionizable groups in proteins: atomic detail from continuum electrostatic model. Biochemistry, 29, 10219-10225. Blasie, C. A. & Berg, J. M. (1997). Electrostatic interactions across a b-sheet. Biochemistry, 36, 62186222.
Importance of Electrostatic Interactions in NTL9 Cavagnero, S., Debe, D. A., Zhou, Z. H., Adams, M. W. W. & Chan, S. I. (1998). Kinetic role of electrostatic interactions in the unfolding of hyperthermophilic and mesophilic rubredoxin. Biochemistry, 37, 3369-3376. Gilson, M. K., Rashin, A., Fine, R. & Honig, B. H. (1985). On the calculation of electrostatic interactions in proteins. J. Mol. Biol. 183, 503-516. Hendsch, Z. S. & Tidor, B. (1994). Do salt bridges stabilize proteins? A continuum electrostatic analysis. Protein Sci. 3, 211-226. Hollecker, M. & Creighton, T. E. (1982). Effect on protein stability of reversing the charge on aminogroups Biochim. Biophys. Acta, 701, 395-404. Honig, B. & Nicholls, A. (1995). Classical electrostatics in biology and chemistry. Science, 268, 1144-1149. Horovitz, A., Serrano, L., Avron, B., Bycroft, M. & Fersht, A. R. (1990). Strength and co-operativity of contributions of surface salt bridges to protein stability. J. Mol. Biol. 216, 1031-1044. Jackson, S. E. & Fersht, A. R. (1991). Folding of chymotrypsin inhibitor 2. Evidence for a two state folding transition. Biochemistry, 30, 10436-10443. Kohn, W. D., Kay, C. M. & Hodges, R. S. (1997). Salt effects on protein stability: Two-stranded a-helical coiled-coils containing inter- or intrahelical ion pairs. J. Mol. Biol. 267, 1039-1052. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structure. J. Appl. Crystallog. 24, 946-950. Kuhlman, B., Boice, J. A., Fairman, R. & Raleigh, D. P. (1998a). Structure and stability of the N-terminal domain of the ribosomal protein L9: evidence for the rapid two-state folding. Biochemistry, 37, 10251032. Kuhlman, B., Luisi, D. L., Evans, P. A. & Raleigh, D. P. (1998b). Global analysis of the effects of temperature and denaturant on the folding and unfolding kinetics of the N-terminal domain of the protein L9. J. Mol. Biol. 284, 1661-1670. Kuhlman, B., Luisi, D. L., Young, P. & Raleigh, D. P. (1999). pKa values and the pH dependent stability of the N-terminal domain of L9 as probes of electrostatic interactions in the denatured state. Differentiation between local and nonlocal interactions. Biochemistry, 38, 4896-4903. Kuhlman, B. & Raleigh, D. P. (1998). Global analysis of the thermal and chemical denaturation of the N-terminal domain of the ribosomal protein L9 in H2O and D2O. Determination of the thermodynamic parameters, Ho, So, and Cop, and evaluation of solvent isotope effects. Protein Sci. 7, 24052412. Kumar, S. & Nussivov, R. (1999). Salt bridge stability in monomeric proteins. J. Mol. Biol. 293, 1241-1255. LinderstroÈm-Lang, C. R. (1924). The ionization of proteins. Compt. Rend. Trav. Lab. Carlsberg, Ser. Chim. 15, 29. LoÂpez-Arenas, L., SolõÂs-Mendiola, S. & HernaÂndezArana, A. (1999). Estimating the degree of expansion in the transition state for protein unfolding: analysis of the pH dependence of the rate constant for caricain denaturation. Biochemistry, 38, 1593615943. Luisi, D. L., Kuhlman, B., Sideras, K., Evans, P. A. & Raleigh, D. P. (1999). Effects of varying the local propensity to form secondary structure on the stability and the folding kinetics of a rapid folding mixed a/b protein: characterization of a truncation
1099 mutant of the N-terminal domain of the ribosomal protein L9. J. Mol. Biol. 289, 167-174. Lumb, K. J. & Kim, P. S. (1995). Measurement of interhelical electrostatic interactions in the GCN4 leucine zipper. Science, 268, 436-439. Lyu, P. C., Gans, P. J. & Kallenbach, N. R. (1992). Energetic contribution of solvent-exposed ion pairs to alpha-helix structure. J. Mol. Biol. 223, 343350. Myers, J. K., Pace, C. N. & Scholtz, J. M. (1995). Denaturant m values and heat capacity changes: relation to changes in accessible surface areas of protein unfolding. Protein Sci. 4, 2138-2148. Oliveberg, M. & Fersht, A. R. (1996). Formation of electrostatic interactions on the protein-folding pathway. Biochemistry, 35, 2726-2737. Pace, C. N., Laurents, D. V. & Thomson, J. A. (1990). pH dependence of urea and guanidine hydrochloride denaturation of ribonuclease A and ribonuclease T1. Biochemistry, 29, 2564-2572. Pace, C. N., Shirely, B. A. & Thomson, J. A. (1989). Measuring the conformational stability of a protein. In Protein Structure: A Practical Approach (Creighton, T. E., ed.), IRL press, Oxford. Perutz, M. F. (1978). Electrostatic effects in proteins. Science, 201, 1187-1191. Sali, D., Bycroft, M. & Fersht, A. R. (1991). Surface electrostatic interactions contribute little to the stability of barnase. J. Mol. Biol. 220, 779-788. Serrano, L. (1994). Protein engineering and the dissection of protein folding pathways. Curr. Opin. Struct. Biol. 4, 107-111. Shaller, W. & Robertson, A. D. (1995). pH, ionic strength, and temperature dependences of ionization equilibria for the carboxyl groups in turkey ovomucoid third domain. Biochemistry, 34, 47144723. Shoemaker, D. P., Garland, C. W. & Nibler, J. W. (1989). Experiments in Physical Chemistry, 5th edit., pp. 819820, McGraw-Hill Inc., New York. SolõÂs-Mendiola, S., GutieÂrrez-GonzaÂlez, L. H., ArroyoReyna, A., Padilla-ZuÂnÄiga, J., Rojo-DomõÂnguez, A. & HernaÂndez-Arana, A. (1998). pH dependence of the activation parameters for chymopapain unfolding: in¯uence of ion pairs on the kinetic stability of proteins. Biochim. Biophys. Acta, 1388, 363-372. Spector, S., Wang, M., Carp, S. A., Robblee, J., Hendsch, Z. S., Fairman, R., Tidor, B. & Raleigh, D. P. (2000). Rational modi®cation of protein stability by the mutation of charged surface residues. Biochemistry, 39, 872-879. Strop, P. & Mayo, S. L. (2000). Contribution of surface salt bridges to protein stability. Biochemistry, 39, 1251-1255. Tan, Y.-J., Oliveberg, M. & Fersht, A. (1996). Titration properties and thermodynamics of the transition state for folding: comparison of two-state and multi-state folding pathways. J. Mol. Biol. 264, 377389. Tanford, C. (1970). Protein denaturation. Part C. Theoretical models for the mechanism of denaturation. Advan. Protein Chem. 24, 1-95. Tissot, A. C., Vuilleumier, S. & Fersht, A. R. (1996). Importance of two buried salt bridges in the stability and folding pathway of Barnase. Biochemistry, 35, 6786-6794. Waldburger, C. D., Schildbach, J. F. & Sauer, R. T. (1995). Are buried salt bridges important for protein
1100
Importance of Electrostatic Interactions in NTL9
stability and conformational speci®city? Nature Struct. Biol. 2, 122-128. Warshel, A. & Aqvist, J. (1991). Electrostatic energy and macromolecular function. Annu. Rev. Biophys. Biophys. Chem. 20, 267-298.
Wyman, J., Jr. (1964). Linked functions and reciprocal effects in hemoglobin: a second look. Advan. Protein Chem. 16, 223-245. Yang, A.-S. & Honig, B. (1993). On the pH dependence of protein stability. J. Mol. Biol. 231, 459-474.
Edited by C. R. Matthews (Received 21 December 1999; received in revised form 29 March 2000; accepted 31 March 2000)