- Email: [email protected]
Weak Cooperativity in the Core Causes a Switch in Folding Mechanism Between Two Proteins of the cks Family
doi:10.1016/S0022-2836(02)01202-0
J. Mol. Biol. (2003) 325, 189–199
Weak Cooperativity in the Core Causes a Switch in Folding Mechanism Between Two Proteins of the cks Family M. A. Seeliger, S. E. Breward and L. S. Itzhaki* MRC Centre for Protein Engineering University Chemical Laboratory, Lensfield Road Cambridge CB2 1EW, UK
The human protein ckshs1 (cks1) is a 79 residue a/b protein with low thermodynamic and kinetic stability. Its folding mechanism was probed by mutation at sites throughout the structure. Many of the mutations caused changes in the slope of the unfolding arm of the chevron plot. The effects can be rationalised in terms of either transition-state movement or native-state “breathing”, and in either case, the magnitude of the effect enables the sequence of events in the folding reaction to be determined. Those sites that fold early exhibit a small perturbation, whilst those sites that fold late exhibit a large perturbation. The results show that cks1 folds sequential pairs of b-strands first; b1/b2 and b3/b4. Subsequently, these pairs pack against each other and onto the a-helical region to form the core. The folding process of cks1 contrasts with that of the homologue, suc1. The 113 residue suc1 has the same b-sheet core structure but, additionally, two large insertions that confer much greater thermodynamic and kinetic stability. The more extensive network of tertiary interactions in suc1 provides sufficient enthalpic gain to overcome the entropic cost of forming the core and thus tips the balance in favour of non-local interactions: the non-local, central b-strand pair, b2/b4, forms first and the periphery strands pack on later. Moreover, the greater cooperativity of the core of suc1 protects its folding from perturbation and consequently the slope of the unfolding arm of the chevron plot is much less sensitive to mutation. q 2002 Elsevier Science Ltd. All rights reserved
*Corresponding author
Keywords: protein folding; protein stability; transition state; topology; F-values
Introduction The human protein ckshs1 (cks1) is a member of the cks family of proteins that are essential for regulation of the eukaryotic cell cycle.1 There are two human homologues, ckshs1 and ckshs2, having 81% sequence identity, with ckshs1 having evolved an additional function compared with other members of the family.2 Three-dimensional domain swapping3 has been observed for several cks proteins, with the C-terminal b-strand, b4, exchanging with another molecule to form an inter-twined dimer.4,5 The process is mediated by a highly conserved “hinge loop” or “hinge” sequence and two proline residues in the hinge Abbreviations used: cks, cyclin-dependent kinase subunit; ckshs, human cks. E-mail address of the corresponding author: [email protected]
loop are the main determinants of whether the protein adopts the monomer or the dimer form.6,7 We have been using the different members of the cks family to investigate the phenomenon of domain swapping.7 – 9 Here, we describe the folding kinetics of cks1 and compare it with that of the Schizosaccharomyces pombe homologue suc1. cks1 has a much lower free energy of unfolding than suc1 (3.5 kcal mol21 versus 8 kcal mol21 at 25 8C).9 It also has a much lower kinetic stability (the rate of unfolding in water is , 1 s21 at 25 8C compared with , 1 £ 1025 s21 for suc1), and cks1 folds in a two-state manner without the population of intermediates, whilst suc1 folds via an intermediate.8 Both the intermediate and the rate-determining transition state for folding of suc1 were characterised previously using F-value analysis.10 We find that the position along the reaction coordinate of the rate-determining transition state for folding/ unfolding of cks1 is very sensitive to changes in
0022-2836/03/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved
190
Folding Mechanism of Human cks1
rationalised in terms of the structure and cooperativity of the native states of the two proteins and they illustrate nicely the enthalpy –entropy balance in protein folding.
Results and Discussion cks1 structure and mutant design cks1 has 79 residues and an a/b fold consisting of a four-stranded b-sheet capped at one end by two short a-helices (Figure 1).14 The hydrophobic core is composed of side-chains in the central strands of the b-sheet (b2 and b4) and the a-helical region. Truncation mutants were constructed that probed the formation of secondary structure elements (11 mutants for the sheet and three mutants for the helices) or the formation of the hydrophobic core (five mutants) (Table 1). The loops were probed and the b-turn or so-called hinge between strands 3 and 4 that mediates domain swapping. Analysis of equilibrium and kinetic data Figure 1. (a). Representations of the structure of cks114 on the left, with the structure of suc1 on the right for comparison.39 (b). Topologies of cks1 and of suc1. Shown in red are the insertions in suc1 compared with cks1.
solvent conditions and to mutation. These movements preclude a full and quantitative F-value analysis of the transition state structure. Further, the perturbations are so large that they preclude a “continuous F-value” analysis, an approach developed by Oliveberg and co-workers.11 Using this approach, rate constants for wild-type and mutant are compared at the same b-Tanford value and continuous F-values can then be obtained as a function of the b-Tanford value. Thus, a picture is built up of both early and later parts of the transition-state energy barrier. For many cks1 mutants this approach would require the use of rate constants at negative concentrations of urea. Instead of these approaches, we use the change upon mutation in the slope of the unfolding arm of chevron plot as a qualitative measure of structure formation during the folding process. A similar approach was used recently for the protein S612 and for the 27th Ig domain of titin.13 We compare the results obtained from the unfolding data with the F-values determined from the rates of refolding. Finally, the data are consistent with an alternative, ground state explanation for the observed perturbations, involving “breathing” of the native structure. The folding mechanism of cks1 revealed by the analysis is in striking contrast to that of its homologue, suc1. Moreover, unlike cks1, suc1 does not exhibit large perturbations in folding behaviour upon mutation. The findings can be
The following parameters were obtained from the equilibrium data (Table 1) and kinetic data (Table 2): From the equilibrium data, we obtained the urea midpoint of unfolding ([U]50%), the equilibrium m-value (meq) the free energy of unfolding ðDGD-N Þ and the change in the free energy of unfolding upon mutation ðDDGD-N Þ: Representative equilibrium denaturation curves are shown in Figure 2. Truncation of the completely buried residues in the core leads to a large destabilisation of the native state. Two of these mutants, LA24 and FL69, could not be expressed in sufficient quantities for equilibrium measurement, most likely because they are very unstable, and therefore only the kinetic data are given. From the refolding kinetics we obtained the rate of refolding in water (kf), the slope of the refolding arm of the chevron (mf), and the F-value in water calculated using the refolding rate (Ff). From the unfolding kinetics, we obtained the slope of the unfolding arm of the chevron (mu), and the rate constants for unfolding in the absence of denaturant (ku). F-values were calculated from the unfolding data for those mutants where the change in mu compared with the wild-type value was less than 5%. Finally, the free energy of unfolding was calculated using the kinetic data ðDGkin D-N Þ from the ratio of refolding and unfolding rate constants. Categorisation of mutants according to experimental kinetic parameters Representative chevron plots are shown in Figure 3, fit according to a two-state model (see Materials and Methods). Wild-type cks1 and the mutants display curvature in the unfolding arm of the chevron plots. This second-order term was
191
Folding Mechanism of Human cks1
Table 1. Location of mutated sites, interactions deleted upon mutation, and equilibrium data Mutant
Location of mutated residue—location of interactions deleted upon mutation
m (kcal mol21 M21)
[U]50% (M)
DGD-N (kcal mol21)
DDG½U50% D-N (kcal mol21)
WT IV6 YA8 SA9 KA11 FL17 EA18 VA22 LA24b VA32 SA39 LA46 VA55 VG55 ML58 PA62 PA64 HA65 LA67 FL69b RA70 RA71
b1/b2 (corea) b1/b2 b1/b2 b1/b2 b2/b4 b2/b4 b2/b1 (corea) Loop/b4 a1/a1,a2 a2/a2 a2/a1,a2 (corea) b3/b4 b3/b4 b3/b4 Hinge loop Hinge loop Hinge loop b4/b2 b4/a2,b3 (corea) b4/b2,b3 b4/b2,b3
1.01 ^ 0.04 0.91 ^ 0.05 0.93 ^ 0.03 0.95 ^ 0.04 1.07 ^ 0.06 1.19 ^ 0.02 1.30 ^ 0.02 1.11 ^ 0.03 – 1.03 ^ 0.03 1.14 ^ 0.04 0.98 ^ 0.03 1.21 ^ 0.05 1.00 ^ 0.04 1.17 ^ 0.06 1.00 ^ 0.04 1.01 ^ 0.03 1.05 ^ 0.03 0.99 ^ 0.03 – 0.86 ^ 0.03 1.05 ^ 0.03
3.85 ^ 0.03 3.47 ^ 0.06 3.33 ^ 0.04 3.54 ^ 0.03 4.32 ^ 0.04 1.26 ^ 0.01 2.69 ^ 0.06 2.43 ^ 0.02 – 2.80 ^ 0.03 3.08 ^ 0.03 1.92 ^ 0.02 2.87 ^ 0.03 1.90 ^ 0.02 3.37 ^ 0.04 5.16 ^ 0.03 3.47 ^ 0.02 3.30 ^ 0.03 1.97 ^ 0.02 – 1.25 ^ 0.03 3.22 ^ 0.02
4.14 ^ 0.07 3.58 ^ 0.08 3.80 ^ 0.06 3.80 ^ 0.07 4.65 ^ 0.08 1.35 ^ 0.02 2.89 ^ 0.04 2.61 ^ 0.05 – 3.01 ^ 0.06 3.31 ^ 0.06 2.07 ^ 0.04 3.08 ^ 0.06 2.04 ^ 0.04 3.62 ^ 0.07 5.54 ^ 0.09 3.72 ^ 0.06 3.54 ^ 0.06 2.11 ^ 0.04 – 1.34 ^ 0.04 3.46 ^ 0.06
– 0.40 ^ 0.10 0.56 ^ 0.11 0.33 ^ 0.09 20.51 ^ 0.10 2.78 ^ 0.07 1.25 ^ 0.08 1.53 ^ 0.08 – 1.13 ^ 0.09 0.83 ^ 0.09 2.07 ^ 0.08 1.06 ^ 0.09 2.10 ^ 0.08 0.52 ^ 0.10 21.40 ^ 0.11 0.41 ^ 0.10 0.59 ^ 0.09 2.02 ^ 0.08 – 2.79 ^ 0.08 0.68 ^ 0.09
All measurements were made at 10 8C. The data were fitted to a two-state model to give [U]50%, the concentration of urea at which 50% of the protein is denatured, and m, a constant of proportionality that is related to the change in solvent exposure of hydrophobic is the free energy of unfolding, calculated using the equation: side-chains upon unfolding. DG½U50% D-N DGD-N ¼ kml½U50% and DDG½U50% is the change in the free energy of unfolding upon mutation, calculated using the equation: D-N DDG½U50% ¼ kmlD½U50% D-N where D½U50% is the difference in the concentrations of urea at which 50% of the wild-type and mutant proteins are denatured. kml, the weighted average of m-value for wild-type and all mutants, is 1:07 ^ 0:02: a core indicates that the mutated residue is located in the hydrophobic core. b The yields of these mutants were too low for equilibrium measurement to be performed.
similar for wild-type and mutants. Because it is difficult to fit this term accurately, a weighted average value was obtained from the wild-type and mutants, and the chevron plots were then refit with the second-order term fixed at the average value ð0:054 ^ 0:010Þ: Seven non-exclusive types of folding behaviour were noted and the mutants were categorised accordingly. Type I mu mutant , mu wild-type: YA8, SA9, KA11, EA18, PA62 and HA65. Type II mu mutant , mu wild-type: IV6, VA22, VA32, SA39, VA55, ML58, PA64 and RA71. Type III mu mutant p mu wild-type: FL17, LA24, LA46, VG55, LA67, FL69 and RA70. Type IV High Ff (. 0.3): IV6, YA8, SA9, FL17, EA18, VA22, ML58, PA62, HA65 and RA70. Type V Low Ff (, 0.3): VA32, SA39, LA46, VA55, VG55, PA64, LA67 and RA71. Type VI DGkin D-N , eq DGD-N : YA8, SA9, KA11, EA18, LA46, ML58, PA62 eq and HA65. Type VII DGkin D-N p DGD-N : IV6, FL17, VA22, VA32, SA39, VA55, VG55, PA64, LA67, RA70 and RA71. Transition-state movement provides a picture of the sequence of folding events Here, we have used the magnitude of the movement of the position of the transition state along
the reaction coordinate, probed experimentally by the kinetic m-values, as a measure of structure formation. The rationale is as follows (Figure 4(a)): the transition-state movements observed for cks1 can be explained if the energy barrier is viewed as broad and smooth. Consequently, if a mutation is made in a part of the structure that forms late in the folding process, on the downhill side of the wild-type transition state, then the late regions of the energy barrier, but not the early regions, are destabilised. Consequently, the peak of the energy barrier is shifted towards the native state and the observed mu value decreases. Conversely, if a mutation is made in a part of the structure that forms early in the folding process, on the uphill side of the wild-type transition state, then the whole of the energy barrier is destabilised and the peak does not shift its position along the reaction coordinate. The curvature observed in the unfolding arm of the chevrons is explained also by transition-state movement, in accord with Hammond-postulate behaviour.15 At high concentrations of urea, the late part of the energy barrier is destabilised more than the early part, because the sensitivity to denaturant increases the further along the reaction coordinate the species is located.
192
Folding Mechanism of Human cks1
Table 2. Kinetic parameters for cks1 wild-type and mutants Protein Wild-type IV6 YA8 SA9 KA11 FL17 EA18 VA22 LA24 VA32 SA39 LA46 VA55 VG55 ML58 PA62 PA64 HA65 LA67 FL69 RA70 RA71
kf (s21)
mf (M21)
O FH F
ku (s21)
mu (M21)
O 1 2 FH U
21 DGkin D-N (kcal mol )
92.1 ^ 2.6 64.5 ^ 2.5 68.3 ^ 2.3 70.4 ^ 1.5 74.4 ^ 1.6 8.4 ^ 0.2 7.9 ^ 1.5 38.8 ^ 4.7 36.4 ^ 7.0 67.4 ^ 2.1 61.6 ^ 2.7 59.9 ^ 1.1 71.9 ^ 6.8 65.4 ^ 6.1 63.2 ^ 3.0 222 ^ 16 76.7 ^ 6.2 53.4 ^ 1.6 45.0 ^ 1.8 66.9 ^ 14.4 18.4 ^ 1.5 89.0 ^ 3.6
0.61 ^ 0.06 0.57 ^ 0.05 0.45 ^ 0.04 0.47 ^ 0.03 0.47 ^ 0.03 –b 0.70 ^ 0.30 0.69 ^ 0.22 –b 0.49 ^ 0.05 0.55 ^ 0.09 0.42 ^ 0.33 0.65 ^ 0.19 0.62 ^ 0.19 0.51 ^ 0.08 0.59 ^ 0.04 0.22 ^ 0.09 0.59 ^ 0.09 0.51 ^ 0.34 0.80 ^ 0.42 0.87 ^ 0.22 0.49 ^ 0.06
0.49 0.30 0.45 20.23 0.48 1.10 0.32 –c 0.15 0.27 0.12 0.13 0.09 0.41 0.35 0.25 0.51 0.20 –c 0.32 0.03
0.065 ^ 0.018 0.27 ^ 0.03 0.114 ^ 0.024 0.073 ^ 0.013 0.039 ^ 0.010 3.45 ^ 0.35 0.080 ^ 0.015 0.62 ^ 0.16 9.01 ^ 0.70 0.59 ^ 0.04 0.58 ^ 0.06 1.88 ^ 0.43 0.95 ^ 0.23 4.57 ^ 0.50 0.29 ^ 0.05 0.019 ^ 0.008 0.32 ^ 0.12 0.058 ^ 0.011 2.32 ^ 0.33 4.09 ^ 0.82 4.94 ^ 0.17 0.66 ^ 0.11
1.37 ^ 0.04 1.21 ^ 0.01 1.33 ^ 0.03 1.37 ^ 0.03 1.40 ^ 0.04 0.97 ^ 0.02 1.32 ^ 0.02 1.12 ^ 0.04 0.92 ^ 0.01 1.17 ^ 0.01 1.15 ^ 0.01 1.09 ^ 0.03 1.12 ^ 0.04 0.94 ^ 0.02 1.21 ^ 0.02 1.39 ^ 0.07 1.21 ^ 0.05 1.40 ^ 0.03 1.06 ^ 0.02 1.01 ^ 0.03 0.89 ^ 0.01 1.17 ^ 0.02
–a 0.44 0.80 0.44 –a 0.91 –a –c –a –a –a –a –a –a 0.51 –a 1.11 –a –c –a –a
4.06 ^ 0.16 3.07 ^ 0.07 3.58 ^ 0.12 3.85 ^ 0.10 4.23 ^ 0.14 0.50 ^ 0.06 2.57 ^ 0.15 2.32 ^ 0.15 0.78 ^ 0.12 2.65 ^ 0.04 2.61 ^ 0.06 1.94 ^ 0.13 2.42 ^ 0.14 1.49 ^ 0.08 3.02 ^ 0.10 5.24 ^ 0.24 3.07 ^ 0.21 3.82 ^ 0.11 1.66 ^ 0.08 1.57 ^ 0.16 0.74 ^ 0.05 2.75 ^ 0.10
2
2
All measurements were made at 10 8C. The rate constants and their respective m-values were obtained by fitting the chevron plots according to a two-state model (as described in Materials and Methods) using the equation: 2O 2O ðexpð2mf ½U 2 0:054½U2 Þ þ kH ðexpðmu ½U 2 0:054½U2 Þ kobs ¼ kH u f
DGkin D-N is the free energy of unfolding calculated from the kinetic data using the equation: H2 O H2 O DGkin =ku Þ D-N ¼ RT lnðkf
where R is the gas constant and T is the temperature (in Kelvin). a Fu was not calculated in the case of mutants for which there was a significant change in mu compared with the wild-type value. b The error was larger than the value for these mutants because the refolding arm of the chevron was very short. c F-values could not be calculated for these mutants because the protein yields were too low to enable equilibrium measurement to be made.
Ground-state versus transition-state effects An alternative explanation for the changes in mu upon mutation is movement of the ground (native)
state (Figure 4(b)). A mutant that causes a decrease in mu can do so either by shifting the peak of the transition-state barrier towards the native state or by destabilising the native state sufficiently for it
Figure 2. Equilibrium denaturation curves for wild-type and representative mutants. Measurements were made at 10 8C in 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, and with a protein concentration of 4 mM. The data were fit according to a two-state model.
193
Folding Mechanism of Human cks1
Figure 4. Two alternative explanations for the observed decrease in mu upon mutation. The energy profile of wild-type is shown with a continuous line and with a broken line for the mutant. (a) Transition-state movement. If the structure at the site of mutation is formed late in the folding reaction, only the late part of the transition-state energy barrier is destabilised and therefore the top of the energy barrier is shifted towards the native state along the reaction coordinate. (b) Ground-state movement. Mutation at a site that forms late in the folding reaction causes the native state to be destabilised, leading to local unfolding or expansion, and the native state is consequently shifted towards the transition state along the reaction.
mediate. Therefore, whether the changes in mu arise from native-state fraying or transition-state movement, the implications for the order for structure formation during the folding processes are the same; i.e. those parts of the structure at which mutation has such an effect must consolidate late in folding. Order of structure formation during folding: sites that fold early Figure 3. Chevron plots for representative mutants at sites that form (a) early, YA8 (W), SA9 (L), KA11 (A), EA18 (B), PA62 ( £ ), HA65 (V), (b) at an intermediate stage, IV6 (W), VA22 (A), VA32 (L), SA39 ( £ ), and (c) late in the folding process, LA46 (A), VG55 (L), LA67 ( £ ), FL69 (V) and RA70 (W). Wild-type (X) is shown in each plot for comparison. Measurements were made at 10 8C in conditions, after mixing in the stopped-flow apparatus, of 50 mM Tris –HCl (pH 7.5), 1 mM EDTA, and a protein concentration of 2 mM. The fits of the data according to a two-state model are shown.
to “fray”, becoming less compact in the process. There may be a continuum of expanded nativelike species or, alternatively, a discrete unfolding intermediate populated in the dead-time of the kinetic experiment. Only mutants at those sites that form late in the folding reaction will destabilise the native state relative to the native-like inter-
Mutation of only six residues has no effect on mu (type I): YA8 (b1), SA9 (b1), KA11 (b1), EA18 (b2), PA62 (hinge) and HA65 (hinge) (Figure 3(a); Table 2). They are the only mutants that unfold at a rate similar to that of the wild-type. All except KA11 refold significantly more slowly than wildtype (type IV) and the F-values calculated from either the refolding or the unfolding rates are high for these mutants (between 0.3 and 1). The mutant PA62 also falls into this category (Figure 3(a)). Although it displays pronounced rollover in the refolding arm, as discussed in more detail later (see section on The stabilised mutant PA62 changes the refolding arm of the chevron), a F-value can be calculated from the data measured above 1 M urea where the chevron plot runs parallel with that of the wild-type. The F-value obtained in this way is high (0.4) and it is in good agreement with the F-value obtained using the unfolding data (0.5). Finally, the mutant KA11 behaves like PA62—it is significantly more stable than wild-type and
194
displays slight rollover in the refolding arm. As a result of the rollover, the rate of refolding in water is smaller than that of wild-type, and this gives an anomalous F-value of less than zero, but above 1 M urea the refolding rate is greater than wildtype.
Folding Mechanism of Human cks1
Folding and unfolding data give consistent results The sequence of events revealed by the refolding data is generally in good agreement with that revealed by the unfolding data. Sites displaying low F-values also display low values of mu, while residues with high F-values have values of mu that are similar to that of the wild-type.
Sites that fold late The majority of the cks1 mutations change mu significantly compared with the wild-type value, suggesting that the sites fold late and unfold early (Figure 3(c)). This observation is consistent with the low b-Tanford value (b‡) of the wild-type protein in water of 0.27 (calculated as 1 2 ðmu =meq Þ rather than mf =meq ; because mu could be determined with greater accuracy than mf), indicating a very expanded transition state. Also, the majority of the mutants unfold faster than the wild-type over the whole range of concentrations of denaturant sampled, again, indicating that there is weak structure formation in the transition state for folding. The mutants with the lowest values of mu (type III) are FL17 (b2), LA24 (loop/ core), LA46 (loop/core), VG55 (b3), LA67 (b4), FL69 (b4/core) and RA70 (b4). Finally, the majority of the mutants probed in cks1 refold at a rate similar to that of the wild-type (type V), again consistent with only weak structure formation in the transition state. The F-values calculated from the refolding rates for these mutants are all less than 0.3. To summarise, there is reasonable agreement between type III and type V mutants, with both categories, indicating regions of the protein that form late in the folding reaction. There are two exceptions; FL17 and RA70 are mutants that refold significantly more slowly than wild-type but that exhibit very low mu values (Figure 3(c); Table 2). However, the low refolding rates of these mutants may not indicate early structure formation at these sites, but rather they may be an artefact due to off-pathway aggregation during refolding. This behaviour was observed for the wild-type at high concentrations of protein,9 and the two mutants were observed to increase greatly this aggregation propensity.
Partly structured sites A few mutations have a small effect on mu (type II): IV6 (b1), and ML58 (b3) and PA64 (hinge) (Figure 3(b); Table 2). The F-values, calculated from the refolding rates, are between 0.4 and 0.5 for these mutants (classed as type IV). Mutations VA22, VA32, SA39, VA55 and RA71 have a slightly greater effect on mu, indicating that these sites form somewhat later in the folding reaction. Consistent with this, the F-values, calculated from the refolding rates, are between 0 and 0.3 for these mutants (type V).
Increase in equilibrium m-values upon mutation indicates ground-state movement One way to assess whether the perturbations observed in the unfolding kinetics are due to ground-state or to transition-state movement is to look for changes upon mutation in the equilibrium m-values and whether any changes correlate with changes in the kinetic m-values. There is no correlation between mu and meq, and there is a weak correlation between mf and meq, The increase in mf with increasing meq suggests that the denatured, ground state for refolding becomes expanded upon mutation at certain sites. However, the absence of a correlation between mu and meq suggests that the changes in mu upon mutation are due to transition-state movement rather than native, ground-state movements. Alternatively, it is possible that native-state movement is apparent only at the very high concentrations of denaturant used for determining the unfolding kinetics, when the late part of the reaction coordinate is already highly destabilised, but that it is not apparent at the lower concentrations of denaturant around the transition midpoint where meq is measured. Discrepancies between equilibrium and kinetic data highlight the changes in the free-energy barrier over the range of concentrations of denaturant In a two-state system, the ratio of the folding and unfolding rate constants in water can be used to calculate the overall free energy of unfolding. For cks1 wild-type and a number of mutants (type VI), there is good agreement between this value measured (DGkin D-N ; given in Table 2) and the value eq by equilibrium denaturation (DGD-N given in Table 1). However, for many mutants (type VII: IV6, FL17, VA22, VA32, SA39, VA55, VG55, PA64, LA67, RA70 and RA71), DGkin D-N is significantly eq lower than DGD-N : The curvature used to fit the unfolding arm of the chevron implies a smooth movement of the transition state with changing concentration of urea. The non-coincidence between the kinetic and equilibrium data suggests that this extrapolation of the kinetic unfolding data to water is not correct for this subset of mutants. The mutants are all located in parts of the structure that form late, as judged by the low mu values. It is possible that only at high concentrations of denaturant, when the late part of the energy barrier is already very destabilised relative
195
Folding Mechanism of Human cks1
Comparison with suc1: structural details The sequence of events in the folding reaction, as revealed by the m-value and F-value analysis, is also different in the two proteins. The F-value analysis of suc1 revealed that the central strands of the sheet, b2 and b4, are the most structured in the transition state (F ¼ 0.6– 0.9) whilst the outer strands, b1 and b3, and the loops, are less structured (F ¼ 0 – 0.3). Thus, suc1 folds from the centre of the hydrophobic core outwards. For cks1, in contrast, the F-values are highest for the end of b1/ start of b2 and the end of b3/hinge/start of b4 (Figure 5). Thus, strand pairing between adjacent b-strands in the sequence, i.e. b1/b2 and b3/b4, occurs first in the folding of cks1. Comparison with other proteins of low stability
Figure 5. Location in the structure of sites that fold early (blue) and late (red) in the folding reaction of cks1.
to the early part, can these mutations cause a shift in position of the top of the transition-state energy barrier. Thus, the unfolding arm could be kinked at intermediate concentrations of denaturant, rather than being curved smoothly over the entire range of concentrations. The observed eq discrepancy between DGkin D-N and DGD-N can be explained in terms of the ground-state effects. For these mutants, the ground state for the unfolding reaction of the mutants at low concentrations of urea may be as compact as the wild-type native state, and it may become expanded only at very high concentrations of urea.
Comparison of cks1 with suc1: the global characteristics of the energy barrier for folding The perturbations in the folding behaviour of cks1 upon mutation preclude a comprehensive and quantitative F-value analysis. However, a number of features emerge from the qualitative analysis of the data described above, and it is particularly revealing to compare cks1 with its homologue suc1, for which full F-value analysis has been possible.10 The rate-determining transition state for cks1 is less compact than that of suc1: b‡ for cks1 is 0.27 compared with 0.60 for suc1. A folding intermediate was analysed for suc1 using F-values and this species is reasonably compact (b-Tanford is 0.45). The position of the transition state for cks1 is much more susceptible to perturbation by mutation than that for suc1, although the sensitivity to denaturant, as reflected in the curvature of the unfolding arm, appears similar in the two proteins.
How do the transition-state movements compare with those observed for other proteins? Can the explanation for the behaviour of cks1 lie, in part, in its very low thermodynamic and/or kinetic stability ? With 156 residues, p16 is a much larger protein than cks1, but the free energies of unfolding are similar (3.1 kcal mol21 for p16 compared with 3.5 kcal mol21 for cks1 at 25 8C) as are their kinetic stabilities (ku in water is , 1 s21 for both proteins at 25 8C).16 However, the transition state for folding/unfolding of p16 changes very little upon mutation. The reason for this becomes apparent when the extent of structure in the transition state is considered. The transition state is very native-like: b‡ is 0.9 in water and the F-values for the majority of mutants are close to 1.17 Therefore, again viewing the transition state as a broad barrier, most mutations cause the whole barrier to be destabilised and so there is no overall shift in the top of the barrier. In contrast, the transition state of cks1 is much less structured and therefore the majority of mutations destabilise only the later part of the transition-state barrier, causing a shift in the top of the barrier. Thus, the sensitivity of the transition state is likely to be influenced by a number of factors that include the stability of the native state and whether the rate-determining step is early or late in the folding process (i.e. how native-like it is). Viewed as a ground-state effect, again for a protein such as p16, with a transition state lying very close to the native state, there is much less potential for fraying of the native state. Weakening the cooperativity of the core leads to a change in folding mechanism In suc1, the non-sequential pair of strands, b2 and b4, that constitute the centre of the b-sheet and of the core of the protein, form first. The residues with highest F-values are located in b2 and b4, and residues that interact with these two strands. In cks1, the highest F-values (and mu values) are shifted away from the central strands to the peripheral strands. Strand pairs that are
196
tandem in sequence form together (i.e. b1 with b2 and b3 with b4). Thus, local interactions initiate folding in cks1 and non-local ones in suc1. In terms of sequence separation, it is not more costly to bring together tandem pairs in suc1 than it is in cks1. In fact, the additional sequence present in suc1 is an a-helix located at the N terminus and a loop between b2 and b3. Consequently, the entropic cost of bringing together b2 and b4 in suc1 should be greater than in cks1. The reason why these are the first to form in suc1, despite the greater entropic cost, may be due to the greater enthalpic gain of forming this pair in suc1 compared with cks1. The b2/b4 strand pair is stabilised by the extensive network of interactions that comprises the hydrophobic core of suc1. In cks1, the core is much less extensive. It consists mainly of residues in the b-sheet and lacks the further packing interactions between the end of the b-sheet and the a-helical region that are present in suc1. Thus, the four strands are balanced more equally in terms of enthalpy, so the closest strands form first. Does stabilisation of the native state of cks1 induce suc1-like folding behaviour? The stability of the native state of cks1 can be increased in two ways, first by mutation (PA62 is 1.4 kcal mol21 more stable than wild-type), and second, by the addition of solutes. Addition of 500 mM sodium phosphate increases the stability of wild-type by 1.9 kcal mol21 (data not shown). We examined the effect of a mutation that causes a large change in mu (RA70) on the unfolding kinetics of these two stabilised proteins. In spite of the additional stability, conferred either by mutation or by addition of solute, the mutation RA70 still resulted in a large change in mu. Thus, the susceptibility of the folding of cks1 to perturbation does not arise from its overall low thermodynamic stability per se. Rather it is the weak cooperativity of the cks1 core that causes its vulnerability and results in a change in the folding sequence compared with suc1. The stabilised mutant PA62 changes the refolding arm of the chevron As stated previously, the mutant PA62 is stabilised by 1.4 kcal mol21 relative to wild-type and its chevron exhibits rollover in the refolding arm (see Figure 3(a)). Again, the susceptibility of the folding behaviour of cks1 is apparent. In contrast to the effect that a destabilising mutation has on the unfolding arm of the chevron if it is in a region of the structure that forms late in the folding reaction, a stabilising mutation in a region that forms early perturbs the refolding arm of the chevron. In regions that form late, a destabilising mutation may allow an unfolding intermediate to be populated; and in regions that form early, a stabilising mutation may allow a refolding intermediate to be
Folding Mechanism of Human cks1
populated. Thus, the rollover observed for PA62 can be interpreted as stabilisation of a refolding intermediate before the transition state. The stabilised mutant KA11 shows behaviour similar to that of PA62.
Conclusions The entropy –enthalpy balance in protein folding appears to be resolved by forming the best contacts between residues that are as close in sequence as possible,18 – 21 and this basic principle has enabled the prediction of folding mechanisms from the native-state topology22 – 24 Less predictable is the case of highly symmetrical structures, for example protein G and protein L, and the specific distribution of interaction energies throughout the protein can then tip the balance in favour of one alternative folding route rather than another.25 – 27 One type of study that illustrates this principle of entropy versus enthalpy, or topological constraints versus energetic factors, is the folding of circular permutants where the chain connectivity is changed relative to the wild-type, whilst the strength of the interactions is unaltered.28 – 35 If the chain is cut in the part of the structure that folds early, then it becomes energetically unfavourable to fold this part first in the permutant and this can be manifest in a shift in the transition-state structure. A related example is the folding of the dimeric coiled-coil protein GCN4 compared with that of a single-chain monomeric variant that was cross-linked at one end.36 However, some transition-state structures appear to be much more resilient to these gross perturbations in topology, such as that of chymotrypsin inhibitor 2, which does not change upon permutation.33 The folding mechanisms of suc1 and cks1 illustrate the entropy– enthalpy balance. Although it is difficult to obtain a quantitative description of the folding mechanism of cks1 without a complete F-value analysis, the available data show clearly that the order of structure formation is different for cks1 compared with the homologue suc1. This is somewhat surprising, since cks1 has the same b-sheet core structure, the centre of which was shown previously to initiate folding in suc1. Moreover, the level of sequence identity is very high. However, cks1 is much smaller than suc1 (79 residues compared with 113). It lacks 20 amino acid residues at the N terminus, ten of which form an a-helix, and a nine residue unstructured loop located between b2 and b3. Thus, there is a greater entropic cost of forming the b2/b4 strand pair in suc1 compared with cks1, yet this region forms early in suc1 and late in cks1. In contrast, the structure of cks1 is less cooperative and the entropy/ enthalpy balance tips the folding route in favour of sequential rather than non-sequential strand pairing. Remarkably, even when the folding nucleus of suc1 is drastically disfavoured entropically, as occurs in the domain-swapped dimer
197
Folding Mechanism of Human cks1
where the b2/b4 pairing is made inter-molecularly, the folding mechanism is not altered.37 The folding nucleus does not shift, it remains in the same part of the structure but it recruits more interactions and thus a greater enthalpic gain in order to balance the greater entropic penalty of forming the dimer. The reason for this tolerance in suc1 may be that there is a much greater enthalpic gain in forming the core of suc1 compared with cks1. With an increasing number of protein engineering studies of folding now available, it is becoming clear that small movements of the free-energy barrier for folding/unfolding with changes in concentration of denaturant and mutation are normal rather than being exceptions.20 However, cks1 lies at the extreme end of the spectrum of this type of behaviour. Why is this protein so sensitive whilst many others are quite robust in their response to mutation? Two factors appear to contribute. The first factor is reflected by the low thermodynamic stability of cks1 and the second factor is that the rate-determining step in the folding of cks1 is very early in the reaction. A comparison with two other proteins illustrates these contributions. The first is suc1, a homologue of cks1 that has high level of sequence identity but is a much larger protein. The much greater stability conferred by the additional structural elements protects it from large perturbation upon mutation. The second protein is p16, which has a stability similar to that of cks1 but its transition state is much more nativelike.16 Again, the consequence is little or no perturbation upon mutation.17 The kinetic behaviour of wild-type and mutants of cks1 can be rationalised in terms of movement of the top of the broad energy barrier for folding.12 An alternative explanation is that the mutations cause expansion of the ground (native) state, and there is evidence that the denatured state, the ground state in the refolding direction, is susceptible to mutation. In the case of cks1, this expansion of the native state is described better by the term breathing rather than fraying, since it is the centre of its structure, rather than the outer b-strands, that are expanding.
Materials and Methods Mutagenesis and protein expression and purification Site-directed mutagenesis was performed using the QuikChange kit (Stratagene) and mutant plasmids were identified by sequencing (Oswell, University of Southampton, UK). Mutant proteins were expressed and purified as described for the wild-type.9 Protein identity and purity were confirmed by mass spectrometry.
were performed at 10 8C. The low temperature was used because the rates of unfolding (described below) at 25 8C were too fast to be measured with accuracy. The data were fit to a two-state transition with the fluorescence of the native and denatured states dependent on the concentration of denaturant. Data fitting was carried out using the non-linear least-squares algorithm provided with the program Kaleidagraph (Abelbeck Software). The two-state fit gives an m-value (a constant of proportionality that is related to the change in solvent exposure of hydrophobic side-chains upon unfolding) and [U]50% (the concentration of urea at which 50% of the ½U50% protein is denatured).38 Thus, DDGD-N ¼ kmlD½U50% ; ½U50% where D½U50% and DDGD-N are the difference in the concentrations of urea at which 50% of the wild-type and mutant proteins are denatured and the change in the free energy of unfolding upon mutation, respectively. kml; the weighted average value of m obtained for wildtype and all the mutants, was 1:07 ^ 0:02: Kinetic analysis Kinetic experiments were performed using an Applied Photophysics SX-17MV stopped-flow fluorescence-detected stopped-flow instrument at 10 8C. The excitation wavelength was 280 nm and fluorescence was collected using a 325 nm cut-off filter. Unfolding was initiated by mixing one volume of protein in 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, with ten volumes of urea in 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, to give final conditions of 50 mM Tris – HCl (pH 7.5), 1 mM EDTA. The concentration of protein after mixing was 2 mM. Under strong unfolding conditions outside the unfolding transition region, the kinetic data could be fit to a single-exponential function. The urea-dependence of the rate constant of unfolding shows slight downward curvature over the wide range of concentrations of urea according to Hammond behaviour and the data were fit to the second-order polynomial equation: H2 O ln k½U þ mu þ mpu ½U2 u ¼ ln ku
where ln ku½U and ln kuH2 O are the rate constants in a concentration of urea, [U], and in water, respectively, mu is the kinetic m-value and mpu is the second-order term. In order to minimise the error, a weighted average value of mpu (20.054 ^ 0.010) was calculated using the wild-type and all the mutants. This value was then used to refit the data to a second-order polynomial with fixed curvature. Refolding was initiated by mixing acid-denatured protein in 30 mM HCl with an equal volume of 100 mM Tris – HCl buffer (pH 8.1), 2 mM EDTA, to give final conditions of 50 mM Tris –HCl (pH 7.5), 1 mM EDTA. The concentration of protein after mixing was 2 mM. The data were fit to the sum of two exponential phases, representing a fast-refolding phase accounting for most of the amplitude and a slow-refolding phases involving proline isomerisation, as reported.9 The fast-refolding rate constant, kf, is used for the F-value analysis. The kinetic data in the form of chevron plots were fitted to a two-state model using the equation: 2O 2O kobs ¼ kH ðexpð2mf ½U 2 0:054½U2 Þ þ kH ðexpðmu ½U u f
Equilibrium denaturation Urea-induced equilibrium denaturation was monitored by fluorescence spectroscopy as described.9 The protein concentration was 4 mM, the buffer was 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, and all experiments
2 0:054½U2 Þ using the non-linear, least-squares algorithm provided O O with the program Kaleidagraph. kH and kH are, f u respectively, the rate constants for refolding and 2
2
198
Folding Mechanism of Human cks1
unfolding in water. mf and mu are, respectively, the kinetic refolding and unfolding m-values in water. The average value of the curvature, 0.054, was used. The change upon mutation in the free energy of the transition state relative to the native state, DDG‡-F ; is calculated 2O using the equation: DDG‡-N ¼ 2RTðln kH 2 ln k0u H2 O Þ; u O 0 O where kH and kH are the rate constants of unfolding u u of wild-type and the mutant respectively. The change upon mutation in the free energy of the transition state relative to the denatured state, DDGU-‡ ; is calculated 2O using the equation: DDGU-‡ ¼ RTðln kH 2 ln k0f H2 O Þ; f HO H 0 O where kf and kf are the rate constants of refolding of wild-type and the mutant respectively. The ratio between DDG‡-N and DDGD-N is defined as FU. The ratio of DDGD-‡ and DDGD-N gives the F-value in the refolding direction, FF, where FF ¼ 1FU. A FF-value of 0 means that the interactions deleted upon mutation are formed as weakly in the transition state as it is in the denatured state. A F-value of 1 means that the interactions are as energetically well formed in the transition state as in the native structure. 2
2
9.
2
10.
2
11.
12.
13.
14.
Acknowledgements This work was supported by the Medical Research Council of the UK (MRC). L.S.I. was supported by a Career Development Award from the MRC. M.A.S. was supported by an External Research Studentship from Trinity College, Cambridge, UK.
15. 16.
17.
References 1. Hayles, J., Aves, S. & Nurse, P. (1986). suc1 is an essential gene involved in both the cell cycle and growth in fission yeast. EMBO J. 5, 3373– 3379. 2. Ganoth, D., Bornstein, G., Ko, T. K., Larsen, B., Tyers, M., Pagano, M. & Hershko, A. (2001). The cell-cycle regulatory protein Cks1 is required for SCFSkp2mediated ubiquitinylation of p27. Nature Cell Biol. 3, 321– 324. 3. Bennett, M. J., Choe, S. & Eisenberg, D. (1994). Domain swapping: entangling alliances between proteins. Proc. Natl Acad. Sci. USA, 91, 3127– 3131. 4. Bourne, Y., Arvai, A. S., Bernstein, S. L., Watson, M. H., Reed, S. I., Endicott, J. A. et al. (1995). Crystal structure of the cell cycle-regulatory protein suc1 reveals a b-hinge conformational switch. Proc. Natl Acad. Sci. USA, 92, 10232– 10236. 5. Parge, H. E., Arvai, A. S., Murtari, D. J., Reed, S. I. & Tainer, J. A. (1993). Human CksHs2 atomic structure: a role for its hexameric assembly in cell cycle control. Science, 262, 387– 394. 6. Rousseau, F., Schymkowitz, J. W. H., Wilkinson, H. R. & Itzhaki, L. S. (2001). Three-dimensional domain swapping in p13suc1 occurs in the denatured state and is controlled by conserved proline residues. Proc. Natl Acad. Sci. USA, 98, 5596 –5601. 7. Schymkowitz, J. W. H., Rosseau, F., Wilkinson, H. R., Friedler, A. & Itzhaki, L. S. (2001). Observation of signal transduction in three-dimensional domain swapping. Nature Struct. Biol. 8, 888– 892. 8. Rousseau, F., Schymkowitz, J. W. H., Sanchez del Pino, M. & Itzhaki, L. S. (1998). Stability and folding
18. 19.
20.
21.
22.
23.
24.
25.
of the cell cycle regulatory protein, p13suc1. J. Mol. Biol. 284, 503– 519. Seeliger, M. A., Schymkowitz, J. W. H., Rousseau, F., Wilkinson, H. R. & Itzhaki, L. S. (2002). Folding and association of the human cell cycle regulatory proteins ckshs1 and ckshs2. Biochemistry, 41, 1202– 1210. Schymkowitz, J. W. H., Rousseau, F., Irvine, L. R. & Itzhaki, L. S. (2000). The folding pathway of the cellcycle regulatory protein p13suc1: clues for the mechanism of domain swapping. Struct. Fold. Des. 8, 89 –100. Ternstrom, T., Mayor, U., Akke, M. & Oliveberg, M. (1999). From snapshot to movie: F value analysis of protein folding transition states taken one step further. Proc. Natl Acad. Sci. USA, 96, 14854– 14859. Otzen, D. E. & Oliveberg, M. (2002). Conformational plasticity in folding of the split b –a – b protein S6: evidence for burst-phase disruption of the native state. J. Mol. Biol. 317, 613– 627. Fowler, S. B. & Clarke, J. (2001). Mapping the folding pathway of an immunoglobulin domain: structural detail from phi value analysis and movement of the transition state. Structure, 9, 355– 366. Arvai, A. S., Bourne, Y., Hickey, M. J. & Tainer, J. A. (1995). Crystal structure of the human cell cycle protein CksHs1: single domain fold with similarity to kinase N-lobe domain. J. Mol. Biol. 249, 835– 842. Hammond, G. S. (1953). A correlation of reaction rates. J. Am. Chem. Soc. 77, 334– 338. Tang, K. S., Guralnick, B. J., Wang, W. K., Fersht, A. R. & Itzhaki, L. S. (1999). Stability and folding of the tumour suppressor protein p16. J. Mol. Biol. 285, 1869– 1886. Tang, K. S., Fersht, A. R. & Itzhaki, L. S. (2002). Sequential unfolding of ankyrin repeats in tumour suppressor p16. Structure. In press. Dinner, A. R. & Karplus, M. (2001). The roles of stability and contact order in determining proten folding rates. Nature Struct. Biol. 8, 21 –22. Fersht, A. R. (2000). Transition-state structure as a unifying basis in protein-folding mechanisms: contact order, chain topology, stability and the extended nucleus mechanism. Proc. Natl Acad. Sci. USA, 97, 1525– 1529. Oliveberg, M. (2001). Characterisation of the transition state for protein folding: towards a new level of mechanistic detail in protein engineering analysis. Curr. Opin. Struct. Biol. 11, 94 – 100. Plaxco, K. W., Simons, K. T., Ruczhinski, I. & Baker, D. (2000). Topology, stability, sequence, and length: defining the determinants of two-state protein folding kinetics. Biochemistry, 39, 11177– 11183. Alm, R. & Baker, D. (1999). Prediction of proteinfolding mechanisms from free-energy landscapes derived from native structures. Proc. Natl Acad. Sci. USA, 96, 11305– 11310. Galzitskaya, O. & Finkelstein, A. (1999). A theoretical search for folding/unfolding nuclei in three-dimensional protein structures. Proc. Natl Acad. Sci. USA, 96, 11299– 11304. Munoz, V. & Eaton, W. (1999). A simple model for calculating the kinetics of protein folding from three-dimensional structures. Proc. Natl Acad. Sci. USA, 96, 11311 – 11316. Kim, D., Fisher, C. & Baker, D. (2000). A breakdown of symmetry in the folding transition state of protein L. J. Mol. Biol. 298, 971– 984.
Folding Mechanism of Human cks1
26. McCallister, E. L., Alm, E. & Baker, D. (2000). Critical role of beta-hairpin formation in protein G folding. Nature Struct. Biol. 7, 669– 673. 27. Nauli, S., Kuhlman, B. & Baker, D. (2001). Computerbased redesign of a protein folding pathway. Nature Struct. Biol. 8, 602– 605. 28. Clementi, C., Jennings, P. A. & Onuchic, J. N. (2001). Prediction of folding mechanism for circularpermuted proteins. J. Mol. Biol. 311, 879– 890. 29. Grantcharova, V., Riddle, D., Santiago, J., Alm, E., Tsai, J. & Baker, D. (1999). Experiment and theory highlight role of native state topology in SH3 folding. Nature Struct. Biol. 6, 1016– 1024. 30. Grantcharova, V. P. & Baker, D. (2001). Circularization changes the folding transition state of the src Sh3 domain. J. Mol. Biol. 306, 555– 563. 31. Lindberg, M. O., Tangrot, J., Otzen, D. E., Dolgikh, D. A., Finkelstein, A. V. & Oliveberg, M. (2001). Folding of circular permutants with decreased contact order: general trend balanced by protein stability. J. Mol. Biol. 314, 891– 900. 32. Miller, E. J., Fischer, K. F. & Marqusee, S. (2002). Experimental evaluation of topological parameters determining protein-folding rates. Proc. Natl Acad. Sci. USA, 99, 10359– 10363. 33. Otzen, D. & Fersht, A. R. (1998). Folding of circular and permuted chymotrypsin inhibitor 2: retention of the folding nucleus. Biochemistry, 37, 8139– 8146.
199
34. Viguera, A., Blanco, F. & Serrano, L. (1995). The order of secondary structure elements does not determine the structure of a protein but does affect its folding kinetics. J. Mol. Biol. 247, 670– 681. 35. Viguera, A. R., Serrano, L. & Wilmanns, M. (1996). Different folding transition-states may result in the same native structure. Nature Struct. Biol. 3, 874– 880. 36. Moran, L. B., Schneider, J. P., Kentsis, A., Reddy, G. A. & Sosnick, T. R. (1999). Transition state heteroeneity in GCN4 coiled coil folding studied by using multisite mutations and crosslinking. Proc. Natl Acad. Sci. USA, 96, 10699– 10704. 37. Rousseau, F., Schymkowitz, J. W. H., Wilkinson, H. R. & Itzhaki, L. S. The structure of the transition state for folding of domain-swapped dimeric p13suc1. Structure, 10, 649– 657. 38. Clarke, J. & Fersht, A. R. (1993). Engineering disulfide bonds as probes of the folding pathway of barnase—increasing the stability of proteins against the rate of denaturation. Biochemistry, 32, 43322 – 44329. 39. Endicott, J. A., Noble, M. E., Garman, E. F., Brown, N., Ramussen, R., Nurse, P. & Johnson, L. N. (1995). The crystal structure of p13suc1, a p34cdc2-interacting cell cycle control protein. EMBO J. 14, 1004 –1014.
Edited by C. R. Matthews (Received 14 August 2002; received in revised form 18 October 2002; accepted 18 October 2002)
J. Mol. Biol. (2003) 325, 189–199
Weak Cooperativity in the Core Causes a Switch in Folding Mechanism Between Two Proteins of the cks Family M. A. Seeliger, S. E. Breward and L. S. Itzhaki* MRC Centre for Protein Engineering University Chemical Laboratory, Lensfield Road Cambridge CB2 1EW, UK
The human protein ckshs1 (cks1) is a 79 residue a/b protein with low thermodynamic and kinetic stability. Its folding mechanism was probed by mutation at sites throughout the structure. Many of the mutations caused changes in the slope of the unfolding arm of the chevron plot. The effects can be rationalised in terms of either transition-state movement or native-state “breathing”, and in either case, the magnitude of the effect enables the sequence of events in the folding reaction to be determined. Those sites that fold early exhibit a small perturbation, whilst those sites that fold late exhibit a large perturbation. The results show that cks1 folds sequential pairs of b-strands first; b1/b2 and b3/b4. Subsequently, these pairs pack against each other and onto the a-helical region to form the core. The folding process of cks1 contrasts with that of the homologue, suc1. The 113 residue suc1 has the same b-sheet core structure but, additionally, two large insertions that confer much greater thermodynamic and kinetic stability. The more extensive network of tertiary interactions in suc1 provides sufficient enthalpic gain to overcome the entropic cost of forming the core and thus tips the balance in favour of non-local interactions: the non-local, central b-strand pair, b2/b4, forms first and the periphery strands pack on later. Moreover, the greater cooperativity of the core of suc1 protects its folding from perturbation and consequently the slope of the unfolding arm of the chevron plot is much less sensitive to mutation. q 2002 Elsevier Science Ltd. All rights reserved
*Corresponding author
Keywords: protein folding; protein stability; transition state; topology; F-values
Introduction The human protein ckshs1 (cks1) is a member of the cks family of proteins that are essential for regulation of the eukaryotic cell cycle.1 There are two human homologues, ckshs1 and ckshs2, having 81% sequence identity, with ckshs1 having evolved an additional function compared with other members of the family.2 Three-dimensional domain swapping3 has been observed for several cks proteins, with the C-terminal b-strand, b4, exchanging with another molecule to form an inter-twined dimer.4,5 The process is mediated by a highly conserved “hinge loop” or “hinge” sequence and two proline residues in the hinge Abbreviations used: cks, cyclin-dependent kinase subunit; ckshs, human cks. E-mail address of the corresponding author: [email protected]
loop are the main determinants of whether the protein adopts the monomer or the dimer form.6,7 We have been using the different members of the cks family to investigate the phenomenon of domain swapping.7 – 9 Here, we describe the folding kinetics of cks1 and compare it with that of the Schizosaccharomyces pombe homologue suc1. cks1 has a much lower free energy of unfolding than suc1 (3.5 kcal mol21 versus 8 kcal mol21 at 25 8C).9 It also has a much lower kinetic stability (the rate of unfolding in water is , 1 s21 at 25 8C compared with , 1 £ 1025 s21 for suc1), and cks1 folds in a two-state manner without the population of intermediates, whilst suc1 folds via an intermediate.8 Both the intermediate and the rate-determining transition state for folding of suc1 were characterised previously using F-value analysis.10 We find that the position along the reaction coordinate of the rate-determining transition state for folding/ unfolding of cks1 is very sensitive to changes in
0022-2836/03/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved
190
Folding Mechanism of Human cks1
rationalised in terms of the structure and cooperativity of the native states of the two proteins and they illustrate nicely the enthalpy –entropy balance in protein folding.
Results and Discussion cks1 structure and mutant design cks1 has 79 residues and an a/b fold consisting of a four-stranded b-sheet capped at one end by two short a-helices (Figure 1).14 The hydrophobic core is composed of side-chains in the central strands of the b-sheet (b2 and b4) and the a-helical region. Truncation mutants were constructed that probed the formation of secondary structure elements (11 mutants for the sheet and three mutants for the helices) or the formation of the hydrophobic core (five mutants) (Table 1). The loops were probed and the b-turn or so-called hinge between strands 3 and 4 that mediates domain swapping. Analysis of equilibrium and kinetic data Figure 1. (a). Representations of the structure of cks114 on the left, with the structure of suc1 on the right for comparison.39 (b). Topologies of cks1 and of suc1. Shown in red are the insertions in suc1 compared with cks1.
solvent conditions and to mutation. These movements preclude a full and quantitative F-value analysis of the transition state structure. Further, the perturbations are so large that they preclude a “continuous F-value” analysis, an approach developed by Oliveberg and co-workers.11 Using this approach, rate constants for wild-type and mutant are compared at the same b-Tanford value and continuous F-values can then be obtained as a function of the b-Tanford value. Thus, a picture is built up of both early and later parts of the transition-state energy barrier. For many cks1 mutants this approach would require the use of rate constants at negative concentrations of urea. Instead of these approaches, we use the change upon mutation in the slope of the unfolding arm of chevron plot as a qualitative measure of structure formation during the folding process. A similar approach was used recently for the protein S612 and for the 27th Ig domain of titin.13 We compare the results obtained from the unfolding data with the F-values determined from the rates of refolding. Finally, the data are consistent with an alternative, ground state explanation for the observed perturbations, involving “breathing” of the native structure. The folding mechanism of cks1 revealed by the analysis is in striking contrast to that of its homologue, suc1. Moreover, unlike cks1, suc1 does not exhibit large perturbations in folding behaviour upon mutation. The findings can be
The following parameters were obtained from the equilibrium data (Table 1) and kinetic data (Table 2): From the equilibrium data, we obtained the urea midpoint of unfolding ([U]50%), the equilibrium m-value (meq) the free energy of unfolding ðDGD-N Þ and the change in the free energy of unfolding upon mutation ðDDGD-N Þ: Representative equilibrium denaturation curves are shown in Figure 2. Truncation of the completely buried residues in the core leads to a large destabilisation of the native state. Two of these mutants, LA24 and FL69, could not be expressed in sufficient quantities for equilibrium measurement, most likely because they are very unstable, and therefore only the kinetic data are given. From the refolding kinetics we obtained the rate of refolding in water (kf), the slope of the refolding arm of the chevron (mf), and the F-value in water calculated using the refolding rate (Ff). From the unfolding kinetics, we obtained the slope of the unfolding arm of the chevron (mu), and the rate constants for unfolding in the absence of denaturant (ku). F-values were calculated from the unfolding data for those mutants where the change in mu compared with the wild-type value was less than 5%. Finally, the free energy of unfolding was calculated using the kinetic data ðDGkin D-N Þ from the ratio of refolding and unfolding rate constants. Categorisation of mutants according to experimental kinetic parameters Representative chevron plots are shown in Figure 3, fit according to a two-state model (see Materials and Methods). Wild-type cks1 and the mutants display curvature in the unfolding arm of the chevron plots. This second-order term was
191
Folding Mechanism of Human cks1
Table 1. Location of mutated sites, interactions deleted upon mutation, and equilibrium data Mutant
Location of mutated residue—location of interactions deleted upon mutation
m (kcal mol21 M21)
[U]50% (M)
DGD-N (kcal mol21)
DDG½U50% D-N (kcal mol21)
WT IV6 YA8 SA9 KA11 FL17 EA18 VA22 LA24b VA32 SA39 LA46 VA55 VG55 ML58 PA62 PA64 HA65 LA67 FL69b RA70 RA71
b1/b2 (corea) b1/b2 b1/b2 b1/b2 b2/b4 b2/b4 b2/b1 (corea) Loop/b4 a1/a1,a2 a2/a2 a2/a1,a2 (corea) b3/b4 b3/b4 b3/b4 Hinge loop Hinge loop Hinge loop b4/b2 b4/a2,b3 (corea) b4/b2,b3 b4/b2,b3
1.01 ^ 0.04 0.91 ^ 0.05 0.93 ^ 0.03 0.95 ^ 0.04 1.07 ^ 0.06 1.19 ^ 0.02 1.30 ^ 0.02 1.11 ^ 0.03 – 1.03 ^ 0.03 1.14 ^ 0.04 0.98 ^ 0.03 1.21 ^ 0.05 1.00 ^ 0.04 1.17 ^ 0.06 1.00 ^ 0.04 1.01 ^ 0.03 1.05 ^ 0.03 0.99 ^ 0.03 – 0.86 ^ 0.03 1.05 ^ 0.03
3.85 ^ 0.03 3.47 ^ 0.06 3.33 ^ 0.04 3.54 ^ 0.03 4.32 ^ 0.04 1.26 ^ 0.01 2.69 ^ 0.06 2.43 ^ 0.02 – 2.80 ^ 0.03 3.08 ^ 0.03 1.92 ^ 0.02 2.87 ^ 0.03 1.90 ^ 0.02 3.37 ^ 0.04 5.16 ^ 0.03 3.47 ^ 0.02 3.30 ^ 0.03 1.97 ^ 0.02 – 1.25 ^ 0.03 3.22 ^ 0.02
4.14 ^ 0.07 3.58 ^ 0.08 3.80 ^ 0.06 3.80 ^ 0.07 4.65 ^ 0.08 1.35 ^ 0.02 2.89 ^ 0.04 2.61 ^ 0.05 – 3.01 ^ 0.06 3.31 ^ 0.06 2.07 ^ 0.04 3.08 ^ 0.06 2.04 ^ 0.04 3.62 ^ 0.07 5.54 ^ 0.09 3.72 ^ 0.06 3.54 ^ 0.06 2.11 ^ 0.04 – 1.34 ^ 0.04 3.46 ^ 0.06
– 0.40 ^ 0.10 0.56 ^ 0.11 0.33 ^ 0.09 20.51 ^ 0.10 2.78 ^ 0.07 1.25 ^ 0.08 1.53 ^ 0.08 – 1.13 ^ 0.09 0.83 ^ 0.09 2.07 ^ 0.08 1.06 ^ 0.09 2.10 ^ 0.08 0.52 ^ 0.10 21.40 ^ 0.11 0.41 ^ 0.10 0.59 ^ 0.09 2.02 ^ 0.08 – 2.79 ^ 0.08 0.68 ^ 0.09
All measurements were made at 10 8C. The data were fitted to a two-state model to give [U]50%, the concentration of urea at which 50% of the protein is denatured, and m, a constant of proportionality that is related to the change in solvent exposure of hydrophobic is the free energy of unfolding, calculated using the equation: side-chains upon unfolding. DG½U50% D-N DGD-N ¼ kml½U50% and DDG½U50% is the change in the free energy of unfolding upon mutation, calculated using the equation: D-N DDG½U50% ¼ kmlD½U50% D-N where D½U50% is the difference in the concentrations of urea at which 50% of the wild-type and mutant proteins are denatured. kml, the weighted average of m-value for wild-type and all mutants, is 1:07 ^ 0:02: a core indicates that the mutated residue is located in the hydrophobic core. b The yields of these mutants were too low for equilibrium measurement to be performed.
similar for wild-type and mutants. Because it is difficult to fit this term accurately, a weighted average value was obtained from the wild-type and mutants, and the chevron plots were then refit with the second-order term fixed at the average value ð0:054 ^ 0:010Þ: Seven non-exclusive types of folding behaviour were noted and the mutants were categorised accordingly. Type I mu mutant , mu wild-type: YA8, SA9, KA11, EA18, PA62 and HA65. Type II mu mutant , mu wild-type: IV6, VA22, VA32, SA39, VA55, ML58, PA64 and RA71. Type III mu mutant p mu wild-type: FL17, LA24, LA46, VG55, LA67, FL69 and RA70. Type IV High Ff (. 0.3): IV6, YA8, SA9, FL17, EA18, VA22, ML58, PA62, HA65 and RA70. Type V Low Ff (, 0.3): VA32, SA39, LA46, VA55, VG55, PA64, LA67 and RA71. Type VI DGkin D-N , eq DGD-N : YA8, SA9, KA11, EA18, LA46, ML58, PA62 eq and HA65. Type VII DGkin D-N p DGD-N : IV6, FL17, VA22, VA32, SA39, VA55, VG55, PA64, LA67, RA70 and RA71. Transition-state movement provides a picture of the sequence of folding events Here, we have used the magnitude of the movement of the position of the transition state along
the reaction coordinate, probed experimentally by the kinetic m-values, as a measure of structure formation. The rationale is as follows (Figure 4(a)): the transition-state movements observed for cks1 can be explained if the energy barrier is viewed as broad and smooth. Consequently, if a mutation is made in a part of the structure that forms late in the folding process, on the downhill side of the wild-type transition state, then the late regions of the energy barrier, but not the early regions, are destabilised. Consequently, the peak of the energy barrier is shifted towards the native state and the observed mu value decreases. Conversely, if a mutation is made in a part of the structure that forms early in the folding process, on the uphill side of the wild-type transition state, then the whole of the energy barrier is destabilised and the peak does not shift its position along the reaction coordinate. The curvature observed in the unfolding arm of the chevrons is explained also by transition-state movement, in accord with Hammond-postulate behaviour.15 At high concentrations of urea, the late part of the energy barrier is destabilised more than the early part, because the sensitivity to denaturant increases the further along the reaction coordinate the species is located.
192
Folding Mechanism of Human cks1
Table 2. Kinetic parameters for cks1 wild-type and mutants Protein Wild-type IV6 YA8 SA9 KA11 FL17 EA18 VA22 LA24 VA32 SA39 LA46 VA55 VG55 ML58 PA62 PA64 HA65 LA67 FL69 RA70 RA71
kf (s21)
mf (M21)
O FH F
ku (s21)
mu (M21)
O 1 2 FH U
21 DGkin D-N (kcal mol )
92.1 ^ 2.6 64.5 ^ 2.5 68.3 ^ 2.3 70.4 ^ 1.5 74.4 ^ 1.6 8.4 ^ 0.2 7.9 ^ 1.5 38.8 ^ 4.7 36.4 ^ 7.0 67.4 ^ 2.1 61.6 ^ 2.7 59.9 ^ 1.1 71.9 ^ 6.8 65.4 ^ 6.1 63.2 ^ 3.0 222 ^ 16 76.7 ^ 6.2 53.4 ^ 1.6 45.0 ^ 1.8 66.9 ^ 14.4 18.4 ^ 1.5 89.0 ^ 3.6
0.61 ^ 0.06 0.57 ^ 0.05 0.45 ^ 0.04 0.47 ^ 0.03 0.47 ^ 0.03 –b 0.70 ^ 0.30 0.69 ^ 0.22 –b 0.49 ^ 0.05 0.55 ^ 0.09 0.42 ^ 0.33 0.65 ^ 0.19 0.62 ^ 0.19 0.51 ^ 0.08 0.59 ^ 0.04 0.22 ^ 0.09 0.59 ^ 0.09 0.51 ^ 0.34 0.80 ^ 0.42 0.87 ^ 0.22 0.49 ^ 0.06
0.49 0.30 0.45 20.23 0.48 1.10 0.32 –c 0.15 0.27 0.12 0.13 0.09 0.41 0.35 0.25 0.51 0.20 –c 0.32 0.03
0.065 ^ 0.018 0.27 ^ 0.03 0.114 ^ 0.024 0.073 ^ 0.013 0.039 ^ 0.010 3.45 ^ 0.35 0.080 ^ 0.015 0.62 ^ 0.16 9.01 ^ 0.70 0.59 ^ 0.04 0.58 ^ 0.06 1.88 ^ 0.43 0.95 ^ 0.23 4.57 ^ 0.50 0.29 ^ 0.05 0.019 ^ 0.008 0.32 ^ 0.12 0.058 ^ 0.011 2.32 ^ 0.33 4.09 ^ 0.82 4.94 ^ 0.17 0.66 ^ 0.11
1.37 ^ 0.04 1.21 ^ 0.01 1.33 ^ 0.03 1.37 ^ 0.03 1.40 ^ 0.04 0.97 ^ 0.02 1.32 ^ 0.02 1.12 ^ 0.04 0.92 ^ 0.01 1.17 ^ 0.01 1.15 ^ 0.01 1.09 ^ 0.03 1.12 ^ 0.04 0.94 ^ 0.02 1.21 ^ 0.02 1.39 ^ 0.07 1.21 ^ 0.05 1.40 ^ 0.03 1.06 ^ 0.02 1.01 ^ 0.03 0.89 ^ 0.01 1.17 ^ 0.02
–a 0.44 0.80 0.44 –a 0.91 –a –c –a –a –a –a –a –a 0.51 –a 1.11 –a –c –a –a
4.06 ^ 0.16 3.07 ^ 0.07 3.58 ^ 0.12 3.85 ^ 0.10 4.23 ^ 0.14 0.50 ^ 0.06 2.57 ^ 0.15 2.32 ^ 0.15 0.78 ^ 0.12 2.65 ^ 0.04 2.61 ^ 0.06 1.94 ^ 0.13 2.42 ^ 0.14 1.49 ^ 0.08 3.02 ^ 0.10 5.24 ^ 0.24 3.07 ^ 0.21 3.82 ^ 0.11 1.66 ^ 0.08 1.57 ^ 0.16 0.74 ^ 0.05 2.75 ^ 0.10
2
2
All measurements were made at 10 8C. The rate constants and their respective m-values were obtained by fitting the chevron plots according to a two-state model (as described in Materials and Methods) using the equation: 2O 2O ðexpð2mf ½U 2 0:054½U2 Þ þ kH ðexpðmu ½U 2 0:054½U2 Þ kobs ¼ kH u f
DGkin D-N is the free energy of unfolding calculated from the kinetic data using the equation: H2 O H2 O DGkin =ku Þ D-N ¼ RT lnðkf
where R is the gas constant and T is the temperature (in Kelvin). a Fu was not calculated in the case of mutants for which there was a significant change in mu compared with the wild-type value. b The error was larger than the value for these mutants because the refolding arm of the chevron was very short. c F-values could not be calculated for these mutants because the protein yields were too low to enable equilibrium measurement to be made.
Ground-state versus transition-state effects An alternative explanation for the changes in mu upon mutation is movement of the ground (native)
state (Figure 4(b)). A mutant that causes a decrease in mu can do so either by shifting the peak of the transition-state barrier towards the native state or by destabilising the native state sufficiently for it
Figure 2. Equilibrium denaturation curves for wild-type and representative mutants. Measurements were made at 10 8C in 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, and with a protein concentration of 4 mM. The data were fit according to a two-state model.
193
Folding Mechanism of Human cks1
Figure 4. Two alternative explanations for the observed decrease in mu upon mutation. The energy profile of wild-type is shown with a continuous line and with a broken line for the mutant. (a) Transition-state movement. If the structure at the site of mutation is formed late in the folding reaction, only the late part of the transition-state energy barrier is destabilised and therefore the top of the energy barrier is shifted towards the native state along the reaction coordinate. (b) Ground-state movement. Mutation at a site that forms late in the folding reaction causes the native state to be destabilised, leading to local unfolding or expansion, and the native state is consequently shifted towards the transition state along the reaction.
mediate. Therefore, whether the changes in mu arise from native-state fraying or transition-state movement, the implications for the order for structure formation during the folding processes are the same; i.e. those parts of the structure at which mutation has such an effect must consolidate late in folding. Order of structure formation during folding: sites that fold early Figure 3. Chevron plots for representative mutants at sites that form (a) early, YA8 (W), SA9 (L), KA11 (A), EA18 (B), PA62 ( £ ), HA65 (V), (b) at an intermediate stage, IV6 (W), VA22 (A), VA32 (L), SA39 ( £ ), and (c) late in the folding process, LA46 (A), VG55 (L), LA67 ( £ ), FL69 (V) and RA70 (W). Wild-type (X) is shown in each plot for comparison. Measurements were made at 10 8C in conditions, after mixing in the stopped-flow apparatus, of 50 mM Tris –HCl (pH 7.5), 1 mM EDTA, and a protein concentration of 2 mM. The fits of the data according to a two-state model are shown.
to “fray”, becoming less compact in the process. There may be a continuum of expanded nativelike species or, alternatively, a discrete unfolding intermediate populated in the dead-time of the kinetic experiment. Only mutants at those sites that form late in the folding reaction will destabilise the native state relative to the native-like inter-
Mutation of only six residues has no effect on mu (type I): YA8 (b1), SA9 (b1), KA11 (b1), EA18 (b2), PA62 (hinge) and HA65 (hinge) (Figure 3(a); Table 2). They are the only mutants that unfold at a rate similar to that of the wild-type. All except KA11 refold significantly more slowly than wildtype (type IV) and the F-values calculated from either the refolding or the unfolding rates are high for these mutants (between 0.3 and 1). The mutant PA62 also falls into this category (Figure 3(a)). Although it displays pronounced rollover in the refolding arm, as discussed in more detail later (see section on The stabilised mutant PA62 changes the refolding arm of the chevron), a F-value can be calculated from the data measured above 1 M urea where the chevron plot runs parallel with that of the wild-type. The F-value obtained in this way is high (0.4) and it is in good agreement with the F-value obtained using the unfolding data (0.5). Finally, the mutant KA11 behaves like PA62—it is significantly more stable than wild-type and
194
displays slight rollover in the refolding arm. As a result of the rollover, the rate of refolding in water is smaller than that of wild-type, and this gives an anomalous F-value of less than zero, but above 1 M urea the refolding rate is greater than wildtype.
Folding Mechanism of Human cks1
Folding and unfolding data give consistent results The sequence of events revealed by the refolding data is generally in good agreement with that revealed by the unfolding data. Sites displaying low F-values also display low values of mu, while residues with high F-values have values of mu that are similar to that of the wild-type.
Sites that fold late The majority of the cks1 mutations change mu significantly compared with the wild-type value, suggesting that the sites fold late and unfold early (Figure 3(c)). This observation is consistent with the low b-Tanford value (b‡) of the wild-type protein in water of 0.27 (calculated as 1 2 ðmu =meq Þ rather than mf =meq ; because mu could be determined with greater accuracy than mf), indicating a very expanded transition state. Also, the majority of the mutants unfold faster than the wild-type over the whole range of concentrations of denaturant sampled, again, indicating that there is weak structure formation in the transition state for folding. The mutants with the lowest values of mu (type III) are FL17 (b2), LA24 (loop/ core), LA46 (loop/core), VG55 (b3), LA67 (b4), FL69 (b4/core) and RA70 (b4). Finally, the majority of the mutants probed in cks1 refold at a rate similar to that of the wild-type (type V), again consistent with only weak structure formation in the transition state. The F-values calculated from the refolding rates for these mutants are all less than 0.3. To summarise, there is reasonable agreement between type III and type V mutants, with both categories, indicating regions of the protein that form late in the folding reaction. There are two exceptions; FL17 and RA70 are mutants that refold significantly more slowly than wild-type but that exhibit very low mu values (Figure 3(c); Table 2). However, the low refolding rates of these mutants may not indicate early structure formation at these sites, but rather they may be an artefact due to off-pathway aggregation during refolding. This behaviour was observed for the wild-type at high concentrations of protein,9 and the two mutants were observed to increase greatly this aggregation propensity.
Partly structured sites A few mutations have a small effect on mu (type II): IV6 (b1), and ML58 (b3) and PA64 (hinge) (Figure 3(b); Table 2). The F-values, calculated from the refolding rates, are between 0.4 and 0.5 for these mutants (classed as type IV). Mutations VA22, VA32, SA39, VA55 and RA71 have a slightly greater effect on mu, indicating that these sites form somewhat later in the folding reaction. Consistent with this, the F-values, calculated from the refolding rates, are between 0 and 0.3 for these mutants (type V).
Increase in equilibrium m-values upon mutation indicates ground-state movement One way to assess whether the perturbations observed in the unfolding kinetics are due to ground-state or to transition-state movement is to look for changes upon mutation in the equilibrium m-values and whether any changes correlate with changes in the kinetic m-values. There is no correlation between mu and meq, and there is a weak correlation between mf and meq, The increase in mf with increasing meq suggests that the denatured, ground state for refolding becomes expanded upon mutation at certain sites. However, the absence of a correlation between mu and meq suggests that the changes in mu upon mutation are due to transition-state movement rather than native, ground-state movements. Alternatively, it is possible that native-state movement is apparent only at the very high concentrations of denaturant used for determining the unfolding kinetics, when the late part of the reaction coordinate is already highly destabilised, but that it is not apparent at the lower concentrations of denaturant around the transition midpoint where meq is measured. Discrepancies between equilibrium and kinetic data highlight the changes in the free-energy barrier over the range of concentrations of denaturant In a two-state system, the ratio of the folding and unfolding rate constants in water can be used to calculate the overall free energy of unfolding. For cks1 wild-type and a number of mutants (type VI), there is good agreement between this value measured (DGkin D-N ; given in Table 2) and the value eq by equilibrium denaturation (DGD-N given in Table 1). However, for many mutants (type VII: IV6, FL17, VA22, VA32, SA39, VA55, VG55, PA64, LA67, RA70 and RA71), DGkin D-N is significantly eq lower than DGD-N : The curvature used to fit the unfolding arm of the chevron implies a smooth movement of the transition state with changing concentration of urea. The non-coincidence between the kinetic and equilibrium data suggests that this extrapolation of the kinetic unfolding data to water is not correct for this subset of mutants. The mutants are all located in parts of the structure that form late, as judged by the low mu values. It is possible that only at high concentrations of denaturant, when the late part of the energy barrier is already very destabilised relative
195
Folding Mechanism of Human cks1
Comparison with suc1: structural details The sequence of events in the folding reaction, as revealed by the m-value and F-value analysis, is also different in the two proteins. The F-value analysis of suc1 revealed that the central strands of the sheet, b2 and b4, are the most structured in the transition state (F ¼ 0.6– 0.9) whilst the outer strands, b1 and b3, and the loops, are less structured (F ¼ 0 – 0.3). Thus, suc1 folds from the centre of the hydrophobic core outwards. For cks1, in contrast, the F-values are highest for the end of b1/ start of b2 and the end of b3/hinge/start of b4 (Figure 5). Thus, strand pairing between adjacent b-strands in the sequence, i.e. b1/b2 and b3/b4, occurs first in the folding of cks1. Comparison with other proteins of low stability
Figure 5. Location in the structure of sites that fold early (blue) and late (red) in the folding reaction of cks1.
to the early part, can these mutations cause a shift in position of the top of the transition-state energy barrier. Thus, the unfolding arm could be kinked at intermediate concentrations of denaturant, rather than being curved smoothly over the entire range of concentrations. The observed eq discrepancy between DGkin D-N and DGD-N can be explained in terms of the ground-state effects. For these mutants, the ground state for the unfolding reaction of the mutants at low concentrations of urea may be as compact as the wild-type native state, and it may become expanded only at very high concentrations of urea.
Comparison of cks1 with suc1: the global characteristics of the energy barrier for folding The perturbations in the folding behaviour of cks1 upon mutation preclude a comprehensive and quantitative F-value analysis. However, a number of features emerge from the qualitative analysis of the data described above, and it is particularly revealing to compare cks1 with its homologue suc1, for which full F-value analysis has been possible.10 The rate-determining transition state for cks1 is less compact than that of suc1: b‡ for cks1 is 0.27 compared with 0.60 for suc1. A folding intermediate was analysed for suc1 using F-values and this species is reasonably compact (b-Tanford is 0.45). The position of the transition state for cks1 is much more susceptible to perturbation by mutation than that for suc1, although the sensitivity to denaturant, as reflected in the curvature of the unfolding arm, appears similar in the two proteins.
How do the transition-state movements compare with those observed for other proteins? Can the explanation for the behaviour of cks1 lie, in part, in its very low thermodynamic and/or kinetic stability ? With 156 residues, p16 is a much larger protein than cks1, but the free energies of unfolding are similar (3.1 kcal mol21 for p16 compared with 3.5 kcal mol21 for cks1 at 25 8C) as are their kinetic stabilities (ku in water is , 1 s21 for both proteins at 25 8C).16 However, the transition state for folding/unfolding of p16 changes very little upon mutation. The reason for this becomes apparent when the extent of structure in the transition state is considered. The transition state is very native-like: b‡ is 0.9 in water and the F-values for the majority of mutants are close to 1.17 Therefore, again viewing the transition state as a broad barrier, most mutations cause the whole barrier to be destabilised and so there is no overall shift in the top of the barrier. In contrast, the transition state of cks1 is much less structured and therefore the majority of mutations destabilise only the later part of the transition-state barrier, causing a shift in the top of the barrier. Thus, the sensitivity of the transition state is likely to be influenced by a number of factors that include the stability of the native state and whether the rate-determining step is early or late in the folding process (i.e. how native-like it is). Viewed as a ground-state effect, again for a protein such as p16, with a transition state lying very close to the native state, there is much less potential for fraying of the native state. Weakening the cooperativity of the core leads to a change in folding mechanism In suc1, the non-sequential pair of strands, b2 and b4, that constitute the centre of the b-sheet and of the core of the protein, form first. The residues with highest F-values are located in b2 and b4, and residues that interact with these two strands. In cks1, the highest F-values (and mu values) are shifted away from the central strands to the peripheral strands. Strand pairs that are
196
tandem in sequence form together (i.e. b1 with b2 and b3 with b4). Thus, local interactions initiate folding in cks1 and non-local ones in suc1. In terms of sequence separation, it is not more costly to bring together tandem pairs in suc1 than it is in cks1. In fact, the additional sequence present in suc1 is an a-helix located at the N terminus and a loop between b2 and b3. Consequently, the entropic cost of bringing together b2 and b4 in suc1 should be greater than in cks1. The reason why these are the first to form in suc1, despite the greater entropic cost, may be due to the greater enthalpic gain of forming this pair in suc1 compared with cks1. The b2/b4 strand pair is stabilised by the extensive network of interactions that comprises the hydrophobic core of suc1. In cks1, the core is much less extensive. It consists mainly of residues in the b-sheet and lacks the further packing interactions between the end of the b-sheet and the a-helical region that are present in suc1. Thus, the four strands are balanced more equally in terms of enthalpy, so the closest strands form first. Does stabilisation of the native state of cks1 induce suc1-like folding behaviour? The stability of the native state of cks1 can be increased in two ways, first by mutation (PA62 is 1.4 kcal mol21 more stable than wild-type), and second, by the addition of solutes. Addition of 500 mM sodium phosphate increases the stability of wild-type by 1.9 kcal mol21 (data not shown). We examined the effect of a mutation that causes a large change in mu (RA70) on the unfolding kinetics of these two stabilised proteins. In spite of the additional stability, conferred either by mutation or by addition of solute, the mutation RA70 still resulted in a large change in mu. Thus, the susceptibility of the folding of cks1 to perturbation does not arise from its overall low thermodynamic stability per se. Rather it is the weak cooperativity of the cks1 core that causes its vulnerability and results in a change in the folding sequence compared with suc1. The stabilised mutant PA62 changes the refolding arm of the chevron As stated previously, the mutant PA62 is stabilised by 1.4 kcal mol21 relative to wild-type and its chevron exhibits rollover in the refolding arm (see Figure 3(a)). Again, the susceptibility of the folding behaviour of cks1 is apparent. In contrast to the effect that a destabilising mutation has on the unfolding arm of the chevron if it is in a region of the structure that forms late in the folding reaction, a stabilising mutation in a region that forms early perturbs the refolding arm of the chevron. In regions that form late, a destabilising mutation may allow an unfolding intermediate to be populated; and in regions that form early, a stabilising mutation may allow a refolding intermediate to be
Folding Mechanism of Human cks1
populated. Thus, the rollover observed for PA62 can be interpreted as stabilisation of a refolding intermediate before the transition state. The stabilised mutant KA11 shows behaviour similar to that of PA62.
Conclusions The entropy –enthalpy balance in protein folding appears to be resolved by forming the best contacts between residues that are as close in sequence as possible,18 – 21 and this basic principle has enabled the prediction of folding mechanisms from the native-state topology22 – 24 Less predictable is the case of highly symmetrical structures, for example protein G and protein L, and the specific distribution of interaction energies throughout the protein can then tip the balance in favour of one alternative folding route rather than another.25 – 27 One type of study that illustrates this principle of entropy versus enthalpy, or topological constraints versus energetic factors, is the folding of circular permutants where the chain connectivity is changed relative to the wild-type, whilst the strength of the interactions is unaltered.28 – 35 If the chain is cut in the part of the structure that folds early, then it becomes energetically unfavourable to fold this part first in the permutant and this can be manifest in a shift in the transition-state structure. A related example is the folding of the dimeric coiled-coil protein GCN4 compared with that of a single-chain monomeric variant that was cross-linked at one end.36 However, some transition-state structures appear to be much more resilient to these gross perturbations in topology, such as that of chymotrypsin inhibitor 2, which does not change upon permutation.33 The folding mechanisms of suc1 and cks1 illustrate the entropy– enthalpy balance. Although it is difficult to obtain a quantitative description of the folding mechanism of cks1 without a complete F-value analysis, the available data show clearly that the order of structure formation is different for cks1 compared with the homologue suc1. This is somewhat surprising, since cks1 has the same b-sheet core structure, the centre of which was shown previously to initiate folding in suc1. Moreover, the level of sequence identity is very high. However, cks1 is much smaller than suc1 (79 residues compared with 113). It lacks 20 amino acid residues at the N terminus, ten of which form an a-helix, and a nine residue unstructured loop located between b2 and b3. Thus, there is a greater entropic cost of forming the b2/b4 strand pair in suc1 compared with cks1, yet this region forms early in suc1 and late in cks1. In contrast, the structure of cks1 is less cooperative and the entropy/ enthalpy balance tips the folding route in favour of sequential rather than non-sequential strand pairing. Remarkably, even when the folding nucleus of suc1 is drastically disfavoured entropically, as occurs in the domain-swapped dimer
197
Folding Mechanism of Human cks1
where the b2/b4 pairing is made inter-molecularly, the folding mechanism is not altered.37 The folding nucleus does not shift, it remains in the same part of the structure but it recruits more interactions and thus a greater enthalpic gain in order to balance the greater entropic penalty of forming the dimer. The reason for this tolerance in suc1 may be that there is a much greater enthalpic gain in forming the core of suc1 compared with cks1. With an increasing number of protein engineering studies of folding now available, it is becoming clear that small movements of the free-energy barrier for folding/unfolding with changes in concentration of denaturant and mutation are normal rather than being exceptions.20 However, cks1 lies at the extreme end of the spectrum of this type of behaviour. Why is this protein so sensitive whilst many others are quite robust in their response to mutation? Two factors appear to contribute. The first factor is reflected by the low thermodynamic stability of cks1 and the second factor is that the rate-determining step in the folding of cks1 is very early in the reaction. A comparison with two other proteins illustrates these contributions. The first is suc1, a homologue of cks1 that has high level of sequence identity but is a much larger protein. The much greater stability conferred by the additional structural elements protects it from large perturbation upon mutation. The second protein is p16, which has a stability similar to that of cks1 but its transition state is much more nativelike.16 Again, the consequence is little or no perturbation upon mutation.17 The kinetic behaviour of wild-type and mutants of cks1 can be rationalised in terms of movement of the top of the broad energy barrier for folding.12 An alternative explanation is that the mutations cause expansion of the ground (native) state, and there is evidence that the denatured state, the ground state in the refolding direction, is susceptible to mutation. In the case of cks1, this expansion of the native state is described better by the term breathing rather than fraying, since it is the centre of its structure, rather than the outer b-strands, that are expanding.
Materials and Methods Mutagenesis and protein expression and purification Site-directed mutagenesis was performed using the QuikChange kit (Stratagene) and mutant plasmids were identified by sequencing (Oswell, University of Southampton, UK). Mutant proteins were expressed and purified as described for the wild-type.9 Protein identity and purity were confirmed by mass spectrometry.
were performed at 10 8C. The low temperature was used because the rates of unfolding (described below) at 25 8C were too fast to be measured with accuracy. The data were fit to a two-state transition with the fluorescence of the native and denatured states dependent on the concentration of denaturant. Data fitting was carried out using the non-linear least-squares algorithm provided with the program Kaleidagraph (Abelbeck Software). The two-state fit gives an m-value (a constant of proportionality that is related to the change in solvent exposure of hydrophobic side-chains upon unfolding) and [U]50% (the concentration of urea at which 50% of the ½U50% protein is denatured).38 Thus, DDGD-N ¼ kmlD½U50% ; ½U50% where D½U50% and DDGD-N are the difference in the concentrations of urea at which 50% of the wild-type and mutant proteins are denatured and the change in the free energy of unfolding upon mutation, respectively. kml; the weighted average value of m obtained for wildtype and all the mutants, was 1:07 ^ 0:02: Kinetic analysis Kinetic experiments were performed using an Applied Photophysics SX-17MV stopped-flow fluorescence-detected stopped-flow instrument at 10 8C. The excitation wavelength was 280 nm and fluorescence was collected using a 325 nm cut-off filter. Unfolding was initiated by mixing one volume of protein in 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, with ten volumes of urea in 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, to give final conditions of 50 mM Tris – HCl (pH 7.5), 1 mM EDTA. The concentration of protein after mixing was 2 mM. Under strong unfolding conditions outside the unfolding transition region, the kinetic data could be fit to a single-exponential function. The urea-dependence of the rate constant of unfolding shows slight downward curvature over the wide range of concentrations of urea according to Hammond behaviour and the data were fit to the second-order polynomial equation: H2 O ln k½U þ mu þ mpu ½U2 u ¼ ln ku
where ln ku½U and ln kuH2 O are the rate constants in a concentration of urea, [U], and in water, respectively, mu is the kinetic m-value and mpu is the second-order term. In order to minimise the error, a weighted average value of mpu (20.054 ^ 0.010) was calculated using the wild-type and all the mutants. This value was then used to refit the data to a second-order polynomial with fixed curvature. Refolding was initiated by mixing acid-denatured protein in 30 mM HCl with an equal volume of 100 mM Tris – HCl buffer (pH 8.1), 2 mM EDTA, to give final conditions of 50 mM Tris –HCl (pH 7.5), 1 mM EDTA. The concentration of protein after mixing was 2 mM. The data were fit to the sum of two exponential phases, representing a fast-refolding phase accounting for most of the amplitude and a slow-refolding phases involving proline isomerisation, as reported.9 The fast-refolding rate constant, kf, is used for the F-value analysis. The kinetic data in the form of chevron plots were fitted to a two-state model using the equation: 2O 2O kobs ¼ kH ðexpð2mf ½U 2 0:054½U2 Þ þ kH ðexpðmu ½U u f
Equilibrium denaturation Urea-induced equilibrium denaturation was monitored by fluorescence spectroscopy as described.9 The protein concentration was 4 mM, the buffer was 50 mM Tris – HCl (pH 7.5), 1 mM EDTA, and all experiments
2 0:054½U2 Þ using the non-linear, least-squares algorithm provided O O with the program Kaleidagraph. kH and kH are, f u respectively, the rate constants for refolding and 2
2
198
Folding Mechanism of Human cks1
unfolding in water. mf and mu are, respectively, the kinetic refolding and unfolding m-values in water. The average value of the curvature, 0.054, was used. The change upon mutation in the free energy of the transition state relative to the native state, DDG‡-F ; is calculated 2O using the equation: DDG‡-N ¼ 2RTðln kH 2 ln k0u H2 O Þ; u O 0 O where kH and kH are the rate constants of unfolding u u of wild-type and the mutant respectively. The change upon mutation in the free energy of the transition state relative to the denatured state, DDGU-‡ ; is calculated 2O using the equation: DDGU-‡ ¼ RTðln kH 2 ln k0f H2 O Þ; f HO H 0 O where kf and kf are the rate constants of refolding of wild-type and the mutant respectively. The ratio between DDG‡-N and DDGD-N is defined as FU. The ratio of DDGD-‡ and DDGD-N gives the F-value in the refolding direction, FF, where FF ¼ 1FU. A FF-value of 0 means that the interactions deleted upon mutation are formed as weakly in the transition state as it is in the denatured state. A F-value of 1 means that the interactions are as energetically well formed in the transition state as in the native structure. 2
2
9.
2
10.
2
11.
12.
13.
14.
Acknowledgements This work was supported by the Medical Research Council of the UK (MRC). L.S.I. was supported by a Career Development Award from the MRC. M.A.S. was supported by an External Research Studentship from Trinity College, Cambridge, UK.
15. 16.
17.
References 1. Hayles, J., Aves, S. & Nurse, P. (1986). suc1 is an essential gene involved in both the cell cycle and growth in fission yeast. EMBO J. 5, 3373– 3379. 2. Ganoth, D., Bornstein, G., Ko, T. K., Larsen, B., Tyers, M., Pagano, M. & Hershko, A. (2001). The cell-cycle regulatory protein Cks1 is required for SCFSkp2mediated ubiquitinylation of p27. Nature Cell Biol. 3, 321– 324. 3. Bennett, M. J., Choe, S. & Eisenberg, D. (1994). Domain swapping: entangling alliances between proteins. Proc. Natl Acad. Sci. USA, 91, 3127– 3131. 4. Bourne, Y., Arvai, A. S., Bernstein, S. L., Watson, M. H., Reed, S. I., Endicott, J. A. et al. (1995). Crystal structure of the cell cycle-regulatory protein suc1 reveals a b-hinge conformational switch. Proc. Natl Acad. Sci. USA, 92, 10232– 10236. 5. Parge, H. E., Arvai, A. S., Murtari, D. J., Reed, S. I. & Tainer, J. A. (1993). Human CksHs2 atomic structure: a role for its hexameric assembly in cell cycle control. Science, 262, 387– 394. 6. Rousseau, F., Schymkowitz, J. W. H., Wilkinson, H. R. & Itzhaki, L. S. (2001). Three-dimensional domain swapping in p13suc1 occurs in the denatured state and is controlled by conserved proline residues. Proc. Natl Acad. Sci. USA, 98, 5596 –5601. 7. Schymkowitz, J. W. H., Rosseau, F., Wilkinson, H. R., Friedler, A. & Itzhaki, L. S. (2001). Observation of signal transduction in three-dimensional domain swapping. Nature Struct. Biol. 8, 888– 892. 8. Rousseau, F., Schymkowitz, J. W. H., Sanchez del Pino, M. & Itzhaki, L. S. (1998). Stability and folding
18. 19.
20.
21.
22.
23.
24.
25.
of the cell cycle regulatory protein, p13suc1. J. Mol. Biol. 284, 503– 519. Seeliger, M. A., Schymkowitz, J. W. H., Rousseau, F., Wilkinson, H. R. & Itzhaki, L. S. (2002). Folding and association of the human cell cycle regulatory proteins ckshs1 and ckshs2. Biochemistry, 41, 1202– 1210. Schymkowitz, J. W. H., Rousseau, F., Irvine, L. R. & Itzhaki, L. S. (2000). The folding pathway of the cellcycle regulatory protein p13suc1: clues for the mechanism of domain swapping. Struct. Fold. Des. 8, 89 –100. Ternstrom, T., Mayor, U., Akke, M. & Oliveberg, M. (1999). From snapshot to movie: F value analysis of protein folding transition states taken one step further. Proc. Natl Acad. Sci. USA, 96, 14854– 14859. Otzen, D. E. & Oliveberg, M. (2002). Conformational plasticity in folding of the split b –a – b protein S6: evidence for burst-phase disruption of the native state. J. Mol. Biol. 317, 613– 627. Fowler, S. B. & Clarke, J. (2001). Mapping the folding pathway of an immunoglobulin domain: structural detail from phi value analysis and movement of the transition state. Structure, 9, 355– 366. Arvai, A. S., Bourne, Y., Hickey, M. J. & Tainer, J. A. (1995). Crystal structure of the human cell cycle protein CksHs1: single domain fold with similarity to kinase N-lobe domain. J. Mol. Biol. 249, 835– 842. Hammond, G. S. (1953). A correlation of reaction rates. J. Am. Chem. Soc. 77, 334– 338. Tang, K. S., Guralnick, B. J., Wang, W. K., Fersht, A. R. & Itzhaki, L. S. (1999). Stability and folding of the tumour suppressor protein p16. J. Mol. Biol. 285, 1869– 1886. Tang, K. S., Fersht, A. R. & Itzhaki, L. S. (2002). Sequential unfolding of ankyrin repeats in tumour suppressor p16. Structure. In press. Dinner, A. R. & Karplus, M. (2001). The roles of stability and contact order in determining proten folding rates. Nature Struct. Biol. 8, 21 –22. Fersht, A. R. (2000). Transition-state structure as a unifying basis in protein-folding mechanisms: contact order, chain topology, stability and the extended nucleus mechanism. Proc. Natl Acad. Sci. USA, 97, 1525– 1529. Oliveberg, M. (2001). Characterisation of the transition state for protein folding: towards a new level of mechanistic detail in protein engineering analysis. Curr. Opin. Struct. Biol. 11, 94 – 100. Plaxco, K. W., Simons, K. T., Ruczhinski, I. & Baker, D. (2000). Topology, stability, sequence, and length: defining the determinants of two-state protein folding kinetics. Biochemistry, 39, 11177– 11183. Alm, R. & Baker, D. (1999). Prediction of proteinfolding mechanisms from free-energy landscapes derived from native structures. Proc. Natl Acad. Sci. USA, 96, 11305– 11310. Galzitskaya, O. & Finkelstein, A. (1999). A theoretical search for folding/unfolding nuclei in three-dimensional protein structures. Proc. Natl Acad. Sci. USA, 96, 11299– 11304. Munoz, V. & Eaton, W. (1999). A simple model for calculating the kinetics of protein folding from three-dimensional structures. Proc. Natl Acad. Sci. USA, 96, 11311 – 11316. Kim, D., Fisher, C. & Baker, D. (2000). A breakdown of symmetry in the folding transition state of protein L. J. Mol. Biol. 298, 971– 984.
Folding Mechanism of Human cks1
26. McCallister, E. L., Alm, E. & Baker, D. (2000). Critical role of beta-hairpin formation in protein G folding. Nature Struct. Biol. 7, 669– 673. 27. Nauli, S., Kuhlman, B. & Baker, D. (2001). Computerbased redesign of a protein folding pathway. Nature Struct. Biol. 8, 602– 605. 28. Clementi, C., Jennings, P. A. & Onuchic, J. N. (2001). Prediction of folding mechanism for circularpermuted proteins. J. Mol. Biol. 311, 879– 890. 29. Grantcharova, V., Riddle, D., Santiago, J., Alm, E., Tsai, J. & Baker, D. (1999). Experiment and theory highlight role of native state topology in SH3 folding. Nature Struct. Biol. 6, 1016– 1024. 30. Grantcharova, V. P. & Baker, D. (2001). Circularization changes the folding transition state of the src Sh3 domain. J. Mol. Biol. 306, 555– 563. 31. Lindberg, M. O., Tangrot, J., Otzen, D. E., Dolgikh, D. A., Finkelstein, A. V. & Oliveberg, M. (2001). Folding of circular permutants with decreased contact order: general trend balanced by protein stability. J. Mol. Biol. 314, 891– 900. 32. Miller, E. J., Fischer, K. F. & Marqusee, S. (2002). Experimental evaluation of topological parameters determining protein-folding rates. Proc. Natl Acad. Sci. USA, 99, 10359– 10363. 33. Otzen, D. & Fersht, A. R. (1998). Folding of circular and permuted chymotrypsin inhibitor 2: retention of the folding nucleus. Biochemistry, 37, 8139– 8146.
199
34. Viguera, A., Blanco, F. & Serrano, L. (1995). The order of secondary structure elements does not determine the structure of a protein but does affect its folding kinetics. J. Mol. Biol. 247, 670– 681. 35. Viguera, A. R., Serrano, L. & Wilmanns, M. (1996). Different folding transition-states may result in the same native structure. Nature Struct. Biol. 3, 874– 880. 36. Moran, L. B., Schneider, J. P., Kentsis, A., Reddy, G. A. & Sosnick, T. R. (1999). Transition state heteroeneity in GCN4 coiled coil folding studied by using multisite mutations and crosslinking. Proc. Natl Acad. Sci. USA, 96, 10699– 10704. 37. Rousseau, F., Schymkowitz, J. W. H., Wilkinson, H. R. & Itzhaki, L. S. The structure of the transition state for folding of domain-swapped dimeric p13suc1. Structure, 10, 649– 657. 38. Clarke, J. & Fersht, A. R. (1993). Engineering disulfide bonds as probes of the folding pathway of barnase—increasing the stability of proteins against the rate of denaturation. Biochemistry, 32, 43322 – 44329. 39. Endicott, J. A., Noble, M. E., Garman, E. F., Brown, N., Ramussen, R., Nurse, P. & Johnson, L. N. (1995). The crystal structure of p13suc1, a p34cdc2-interacting cell cycle control protein. EMBO J. 14, 1004 –1014.
Edited by C. R. Matthews (Received 14 August 2002; received in revised form 18 October 2002; accepted 18 October 2002)