Methionine oxidation of monomeric λ repressor: The denatured state ensemble under nondenaturing conditions

Abstract

Although poorly understood, the properties of the denatured state ensemble are critical to the thermodynamics and the kinetics of protein folding. The most relevant conformations to cellular protein folding are the ones populated under physiological conditions. To avoid the problem of low expression that is seen with unstable variants, we used methionine oxidation to destabilize monomeric λ repressor and predominantly populate the denatured state under nondenaturing buffer conditions. The denatured ensemble populated under these conditions comprises conformations that are compact. Analytical ultracentrifugation sedimentation velocity experiments indicate a small increase in Stokes radius over that of the native state. A significant degree of α-helical structure in these conformations is detected by far-UV circular dichroism, and some tertiary interactions are suggested by near-UV circular dichroism. The characteristics of the denatured state populated by methionine oxidation in nondenaturing buffer are very different from those found in chemical denaturant.

Many small single-domain proteins required for cellular function fold reversibly without the aid of chaperones (Jackson 1998). The starting point for reversible folding is the denatured state within the cell. A detailed description of this denatured state is necessary to understand the thermodynamics and kinetics of protein folding (Shortle 2002). For example, residual structure in the denatured state has been shown to differentially affect the thermodynamic stability of RNase H from Escherichia coli and Thermus thermophilus (Robic et al. 2003). The structural properties of the denatured state also provide clues to early events in protein folding (Shortle 2002). While it is impossible to duplicate the cellular environment in vitro, the denatured state under physiological conditions is most relevant. Traditional methods used to populate the denatured state, including high concentrations of chemical denaturants, high temperatures, and low pH are poor models for denatured proteins in vivo because of the difference in solvent conditions compared to cellular conditions. In this paper we present evidence that the denatured state of the monomeric λ repressor in solvent conditions that are more like the cellular environment is significantly different from the one populated by addition of chemical denaturant.

Amino acid substitutions can be used to locally perturb the energy of a protein, thereby shifting the folding equilibrium to favor the denatured state under conditions that normally favor the folded native state. However, there can be problems with expression or purification of such low stability variants. An in vivo modification that has been observed to destabilize proteins, including calmodulin and thrombomodulin, is oxidation of methionine residues (Gao et al. 1998; Wood et al. 2003). Methionine oxidation involves a reaction between a methionine side chain and a reactive oxygen species (ROS) to yield a methionine sulfoxide. The cycle of methionine oxidation by ROS and reduction by methionine sulfoxide reductase has been implicated in many biological processes including enzyme regulation, antioxidant activity, and aging (Stadtman et al. 2002). The destabilization caused by methionine oxidation in staphylococcal nuclease is quantifiable, and varies depending on the site of the methionine (Kim et al. 2001). In the work presented here, we successfully used this chemical modification to predominantly populate the denatured state of the monomeric λ repressor in nondenaturing buffer at 25°C. Thus, we have created an experimentally tractable model of the physiological denatured state for structural studies.

Denatured proteins have been traditionally modeled as statistical random coils (Tanford 1968). In fact, as shown in a recent review of small-angle X-ray scattering experiments, the radii of gyration of chemically denatured proteins correspond well with the values expected for random coils (Kohn et al. 2004). However, many denatured proteins have been shown to retain a significant amount of native-like structure even in the presence of chemical denaturant (Neri et al. 1992; Logan et al. 1994; Ackerman and Shortle 2002; Shortle 2002). An explanation for this paradox is offered by Fitzkee and Rose (2004), who used simple models to show that global random coil behavior is not inconsistent with a significant amount of local native-like secondary structure. Additionally, using distributed computing to model the denatured state ensembles of five proteins under native conditions, Zagrovic and Pande (2003) showed that the average properties of the denatured ensemble can be native-like even if no individual member of the ensemble is native-like. Neri et al. (1992) and Logan et al. (1994) were among the first to report residual structure in proteins in high concentrations of denaturant. Within the last decade, proteins studied under more physiological conditions have also been shown to have nonrandom structural properties (Alexandrescu et al. 1994; Zhang and Forman-Kay 1995; Ochsenbein et al. 2001; Yao et al. 2001; Mayor et al. 2003; Li et al. 2005). Many methods have been used to detect the structural properties of denatured state ensembles, including NMR distance contraints from NOEs (Neri et al. 1992; Logan et al. 1994; Zhang and Forman-Kay 1995; Yao et al. 2001) and paramagnetic relaxation enhancement (Gillespie and Shortle 1997a,b), residual dipolar couplings (Shortle and Ackerman 2001), and perturbed pKas (Oliveberg et al. 1995; Tan et al. 1995; Forsyth et al. 1998; Kuhlman et al. 1999).

To fully understand the folding process of the N-terminal domain of the λ repressor, we require a model of the starting point for folding: the denatured state under physiological conditions. An appropriate model was achieved using methionine oxidation to destabilize the protein. To improve the solubility of this model, we added back wild-type N terminus (with three additional lysines) and changed two surface hydrophobic residues to lysine to create a more soluble version of the protein. To estimate the magnitude of the destabilization caused by the oxidation, we made a hyperstable variant (Q33L), oxidized its methionines, and forced it to refold in sucrose. To fully populate the denatured state upon methionine oxidation, it was necessary to further destabilize the protein through substitution of Ser77 with Ala. We have found that the methionine-oxidized denatured state in nondenaturing buffer is compact and contains a significant amount of α-helical secondary structure and some tertiary interactions.

Results and Discussion

Design of a soluble physiological denatured state

The monomeric version of the N-terminal domain of λ repressor consists of five helices with the fifth helix truncated to abolish the dimerization interface (Huang and Oas 1995; Burton et al. 1996). This variant contains two native methionine residues at positions 40 and 42 (Fig. 1). These methionines are easily oxidized using a low concentration of hydrogen peroxide to methionine sulfoxide (see Materials and Methods). Oxidation of the methionine side chains is reversible using a reducing agent such as dithiothretol (DTT) (Creighton 1993).

Figure 1.

Ribbon diagram of residues 1–85 of the N-terminal domain of λ repressor based on the crystal structure of Beamer and Pabo (1992). The native methionine residues at positions 40 and 42 are indicated using ball-and-stick figures. Additionally, the residues substituted at positions 54, 56, and 77 are indicated using ball-and-stick figures, and the first five residues are shown in white. The figure was generated using MOLSCRIPT (Kraulis 1991).

Denatured and partially denatured proteins tend to aggregate in nondenaturing solution. When previously studied variants of the monomeric λ repressor (Huang and Oas 1995; Burton et al. 1996) were oxidized, in the absence of denaturant, they aggregated at low micromolar concentrations (J.K. Myers, unpubl.). To increase the solubility of the denatured protein in physiological buffer, we used a variant of the monomeric λ repressor with two classes of alterations. The first was the restoration of the five N-terminal residues, three of which are lysines, found in the full-length λ phage cI gene (Pabo et al. 1982) that had been removed in previous versions of the monomeric N-terminal domain (Huang and Oas 1995). The second was the substitution of two aliphatic residues, Ile 54 and Ala 56, in the loop between helices 3 and 4 with lysine. These residues have side-chain solvent accessibilities of 101% and 44%, respectively, and contribute significantly to the predicted hydrophobicity of this region of the amino acid sequence (Kyte and Doolittle 1982). These modifications yield a protein whose stability is slightly lower that of the original version of the monomeric λ repressor, ΔΔG = 3.6 ± 0.1 kcal/mol (see Supplementary Fig. 1). Sedimentation equilibrium analysis indicates that when oxidized, this variant, henceforth referred to as λ_S, can be well fit by molecular weights between 7300 and 9700 kDa at concentrations from 200 μM to 600 μM (for representative data, see Supplementary Fig. 2). The actual molecular weight is 9390 kDa. These results indicate that the oxidized protein is predominantly monomeric at concentrations suitable for NMR experiments.

Compared to the unoxidized protein, MetO-λ_S (methionine-oxidized λ repressor variants will be referred to hereafter using the prefix MetO) is much less stable. The far-UV circular dichroism spectrum has diminished minima at 208 nm and 222 nm, and circular dichroism-detected urea-denaturation displays no native baseline (see Supplementary Fig. 1).

The ¹H,¹⁵N-heteronuclear single quantum coherence (HSQC) spectrum of MetO-λ_S has broad and missing peaks (Fig. 2). Broad or missing ¹H,¹⁵N-HSQC peaks have been used previously as an indication of a molten globule state with significant secondary structure and little tertiary structure (Redfield 2004). However, our previous kinetic studies of monomeric λ repressor folding indicate that the native and denatured resonances of many residues are in exchange in the intermediate NMR exchange regime. If the rate of exchange between native and denatured states is similar to the difference in chemical shift of the resonances, then the broad peaks indicate significant populations of both native and denatured ensembles. Indeed, we found that the aromatic region of the ¹H NMR spectrum of MetO-λ_S could be simulated using the ALASKA package (Burton et al. 1998) by a simple two-state equilibrium between the native and the denatured states and estimates of 830 sec⁻¹ and 980 sec⁻¹ for the folding and unfolding rate constants, respectively (data not shown). These rate constants lead to an estimate for the native population of 43%, so the modification destabilizes MetO-λ_S enough to significantly populate both native and denatured ensembles.

Figure 2.

600-MHz ¹H-¹⁵N HSQC spectrum of MetO-λ_S in 20 mM phosphate buffer at 25°C and pH 6.

To further depopulate the native ensemble of λ_S, we introduced a previously studied substitution, S77A, which abolishes the buried hydrogen bond between S77 and D14 and destabilizes the 1–92 and 6–85 versions by 1.5 kcal/mol (Marqusee and Sauer 1994; Myers and Oas 1999). The S77A variant of λ_S (referred to hereafter as λ_LS) is destabilized by an equivalent amount, as measured by urea denaturation detected by far-UV circular dichroism signal at 222 nm, which gives a stability of 2.4 ± 0.1 kcal/mol at 25°C (Fig. 3). To measure the folding kinetics of λ_LS, we used NMR lineshape analysis (Burton et al. 1996) with the ALASKA Mathematica package (Burton et al. 1998; Fig. 4). The extrapolated folding rate in the absence of urea is 15,600 ± 1000 sec⁻¹ and the extrapolated unfolding rate is 68 ± 4 sec⁻¹ (Table 1). The extrapolated folding and unfolding rates for the previously studied S77A variant are 62,700 ± 28,000 sec⁻¹ and 7.43 ± 2.9 sec⁻¹, respectively (Myers and Oas 1999). Therefore, the extra five amino acids and the I54K/A56K substitutions decrease the folding rate and increase the unfolding rate. The urea dependence of the folding and unfolding rate constants can be used to determine the β_T value, which describes the fractional decrease in the solvent accessibility of the transition state ensemble relative to that of the native state (Matouschek et al. 1995). The β_T value for λ_LS is 0.77 ± 0.04, which is within experimental error of the β_T value of 0.77 ± 0.2 found by Myers and Oas for the S77A substitution in the WT* variant (Myers and Oas 1999). Therefore, the transition state ensemble of this variant with the extra five residues and lysine substitutions has the same solvent-accessible surface area as the WT* variant transition state, suggesting that the major folding pathway is unchanged by the modifications.

Figure 3.

Urea-induced denaturation followed by circular dichroism at 222 nm of λ_LS (•) and MetO-λ_LS (○) in 20 mM phosphate buffer at 25°C and pH 7. The solid line is a nonlinear least-squares fit to an equation combining the two-state assumption with the linear extrapolation method (Nicholson and Scholtz 1996).

Figure 4.

Plot of kinetic rate constants of folding (•) and unfolding (▪) vs. urea for λ_LS in 20 mM phosphate buffer at 25°C and pH 6. Lines are fits to a two-state model.

Table 1.

Kinetic parameters for folding of λLS determined by NMR lineshape analysis

Kf × 103(sec−1)a	Ku(sec−1)a	−mf	mu	βTb
15.6 ± 1.0	68 ± 4	1.64 ± 0.02	0.48 ± 0.02	0.77 ± 0.04

^a Rates extrapolated to 0 M urea.

^b β_T = (m_f/(m_f −m_u).

While NMR assignments for backbone atoms have been published for λ_6–85 (Huang and Oas 1995), the extra five residues and three substitutions changed the spectrum sufficiently that many assignments could not be transferred. Standard three-dimensional NMR experiments were performed to assign all crosspeaks in the ¹H,¹⁵N-HSQC spectrum (Fig. 5A). The residues whose ¹H and ¹⁵N chemical shifts changed the most from the wild-type spectrum are those that were substituted or are near the sites of sequence change, suggesting that the several substitutions in λ_LS do not significantly perturb the native structure (see Supplementary Fig. 3).

Figure 5.

600-MHz ¹H-¹⁵N HSQC spectrum of λ_LS (A) and MetO-λ_LS (B) in 20 mM phosphate buffer at 25°C and pH 6.

Oxidized λ repressor populates the denatured state

When MetO-λ_LS is oxidized, its far-UV CD spectrum has diminished minima at 208 nm and 222 nm (Fig. 7A, below). Like MetO-λ_S, the urea-induced denaturation monitored by CD signal at 222 nm of MetO-λ_LS lacks a native baseline (Fig. 3). However, in contrast to that of MetO-λ_S, the ¹H,¹⁵N-HSQC spectrum of MetO-λ_LS has sharp peaks with little dispersion (Fig. 5B). Although the native and the denatured resonances are presumably still in the intermediate exchange regime, the S77A substitution decreases the native state population sufficiently to eliminate exchange broadening and sharpen the ¹H,¹⁵N-HSQC cross-peaks.

Figure 7.

(A)Far-UV circular dichroism spectra of λ_LS (•),MetO-λ_LS (○), and λ_LS in 9.7 M urea (▴) in 20 mM phosphate buffer at pH 7 and 25°C. (B) Near-UV circular dichroism spectra of λ_LS (•), MetO-λ_LS (○), and λ_LS in 8.6 M urea (▴) in 20 mM phosphate buffer at pH 7 and 25°C.

The magnitude of destabilization by methionine oxidation

The stability of λ_LS was determined to be 2.4 ± 0.1 kcal/mol. This stability corresponds to a denatured state population of ~1% in the absence of denaturant. However, the oxidized form, MetO-λ_LS, is predominantly denatured under the same conditions, indicating that the destabilization of oxidation, ΔΔG_ox, is quite large. This destabilization is the difference between the standard unfolding free energies of the unoxidized and the oxidized forms. Unfortunately, the unfolding free energy of MetO-λ_LS cannot be quantified using urea-induced denaturation because the curve cannot be fit by the linear extrapolation model (Santoro and Bolen 1988) because it lacks a native baseline. Small osmolytes, such as TMAO, can be used to refold unstable proteins (Baskakov and Bolen 1998). However, at high concentrations of TMAO, MetO-λ_LS is not soluble.

To measure the extent of destabilization caused by methionine oxidation, we made a variant of λ_S that was predominantly native even when oxidized. A previously characterized stabilizing substitution, Q33L, was introduced into the λ_S variant (G.T. Kapp, unpubl.). Although the Q33L substitution increases the stability of λ_S by 2.2 kcal/mol, the urea titration followed by CD signal at 222 nm shows that MetO-Q33L is not fully native in the absence of denaturant at 25°C (Fig. 6A). It is not possible to determine the stability of a partially folded protein from one chemical denaturation curve. Mello and Barrick (2003) recently developed a method to determine the stability of a partially folded protein using multiple denaturation curves collected in the presence of various concentrations of an osmolyte and extrapolating the stability in the absence of osmolyte. Following this approach, we performed urea titrations at several concentrations of sucrose to obtain the stability of MetO-Q33L versus sucrose concentration. The relationship between stability and osmolyte concentration is linear, and the presence of osmolyte seemed to have no effect on the denaturant m value (Fig. 6B), as seen by Mello and Barrick. Therefore, the stability of MetO-Q33L in the absence of osmolyte could be reliably extrapolated to 0 M sucrose.

Figure 6.

Determination of the ΔG_N-D of MetO-Q33L. (A) Urea denaturation curves monitored by CD at 222 nm of MetO-Q33L in 20 mM phosphate buffer at pH 7 and 25°C and increasing sucrose concentrations (from left to right: 0 M, 0.7 M, 0.75 M, 0.8 M, and 0.9 M sucrose). (B) Dependence of unfolding free energy (•) and m-values (○) on sucrose. The data were fitted using weighted linear regression.

Under these conditions, the standard free energy of denaturation (ΔG_N-D) of Q33L is 5.92 ± 0.05 kcal/mol and the extrapolated ΔG_N-D for MetO-Q33L is −0.12 ± 0.05 kcal/mol (see Supplementary Fig. 4). The destabilization caused by methionine oxidation (ΔΔG_ox) in this variant is therefore 5.8 ± 0.1 kcal/mol. Although Lim and Sauer (1991) found that substitution of a polar residue into the hydrophobic core of the N-terminal domain of the λ repressor is highly unfavorable, this degree of destabilization is quite large considering that only two atoms are added to the protein upon oxidation. Assuming that the ΔΔG_ox is equally large in λ_LS, the unfolding free energy of MetO-λ_LS is −3.3 kcal/mol and the native state population is ~0.5% at 25°C.

The energetic effect of oxidation of methionines should be comparable to substituting the methionine with a charged residue because the hydrophobicity of methionine sulfoxide is similar to lysine (Black and Mould 1991). We have determined that this modification has a large thermodynamic consequence. However, the structure of the denatured state, which lacks the native hydrophobic core, should be perturbed only in the immediate vicinity of the Met side chains. The evidence that oxidation does not have a large structural consequence on either the native or denatured states comes from the denaturation titrations in increasing sucrose concentration. The m-value determined for the oxidized protein in sucrose is within experimental error of the m-value for the unoxidized protein. Since the m-value is related to the change in accessible surface area upon unfolding (Myers et al. 1995), this result suggests that the conformational differences between native and denatured ensembles are similar in the unoxidized and oxidized forms of the protein. For this reason, we conclude that the conformational ensemble of MetO-λ_LS is a good model for the unfolded ensemble of the N-terminal domain of the wild-type λ repressor.

Structural properties of MetO-λ_LS are different from urea-denatured λ_LS

The denatured state is an ensemble of conformations that interchange among themselves more rapidly than with the native state ensemble (Zwanzig 1995). The populations of these individual substates can be altered depending on the physical and chemical conditions of the sample. Zhang and Forman-Kay (1995) have shown that the denatured form of the marginally stable N-terminal SH3 domain of drk has very different conformational properties in physiological buffer than it does in 2 M guanidine chloride. Similarly, we have found that λ_LS denatured with urea has different properties from MetO-λ_LS.

MetO-λ_LS is more compact than a random coil. The Stokes radius of a molecule is related, through its frictional coefficient, to its sedimentation velocity. Sedimentation coefficients (s) for λ_LS and MetO-λ_LS were determined by sedimentation velocity analytical ultracentrifugtion experiments. The sedimentation coefficients determined for λ_LS and MetO-λ_LS are 1.09 ± 0.05 S and 1.01 ± 0.02 S, respectively (see Table 2; Supplementary Fig. 5). Any concentration dependence of the sedimentation coefficient was within the experimental error. Using the crystal structure of residues 1–85 of the λ repressor (Beamer and Pabo 1992) we estimate an R_h for the native protein of 18.6 Å (Garcia De La Torre et al. 2000); and the ratio of sedimentation coefficients predicts an R_h of 20.1 Å for the oxidized unfolded protein. The compact R_h of the MetO-λ_LS ensemble indicates that this form of the denatured protein is significantly different from the form under chemical denaturation conditions, for which the predicted Stokes radius would be 27.8 Å (Wilkins et al. 1999).

Table 2.

Hydrodynamic measurements from sedimentation velocity

Protein concentration (μM)	s20,w for λLS (S)	s20,w for MetO-λLS (S)
100	1.069	1.029
200	1.148	0.998
400	1.048	1.014
Average	1.09 ± 0.05a	1.01 ± 0.02a

^a Errors are standard deviations from all concentrations.

The increase in R_h for the oxidized protein seems small, but corresponds to a volume of 7061 ų, sufficient to accommodate 230 additional water molecules, based on the density of water. That the physiological denatured state is well solvated even though it is compact is also supported by the low dispersion in chemical shift seen in the ¹H,¹⁵N-HSQC. The lack of dispersion indicates that each residue experiences a very similar solution environment implying that each residue is well solvated (Yao et al. 1997).

The compactness of the oxidized denatured state can be explained by the secondary and tertiary structure detected by circular dichroism spectroscopy. Far-UV circular dichroism is used to measure secondary structure in proteins (Sreerama and Woody 2004). The magnitude of the signal at 222 nm can be used to estimate a fractional helicity (Luo and Baldwin 1997). The far-UV CD spectrum of MetO-λ_LS has a significantly stronger signal at 222 nm, −4500 deg·cm²·dmol⁻¹, than the signal of λ_LS denatured in 8 M urea, −1000 deg·cm²·dmol⁻¹ (Fig. 7A). Since CD is an ensemble average technique, it is not possible to determine whether some parts of the protein remain highly α-helical in the denatured state or if all parts of the protein sample α-helical conformations. In either case, the CD spectrum of MetO-λ_LS is consistent with significant bias toward α-helical backbone conformations in this denatured state ensemble.

The signal in the near-UV circular dichroism spectrum arises from an optically active environment around the aromatic side chains (Sreerama and Woody 2004). Whereas the urea-denatured protein has almost no CD signal in near-UV, the CD signal from MetO- λ_LS is almost half that of native λ_LS (Fig. 7B). This significant CD signal in the near-UV region indicates that the aromatic side chains are not fully solvated in MetO-λ_LS. Our conclusion from the sedimentation experiments and circular dichroism measurements is that the denatured state populated in nondenaturing buffer by methionine oxidation is compact due to secondary structure formation and collapse that may be due to native-like or nonnative like tertiary structure. The solvent-dependent differences in structural properties we detected in the denatured state ensemble highlight the importance of using buffers that mimic the important properties of the cellular environment.

Conclusions

Our goal in using methionine oxidation to destabilize the N-terminal domain of the λ repressor was to develop a method that would populate a realistic denatured state ensemble in a physiologically relevant buffer while minimally perturbing the protein. Sample conditions previously used to populate denatured states including high concentrations of denaturant, low pH, and extremes of temperature may produce distorted denatured state ensembles whose substate populations may differ greatly from the physiological ensemble. Previous structural and dynamic studies on nonnative proteins under nondenaturing conditions have focused on truncation variants (Alexandrescu et al. 1994) and marginally or completely unstable ligand-binding domains (Zhang and Forman-Kay 1995; Ochsenbein et al. 2001; Yao et al. 2001). Oxidation of buried methionines provides a method to populate the denatured state of any stable protein under nondenaturing conditions. The initial structural characterization of the denatured state of the monomeric N-terminal λ repressor shows that under physiological conditions, the ensemble is populated by conformations that are significantly different from those in chemical denaturant. As the starting point for protein folding, it is essential to understand the features of this denatured state and its role in the early events in folding.

We believe that this method for unfolding proteins in physiological buffer will be applicable to many proteins. We have successfully produced a tractable denatured state model through judicious addition of lysine residues, which should be a generally applicable approach. For proteins with no native methionines, substitution of appropriate buried residues with methionine may be effective. Methionine side chains occupy roughly the same volume as leucine, isoleucine, and phenylalanine side chains. In fact, Gassner et al. (1996^{, 2003)} were able to substitute up to 10 hydrophobic residues with methionine in the core of T4 lysozyme without altering its native structure or activity. Leu was used as a stability neutral substitution for Met by Lim and Sauer (1991) in the N-terminal domain of the λ repressor. Leu and Ile side chains were used as substitutions for Met in staphylococcal nuclease by Kim et al. (2001). Therefore, methionine could be reasonably substituted for a buried leucine or isoleucine residue. However, oxidation of different positions in a protein will affect its stability differently. Therefore, it would be necessary to test multiple sites for their effectiveness. We are currently in the process of applying this method to other proteins in our laboratory.

Materials and methods

Mutagenesis, overexpression, and protein purification

Substitutions were introduced using QuikChange into the N-terminal domain of the λ repressor containing residues 1–85 and the stabilizing G46A and G48A substitutions cloned into the pET9a vector (Novagen) using the EcoRI and BamHI sites. Each substitution was then verified by DNA sequencing of the entire gene and promoter region.

λ_1–85 variants were expressed in BL21(DE3) (Stratagene) cells in 1 L of rich media (20 g/L bactotrytone, 10 g/L yeast extract, 0.4 g/L glucose, 0.36 g/L NH₄Cl, 2.16 g/L Na₂HPO₄, 1.08 g/L KH₂PO₄, 5.18 g/L NaCl) at 37°C to an A600 of 0.8. Expression was induced by addition of IPTG to a concentration of 0.4 mM. The cells were incubated at 37°C for a further 3 to 4 h before harvesting by centrifugation, and resuspended in lysis buffer (50 mM Tris-HCl, 10 mM EDTA at pH 8.0).

λ_1–85 variants were purified by a modified version of an earlier protocol (Burton et al. 1996). Cell lysates were generated in a French pressure cell at 12,000 lb/in². The lysates were run through a DEAE Sephacel column (Amersham Biosciences) to remove cell debris and other proteins. The flow-through was collected and applied directly to an Affigel Blue (BioRad) column that was run in phosphate buffer at pH 7. The protein was eluted from the column in 300 mM KCl. The eluted fraction was dialyzed against degassed distilled H₂O overnight, then lyophilized. The lyophilized protein was redissolved in distilled and deionized H₂O and further purified by size-exclusion chromatography on a Sephadex G-50 column (Amersham Biosciences). The final purity of each protein was checked by reverse-phase HPLC and electrospray mass spectrometry. For each protein, the MS-determined mass was within 1 AMU of the expected mass, further confirming the proper amino acid substitutions. Both HPLC and mass spectrometry indicated no significant oxidation of the methionine residues. All protein concentrations were determined using the method of Edelhoch (1967).

Protein oxidation

Lyophilized protein was hydrated in deionized water to a concentration of 1 mg/mL. To the protein solution was added ~20 μL/mL 0.1 M perchloric acid to bring the pH to <3 and 30% hydrogen peroxide to a final concentration of 0.05%. This solution was incubated on the bench for 45 min before lypohilization. Methionine oxidation to sulfoxide was checked by reverse-phase HPLC and electrospray mass spectrometry.

Circular dichroism

Samples for circular dichroism consisted of 10–50 μM protein in a buffer of 20 mM sodium phosphate and 100 mM NaCl at pH 7.0. Far-UV spectra were collected using an Aviv 202 circular dichroism spectrometer with 10 μM protein in a 1-cm pathlength thermostated quartz cell maintained at 25°C, and were signal averaged at each wavelength for 3 sec. Near-UV spectra were collected with 50 μM protein in a 10-cm pathlength jacketed quartz cell maintained at 25°C by a circulating water bath and were signal averaged for 10 sec. The equilibrium stability of the monomeric λ repressor was measured by automated urea titration with the CD spectrometer interfaced to a Hamilton Microlab 500 titrator. A titrant solution was made with urea (Nacalai Tesque, Kyoto, Japan). The urea concentrations of the starting sample and titrant solution were determined from refractive index measurements (Pace 1986). The CD sample was maintained at a constant temperature of 25 ± 0.1°C throughout the titration. At each titration point, the urea concentration was increased by the titrator, the sample was mixed for at least 30 sec, and the signal at 222 nm was averaged for 30 sec. Urea titrations followed by CD were fit to an equation (Nicholson and Scholtz 1996) combining the two-state assumption and the linear extrapolation method (Santoro and Bolen 1988).

NMR lineshape analysis

Twenty-one samples were prepared containing ~300 μM protein, 20 mM potassium phosphate, pD 6.0 (uncorrected pH meter reading), 100 mM NaCl, 1 mM sodium azide, 1 mM TMSP, in 99% D₂O, and various concentrations of deuterated urea. The urea concentration of each sample was determined from refractive index measurements (Pace 1986). ¹H-NMR spectra of protein in various concentrations of urea were recorded at 25°C on a Varian Unity 500 MHz with 256 scans, and a spectral width of 6500 Hz and 8192 points. The pulse sequence included water and urea saturation. The aromatic regions of the spectra were analyzed with the ALASKA package as described previously to obtain folding and unfolding rates (Burton et al. 1998). Uncertainties are obtained from the global fit.

NMR spectroscopy

All three-dimensional NMR spectra were collected at 25°C on a Varian Unity 600 spectrometer with a triple resonance z-shielded gradient probe on samples at concentrations of 0.6–1.2 mM in 20 mM sodium phosphate buffer at pH 7. Two-dimensional gradient-enhanced sensitivity-enhanced ¹H,¹⁵N-HSQC experiments (Kay et al. 1992) were recorded with a spectral width in the ¹H dimension of 7000 Hz and 1024 complex points and a spectral width in the ¹⁵N dimension of 1500 Hz and 128 complex points. Native state resonance assignments were made by using a suite of triple resonance experiments including HNCO (Grzesiek and Bax 1992), HNCACB (Wittekind and Mueller 1993), and HN(CO)CACB (Yamazaki et al. 1994). In all of the three-dimensional NMR spectra of the native state the spectral width in the ¹H dimension was 8000 Hz with 1024 complex points. The spectral width in the ¹⁵N dimension was 1500 Hz with 34 or 37 complex points. The spectral width in the ¹³C dimension was 3000 Hz with 40 complex points for the HNCO. The spectral width was 10,000 Hz with 70 complex points for the HNCACB and HN(CO)CACB. All NMR spectra were processed using NMRPipe (Delaglio et al. 1995). Three-dimensional spectra were analyzed using the NMRView software (Johnson and Blevins 1994).

Sedimentation equilibrium and sedimentation velocity

Sedimentation equilibrium analysis was performed at 25°C at 20,000 and 30,000 rpm using a Beckman OptimaXL-A analytical ultracentrifuge equipped with a 60Ti rotor and six channel centerpieces. Protein samples were in 20 mM phosphate buffer (pH 6.0). Buffer alone was used in the reference cell positions. Cells were scanned at 6-h intervals at 250 nm until consecutive scans (typically three) were unchanged and the system was judged to be at equilibrium (24 h). The data were fitted by the Ideal-1 program (Beckman Instruments) using a value of 0.75 for the partial specific volume of the monomeric λ repressor based on the amino acid composition using the program SEDNTERP. The density of the solvent was 1.001, as calculated by the program SEDNTERP.

Sedimentation velocity analysis was performed at 20°C at 50,000 rpm using a two-channel centerpieces. Protein samples were in 20 mM phosphate buffer (pH 6.0). Buffer alone was used in the reference cell positions. Scans at 280 nm were taken at 15-min intervals. Data were fit by second moment analysis (Beckman program) to obtain sedimentation coefficients.

Electronic supplemental material

The supplementary materials consist of five figures: Supplementary Figure 1, urea-induced denaturation of λ_S and MetO-λ_S followed by circular dichroism at 222 nm; Supplementary Figure 2, sedimentation equilibrium analysis of MetO-λ_LS; Supplementary Figure 3, chemical shift differences between previously published WT* and λ_LS variants of monomeric λ repressor; Supplementary Figure 4, urea-induced denaturation of Q33L followed by circular dichroism at 222 nm; Supplementary Figure 5, sedimentation velocity analysis of λ_LS and MetO-λ_LS.

Abbreviations

• CD, circular dichroism

• HSQC, heteronuclear single quantum coherence

• R_h, hydrodynamic radius

• λ_S, monomeric λ repressor (residues 1–85)

• ROS, reactive oxygen species

Article published online ahead of print. Article and publication date are at http://www.proteinscience.org/cgi/doi/10.1110/ps.051856406.