Enthalpic stabilization of an SH3 domain by D₂O

Abstract

The stability of a protein is vital for its biological function, and proper folding is partially driven by intermolecular interactions between protein and water. In many studies, H₂O is replaced by D₂O because H₂O interferes with the protein signal. Even this small perturbation, however, affects protein stability. Studies in isotopic waters also might provide insight into the role of solvation and hydrogen bonding in protein folding. Here, we report a complete thermodynamic analysis of the reversible, two‐state, thermal unfolding of the metastable, 7‐kDa N‐terminal src‐homology 3 domain of the Drosophila signal transduction protein drk in H₂O and D₂O using one‐dimensional ¹⁹F NMR spectroscopy. The stabilizing effect of D₂O compared with H₂O is enthalpic and has a small to insignificant effect on the temperature of maximum stability, the entropy, and the heat capacity of unfolding. We also provide a concise summary of the literature about the effects of D₂O on protein stability and integrate our results into this body of data.

Introduction

Water, arguably the most important molecule for life on Earth,1, 2 is essential for the stability, folding, and structure of proteins that drive biology. The hydrogen bonds between protein and water help shape the free energy landscape of folding, guiding a protein towards its stable, folded state. In many experimental techniques, however, the signal from H₂O interferes with that from the protein, and D₂O is used as the solvent. Nuclear magnetic resonance spectroscopy (NMR), infrared spectroscopy, small angle X‐ray scattering, and small angle neutron scattering are techniques that often incorporate this substitution. Additionally, examining proteins in D₂O, which is only a small perturbation of the system, can provide insight into the role of hydration and hydrogen‐bonding in protein folding.3, 4, 5, 6, 7, 8 Nonetheless, this solvent substitution can affect proteins,9 and it is important to be cognizant of these effects when conducting experiments in D₂O. The aim of the present study is three‐fold: provide a complete thermodynamic analysis of globular protein folding in H₂O and D₂O, concisely summarize similar literature studies, and compare our results to those in the literature.

We chose the metastable, 7‐kDa N‐terminal src homology 3 domain of the Drosophila signal transduction protein drk (SH3) to probe solvent isotope effects on protein stability. Even under non‐denaturing conditions, a large population of SH3 is unfolded.10 SH3 has one tryptophan, which we labeled with a fluorine atom.11 This residue experiences different solvent exposure in the folded and unfolded states, resulting in two ¹⁹F resonances in slow exchange on the NMR timescale,10, 12 one for the folded state and one for the unfolded ensemble (Fig. 1). The presence of only two resonances is consistent with two‐state folding. The areas under the resonances can be integrated to obtain the relative populations of each state and thus a modified standard‐state free energy of unfolding:

$$\mathrm{\Delta }{G}_U^{{°}^{\prime }}=-\mathit{\text{RTln}}\frac{\mathit{\text{population unfolded}}}{\mathit{\text{population folded}}}$$ (1)

Figure 1.

Two‐state, temperature‐dependent, reversible SH3 folding in H₂O and D₂O monitored with ¹⁹F NMR. (A) One‐dimensional spectra in H₂O at nine temperatures normalized to the intensity of the folded state peak at 5°C. Subsequent spectra are offset by −0.2 ppm to aid visualization. The downfield resonance is the folded state (F), and the upfield resonance is the unfolded ensemble (U). (B) Spectra at 25°C before (black) and after (gray) an experiment where the temperature ranged from 5°C to 45°C over approximately 180 min indicating the reversibility of denaturation. (C) Spectra in H₂O (black) and D₂O (red) at 45°C. The D₂O spectrum is normalized to the chemical shift and intensity of the folded state resonance in H₂O.

where R is the gas constant and T is the absolute temperature. The temperature dependence of $\mathrm{\Delta }{G}_U^{{°}^{\prime }}$ is used to construct a protein stability curve (Fig. 2)13 which provides a complete thermodynamic picture of folding in H₂O and D₂O via the integrated Gibbs–Helmholtz equation:

$$\mathrm{\Delta }{G}_{\mathrm{U}}^{{°}^{\prime }}=\mathrm{\Delta }{H}_{\mathrm{U}}^{{°}^{\prime }}-T\mathrm{\Delta }{S}_{\mathrm{U}}^{{°}^{\prime }}+\mathrm{\Delta }{C}_p^{{°}^{\prime }}\left[T-T_{\mathrm{ref}}-Tln\left(\frac{T}{T_{\mathrm{ref}}}\right)\right]$$ (2)

Figure 2.

Temperature dependence of SH3 stability in H₂O and D₂O. Error bars represent standard errors of the mean from three independent trials.

where T_ref is a reference temperature, $\mathrm{\Delta }{H}_{\mathrm{U}}^{{°}^{\prime }}$ and $\mathrm{\Delta }{S}_{\mathrm{U}}^{{°}^{\prime }}$are the modified standard state enthalpy and entropy of unfolding, respectively, and $\mathrm{\Delta }{C}_p^{{°}^{\prime }}$ is the modified standard state heat capacity of unfolding which is assumed to be temperature independent within the range studied.

Results and Discussion

Heat‐induced unfolding of SH3 in H₂O and D₂O

The metastability of this SH3 domain allows stability curves to be constructed and analyzed13, 14, 15 at reasonable temperatures (5°C–45°C). The stability curve in D₂O is simply shifted above the curve in H₂O (Fig. 2). This shift indicates that SH3 stability is greater in D₂O at all temperatures and that the effects of D₂O are mainly enthalpic.9, 14 $\mathrm{\Delta }{C}_p^{{°}^{\prime }}$, as well as $\mathrm{\Delta }{G}_{\mathrm{U}}^{{°}^{\prime }}$, $\mathrm{\Delta }{S}_{\mathrm{U}}^{{°}^{\prime }}$, and $\mathrm{\Delta }{H}_{\mathrm{U}}^{{°}^{\prime }}$ at any temperature can be calculated by fitting the data to Equation (2).

The thermodynamic parameters (Table 1) paint a picture of the solvent isotope effect. At 318 K, approximately half way between T_m in D₂O and H₂O, the heavy water stabilizes the protein by nearly 1 kcal/mol [Fig. 1(C)]. This increase is also visible by comparing the T_m (the melting temperature, where $\mathrm{\Delta }{G}_{\mathrm{U}}^{{°}^{\prime }}=0$). D₂O increases this value by approximately 12 K, indicating an increased thermal stability. The higher stability in D₂O is often rationalized in terms of the increased difficulty of cavity creation in D₂O compared with H₂O.9, 16, 17

Table 1.

Equilibrium thermodynamic parameters for SH3 in H₂O and D₂O

Condition	$\mathrm{\Delta }{G}_{\mathrm{U},318.15K}^{{°}^{\prime }}$ (kcal/mol)a	$\mathrm{\Delta }{H}_{\mathrm{U},T_{\mathrm{m}}}^{{°}^{\prime }}$ (kcal/mol)b	$\mathrm{\Delta }{H}_{\mathrm{U},T_{\mathrm{s}}}^{{°}^{\prime }}$ (kcal/mol)b	$\mathrm{\Delta }{H}_{\mathrm{U},318.15K}^{{°}^{\prime }}$ (kcal/mol)c	$\mathrm{T\Delta }{S}_{\mathrm{U},318.15K}^{{°}^{\prime }}$ (kcal/mol)c	$\mathrm{\Delta }{C}_P^{{°}^{\prime }}$ (kcal/mol/K)b	Tm (K)b	Ts (K)b
H2O	−0.52 ±0.02	22 ±2	0.89 ±0.03	28 ±1	28 ±1	0.86 ±0.09	311 ±1	287 ±1
D2O	0.46 ±0.04	33 ±2	1.8 ±0.1	29 ±1	28 ±1	0.89 ±0.07	323 ±1	288 ±1

^a Uncertainties determined from standard error of the mean from triplicate experimental analysis.

^b Uncertainties determined from 95% confidence intervals of fit to the integrated Gibbs–Helmholtz equation.

^c The temperature half way between T_m in H₂O and D₂O. Values from Kirchhoff's equations and uncertainties by error propagation from the uncertainties in $\mathrm{\Delta }{C}_p^{{°}^{\prime }}$, T_ref (T_m or T_s), and $\mathrm{\Delta }{H}_{\mathrm{U},\mathrm{ref}}^{{°}^{\prime }}$ ($\mathrm{\Delta }{H}_{\mathrm{U},{\mathrm{T}}_{\mathrm{m}}}^{{°}^{\prime }}$ or $\mathrm{\Delta }{H}_{\mathrm{U},T_{\mathrm{s}}}^{{°}^{\prime }}$)

Breaking $\mathrm{\Delta }{G}_{\mathrm{U}}^{{°}^{\prime }}$ into its enthalpic and entropic components shows only an effect on the enthalpy of unfolding. In both H₂O and D₂O, the enthalpy and entropy of unfolding are large and positive making it difficult to determine which one dominates at 318 K. Examining the enthalpy at the temperature of maximum stability, T_s, is more useful due to the small uncertainty in the value at this temperature. At T_s the entropy is zero,13 and therefore differences in enthalpy are identified. Additionally, T_s is nearly the same in H₂O and D₂O. The $\mathrm{\Delta }{H}_{\mathrm{U},T_{\mathrm{S}}}^{{°}^{\prime }}$ in H₂O is 0.89 ±0.03 kcal/mol, whereas in D₂O, it is twice as large (1.8 ±0.1 kcal/mol). The curvature is the same in both solvents. This is quantitatively demonstrated by a minimal change in $\mathrm{\Delta }{C}_p^{{°}^{\prime }}$. In summary, at all temperatures, the increased stability of SH3 in D₂O is dominated by $\mathrm{\Delta }{H}_{\mathrm{U}}^{{°}^{\prime }}$.

Origins of the D₂O effect

The molecular origins of this enthalpic stabilization are difficult to pinpoint, especially since they cannot be directly measured. D₂O stabilization of proteins is often attributed to an increase in hydrogen bond strength in heavy water,18, 19, 20, 21, 22 which is also consistent with the observation that D₂O reduces protein flexibility.23, 24 Although a change in solvent hydrogen bond strength would be reflected in a change in enthalpy because there are numerous solvent–solvent, solvent–protein, and protein–protein hydrogen bonds formed during protein folding, there are multiple contributions to the enthalpy of unfolding.25 In addition to the enthalpy from hydrogen bond formation and breakage, solvation enthalpy also plays a significant role in protein folding.9 Protein unfolding involves solvation of groups that are buried in the folded state. The enthalpy of solvation for a particular protein is, therefore, based on its sequence, structure, and changes in the solvent accessible surface area upon unfolding. Solvation enthalpies are typically based on studies of small molecules, which are then used in combination to calculate a net enthalpy of protein solvation.9, 25 These solvation enthalpies are large in magnitude and opposite in sign for apolar versus polar groups, often resulting in small estimated net enthalpy changes of both signs for an entire protein. It is likely that the observed increase in the enthalpy of SH3 unfolding in D₂O arises from a combination of solvation and hydrogen‐bond effects.

The heat capacity of unfolding, $\mathrm{\Delta }{C}_p^{{°}^{\prime }},$ which is related to solvation changes upon unfolding, is often difficult to quantitatively interpret due to the relatively large uncertainty in its value (~10%, Table 1).9 Additionally, we assume that $\mathrm{\Delta }{C}_p^{{°}^{\prime }}$ does not change with temperature, which is not necessarily true,13, 25, 26 but a good assumption over a small range, like the one used here.13, 27 For SH3, $\mathrm{\Delta \Delta }{C}_p^{{°}^{\prime }}$ ($\mathrm{\Delta }{C}_{p,D_{2}O}^{{°}^{\prime }}-\mathrm{\Delta }{C}_{p,H_{2}O}^{{°}^{\prime }}$) is 0.0 ±0.1 kcal/mol/K, meaning that any change is too small to interpret, which suggests that $\mathrm{\Delta }{C}_p^{{°}^{\prime }}$ is constant from 5 °C to 45 °C. This conclusion is consistent with other observations.9, 22, 28 As suggested in the previous paragraph, the change in the heat capacity of transfer of hydrophilic versus hydrophobic protein groups from light‐ to heavy‐water are often large and opposite in sign,9 resulting in a minimal and uncertain change in $\mathrm{\Delta }{C}_p^{{°}^{\prime }}$.

Literature studies find D₂O is primarily stabilizing

The effects of D₂O on protein stability have been of interest for decades (Table 2). The purpose of Table 2 is to highlight peer‐reviewed publications in which the stability of a protein is directly compared in H₂O and D₂O. The majority of studies reveal that D₂O stabilizes proteins. The parameter most used to assess the influence of heavy water is its effect on the melting temperature, T_m. Although the degree to which D₂O increases the T_m of a particular protein varies, our data is in accord with the literature19, 21, 22, 28, 30, 31, 32, 36, 37 in that the T_m of a protein increases upon changing H₂O to D₂O. In all studies, but one,32 that report $\mathrm{\Delta }{G}_{\mathrm{U}}^{{°}^{\prime }}$ along with T_m, an increased melting temperature is accompanied by an increase in the free energy of unfolding.22, 28, 35 In the case of phycocyanin,30 an increase in the activation free energy of denaturation is observed.

Table 2.

Effects of D₂O on Protein Stability

Protein	Method	Effect of D2O	Parameter(s) examined
Ovalbumin29	Urea, polarimetry	Stabilizing	t 1/2
Ribonuclease19, 21	Heat, polarimetry	Stabilizing	T m
Phycocyanin30	Heat, absorbance spectroscopy, fluorescence quenching	Stabilizing	T m , ΔH°′ ‡ , ΔS°′ ‡ , ΔG°′ ‡
Staphylococcal nuclease31	Heat, GdnSCN, GdnHCl, NMR, CD, fluorescence spectroscopy	Stabilizing	T m
Bovine ribonuclease A32	DSC	Small	ΔH°′, ΔG°′, T m
Hen egg lysozyme32	DSC	Destabilizing	ΔH°′, ΔG°′, T m
Cytochrome c 32	DSC	Destabilizing	ΔH°′, ΔG°′, T m
Malate dehydrogenase33	Enzyme assay	Stabilizing	Residual enzyme activity
Staphylococcal nuclease34	Urea, CD, FTIR	Stabilizing	m‐Value, ΔG°’
Domain 1 of rat CD235	GdnHCl, stopped‐flow fluorescence spectroscopy	Stabilizing	ΔG°′, ΔG°′ ‡, k I‐F , k F‐I , m‐values
NTL928	Heat, urea, GdnHCl, far‐UV CD	Stabilizing	T m , ΔG°′, ΔH°′, ΔS°′, m‐values, ΔC p °′
β‐lactoglobulin36	DSC, DLS	Stabilizing	T m , ΔH°′
Ribonuclease A37	Heat, urea, CD, fluorescence spectroscopy, HDX NMR	Stabilizing	T m , ΔH°, ΔC p °′
Ribonuclease T137	Heat, urea, CD, fluorescence spectroscopy, HDX NMR	Stabilizing	T m , ΔH°, ΔC p °’
Polyproline type II helix38	CD	Stabilizing	Polyproline II content
Hen egg lysozyme22	DSC	Stabilizing	T m , ΔG°′, ΔH°′, ΔS°′, ΔC p °’
Bovine serum albumin22	DSC	Stabilizing	T m , ΔG°′, ΔH°′, ΔS°′, ΔC p °’
Bovine serum albumin39	Heat, far‐UV CD	Stabilizing	Molar ellipticity

Abbreviations: t _1/2, half‐time of denaturation reaction; GdnSCN, guanidine thiocyanate; GdnHCl, guanidine hydrochloride; NMR, nuclear magnetic resonance spectroscopy; CD, circular dichroism spectropolarimetry; k _I‐F, rate of intermediate to folded state reaction; k _F‐I, rate of folded state to intermediate reaction; NTL9, N‐terminal domain of the ribosomal protein L9; UV, ultraviolet; ΔC _p °′, change in heat capacity; DSC, differential scanning calorimetry; DLS, dynamic light scattering; HDX, hydrogen–deuterium exchange.

Additional equilibrium thermodynamic parameters describing protein stability became more commonplace with the advent of highly accurate calorimeters. Differential scanning calorimetry is particularly useful because the melting temperature, enthalpy, and heat capacity of unfolding can be measured.40, 41 All but one study32 reports results similar to ours: D₂O increases the enthalpy of unfolding.22, 30, 36, 40 One report28 shows minimal changes to $\mathrm{\Delta }{H}_{\mathrm{U}}^{{°}^{\prime }}$ and increases in both T_m and $\mathrm{\Delta }{G}_{\mathrm{U}}^{{°}^{\prime }}$. Attempts have been made to describe the molecular basis of this change in enthalpy, often attributed to increased hydrogen‐bond strength. As described here, however, there are likely multiple contributions making it difficult to ascribe its effects to hydrogen bonding or solvation alone. Some investigators also report the entropy of unfolding28, 30; but like enthalpy, it contains multiple contributions.25 In summary, our results correspond to almost all published observations: D₂O increases the stability ($\mathrm{\Delta }{G}_{\mathrm{U}}^{{°}^{\prime }}$),22, 28, 30, 34, 35, 37 the melting temperature (T_m),19, 21, 22, 28, 30, 31, 32, 36, 37 and the enthalpy of protein unfolding ($\mathrm{\Delta }{H}_{\mathrm{U}}^{{°}^{\prime }}$).22, 30, 36, 37

D₂O affects many biological processes

Although we focus on protein stability, the effects of D₂O on many biological processes have been investigated with the potential for widespread impact on the fundamental roles of water in biology and therapeutics. D₂O affects protein–carbohydrate, protein–peptide, and protein–nucleic acid interactions,42 with effects on the enthalpy of binding. In addition to binding, D₂O enhances protein oligomerization and aggregation,36, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52 by what has been suggested to be the promotion of hydrophobic interactions. Given the influence of heavy water on biomolecular reactions, D₂O is also expected to affect whole organisms. Research in this field began as soon as the deuterium isotope was discovered53 and isolated54 in the 1930s. High concentrations of D₂O have deleterious effects on organismal growth and survival from microorganisms like Escherichia coli 55, 56 and yeast57, 58 to algaes,59, 60 plants,58, 61, 62 and animals such as mice58, 60, 63 and dogs55. However, the natural abundance of deuterium in nature is approximately 156 ppm.64, 65 More recent studies show that low concentrations may be necessary and even beneficial,66, 67 with interesting recent hypotheses on the use of heavy isotopes for increasing human longevity.68 Finally, there is some interest in the use of D₂O as an excipient52, 69 because of the observation that D₂O can stabilize vaccines.70

Understanding the effects of isotopic waters is key to understanding biology, including protein folding. We focused on the equilibrium thermodynamics of D₂O on protein stability, because stably folded proteins are often a pre‐requisite to proper biological function. We anticipate that our results and those of others compiled here will be of use for understanding the effects of the molecule that unites all of life on Earth.

Materials and Methods

Protein expression and purification

5‐Fluorotryptophan‐labeled SH3 was expressed and purified as described.71, 72

NMR

NMR samples were prepared as described.14, 15, 71 Briefly, 1 mg of fluorine‐labeled, SH3 was resuspended in NMR buffer (50 mM HEPES, bis–tris propane, sodium acetate/acetic acid, pH 7.2) made using H₂O or 99.9% D₂O. pH readings are direct measurements and uncorrected for the D₂O isotope effect.73 For samples prepared in H₂O, a coaxial‐insert containing D₂O was used to lock the spectrometer. 4,4‐Dimethyl‐4‐silapentane‐1‐sulfonic acid (DSS, Cambridge Isotope Laboratories, Cambridge, UK) was used as a chemical shift reference. One‐dimensional ¹⁹F spectra were acquired at 5°C increments between 5°C and 45°C on a Bruker Avance III HD spectrometer operating at a ¹⁹F Larmor frequency of 470 MHz equipped with a Bruker QCI cryoprobe.

Data processing and analysis

Data were processed as described using Topspin 3.2.14, 15, 71 The parameters shown in Table 1 were calculated using Kirchhoff's equations and the integrated Gibbs‐Helmholtz equation as described14 using MATLAB R2016a.

Note added in proof. After our manuscript was accepted, we learned about work that more completely explains the stabilizing effect of D2O. [Pica A, Graziano G (2018) Effect of heavy water on the conformational stability of globular proteins. Biopolymers, 2017].