Atomic view of cosolute-induced protein denaturation probed by NMR solvent paramagnetic relaxation enhancement

Abstract

The cosolvent effect arises from the interaction of cosolute molecules with a protein and alters the equilibrium between native and unfolded states. Denaturants shift the equilibrium toward the latter, while osmolytes stabilize the former. The molecular mechanism whereby cosolutes perturb protein stability is still the subject of considerable debate. Probing the molecular details of the cosolvent effect is experimentally challenging as the interactions are very weak and transient, rendering them invisible to most conventional biophysical techniques. Here, we probe cosolute-protein interactions by means of NMR solvent paramagnetic relaxation enhancement together with a formalism we recently developed to quantitatively describe, at atomic resolution, the energetics and dynamics of cosolute-protein interactions in terms of a concentration normalized equilibrium average of the interspin distance, [Formula: see text], and an effective correlation time, τ_c The system studied is the metastable drkN SH3 domain, which exists in dynamic equilibrium between native and unfolded states, thereby permitting us to probe the interactions of cosolutes with both states simultaneously under the same conditions. Two paramagnetic cosolute denaturants were investigated, one neutral and the other negatively charged, differing in the presence of a carboxyamide group versus a carboxylate. Our results demonstrate that attractive cosolute-protein backbone interactions occur largely in the unfolded state and some loop regions in the native state, electrostatic interactions reduce the [Formula: see text] values, and temperature predominantly impacts interactions with the unfolded state. Thus, destabilization of the native state in this instance arises predominantly as a consequence of interactions of the cosolutes with the unfolded state.

Keywords: NMR relaxation; drkN SH3 native and unfolded states; protein–cosolute interactions; replica exchange molecular dynamics; transient states.

Fig. 1.

Summary of overall strategy employed to study cosolute interactions with native and unfolded drkN SH3 at atomic resolution. The native and unfolded states are in slow exchange on the chemical shift timescale enabling solvent PREs (sPREs) arising from paramagnetic cosolutes (3-carbamoyl and 3-carboxy PROXYL) to be measured simultaneously on individual backbone amide protons for both states. The experimental spectral density function, J(ω), is mapped at several frequencies by measuring the transverse (zero frequency) and longitudinal (at several spectrometer frequencies) sPREs on each amide proton. The experimental data are then fitted to an ansatz spectral density function, J_approx (ω), integration of which yields two residue-specific parameters that describe the energetics (measured as a concentration normalized equilibrium average of the interspin distance between the electron spin on the cosolute and the ¹H spin on the protein, ${\langle r^{-6}\rangle }_{\text{norm}}$) and dynamics (provided by an effective correlation time τ_c) of cosolute–protein interactions at atomic resolution.

Fig. 2.

Probing the effect of neutral and negatively charged paramagnetic cosolutes on the folding/unfolding equilibrium of drkN SH3 by smFRET. (A) Structures of the two paramagnetic cosolutes employed in the current work: 3-carbamoyl-PROXYL (3CY, neutral) and 3-carboxy-PROXYL (3CX, negatively charged). The location of the unpaired electron is indicated by the black dot. (B) smFRET efficiency histogram as a function of paramagnetic cosolute concentration: Left, 3-carbamoyl PROXYL; Right, 3-carboxy PROXYL. (C) Fraction native population as a function of paramagnetic cosolute concentration (blue, 3-carbamoyl PROXYL; red, 3-carboxy PROXYL) derived from analysis of the smFRET data. Details of the experimental conditions and analysis are provided in SI Appendix.

Fig. 3.

Experimental backbone amide sPRE profiles for native and unfolded drkN SH3 in the presence of neutral and negatively charged paramagnetic cosolutes. sPRE data were obtained for 0.2 mM ²H/¹⁵N-labeled drkN SH3 in the presence of 25 mM (A) 3-carbamoyl PROXYL and (B) 3-carboxy PROXYL paramagnetic cosolutes at 298 (Left) and 277 (Right) K. Top and Bottom display the experimental ¹H_N-Γ₂ and ¹H_N-Γ₁ profiles, respectively, measured at 500 (blue), 800 (green), and 900 (red) MHz. The ¹H_N-R₁ and R₂ rates at 298 K, used to calculate the sPRE values, are corrected to take into account chemical exchange between the native and unfolded states (SI Appendix and SI Appendix, Fig. S3); at 277 K exchange between native and unfolded states is too slow to impact the measured ¹H_N relaxation data and calculated ${\langle r^{-6}\rangle }_{\text{norm}}$ values (SI Appendix, Fig. S4). The location of the β-strands in the native state is shown at the Top of A: strands β1, β2, β3, β4, and β5 comprise residues 2–5, 23–29, 36–41, 44–49, and 53–55, respectively.

Fig. 4.

Experimentally derived ${\langle r^{-6}\rangle }_{\text{norm}}$ and τ_C profiles obtained from the analysis of sPRE measurements on native and unfolded drkN SH3 in the presence of neutral and negatively charged paramagnetic cosolutes. Data were acquired at two temperatures: (A) 298 and (B) 277 K. The experimentally derived values are shown as circles, with the exception of those for the side-chain N_ε1H indole proton of Trp36, which are represented by a star. ${\langle r^{-6}\rangle }_{\text{norm}}$ values with errors > 7 × 10²⁸ m⁻³ and τ_C values with errors > 0.75 ns have been excluded from the plots. The sPRE data used to calculated ${\langle r^{-6}\rangle }_{\text{norm}}$ and τ_c were collected on 0.2 mM ²H/¹⁵N-labeled drkN SH3 in the presence of 25 mM 3-carbamoyl (3CY, neutral) or 3-carboxy (3CX, negative) PROXYL. For comparison, the ${\langle r^{-6}\rangle }_{\text{norm}}^{\text{exc}}$ profiles calculated directly from the coordinates of drkN SH3 in the native [PDB ID code 2AZS (32)] and unfolded states (see SI Appendix for details), taking into account only the excluded volume with no intermolecular forces between protein and cosolute, are displayed as continuous black and gray lines, respectively. The ${\langle r^{-6}\rangle }_{\text{norm}}^{\text{exc}}$ values in the native state were calculated for the 10 NMR structures deposited in the 2AZS coordinates to account for different surface side-chain conformers, and the error bars (black vertical lines) represent the SDs among these 10 structures. Similarly, for the unfolded state ${\langle r^{-6}\rangle }_{\text{norm}}^{\text{exc}}$ profiles, the error bars represent the SDs among the 505 structure snapshots obtained from the replica exchange MD trajectory at either 298 or 277 K (see SI Appendix for details; also note there is no noticeable difference in the ${\langle r^{-6}\rangle }_{\text{norm}}^{\text{exc}}$ values of the unfolded state ensemble calculated from the MD simulations at 298 and 277 K; see SI Appendix, Fig. S7B).

Fig. 5.

Mapping sites of preferential interaction of neutral and negatively charge paramagnetic cosolutes with the native and unfolded states of drkN SH3. (A) $\left[{⟨r^{-6}⟩}_{\text{norm}}-{⟨r^{-6}⟩}_{\text{norm}}^{\text{exc}}\right]$ profiles obtained for (A) 3-carbamoyl PROXYL (3CY, neutral) and (B) 3-carboxy PROXYL (3CX, negative) at 298 and 277 K. Correlation between ${\langle r^{-6}\rangle }_{\text{norm}}$ values obtained at 298 and 277 K for the (C) native and (D) unfolded states of drkN SH3. The experimentally derived values are shown as circles, with the exception of those for the side-chain N_ε1H indole proton of Trp36, which are represented by a star. The continuous lines in panels (C and D) are the best-fits to the linear relationship y = mx.

Fig. 6.

Correlation of ${\langle r^{-6}\rangle }_{\text{norm}}$ and τ_C values obtained with 3-carbamoyl and 3-carboxy PROXYL paramagnetic cosolutes. Correlations for the (A) native and (B) unfolded states of drkN SH3 at two temperatures (298 and 277 K). The continuous lines are the best-fits to the linear relationship y = mx.

Fig. 7.

Comparison of the local preferential interaction coefficient, Λ_IP, between Cl⁻ ions and backbone amide protons determined from MD simulations and the difference in ${\langle r^{-6}\rangle }_{\text{norm}}$ profiles between neutral and negatively charged paramagnetic cosolutes for the native and unfolded states of drkN SH3. (A) Λ_IP (Top) and Δ${\langle r^{-6}\rangle }_{\text{norm}}$ (Bottom) profiles at two temperatures (298 and 277 K). Λ_IP is defined by Eq. 11; Δ${\langle r^{-6}\rangle }_{\text{norm}}$ is defined as [${\langle r^{-6}\rangle }_{\text{norm}}$(3CX) − ${\langle r^{-6}\rangle }_{\text{norm}}$(3CY)], where 3CY and 3CX are 3-carbamoyl and 3-carboxy PROXYL, respectively. (B) Δ${\langle r^{-6}\rangle }_{\text{norm}}$ color graded on the molecular surface of native drkN SH3 from orange (−1.2 × 10²⁹ m⁻³ molecule) through white (0 × 10²⁹ m⁻³ molecule) to green (+1.2 × 10²⁹ m⁻³ molecule) (C) Λ_IP are color graded on the molecular surface of native drkN SH3 from red (−0.0035 molecule) through white (0 molecule) to blue (−0.0035 molecule). Residues that were not analyzed are shown in gray. The coordinates of native drkN SH3 are taken from PDB 2AZS (32), and the molecular surfaces were generated in the PyMol Molecular Graphics System (Version 2.0; Schröder LLC).