Unique Features of the Folding Landscape of a Repeat Protein Revealed by Pressure Perturbation

Abstract

The volumetric properties of proteins yield information about the changes in packing and hydration between various states along the folding reaction coordinate and are also intimately linked to the energetics and dynamics of these conformations. These volumetric characteristics can be accessed via pressure perturbation methods. In this work, we report high-pressure unfolding studies of the ankyrin domain of the Notch receptor (Nank1–7) using fluorescence, small-angle x-ray scattering, and Fourier transform infrared spectroscopy. Both equilibrium and pressure-jump kinetic fluorescence experiments were consistent with a simple two-state folding/unfolding transition under pressure, with a rather small volume change for unfolding compared to proteins of similar molecular weight. High-pressure fluorescence, Fourier transform infrared spectroscopy, and small-angle x-ray scattering measurements revealed that increasing urea over a very small range leads to a more expanded pressure unfolded state with a significant decrease in helical content. These observations underscore the conformational diversity of the unfolded-state basin. The temperature dependence of pressure-jump fluorescence relaxation measurements demonstrated that at low temperatures, the folding transition state ensemble (TSE) lies close in volume to the folded state, consistent with significant dehydration at the barrier. In contrast, the thermal expansivity of the TSE was found to be equivalent to that of the unfolded state, indicating that the interactions that constrain the folded-state thermal expansivity have not been established at the folding barrier. This behavior reveals a high degree of plasticity of the TSE of Nank1–7.

Introduction

One of the major issues remaining to be explained in protein folding involves the characterization of the thermodynamics and dynamics linked to changes in solvation that occur concomitant with protein conformational transitions (1). In particular, the volumetric properties of proteins, which can be accessed via pressure perturbation methods, yield information about the changes in relative solvation and packing between various states along the folding reaction coordinate (2–5). These volumetric properties are also intimately linked to the energetics and dynamics of these conformations. As Kauzmann pointed out several years ago (6), “… volume and enthalpy changes are equally fundamental properties of the unfolding process, and no model can be considered acceptable unless it accounts for the entire thermodynamic behavior.” We have studied for a number of years the pressure-induced unfolding of a model globular protein, staphylococcal nuclease (Snase) (7–12). Based on pressure perturbation calorimetric measurements and densitometric and high-pressure studies on Snase, we have proposed a qualitative physical model to explain its pressure-temperature phase diagram (2,8,10,11). According to our model, the magnitude of pressure effects depends primarily upon the packing efficiency of the protein core and the relative thermal expansivity and energetics of the various states on the landscape. Our model further assumes that due to the opposite signs of volume changes caused by polar and hydrophobic hydration, solvent density changes (relative to the bulk) linked to hydration of the newly exposed protein surfaces upon unfolding provide a relatively small contribution to ΔV_u. A number of other studies of pressure-induced unfolding have been reported (see Royer and Winter (3–5) for an overview), but given the relative paucity of groups working in this field, our understanding of pressure effects, despite recent progress (13), lags behind our understanding of protein folding via alternate perturbations (temperature, denaturant).

Here, we report a thorough study of the pressure-induced unfolding of the ankyrin repeat domain of the Notch receptor (14) (Nank1–7). This protein provides an interesting model system to study pressure effects on proteins. Its folding has been thoroughly characterized at atmospheric pressure (14–20). Moreover, it is a rather large (213 residues), all α-helical protein exhibiting the elongated, modular hydrophobic core common to repeat proteins. Using fluorescence-based approaches, we have characterized its equilibrium and kinetic pressure responses as a function of both urea concentration and temperature, allowing direct comparison with previous studies of denaturant-induced unfolding at atmospheric pressure. We also have probed, by high-pressure Fourier transform infrared spectroscopy (FTIR) and small-angle x-ray scattering (SAXS), the effect of denaturants on the helical content and the degree of chain condensation of the pressure-unfolded states of Nank1–7. Our results reveal that the volume change during unfolding for this protein is rather small in magnitude given the rather large size of the protein. We find that the pressure-unfolded state is more compact and retains considerable helical content compared to the urea-induced unfolded state. Finally, we observe a high degree of plasticity of both the unfolded state ensemble and the transition state ensemble (TSE) of Nank1–7 that contrasts with that of the globular protein model Snase, reported previously (10).

Material and Methods

Protein production and purification

Nank1–7 was produced in Escherichia coli transformed via a pET15b (Novagen, Madison, WI) plasmid containing the seven-motif ankyrin repeat sequence as previously described (20). The His-tagged ankyrin peptide was purified using nickel-affinity (HiTrap, GE Healthcare, Chalfont St. Giles, United Kingdom) and size-exclusion (HiLoad 26/60 S75, GE Healthcare) chromatography, also as previously described (20). Protein concentration was estimated by absorbance at 280 nm (extinction coefficient, ɛ = 3550 L mol⁻¹ cm⁻¹).

High-pressure fluorescence spectroscopy

High-pressure fluorescence measurements were carried out as previously described (7). Details are provided in the Supporting Material. Briefly, the protein solution at ∼50 μM in Tris buffer is loaded into a 500-μL cell fitted with a DuraSeal cap held in place by an O-ring. The cell is placed in a high-pressure vessel equipped with four sapphire windows and connected to a pump and gauge. The pressure-transducing liquid is 18 MOhm MilliQ water. The excitation light is provided from a xenon lamp and monochromator via an optical fiber. Emission is collected at 90° through another monochromator on a photomultiplier tube. For each fluorescence emission spectrum, the center of spectral mass was calculated as

$${\langle \lambda \rangle }_j=\sum _j{F}_j{\lambda }_j/\sum _j{F}_j,$$ (1)

where F_j is the fluorescence intensity at wavelength λ_j = 320, 321, 322, …, 450 nm.

For each pressure-induced unfolding measurement, the center of spectral mass was fitted as a function of pressure for values of ΔG°_u and ΔV°_u using the BioEQS program, as previously described (Supporting Material (7,21)), taking into account the correction by the relative quantum yield of unfolded and folded states (0.5) obtained from the ratio of the high- and low-pressure spectra. For pressure-induced equilibrium unfolding analysis, the free energy of unfolding is assumed to evolve linearly with the pressure, p:

$$\Delta G_{\text{u}}^{\text{0}}\left(p\right)=\Delta G_{\text{u}}^{\text{0}}\left(p\right)+p\times \Delta V_{\text{u}}^{\text{0}},$$ (2)

where

$$\Delta G_{\text{u}}^{\text{o}}\left(p\right)=-RT\ln K_{\text{f}}\left(p\right)$$ (3)

and

$$K_f\left(p\right)=\frac{(\langle {\lambda }_F\rangle -\langle \lambda \left(p\right)\rangle )}{(\langle \lambda \left(p\right)\rangle -\langle {\lambda }_U\rangle )}\text{.}$$ (4)

The kinetic data were fit for the relaxation time (τ) and the activation volumes and rate constants for folding and unfolding at atmospheric pressure were obtained from the values of the natural logarithm of the relaxation time as a function of pressure, as previously described (7). Briefly, the natural logarithm of the relaxation time can be expressed as

$$\ln \left(\tau \right)\left(p\right)=\ln \left[\frac{1}{k_{\text{f}}\left(p\right)+k_{\text{u}}\left(p\right)}\right],$$ (5)

where $k_{\text{f}\left(p\right)}$ and $k_{\text{u}\left(p\right)}$ are the rate constants of folding and unfolding, respectively, at pressure p, and can be expressed in terms of the apparent equilibrium constant in Eq. 3:

$$K_{\text{f}}\left(p\right)=k_{\text{f}}\left(p\right)/k_{\text{u}}\left(p\right)\text{.}$$ (6)

The variation of the rate constants with pressure is expressed in terms of their values at atmospheric pressure, $k_{\text{f}}^0$ and $k_{\text{u}}^0$:

$$k_{\text{f}}\left(p\right)=k_{\text{f}}^0\times e^{-p\times \Delta V_{\text{f}}^{\text{\#}}/RT}$$ (7)

and

$$k_{\text{u}}\left(p\right)=k_{\text{u}}^0\times e^{-p\times \Delta V_{\text{u}}^{\text{\#}}/RT},$$ (8)

where $\Delta V_{\text{f}}^{\text{\#}}$ and $\Delta V_{\text{u}}^{\text{\#}}$ are the activation volumes for folding and unfolding, respectively, which define the global volume change during folding, $\Delta V_{\text{f}}$, by the equation

$$\Delta V_{\text{f}}=\Delta V_{\text{f}}^{\text{\#}}-\Delta V_{\text{u}}^{\text{\#}}\text{.}$$ (9)

Plots of ln(τ) versus pressure were fit for $k_{\text{f}}^0$ and $\Delta V_{\text{f}}^{\text{\#}}$ by substituting Eqs. 7–9 into Eq. 3.

SAXS

SAXS (22–24) was carried out on Nank1–7 solutions at several pressures and urea concentrations. Protein solutions were prepared immediately before the measurements. To avoid pressure-induced changes of the pH value, 50 mM TrisHCl buffer was used at a pH value of 7.5 (25). Buffers were prepared using deionized water and adding the appropriate amount of buffer salt and urea (2.0 M and 2.2 M, respectively). The solution concentration range was 7.8–11.0 mg/mL. Under these conditions, the solution is diluted enough so that the single scattering approximation is valid (26) yet provides a scattering signal with a good signal/noise ratio.

The SAXS experiments were performed at beamline BL9 of DELTA (TU Dortmund, Dortmund, Germany) (27) and at beamline BW4 of HASYLAB (DESY, Hamburg, Germany) (28). High-pressure SAXS (up to 3 kbar) was carried out in a special sample cell employing two flat diamond windows (29). At BW4 (λ = 1.3808 Å), using the MAR CCD detector, a q-range of 0.3–3.0 nm⁻¹ could be covered. The exposure time was 20 min. A similar exposure time was used at BL9 by employing a MAR345 image plate detector. A wavelength of λ = 1.239 Å allowed for coverage of a q-range of 0.3–3.5 nm⁻¹. We note that the loss of scattering intensity with pressure is due to the loss of contrast between protein and water, as the density of the water increases with pressure.

The Guinier approximation (30) was used to derive the radius of gyration, R_g, of the protein in solution. The approximation is valid for q-values (q momentum transfer) only up to q_max ≤ 1.3/R_g (22). The radius of gyration, R_g, was also derived from the pair-distribution function p(r), which characterizes the shape and size of the particle (23) (Supporting Material). After azimuthal averaging of the SAXS spectra, the scattering signals were normalized and corrected for solvent and background scattering by subtracting the scattering curve of the corresponding pure buffer solution. To obtain the radii of gyration, the low q-range showing a linear curve shape in the log (I(q)) vs. q² representation was fitted according to Guinier's approximation. The pair-distance distribution functions were calculated by fitting the scattering data using the program GNOM (31). To reveal whether the proteins are folded or denatured, SAXS curves are shown in the so-called Kratky plot, i.e., q²I(q) vs. q. In the case of a globular, folded protein, the scattering curve exhibits a distinct maximum, whereas in the case of an unfolded polypetide, the curve displays a flat plateau (22).

High-pressure FTIR spectroscopy

A 2% (w/v) solution of Nank1–7 was prepared in a 50 mM Tris, 150 mM NaCl buffer solution in D₂O (pH 8.0) with or without ¹³C-urea (Sigma, St. Louis, MO) as denaturant. FTIR spectra were recorded in a diamond anvil cell with a Nicolet MAGNA 550 spectrometer. For each pressure step, 256 scans were summed up. Spectral processing and deconvolution of the amide-I′-band was performed with GRAMS/AI 8.0 software. For comparison, all peak areas were normalized to unity.

Results

Pressure-induced equilibrium unfolding

The pressure-induced equilibrium unfolding of Nank1–7 was characterized as a function of both urea concentration and temperature, by measuring the average wavelength of the fluorescence emission of the single tryptophan residue of the protein (Fig. 1), located in the 5th ankyrin repeat (14). The pressure profiles were collected at urea concentrations ranging from 1.9 to 2.4 M, depending upon the temperature, to observe unfolding over our experimentally accessible pressure range (0.1–3500 bar). We note that at the micromolar protein concentrations used for the fluorescence experiments, these transitions were entirely reversible. At all temperatures, the application of pressure results in a sigmoidal red shift in the fluorescence emission (Fig. 2, A–C) that exhibited a distinct high-pressure plateau for the average emission wavelength. However, the magnitude of the pressure-induced red shift was strongly dependent upon the concentration of urea. The data were fit at each temperature either individually or globally, linking the volume change across all urea concentrations, for the volume change upon folding (ΔV°_f) and the free-energy change of unfolding (ΔG°_u) (Table S1 in the Supporting Material). In all cases, the volume change was found to be independent of the urea concentration, and the values reported are those from the global fits.

Figure 1.

Ribbon diagram of the 3D structure of the ankyrin repeat domain of the Notch receptor, Nank1–7. Each ankyrin repeat is differently colored. The tryptophan residue is shown in blue.

Figure 2.

Average emission wavelength of the tryptophan fluorescence spectrum of Nank1–7 as a function of pressure and urea concentration (indicated in molar units) at 12°C (A), 20°C (B), and 24°C (C).

The increase of the plateau value of the average emission wavelength with urea concentration reveals increased solvent relaxation around the excited state of the tryptophan at the high-pressure plateau, indicative of an increased degree of solvent accessibility. At the highest urea concentrations used, at all temperatures, the high-pressure plateau corresponds to the average emission wavelength observed for N-acetyl-tryptophanamide in water, i.e., a completely accessible tryptophan moiety. We hypothesize that the unfolding of Nank1–7 by pressure conforms to a two-state model, with a different degree of tryptophan exposure in the unfolded state depending upon the urea concentration. The values of the ΔG°_u at atmospheric pressure and in the absence of urea are calculated from the linear regression of the plots of ΔG°_u(i) at atmospheric pressure obtained from the analysis of the pressure unfolding profiles at each urea concentration, i, for a given temperature (Fig. S1 A). Although the uncertainty of these extrapolated values is large, given the small number of points (four to five urea concentrations) and the long extrapolation (2.3–0 M urea), the values calculated for ΔG°_u at 16° and 20°C (6.9 ± 1.0 and 7.0 ± 1.25 kcal/mol, respectively) are within the error of those previously reported from urea unfolding studies at atmospheric pressure at 16° and 20°C (8.09 ± 0.07 and 7.52 ± 0.04 kcal/mol, respectively (32)). The nonlinear temperature dependence of ΔG°_u in the absence of urea (Fig. S1 B) is consistent with a significant heat capacity change upon unfolding, which is typical of proteins (33). We also find a strong dependence of ΔV°_f upon temperature, ∼−1.7 mL/mol K, (Fig. S1 C), which arises, as noted above, from the difference in thermal expansivity, Δα_u, between the unfolded and folded states. The p-T phase diagram of Nank1–7 (Fig. S1 D) calculated from the values of ΔG°_u and ΔV°_u obtained at 2.1 M urea presents a shape that is similar to that of previously reported p-T phase diagrams for proteins (10,34–40).

Chain condensation in the pressure unfolded state

To investigate the structural changes of Nank1–7 associated with changes in pressure and urea concentration, we performed pressure-dependent SAXS experiments at 24°C. We note that, although no aggregation was observed under pressure, these experiments were irreversible upon release of pressure due to aggregation at the high concentrations required for SAXS measurements. In the absence of urea, no appreciable change in the scattering profiles was observed between atmospheric pressure and 3 kbar (Fig. S2 A and Fig. 3 C). However, in the presence of 2.0 or 2.2 M urea, both the shape and the intensity of the scattering profiles are strongly pressure-dependent (Fig. S2, B and C), and the pair-distribution functions (Fig. 3, A and B) reveal a significant increase in their average and largest pairwise distance with increasing pressure and urea concentration. Fig. 3 C shows a comparison of the pair-distribution functions for 0, 2, and 2.2 M urea obtained at 3 kbar, corresponding to the high-pressure plateau observed in the fluorescence profiles. The pair-distribution function is much broader at the highest urea concentration, indicating a significantly more expanded chain. Kratky plots at increasing pressure for 0, 2, and 2.2 M urea (Fig. 4, A–C) indicate that the chain retains an organized structural character at 3 kbar in the absence of urea but behaves as a Gaussian chain at high pressure for both urea concentrations.

Figure 3.

(A and B) Pressure-dependent pair distribution functions obtained from analysis of the SAXS data on Nank1–7 at 24°C at 2 M (A) and 2.2 M (B) urea. Pressures are as indicated in the figures. (C) Comparison of the pair-distribution functions at 1 bar and 0 M urea (red line), and at 3 kbar in the presence of 0 M urea (yellow dotted line), 2 M urea (green dotted line), and 2.2 M urea (blue line).

Figure 4.

Kratky plots obtained from analysis of the SAXS data on Nank1–7 at 24°C for 0, 2, and 2.2 M urea, A–C respectively, at the pressures indicated.

Secondary structure in the pressure-unfolded state

To assess the evolution of secondary structure accompanying the fluorescence- and SAXS-detected unfolding transitions, high-pressure FTIR experiments were carried out in the absence or presence of varying concentrations of urea at 20°C. So that the urea bands would not obscure the protein amide bands, we used ¹³C-labeled urea as a denaturant (41,42). The application of pressure to Nank1–7, in the absence of urea and in the presence of 2 and 2.35 M ¹³C-urea, led to a red shift in the amide I absorption band, indicative of a change in secondary structural content (Fig. S3, A–C). Analysis of the shift as a function of pressure, as described in the Materials and Methods section, revealed a decrease in helical content (Fig. 4 A), although according to the deconvolution software used, residual helical structure remained at the highest pressures, as well as at the 2.35 M urea concentration. Like the SAXS experiments, the FTIR experiments were irreversible upon return to atmospheric pressure due to aggregation at low pressure, but again no aggregation occurred under pressure, because pressure destabilizes protein oligomers and aggregates (43,44). In the absence of urea (Fig. 5 A), the pressure-induced loss of ∼25% α-helical content took place over a very broad pressure range (between ∼2.5 and 6 kbar), mostly beyond the range of our fluorescence and SAXS high-pressure cells. The lack of any change in secondary structural content below ∼3 kbar in the absence of urea is consistent with the insignificant changes in fluorescence and SAXS over the same pressure range. In the presence of 2 M urea (Fig. 5 B), a ∼25% loss in α-helical content occurred between 1.5 and 4.5 kbar, with about half of the change occurring below 3 kbar, corresponding to the plateau observed in the fluorescence and SAXS experiments. Due to the irreversible nature of the FTIR transition, calculation of an equilibrium volume change is not possible, although we note that the sensitivity of the FTIR transition to pressure is substantially less (by ∼50%) than that observed in the reversible fluorescence-monitored transitions. At 2.35 M urea (Fig. 5 C), we observed a lower helical content already at atmospheric pressure, and the ∼35% pressure-induced loss of α-helix occurred between 0 and 4 kbar. At 2.2 kbar and 2.35 M urea, the fluorescence spectrum of Nank1–7 is as red-shifted as possible (equivalent to N-acetyl L-tryptophanamide in water), and the chain expansion, as observed by SAXS, has also reached a maximum. Therefore, the FTIR results indicate that the Nank1–7 α-helices resist despite considerable disruption of the tertiary structure by pressure. Indeed, it has been shown previously that pressure has very little effect on isolated α-helices (45,46).

Figure 5.

Pressure dependence of the secondary structural content for Nank1–7 at 20°C as a function of pressure for 0 (A), 2 (B), and 2.35 M urea (C), where the structural elements are the α-helix (solid triangles), the random coil (solid squares), the β-sheet (open squares), • and ○ - various turn structures. There is an estimated 5% uncertainty in the calculated secondary structure content.

Pressure-jump kinetics

Pressure-jump relaxation profiles were obtained as a function of temperature and urea concentration by monitoring the fluorescence intensity at the maximum of the folded-state fluorescence emission. The relaxation time under all these conditions was consistent with a single exponential decay of the profiles of intensity versus time after the pressure jump. Examination of the plots of the natural logarithm of the relaxation times versus pressure (Fig. 6, A–C) reveals that, as in the case of the equilibrium unfolding, urea has no effect on the volume change of unfolding. However, it can be seen that the high-pressure slopes of the profiles are strongly dependent on temperature, with a distinct downward curvature (Chevron-like behavior) observed at low temperature and high pressure but fairly linear profiles over the entire unfolding transition at higher temperatures. The data obtained at all temperatures were analyzed for the activation volume and atmospheric rate constant for folding, ΔV^#_f and k°_f (Table S2 and Materials and Methods). The corresponding parameters for unfolding (ΔV^#_u and k°_u) were derived from the equilibrium free-energy and volume changes. At low temperature, the specific volume of the TSE was found to be significantly closer to that of the folded state than to that of the unfolded state (ΔV^∗_f = +50 mL/mol, ΔV^#_u = −18 mL/mol), indicating that at the barrier, a significant degree of dehydration occurs. According to our model, at the barrier, the protein is collapsed, forming a significant proportion of the native-state solvent-excluded volume, and is thus relatively “dry” (Fig. 7 B).

Figure 6.

Natural logarithm of the pressure-jump fluorescence-detected relaxation time for Nank1–7 as a function of pressure for the urea concentrations indicated at 12°C (A), 20°C (B), and 28°C (C). Lines through the points correspond to fits of the data, as described in Materials and Methods. Open symbols represent data points below the transition midpoint and solid symbols those above the transition midpoint.

Figure 7.

Volumetric properties of Nank1–7. (A) Temperature dependence of the equilibrium volume change for unfolding, ΔV°_u (■, green), and the activation volumes for folding, ΔV^#_f (Δ, red) and unfolding, ΔV^#_u (▿, blue). Points correspond to the values obtained from fits of the data, and lines represent the linear regressions. (B) Volumetric diagram (V_M is molar volume) for Nank1–7 at 12° and 28°C, showing the relationships between the specific molar volumes as well as the thermal expansivity of the folded state, F, the unfolded state, U, and the TSE.

Comparison of the temperature dependence of the activation volumes for folding and unfolding with that of the equilibrium volume change (Fig. 7 A) reveals that the entire temperature dependence of the ΔV°_f resides in the activation volume for unfolding, ΔV^#_u. Moreover, the value of ΔV^∗_u is negative at low temperature and changes sign to become positive at high temperature. Indeed, unlike denaturant-based unfolding, pressure-induced unfolding need not display Chevron-like behavior, as the sign of the activation volumes can be either positive or negative. The temperature dependence of the activation volumes, like that of the equilibrium volume change, is equivalent to the difference in thermal expansivity between the two states implicated in the reaction. We observe no dependence of the ΔV^#_f on temperature (0.12 ± 0.54 mL/mol K). In contrast, we observe a strong dependence of ΔV^#_u on temperature (2.08 ± 0.46 mL/mol K), highlighting a significant difference in thermal expansivity between the TSE and the folded state. This value is equivalent, within error, to that (1.72 ± 0.25 mL/mol K) observed for the ΔV°_u (the difference in thermal expansivity between the unfolded and folded states). Hence, we deduce that the thermal expansivity of the TSE is equivalent to that of the unfolded state (Fig. 7 B).

Discussion

The folding energetics and kinetics of Nank1–7 have been characterized and linked to structural folding pathways (17,18,47), and prior results provide a solid basis for comparison and interpretation of the high-pressure studies presented here. These SAXS and fluorescence studies reveal pressure-induced unfolding with a plateau near 3 kbar. Nank1–7, with a value of ∼45 ml/mol, exhibits a rather small pressure sensitivity (3) given its size (213 residues).

Comparison of the fluorescence, SAXS, and FTIR data reveals significant effects of urea on the physical properties of pressure-unfolded Nank1–7. In Fig. 8, the structural properties of the pressure-unfolded state of Nank1–7 are presented under different conditions of temperature and urea concentration. At moderate urea concentration, the pressure-unfolded ensemble, although a Gaussian chain, is not completely expanded and retains a major fraction of the native secondary structure. Indeed, the chain becomes much more expanded and the tryptophan significantly more exposed to solvent, and the secondary structure content decreases to ∼50% that of the native protein at the high-pressure plateau reached in the presence of slightly higher urea concentrations. It is not surprising that the secondary structure persists at high pressure. It has been shown that pressure has little effect on α-helices (45,46,48), and hence, the loss of secondary structure under pressure typically occurs only if the helices are not intrinsically stable in the absence of tertiary interactions. Estimation of helix stability for the Nank helices using the program Agadir (49), revealed that in comparison to the ∼60–65% helicity of the peptide described by Baldwin and co-workers (AAAAKAAAAKAAAAKA) (50), the Nank1–7 helices were highly unstable. However, the residual interactions responsible for the relative chain compaction and blue tryptophan emission at low urea likely stabilize the helices. Results of recent experimental and theoretical studies on volume changes suggest that pressure-induced unfolding results from internal voids in the native state. Since these voids are not always homogeneously distributed throughout the structure, certain regions can be less pressure-sensitive than others.

Figure 8.

Schematic of the temperature and urea concentration dependence of the physical properties of Nank1–7 at the 3-kbar pressure plateau.

It is interesting that fits of the fluorescence equilibrium profiles to a two-state unfolding model yielded equivalent volume changes for unfolding, despite the larger plateau values for the average emission wavelength of the tryptophan with increasing urea. We and others have shown that denaturant itself does not influence the value of the volume change upon unfolding (51,52). However, we observe here that increasing urea leads to greater solvent exposure of the tryptophan residue and a more expanded Gaussian chain at high pressure, without any change in molar volume. Such an observation is consistent with our working model, in which disappearance of internal voids upon unfolding, rather than density changes due to differential (polar versus hydrophobic) solvation, provide the bulk of the volume change upon unfolding. Once the protein has expanded sufficiently to allow occupation of these internal voids by water, pressure has no further effect. Hence, we suggest that the pressure-unfolded state, which is not maximally expanded or devoid of α-helices, may more closely resemble the unfolded state that is sampled by this protein under native conditions than those unfolded by denaturant or temperature perturbation.

Consistent with the lack of energetic barriers between these unfolded conformations, extrapolation to atmospheric pressure and 0 M urea of the urea- and pressure-dependent fluorescence free energies of unfolding yields values that are in reasonably good agreement with the global stability of the protein previously determined from urea denaturation profiles under similar conditions (32). For this transition from the folded state to the compact denatured state, the p-T phase behavior of Nank1–7 displays the shape observed for the few globular proteins that have been studied to date (10,34–40). Indeed, it has already been shown that Nank1–7 exhibits a large heat capacity upon unfolding (19), and we show here that it also exhibits a significant change in thermal expansivity, the second most important parameter giving rise to the curvature in the p-T plane.

Our pressure-jump kinetics measurements on Nank1–7 fit well to a two-state unfolding/refolding model. In contrast, thorough studies of Nank1–7 folding and unfolding kinetics using denaturant revealed multiple kinetic phases, resulting from denatured-state proline isomerization and an on-pathway unfolding intermediate (18). This intermediate displayed tryptophan emission and circular dichroism similar to the folded state, a modest free-energy gain, and ∼40% loss of solvent-accessible surface area compared to the unfolded state. Proline isomerization reactions, albeit quite slow, are not rate-limiting at high pressure (53), due to the large positive activation volume for folding, which slows the main folding transition even further (12,53). Hence, it is not surprising that the proline-based kinetic complexity was not observed in our pressure-jump studies. Likewise, the on-pathway intermediate was not detected in our pressure-jump studies, either because it was not highly populated enough to be observable in our experiments or because of the pressure-dependent increase in the main folding barrier.

The TSE for the single transition that we observe exhibits a molar volume at low temperature quite close to that of the folded state. According to our working model of pressure effects, this indicates that the bulk of the solvent-excluded packing defects have formed in this TSE. In contrast, the thermal expansivity of the TSE is equivalent to that of the unfolded state, indicating that most of the native contacts that limit the thermal expansion of the folded state are not formed at the barrier. Previous studies on Nank1–7 folding indicated that it folds inside-out, with the main barrier corresponding to the formation of the central repeats (3–5,54). Our expansivity data suggest that this TSE remains quite malleable. An expansivity for the TSE of Nank1–7 equivalent to that of the unfolded state is in contrast with that found for the TSE of Snase (the only other protein for which this quantity has been measured) (10), which was nearly equivalent to that of the folded state, consistent with a substantial degree of structure at the barrier for this globular protein. That the tertiary interactions in Nank1–7 are not locked in at the folding barrier may derive from the elongated and modular nature of the hydrophobic core of Nank1–7, which imposes fewer spatial constraints on the structure.

In conclusion, these pressure perturbation studies on the ankyrin repeat domain of the Notch receptor reveal a high degree of plasticity in both the unfolded-state ensemble and the TSE. The physical properties of these states, such as chain condensation and thermal expansion, are strongly dependent upon conditions. Overall, the observation of a less drastic effect of pressure, as compared to chemical denaturants, can be understood in light of the different physical basis for their perturbations. Urea interacts with the protein, leading to more expanded chains in the unfolded state via an increase in the persistence length. Pressure effects arise simply due to the smaller system volume when the protein is in the unfolded state, and therefore do not necessarily result in maximal expansion of the chain, or maximal loss of helical content. Our understanding of protein folding mechanisms is dependent upon our ability to characterize the conformational states on the protein folding landscape. Here, we show that pressure can reveal a unique set of conformational states with distinct structural and thermodynamic properties, thus increasing our knowledge and broadening our perspective of this complex process.

Supporting Material

Two tables and three figures are available at http://www.biophysj.org/biophysj/supplemental/S0006-3495(10)00324-3.