Thermodynamic Characterization of the Coupled Folding and Association of Heterodimeric Coiled Coils (Leucine Zippers)

J. Mol. Biol. (1996) 263, 344–358

Thermodynamic Characterization of the Coupled Folding and Association of Heterodimeric Coiled Coils (Leucine Zippers) Ilian Jelesarov* and Hans Rudolf Bosshard Biochemisches Institut der Universita¨t Zu¨rich Winterthurerstrasse 190 CH-8057 Zu¨rich, Switzerland

Folding thermodynamics of nine heterodimeric, parallel coiled coils were studied by isothermal titration calorimetry (ITC) and thermal unfolding circular dichroism measurements. The heterodimers were composed of an acidic and a basic 30-residue peptide, which when in isolation were monomeric and essentially unstructured. The reaction followed a two-state mechanism indicating that folding and association were coupled. DHfold , DSfold and DCP normalized per mol of residue were of the same magnitude as for monomeric globular proteins, hence the energetics of folding and association of the heterodimeric coiled coils was balanced similarly to the folding of a single polypeptide chain. Cavity creating Leu/Ala substitutions revealed strong and position-dependent energetic coupling between leucine residues in the hydrophobic core of the coiled coil. DGunfold (equivalent to −DGfold in the two-state reaction) was determined from thermal unfolding. Global stability curves were calculated according to the Gibbs-Helmholtz equation and using the combined free energy data from ITC and thermal unfolding. Maximum stabilities were between 15 and 37°C and cold denaturation could be demonstrated by direct calorimetry. The stability curves were based on free energies of folding measured between 10 and 85°C and under identical solvent conditions. This represents a novel experimental approach which circumvents the use of varying solvent conditions as is typically required to measure protein stability curves. Discrepancies were noticed between van’t Hoff enthalpies deduced from thermal unfolding and measured by direct calorimetry. The discrepancies are thought to be due to residual ordered structure in the denatured single chains around room temperature but not near the transition midpoint temperature Tm . This demonstrates that over an extended temperature range the assumption of a common denatured state implicit in the van’t Hoff analysis may not always be valid. 7 1996 Academic Press Limited

*Corresponding author

Keywords: isothermal titration calorimetry; protein stability curve; thermal unfolding; folding enthalpy; co-operativity

Abbreviations used: ANS, 1-anilino-8-naphthalenesulphonate; CD, circular dichroism; ITC, isothermal titration calorimetry; DEfold (E = G, H, S), free energy, enthalpy and entropy of folding/association determined by ITC; DEunfold (E = G, H, S), free energy, enthalpy and entropy of unfolding/dissociation determined from temperature unfolding curves; DCp (ITC), DCp (fit), DCp (CD), heat capacity change determined by ITC, by a global fit of all DG(T) values measured in the temperature range 10 to 85°C to the Gibbs-Helmhotz equation, and by van’t Hoff analysis of the temperature unfolding curve near Tm , respectively; Tm , transition midpoint temperature at which the fraction of unfolded coiled coil equals 0.5. 0022–2836/96/420344–15 $25.00/0

Introduction The stability of the native fold of a protein results from a delicate balance between all the interatomic interactions in the folded and the unfolded state. From the point of view of equilibrium thermodynamics, a precisely tuned interplay between enthalpic and entropic factors determines the stability of the native state within a relatively narrow range of temperature, pH and solvent composition. A large number of monomeric proteins of mainly globular shape has been thermodynamically characterized. This has allowed the definition of general thermodynamic principles of confor7 1996 Academic Press Limited

345

Thermodynamics of Leucine Zipper Folding

mational stability and to link macroscopic thermodynamic quantities with molecular details of protein structure (reviews by Murphy & Freire, 1992; Makhatadze & Privalov, 1995). When expressed per mole of residue or per unit of mass, DG, DH and DS of unfolding and the temperature dependence of these quantities are similar for a large number of small, monomeric and compact (globular) proteins or protein domains that fold as a cooperative unit (Makhatadze & Privalov, 1995). Larger differences in thermodynamic folding/unfolding parameters were observed for rod-shaped (collagen; Privalov, 1982), very small (neurotoxins; Privalov & Gill, 1988), loosely packed (histones; Tiktopulo et al., 1982), or conformationally very flexible proteins (barstar; Wintrode et al., 1995). Very few studies deal with the thermodynamics of folding and assembly of dimeric proteins. In principle, the thermodynamics of folding of a dimeric (or oligomeric) protein may be divided into the folding of the individual monomer and the association of the pre-folded monomers. Often, however, folding and assembly are tightly coupled processes, that is, the individual subunits in isolation are unfolded, folding being concomitant to association. Coupled folding and assembly has been observed for dimeric regulatory proteins (Bowie & Sauer, 1989; De Francesco et al., 1991; Johnson et al., 1992; Mann et al., 1993; Thompson et al., 1993; Eftink et al., 1994; Steif et al., 1995). Here we report on the thermodynamics of the coupled association and folding of a set of engineered heterodimeric coiled coils or leucine zippers. Thermodynamic parameters and protein stability curves are presented for the large temperature range of 10 to 85°C. This was accomplished by a combination of isothermal titration calorimetry (ITC) and thermal unfolding experiments. To our knowledge it is the first study of a cooperative conformational transition in a heterodimeric protein molecule, represented here by a leucine zipper, performed under identical solvent conditions of pH and ionic strength. This approach is unique to a heterodimeric system and cannot be applied to the study of a monomeric or homodimeric protein for which DGunfold in the temperature region of maximum protein stability is obtained conventionally by linear extrapolation of denaturant unfolding curves (Pace, 1986), and near the transition temperature, Tm , by variation of pH or another solvent parameter. Two-stranded coiled coils are dimeric folding motifs found in many proteins of widely different function (Hurst, 1995). The structure is based on the sequential repetition of a seven-residue motif, (abcdefg)n , in which Leu, Val or Ile are frequent in positions a and d. In a parallel dimeric coiled coil with the two chains in register, alternate layers of residues in d and e and g and a positions interact side-by-side to form a hydrophobic interface (reviews by Alber, 1992; Ellenberger, 1994; Hurst, 1995). In many naturally occurring leucine zippers, which form the dimerization domain of bZIP

transcription factors (Landschulz et al., 1988), a pair of Asn residues at a central a position interrupts the hydrophobic core. The Asn pair partially destabilizes the coiled coil and constrains the helices to be parallel and in register (O’Shea et al., 1991; Lumb & Kim, 1995; Betz et al., 1995; Wendt et al., 1995; Junius et al., 1995, 1996). Equally charged residues at g and e positions lead to interhelical ionic repulsion in the homodimeric state, shifting the monomer/coiled coil equilibrium to the side of the unstructured monomer. This feature has been exploited to design heterodimeric coiled coils composed of an acidic chain (Glu in positions g and e) and a basic chain (Lys in positions g and e). The destabilizing repulsive charges are neutralized in the heterodimer, which may be orders of magnitude more stable than the homodimers (Graddis et al., 1993; O’Shea et al., 1993; Myszka & Chaiken, 1994; Kohn et al., 1995). We have used this design principle in synthesizing three acidic peptides and three basic peptides containing four heptad repeats that differ only by Leu to Ala substitutions at central d positions. The peptides were combined to form nine different heterodimers. Because each peptide in isolation was unfolded, the association-coupled folding reaction could be studied by mixing two oppositely charged peptide chains in the isothermal titration calorimeter (Wiseman et al., 1989; Freire et al., 1990). In this way, DGfold , DHfold and DSfold could be measured at temperatures in the region of maximum stability of the heterodimer, i.e. far below the unfolding transition temperature region. Strong energetic coupling between residues of the hydrophobic core was deduced from thermodynamic cycles (Horovitz & Fersht, 1990) performed on heterodimers containing cavity-creating double Leu/Ala substitutions. In a second set of experiments, thermal unfolding of the heterodimers was followed by circular dichroism (CD) to obtain DGunfold in the region of Tm . Because of the dimeric structure, Tm of the coiled coil was concentrationdependent. The folding/unfolding transition followed a simple two-state mechanism and, therefore, DGunfold was equivalent to −DGfold . The data from ITC obtained at temperatures between 10 and 40°C and the data from thermal unfolding in the range of 50 to 85°C could be combined to obtain stability curves (Becktel & Schellman, 1987).

Results Design of heterodimeric coiled coils The coiled coils were composed of the acidic peptide Ac-EYQALEKEVAQ(L/A)EAENQA(L/A)EKEVAQLEHEG-amide and the basic peptide Ac-EYQALKKKVAQ(L/A)KAKNQA(L/A)KKKVAQLKHKG-amide (residues at a and d positions are in bold and charged residues at e and g positions are underlined). At neutral pH, the acidic peptide had a calculated net charge of −7 and the basic peptide of +9. The four consecutive heptads of each

346

Thermodynamics of Leucine Zipper Folding

Figure 1. Helix wheel representation of dimer AB with the two helices oriented parallel and the N termini facing the viewer. The double headed arrows represent interhelical hydrophobic interactions between residues in a, a' and d, d ' positions, respectively. Potential electrostatic side-chain interactions between acidic and basic residues are indicated by dotted lines.

peptide were flanked by N-terminal Ac-Glu and C-terminal Gly-amide. Leu or Ala occupied d position 12 and 19, respectively. The parent (‘‘wild type’’) peptides had Leu in positions 12 and 19 and were called peptide A and B, respectively. Peptides with Ala in position 12 or 19 were abbreviated A12, A19, B12, and B19, respectively. The six peptides could be combined to nine heterodimers: the parent heterodimer AB, the singly substituted heterodimers A12B, A19B, AB12, AB19, and the doubly substituted heterodimers A12B12, A19B12, A12B19 and A19B19. The pattern of hydrophobic residues in the a and d positions of peptides A and B was that of the leucine zipper domain of the transcription factor GCN4, except for the first a position which was Tyr and served as a chromophore to determine peptide concentration. The pattern LdVaLdNa LdVaLd (superscripts indicate position in heptad) stabilizes a parallel and in-register orientation of the dimer (O’Shea et al., 1991; Alber, 1992; Lumb & Kim, 1995; Betz et al., 1995; Junius et al., 1995, 1996). Helix stabilizing residues were chosen for positions b, c, and f located on the solvent-exposed surface of the coiled coil. Figure 1 shows a helix wheel representation of dimer AB with the strands oriented parallel. Acidic and basic peptides associate to stable, parallel heterodimeric coiled coils Earlier sedimentation equilibrium analysis of equimolar amounts of peptide A and B (total chain concentration 100 to 300 mM in 0.1 M phosphate buffer, pH 7.2) revealed a homogeneous population of molecules sedimenting with an apparent

molecular mass of 6710(285) Da, close to the calculated mass of 6921 Da. Peptides A and B alone sedimented with the mass of the monomer (Thomas et al., 1995). These experiments confirmed that the heterodimer AB is the only significantly populated species at micromolar peptide concentrations. In the two-stranded parallel coiled coil AB (Figure 1), the hydrophobic residues in the a and d positions in one strand interact with the residues in the a' and d ' positions, respectively, in the other strand. In the antiparallel orientation, the major hydrophobic interhelical interactions occur between a and d ', and d and a' positions, respectively. Because peptides A and B contain Asn in the central a position 16, the antiparallel orientation produces two energetically very unfavourable interactions by juxtaposing Asn16 (a or a' position) with Leu12 (d ' or d position). This is the reason why Asn in an interior a position always enforces a parallel orientation of strands, as was widely recognized in natural and artificial leucine zippers (Ellenberger et al., 1992; Alber, 1992; Lumb & Kim, 1995; Betz et al., 1995; Wendt et al., 1995; Junius et al., 1995, 1996). The question arises whether the parallel orientation is also favoured in the coiled coil mutants with one or two cavity-creating Ala substitutions in d positions 12 and/or 19? In the parallel oriented mutant leucine zippers, a single Ala substitution introduces a single small cavity. Double Ala substitutions either produce two small cavities (peptides A12B19 and A19B12) or a single large cavity (peptides A12B12 and A19B19). If the mutants had an antiparallel orientation, all the Asn-Asn interactions, which are stabilized by a

Thermodynamics of Leucine Zipper Folding

347

Figure 2. Temperature-induced changes in the CD spectra of the parent heterodimer AB, single mutant AB19, double mutants A12B12 and A12B19 and peptides A and B alone. Spectra were measured in 10 mM sodium phosphate, 75 mM NaCl (pH 7.2) at (bottom to top traces) 7, 15, 25, 35, 45, 55, 65 and 75°C. Total peptide concentration was 100 mM.

hydrogen bond between the amide side chains (O’Shea et al., 1991; Junius et al., 1996) would be lost. Instead there would be either two highly unfavourable Asn-Leu juxtapositions in peptides A19B and AB19, an Asn-Leu and an Asn-Ala juxtaposition in A12B, AB12, A12B19, and A19B12, or two Asn-Ala juxtapositions in A12B12. Assuming all the other interactions are energetically equivalent in the parallel and antiparallel orientation, the replacement in all the mutants of the Asn-Asn interaction by two unfavourable interactions between Asn and Leu and/or Asn and Ala makes the antiparallel orientation energetically unfavourable. The predicted parallel orientation of the coiled coils was tested by fluorescence quenching experiments as described in detail before (Wendt et al., 1995). N-terminal extension of coiled coil peptides by fluorescein-Gly-Gly-Gly produces fluorescent derivatives in which the fluorescence is strongly quenched if the arrangement of the strands is parallel, but not if antiparallel. The N-terminal fluorescein group attached to the coiled coil through a triglycine spacer does not change the equilibrium association constant (Wendt et al., 1995). When a fluorescent coiled coil is mixed with the corresponding non-fluorescent coiled coil the fluorescence increases because of the exchange of strands without a shift in the monomer/dimer equilibrium (Wendt et al., 1995). In contrast, there is no fluorescence increase in the case of strand

exchange between antiparallel coiled coils because the microenvironment of the ‘‘lonely’’ N-terminal fluorescent group does not change on exchanging strands in an antiparallel coiled coil. The nine heterodimeric coiled coils were synthesized with N-terminal fluorescein-Gly-Gly-Gly (triglycine served as a spacer). Equally concentrated solutions of a fluorescent labelled and the corresponding non-labelled coiled coil were mixed and the change of fluorescence was monitored. In each case, fluorescence increased significantly and as observed before (Wendt et al., 1995), the magnitude of the fluorescence increase depended on the monomer/dimer ratio at the peptide concentration employed (5 to 20 mM). The observed fluorescence quenching confirmed the parallel orientation of the nine heterodimeric coiled coils (data not shown). Increasing the temperature changed the CD spectrum of the coiled coils (Figure 2). At low temperature, equimolar mixtures of acidic and basic peptide exhibited the characteristic spectral signature of the a-helical conformation. The ratio of the two minima centred at 221 nm and 208 nm was 1.04 for the heterodimer AB. This value is thought to be typical for interacting parallel a-helices (Zhou et al., 1992a). Raising the temperature gradually changed the spectra with a well defined isodichroic point at 203 nm, consistent with a co-operative two-state helix-to-random coil transition (Greenfield & Fasman, 1969). CD-spectra of peptides A and B in isolation indicated non-helical,

348

Figure 3. Calorimetric binding isotherms describing the association-coupled folding of heterodimer AB (filled squares) and A19B12 (open circles) in 10 mM sodium phosphate (pH 7.2) at 20°C. 12.5 mM peptide A was titrated with 10 ml aliquots of peptide B from a 300 mM solution. 20 mM peptide A19 was titrated with 10 ml aliquots of peptide B12 from a 500 mM solution. The folding enthalpy for each injection (corrected for unspecific dilution effects, see also Materials and Methods) is plotted versus the molar ratio of acidic chain to basic chain. The continuous lines are best fits to a 1:1 association model.

quasi random coil conformation. The ellipticity of peptide A in the region 205 to 240 nm was almost unchanged between 7 and 85°C and the spectrum recorded at 7°C was identical to that of the fully unfolded AB heterodimer at 85°C. Peptide B showed a very weak signal in the a-helix region of the CD-spectrum, which disappeared on heating to 30°C. A weak negative ellipticity in the region 210 to 230 nm has been found in another basic peptide that together with an acidic peptide forms a coiled coil (O’Shea et al., 1993). The finding has been interpreted to reflect less destabilizing interhelical repulsion and differences in side-chain packing in the potential BB homodimer as compared to the potential AA homodimer (O’Shea et al., 1993). In the far UV region around 200 nm, the CD spectrum changed gradually on heating the isolated chains from 7 to 75°C. This was taken to indicate temperature induced alterations in the polypeptide backbone of the single chains (see Discussion). Isothermal microcalorimetric titration experiments ITC was used to measure the thermodynamic parameters of folding. One peptide was placed in the reaction cell of the calorimeter and was titrated with small aliquots of a concentrated solution of the other peptide. The heat of reaction was measured after each addition. Results did not depend on the order of titration: acidic chain to basic chain or vice versa. Two representative titrations are shown in Figure 3. The apparent folding constant, Kfold , and the apparent molar enthalpy of folding, DHfold , were

Thermodynamics of Leucine Zipper Folding

obtained by non-linear regression analysis to a 1:1 association model. Free energies and entropies of folding were calculated from DGfold = −RT ln Kfold and DSfold = (DHfold − DGfold )/T. DHfold was also measured in independent experiments by titrating one peptide into a large excess of the other so that after each addition the peptide in low concentration was completely associated to the peptide in excess. Kfold could not be obtained under these conditions but the determination of DHfold was more accurate (Wiseman et al., 1989). ITC experiments were performed for all possible combinations of peptides at 20°C in 10 mM sodium phosphate (pH 7.2) ionic strength 22 mM. The four single mutants and the double mutant A12B12 were also measured at 10, 30 and 40°C; the other double mutants were too unstable to be analysed by ITC at the higher temperatures. ITC experiments were repeated in Tris-HCl buffer of pH 7.2 and 22 mM ionic strength instead of phosphate buffer. The same values of Kfold and DHfold were obtained. Tris and phosphate buffer have very different heats of ionization, 11.4 kcal mol−1 and 1.22 kcal mol−1, respectively (Jelesarov & Bosshard, 1994; Morin & Freire, 1988). Identical values of DHfold in both buffers demonstrated that at pH 7.2 there was no significant change in the protonation state of any protonable group on going from the unfolded monomers to the folded heterodimer. Therefore, the enthalpy change measured by ITC was equivalent to DHfold . Table 1 summarizes the thermodynamic parameters measured by ITC in the range 10 to 40°C. Replacing Leu by Ala at the two central d positions destabilized the coiled coil. The thermodynamic quantities for the different heterodimers measured at 20°C are compared in Figure 4. Among the single mutants, AB12 was almost as stable as AB (DDG293 = 0.13 kcal mol−1 ) and was even more stable than AB at 10°C (DDG283 = −0.31 kcal mol−1 ). The stability of the single mutants decreased in the order AB12 > A19B > A12B > AB19. The Ala substitution at d position 19 of the basic chain was more destabilizing than the other substitutions. This also followed from the order of stabilities of the double mutants: A12B12 > A19B12 > A12B19 > A19B19. Stability changes were not additive. For example, the parent dimer AB and the double mutant A19B19 differed by DDG = 3.3 kcal mol−1, while the DDGs of the single mutants A19B and AB19 added up to DDG = 2.22 kcal mol−1 (values for 293 K in Table 1). This observation indicated energetic coupling or co-operativity between the d positions in the centre of the coiled coil (see Discussion). The folding of the AB heterodimer was driven by a favourable enthalpy change and opposed by a negative change in entropy (Table 1). Leu to Ala substitutions always led to enthalpic destabilization, which was partly compensated by a favourable increase in the overall folding entropy. Differences in the enthalpy and entropy changes

349

Thermodynamics of Leucine Zipper Folding

Table 1. Thermodynamic parameters of folding determined by isothermal titration calorimetry (10 mM sodium phosphate (pH 7.2), ionic strength 22 mM) T (K) 283

293

303

313

Coiled coil

Kfold × 10−6 (l mol−1 )a

−DGfold (kcal mol−1 )a

−DHfold (kcal mol−1 )b

−TDSfold (kcal mol−1 )

AB AB12 A12B AB19 A19B A12B12 AB AB12 A12B AB19 A19B A12B12 A19B19 A12B19 A19B12 AB AB12 A12B AB19 A19B A12B12 AB AB12 A12B AB19 A19B A12B12

36.0 2 9.0 63.0 2 19.0 22.0 2 4.0 15.0 2 6.0 25.0 2 4.0 5.3 2 0.6 72.0 2 12.0 58.0 2 9.0 12.0 2 1.0 6.3 2 0.7 18.2 2 2.2 2.5 2 0.2 0.25 2 0.02 0.36 2 0.05 1.70 2 0.10 50.0 2 8.0 37.0 2 4.0 10.0 2 1.0 3.4 2 0.2 4.9 2 0.3 1.2 2 0.5 35.0 2 5.0 12.0 2 2.0 1.9 2 0.1 0.95 2 0.10 1.35 2 0.10 0.40 2 0.05

9.8 2 0.20 10.1 2 0.20 9.52 2 0.11 9.30 2 0.29 9.59 2 0.10 8.72 2 0.10 10.55 2 0.11 10.42 2 0.10 9.50 2 0.11 9.13 2 0.13 9.75 2 0.09 8.59 2 0.10 7.25 2 0.10 7.46 2 0.10 7.75 2 0.05 10.69 2 0.10 10.51 2 0.10 9.72 2 0.10 9.07 2 0.10 9.29 2 0.05 8.44 2 0.32 10.82 2 0.10 10.15 2 0.10 9.02 2 0.10 8.57 2 0.10 8.79 2 0.06 8.03 2 0.10

17.60 2 0.37 16.80 2 0.22 14.40 2 0.37 14.20 2 0.50 15.80 2 0.50 14.00 2 0.30 24.70 2 0.43 23.50 2 0.39 21.50 2 0.43 20.70 2 0.50 21.6 2 0.70 21.4 2 0.32 15.3 2 1.0c 15.5 2 1.0c 20.2 2 0.3 31.90 2 0.38 29.30 2 0.27 27.10 2 0.21 26.80 2 0.15 28.40 2 0.22 28.00 2 0.30

7.80 6.70 4.90 4.90 6.21 5.28 14.15 13.10 12.00 11.57 11.90 12.81 8.05 8.04 12.45 21.20 19.10 17.37 17.73 19.11 19.56

a Numbers are mean 2 maximum experimental error. Standard deviations of the mean were always lower than the maximum experimental error. b Mean 2 SDM from eten experiments performed by injection of one chain into a large excess of the other chain under conditions of total association. c Enthalpy obtained from the experiment used to determine Kfold .

were of the same sign and =DDH = > =DDG =, a thermodynamic behaviour that has been observed often in protein systems. The correlation between the changes in enthalpy and entropy was very good (Figure 5). Incidentally, the strong linear correlation is another argument in favour of the uniform orientation of the peptide chains. If some of the mutant peptides would be composed of a mixture of parallel and antiparallel oriented strands, the linear correlation most probably would be lower.

Figure 4. Effect of single and double Leu/Ala substitutions on the thermodynamic parameters of folding at 293 K determined by ITC. Bars represent DDE = DDEmutant − DDEAB , where E = G, H, S.

Heat capacity changes were calculated from DH(T ) = DH(TR ) + DCp (T − TR ) using the reference temperature TR = 293 K and assuming that DCp was temperature independent in the interval 283 to 313 K. The results are listed in Table 2 as DCp (ITC).

Figure 5. Plot of DHfold versus DSfold between 283 and 303 K at pH 7.2 and 22 mM ionic strength for heterodimer AB and eight mutant heterodimers. Values from Table 1. The broken line is a linear least-squares fit to the data (correlation coefficient 0.996).

350.9 343.2 334.5 325.0 328.9 323.6

AB AB12 AB12B AB19 A19B A12B12

Tm DS(Tm )c (kcal/mol) 45.5 42.5 34.6 36.8 35.4 29.4

DH(Tm )b (kcal/mol)

53.6 2 1.0 50.3 2 1.0 42.0 2 1.3 44.1 2 1.2 42.8 2 0.8 36.2 2 1.7 8.08 (7.88) 7.83 (7.71) 7.36 (7.51) 7.30 (7.30) 7.38 (7.39) 6.83 (7.27)

DG(Tm )d (kcal/mol) 0.72 0.63 0.64 0.63 0.63 0.70

DCP (ITC)e (kcal/mol K) 1.02 2 0.1 0.86 2 0.08 0.85 2 0.08 1.00 2 0.1 0.83 2 0.08 1.10 2 0.1

DCP (fit)f (kcal/mol K) 1.4 2 0.1 1.0 2 0.4 0.9 2 0.5 2.1 2 0.4 1.2 2 0.3 2.2 2 0.3

DCP (CD)g (kcal/mol K) 10.48 (10.55) 10.43 (10.42) 9.65 (9.50) 9.29 (9.13) 9.64 (9.75) 8.74 (8.59)

DG(293)h (kcal/mol) 393.3 388.5 379.3 366.9 374 361.7

Tgi (K)

231.1 217.8 215.8 219.8 210.0 225.7

Tg'i (K)

308.7 299.1 293.7 290.3 288.0 291.1

Tsj (K)

298.0 286.9 282.3 281.0 276.3 283.1

Thk (K)

Tm and values of DE(Tm ) refer to experiments with 50 mM total peptide concentration (Figure 7). a The fitting error of Tm was <0.5 K. Values of Tm determined by taking the minimum of the derivative of the CD curves in Figure 6 differed by 21 K. b Values with fitting error; fit by equation (5). Values depended slightly on the magnitude of the temperature interval around Tm where the fitting was done. Probably a better estimate of the error of DH(Tm ) was the mean deviation from the van’t Hoff enthalpy; this error was 10%. c Calculated as Tm DS(Tm ) = DH(Tm ) − DG(Tm ). d Error <0.2 kcal/mol. Numbers in parentheses are calculated from DG(Tm ) = −RTm ln(0.25Ct ); see equation (3). e From the change of −DHfold from ITC with temperature (Table 1), estimated error 210%. f From the global fit of equation (7) to the data shown in Figure 9. g From plotting DH(Tm ) versus Tm using the data for 3 to 4 different peptide concentrations. h Values from stability curve shown in Figure 9. Values calculated from −Kfold at 293 K measured by ITC are given in parentheses. i Tg and Tg' are the concentration-independent temperatures where DGunfold = 0. j Ts , the temperature of maximum stability, was calculated from ln(Tg /Ts ) = [DS(Tg )]/DCP (fit), where DS(Tg ) = [DH(Tg )]/Tg ; Tg , DH(Tg ) and DCP (fit) are the fitting parameters from the curve shown in Figure 9. k Th , the temperature at which DHunfold = 0, was calculated from Th = Tg − [DH(Tg )]/DCP (fit).

Tma (K)

Coiled coil

Table 2. Thermodynamic parameters of the stability curves shown in Figures 7 and 9

351

Thermodynamics of Leucine Zipper Folding

Figure 6. Temperature-induced unfolding of heterodimeric leucine zippers. The unfolding transition was monitored by the change in the CD signal at 221 nm. The data are presented as fraction dimer (equation (1)).Total peptide concentration 50 mM; 10 mM sodium phosphate (pH 7.2).

Thermal unfolding experiments To better understand the energetics of the association-coupled folding transition and to calculate stability curves, data had to be collected in a wider temperature range than was accessible by ITC. This was achieved by thermal denaturation and analysis of the negative CD signal at 221 nm for a two-state monomer-dimer equilibrium. A twostate transition was supported by (1) the sigmoidal shape of the denaturation curves (Figure 6), (2) the well defined isodichroic point at 203 nm (Figure 2) and (3) the linearity of plots described by equation (6) and shown in Figure 8. Thermal unfolding curves were measured for AB, A12B, AB12, A19B, AB19 and A12B12. The raw data (ellipticity at 221 nm versus temperature) were normalized to fraction dimer ( fD ) by: fD =

uobs − (uM + mM·T ) (uN + mN·T ) − (uM + mM·T )

Figure 7. Data used to determine Tm and the unfolding enthalpy DHm at Tm . Data points above 320 K are DGunfold calculated from Kunfold defined by equation (2) and measured in the transition region of the temperature denaturation curves shown in Figure 6 (total peptide concentration 50 mM). Data below 320 K are DGunfold = −DGfold measured by ITC (values from Table 1). The lines are best non-linear least-squares fits according to equation (5). Symbols as for Figure 6.

Ct = A0 + B0 is the total concentration of peptide chains. Kunfold is concentration-dependent and at Tm (fD = 0.5) is: Kunfold (Tm ) = 0.25·Ct

DGunfold = −RT ln Kunfold was calculated for temperatures around Tm (0.4 < fD < 0.6). For a reversible two-state transition, DGunfold is equivalent to −DGfold . The two sets of DG values from ITC and thermal unfolding can be combined and described by the Gibbs-Helmholtz equation: DG(T ) = DH(Tm ) + DCP (T − Tm )

$

− T DS(Tm ) + DCP ln

(1)

uM and uN are the ellipticity value of the fully unfolded monomer and the fully folded, native dimer, respectively, at a lower reference temperature, e.g. 0°C. mM and mN are the slopes of the linear pre-transition and post-transition temperature dependence of uM and uN , respectively†. For a two-state equilibrium A + B F AB and equivalent total peptide concentrations, A0 = B0 , the equilibrium unfolding constant is (Marky & Bresslauer, 1987): (1 − fD )2·Ct Kunfold = (2) 2·fD

† For the less stable heterodimers A12B19, A19B12 and A19B19, the quality of the pre- and post-transition baseline was not sufficient to calculate fD by equation (1).

(3)

0 1% T Tm

(4)

Equation (4) does not account explicitly for the concentration dependence of DG in the case of a dimeric system and was modified by substituting Tm DS(Tm ) = DH(Tm ) − RTm ln(0.25·Ct ). After rearranging, one obtains:

0

DG(T ) = DH(Tm )· 1 −

0

T Tm

+ DCP T − Tm − T·ln

1 1

T − RT·ln(0.25·Ct ) Tm

(5)

DH(Tm ) is the transition enthalpy of unfolding and DCp is the difference in heat capacity between the native heterodimeric state and the denatured monomeric state. The last term describes the concentration dependence. Equation (5) was fitted to −DGfold (T ) from ITC and to DGunfold (T ) from

352

Thermodynamics of Leucine Zipper Folding

Figure 8. Plots according to equation (6). The linearity of these plots is in agreement with a simple two-state transition reaction. Symbols and conditions are as for Figure 6.

thermal unfolding. Figure 7 shows the values of DG(T ) for 50 mM total peptide concentration together with a best fit by equation (5). As a further test of the validity of the two-state transition described by equations (2) and (3), ln Ct was plotted against 1/Tm . The resulting plots were linear (Figure 8), in agreement with equation (6) (Marky & Breslauer, 1987): 1 R DS° + R ln 4 = − ·ln Ct + Tm DH° DH°

(6)

DH° and DS° refer to the temperature-independent change in enthalpy and entropy, respectively. Stability curves A protein stability curve is defined as the temperature variation of DGunfold and represents a best fit of a set of experimental values of DGunfold (T ) to a particular thermodynamic model of unfolding (Becktel & Schellman, 1987). The stability curves for six heterodimers are depicted in Figure 9 and were constructed as follows. Data points at lower temperatures were obtained by ITC performed between 10 and 40°C (Table 1). Data points at higher temperatures are DGunfold (Tm ) for different peptide concentrations and were obtained from fits of DG(T ) to equation (5). The total peptide concentration Ct was varied up to fourfold to shift Tm by up to 10 deg.C. DGunfold (Tm ) obtained by the fitting procedure and DGunfold (Tm ) calculated by equation (3) differed by less than 0.2 kcal mol−1. Finally, a global fit of the data was obtained using the equation:

0 1 0

DGunfold (T ) = DH(Tg )· 1 −

T Tg

+ DCP T − Tg − T·ln

1

T Tg

(7)

Figure 9. Stability curves of heterodimeric leucine zippers in 10 mM sodium phosphate buffer (pH 7.2). Below 320 K, the values of DGunfold were obtained from ITC (Table 1). At higher temperatures, DGunfold was obtained from temperature-induced unfolding curves at different total peptide concentrations. The lines are best non-linear least-squares fit of equation (7) to the combined data set. Fitting parameters are listed in Table 2. Solid segments indicate the temperature region where experimental data were collected.

Tg is the concentration-independent reference temperature at which DGunfold = 0. The parameters deduced from the fits in Figure 9 are listed in Table 2. DCp obtained from the fit by equation (7) is tabulated as DCp (fit) in Table 2. DCp was also calculated from dDH(Tm )/d(Tm ), using DH(Tm ) and Tm determined for different total peptide concentration. The corresponding values are called DCp (CD) and are also listed in Table 2.

Discussion Structure of the heterodimeric leucine zippers Peptides A and B were designed to assemble to a parallel and in-register coiled coil by choosing the pattern of a and d residues of the parallel and dimeric leucine zipper domain of the GCN4 transcription factor, including a pair of Asn residues in a central a position. The Asn pair is the main determinant of the parallel and in-register orientation (O’Shea et al., 1991; Lumb & Kim, 1995; Betz et al., 1995; Junius et al., 1995, 1996). The parallel orientation of the acidic and basic strand of all the nine heterodimers was confirmed by fluorescence quenching experiments based on our previous finding that in a parallel coiled coil, N-terminal fluorescein groups are strongly quenched (Wendt et al., 1995). In addition, model building shows that the antiparallel orientation is energetically disfavoured because the characteristic Asn-Asn interaction at position 16 is replaced by unfavourable juxtaposition of Asn with Leu and/or Ala. To test if the hydrophobic core of the coiled coil

353

Thermodynamics of Leucine Zipper Folding

structure was tight and well packed, binding of the fluorescent dye 1-anilino-8-naphthalenesulphonate (ANS) was tested. This dye binds to hydrophobic patches and to clusters of loosely packed hydrophobic residues (Stryer, 1965; Semisotnov et al., 1991). Neither the parent AB dimer nor any of the less stable mutant dimers bound ANS. The increase in fluorescence on addition of excess ANS was always less than 5% of the total fluorescence signal (not shown). The inability to bind ANS, the overall thermodynamic parameters to be discussed next, and the highly co-operative folding behaviour demonstrated that the heterodimeric coiled coils had a native-like structure and could be regarded as native ‘‘mini proteins’’ (Betz et al., 1995; Zhou et al., 1992b). Thermodynamics of coiled coil folding measured by ITC When normalized per mol of residue, the enthalpies of folding at 20°C for AB and the eight single and double mutants fell in the range of −0.26 to −0.42 kcal/(mol of residue). The entropic contributions (TDS) at 20°C were −0.13 to −0.24 kcal/(mol of residue). These values are well within the range of −DHunfold and −TDSunfold reported for a reference set of 20 proteins of known 3D structure (Table III and Figures 7 and 8 of Makhatadze & Privalov, 1995). At Tm , which ranged from 50°C (A12B12) to 78°C (AB), DHunfold (Tm ) varied between 0.6 and 0.89 kcal per mol of residue, again typical values found also in the reference set of 20 proteins (Makhatadze & Privalov, 1995). The similarity of the normalized parameters indicates that the energetic factors which drive folding and association of two polypeptide chains are balanced very similarly to the folding of a single polypeptide chain to the native state. Large positive heat capacity changes are the hallmark of protein unfolding transitions (Sturtevant, 1977; Becktel & Schellman, 1987; Privalov & Gill, 1988). DCP reflects mainly the exposure of buried hydrophobic residues to water upon unfolding and empirically correlates with the increase of solvent-accessible apolar and polar surface on going from the native compact state to the denatured state (Murphy & Freire, 1992). DCP (ITC) for the unfolding of AB was 715 cal K−1 (mol dimer)−1 (Table 2), corresponding to 12 cal K−1 (mol residue)−1. This is again within the range of values measured for the unfolding of several monomeric proteins (Privalov & Gill, 1988). Hence, in spite of their small size, the short rod-shaped coiled coils possess a tightly packed hydrophobic core, whose contribution to thermodynamic stability is comparable to globular proteins and helps to overcome an unfavourable surface to volume ratio. A very similar value of 740 cal K−1 (mol dimer)−1 was reported for B1-Dim, a homodimeric coiledcoil of similar size (64 residues) and similar stability (De Francesco et al., 1991). DCp of unfolding of the leucine zipper peptide GCN4-56 measured by

Figure 10. Free energies of coupling between leucine residues at central d positions. Boxed numbers are DGc (kcal/mol) at 20°C for residues connected by double headed arrows. DGc was calculated by equation (8) using values of −DGfold of Table 1.

differential scanning calorimetry is 0300 cal K−1 (mol dimer)−1, less than DCp (ITC) of coiled coil AB (Thompson et al., 1993). DCp (ITC) of most single and double mutants was slightly lower than DCp of the parent dimer AB. A lower DCp for destabilizing mutations has been observed before (Kuroki et al., 1992; Cooper et al., 1992; Steif et al., 1995). It has been argued that destabilizing cavity creating mutations tend to increase the heat capacity of the native state but do not affect the unfolded state and in this way reduce the apparent heat capacity change of unfolding (Steif et al., 1995). Disruption of intermolecular contacts on creating a cavity may increase the vibrational content of the native state due to the tendency to reduce the cavity and, thus, could decrease DCp of unfolding. The present data support this view. Surprisingly, however, DCp (ITC) for AB and A12B12 were the same within error. Side-chain rearrangements and main-chain flexibility may optimize the local geometry on inner core mutations (Baldwin et al., 1993; Lim et al., 1994). Co-operative interactions in the hydrophobic core The decrease in DGfold in going from single to double Leu/Ala substitutions was not additive, indicating co-operative interactions between residues of the hydrophobic core. The co-operativity can be expressed by the free energy of coupling, DGc , as described by Horovitz & Fersht (1990): DGc = (DGfold,sm 1 − DGfold,AB ) + (DGfold,sm 2 − DGfold,AB ) − (DGfold,dm − DGfold,AB )

(8)

DGfold,AB is the free energy of folding of the parent coiled coil AB, DGfold,sm 1 and DGfold,sm 2 are the free energies of folding of two single mutants (e.g. A12B and AB12), and DGfold,dm is the free energy of folding

354 of the corresponding double mutant (e.g. A12B12)†. Coupling free energies between d residues that are in direct van der Waals contact (A12B12, A19B19) and between d residues that are separated by the central Asn pair (A12B19, A19B12) are shown in Figure 10. Specific structurebased interpretations of the coupling free energies are not possible in the absence of high resolution structural data and only generalized explanations can be inferred. The arrangement of stability-determining hydrophobic residues was sequentially symmetrical but the peptide chains were not identical. a and b-carbon atoms of residues in e and g positions also contributed to the hydrophobic core. Molecular modelling demonstrated that charged residues in e and g positions could build a complicated network of electrostatic interactions and may have influenced the strength of interactions in the inner core of the dimer, both directly via the packing of the aliphatic parts of the side-chains and indirectly by restricting the mobility of the main-chain. The two central d positions were energetically non-equivalent in the heterodimeric coiled coil and mutations in equivalent positions of the acidic and basic chain differed in stability (compare A19B and AB19, Figure 4). The energetic coupling propagated through the hydrophobic core in an asymmetric manner. The observed cross-wise interaction between leucine residues in positions 12 and 19 (Cg to Cg distance ˚ ) was mediated by the intervening Asn pair. 012 A This pair of polar residues, which is typical for naturally occurring leucine zippers, not only guided the chains to be parallel and in register but also helped to fine tune polar and non-polar interactions in the hydrophobic core and thereby influenced thermodynamic stability. Protein stability curves For a co-operative two-state transition, the temperature dependence of the Gibbs free energy change is referred to as the stability curve (Becktel & Schellman, 1987). Knowledge of DCp is necessary to calculate a stability curve. An interesting possibility to construct a protein stability curve without precise DCP data was proposed by Pace & Laurents (1989). The idea is to measure DGunfold in a wide temperature range, including the region of maximal stability where the curvature of the stability curve is most evident. DCP , DH(Tm ) and DS(Tm ) are then obtained from fitting of all the measured values of DGunfold (T ) to the Gibbs-

† A mutation in one unfolded chain (e.g. A : A12) is independent of a mutation in the other unfolded chain (e.g. B : B12) because the two unfolded chains are independent of each other; this is different to the case of a monomeric protein where also in the unfolded polypeptide chain one mutation may, in principle, influence the energetic cost of a nearby mutation.

Thermodynamics of Leucine Zipper Folding

Helmholtz equation. The difficulty with this approach is that in most cases DGunfold (T ) has to be measured under different solvent conditions (pH, ionic strength, buffer salt, denaturant). For example, DGunfold near Tm , Tm itself, and DHm are extracted from temperature unfolding curves measured under varying solvent conditions. DGunfold far below Tm is extrapolated linearly from denaturant unfolding curves to water (Pace, 1986). The validity of the linear extrapolation method may be questionable (Yao & Bolen, 1995). The denatured state of a protein may depend on solvent composition and pH (Arcus et al., 1995; Shortle, 1996). As a result, different values of DGunfold (T ) measured in solutions of different composition may not necessarily pertain to the same stability curve. These difficulties could be overcome in our analysis of heterodimeric coiled coils for which we could measure DG(T ) under identical solvent conditions. Since the monomer/dimer equilibrium was concentration-dependent, variation of Tm and DHm was accomplished by variation of the peptide concentration. At temperatures where the native state predominates, DGfold = −DGunfold was measured by ITC. The parent coiled coil AB was most stable at 37°C. At 20°C, the free energy of unfolding was 10.5 kcal mol−1, comparable to the unfolding free energy of two other dimeric systems, the Arc repressor protein (11 kcal mol−1; Bowie & Sauer, 1989) and the B1-Dim peptide (011.5 kcal mol−1; De Francesco et al., 1991). The temperature of maximum stability of the most stable mutant AB12 was 10 degrees lower than that of AB. Maximum stability of the other mutants was 15 to 20 degrees lower than that of AB (Figure 9 and Ts in Table 2). The contribution of a methylene group at the hydrophobic core to coiled coil stability varied in the range of 0.1 to 1 kcal/mol (mean 0.42(20.15) at 293 K and 0.82(20.40) at 334 K, the mean Tm ). Values around 0.5 kcal per mol of methylene groups at room temperature have been reported for Leu to Ala mutations in disulphide linked synthetic leucine zippers (Zhou et al., 1992a). Data from protein engineering indicate larger average stabilizing effects of 1.3 to 1.5 kcal per mol of methylene group (Matthews, 1993; Fersht & Serrano, 1993). Depending on side-chain rearrangement and compensatory deformation of the protein backbone following the creation of a cavity, the measured loss in stability per methylene group may vary considerably (Lee, 1993; Matthews, 1993). The low energetic cost of Leu to Ala mutations in coiled coils could be caused by compensatory conformational adaptation of the peptide backbone. This argument is supported by our previous CD studies reflecting a ‘‘non-ideal’’ a-helical backbone conformation in mutant coiled coils: substitution of Leu by Ala reduced the molar ellipticity per residue in the a-helical region of the CD spectrum without changing the overall dimeric stability (Wendt et al., 1995).

355

Thermodynamics of Leucine Zipper Folding

Discrepancy between calorimetric and van’t Hoff enthalpy For a protein system that obeys the reversible two-state transition model, the maximum stability is reached at temperature Ts , where dDG/ dT = −DS = 0 and the system is purely enthalpically stabilized. The second characteristic temperature Th at which DH = 0 and changes sign is shifted by several degrees to the left of Ts on the temperature scale (see Table 2 for calculation of Ts and Th ). Accordingly, one predicts from the stability curve of AB (Figure 9) that DH changes sign at 25°C. However, DHfold for AB was large and negative when measured directly by ITC. Also for the Leu/Ala substituted heterodimers, DH is predicted to change sign around room temperature but is negative in the temperature region of Th if measured directly (Table 1). This discrepancy between −DHfold measured calorimetrically from 10 to 40°C and calculated for the same temperature range with the help of DCp (fit) (equation (7), Figure 9) came as a surprise. In principle, several explanations are possible. For example, the assumption of a two-state equilibrium may be wrong. However, the evidence for a two-state transition is very strong, as discussed above. In further support we found that the kinetics of association and folding are described by a single concentration-dependent association rate constant and a single concentration-independent dissociation rate constant, in accord with the mechanism A + B F AB (unpublished results). The kinetics of folding of the 33-residue GCN4-p1 leucine zipper also follows a simple two-state mechanism (Zitzewitz et al., 1995) and calorimetry of a 56-residue fragment of the GCN4 basic leucine zipper domain shows no evidence for significantly populated intermediate states (Thompson et al., 1993). Another explanation is a heat effect caused by differential binding of buffer ions to the native and the unfolded state and contributing to the overall heat change measured by ITC. This possibility can also be excluded because the measured enthalpies were the same in buffers of largely different composition. As a third and, we believe, likely explanation we propose that the unfolded (denatured) states of the single A and B chains were not equivalent over the entire temperature range of the measurements. This is to say that the denatured state at low temperature possessed ‘‘excess’’ enthalpy that favoured folding. The unfolded state of a protein is only operationally defined and poorly understood, especially under conditions where the native state predominates and the unfolded state is marginally populated (Shortle, 1996). The term unfolded has solely theoretical meaning and, in practice, we deal with the denatured state. To what extent the denatured state approximates the unfolded state depends on the system under study. Denatured states of the same protein induced by different physical means and chemical agents may differ (Arcus et al., 1995;

Shortle, 1996 ). If the denatured macro-states differ depending on solvent conditions, temperature, etc. the energetics of the transition between the native and the denatured state will differ as well. Here, we envisage some residual order of the single peptide chains at low temperatures but not at temperatures around Tm . The CD spectrum between 190 and 205 nm changed when either of the isolated chains was heated from room temperature (initial denatured state in the ITC experiment) to over 85°C (final denatured state in the thermal unfolding experiment). The spectral changes, which featured no isosbestic point(s), hint at a temperature-induced gradual shift between differently populated families of backbone conformations. The difference in the heat capacity change from direct calorimetric measurement and van’t Hoff analysis is in line with this explanation: DCP (CD) was larger than DCP (ITC) indicating a transition to a less compact structure of the isolated chains near Tm . Accordingly, DCp (fit) had to be intermediate between DCp (ITC) and DCp (CD), as observed. Discrepancies between calorimetric and van’t Hoff enthalpies were discussed by Sturtevant and co-workers (Naghibi et al., 1995; Liu & Sturtevant, 1995). We know of one report describing an ‘‘anomalous’’ favourable enthalpy contribution to the formation of a DNA heteroduplex at 25°C because the single strand ‘‘unfolded’’ DNA has an ordered conformation (Vesnaver & Breslauer, 1991). In general, calorimetric and van’t Hoff enthalpies may differ whenever residual interactions occur in the denatured state and if such partly ordered or partly compact structures change with temperature. This will influence the thermodynamic profile of a conformational transition. To our knowledge, such discrepancies have not yet been reported for conformational transitions in monomeric proteins, perhaps because for a monomeric protein, DHfold and DHunfold cannot be measured directly and under identical solvent conditions over a large enough temperature range, including the region of high stability where the denatured state is only scarcely populated.

Materials and Methods Peptide synthesis and purification Peptides were synthesized on a multiple automated peptide synthesizer (MultiSynth Technology, Bochum, Germany) and on a 433A peptide synthesizer (Applied Biosystems), using the Rink amide MBHA resin from Novabiochem, the 9-fluorenylmethyloxycarbonyl (NaFmoc) protection strategy and carboxyl group activation by O-benzotriazol-1-yl-N, N, N ', N '-tetramethyl-uroniumhexafluorophosphate/N, N-diisopropylethylamine. The N-terminal residue was acetylated by reaction of the resin-bound and side chain-protected peptide with a tenfold molar excess of acetic anhydride and a fivefold molar excess of diisopropylethylamine in dichloromethane. Deprotection of the side-chains and cleavage from the resin was achieved with a mixture of 100 ml thioanisol + thiocresol (1:1) and 900 ml trifluoroacetic

356 acid, or with a mixture of 0.37 g phenol, 250 ml water and 3 ml trifluoroacetic acid. Peptides were released as C-terminal amides. The material was desalted on a Sephadex-G25 column in 1 M acetic acid. Final purification was achieved by reversed phase high performance liquid chromatography on a semi-preparative C8 column (Machery & Nagel) eluted with binary gradients of acetonitrile/water containing 0.1% or 0.085% (v/v) trifluoroacetic acid. Purity of peptides was controlled by amino acid analysis and ion spray mass spectrometry. Peptide concentrations were determined by amino acid analysis on an on-line PTH-amino acid analyser (Applied Biosystems Inc., model 120A) and by UV-absorption in −1 6 M guanidinium hydrochloride (e275,3 cm−1; mM = 1.45 mM Brandts & Kaplan, 1973). Fluorescein-labelled peptides were prepared as before (Wendt et al., 1995). CD spectroscopy CD spectra and thermal denaturation curves were measured on a JASCO J-500 spectropolarimeter in a thermostated cuvette of 1 mm pathlength. Wavelength scans were performed at a constant temperature with a 1 nm bandwidth and a 1 nm min−1 scan speed. Temperature was maintained using a JULABO F10 circulating water bath and was monitored by a Pt100 temperature probe in physical contact with the peptide solution. Temperature scans were performed by scanning continuously from 7 to 90°C at scan rates of 0.5 deg.min−1 or 1 deg.min−1 and the ellipticity at 221 nm was measured at discrete intervals. Typically, 180 equally spaced data points per curve were subjected to data analysis. Reversibility was checked by repeated scans. Temperature gradients were reproducible to within 0.1 deg.C. Total peptide concentration was varied between 3 and 200 mM.

Thermodynamics of Leucine Zipper Folding

small aliquots from the rotating injection syringe. To measure the equilibrium folding constant Kfold , 5 to 50 mM peptide in the reaction cell was titrated with aliquots of a 25-fold more concentrated solution of the other peptide until saturation was reached. A typical experiment consisted of 14 injections, each of 10 ml volume and 15 seconds duration, with a five minute interval between injections. In these experiments, the product of the total peptide concentration times Kfold , the c-value as defined by Wiseman et al. (1989), was 15 to 350. The titration data were corrected for the small heat changes observed upon injections after saturation. For a more accurate determination of the folding enthalpy, one peptide was injected in 5 to 10 ml portions into a large excess of the other peptide placed in the reaction cell so that the folding reaction was complete at very low partial saturation (the ratio of the total peptide concentrations was <0.1). In these experiments, the c-values were >1000. To correct for heat effects not directly related to folding, control experiments were carried out by repeating the injection scheme of the actual experiment with one peptide injected into the sample cell containing only buffer. Data analysis was performed with the software provided with the instrument.

Acknowledgements We thank Christine Berger and Dr Hans Wendt for many helpful discussions and Dr Anette Beck-Sickinger for help with the synthesis of some peptides. This work was supported in part by the Swiss National Science Foundation, the Kommission fu¨r wissenschaftliche Forschung, the Ciba-Geigy Jubila¨ums-Stiftung, the EMDO-Stiftung, and the Hartmann-Mu¨ller Stiftung.

Fluorescence measurements Fluorescein-labelled heterodimeric coiled coils (5 to 20 mM in 10 mM sodium phosphate, pH 7.2) were mixed with equally concentrated solutions of non-labelled peptides at 20°C. The change in fluorescence emission was measured between 500 and 600 nm (excitation 360 nm, 1 cm path length, 1 nm steps, one second integration time) in a Spex Fluorolog spectrofluorimeter. Binding of ANS was tested by addition of 2 to 20 ml of 10 mM aqueous dye solution to 2 ml of 50 mM solutions of coiled coils and measuring the change in fluorescence emission between 400 and 600 nm (excitation 350 nm). Other conditions as above. Isothermal titration calorimetry (ITC) All calorimetric experiments were performed with an OMEGA titration calorimeter (Microcal Inc., Northampton, MA) equipped with a nanovolt preamplifier to reduce electrical noise. The calorimeter was calibrated with electrically generated heat pulses as recommended by the manufacturer. To improve baseline stability, the temperature of the system was kept about 5 deg.C below the working temperature of the actual experiment with the help of a circulating water bath. All solutions were thoroughly degassed by stirring under vacuum before use. Peptide samples were prepared from concentrated stock solutions by dilution with buffer of the same batch to minimize artifacts due to minor differences in buffer composition. One peptide was placed in the reaction cell (1.34 ml volume) and the other peptide was injected in

References Alber, T. (1992). Structure of the leucine zipper. Curr. Opin. Genet. Dev. 2, 205–210. Arcus, V. L., Vuilleumier, S., Freund, S. M. V., Bycroft, M. & Fersht, A. R. (1995). A comparison of the pH, urea and temperature-denatured states of barnase by heteronuclear NMR: implications for the initiation of protein folding. J. Mol. Biol. 254, 305–321. Baldwin, E. P., Hajiseyedjavadi, O., Baase, W. A. & Matthews, B. W. (1993). The role of backbone flexibility in the accommodation of variants that repack the core of T4 lysozyme. Science, 262, 1715–1718. Becktel, W. J. & Schellman, J. A. (1987). Protein stability curves. Biopolymers, 26, 1859–1877. Betz, S. F., Bryson, J. W. & DeGrado, W. F. (1995). Native-like and structurally characterized designed a-helical bundles. Curr. Opin. Struct. Biol. 5, 457–463. Bowie, J. U. & Sauer, R. T. (1989). Equilibrium dissociation and unfolding of the Arc repressor dimer. Biochemistry, 28, 7139–7143. Brandts, J. F. & Kaplan, L. J. (1973). Derivative spectroscopy applied to tyrosyl chromophores. Studies on ribonuclease, lima bean inhibitor, insulin and pancreatic trypsin inhibitor. Biochemistry, 12, 2011–2024. Cooper, A., Eyles, S. J., Radford, S. E. & Dobson, C. M. (1992). Thermodynamic consequences of the removal of a disulphide bridge from hen lysozyme. J. Mol. Biol. 225, 939–943.

Thermodynamics of Leucine Zipper Folding

De Francesco, R., Pastore, A., Vecchio, G. & Cortese, R. (1991). Circular dichroism study on the conformational stability of the dimerization domain of transcription factor LFB1. Biochemistry, 30, 143–147. Eftink, M. R., Helton, K. J., Beavers, A. & Ramsay, G. D. (1994). The unfolding of trp aporepressor as a function of pH: evidence for an unfolding intermediate. Biochemistry, 33, 10220–10228. Ellenberger, T. (1994). Getting a grip on DNA recognition: structures of the basic region leucine zipper, and the basic region helix-loop-helix DNA-binding domains. Curr. Opin. Struct. Biol. 4, 12-21. Ellenberger, T. E., Brandl, C. J., Struhl, K. & Harrison, S. C. (1992). The GCN4 basic region leucine zipper binds DNA as a dimer of uninterrupted alpha helices: crystal structure of the protein-DNA complex. Cell, 71, 1223–1237. Fersht, A. R. & Serrano, L. (1993). Principles of protein stability derived from protein engineering experiments. Curr. Opin. Struct. Biol. 3, 75–83. Freire, E., Mayorga, O. L. & Straume, M. (1990). Isothermal titration calorimetry. Anal. Chem. 62, 950A–959A. Graddis, T. J., Myszka, D. G. & Chaiken, I. M. (1993). Controlled formation of model mono- and heterodimer coiled coil peptides. Biochemistry, 32, 12664–12671. Greenfield, N. & Fasman, G. D. (1969). Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry, 8, 4108–4116. Horovitz, A. & Fersht, A. R. (1990). Strategy for analysing the co-operativity of intramolecular interactions in peptides and proteins. J. Mol. Biol. 214, 613–617. Hurst, H. C. (1995). Transcription factors 1: bZIP proteins. Protein Profile, 2, 101–168. Jelesarov, I. & Bosshard, H. R. (1994). Thermodynamics of ferredoxin binding to ferredoxin:NADP+ reductase and the role of water at the complex interface. Biochemistry, 33, 13321–13328. Johnson, C. M., Cooper, A. & Stockley, P. G. (1992). Differential scanning calorimetry of thermal unfolding of the methionine repressor protein (MetJ) from Escherichia coli. Biochemistry, 31, 9717–9724. Junius, F. K., Mackay, J. P., Bubb, W. A., Jensen, S. A., Weiss, A. S. & King, G. F. (1995). Nuclear magnetic resonance characterization of the Jun leucine zipper domain: unusual properties of coiled-coil interfacial polar residues. Biochemistry, 34, 6164–6174. Junius, F. K., O’Donoghue, S. I., Nilges, M., Weiss, A. S. & King, G. F. (1996). High resolution NMR solution structure of the leucine zipper domain of the c-Jun homodimer. J. Biol. Chem. 271, 13663–13667. Kohn, W. D., Monera, O. D., Kay, C. M. & Hodges, R. S. (1995). The effects of interhelical electrostatic repulsions between glutamic acid residues in controlling the dimerization and stability of twostranded a-helical coiled-coils. J. Biol. Chem. 270, 25495–25506. Kuroki, R., Inaka, K., Taniyama, Y., Kidokoro, S., Matsushima, M., Kikuchi, M. & Yutani, K. (1992). Enthalpic destabilization of a mutant human lysozyme lacking a disulphide bridge between cysteine-77 and cysteine-95. Biochemistry, 31, 8323– 8328. Landschulz, W. H., Johnson, P. F. & McKnight, S. L. (1988). The leucine zipper: a hypothetical structure common to a new class of DNA binding proteins. Science, 240, 1759–1764.

357 Lee, B. (1993). Estimation of the maximum change in stability of globular proteins upon mutation of a hydrophobic residue to another of smaller size. Protein Sci. 2, 733–738. Lim, W. A., Hodel, A., Sauer, R. T. & Richards, F. M. (1994). The crystal structure of a mutant protein with altered but improved hydrophobic core packing. Proc. Natl Acad. Sci. USA, 91, 423–427. Liu, Y. & Sturtevant, J. M. (1995). Significant discrepancies between van’t Hoff and calorimetric enthalpies II. Protein Sci. 4, 2559–2561. Lumb, K. L. & Kim, P. S. (1995). A buried polar interaction imparts structural uniqueness in a designed heterodimeric coiled coil. Biochemistry, 34, 8642–9648. Makhatadze, G. I. & Privalov, P. L. (1995). Energetics of protein structure. Advan. Protein Chem. 47, 308–425. Mann, C. J., Royer, C. A. & Matthews, C. R. (1993). Tryptophan replacements in the trp aporepressor from Escherichia coli: probing the equilibrium and kinetic folding models. Protein Sci. 2, 1853–1861. Marky, L. A. & Bresslauer, K. J. (1987). Calculating thermodynamic data for transitions of any molecularity from equilibrium melting curves. Biopolymers, 26, 1601–1620. Matthews, B. W. (1993). Structural and genetic analysis of protein stability. Annu. Rev. Biochem. 62, 139–160. Morin, P. E. & Freire, E. (1988). Direct calorimetric analysis of the enzymatic activity of yeast cytochrome c oxidase. Biochemistry, 30, 8494–8500. Murphy, K. P. & Freire, E. (1992). Thermodynamics of structural stability and cooperative folding behavior in proteins. Advan. Protein Chem. 43, 313–361. Myszka, D. G. & Chaiken, I. M. (1994). Design and characterization of an intramolecular antiparallel coiled coil peptide. Biochemistry, 33, 2368–2372. Naghibi, H., Tamura, A. & Sturtevant, J. M. (1995). Significant discrepancies between van’t Hoff and calorimetric enthalpies. Proc. Natl Acad. Sci. USA, 92, 5597–5599. O’Shea, E. K., Klemm, J. D., Kim, P. S. & Alber, T. (1991). X-ray structure of the GCN4 leucine zipper, a two-stranded, parallel coiled coil. Science, 254, 539–544. O’Shea, E. K., Lumb, K. J. & Kim, P. S. (1993). Peptide ‘‘Velcro’’: design of a heterodimeric coiled coil. Curr. Biol. 3, 658–667. Pace, C. N. (1986). Determination and analysis of urea and guanidine hydrochloride denaturation curves. Methods Enzymol. 131, 266–280. Pace, C. N. & Laurents, D. V. (1989). A new method for determining the heat capacity change for protein folding. Biochemistry, 28, 2520–2525. Privalov, P. L. (1982). Stability of proteins. Proteins which do not present a single cooperative system. Advan. Protein Chem. 35, 1–104. Privalov, P. L. & Gill, S. J. (1988). Stability of protein structure and hydrophobic interaction. Advan. Protein Chem. 39, 191–234. Semisotnov, G. V., Rodionova, N. A., Razgulayev, O. I., Uversky, V. N., Gropas, A. F. & Gilmanshin, R. I. (1991). Study of the ‘‘molten globule’’ intermediate state in protein folding by hydrophobic fluorescent probe. Biopolymers, 31, 119–128. Shortle, D. (1996). The denatured state (the other half of the folding equation) and its role in protein stability. FASEB J. 10, 27–34. Steif, C., Hinz, H. J. & Cesareni, G. (1995). Effects of cavity-creating mutations on conformational

358

Thermodynamics of Leucine Zipper Folding

stability and structure of the dimeric 4-a-helical protein ROP: thermal unfolding studies. Proteins: Struct. Funct. Genet. 23, 83–96. Stryer, L. (1965). The interaction of naphtalene dye with apomyoglobin and apohemoglobin. A fluorescent probe of non-polar binding sites. J. Mol. Biol. 13, 482–495. Sturtevant, J. M. (1977). Heat capacity and entropy changes in processes involving proteins. Proc. Natl Acad. Sci. USA, 74, 2236–2241. Thomas, R. M., Wendt, H., Zampieri, A. & Bosshard, H. R. (1995). a-Helical coiled coils: simple models for self-associating peptide and protein systems. Prog. Coll. Polymer. Sci. 99, 24–30. Thompson, K. S., Vinson, C. R. & Freire, E. (1993). Thermodynamic characterization of the structural stability of the coiled-coil region of the bZIP transcription factor GCN4. Biochemistry, 32, 5491–5496. Tiktopulo, E. I., Privalov, P. L., Odintsova, T. I., Ermokhina, T. M., Krasheninnikov, I. A., Aviles, F. X., Cary, P. D. & Crane-Robinson, C. (1982). The central tryptic fragment of histones H1 and H5 is a fully compact domain and is the only folded region in the polypeptide chain. A thermodynamic study. Eur. J. Biochem. 122, 327–331. Vesnaver, G. & Breslauer, K. J. (1991). The contribution of DNA single-stranded order to the thermodynamics of duplex formation. Proc. Natl Acad. Sci. USA, 88, 3569–3573. Wendt, H., Berger, C., Baici, A., Thomas, R. M. &

Bosshard, H. R. (1995). Kinetics of folding of leucine zipper domains. Biochemistry, 34, 4097–4107. Wintrode, P. L., Griko, Y. V. & Privalov, P. L. (1995). Structural energetics of barstar studied by differential scanning microcalorimetry. Protein Sci. 4, 1528–1534. Wiseman, T., Williston, S., Brandts, J. F. & Lin, L. N. (1989). Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal. Biochem. 179, 131–137. Yao, M. & Bolen, D. W. (1995). How valid are denaturant-induced unfolding free energy measurements? Level of conformance to common assumptions over an extended range of ribonuclease A stability. Biochemistry, 34, 3771–3781. Zhou, N. E., Kay, C. M. & Hodges, R. S. (1992a). Synthetic model proteins: the relative contribution of leucine residues at the nonequivalent positions of the 3-4 hydrophobic repeat to the stability of the twostranded alpha-helical coiled-coil. Biochemistry, 31, 5739–5746. Zhou, N. E., Zhu, B. Y., Kay, C. M. & Hodges, R. S. (1992b). The two-stranded alpha-helical coiled coil is an ideal model for studying protein stability and subunit interactions. Biopolymers, 32, 419–426. Zitzewitz, J. A., Bilsel, O., Luo, J., Jones, B. E. & Matthews, C. R. (1995). Probing the folding mechanism of a leucine zipper peptide by stopped-flow circular dichroism spectroscopy. Biochemistry, 34, 12812– 12819.

Edited by F. E. Cohen (Received 4 March 1996; received in revised form 25 July 1996; accepted 12 August 1996)