- Email: [email protected]
[26] Effect of point mutations of the folding of globular proteins
498
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
could also reproduce the properties of the natural sequence. 30In this way, the role of the amphiphilic structure has been more thoroughly defined, and evidence suggesting that this structure is the regular a helix and not a 7r helix or other less regular structure has been obtained. Differences observed in the physicochemical properties of the model peptide described here compared to those of/3-endorphin provide a rationale for some of the observed differences in its behavior in the biological assays. Thus, tighter binding to 8- and/z-opiate receptors, which may not have much specificity for the carboxy-terminal portion of/3-endorphin, as well as its strong nonspecific binding to brain membranes and greater resistance to the degradative actions of proteolytic enzymes are probably consequences of its enhanced stability at amphiphilic interfaces relative to fl-endorphin. Just as these differences are a direct result of the model peptide design, as discussed previously, so too are the lower potencies of the model peptide in the most specific pharmacological assays employed. The opiate receptors of the rat vas deferens and the opiate receptors mediating the analgesic action of/3-endorphin in the central nervous system clearly have different additional specificities for individual side chains o n the surface of the amphiphilic helical structure which the simplified structure of the model peptide does not provide. The present results, therefore, allow a rational approach to identifying these residues to be adopted so that opioid peptides with a high specificity for different opiate receptors and potent long-lasting in vivo activities may be developed. The common occurrence of potential amphiphilic o~ helices in peptide hormones 2 indicate that this approach should be widely applicable.
[26] E f f e c t o f P o i n t M u t a t i o n s on t h e F o l d i n g of Globular Proteins By C. ROBERT MATTHEWS
General Introduction Kinetic studies of protein folding have proven to be a richer source of information on the mechanism of folding than their equilibrium counterparts. The detection of multiple unfolded forms I and transient intermedii B. T. Nail,
Comments Mol. Cell. Biophys. 3, 123 (1985).
METHODS IN ENZYMOLOGY, VOL. 154
Copyright © 1987by AcademicPress, Inc. All rightsof reproductionin any form reserved.
[26]
POINT MUTATIONS AND PROTEIN FOLDING
499
has improved our understanding of the types of structures that play significant roles in this complex process. An important goal of current experiments is to elucidate the conformations of folding intermediates or, equivalently, the details of the conformational changes that link native, intermediate, and unfolded forms. Available spectroscopic methods that are sufficiently sensitive to detect the intermediates, e.g., difference ultraviolet or fluorescence spectroscopy, cannot, in general, provide detailed structural information. Methods that could provide such information, including X-ray crystallography or Fourier transform NMR spectroscopy, do not have the sensitivity to elucidate the conformations of transient species whose lifetimes may be in the millisecond time range. As discussed in a recent contribution to this series, 3 hydrogen exchange methods have the potential to follow the formation of secondary structure in identifiable segments of protein. However, the application of these methods to large proteins is sufficiently complex that the development of alternative methods would be desirable. An approach that we have been developing in our laboratory is to study the effect of single amino acid replacements on the stability and folding of globular proteins. By identifying amino acids that play key roles in folding and stability, we hope to elucidate the structural basis for ratelimiting steps in folding. The ability to specifically replace a given amino acid using recombinant DNA technology makes such an approach feasible. ates 2
Principle The rationale for this approach is based on the now well-accepted hypothesis of Anfinsen that the amino acid sequence of a protein determines its tertiary structure. 4 This hypothesis can be extended to propose that the primary sequence also determines the folding pathway and that replacements of amino acids which play key roles in this process will have observable effects on the folding rate and perhaps the stability of the native conformation. Quantitative comparisons of the changes in the free energy of folding and the rate constants for unfolding and refolding can reveal whether a given replacement alters the stability or a rate-limiting step in folding or both. This information can be used to map the residues involved in the rate-limiting steps and to eliminate potential structural 2 p. S. Kim and R. L. Baldwin, Annu. Rev. Biochem. 51, 459 (1982). 3 p. S. Kim, this series, Vol. 131, p. 136. 4 C. B. Antinsen, Science 181, 223 (1973).
500
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
models for conformational changes. Consideration of these results in terms of an X-ray crystal structure should eventually lead to a better understanding of the conformational changes involved in folding. Materials and Reagents Ultra-pure or the equivalent grade urea and guanidine hydrochloride should be used in unfolding studies and are commercially available. In some systems, the choice of denaturant will be dictated by the limited solubility of folding intermediates which are easily salted out by ionic denaturants. Reversible unfolding for proteins with the free sulhydryl groups usually necessitates the addition of a reducing agent in the buffer. For difference ultraviolet studies, 2-mercaptoethanol is a better choice than dithioerythritol because the oxidized form of 2-mercaptoethanol does not absorb strongly in the range 280-300 nm where the protein difference spectrum arises. Oxidation of the free sulfhydryls can also be minimized by degassing all buffers and by maintaining a low concentration, e.g., 0.1 mM, of ethylenediaminetetraacetic acid in all buffers. Purified protein can usually be stored at 4° in a closed vial as an ammonium sulfate precipitate for a period of several months before a measurable loss of enzymatic activity occurs. The precipitate is suspended in a minimum amount of appropriate buffer and is dialyzed extensively to remove residual ammonium sulfate. The protein is then concentrated prior to folding studies by one of several commercially available products. Method Although the emphasis in this chapter is on the effects of amino acid replacements on the kinetics of folding, it is important to measure the effects on the equilibrium unfolding transition as well. The latter information is required for identification of an appropriate folding model, for designing kinetic experiments and for the correct interpretation of the results of kinetic studies.
Equilibrium Studies The equilibrium unfolding transition for the wild-type protein can be monitored by a variety of spectroscopic techniques. Those with sufficient sensitivity to require protein concentrations of approximately 1 mg ml -~ or less are advantageous because intermolecular effects such as aggregation are minimized and because protein usage is low. Difference ultraviolet, fluorescence, and circular dichroism spectroscopies all satisfy this
[26]
POINT MUTATIONS AND PROTEIN FOLDING
501
requirement; however, the first two methods are preferable because stopped-flow kinetic studies often necessary for this analysis can be performed on commercially available instruments. Folding studies in our laboratory have relied principally on difference ultraviolet spectroscopy because these dual beam instruments have the stability to monitor small changes in absorbance over a period of hours. The principle contributions to the difference spectrum in the near ultraviolet region, 275-300 nm, come from changes in the exposure to solvent of buried tyrosine and tryptophan residues that occur on unfolding. 5 Therefore, the technique is applicable to a wide variety of proteins. The unfolding can be induced by chemical denaturants, increases or decreases in pH, or by increase in the temperature. Because reliable analysis of the equlibrium data requires that the unfolding reaction be reversible, chemical denaturants, e.g., urea or guanidine hydrochloride, or decreases in pH are preferred over increases in pH or temperature. At alkaline pH or elevated temperatures, protein folding is often irreversible due, in part, to chemical damage, e.g., amide hydrolysis. For proteins with isoelectric points below pH 7, acid-induced unfolding can lead to potential problems with aggregation that can occur as the isoelectric point is reached during the titration. Therefore, the most generally useful method involves chemical denaturants. The procedures that are used to obtain difference ultraviolet spectra for denaturant-induced protein unfolding have been described, 5,6 as has the procedure for determining the free energy of unfolding in the absence of denaturant, i.e., the stability .7 The only additional comment to be made concerns the form of the dependence of the apparent free energy of unfolding, AGapp, on the denaturant concentration. Of the three models presented by Pace, 8 we prefer the linear dependence of AGapp on denaturant concentration because a thermodynamic justification has been provided. 9 In this model, AGapp = AG~20 +
A[denaturant]
where AG~uz° is the free energy of unfolding in water and A is an empirical parameter that describes the dependence of AGapp on denaturant concentration. Estimates of the stability in water are obtained by linear extrapolation of AGapp to zero molar denaturant. Once a satisfactory fit of the equilibrium unfolding data has been ob5 T. T. Herskovits, this series, Vol. 11, p. 748. 6 j. W. Donovan, this series, Vol. 27, p. 497. 7 C. N. Pace, this series, Vol. 131, p. 266. 8 C. N. Pace, CRC Crit. Rev. Biochem. 3, 1 (1975). 9 j. A. Schellman, Biopolymers 17, 1305 (1978).
502
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
tained, the effect of amino acid replacements on the stability can be determined. In principle, one would prefer to compare the stabilities of wild-type and mutant proteins in the absence of denaturants. In practice, the necessity of extrapolating data which can only be measured accurately in the transition zone over several molar units of denaturant may lead to errors which are comparable to the differences in stability between wild-type and mutant proteins. Also, Pace has tested the validity of a linear extrapolation and has found that deviations of up to 30% can occur near zero molar denaturant.I° An alternative approach is to compare the free energies of unfolding at the concentration of denaturant corresponding to the midpoint of the unfolding transition for the wild-type protein. 1~ This has the advantage that the comparisons are made where the free energies can be most accurately measured; however, it is strictly empirical. Kinetic Studies
To obtain useful results from studies of the kinetics of folding of mutant proteins, one must establish a mechanism of folding for the wild-type protein. Mechanisms for a number of proteins have been proposed and the procedures by which these mechanisms were defined have been described. 12 Of importance to the present discussion are steps which actually involve folding, i.e., the interconversion of native, intermediate, and unfolded forms. Reactions that involve interconversions of multiple unfolded forms have been discussed in this series 1 and are not of interest here. The data essential for this analysis are the relaxation times for unfolding and refolding and the dependence of these relaxation times on the final denaturant concentration. For unfolding studies, the protein is initially maintained in a buffered solution at neutral pH where the protein is known to be in the native conformation. Unfolding is than initiated by volumetrically diluting this sample into a buffered solution containing a high denaturant concentration. The procedure for refolding is similar except that the protein is first equilibrated in a high denaturant concentration where the unfolded form is favored. The equilibrium data define the appropriate conditions. Dilution to varying final denaturant concentrations and constant protein concentration is achieved by volumetrically adding this sample to solutions containing the required amounts of dena10 C. N. Pace and K. E. Vanderburg, Biochemistry 18, 288 (1979). 11 j. F. Cupo and C. N. Pace, Biochemistry 22, 2654 (1983). n H. Utiyama and R. L. Baldwin, this series, Vol. 131, p. 51.
[26]
POINT MUTATIONS AND PROTEIN FOLDING
503
turant. The choice of the initial denaturant concentration for refolding studies does not affect the observed relaxation times because they are determined by the final solution conditions. 13 The accuracy of the measurements is usually improved by starting with the fully unfolded protein so as to maximize the amplitudes of the folding phases. For reactions with relaxation times longer than 10 sec, manual mixing is satisfactory; faster reactions require stopped-flow instrumentation. The signal is recorded, and the relaxation times ~'i (or equivalently, the apparent rate constants, ki, where k / = 1/zi) are extracted from the data. The transient response of a first-order system that undergoes a displacement in the equilibrium distribution can be described by a sum of exponentials as follows:
A(t) = ~ Ai exp(-t/~i) + A(~) i
where A(t) is the absorbance at time t, Ai the amplitude of phase i with relaxation time ~'i, and A(~) the absorbance at infinite time. In simple cases where only two relaxation times that differ by at least a factor of 3 and preferably a factor of 5 or more are involved, exponential stripping t4 can be used with reasonable confidence. In more complex cases, a nonlinear least squares computer-fit of the data is advisable. ~5Appropriate software is normally available at major computational facilities and can be purchased commercially. For phases that correspond to actual folding reactions, the relaxation times have been observed to have a characteristic dependence on the denaturant concentration (Fig. 1). The logarithm of the relaxation time first increases linearly with denaturant concentration, reaches a maximum, and then decreases with further increases in denaturant concentration. This dependence can be understood in terms of the expected behavior for a simple two-state reaction, e.g., N.
kv kg
"U
where N and U are the native and unfolded forms, respectively, and ku and kR are the rate constants for unfolding and refolding, respectively. For this system, the observed relaxation time is related to the rate constants by ~.-i = ku + kR ~3 C. Tanford, Adv. Protein Chem. 23, 121 (1968). ~4p. j. Hagerman and R. L. Baldwin, Biochemistry 15, 1462 (1976). ~5M. C. Johnson and S. G. Frasier, this series, Vol. 117, p. 301.
504
MUTAGENESIS AND PROTEIN ENGINEERING
[26l
ILl n
I-Z
o
l-X ,-I ILl n-
I
-
O
o
.J
[DENATURANT ]
FIG. 1. Expected dependence of the logarithm of the relaxation time on the final denaturant concentration for a simple two-state folding reaction.
The equilibrium constant Ibr unfolding, Ku, is defined in the usual way: Ku = [U]/[N] = ku/kR. For unfolding jumps where the unfolded form is favored, Ku > > 1 or equivalently ku > > kR. Under these conditions, ~.-1 ~ ku, and the observed relaxation time is a measure of the unfolding rate constant. The progressive decrease in ~"at high denaturant concentration reflects the shift of the equilibrium to favor the unfolded form. For refoldingjumps where the native form is favored, Ku < < 1 and ka > > ku. Under these conditions, ~.-1 _ ka, and the observed relaxation time is a measure of the refolding rate constant. The progressive decrease at low denaturant concentration reflects the shift of the equilibrium to favor the native form. At intermediate denaturant concentrations, the relaxation time is a composite of the two rate constants and proceeds through a maximum which corresponds closely to the midpoint of the appropriate equilibrium transition. The relaxation times for the wild-type protein serve as a basis for comparison with the mutant proteins. The advantage of measuring the effects of amino acid replacements at a series of denaturant concentrations is that it improves the confidence in measurements of small differences in relaxation times. It also makes it possible to observe changes in the slopes of the plots of log ~- versus denaturant concentration. Tanford has related this slope to the position of the transition state along the reaction coordinate, but no structural interpretations have been offered. 16 Studies on mutant proteins may shed light on this matter. ~6C. Tanford, Adv. Protein Chem. 24, 1 (1970).
[26]
POINT MUTATIONS AND PROTEIN FOLDING
505
Analysis
Comparison of the equilibrium and kinetic properties of folding for the wild-type and mutant proteins permits one to determine if the amino acid in question plays a key role in the rate-limiting step, alters the stability, or both. The scheme that we have developed for this analysis employs reaction coordinate diagrams and transition state theory. Such a diagram for a simple, two state N ~ U folding reaction is shown in Fig. 2. The stability of the protein, in the absence of denaturant, is indicated by the difference in free energy between the native and unfolded forms. The relaxation time for unfolding or, equivalently, the rate constant for unfolding is related to the free energy differences between the native conformation and the transition state, and the corresponding quantities for refolding are related to the free energy difference between the unfolded conformation and the transition state. The quantitative relationships are AGu = - R T In Ku AG~ = - R T ln[(hku/kBT)] AG~R = - R T ln[(hkR/kBT)] where AGE is the free energy of unfolding, AG~ and AG~R the activation free energies for unfolding and refolding, respectively, R the gas constant, T the absolute temperature, and h and kB the Planck and Boltzmann
HzO :3U
,
N
Refolding
Unfolding
=-
R E A C T I O N COORDINATE
FIG. 2. Reaction coordinate diagram for a simple two-state folding reaction, Axes not drawn to scale.
506
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
constants, respectively. Therefore, measurements of the equilibrium and kinetic properties of folding permit one to measure the differences in free energy between the native, transition state, and unfolded forms. Because Ko, ktj, and kR and therefore AGu, AG~, and AG*Rall depend on the denaturant concentration, the question arises as to the appropriate c o w entration for comparing the wild-type and mutant proteins. The procedure described above for obtaining the free energy of folding in the absence of denaturant by extrapolating the linear dependence of free energy to zero molar denaturant can also be applied to the activation free energies to obtain the rate constants in the absence of denaturant. This assumption is based on the presumption from transition state theory that the transition state is in equilibrium with each of the stable conformations. Then, using Schellman's treatment, 9 the activation free energies should be a linear function of the denaturant concentration: AG*u = AG*uH20 + Au[denaturant] AG*a = AG*Rtt20 + AR[denaturant] The observation that the log r (or log k) is a linear function of the denaturant concentration for a number of proteins z3,17,~8supports this assumption. Therefore, the equilibrium and kinetic data for folding can be used to construct a reaction coordinate diagram for each particular protein in the absence of denaturant. In practice, the extrapolation of AG* to zero molar denaturant has the same uncertainties as described above for the free energy difference. It is our opinion that comparisons of relaxation times for wild-type and mutant proteins are best made at urea concentrations where the measurements are made. Figures 3A, 4A, and 5A show several possible effects of amino acid replacements on reaction coordinate diagrams for folding reactions. In Fig. 3A, the net effect of the replacement is to decrease the free energy of the native conformation with respect to the energies of the transition state and unfolded forms. The free energy of unfolding, i.e., the stability is increased (Fig. 3B), the relaxation time for unfolding is increased, and the relaxation time for refolding is unchanged (Fig. 3C). This behavior, where the free energy of one of the stable states is selectively altered, can be classified as an effect on the equilibrium properties and such mutant proteins termed equilibrium mutants. In this case, the amino acid involved plays a role in stabilizing the protein but is not involved in the rate-limiting step. 77 R. R. Kelley, J. Wilson, C. Bryant, and E. Stcllwagcn, Biochemistry 25, 728 (1986). ~s M. M. Crisanti and C. R. Matthews, Biochemistry 20, 2700 (1981).
[26]
POINT MUTATIONS AND PROTEIN FOLDING
507
Energy N
U
~Refolding Unfolding---~
-or
Fapp
0.0 --
J/~'/
I: Denaturant ]
~n T
"
[ Denaturant ]
FIG. 3. Equilibrium and kinetic properties of the wild-type (--) and equilibrium mutant ( - - - ) proteins. (A) The reaction coordinate diagrams. (B) The equilibrium unfolding transition. F~p represents the fraction of unfolded protein. (C) The relaxation time as a function of the final denaturant concentration.
Energy N
i.? /ll
U
~l--Refokling Unfol~ng --~
'L/
[ Denaturant]
Znr [ Denaturant ]
FIG. 4. Equilibrium and kinetic properties of the wild-type (--) and kinetic mutant (----) proteins. (A) The reaction coordinate diagrams. (B) The equilibrium unfolding transition. F~oprepresents the fraction of unfolded protein. (C) The relaxation time as a function of the final denaturant concentration.
508
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
Energy N
U
,=--Refolding Unfolding--~,
"°r :,,-7' o.0"
"/
"/
[Denaturant]
[ Denaturant ]
FIG. 5. Equilibrium and kinetic properties of the wild-type (--) and mixed equilibriumkinetic mutant ( - - - ) proteins. (A) The reaction coordinate diagrams. (B) The equilibrium unfolding transition. Fapprepresents the fraction of unfolded protein. (C) The relaxation time as a function of the final denaturant concentration.
Another possible outcome for an amino acid replacement is shown in Fig. 4A. In this case, the free energy of the transition state is selectively altered. The stability of the protein is not changed (Fig. 4B); however, the relaxation times for both unfolding and refolding are increased. Such mutants can be classified as kinetic mutants and serve to identify amino acids that play key roles in the rate-limiting step in folding. The third general class of mutants are those that affect both the equilibrium and kinetic properties, as shown in Fig. 5A-C. The free energies of the native conformation, transition state, and unfolded conformation are all altered with respect to each other. In the case shown, the stability is increased, the relaxation time for unfolding is increased, and the relaxation time for refolding is decreased. This behavior can be classified as a mixed equilibrium-kinetic mutant and, like the kinetic mutant described above, implicates the amino acid in the rate-limiting step in folding. The reaction coordinate diagrams shown in Fig. 3-5 have been arbitrarily aligned by equating the free energies of the wild-type and mutant proteins in the unfolded state. The analysis of the equilibrium and kinetic data provides information on the differences in free energy between the various states. However, they do not provide information on the absolute free energy of any state. The critical factor in determining the role of a particular amino acid in folding is the differential effect of the replacement
[26]
POINT MUTATIONS AND PROTEIN FOLDING
509
on the relative free energies of the two stable states and the transition state. Therefore, the alignment of the reaction coordinate diagrams is arbitrary.
Application As an example of how amino acid replacements can be used to obtain structural information on folding reactions, a problem from the folding of the t~ subunit of tryptophan synthase from Escherichia coli will be described. The a subunit is a 29,000-dalton protein that has a larger, more stable amino domain, residues 1-188, and a small, less stable carboxyl domain, residues 189-268.19 Studies on the mechanism of folding suggested that the final step in folding involves the conversion of a stable intermediate, which has a folded amino domain and an unfolded carboxyl domain, to the fully folded native conformation. 2° Left unanswered was the nature of the rate-limiting step in this process: Is it limited by the folding of the carboxyl domain or by the association of the two folded domains? This problem was resolved by examining the effect of single amino acid replacements in both the amino and carboxyl domains on the unfolding and refolding relaxation times for the intermediate-to-native step. If the folding of the carboxyl domain is rate-limiting, only mutations in the carboxyl domain can have a real effect on the kinetics of this reaction. If domain association is rate limiting then mutations in either domain can act as kinetic or mixed equilibrium-kinetic mutants. The effects of the Phe 22 ~ Leu and Gly 211 ~ Glu replacements on the relaxation times for unfolding and refolding are shown in Fig. 6. 21 The replacement in the amino domain, Phe 22 ~ Leu, is very close to a pure kinetic mutant; the unfolding relaxation time increases by 4-fold at 6 M urea while that for refolding increased by 2.5-fold at 2 M urea. The equilibrium data confirm the near absence of an effect on stability. The replacement in the carboxyl domain, Gly 211 ~ Glu, is clearly a mixed equilibrium-kinetic mutant; the relaxation time for unfolding increases 8fold at 6 M urea and that for refolding increases 4-fold at 2 M urea. Consistent with these results, the equilibrium data show that this replacement causes a small increase in stability. Because replacements in both ~9 E. W. Miles, K. Yutani, and K. Ogaschara, Biochemistry 21, 2586 (1982). 2o C. R. Matthews, M. M. Crisanti, J. T. Manz, and G. L. Gepner, Biochemistry 22, 1445 (1983). 2~ A. M. Beasty, M. R. Hurle, J. T. Manz, T. Stackhouse, J. J. Onuffer, and C. R. Matthews, Biochemistry 25, 2965 (1986).
510
MUTAGENESIS AND PROTEIN ENGINEERING 5000
I
I
I
I
I
[26]
I
?i"e
t",, ,°°° 500
I...~..~ •. ~.. , . . i f
o /
lit
.~ I *
t/i,,¢7.1
,oo.
oo~
,e,, ~,,--o,o
"~..._~ -,,,,~_,... \ \
X
'.
%
",:,.
\
\ "., \ w,,,,,,X "..,, ', \-.
'.,
-
°°°%1 IC
I I
I 2
I I 3 4 [UREA] (M)
5
I
6
7
FIG. 6. Relaxation time at the indicated final urea concentration for the single phase in unfolding and two slow phases in refolding for the Phe 22 --~ Leu (r-q,It, O) and the Gly 211 --* Gin (A, &, O) mutant a subunits from tryptophan synthase. 21 Open symbols represent unfolding data and filled symbols represent refolding data. For comparison, the relaxation times for the wild-type a subunit under the same conditions (25°, pH 7.8) are shown as the solid lines. the amino and c a r b o x y l domains affect the rate-limiting step, one can conclude that the reaction m u s t c o r r e s p o n d to domain association or, p e r h a p s , to s o m e o t h e r type of molecule-wide reaction. T h e u r e a - i n d e p e n d e n t relaxation times o b s e r v e d at urea concentrations b e l o w 1 M in Fig. 6 are thought to reflect Pro isomerization reactions which b e c o m e rate limiting at low urea concentrations. A detailed account of their i n v o l v e m e n t in the folding m e c h a n i s m has b e e n presented elsewhere. ~s Comments T h e m e t h o d p r e s e n t e d for analyzing the effects of amino acid replacem e n t s on protein folding and stability is applicable to folding reactions in
[27]
S T R U C T U R E A N D T H E R M A L S T A B I L I T Y OF T 4 LYSOZYME
511
which both the unfolding and refolding rate constants or, equivalently, one of the rate constants and the associated equilibrium constant can be determined. Therefore, certain kinetic schemes may preclude its use. Als0, it may be useful only for folding reactions that occur late in the folding pathway, near the native conformation, where the protein is highly constrained and is effectively moving from one particular conformation to another. Early folding reactions, nearer to the unfolded conformation, may proceed along a variety of pathways at nearly equal rates and preclude use of the method. Obviously, more experiments are required to test the extent of its applicability. The results described above, however, demonstrate its potential in elucidating the structural basis for reactions that occur during protein folding. It is our expectation that mutagenesis will dramatically improve our understanding of the mechanisms of protein folding and improve the possibility of predicting the three-dimensional structure of a protein from its amino acid sequence.
[27] S t r u c t u r e a n d T h e r m a l Stability of P h a g e T 4 L y s o z y m e
By TOM ALBER and BRIAN W. MATTHEWS Introduction
Understanding the physical basis of the thermal stability of proteins is a major problem in molecular biology and a prerequisite for rational protein design. While the three-dimensional structure of a protein can be determined relatively accurately, the strategies for designing an amino acid sequence to stabilize that structure remain mysterious. Reversible protein denaturation has proven to be an extremely complex reaction.l-3 Dramatic changes in the solvation and flexibility of the polypeptide chain are accompanied by compensating changes in the enthalpy and entropy of the system. These thermodynamic state functions vary steeply with temperature. At high temperature, denaturation results in a large increase in entropy, presumably due to the added flexibility of the protein, and in a compensating increase in enthalpy, attributed to changes in interactions in the protein and solvent. Surprisingly, proteins can also denature at low temperature, and the system actually loses eni p. L. Privalov, Ado. Protein Chem. 33, 198 (1979). 2 j. A. Schellman, M. Lindorfer, R. Hawkes, and M. GrOtter, Biopolymers 20, 1989 (1981). 3 p. L. Privalov, Y. V. Griko, S. Venyaminov, and V. P. Kutyshenko, J. Mol. Biol. 190, 487 (1986).
METHODS IN ENZYMOLOGY, VOL. 154
Copyright © 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
could also reproduce the properties of the natural sequence. 30In this way, the role of the amphiphilic structure has been more thoroughly defined, and evidence suggesting that this structure is the regular a helix and not a 7r helix or other less regular structure has been obtained. Differences observed in the physicochemical properties of the model peptide described here compared to those of/3-endorphin provide a rationale for some of the observed differences in its behavior in the biological assays. Thus, tighter binding to 8- and/z-opiate receptors, which may not have much specificity for the carboxy-terminal portion of/3-endorphin, as well as its strong nonspecific binding to brain membranes and greater resistance to the degradative actions of proteolytic enzymes are probably consequences of its enhanced stability at amphiphilic interfaces relative to fl-endorphin. Just as these differences are a direct result of the model peptide design, as discussed previously, so too are the lower potencies of the model peptide in the most specific pharmacological assays employed. The opiate receptors of the rat vas deferens and the opiate receptors mediating the analgesic action of/3-endorphin in the central nervous system clearly have different additional specificities for individual side chains o n the surface of the amphiphilic helical structure which the simplified structure of the model peptide does not provide. The present results, therefore, allow a rational approach to identifying these residues to be adopted so that opioid peptides with a high specificity for different opiate receptors and potent long-lasting in vivo activities may be developed. The common occurrence of potential amphiphilic o~ helices in peptide hormones 2 indicate that this approach should be widely applicable.
[26] E f f e c t o f P o i n t M u t a t i o n s on t h e F o l d i n g of Globular Proteins By C. ROBERT MATTHEWS
General Introduction Kinetic studies of protein folding have proven to be a richer source of information on the mechanism of folding than their equilibrium counterparts. The detection of multiple unfolded forms I and transient intermedii B. T. Nail,
Comments Mol. Cell. Biophys. 3, 123 (1985).
METHODS IN ENZYMOLOGY, VOL. 154
Copyright © 1987by AcademicPress, Inc. All rightsof reproductionin any form reserved.
[26]
POINT MUTATIONS AND PROTEIN FOLDING
499
has improved our understanding of the types of structures that play significant roles in this complex process. An important goal of current experiments is to elucidate the conformations of folding intermediates or, equivalently, the details of the conformational changes that link native, intermediate, and unfolded forms. Available spectroscopic methods that are sufficiently sensitive to detect the intermediates, e.g., difference ultraviolet or fluorescence spectroscopy, cannot, in general, provide detailed structural information. Methods that could provide such information, including X-ray crystallography or Fourier transform NMR spectroscopy, do not have the sensitivity to elucidate the conformations of transient species whose lifetimes may be in the millisecond time range. As discussed in a recent contribution to this series, 3 hydrogen exchange methods have the potential to follow the formation of secondary structure in identifiable segments of protein. However, the application of these methods to large proteins is sufficiently complex that the development of alternative methods would be desirable. An approach that we have been developing in our laboratory is to study the effect of single amino acid replacements on the stability and folding of globular proteins. By identifying amino acids that play key roles in folding and stability, we hope to elucidate the structural basis for ratelimiting steps in folding. The ability to specifically replace a given amino acid using recombinant DNA technology makes such an approach feasible. ates 2
Principle The rationale for this approach is based on the now well-accepted hypothesis of Anfinsen that the amino acid sequence of a protein determines its tertiary structure. 4 This hypothesis can be extended to propose that the primary sequence also determines the folding pathway and that replacements of amino acids which play key roles in this process will have observable effects on the folding rate and perhaps the stability of the native conformation. Quantitative comparisons of the changes in the free energy of folding and the rate constants for unfolding and refolding can reveal whether a given replacement alters the stability or a rate-limiting step in folding or both. This information can be used to map the residues involved in the rate-limiting steps and to eliminate potential structural 2 p. S. Kim and R. L. Baldwin, Annu. Rev. Biochem. 51, 459 (1982). 3 p. S. Kim, this series, Vol. 131, p. 136. 4 C. B. Antinsen, Science 181, 223 (1973).
500
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
models for conformational changes. Consideration of these results in terms of an X-ray crystal structure should eventually lead to a better understanding of the conformational changes involved in folding. Materials and Reagents Ultra-pure or the equivalent grade urea and guanidine hydrochloride should be used in unfolding studies and are commercially available. In some systems, the choice of denaturant will be dictated by the limited solubility of folding intermediates which are easily salted out by ionic denaturants. Reversible unfolding for proteins with the free sulhydryl groups usually necessitates the addition of a reducing agent in the buffer. For difference ultraviolet studies, 2-mercaptoethanol is a better choice than dithioerythritol because the oxidized form of 2-mercaptoethanol does not absorb strongly in the range 280-300 nm where the protein difference spectrum arises. Oxidation of the free sulfhydryls can also be minimized by degassing all buffers and by maintaining a low concentration, e.g., 0.1 mM, of ethylenediaminetetraacetic acid in all buffers. Purified protein can usually be stored at 4° in a closed vial as an ammonium sulfate precipitate for a period of several months before a measurable loss of enzymatic activity occurs. The precipitate is suspended in a minimum amount of appropriate buffer and is dialyzed extensively to remove residual ammonium sulfate. The protein is then concentrated prior to folding studies by one of several commercially available products. Method Although the emphasis in this chapter is on the effects of amino acid replacements on the kinetics of folding, it is important to measure the effects on the equilibrium unfolding transition as well. The latter information is required for identification of an appropriate folding model, for designing kinetic experiments and for the correct interpretation of the results of kinetic studies.
Equilibrium Studies The equilibrium unfolding transition for the wild-type protein can be monitored by a variety of spectroscopic techniques. Those with sufficient sensitivity to require protein concentrations of approximately 1 mg ml -~ or less are advantageous because intermolecular effects such as aggregation are minimized and because protein usage is low. Difference ultraviolet, fluorescence, and circular dichroism spectroscopies all satisfy this
[26]
POINT MUTATIONS AND PROTEIN FOLDING
501
requirement; however, the first two methods are preferable because stopped-flow kinetic studies often necessary for this analysis can be performed on commercially available instruments. Folding studies in our laboratory have relied principally on difference ultraviolet spectroscopy because these dual beam instruments have the stability to monitor small changes in absorbance over a period of hours. The principle contributions to the difference spectrum in the near ultraviolet region, 275-300 nm, come from changes in the exposure to solvent of buried tyrosine and tryptophan residues that occur on unfolding. 5 Therefore, the technique is applicable to a wide variety of proteins. The unfolding can be induced by chemical denaturants, increases or decreases in pH, or by increase in the temperature. Because reliable analysis of the equlibrium data requires that the unfolding reaction be reversible, chemical denaturants, e.g., urea or guanidine hydrochloride, or decreases in pH are preferred over increases in pH or temperature. At alkaline pH or elevated temperatures, protein folding is often irreversible due, in part, to chemical damage, e.g., amide hydrolysis. For proteins with isoelectric points below pH 7, acid-induced unfolding can lead to potential problems with aggregation that can occur as the isoelectric point is reached during the titration. Therefore, the most generally useful method involves chemical denaturants. The procedures that are used to obtain difference ultraviolet spectra for denaturant-induced protein unfolding have been described, 5,6 as has the procedure for determining the free energy of unfolding in the absence of denaturant, i.e., the stability .7 The only additional comment to be made concerns the form of the dependence of the apparent free energy of unfolding, AGapp, on the denaturant concentration. Of the three models presented by Pace, 8 we prefer the linear dependence of AGapp on denaturant concentration because a thermodynamic justification has been provided. 9 In this model, AGapp = AG~20 +
A[denaturant]
where AG~uz° is the free energy of unfolding in water and A is an empirical parameter that describes the dependence of AGapp on denaturant concentration. Estimates of the stability in water are obtained by linear extrapolation of AGapp to zero molar denaturant. Once a satisfactory fit of the equilibrium unfolding data has been ob5 T. T. Herskovits, this series, Vol. 11, p. 748. 6 j. W. Donovan, this series, Vol. 27, p. 497. 7 C. N. Pace, this series, Vol. 131, p. 266. 8 C. N. Pace, CRC Crit. Rev. Biochem. 3, 1 (1975). 9 j. A. Schellman, Biopolymers 17, 1305 (1978).
502
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
tained, the effect of amino acid replacements on the stability can be determined. In principle, one would prefer to compare the stabilities of wild-type and mutant proteins in the absence of denaturants. In practice, the necessity of extrapolating data which can only be measured accurately in the transition zone over several molar units of denaturant may lead to errors which are comparable to the differences in stability between wild-type and mutant proteins. Also, Pace has tested the validity of a linear extrapolation and has found that deviations of up to 30% can occur near zero molar denaturant.I° An alternative approach is to compare the free energies of unfolding at the concentration of denaturant corresponding to the midpoint of the unfolding transition for the wild-type protein. 1~ This has the advantage that the comparisons are made where the free energies can be most accurately measured; however, it is strictly empirical. Kinetic Studies
To obtain useful results from studies of the kinetics of folding of mutant proteins, one must establish a mechanism of folding for the wild-type protein. Mechanisms for a number of proteins have been proposed and the procedures by which these mechanisms were defined have been described. 12 Of importance to the present discussion are steps which actually involve folding, i.e., the interconversion of native, intermediate, and unfolded forms. Reactions that involve interconversions of multiple unfolded forms have been discussed in this series 1 and are not of interest here. The data essential for this analysis are the relaxation times for unfolding and refolding and the dependence of these relaxation times on the final denaturant concentration. For unfolding studies, the protein is initially maintained in a buffered solution at neutral pH where the protein is known to be in the native conformation. Unfolding is than initiated by volumetrically diluting this sample into a buffered solution containing a high denaturant concentration. The procedure for refolding is similar except that the protein is first equilibrated in a high denaturant concentration where the unfolded form is favored. The equilibrium data define the appropriate conditions. Dilution to varying final denaturant concentrations and constant protein concentration is achieved by volumetrically adding this sample to solutions containing the required amounts of dena10 C. N. Pace and K. E. Vanderburg, Biochemistry 18, 288 (1979). 11 j. F. Cupo and C. N. Pace, Biochemistry 22, 2654 (1983). n H. Utiyama and R. L. Baldwin, this series, Vol. 131, p. 51.
[26]
POINT MUTATIONS AND PROTEIN FOLDING
503
turant. The choice of the initial denaturant concentration for refolding studies does not affect the observed relaxation times because they are determined by the final solution conditions. 13 The accuracy of the measurements is usually improved by starting with the fully unfolded protein so as to maximize the amplitudes of the folding phases. For reactions with relaxation times longer than 10 sec, manual mixing is satisfactory; faster reactions require stopped-flow instrumentation. The signal is recorded, and the relaxation times ~'i (or equivalently, the apparent rate constants, ki, where k / = 1/zi) are extracted from the data. The transient response of a first-order system that undergoes a displacement in the equilibrium distribution can be described by a sum of exponentials as follows:
A(t) = ~ Ai exp(-t/~i) + A(~) i
where A(t) is the absorbance at time t, Ai the amplitude of phase i with relaxation time ~'i, and A(~) the absorbance at infinite time. In simple cases where only two relaxation times that differ by at least a factor of 3 and preferably a factor of 5 or more are involved, exponential stripping t4 can be used with reasonable confidence. In more complex cases, a nonlinear least squares computer-fit of the data is advisable. ~5Appropriate software is normally available at major computational facilities and can be purchased commercially. For phases that correspond to actual folding reactions, the relaxation times have been observed to have a characteristic dependence on the denaturant concentration (Fig. 1). The logarithm of the relaxation time first increases linearly with denaturant concentration, reaches a maximum, and then decreases with further increases in denaturant concentration. This dependence can be understood in terms of the expected behavior for a simple two-state reaction, e.g., N.
kv kg
"U
where N and U are the native and unfolded forms, respectively, and ku and kR are the rate constants for unfolding and refolding, respectively. For this system, the observed relaxation time is related to the rate constants by ~.-i = ku + kR ~3 C. Tanford, Adv. Protein Chem. 23, 121 (1968). ~4p. j. Hagerman and R. L. Baldwin, Biochemistry 15, 1462 (1976). ~5M. C. Johnson and S. G. Frasier, this series, Vol. 117, p. 301.
504
MUTAGENESIS AND PROTEIN ENGINEERING
[26l
ILl n
I-Z
o
l-X ,-I ILl n-
I
-
O
o
.J
[DENATURANT ]
FIG. 1. Expected dependence of the logarithm of the relaxation time on the final denaturant concentration for a simple two-state folding reaction.
The equilibrium constant Ibr unfolding, Ku, is defined in the usual way: Ku = [U]/[N] = ku/kR. For unfolding jumps where the unfolded form is favored, Ku > > 1 or equivalently ku > > kR. Under these conditions, ~.-1 ~ ku, and the observed relaxation time is a measure of the unfolding rate constant. The progressive decrease in ~"at high denaturant concentration reflects the shift of the equilibrium to favor the unfolded form. For refoldingjumps where the native form is favored, Ku < < 1 and ka > > ku. Under these conditions, ~.-1 _ ka, and the observed relaxation time is a measure of the refolding rate constant. The progressive decrease at low denaturant concentration reflects the shift of the equilibrium to favor the native form. At intermediate denaturant concentrations, the relaxation time is a composite of the two rate constants and proceeds through a maximum which corresponds closely to the midpoint of the appropriate equilibrium transition. The relaxation times for the wild-type protein serve as a basis for comparison with the mutant proteins. The advantage of measuring the effects of amino acid replacements at a series of denaturant concentrations is that it improves the confidence in measurements of small differences in relaxation times. It also makes it possible to observe changes in the slopes of the plots of log ~- versus denaturant concentration. Tanford has related this slope to the position of the transition state along the reaction coordinate, but no structural interpretations have been offered. 16 Studies on mutant proteins may shed light on this matter. ~6C. Tanford, Adv. Protein Chem. 24, 1 (1970).
[26]
POINT MUTATIONS AND PROTEIN FOLDING
505
Analysis
Comparison of the equilibrium and kinetic properties of folding for the wild-type and mutant proteins permits one to determine if the amino acid in question plays a key role in the rate-limiting step, alters the stability, or both. The scheme that we have developed for this analysis employs reaction coordinate diagrams and transition state theory. Such a diagram for a simple, two state N ~ U folding reaction is shown in Fig. 2. The stability of the protein, in the absence of denaturant, is indicated by the difference in free energy between the native and unfolded forms. The relaxation time for unfolding or, equivalently, the rate constant for unfolding is related to the free energy differences between the native conformation and the transition state, and the corresponding quantities for refolding are related to the free energy difference between the unfolded conformation and the transition state. The quantitative relationships are AGu = - R T In Ku AG~ = - R T ln[(hku/kBT)] AG~R = - R T ln[(hkR/kBT)] where AGE is the free energy of unfolding, AG~ and AG~R the activation free energies for unfolding and refolding, respectively, R the gas constant, T the absolute temperature, and h and kB the Planck and Boltzmann
HzO :3U
,
N
Refolding
Unfolding
=-
R E A C T I O N COORDINATE
FIG. 2. Reaction coordinate diagram for a simple two-state folding reaction, Axes not drawn to scale.
506
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
constants, respectively. Therefore, measurements of the equilibrium and kinetic properties of folding permit one to measure the differences in free energy between the native, transition state, and unfolded forms. Because Ko, ktj, and kR and therefore AGu, AG~, and AG*Rall depend on the denaturant concentration, the question arises as to the appropriate c o w entration for comparing the wild-type and mutant proteins. The procedure described above for obtaining the free energy of folding in the absence of denaturant by extrapolating the linear dependence of free energy to zero molar denaturant can also be applied to the activation free energies to obtain the rate constants in the absence of denaturant. This assumption is based on the presumption from transition state theory that the transition state is in equilibrium with each of the stable conformations. Then, using Schellman's treatment, 9 the activation free energies should be a linear function of the denaturant concentration: AG*u = AG*uH20 + Au[denaturant] AG*a = AG*Rtt20 + AR[denaturant] The observation that the log r (or log k) is a linear function of the denaturant concentration for a number of proteins z3,17,~8supports this assumption. Therefore, the equilibrium and kinetic data for folding can be used to construct a reaction coordinate diagram for each particular protein in the absence of denaturant. In practice, the extrapolation of AG* to zero molar denaturant has the same uncertainties as described above for the free energy difference. It is our opinion that comparisons of relaxation times for wild-type and mutant proteins are best made at urea concentrations where the measurements are made. Figures 3A, 4A, and 5A show several possible effects of amino acid replacements on reaction coordinate diagrams for folding reactions. In Fig. 3A, the net effect of the replacement is to decrease the free energy of the native conformation with respect to the energies of the transition state and unfolded forms. The free energy of unfolding, i.e., the stability is increased (Fig. 3B), the relaxation time for unfolding is increased, and the relaxation time for refolding is unchanged (Fig. 3C). This behavior, where the free energy of one of the stable states is selectively altered, can be classified as an effect on the equilibrium properties and such mutant proteins termed equilibrium mutants. In this case, the amino acid involved plays a role in stabilizing the protein but is not involved in the rate-limiting step. 77 R. R. Kelley, J. Wilson, C. Bryant, and E. Stcllwagcn, Biochemistry 25, 728 (1986). ~s M. M. Crisanti and C. R. Matthews, Biochemistry 20, 2700 (1981).
[26]
POINT MUTATIONS AND PROTEIN FOLDING
507
Energy N
U
~Refolding Unfolding---~
-or
Fapp
0.0 --
J/~'/
I: Denaturant ]
~n T
"
[ Denaturant ]
FIG. 3. Equilibrium and kinetic properties of the wild-type (--) and equilibrium mutant ( - - - ) proteins. (A) The reaction coordinate diagrams. (B) The equilibrium unfolding transition. F~p represents the fraction of unfolded protein. (C) The relaxation time as a function of the final denaturant concentration.
Energy N
i.? /ll
U
~l--Refokling Unfol~ng --~
'L/
[ Denaturant]
Znr [ Denaturant ]
FIG. 4. Equilibrium and kinetic properties of the wild-type (--) and kinetic mutant (----) proteins. (A) The reaction coordinate diagrams. (B) The equilibrium unfolding transition. F~oprepresents the fraction of unfolded protein. (C) The relaxation time as a function of the final denaturant concentration.
508
MUTAGENESIS AND PROTEIN ENGINEERING
[26]
Energy N
U
,=--Refolding Unfolding--~,
"°r :,,-7' o.0"
"/
"/
[Denaturant]
[ Denaturant ]
FIG. 5. Equilibrium and kinetic properties of the wild-type (--) and mixed equilibriumkinetic mutant ( - - - ) proteins. (A) The reaction coordinate diagrams. (B) The equilibrium unfolding transition. Fapprepresents the fraction of unfolded protein. (C) The relaxation time as a function of the final denaturant concentration.
Another possible outcome for an amino acid replacement is shown in Fig. 4A. In this case, the free energy of the transition state is selectively altered. The stability of the protein is not changed (Fig. 4B); however, the relaxation times for both unfolding and refolding are increased. Such mutants can be classified as kinetic mutants and serve to identify amino acids that play key roles in the rate-limiting step in folding. The third general class of mutants are those that affect both the equilibrium and kinetic properties, as shown in Fig. 5A-C. The free energies of the native conformation, transition state, and unfolded conformation are all altered with respect to each other. In the case shown, the stability is increased, the relaxation time for unfolding is increased, and the relaxation time for refolding is decreased. This behavior can be classified as a mixed equilibrium-kinetic mutant and, like the kinetic mutant described above, implicates the amino acid in the rate-limiting step in folding. The reaction coordinate diagrams shown in Fig. 3-5 have been arbitrarily aligned by equating the free energies of the wild-type and mutant proteins in the unfolded state. The analysis of the equilibrium and kinetic data provides information on the differences in free energy between the various states. However, they do not provide information on the absolute free energy of any state. The critical factor in determining the role of a particular amino acid in folding is the differential effect of the replacement
[26]
POINT MUTATIONS AND PROTEIN FOLDING
509
on the relative free energies of the two stable states and the transition state. Therefore, the alignment of the reaction coordinate diagrams is arbitrary.
Application As an example of how amino acid replacements can be used to obtain structural information on folding reactions, a problem from the folding of the t~ subunit of tryptophan synthase from Escherichia coli will be described. The a subunit is a 29,000-dalton protein that has a larger, more stable amino domain, residues 1-188, and a small, less stable carboxyl domain, residues 189-268.19 Studies on the mechanism of folding suggested that the final step in folding involves the conversion of a stable intermediate, which has a folded amino domain and an unfolded carboxyl domain, to the fully folded native conformation. 2° Left unanswered was the nature of the rate-limiting step in this process: Is it limited by the folding of the carboxyl domain or by the association of the two folded domains? This problem was resolved by examining the effect of single amino acid replacements in both the amino and carboxyl domains on the unfolding and refolding relaxation times for the intermediate-to-native step. If the folding of the carboxyl domain is rate-limiting, only mutations in the carboxyl domain can have a real effect on the kinetics of this reaction. If domain association is rate limiting then mutations in either domain can act as kinetic or mixed equilibrium-kinetic mutants. The effects of the Phe 22 ~ Leu and Gly 211 ~ Glu replacements on the relaxation times for unfolding and refolding are shown in Fig. 6. 21 The replacement in the amino domain, Phe 22 ~ Leu, is very close to a pure kinetic mutant; the unfolding relaxation time increases by 4-fold at 6 M urea while that for refolding increased by 2.5-fold at 2 M urea. The equilibrium data confirm the near absence of an effect on stability. The replacement in the carboxyl domain, Gly 211 ~ Glu, is clearly a mixed equilibrium-kinetic mutant; the relaxation time for unfolding increases 8fold at 6 M urea and that for refolding increases 4-fold at 2 M urea. Consistent with these results, the equilibrium data show that this replacement causes a small increase in stability. Because replacements in both ~9 E. W. Miles, K. Yutani, and K. Ogaschara, Biochemistry 21, 2586 (1982). 2o C. R. Matthews, M. M. Crisanti, J. T. Manz, and G. L. Gepner, Biochemistry 22, 1445 (1983). 2~ A. M. Beasty, M. R. Hurle, J. T. Manz, T. Stackhouse, J. J. Onuffer, and C. R. Matthews, Biochemistry 25, 2965 (1986).
510
MUTAGENESIS AND PROTEIN ENGINEERING 5000
I
I
I
I
I
[26]
I
?i"e
t",, ,°°° 500
I...~..~ •. ~.. , . . i f
o /
lit
.~ I *
t/i,,¢7.1
,oo.
oo~
,e,, ~,,--o,o
"~..._~ -,,,,~_,... \ \
X
'.
%
",:,.
\
\ "., \ w,,,,,,X "..,, ', \-.
'.,
-
°°°%1 IC
I I
I 2
I I 3 4 [UREA] (M)
5
I
6
7
FIG. 6. Relaxation time at the indicated final urea concentration for the single phase in unfolding and two slow phases in refolding for the Phe 22 --~ Leu (r-q,It, O) and the Gly 211 --* Gin (A, &, O) mutant a subunits from tryptophan synthase. 21 Open symbols represent unfolding data and filled symbols represent refolding data. For comparison, the relaxation times for the wild-type a subunit under the same conditions (25°, pH 7.8) are shown as the solid lines. the amino and c a r b o x y l domains affect the rate-limiting step, one can conclude that the reaction m u s t c o r r e s p o n d to domain association or, p e r h a p s , to s o m e o t h e r type of molecule-wide reaction. T h e u r e a - i n d e p e n d e n t relaxation times o b s e r v e d at urea concentrations b e l o w 1 M in Fig. 6 are thought to reflect Pro isomerization reactions which b e c o m e rate limiting at low urea concentrations. A detailed account of their i n v o l v e m e n t in the folding m e c h a n i s m has b e e n presented elsewhere. ~s Comments T h e m e t h o d p r e s e n t e d for analyzing the effects of amino acid replacem e n t s on protein folding and stability is applicable to folding reactions in
[27]
S T R U C T U R E A N D T H E R M A L S T A B I L I T Y OF T 4 LYSOZYME
511
which both the unfolding and refolding rate constants or, equivalently, one of the rate constants and the associated equilibrium constant can be determined. Therefore, certain kinetic schemes may preclude its use. Als0, it may be useful only for folding reactions that occur late in the folding pathway, near the native conformation, where the protein is highly constrained and is effectively moving from one particular conformation to another. Early folding reactions, nearer to the unfolded conformation, may proceed along a variety of pathways at nearly equal rates and preclude use of the method. Obviously, more experiments are required to test the extent of its applicability. The results described above, however, demonstrate its potential in elucidating the structural basis for reactions that occur during protein folding. It is our expectation that mutagenesis will dramatically improve our understanding of the mechanisms of protein folding and improve the possibility of predicting the three-dimensional structure of a protein from its amino acid sequence.
[27] S t r u c t u r e a n d T h e r m a l Stability of P h a g e T 4 L y s o z y m e
By TOM ALBER and BRIAN W. MATTHEWS Introduction
Understanding the physical basis of the thermal stability of proteins is a major problem in molecular biology and a prerequisite for rational protein design. While the three-dimensional structure of a protein can be determined relatively accurately, the strategies for designing an amino acid sequence to stabilize that structure remain mysterious. Reversible protein denaturation has proven to be an extremely complex reaction.l-3 Dramatic changes in the solvation and flexibility of the polypeptide chain are accompanied by compensating changes in the enthalpy and entropy of the system. These thermodynamic state functions vary steeply with temperature. At high temperature, denaturation results in a large increase in entropy, presumably due to the added flexibility of the protein, and in a compensating increase in enthalpy, attributed to changes in interactions in the protein and solvent. Surprisingly, proteins can also denature at low temperature, and the system actually loses eni p. L. Privalov, Ado. Protein Chem. 33, 198 (1979). 2 j. A. Schellman, M. Lindorfer, R. Hawkes, and M. GrOtter, Biopolymers 20, 1989 (1981). 3 p. L. Privalov, Y. V. Griko, S. Venyaminov, and V. P. Kutyshenko, J. Mol. Biol. 190, 487 (1986).
METHODS IN ENZYMOLOGY, VOL. 154
Copyright © 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.