Stability of monomeric Cro variants: Isoenergetic transformation of a type I′ to a type II′ β-hairpin by single amino acid replacements

Abstract

The thermodynamic stabilities of three monomeric variants of the bacteriophage λ Cro repressor that differ only in the sequence of two amino acids at the apex of an engineered β-hairpin have been determined. The sequences of the turns are EVK-XX-EVK, where the two central residues are DG, GG, and GT, respectively. Standard-state unfolding free energies, determined from circular dichroism measurements as a function of urea concentration, range from 2.4 to 2.7 kcal/mole, while those determined from guanidine hydrochloride range from 2.8 to 3.3 kcal/mole for the three proteins. Thermal denaturation yields van’t Hoff unfolding enthalpies of 36 to 40 kcal /mole at midpoint temperatures in the range of 53 to 58°C. Extrapolation of the thermal denaturation free energies with heat capacities of 400 to 600 cal/mole deg gives good agreement with the parameters determined in denaturant titrations. As predicted from statistical surveys of amino acid replacements in β-hairpins, energetic barriers to transformation from a type I′ turn (DG) to a type II′ turn (GT) can be quite small.

β-Hairpins are common features in proteins (Sibanda et al. 1989) and common targets for the design of minimal peptide models (Searle 2001) of protein folding. Statistical studies of turns found in the database of protein structures show clear conformational preferences for certain sequences (Hutchinson and Thornton 1994). Several studies have identified homologous structures in the protein database that differ in the conformation of β-hairpins (Sibanda and Thornton 1993; Gunasekaran et al. 1998), indicating that conversions can accumulate in evolution without long-range structural consequences. Computational studies have also addressed the relative energetics of hairpin turn conformations (Mattos et al. 1994; Yang et al. 1996). This combination of approaches to understanding β-hairpins as model elements of protein structure may eventually aid in predicting protein topology from sequence (de la Cruz et al. 2002). Here we describe the energetic contributions of three turn sequences to the stability of an engineered monomer of the λ Cro repressor.

A series of Cro variants were constructed in which the intermolecular antiparallel β-ribbon that forms the dimer interface was replaced by an intramolecular β-hairpin (Mossing and Sauer 1990). Each of these engineered monomers has five additional amino acids inserted after position 56 of wild-type (wt) Cro. Codons 56a and 56b were originally synthesized as degenerate codons that encoded, among others, all combinations of residues commonly found in type I′ and type II′ β-hairpins (Sibanda et al. 1989). Codons 56c, 56d, and 56e encode duplicates of amino acids E54, V55, and K56, and replace amino acids normally provided by the opposite subunit in the dimer (Fig. 1 ▶). The level of accumulation of proteins in crude cell lysates correlated well with their sequences’ match to the consensus type I′ and II′ turn sequences. In fact, the low expression levels of proteins containing nonconsensus turn sequence made it difficult to purify them in quantities sufficient for biophysical characterization. Subsequent crystallographic (Albright et al. 1996) and NMR (Mossing 1998) analysis of the Cro K56[DGEVK] (DG) protein confirmed that the engineered turn was of the type I′ class, largely solvent exposed, and relatively independent of the influence of nonsequentially contiguous side chains (Fig. 1C ▶). Apart from the resonances of the turn residues, homonuclear and heteronuclear ¹H-¹⁵N NMR spectra of Cro K56[GGEVK] (GG) and Cro K56[GTEVK] (GT) are nearly superimposable on the DG spectra (data not shown). NH-NH NOE cross-peaks from a 3D ¹⁵N-¹H HMQC-NOESY-HMQC (Ikura et al. 1990) experiment for uniformly ¹⁵N-labeled GT protein shown in Figure 1E ▶ are consistent with the type II′ turn. Strips containing amide resonances of residues K56, G56a, T56b, and E56c (residues i through [i + 3] in the β hairpin nomenclature; Sibanda and Thornton 1991) are drawn from the proton planes indicated at the top of the figure and show ¹⁵N-¹⁵N correlations. The strong cross-peak in each strip is the amide resonance from the diagonal; only T56b and E56c show a cross-peak indicative of a short NH–NH distance. The NH(i + 1) to NH (i + 2) cross-peak that is characteristically present in type I′ turns and absent in type II′ turns (Wüthrich 1986) is not observed. The conformation of the turn in GG has not been characterized.

Figure 1.

Cro dimer interface and engineered monomer hairpin sequence and structure. (A) Dimer interface sequence schematic. Residues 52 to 58 of the wild-type Cro dimer interface are hydrogen bonded (indicated by vertical lines) to residues 52′ to 58′ of a symmetry-related subunit in an antiparallel β-structure. (B) Molscript (Kraulis 1991) diagram of residues 3 to 61 of Cro K56[DGEVK] drawn from the PDB file 1orc (Albright). Residues V55 through P57 are shown in ball-and-stick format with side chain bonds in gray and backbone bonds in white. The residues D56a and G56b are replaced with the dipeptides GG and GT in the two other variants studied. (C) Potential looped structure of residues 52 to 58 in monomeric form of the Cro repressor with F58 occupying an equivalent position to that observed in the dimer. (D) Engineered hairpins of Cro K56[DGEVK], Cro K56[GGEVK], and Cro K56[GTEVK]. The five-residue insertion is indicated in bold type. Hydrogen bonds between residues in the stem of the hairpin from the structure of Cro K56[DGEVK] and presumed to be present in the other two hairpins are indicated by vertical lines. (E) Strips from an HMQC-NOESY-HMQC 3D spectrum of Cro K56[GTEVK]. Residue name and amide proton resonance are indicated at the top of each strip. In this limited spectral bandwidth, 3D experiment the ¹⁵N resonance of K56 is folded back into the spectrum. Dashed lines indicate the single NOESY cross-peak observed for the four turn residues between the amide protons of T56b and E56c.

Recent experiments suggest that a monomeric form of the wt Cro protein may be present, both at equilibrium (Jana et al. 1997) and as a kinetic intermediate in refolding experiments (Satumba and Mossing 2002). Fluorescence characteristics of the F58W protein indicate that the environment of this residue may be very similar in monomeric and dimeric contexts. The most straightforward way to rationalize these results is to infer that there is an alternative conformation of residues 52 to 58 that allows residue 58 to fold back and make intramolecular contacts in the folded monomer as indicated schematically in Figure 1C ▶. Detailed analysis of the folding energetics of the variants reported here, where the insertion has simultaneously stabilized the reverse turn and destabilized the dimer interface, may be useful in understanding the stability of the naturally occurring monomeric form of Cro.

Results and Discussion

Far UV circular dichroism spectra for GG and GT are indistinguishable from the previously published spectrum of DG (Mossing and Sauer 1990). Equilibrium unfolding transitions of DG, GG, and GT were monitored by circular dichroism at 222 nm. Figure 2A ▶ shows the circular dichroism signals of 4-μM solutions of protein at 25°C as a function of the urea concentration. The denaturation profiles of each of the three proteins are similar. As shown in Table 1, GdnHCl denaturation (data not shown) gives slightly larger stabilities than those obtained from urea denaturation. Because Cro is a highly basic protein, the ionic character of the GdnHCl may have a stabilizing effect.

Figure 2.

Chemical and thermal denaturation profiles. (A) Urea denaturation curves for monomeric Cro variants. Circular dichroism signals from 4-μM solutions are plotted as a function of urea concentration. In each of the three panels, data points for proteins with turn sequences DG (open circles), GG (open squares), GT(open down triangles). Fitted curves for the DG (solid), GG (dotted), GT (dashed) proteins are indicated by lines. (B) Thermal denaturation curves for monomeric Cro variants. Circular dichroism signals from 4-μM solutions are plotted as a function of temperature. Symbols are the same as in (A). (C) Stability curves for monomeric Cro variants. Unfolding free energies are plotted as a function of temperature. Symbols are the same as in (A) and (B) except that free energies at 25°C determined with urea as the denaturant are indicated by open symbols, while those determined from guanidine denaturations are indicated by filled symbols.

Table 1.

Thermodynamic parameters from thermal denaturation

Turn sequencea	Tm (°C)	ΔHm (kcal/mole)	ΔGuH2O (T = 25°C) (kcal/mole)
DG	57.7 ± 1.6	38.3 ± 2.5	2.9 ± 0.4
GG	54.8 ± 1.0	39.9 ± 2.0	2.5 ± 0.4
GT	53.4 ± 1.7	35.9 ± 3.2	2.9 ± 0.4

^a Fit with a fixed heat capacity of 500 cal/mole deg. All error ranges represent 95% confidence intervals.

Thermal denaturation profiles of the monomers are shown in Figure 2B ▶. Preliminary thermodynamic parameters for the unfolding transition were obtained by linear extrapolation of the folded and unfolded baselines, calculation of apparent equilibrium unfolding constants as a function of temperature, and estimation of the van’t Hoff enthalpy (ΔH_vH) from the slope of a plot of lnK versus 1/T (Schellman 1987). For the DG protein a T_m of 57°C and ΔH_vH of 40 kcal/mole were estimated. Similar thermal denaturation experiments at pH 3.6, 4.5, and 5.0 (data not shown) gave midpoint temperatures of 37.7, 49.0, and 53.6 and van’t Hoff enthalpies of 31, 34, and 36 kcal/mole, respectively. A plot of ΔH_vH versus T_m yields an estimated ΔC_p of 500 cal /mole deg K. This compares favorably to the ΔC_p observed directly by differential scanning calorimetry of 440 cal /mole deg K (Y. Griko and P. Privalov, unpubl.). The lines in Figure 2B ▶ were generated by fitting the experimental data to the Gibbs-Helmoltz equation as outlined in Materials and Methods, with heat capacities fixed at 500 cal /mole deg K. Fitted parameters are listed in Table 2. In the transition zone, DG (T_m = 56.8°C) is slightly more stable than GG and GT (T_m = 54.8 and 53.4°C, respectively).

Table 2.

Thermodynamic parameters from chemical denaturation

	Urea			Guanidine
Turn sequence	Cm (M)	m	ΔGuH2O (kcal/mole)a	Cm (M)	m	ΔGuH2O (kcal/mole)a
DG	3.5	0.75 ± 0.05	2.66 ± 0.22	1.4	2.14 ± 0.2	2.92 ± 0.35
GG	3.3	0.82 ± 0.07	2.69 ± 0.28	1.6	2.0 ± 0.25	3.26 ± 0.47
GT	3.1	0.78 ± 0.06	2.38 ± 0.23	1.4	1.91 ± 0.18	2.76 ± 0.34

^a ΔG_u^H2O was calculated as described in Santoro and Bolen (1988).

Figure 2C ▶ shows the stability curves of each of the proteins as a function of temperature in the absence of denaturant. Unfolding free energies at 25°C, as extrapolated from guanidine and urea unfolding curves, are plotted as filled and open symbols, respectively. Extrapolation of the free energies determined by denaturant and temperature induced unfolding transitions to a common reference condition of 25°C and 0 M denaturant gave compatible free energies. Although slight variations in the thermodynamic parameters are apparent in individual denaturation profiles, the net change in stability as a result of the sequence changes in the engineered hairpin turns are within 0.2 to 0.4 kcal/mole. These differences are within the error of the determination of the thermodynamic parameters.

Recent studies of β-hairpin stability have focused on isolated hairpin peptides (Searle 2001; Espinosa et al. 2002). Ramirez-Alvarado et al. (1997) compared the intrinsic stability of type I′ hairpin turns that contained DG (36% folded) and GG (~25% folded) sequences in the turn. The free energy difference between the two hairpins can be calculated to be less than 200 cal/mole. This result is similar to the small differences that we report here. Cochran et al. (2001) have recently reported slightly larger stability differences of approximately 0.7 kcal/mole between type I′(NG) and type II′(GN) turns in the context of a hairpin peptide stabilized by cross-strand interactions between tryptophan residues. Although consensus turn sequences for type I′ and II′ turns appear to stabilize the β-hairpins to similar extents, context effects may affect the relative stability in different model systems. Previous studies of alternative turn types in related proteins (Sibanda and Thornton 1993) indicate that energetic barriers to conversion between type I′ and type II′ turns might be small.

The present results are consistent with previous calorimetric studies of wt Cro (Griko et al. 1992; Filimonov and Rogov 1996). The denaturation of the wt Cro dimer is a bimolecular reaction, and thus, stability depends on protein concentration. The stability of the monomers described here is concentration independent, as expected for a unimolecular reaction. The unfolding enthalpy and heat capacity that have been determined for the wt Cro dimer are approximately twice the enthalpy and heat capacity determined here for the engineered monomers, as expected.

Materials and methods

Cro monomers were expressed from tac promoter plasmids pUCro.mDG, pUCro.mGG, and pUCro.mGT (Mossing and Sauer 1990) and purified as previously described (Mollah et al. 1996). The expected molecular masses of the proteins were verified by electrospray mass spectrometry. The measured masses agreed within three mass units of the calculated masses for each of the monomers. All protein concentrations were determined by absorbance at 280 nm using an extinction coefficient of 3990 mole⁻¹ cm⁻¹. NMR sample preparation and experiments were as described for the DG protein (Mossing 1998).

Protein denaturation experiments generally followed the procedures outlined by Pace and Scholtz (1997). Titrations of Cro monomers with guanidine hydrochloride or urea utilized two protein solutions. Solution A had 4 μM of the protein in 20 mM potassium phosphate pH 7.0, 0.1 mM EDTA, and 200 mM KCl. Solution B contained the same concentrations of protein and buffer components and in addition, either 7.5–8.0 M guanidine hydrochloride or 9.0–9.5 M urea. Guanidine hydrochloride and urea concentrations were determined by refractive index (Kawahara and Tanford 1966). Samples were made by adding small amounts of solution B to solution A to obtain the appropriate denaturant concentration. The samples were equilibrated for 2 min and the CD signal was recorded at 25°C, 222 nm, in a 1-cm path length cuvette.

Thermal stability of the monomers was measured by CD, using a 1-cm path length cuvette in an AVIV 62 DS spectrometer. Temperatures were calibrated by monitoring temperatures inside the cell in a test run. The monomers were at 4 μM in 200 mM KCl, 0.1 mM EDTA, 20 mM Potassium Phosphate, pH 7.0. The ellipticity at 222 nm was measured from 5 to 90°C with 2.5°C increments and 2-min equilibration time at each temperature. Also, renaturation experiments (going from 90 to 5°C) were carried out to ensure reversibility. In all cases refolding was more than 80% reversible. Triplicate experiments were done. Additional thermal denaturation experiments with DG were also performed at pH 3.6, 4.5, and 5.0.

Thermodynamic analysis

Both chemical and thermal denaturation profiles were analyzed to extract best-fit parameters for free energy in the transition zone. Preliminary parameter estimates were generated by the following method. Linear parameterization of the temperature or denaturant-dependent signals of the folded (yf) and unfolded (yu) protein were generated from the first and last four points of the data set, respectively.

(1)

(2)

The fraction of the protein unfolded (fu) in the transition zone was calculated from the relationship

(3)

Equilibrium constant (K) and free energy (ΔG) for unfolding were then obtained for data points where fu was between 0.1 and 0.9.

(4)

(5)

For chemical denaturation free energies were modeled as linear functions of the denaturant concentration (Santoro and Bolen 1988).

(6)

In the case of thermal denaturations midpoint, temperatures (T_m) were interpolated and van t’ Hoff enthalpies were estimated from slopes of plots of lnK versus 1/T. The change in heat capacity upon unfolding (ΔC_p) was estimated from the dependence of the van’t Hoff enthalpy as a function of temperature, estimated from thermal denaturation at pH 3.6, pH 4.5, pH 6, and pH 7. Preliminary thermodynamic parameters were used as estimates in full nonlinear least-squares analysis of both temperature and chemical denaturation profiles. For temperature denaturation experiments the Gibbs-Helmholtz equation

(7)

was used to extrapolate the unfolding free energy from the transition zone.

Free energies calculated as a function of temperature (by Equation 7) or denaturant concentration (by Equation 6) were converted to signal intensity by the relationships in Equations 1–5. The sum of the squared differences between the calculated signals and actual signals was minimized in Matlab.

In the fits to both thermal and chemical denaturation transitions there are six adjustable parameters. Four of the six parameters describe the baseline signals (slopes and intercepts for folded and unfolded baseline signals). For thermal denaturation the remaining two parameters are ΔH_m and T_m. ΔC_p was fixed for the least-squares optimization at 500 cal/mole deg K (fixing heat capacities in the range of 400 to 600 cal/mole deg K gave qualitatively similar results). For chemical denaturation profiles m and ΔG were obtained by global analysis (Santoro and Bolen 1988).

Abbreviations

• DG, Cro K56[DGEVK]

• GG, Cro K56[GGEVK]

• GT, Cro K56[GTEVK]

• CD, circular dichroism

• GdnHCl, guanidine hydrochloride