Thermal unfolding of ribonuclease A in phosphate at neutral pH: Deviations from the two–state model

Abstract

The thermal denaturation of ribonuclease A (RNase A) in the presence of phosphate at neutral pH was studied by differential scanning calorimetry (DSC) and a combination of optical spectroscopic techniques to probe the existence of intermediate states. Fourier transform infrared (FTIR) spectra of the amide I′ band and far-uv circular dichroism (CD) spectra were used to monitor changes in the secondary structure. Changes in the tertiary structure were monitored by near-uv CD. Spectral bandshape changes with change in temperature were analyzed using factor analysis. The global unfolding curves obtained from DSC confirmed that structural changes occur in the molecule before the main thermal denaturation transition. The analysis of the far-uv CD and FTIR spectra showed that these lower temperature–induced modifications occur in the secondary structure. No pretransition changes in the tertiary structure (near-uv CD) were observed. The initial changes observed in far-uv CD were attributed to the fraying of the helical segments, which would explain the loss of spectral intensity with almost no modification of spectral bandshape. Separate analyses of different regions of the FTIR amide I′ band indicate that, in addition to α-helix, part of the pretransitional change also occurs in the β-strands.

Ribonuclease A (RNase A; Fig. 1 ▶) is a small, single-domain protein (124 amino acid residues) that catalyzes the hydrolysis of single-stranded RNA. Although it is one of the most extensively studied proteins, its thermal unfolding mechanism remains obscured by seemingly contradictory reports that are based on use of specific techniques for monitoring conformational change. The many approaches to the problem generate different answers depending on the conformational sensitivities of the method.

Fig. 1.

Ribbon diagram of RNase A (PDB entry 3rn3) showing the Tyr residues and the disulfide bonds.

Previous Fourier transform infrared (FTIR) studies, in which the change was monitored at constant frequency (usually 1632 cm⁻¹; Yamamoto and Tasumi 1991; Takeda et al. 1995; Backmann et al. 1996; Reinstadler et al. 1996), and circular dichroism (CD) studies based on single-frequency analysis (Denisov and Halle 1998) or bandshape analysis (Labhardt 1982) using an external basis set indicated that the thermal unfolding at acidic and neutral pH is cooperative and two-state. In contrast, local pretransitional changes, some attributed to stable intermediates, were observed by ¹³C-NMR (Howarth and Lian 1984), ¹H-NMR (Westmoreland and Matthews 1973; Benz and Roberts 1975; Zhang et al. 1995), density and sound velocity measurements (Tamura and Gekko 1995), laser Raman spectroscopy (Chen and Lord 1976), and calorimetry (Tsong et al. 1970).

Thus, improving the analytical procedure so that more subtle variations can be detected in the response to an environmental perturbation should help resolve this question. Because the information content is spread over a spectral region, the most sensitive analyses use variation of the entire bandshape in that region to identify all the spectral changes resulting from an environmental perturbation that can be indicative of a structural change.

The inherently fast timescale of optical spectroscopic measurements makes them suitable for the study of unfolding intermediates, which are often unstable and, therefore, might not be observed with slower techniques like nuclear magnetic resonance (NMR) or X-ray. The problem with optical spectra is lack of resolution of those features assignable to the structural characteristics. To generate more discrimination, it is best to combine data from various spectroscopic techniques that are sensitive to different aspects of the protein structure. Consistent treatment of the samples and the resulting spectra can yield complementary sets of structural information whereby the weaknesses of one technique can be compensated by the strengths of another (Pancoska et al. 1995; Baumruk et al. 1996).

In conjunction with optical spectroscopy, differential scanning calorimetry (DSC) can provide direct measurement of small structural transitions and provide characterization of their thermodynamic aspects. Improvements in DSC instrumentation in the past decade have led to determinations of differential heat capacities on the order of μcal/K mL (Privalov et al. 1995) with small samples.

This paper presents a systematic spectroscopic and calorimetric study of thermal unfolding of RNase A in the presence of phosphate. For this study, we use a novel combination of mathematical treatments of temperature-dependent spectral changes. Following a conventional use of the principal component method of factor analysis to obtain spectral components, the component loadings are parsed into intensity and bandshape contributions and expressed as thermal denaturation curves. These revealed a pretransition at temperatures lower than 50°C, best seen in the far-uv CD and FTIR spectra, which are, in turn, most sensitive to secondary structure changes. The pretransition is confirmed in the DSC-derived thermal variations of the enthalpy and entropy of unfolding.

Results

The solvent-corrected FTIR and far- and near-uv CD spectra of RNase A at different temperatures are presented in Figure 2 ▶ A–C, respectively. Arrows indicate the directions of the intensity changes with increasing temperatures. The spectra all show a distinct change for temperatures between 50°C and 70°C, after remaining fairly consistent, aside from intensity, up to 50°C. The far-uv CD or FTIR do not have well-defined isodichroic or isosbestic points, respectively. The sets of temperature-dependent spectra from FTIR and far- and near-uv CD were each analyzed for the full spectral regions shown in Figure 2 ▶ A–C by the principal component method of factor analysis (PC/FA). The first two component spectra (factors) describe more than 99% of the total bandshape change and, therefore, became the focus of our attention.

Fig. 2.

Temperature variation in the experimental spectra: far-uv CD (A), near-uv CD (B), and FTIR (C). The arrows indicate the directions of change with increasing temperature.

The temperature dependencies of the respective loadings of these factors are presented in Figure 3 ▶; the reduced loadings and norms, in Figure 4 ▶. Because the loadings characterize changes over the whole frequency range, their temperature dependencies reflect changes in the structure over the temperature range. Thus, to the extent the technique can sense the structural change, it is encompassed in the data.

Fig. 3.

Thermal dependence of the PC/FA loadings C_ij: far-uv CD (A), near-uv CD (B), and FTIR (C). The symbols are as follows: (circles) C_i1, (squares) C_i2.

Fig. 4.

Thermal dependence of the PC/FA reduced loadings C^r_ij and norms N_i: far-uv CD (A), FTIR (B). The symbols are as follows: (circles) C^r_i1, (squares) C^r_i2, (diamonds) N_i.

The first loadings in both far-uv CD and in FTIR (Fig. 3 ▶ A,C) displayed biphasic temperature dependencies and were fit to a sum of two sigmoid functions. Only a single sigmoid function (with a linear correction) was required to fit both near-uv loadings. Similarly, the second far-uv loading was fit to one sigmoid. However, the second FTIR loading required a closer analysis to decide between a double sigmoid fit and a sum of a linear and sigmoid function. The fit to a double sigmoid function was found to be better, based on the analysis of the variance and the residuals. The values for the regression coefficient r² for all the fits were 0.99 or better.

In far-uv CD, the reduced loading plots (Fig. 4A ▶) show the same transition but indicate that the biphasic shape is more characteristic of the norms rather than of the C^r_i1 or the bandshape-sensitive reduced loading. In the FTIR (Fig. 4B ▶), both C^r_ij and the norms display pretransitional changes.

Integration of the excess molar heat capacity (Fig. 5A ▶) for the thermal denaturation of RNase A according to Equations 6 and 7 permitted the calculation of ΔH_u(T) and ΔS_u(T) (Fig. 5B,C ▶). Both curves show a biphasic temperature dependence as found in the spectral T-dependent curves.

Fig. 5.

(A) Excess heat capacity of RNase A and temperature dependence of the thermodynamic unfolding parameters, (B) ΔH_u, and (C) ΔS_u.

The thermal denaturation was not completely reversible. Test experiments were run in which the protein was measured at 25°C before and after heating to various intermediate temperatures to see if the irreversibility of the unfolding would affect the low-temperature changes observed. When the protein was heated up to 55°C, the CD spectra before and after heating were completely superimposable, indicating reversibility below 55°C, which encompasses the region of the pretransition. However, as the intermediate temperature was increased above 60°C, small differences in the intensity of the two CD spectral traces measured at 25°C (before and after heating) were observed. Prolonged exposure (several hours) to high temperatures (between 65°C and 75°C) resulted in more significant bandshape differences.

The irreversibility of the denaturation of RNase A is strongly influenced by the presence of phosphate. Calorimetry experiments run in water or in MOPS buffer (data not shown) showed about 27% to 30% irreversible denaturation after the first heating to 85°C, compared with approximately 65% in phosphate buffer. This was judged by the enthalpy of denaturation associated which each heating step. The amount of irreversibly denatured protein depends on the amount of time spent at high temperatures. In far-uv CD, in which experiments were longer and the protein was subjected to high temperature for a longer time, the spectrum of the protein after cooling down was similar to that of the denatured protein at 75°C. In FTIR, the experiments were shorter and concentrations were higher so that partial recovery of the low-temperature spectrum was observed. Despite the irreversibility of the denaturation, the protein solutions were still clear at the end of the experiments and no increase was observed in the FTIR between 1610 and 1620 cm⁻¹, which would indicate aggregation (Jackson and Mantsch 1995).

Discussion

Reversibility of the thermal denaturation of RNase A

In our test experiments, no aggregation or irreversible unfolding occurs up to 55°C for RNase A in phosphate at pH 6.8; even though at higher temperatures and with longer exposure, there is significant irreversibility. These results confirm earlier observations regarding the reversibility of the thermal unfolding of RNase A at neutral pH (Hermans and Scheraga 1961; Tsong et al. 1970; Chen and Lord 1976). Thus, although our complete temperature courses in the unfolding experiments maintained the protein solutions at temperatures above the melting point for approximately 2 hr in FTIR and 3 to 4 hr in CD, irreversibility had no impact on the shape of the unfolding curves in the pretransition (below 55°C) region.

The influence of phosphate on the irreversibility of the thermal denaturation of RNase A seems to be thermodynamically rather than kinetically controlled. The irreversible thermal denaturation of RNase A at pH 6 to 8 has been proposed (Zale and Klibanov 1986) to be caused by a combination of processes: disulfide interchange, deamidation of Asn and Gln residues, and β-elimination of cystine residues, with disulfide scrambling playing the main role in this pH range. Disulfide interchange requires the presence of a reducing agent, free thiols being appropriate sources. The Zale and Klibanov experiments led to the conclusion that free thiols can be generated from proteins, including RNase A, by base catalyzed β-elimination of cystines. To check the possibility of a catalytic effect of phosphate on the β-elimination, calorimetry experiments were run on same-concentration RNase A solutions in 10 mM and 1 mM phosphate at pH 6.8 in which the ionic strength was adjusted to the same value with NaCl. In both experiments the same amount of irreversible denaturation was observed (approximately 65%), ruling out the hypothesis of a phosphate-dependent kinetically controlled process.

In the main transition, the melting temperature observed with the FTIR data was a few degrees higher than that obtained with CD and calorimetry data. This may be attributable in part to an isotope effect caused by deuteration (Makhatadze et al. 1995), and it may also depend on experimental differences (e.g., concentration) in the techniques. However, whatever the source of these variations in the main transition temperature, the major conclusion of the paper regarding the pretransition is unaffected.

Thermal unfolding of the tertiary structure

RNase A has no tryptophan, but it has six tyrosine residues, which experience different environments: Tyr 25, 92, and 97 are buried, whereas Tyr 73, 76, and 115 are exposed (see Fig. 1 ▶) and four cystines. Tyr 25, 73, 76, and 115, together with the disulfide bonds are reported to have the major contribution to the near-uv CD spectra (Horwitz et al. 1970; Kurapkat et al. 1997). The contributions of the other two tyrosines and the phenylalanines are very small.

The near-uv CD loadings result in a relatively sharp, sigmoidal transition, which is consistent with the tertiary structure of RNase A unfolding in a two-state fashion. These spectra monitor changes in the environment and local stereochemistry of Tyr residues but do not directly detect secondary structure. Thermally induced local structural changes not affecting the Tyr 25, 73, 76, and 115 residues would also not be sensed by near-uv CD, which would, in that case, sense only the (major) loss of tertiary structure. The initial linear temperature dependence can be attributed to the temperature dependence of the extinction coefficient (Kodama et al. 1997) and/or to slight modifications of the broad disulfide band (Horwitz et al. 1970).

Similar melting curves to those resulting from the near-uv CD data (Fig. 3B ▶) were obtained in previous RNase A denaturation studies monitored by uv absorption (Hermans and Scheraga 1961; Arnold et al. 1996), fluorescence (Sendak et al. 1996), ir (Reinstadler et al. 1996), and laser Raman spectroscopy (Chen and Lord 1976), all focused on Tyr chromophores. As pH increased, the transition broadened and the denaturation temperature increased (Hermans and Scheraga 1961).

Thermal denaturation of the secondary structure

In contrast with the above results, the denaturation curves from calorimetry, as well as the far-uv CD C_i1 and FTIR C_i1 plots, provide evidence for the existence of pretransitional changes.

The calorimetry melting curves (Fig. 5B,C ▶) describe overall pretransitional conformational changes. The excess unfolding enthalpy and entropy corresponding to the intermediate structure represent about 25% and 26% of the total unfolding values (Table 1). A ΔH_u to ΔH_vH ratio of one is often held to be indicative of a two-state process (Tsong et al. 1970; Privalov and Khechinashvili 1974; Freire and Biltonen 1978; Sturtevant 1987). In our case, this value was 1.3 (Table 1). In a calorimetry study of the thermal unfolding of RNase A at different pH values, deviations from the two-state model also occurred at pH values larger than 5 (Tsong et al. 1970).

Table 1.

Thermodynamic parameters of the thermal unfolding of RNase A at pH 6.8

	ΔHu Kcal/mol	ΔSu Kcal/mol K	TM (°C)	ΔHvH Kcal/mol	ΔHu/ΔHvH
Pre-transition	37.7 ± 0.8a	0.118 ± 0.002a	47.4 ± 0.2c	—	—
Main transition	113.7 ± 0.7a	0.338 ± 0.002a	63.8 ± 0.02c	—	—
Overall	149.8 ± 7.9b	0.444 ± 0.023b	—	115.05d	1.30

^a The corresponding values were determined from the height of the two sigmoid functions from the fit curves. The errors represent standard errors of the fit parameters relative to the experimental data.

^b Experimental values obtained from DSC.

^c Average values from the double sigmoid fits of the thermodynamic parameters.

^d Value calculated using Equation 7 (see text).

When divided into C^r_i1 and N_i, the pretransition seen in the far-uv CD loading C_i1 appears to derive from an intensity contribution, having little effect on the bandshape (Figs. 3A, 4A ▶ ▶). In the far-uv region, the protein spectra have characteristic patterns for dominant secondary structural types such as α-helix and β-sheet that are recognizable in model peptides (Greenfield and Fasman 1969), as well as α- or β- dominant proteins (Johnson Jr. 1988). However, the α-helical contribution to the CD is much more intense than that of the β-sheet and overlaps it spectrally, which leads to an insensitivity to β- and other structures if the protein has a significant helical content (Pancoska et al. 1995; Venyaminov and Yang 1996). Thus, although RNase A is an α/β protein with more sheet than helical residues, its low-temperature far-uv CD spectrum (Fig. 2A ▶) has a similarity in shape (but not in intensity) with the standard α-helix far-uv CD. In our PC/FA analysis, this characteristic was primarily distributed into the first spectral component (data not shown). Hence, the temperature dependence of the corresponding loading (C_i1) predominantly describes the unwinding of the helices.

The far-uv CD intensity is mostly dependent on the fractional amount of α-helical residues in the protein. Consequently, the pretransitional spectral intensity change can be attributed to a loss of helical residues. This could involve all the helices from its initial onset (low temperature), resulting in many microstates, or one helix could unwind preferentially before the others. Noncooperative unraveling of helices resulting in a broad distribution of fluctuating helical segments of different lengths would skew the low-temperature region of the denaturation curves (Schellman 1958), but no new transition would be observed. However, we see a relatively well-defined low-temperature transition (Fig. 3A ▶), which indicates a stable intermediate with a somewhat narrow distribution of microscopic states. In other words, the initial change must be fairly localized (i.e., probably involving only one helix). The initial unraveling is substantial but does not collapse the helices, and hence, the CD bandshape does not undergo a major change in this process. At higher temperatures, in the main transition, the other helices also begin to unwind as the far-uv CD spectra change to a shape more characteristic of random coil structures (but probably mixed with other residual local structures).

However, in FTIR, the pretransition affects both the spectral intensity and bandshape (Figs. 3C, 4B ▶ ▶). In our FTIR experiments, both the reduced loadings and the norm contribute to the pretransition (Fig. 4B ▶), revealing not only a quantitative but also a qualitative change. Analysis of the principal spectral components shows that the first one consists of a broad band covering the whole amide I′ region. The second spectral component has two major peaks, one in the region corresponding to β-sheet (∼1634 cm⁻¹) and the other in a region more characteristic of turn structures (∼1665 cm⁻¹). Both loadings corresponding to these components display pretransitional changes, but the changes in the first component are somewhat clearer.

To seek more structure specific information, especially with respect to the denaturation of β-sheet, the major structural component of RNase A, the amide I′ spectra, were divided into two regions based on the major features observed in the difference spectra: 1616–1640 cm⁻¹ and 1653–1682 cm⁻¹. Although the major FTIR amide I′ spectral change is in the former region, there are significant changes in the latter as well. The double sigmoid feature observed in the denaturation curves from the whole band analysis is still observed in the 1616–1640 cm⁻¹ region (Fig. 6 ▶), although the pretransition is weaker as compared to the full-region analysis (Fig. 3A ▶). If the region is expanded to 1616–1649 cm⁻¹, virtually the same thermal behavior is seen. However, if it is instead restricted to the 1616–1636 cm⁻¹ region, the pretransition is weaker but still evident in C_i1. In the second, higher frequency, region (curves not shown), the magnitude of the spectral change is much smaller (see Fig. 2C ▶) and signal-to-noise ratio (S/N) is poor, making the fitting less reliable. An attempt to band fit all the spectra in the temperature set was performed to separate underlying overlapped components, but a consistent decomposition could not be found over the temperature range studied.

Fig. 6.

Temperature dependence of the PC/FA loadings C_ij (1616–1640 cm⁻¹). The symbols are as follows: (circles) C_i1, (squares) C_i2.

In conclusion, the pretransition observed in the analysis of the whole amide I′ is mostly generated in conjunction with the major structural and spectral changes that are predominantly in the 1616–1649 cm⁻¹ region. This overall region contains contributions from the β-sheet structures. However, the pretransition contributes most to the region between 1636 and 1649 cm⁻¹, which is above the normal β-sheet region and has additional contributions from other components. These latter could include contributions from unordered structures and hydrated and distorted helices, which would be consistent with the far-uv CD result.

Proposed mechanism for the pretransitional structural changes

Because of their non–site-specific sensitivity, our methods do not allow us to pinpoint the location of the pretransitional changes in the primary sequence. However, a combination of previous observations indicates helix II (Asn24-Asn34) and β-sheet segment 43–49 as the more labile fragments of RNase A, as rationalized below.

H/D exchange experiments (Santoro et al. 1993; Wang et al. 1995) showed that helix III and the β-sheet are the most protected and, therefore, most stable regions of the molecule. Helix I is located in the center of the molecule packed against the adjacent β-structures (PDB entry 3rn3). The presence of phosphate is expected to stabilize at least the second half (C-end) of helix I, which contains four phosphate binding sites (de Llorens et al. 1989; Fontecilla-Camps et al. 1994; Nogues et al. 1995; Cuchillo et al. 1997; Fisher et al. 1998; Raines 1998). A disulfide and a H-bond confer stability to the N terminus of helix II, but its C terminus connects to a loop and seems susceptible to modification without altering any Tyr residues. Although unwinding of the N terminus of helix I is possible, to the best of our knowledge, it is not supported by literature data. Unwinding of helix II is also in accord with an earlier proposed mechanism (Burgess and Scheraga 1975) and supported by the pretransitional modifications previously observed (Arnold et al. 1996; Chen and Lord 1976; Howarth and Lian 1984). Pretransitional changes of the β-sheet have also been reported from FTIR (Panick and Winter 2000), NMR, for strand 43–49 (Benz and Roberts 1975; Howarth and Lian 1984), and proteolysis data, strand 43–49 (Arnold et al. 1996). Changes in β-strands 96–111 and 116–124 were also reported (Howarth and Lian 1984), but the regions 106–111 and 116–119 were later found to be particularly stable to H/D exchange (Wang et al. 1995), bringing those proposed changes into question.

It is reasonable that the pretransitional changes we saw in our spectra are determined by the unwinding of the C terminus of helix II correlated with the modification of β-strand 43–49. These residues occupy a relatively central region in the three-dimensional structure, sufficiently remote from the four Tyr residues having major contributions in the near-uv CD so as not to influence their environment.

Structural modifications in RNase A were observed in the calorimetry and far-uv CD experiments (Figs. 5A, 3A ▶ ▶), starting just above 25°C. In FTIR, the lowest temperature changes appear around 35°C. The observed difference is too large to be accounted for by the effect of deuterium or by the experimental differences. However, if the modification of the α-helical segment thermally precedes the changes in the β-strand(s), the data are consistent with the C-terminal of helix II starting to unravel just above room temperature and, as its mobility increases, destabilizing the adjacent β-strand (43–49).

Conclusion

In the present study, we used a combination of calorimetric and optical spectroscopic techniques, with the purpose of compensating for some of the weaknesses of each method. This experimental approach was then associated with full and partial spectral bandshape analysis by the principal component method of factor analysis. The full bandshape analysis evidenced changes not observable when the denaturation was monitored at a single frequency. The major conclusion of our work was the observation of a pretransition that can be associated with unfolding of part of a helical segment and possibly one of the β-strands before the remaining structure unfolds.

The use of CD in the far-uv region combined with FTIR revealed that the temperature-induced pretransitional changes observed in calorimetry are localized changes in the secondary structure and are not accompanied by changes in the tertiary structure as sensed by CD in the near-uv. This observation is inconsistent with proposals that the entire thermal transition is two-state in nature.

Coupling the FTIR and far-uv CD offered complementary information about thermal denaturation of RNase A, because each technique is sensitive to different types of fractional secondary structure. Far-uv CD together with separate analysis of different FTIR regions showed that the pretransitional secondary structure changes observed are consistent with partial unwinding of a helical segment (possibly helix II) and modification of a small β-sheet segment.

Materials and methods

Mathematical methods

Each set of temperature-dependent experimental spectra, Θ_i(v), was decomposed into a linear combination of orthogonal principal components, φ_j(v), according to

1

The components were obtained by factor analysis (Malinowski 1991), where p is the number of significant (nonnoise) components and C_ij are the loadings of each component j in the experimental spectrum obtained at temperature T_i.

To separate intensity and bandshape changes in analyzing the variation of C_ij versus T, we express Θ_i(v) as

2

where the norm, N_i, of each individually measured spectrum is a measure of the overall intensity of Θ_i(v), defined as

3

and the reduced loading, which reflects the bandshape contributions to Θ_i(v), is

4

Plots of norms, N_i, and reduced loadings, C^r_ij, versus T were used to evaluate the independent contributions of the intensity and bandshape changes, respectively, that resulted from the thermal denaturation. The temperature dependence plots of the loadings are equivalent to the unfolding curves. The far-uv CD and FTIR C_ij plots were fit to a sigmoid function or to the sum of two sigmoid functions to probe the two-state and multiple-state models. The near-uv CD loadings displayed a linear temperature dependence before the main transition and were fit to a sum of linear and sigmoidal functions. However, the far-uv CD normalization, N_i, had a different shape, with a minimum occurring around the melting temperature. The minimum appears because the area of the spectrum decreases up to that temperature and increases again afterward. To obtain a melting curve similar to those of the loadings, the posttransitional normalization values were mirror imaged with respect to a horizontal line passing through the minimum of the curve.

In addition, the original FTIR spectra were truncated into two separate regions based on prominent features found in the difference spectra. These regions encompass components that, for RNase A, have been attributed to the following secondary structure types: 1620–1640 cm⁻¹, β-sheet, solvated helices, and unordered structures, and 1650–1683 cm⁻¹, unordered structures, α-helix, turns, and undefined structures (Haris et al. 1986; Olinger et al. 1986; Yamamoto and Tasumi 1991; Martinez and Millhauser 1995). The truncated spectra were, in turn, reanalyzed by factor analysis, as described above.

In the calorimetry experiments, to obtain the thermodynamic parameters, the excess heat capacity curve was integrated between the native and the unfolded states, according to the following formulae:

5

6

where ΔH_u and ΔS_u are the enthalpy and entropy of unfolding, Cp^ex is the excess heat capacity, and N and U are the native and the unfolded states, respectively. The van't Hoff enthalpy ΔH_vH was calculated according to Krishnan and Brandts (Privalov and Khechinashvili 1974; Krishnan and Brandts 1978) as:

7

where T_M is the midpoint of the transition.

Materials

Bovine pancreas RNase A type XIIA was purchased from Sigma. The protein was found to be homogeneous by SDS PAGE electrophoresis and by ion exchange chromatography on CM Sephadex C50 (salt gradient 0.1–0.25 M NaCl in 10 mM Tris buffer at pH 8; Biringer and Fink 1982; Denisov and Halle 1998). Comparison of CD and FTIR spectra of protein purified by ion exchange chromatography and unpurified indicated only minor differences.

For the CD experiments, solutions of approximately 1.5 mg/mL of protein in 10 mM phosphate buffer (pH 6.8) were used. CD spectra were measured on a JASCO J600 spectrometer at 1-nm bandwidth and 2-sec time constant. In the far-uv region, a jacketed cylindrical quartz cell (JASCO, Inc.) with a 0.1-mm pathlength was used, and each spectrum was an average of 10 scans recorded at 20 nm/min. In the near uv region, the spectra were recorded under similar conditions but with a 5-mm pathlength cell. The temperature was controlled using a Fisher Scientific model 9100 circulating bath. All the spectra were baseline corrected by subtraction of the CD spectra of the buffer recorded in the same cell and under the same conditions.

For the FTIR measurements, the protein was H/D exchanged by dissolving 5 to 6 mg protein in 1 mL D₂O and incubating for about 30 min at 57°C to 59°C, followed by rapid cooling. The exchanged protein was lyophilized and then redissolved at a concentration of approximately 5.6 mg/mL in deuterated phosphate buffer (apparent pD 6.8). The sample was sandwiched between two CaF₂ windows separated by a 50-μm Teflon spacer and held in a homemade thermostatted cell holder (Wang 1993). The sample temperature was controlled by a Neslab RTE 110 circulating bath referenced to a probe inserted in the brass outer jacket of the cell holder. After each temperature change, scanning was delayed by 10 to 15 min to achieve thermal equilibration. For each spectrum, 512 scans at 4 cm⁻¹ nominal resolution were averaged using a Digilab FTS-60 FTIR spectrometer. The instrument was equipped with a liquid-nitrogen–cooled MCT detector and continuously purged with dry air. Solvent correction of the protein FTIR spectra was performed by variable subtraction of spectra of the buffer, recorded under similar conditions, to obtain an approximately flat baseline between 1700 and 1900 cm⁻¹. Correction for water vapor absorption was performed by variable subtraction of a water vapor spectrum to eliminate all sharp features in the amide I′ region. The corrected FTIR spectra were truncated to just the amide I′ region for analysis.

The DSC experiments were run on a Nano-Differential Scanning Calorimeter II (Calorimetry Sciences Corp.) equipped with two 24-carat gold cylindrical cells, each having a nominal volume of 0.33 mL. The heating rate was 1°C/min. The concentrations of protein and buffer were the same as in the CD experiments. The protein scans were corrected for buffer by subtraction of buffer scans obtained under the same conditions, but with buffer in both cells. The pretransitional and posttransitional baseline of the buffer corrected protein scans were fit to a linear and polynomial baseline, respectively, to determine the excess heat capacity which was used in the calculation of the unfolding thermodynamic parameters.

The CD and FTIR spectra were analyzed by PC/FA using software developed in this laboratory. The calorimetry results were analyzed using the program CpCalc from Calorimetry Sciences Corp. The protein concentration was determined by measuring the absorbance at 277.5 nm, at which ɛ = 9800 M⁻¹cm⁻¹ (Sela and Anfinsen 1957).

Abbreviations

• RNase A, ribonuclease A

• DSC, differential scanning calorimetry

• FTIR, Fourier transform infrared

• CD, circular dichroism

• NMR, nuclear magnetic resonance

• S/N, signal-to-noise ratio

• PC/FA, the principal component method of factor analysis