Water-Protein Interactions Coupled with Protein Conformational Transition

Abstract

Conformational fluctuations of proteins are crucially important for their functions. However, changes in the location and dynamics of hydrated water in many proteins accompanied by the conformational transition have not been fully understood. Here, we used phase-modulated clean chemical exchange NMR approach to investigate pressure-induced changes in water-to-amide proton exchange occurring at sub-second time scale. With the transition of ubiquitin from its native conformation (N₁) to an alternative conformation (N₂) at 250 MPa, proton exchange rates of residues 32–35, 40–41, and 71, which are located at the C-terminal side of the protein, were significantly increased. These observations can be explained by the destabilization of the hydrogen bonds in the backbone and partial exposure of those amide groups to solvent in N₂. We conclude that phase-modulated clean chemical exchange NMR approach coupled with pressure perturbation will be a useful tool for investigations of more open and hydrated protein structures.

Introduction

Conformational fluctuations of proteins, such as open-close dynamics and partial denaturation, are crucially important for protein functions. Solution NMR spectroscopy provides information about protein structure and dynamics at atomic resolution. Recently, we and other groups have presented evidence that high-pressure NMR spectroscopy is a useful technique to study high-energy conformations, such as open conformation of enzymes and locally disordered conformations of different proteins (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13). Pressure, in general, can cause a population shift in the ensemble of protein conformers from a native conformer to a fully unfolded conformer because protein structure is closely linked to protein partial molar volume. Typically, partial molar volume decreases as the protein unfolds (14, 15).

Although motions of polypeptide chains on nanosecond-to-second timescales can be revealed by solution NMR spectroscopy, the location and dynamics of hydrated water in many proteins have not been fully understood. In particular, the effect of solvent water on the conformational transitions of proteins is still largely unknown. Therefore, protein hydration has been investigated by solution NMR techniques, such as relaxation dispersion of ²H and ¹⁷O (16, 17), intermolecular cross-relaxation (i.e., nuclear Overhauser effect (NOE)) between protein and water (18, 19, 20, 21), and by other techniques, such as x-ray crystallography (22, 23), neutron scattering (24, 25), and molecular dynamics (MD) simulation (25, 26). Furthermore, the native state hydrogen exchange (HX) method combined with NMR spectroscopy or mass spectroscopy is a well-established and widely used approach of studying conformational fluctuations and hydration of proteins (27, 28, 29, 30). However, this approach can be usually used only for those HX reactions that occur on a timescale from minutes to months. Therefore, one can investigate the HX reaction of amide groups in secondary structures and well-occluded regions in proteins. To investigate the HX reactions at sub-second timescales for amide groups located on protein surfaces and in loop regions, the phase-modulated clean chemical exchange (CLEANEX-PM) NMR approach was established by Mori and coworkers (31, 32).

Ubiquitin is a post-translational modifier consisting of 76 amino acid residues, which is involved in many cell functions, including ubiquitin-proteasomal degradation of damaged proteins. In our previous studies, high-pressure NMR spectroscopy revealed the presence of an alternative conformation (N₂) of ubiquitin (33, 34). We also reported NOE-based structure determination of the Q41N mutant of the protein at 250 MPa, where N₂ comprised 97% of all conformers (35). The N₂ state showed a large displacement of the C-terminal β₅-strand (after residue 68) with some displacements of α-helix and connected-loop regions (residues 33–41). Interestingly, the conformation of N₂ matched the changes seen upon ubiquitin binding to the ubiquitin-activating enzyme, E1. Moreover, similar conformational transitions were observed in ubiquitin-like proteins such as NEDD8 (36) and SUMO-2 (37). Therefore, we suggested that the recognition of E1 by ubiquitin is best explained by conformational selection rather than by induced fit motion. In terms of the conformational transition and water-protein interaction of ubiquitin, 1-μs MD simulation successfully described pressure-induced water penetration into the hydrophobic core of the protein, which is coupled with the opening of the protein C-terminal side (26). Here, by using the CLEANEX-PM NMR technique, we studied pressure-induced water-amide proton exchange in ubiquitin at a sub-second timescale. We showed preferential hydration of amide groups located at the C-terminal side of the protein when N₂ conformation was stabilized under high pressure.

Materials and Methods

Sample

Uniformly ¹⁵N-labeled wild-type (WT) ubiquitin was produced by conventional expression in Escherichia coli grown in M9 medium. E. coli cells were lysed by a conventional cell lysis solution, and nucleic acid molecules were removed by treatment with 0.05% polyethyleneimine. Then, the protein solution was purified by acidification at pH 4.5 and subjected to liquid chromatography with SP Sepharose Fast Flow and superdex 75 PG 26/60 columns (GE Healthcare, Chicago, IL). For high-pressure NMR experiments, the ubiquitin solution was diluted to a concentration of 2.2 mM, and its pH was adjusted to 7.36 at 298 K in 20 mM Tris d11-HCl buffer containing 10% ²H₂O. For pH titration experiments, the protein concentration was ∼1.3 mM.

High-pressure NMR measurements and analysis

To obtain accurate rates of chemical proton exchange between water and amide groups, CLEANEX-PM fast heteronuclear single quantum coherence (HSQC) NMR measurements (31, 32) at six mixing times (0.005, 0.010, 0.015, 0.020, 0.025, 0.100 s) were performed at 0.1, 150, 175, 200, 225, and 250 MPa (298 K) using a ¹H 600 MHz NMR spectrometer (DRX-600; Bruker, Billerica, MA) and a pressure-resistant ceramic cell (Daedalus Innovations, Aston, PA). As a reference, fast-HSQC measurement without mixing was obtained. The rate of proton exchange between water and amide was estimated by the following equation,

$$\frac{I}{I_0}=k/\left(R_{1A}+k-R_{1B}\right)\times \left\{\exp \left(-R_{1B}{\tau }_m\right)-\text{exp}\left(-\left[R_{1A}+k\right]{\tau }_m\right)\right\},$$ (1)

where k is the normalized rate constant related to the pseudo-first-order, forward rate constant k_BA (H₂O→NH) = X_B; X_B is the mole fraction of water (∼1); R_1A is the combination of longitudinal and transverse relaxation rates in the CLEANEX-PM measurement; R_1B is the longitudinal relaxation rate for water (0.6 s⁻¹ was used in this experiment) (32). I is the peak intensity in the spectrum at different mixing times τ_m, and I₀ is the latter value in the reference spectrum.

Results and Discussion

We performed CLEANEX-PM NMR experiments at 298 K and pH 7.36 with WT ubiquitin at 0.1, 150, 175, 200, 225, and 250 MPa and detected magnetization transfer from water to amide proton during mixing time (i.e., 0–100 ms). Fig. 1 shows CLEANEX-PM fast-HSQC and its reference spectra at 0.1 MPa (Fig. 1 A) and 250 MPa (Fig. 1 B). Water-amide proton exchanges were observed mainly for amide protons that formed hydrogen bonds with water molecules according to the structure in solution. At 250 MPa, new signals were detected in the CLEANEX-PM spectrum, indicating an increased occurrence of pressure-induced water-amide proton exchanges. Representative plots of peak-intensity ratio (see Eq. 1) as a function of mixing time (∼25 ms) are illustrated in Fig. 2. Data points at short mixing times, namely below 25 ms, are important for determining accurate chemical exchange rates (32). All the water-amide proton exchange rates at different pressures with an assumption of the pseudo-first-order reaction are shown in Fig. 3 and Table S1. The exchange rates at many residues were generally increased to various extents with increasing pressure. There were two remarkable observations: first, water-amide proton exchanges of residues 32–35, 40–41, 62, 66, and 71 were detected only above 150 MPa. Second, HSQC signals of residues 12, 46, and 74 in the reference spectra were remarkably decreased and were almost equal to zero above 150 MPa, although new signals corresponding to a disordered polypeptide chain were not observed in the reference spectra.

Figure 1.

Phase-modulated clean chemical exchange NMR with 100 ms mixing time (red) and reference (black) spectra of WT ubiquitin at 0.1 MPa (A) and 250 MPa (B) at 298 K. Assignments are depicted in panels. To see this figure in color, go online.

Figure 2.

Plots of peak intensity ratio of selected residues as a function of mixing time. The exchange rates of water to amide proton were estimated by Eq. 1. Residue numbers are provided in the panel.

Figure 3.

Water-amide proton exchange rates for ubiquitin at 0.1 MPa (bar), 150 MPa (open squares), 200 MPa (open circles), and 250 MPa (closed circles). Residues 12, 46, and 74 that displayed zero intensity in the reference spectra above 150 MPa are depicted by asterisks. Secondary structures are indicated at the top of the panel by bars (i.e., α-helices) and arrows (i.e., β-strands). Error bars show the root-mean-square-deviation obtained from the nonlinear fit (see Eq. 1).

HX can be explained by the EX2/EX1 model, which has been developed by Linderstrøm-Lang and Hvidt (38, 39). HX between amide H_A and water H_B occurs through the following process:

$$N{H}_A\cdot closed\begin{array}{c}{k}_{open}\\ \rightleftharpoons \\ k_{close}\end{array}N{H}_A\cdot open\stackrel{k_{ex}}{\to }N{H}_B\cdot open$$

In the EX2 limit, where k_close ≫ k_ex, the observed rate constant k_obs is ≈ (k_open / k_close) × k_ex. In the EX1 limit, where k_close ≪ k_ex, k_obs is = k_open. To confirm the exchange mechanism, pH dependence of exchange rates was examined at pH 7.36 and pH 7.85. The rates of proton exchange between water and ubiquitin amide groups were increased 2.1 ± 0.6 (mean ± SD) times on an average with increase in pH by 0.49 units (Table S1). Because pH-dependence matches that of a small peptide (i.e., Δk_ex = 3.1 times) (40) within the data variety, the EX2 mechanism presumably accounts for the observation. Therefore, in these experimental conditions (i.e., 298 K, pH 7.36, 0.1–250 MPa), where ubiquitin maintains its folded conformation, the EX2 mechanism dominates the exchange process according to the NMR measurements.

In the high-pressure experiments, pressure effects should be taken into account for calculations of k_ex. First, solution pH depends on pressure. In a Tris buffer system, a pH change is −0.10 at 98 MPa and −0.14 at 392∼785 MPa, according to the in situ measurement of pH under high pressure (41). In addition, when pressure increases, the reaction accompanied by positive activation volume is decelerated, whereas the reaction with negative activation volume is accelerated. Pressure effects on k_ex can be calculated from the activation volumes of acid-catalyzed, base-catalyzed, and uncatalyzed HX reactions (i.e., ΔV^≠_(H+) = +1.7 mL/mol, ΔV^≠_(OH−) = +11.0 mL/mol, ΔV^≠_(H2O) = −9.0 mL/mol) and from pressure dependence of the ionization constant of water K_W (i.e., ΔV_(KW) = −22.1 mL/mol) (42). At pH 7.36 and 298 K, when the base-catalyzed exchange reaction dominates, ΔV^≠_(app) = ΔV^≠_(OH−) + ΔV_(KW) = −11.1 mL/mol. Taking into account these pressure effects, the ratio of k_{ex(250 MPa)}/k_{ex(0.1 MPa)} was calculated to be ∼2.3. It was similar to the pressure-induced increase in k_obs of many amide groups (residues 2, 8, 10, 11, 14, 16, 20, 39, 47, 49, 60, 63, 72, 76: mean ± SD = 4 ± 1). However, assuming the EX2 mechanism, the remarkable increase in k_obs of residues 32–35, 40–41, 52, 62, 66, and 71 cannot be explained only by pressure effects on k_ex; rather, the equilibrium opening constant K_open (=k_open/k_close) must have been increased at those residues under high pressure.

In the EX2 limit, K_open is also =k_obs/k_ex. The free energy of opening, ΔG_HX, was estimated by the following equation,

$$\mathit{\Delta }{G}_{HX}=-RTln{K}_{open},$$ (2)

where R is the gas constant and T is the absolute temperature. Fig. 4, A and B shows pressure dependence of ΔG_HX for selected amide groups showing remarkable and little pressure-induced increases in k_obs, respectively. The pressure dependence of ΔG_HX provides access to the partial molar volume difference between the closed and opened states, ΔV°_HX, when assuming ΔV_HX has no pressure dependence (43). $\mathit{\Delta }{G}_{HX}^p$ can be expressed by Eq. 3 as a function of pressure at constant temperature,

$$\mathit{\Delta }{G}_{HX}^p=\mathit{\Delta }{G}_{HX}^{°}+\mathit{\Delta }{V}_{HX}^{°}\left(p-p_0\right),$$ (3)

where $\mathit{\Delta }{G}_{HX}^p$ and $\mathit{\Delta }{G}_{HX}^{°}$ are the free-energy of opening at pressure p and p₀ (=0.1 MPa), respectively. Fig. 4, C and D shows ΔV°_HX and ΔG°_HX at 1 bar, respectively, against residue number (see also Table S2). Note that ΔV°_HX was estimated for residues having $\mathit{\Delta }{G}_{HX}^p$ data of at least three different pressures. Residues 32, 33, 35, 40, 52, 62, and 71 showed large negative ΔV°_HX values (i.e., −26 ∼ −12 mL/mol), while the values of other residues were −11 ∼ +10 mL/mol.

Figure 4.

Thermodynamic parameters of ubiquitin at 298 K. (A and B) Pressure induced-changes in ΔG_HX of selected residues are shown. Errors of ΔG_HX in (A) and (B) are assumed to originate with errors of only k_obs. (C and D) ΔV°_HX and ΔG°_HX values, respectively, against residue number are shown. Secondary structures are indicated at the top of (C) by bars (i.e., α-helices) and arrows (i.e., β-strands). Error bars of ΔV°_HX and ΔG°_HX in (C) and (D) show the root-mean-square-deviation obtained from the linear fit (see Eq. 3).

Fig. 5 A shows amide groups with large negative ΔV°_HX. Fig. 5 B shows a superposition of N₁ (WT, 0.1 MPa) and N₂ (Q41N mutant, 250 MPa) models of ubiquitin. Notably, the amide groups showing large negative ΔV_HX, except for residue 52 and 62, matched the regions that determined conformational differences between N₁ and N₂. The magnitude of ΔV°_HX also matched that of the partial molar volume difference between N₁ and N₂, ΔV° (i.e., −25 mL/mol at pH 7.20) (44). Therefore, the remarkable increase in k_obs of the residues 32–41 and 71 can be explained by the destabilization of amide hydrogen bonds and/or partial exposure of those amide groups to the solvent in N₂. These ideas are consistent with the structural characteristics of N₂, namely a slight extension of the connecting loop (i.e., residues 35–40) and a large displacement of the C-terminal β₅-strand (after residue 68) (34, 35). Although residues 52, 62, and 66 did not show large displacements with the N₁-N₂ transition, local exposure can occur under high pressure at these amide groups located in turn and loop regions. In addition, the remarkable increase in k_obs was also observed at similar residues (i.e., residues 31–33, 35, 40, and 41) in the ubiquitin Q41N mutant, in which N₂ comprised 70% of all conformers (44). Thus, it is almost certain that the N₁-to-N₂ transition is accompanied by the destabilization of amide hydrogen bonds and/or the partial exposure of those amide groups to the solvent. Interestingly, 1-μs all-atom MD simulations of the protein in explicit water at high and low pressures (i.e., 0.1, 300, and 600 MPa) showed that the N₁-to-N₂ transition was accompanied by the penetration of water into the void space around the protein C-terminal side within a timescale comparable with the relaxation time of water itself (26). Water penetration and an increase in the number of internal water molecules around the residue sites likely explain the enhancement of the HX reaction under high pressure.

Figure 5.

Pressure-induced changes in HX and structure. (A) Amide groups showing large negative ΔV_HX (i.e., residues 32, 33, 35, 40, 52, 62, and 71) are shown. (B) Superposition of WT ubiquitin at 0.1 MPa, which is a model of N₁ (gray, PDB: 1D3Z), and ubiquitin Q41N mutant at 250 MPa, which is a model of N₂ (red, PDB: 2RU6), are shown. To see this figure in color, go online.

We also considered the origin of ΔV°. According to the N₂ model of ubiquitin Q41N mutant at 250 MPa, where N₂ comprised 97% of all conformers, the N₂ state showed a large displacement of the C-terminal β₅-strand (after residue 68) with some displacements of the α-helix and connected-loop regions (residues 32–41), although other parts of the protein were closely similar to N₁ and N₂ conformations (see Fig. 5 B). In the present experiments, the pressure-induced increase in water-amide proton exchanges of residues 32–41 and 71 were attributed to pressure effects on K_open rather than on k_ex in the EX2 limit, suggesting destabilization of the hydrogen bonds and partial exposure of the amide groups of those residues to the solvent in N₂. Considering the previous and present results, the displacements of the residues and water filling (i.e., water penetration) in the void volume around the C-terminal side of the protein were likely major causes of the negative volume changes (i.e., −25 mL/mol) of the transition from N₁ to N₂.

Even amide groups hydrogen bonded with water have nonzero protection by the nearest neighbor from HX. About 6 kJ/mol of ΔG°_HX at 0.1 MPa seem to correspond to the protection of the amide groups in N₁ from HX (see Fig. 4 D). Large values of ΔG_HX for residues 32, 33, 35, and 40 can be explained by excess free energy of the transition from N₁ to N₂ (i.e., 3.4 kJ/mol) (44). In contrast, negative ΔG°_HX were estimated to residues 8 and 73 (see Fig. 4 D), indicating that these amide sites favor water-accessible open conformations at 0.1 MPa (i.e., k_open > k_close).

Next, we analyzed pressure-induced loss of the reference HSQC signals of residues 12, 46, and 74. All of them showed k_obs values higher (>20 s⁻¹) than those of others at 0.1 MPa and pH 7.36. As discussed above, k_ex should be increased ∼3.1-fold by every increase in pH by 0.49 units. Like under high pressure, the reference HSQC signals of residues 46 and 74 disappeared in the spectrum at high pH of 7.85. Therefore, the pressure-induced loss of the reference HSQC signals was not associated with peak broadening due to pressure-induced enhancements of conformational fluctuation but could be simply explained by the pressure effects on k_ex (e.g., k_{ex(250 MPa)}/k_{ex(0.1 MPa)} = 2.3): k_obs may exceed the NMR timescale (i.e., ∼ms) under higher pressure, when the sample pH is higher than neutral. Notably, k_ex increases more significantly under heat denaturation conditions (e.g., ∼70°C), so that k_ex(70°C)/k_ex(20°C) = 71. Indeed, NMR signals of the majority of amide groups of ubiquitin disappeared in the spectrum at 90°C (45).

Finally, we compared the current results with those of NOE between water and amide protons under high pressure. Hydration sites and dynamics at the surface of ubiquitin in aqueous solution and encapsulated in bis(2-ethylhexyl) sulfosuccinate reverse micelles in pentane were revealed by NOE and rotating-flame NOE measurements (20). In general, it is very difficult to detect dipole-dipole interactions between hydration water and protein protons, because hydration water has extremely shorter residence time than the rotational correlation time of proteins (e.g., ∼ns order). Even in the protein-encapsulated condition, where the reorientational dynamics of encapsulated water is slowed relative to bulk water by approximately one order of magnitude (20), NOE and rotating-flame NOE were not detected or were relatively weak for the residues 34−40, indicating the extremely short residence time of the hydration water around the residues in the N₁ form. Although pressure-induced hydration of ubiquitin was investigated only by ¹⁵N-resolved NOESY experiments, NOE between water and amide protons of the residues was not detected up to 250 MPa (pH 5.0 and 303 K) for the protein both in aqueous solution and encapsulated in bis(2-ethylhexyl) sulfosuccinate reverse micelles in pentane (46). According to the previous results, hydration water around the residues 32−40 seems to have still extremely short residence time in the N₂ form as well as the N₁ form.

Conclusions

Structural and thermodynamic characteristics of the high-energy conformation N₂ of ubiquitin have been revealed by high-pressure NMR spectroscopy. However, the involvement of water molecules in the conformational transition from N₁ to N₂ has been unknown. In our experiments, the CLEANEX-PM NMR approach coupled with the pressure perturbation showed preferential hydration at the C-terminal side of ubiquitin when N₂ was stabilized by pressure. The current results provide a detailed information of water penetration into the void space of the protein C-terminal side as revealed by 1-μs MD simulation (26). Thus, CLEANEX-PM NMR measurements under high pressure will be a useful approach for investigating more open and hydrated forms of protein structure that are more relevant to protein function.

Author Contributions

S.K., Y.A., and T.W. made protein samples. S.K., Y.A., and R.K. performed NMR experiments and spectral and thermodynamic analyses. R.K. designed the research and wrote the article.

Editor: Wendy Shaw.