Regulation of Surface Charge by Biological Osmolytes - Journal of the

Article pubs.acs.org/JACS

Cite This: J. Am. Chem. Soc. 2017, 139, 15013-15021

Regulation of Surface Charge by Biological Osmolytes Roy Govrin,*,‡ Itai Schlesinger,‡ Shani Tcherner, and Uri Sivan* Department of Physics and the Russell Berrie Nanotechnology Institute, Technion−Israel Institute of Technology, Technion City, Haifa 3200003, Israel

Downloaded via UNIV OF FLORIDA on July 22, 2018 at 13:03:49 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Osmolytes, small molecules synthesized by all organisms, play a crucial role in tuning protein stability and function under variable external conditions. Despite their electrical neutrality, osmolyte action is entwined with that of cellular salts and protons in a mechanism only partially understood. To elucidate this mechanism, we utilize an ultrahigh-resolution frequency modulation-AFM for measuring the effect of two biological osmolytes, urea and glycerol, on the surface charge of silica, an archetype protic surface with a pK value similar to that of acidic amino acids. We find that addition of urea, a known protein destabilizer, enhances silica’s surface charge by more than 50%, an effect equivalent to a 4-unit increase of pH. Conversely, addition of glycerol, a protein stabilizer, practically neutralizes the silica surface, an effect equivalent to 2-units’ reduction of pH. Simultaneous measurements of the interfacial liquid viscosity indicate that urea accumulates extensively near the silica surface, while glycerol depletes there. Comparison between the measured surface charge and Gouy−Chapman−Stern model for the silica surface shows that the modification of surface charge is 4 times too large to be explained by the change in dielectric constant upon addition of urea or glycerol. The model hence leads to the conclusion that surface charge is chiefly governed by the effect of osmolytes on the surface reaction constants, namely, on silanol deprotonation and on cation binding. These findings highlight the unexpectedly large effect that neutral osmolytes may have on surface charging and Coulomb interactions.

■

INTRODUCTION

The high concentration of stabilizing osmolytes found in microorganisms adapted to elevated salt concentrations29 or organisms exposed to elevated salt concentrations in the laboratory30 hints that neutral stabilizing osmolytes may compensate for the adverse effect of salt. Early studies showed, indeed, that the effect of osmolytes is entwined with that of cellular salts and protons.1,12,31 Enzymatic activity, inhibited by excess salt, was found to be revived by added osmolytes,12 and the effects (e.g., m-values) of essentially all types2 of osmolytes were found to depend on pH.31−34 Enzymatic activity, protein conformation, and protein hydration35−37 are governed by surface charge and Coulomb interactions. It comes as no surprise then that screening brought by high concentrations of salts inhibits protein activity by disrupting substrate binding to active enzymatic sites. The mechanisms underlying restoration of enzyme function by added neutral osmolytes remain obscure and motivated much of the work presented below. A priori, the correcting effect of stabilizing osmolytes could be assigned to one or more of three distinct mechanisms. The first is counteracting destabilization by salt with an independent stabilizing effect of osmolytes. The second is osmolyte effect on the medium dielectric constant, and hence on coulomb interactions, and the third is osmolyte effect on the immediate environment of protic

Osmolytes are small, usually neutral, soluble molecules that constitute a major fraction of the bulk of solutes in cells.1 They are synthesized by all organisms to control the cell’s osmolarity and help coping with stress conditions such as high salt and low temperatures. Examples include polyols (e.g., glycerol and sorbitol), certain free amino acids and their derivatives (e.g., proline and glycine-betaine), and urea.2 Previous studies identified diverse functions of osmolytes pertaining to structural stability and solubility of protein3−8 and nucleic acid complexes,9−11 protection of enzyme activity against the adverse effect of salt,12 and protection of organisms against extreme temperatures.13−15 Current and prospective applications of osmolytes in industry and medicine further include cryoprotection,16,17 protein separation,18 energy storage,19 improved salinity tolerance in plants,20 drug delivery,21 and artificial kidneys.22 The prevailing view links osmolyte function to their propensity to accumulate or deplete near biological surfaces. Urea, a known protein destabilizer, was shown to accumulate near DNA,10 peptides, and polypeptides.4,23,24 Protein stabilizers (or osmoprotectans), e.g., polyols, were found to deplete near macromolecular surfaces.8,25,26 Models, such as inverse Kirkwood−Buff analysis,27 link the molecular basis of the aforementioned stabilizing effect with unfavorable interactions of stabilizers with the exposed protein backbone.3,25,28 © 2017 American Chemical Society

Received: July 6, 2017 Published: October 3, 2017 15013

DOI: 10.1021/jacs.7b07036 J. Am. Chem. Soc. 2017, 139, 15013−15021

Article

Journal of the American Chemical Society

Due to deprotonation of silanol groups, the surface of silica acquires a negative charge at physiological pH, leading to double layer repulsion between two surfaces. When urea was added to the solution, the surface charge and double layer repulsion were enhanced dramatically, in accord with the denaturing effect of urea on proteins. Addition of glycerol, on the other hand, diminished the repulsion between the two surfaces, concurring with the stabilizing capacity of glycerol on proteins. Force and dissipation rate were measured simultaneously versus tip−sample distance using an ultrahigh-resolution atomic force microscope53 operating in frequency modulation (FM) mode. Silica surface charge densities at varying osmolyte concentrations were extracted by fitting the force by the DLVO model,54 and the effect of osmolytes was compared with the prediction of a 3 pK Gouy−Chapman−Stern model of the silica−solution interface.46,55 In this model, the influence of osmolytes is embodied in their effect on the Debye−Hückel screening length, as well as on the Stern layer capacitance. The increased dielectric constant in the presence of urea led, in accord with the experimental observation, to enhanced surface charge. Conversely, the reduced dielectric constant in the presence of glycerol led to a smaller surface charge, again, concurring with the experimental results. Notwithstanding this qualitative agreement, the overall change in surface charge predicted by the model was significantly smaller than the experimental one, indicating that the regulation of surface charge by osmolytes extends beyond their effect on the medium dielectric constant and on Stern capacitance into modulation of the silanol ionization constants. Force spectroscopy by FM-AFM provides the dissipative component of the force in addition to the conservative one. With the help of a hydrodynamic model56 and an independent measurement of tip radius (Figure S1), the dissipative component yields the local, i.e., interfacial, viscosity coefficient as a function of tip−sample spacing. Small amounts of urea were found to enhance the interfacial viscosity dramatically to values that were 2−3-fold higher than bulk ones, indicating pronounced accumulation of urea near the silica surface. Such accumulation concurs with thermodynamic data and with the denaturing effect of urea on proteins. The interfacial viscosity in the presence of glycerol was found to follow the bulk values up to ∼1.5 M, indicating that glycerol concentration next to the silica surface was similar to that in the bulk. At higher concentrations, the surface viscosity dropped below the bulk values, disclosing glycerol depletion near the silica surface, in accord with the stabilizing capacity of glycerol. Beyond direct measurements of surface charge and medium viscosity our data disclose the interaction between silica surfaces as a function of spacing between them. These data are inaccessible by other methods.

sites, leading to a marked influence on their protonation/ deprotonation reaction constants. The latter mechanism has been studied extensively in the context of solvent effect on soluble acid and base reaction constants. It was found that deprotonation constants of certain acids can change by as much as 5 pH units upon transferring from water to alcohol. Interestingly, the effect on protonation of neutral bases (e.g., NH2 + H+ ⇄ NH3+) was much weaker.38 The effect of solvent on protonation/deprotonation constants has been analyzed successfully using Bjerrum’s theory39 and its modification by Fuoss,40 both predicting proportionality between the change in pK and the change in inverse dielectric constant, ΔpK ∝ Δ(ε−1). We are not aware of comparable analysis of the effect of osmolytes (solvent for this matter) on the charge of protic surfaces. The dielectric properties of the solvation layer surrounding proteins are difficult to measure, especially with proteins in their native state. Yet, the effect of osmolytes on the bulk dielectric constant (78.5 in pure water, ∼70 in 3.5 M (∼30% vol/vol) glycerol,41 and ∼90 in 5 M urea42) suggests that the interfacial dielectric properties43 and reaction constants of acidic and alkaline residues are also affected by added osmolytes. Protein structure is broadly governed by a balance between two opposing forces, hydrophobicity and net charge (often negative).44 While hydrophobicity promotes compaction, an increased net charge drives proteins, under physiological conditions, toward their unstructured state.45 Indeed, we find here that the protein destabilizer urea enhances negative surface charge dramatically, while the protein stabilizer glycerol neutralizes it. We thus identify a clear correlation between the stabilizing/destabilizing power of the two neutral osmolytes and their indirect effect on surface charge. Much of the current understanding of interfacial interactions in saline solutions came from studies of nonbiological surfaces. Fundamental notions, such as electrical double layer, Stern layer, and dispersion forces, were developed in the context of colloidal stability (e.g., the celebrated DLVO theory) and spread from there to biomolecular interactions. Inorganic surfaces offer certain advantages over biological ones. They are rigid, meaning the distance between them is well-defined, and they can easily be modified with desired functional groups, survive broad pH and temperature ranges, and allow for studies in ion concentrations that vary from none to several molars. Most important, inorganic surfaces lend themselves to direct force spectroscopy using surface force apparatus or atomic force microscopy, as well as sophisticated optical spectroscopy. In light of these advantages and the direct technological interest in such surfaces, the effect of salt and pH on the interaction between inorganic surfaces has been studied extensively46−49 in the past few decades. Interestingly, only few experiments explored the effect of different solvents,50−52 and we are not aware of any systematic study of the effect of natural osmolytes on the Coulomb interaction between two charged surfaces in saline solution. In attempt to understand the mechanisms underlying the entwined effect of osmolytes and salts on protein structure and function, we have set out to study the force acting between two silica surfaces in the presence of salt and two natural osmolytes, urea and glycerol. Silica, a weak acid, has been selected for this study since the pKa of its surface silanol groups is not far from that of aspartic and glutamic acids, two abundant amino acids.

■

MATERIALS AND METHODS

1. Solution Preparation. Anhydrous glycerol (Bio-Lab, Jerusalem, Israel), urea, NaCl, NaOH, and HCl (Sigma-Aldrich) were mixed with deionized water (18.2 MΩ·cm, Merck Millipore) to give the concentrations and pH values stated in the Results section. Solutions were left open and shaken overnight to equilibrate the carbonic acid concentration with atmospheric CO2, stabilizing the pH at 5.7. All experiments were conducted at 25 °C. 2. FM-AFM, Substrate, and Cantilever Preparation. The conservative and dissipative forces acting between an oxidized silicon cantilever and an oxidized silicon wafer were measured as a function of tip−sample separation utilizing a home-built, high-resolution AFM53 15014


Article

Journal of the American Chemical Society operating in frequency modulation mode. The setup displayed an 8.5 fm/ Hz optoelectronic noise floor and a 6 pm rms positioning noise in the full operation bandwidth (∼1 kHz). The cantilever was photothermally excited using a power modulated, 405 nm laser diode focused on its base. Cantilevers (ppp-NCHAuD, Nanosensors) had a nominally 42 N/ m spring constant and a measured ∼130 kHz resonance frequency in water. To enhance the signal, the tip radius was increased by sputtering a thin silicon layer (K575 Emitech, 50 mA, 10 μbar, 4 min). Silicon substrates (Si(100), p-type (B), 0.01 Ω·cm, University Wafer) and cantilevers were cleaned and oxidized by oxygen plasma (Diener, 100 W, 0.2 mbar, 1 min) prior to the experiment and immediately immersed in liquid. 2.1. Measurement of the Conservative Force. The cantilever was driven at its resonance frequency using a phase-locked loop (OC4, Nanonis) that maintained a 90° phase lag of the cantilever position relative to the driving force. The resonance frequency shift due to tip− substrate interaction was recorded and converted to the conservative component of the force using the Sader−Jarvis deconvolution method.57 The horizontal axes of figures reporting tip−surface distance denote the minimal approach distance of the oscillating tip. Zero distance in these figures corresponds to a ∼5 kHz frequency-shift threshold and might miss the real tip−surface distance by 1−2 Å. This margin was necessary in order to avoid tip crashing into the surface or potential damage to the surface. 2.2. Measurement of Dissipation Rate. The cantilever oscillation amplitude was kept constant at 0.2 nm (zero-to-peak) during the measurement, using a second feedback loop. The excitation force needed to maintain a constant amplitude was recorded and converted, together with the resonance frequency shift, to yield the measured dissipation power using formula A5 in ref 58,

Pdis(h) =

kA2 ω0(X + Ω(X − 1) − Ω2) 2Q 0

may arise from the substitution of the tip radius of curvature in place of a sphere radius. The extracted tip radius fits nicely radii measured by SEM imaging after the experiment (Figure S1). 2.4. Extraction of Surface Charge Density. The surface charge density, σ, is calculated by fitting the conservative force, F, divided by the tip radius of curvature, R, to the DLVO model.54 Beyond the hydration layer, the experimental force was found to decay exponentially with distance and was hence fitted with the Debye− Hückel (DH) solution to the Poisson−Boltzmann equation with identical surfaces. In this approximation, the DLVO normalized force takes a simple form:54 F(h) 4π 2 −κh H = σ e − 2 R ε0εr κ 6h with κ≡

6πηz(̇ t )R2 z(t )

■

RESULTS Effect of Urea on Surface Charge Density. Figure 1A depicts normalized force vs distance curves measured between an oxidized silica tip and an oxidized silicon wafer in the presence of different water−urea mixtures with 10 mM NaCl (pH ∼5.7). As seen, addition of small amounts of urea readily enhances the repulsive force (F > 0) between the two surfaces. The linearity of the force curves plotted in a semilogarithmic scale in the inset to Figure 1A discloses the expected exponential double layer repulsion with κ−1 ≈ 3 nm. Figure 1B depicts the corresponding growth of silica surface charge density, σ, with added urea, as extracted by fitting the curves of Figure 1A by eq 4. The quality of these fits can be appreciated by comparing the experimental and theoretical curves displayed in Figure 1A. Due to pronounced repulsion, the van der Waals (vdW) attraction is negligible in the fitted range. Figure 1B discloses that the neutral osmolyte urea enhances σ dramatically. The inset shows that the surface charge increases approximately linearly with the logarithm of urea concentration up to ∼2 M. The green circle in Figure 1B presents datum from the same set of measurements, in a solution titrated to pH 8.5 using 1 mM NaOH, with no urea. Addition of 1−2 M urea enhanced the surface charge density by ∼50%, an effect equivalent to increasing the pH by 3−4 units. The addition of urea hardly affected the extracted screening length, κ−1 (Figure S2). Effect of Glycerol on Surface Charge Density. Figure 2A depicts normalized force vs distance curves measured in the presence of different glycerol−water mixtures with 10 mM NaCl (pH ∼5.7). Opposite the case of urea, addition of glycerol suppresses the repulsion between the two surfaces, indicating surface charge neutralization by glycerol. A subtle concavity of the force curves at distances below ∼2.5 nm and high glycerol concentrations (>1.82 M) discloses the expected vdW attraction (eq 4).59 Hydration forces60 are dominant at distances below ∼2 nm, especially at high glycerol concentrations. As evident from the inset, the repulsive force decays exponentially with distance, and the overall force is well fitted by eq 4 with H = 2.9 ± 1.3 pN·nm and κ−1 ≈ 3 nm.

(1)

(2)

with η standing for the local shear viscosity coefficient and ż(t) marking the instantaneous tip velocity. The sphere radius, R, was replaced in the fitting procedure by the tip radius of curvature. The normalized average energy dissipation rate, Pdis/R2, was calculated by averaging the work done by the drag force over one oscillation of the cantilever, leading to

Pdis(h) R2

= 6πηω02((h + h0) + A −

2e 2c∞1000NA /ε0εr kBT

Here, κ is the inverse DH screening length. kB, T, NA, e, and ε0 are the Boltzmann constant, the temperature, the Avogadro number, the elementary charge, and the vacuum permittivity, respectively. c∞ is the bulk molar concentration of 1:1 salt, εr is the medium relative dielectric constant, and H is the Hamaker constant. σ, κ, and H were extracted by fitting the measured normalized force by eq 4.

Here, h is the minimal distance between the oscillating tip and the substrate, ω0 is the unperturbed cantilever resonance frequency, k is the cantilever spring constant, and Q0 is the cantilever quality factor, measured outside the tip−substrate interaction range. X  (Fex(h) − ex ex Fex 0 )/F0 , Ω  (ω(h) − ω0)/ω0, F0 is the amplitude of the excitation force applied to the cantilever outside the interaction range in order to maintain the desired amplitude A, and ω(h) is the cantilever resonance frequency at a distance h from the surface. 2.3. Extraction of Interfacial Viscosity Coefficient by Fitting a Hydrodynamic Model to the Measured Dissipation Rate. The hydrodynamic drag acting on a sphere moving perpendicularly to an infinite plane was calculated for a nonslip boundary condition by Brenner.56 In the limit where the sphere radius, R, is larger than its distance to the plane, z(t), the excess hydrodynamic force due to the presence of a plane is approximately given by

Fdis(z(t )) = −

(4)

(h + h0)2 + 2(h + h0)A ) (3)

The slip-length, h0, corrects Brenner’s result to the case of partial slip. To extract η(h), we first fit the dissipation rate measured in water by eq 3 with ηwater = 0.89 mPa·s. This fit gives R, which is used in later measurements for extracting η in the presence of osmolytes. This procedure eliminates potential geometric factors (of order unity) that 15015


Article


Figure 1. Urea enhances repulsion between two silica surfaces by increasing their surface charge density. (A) Force normalized by tip radius vs tip−substrate separation for different concentrations of urea, pH ∼5.7, and 10 mM NaCl. The curves present an average over 20 force curves measured in approach. Solid lines depict the best fit by eq 4. Inset: Same data plotted in a semilogarithmic scale. (B) Extracted surface charge density vs urea concentration. The green circle depicts the measured surface charge density at pH 8.5 of a 10 mM NaCl solution without urea. Inset: Same data plotted as a function of log(concentration). All data presented in this figure were collected using the same tip (R = 42 ± 10 nm) and substrate.

Figure 2. Glycerol reduces repulsion between two silica surfaces by decreasing their charge densities. (A) Force normalized by tip radius vs tip−surface separation for different concentrations of glycerol, pH ∼5.7, and 10 mM NaCl. The curves present an average over 20 force curves measured in approach. Solid lines depict the best fit by eq 4. Inset: Same data plotted in a semilogarithmic scale. (B) Extracted surface charge density vs glycerol concentration. The blue circle corresponds to surface charge density in a pH 4 solution of 10 mM NaCl without glycerol. Inset: Same data plotted as a function of log(concentration). All data presented in this figure were collected using the same tip (R = 54 ± 18 nm) and substrate.

The extracted surface charge density is plotted in Figure 2B vs glycerol concentration. It shows how the negative surface charge gradually diminishes with increasing glycerol concentration to about 40% of its initial value. As with urea, the inset shows that the surface charge decreases approximately linearly with the logarithm of glycerol concentration. The functional dependence upon glycerol concentration is markedly more gradual than in the case of urea. The blue circle in Figure 2B presents datum from the same set of measurements, in 10 mM NaCl, and pH 4 (0.1 mM HCl) solution, with no glycerol. The effect of 2−3 M glycerol is clearly larger than that of titration to pH 4. Altogether, we find that urea and glycerol, respectively, promote and suppress repulsion between similarly charged bodies. The effect on surface charge is qualitatively similar to direct change of the silanol’s protonation state by varying pH by more than 5 units. Interfacial Viscosity in the Presence of Urea. The measured normalized dissipated power, Pdis(h)/R2, is plotted in Figure 3A vs tip−surface separation for various aqueous urea solutions (pH ∼5.7 and 10 mM NaCl). Data were obtained in

two experiments (see caption). Data set II in Figure 3A and the data presented in Figure 1A were taken simultaneously and respectively depict the dissipative and the conservative components of the force acting on the tip. Dissipation was calculated with eq 1 using the experimental frequency shift, Ω(h), and the excess excitation force, X(h), measured simultaneously. The solid black lines display the best fit of the hydrodynamic model, eq 3, to the experimental data. The impressive fit by eq 3 means that the measured viscosity coefficient, η(h), exhibits a separation-independent value, which changes only with urea concentration. With R determined from the “0 M urea” curve, the local viscosity coefficient, η, was calculated and plotted in Figure 3B for three experiments. For a reference, the blue solid line depicts the bulk viscosity of the corresponding urea concentrations.61,62 The extracted surface viscosity at concentrations higher than 0.4 M saturates to 2.0−3.0 mPa·s, 2−3 times the corresponding bulk values. Such viscosity in bulk mixtures requires urea concentrations in the range of 11−13 M,61 suggesting substantial accumulation of urea near the silica surfaces, far beyond its bulk concentration. 15016


Article


Figure 3. Normalized dissipation rate and interfacial viscosity for different aqueous urea solutions, pH ∼5.7, and 10 mM NaCl. (A) Solid color lines show normalized dissipation rate vs tip−surface separation at different urea concentrations measured in two experiments (“I” and “II”). Solid black lines show best fit of eq 3 to the measured dissipation rate. Set I yielded R = 19 ± 10 nm. Set II was measured simultaneously with the force curves displayed in Figure 1A. Inset: Inverse normalized dissipation rate vs tip−surface separation fitted by eq 3. (B) Symbols represent interfacial viscosity coefficient, η, vs urea concentration extracted from three different experiments. The green, black, and red series were obtained using data sets I, II (A) and a third (dissipation not shown) experiment. Blue line depicts bulk viscosity coefficient vs urea concentration.61,62

Figure 4. Normalized dissipation rate and interfacial viscosity for different water−glycerol mixtures, pH 5.7 and 10 mM NaCl. (A) Solid color lines show measured normalized dissipation power vs tip− surface separation at different glycerol concentrations. Solid black lines show the best fit of eq 3 to the measured dissipation rate. Inset: Inverse normalized dissipation rate vs tip−surface separation, fitted by eq 3. Data collected using the same tip (R = 73 ± 9 nm) and substrate. (B) Symbols represent interfacial viscosity coefficient, η, vs glycerol concentration extracted from two separate experiments. Black circles correspond to the experiment displayed in A, and red circles correspond to a second experiment with R = 45 ± 10 nm. Blue line depicts bulk viscosity coefficient.63

coefficient is plotted vs glycerol concentration in Figure 4B for two experiments, together with bulk viscosity of water−glycerol mixtures (blue solid line).63 Initially, η increases with increasing glycerol content, in correspondence with bulk viscosity. However, at 1.5−2.0 M glycerol, where the surface charge depicted in Figure 2B saturates, the measured viscosity coefficient ceases to follow the bulk one, and lower values are measured. The lower η values relative to those in the bulk suggest that the concentration of glycerol near the surface is lower than its bulk concentration; that is, at high concentrations, glycerol is depleted near the silica surfaces. Such depletion, accompanied by preferential hydration of the surface, is consistent with the stabilizing effect of glycerol on proteins.64

The inverse normalized dissipation power reduces, for A ≪ h, to R2/Pdis(h) = (h + h0)/3πηω02A2

(5) −1

As expected (see inset to Figure 3A), Pdis (h) scales linearly with h for h ≪ R. The extracted slip length, h0 = −1.5 ± 0.3 nm, indicates partial liquid slippage on the surface at all urea concentrations (Figure S3). Interfacial Viscosity in the Presence of Glycerol. The measured normalized dissipated power, Pdis(h)/R2, is plotted in Figure 4A vs tip−surface separation for various glycerol−water mixtures, pH ∼5.7 and 10 mM NaCl. The solid black lines display the best fit of the hydrodynamic model, eq 3, to the experimental data. The inset displays inverse normalized dissipation power curves disclosing the linear dependence of Pdis−1(h) upon h, as predicted by eq 5 for A ≪ h ≪ R. All plots in Figure 4A are accurately described by eq 3, including the “0 M glycerol” one, from which R was determined and used to extract the quantity of interest, η. The interfacial viscosity

■

DISCUSSION Gouy−Chapman-Stern (GCS) Model for Silica. The GCS model46,55 is illustrated in Figure 5. The deprotonated silanols, SiO−, and the doubly protonated ones, SiOH2+, mark the surface plane denoted “0”. The outer Helmholtz plane, labeled “OHP”, marks the boundary between the diffuse and 15017


Article


Figure 5. Gouy−Chapman−Stern model of the silica−solution interface. Water continuously deprotonates and protonates neutral silanols at the “0” plane. The Stern layer and the diffuse layer are separated by the outer Helmholtz plane (“OHP”), on which solvated salt cations accumulate. AFM measures the sum of charges, σ, on the “0” and on the “OHP” planes. Red spheres represent oxygen; gray spheres, hydrogen; green spheres, monovalent cations; purple sphere, monovalent anion.

Stern layers. Its distance from the “0” plane is typically slightly larger than the radius of hydrated common ions, 0.33−0.45 nm.65 The Stern layer is modeled by a fixed capacitance, Cst, whose value varies with osmolyte concentration. The diffuse layer is modeled by solution to the Poisson−Boltzmann equation with the corresponding bulk dielectric constant, εb. The two equations for the electrostatic potentials, φ0 and φOHP, are supplemented with a set of chemical equilibrium equations connecting the surface chemical reactions with the two electrostatic potentials (SI; Gouy−Chapman−Stern model). Urea is known to bind protons and hence increases the solution pH.66 Glycerol, with its self-ionization constant being similar to that of water,67 was shown to increase absorption of CO2 from air and consequently lower the solution pH by carbonic acid accumulation.68 Within the experimental concentration range, urea and glycerol are predicted to modify the solution pH by no more than 0.25 and −0.4 units, respectively. It is difficult to measure the pH change since pH gauges, such as glass ones, are affected by the same phenomena we study here. Solution pH was thus estimated theoretically69 and is included in the GCS calculation of σ. Force spectroscopy by AFM measures the total surface charge up to the OHP, σ = e ({SiOH2+} − {SiO−}), where {..} denotes areal density (adsorbed cations, C+, do not show in the charge since they are represented by a neutral species, SiOC). The best fit of the GCS model to the surface charge density, measured in saline at different pH values, yielded 0.8 Fm−2 Stern capacitance and three reaction constants: silanol deprotonation (pKa = 4.3), double protonation (pKHH = −0.5), and cation binding (pKC = 1.3). Since urea and glycerol are neutral, their effect on σ is indirect. They affect it Coulombically by modifying the bulk and Stern dielectric constants (Figure 5; εb and εi) and nonCoulombically by varying the immediate environment of silanols and, hence, the three reaction constants mentioned above. The computed contributions of screening length and Stern capacitance modification by added osmolytes are compared in Figure 6A with the measured surface charge at

Figure 6. (A) Measured and calculated surface charge density in solutions of varying osmolyte concentration or pH. Urea (red circles, pH 5.7) and glycerol (black circles, pH 5.7) enhance and reduce σ, respectively (Figures 1 and 2). The dashed lines with matching colors to the circles depict the effect of osmolytes on the bulk dielectric constant, and hence on σ, as calculated by the GCS model for pH 5.7. The solid lines add the calculated effect of the Stern layer to the bulk dielectric constant. Green and black squares depict the experimental surface charge at pH 8.5 with no osmolytes and the corresponding prediction of the GCS model, respectively. The blue and black diamonds compare the same quantities for pH 4. The turquoise dashdotted curve depicts the theoretical prediction for the hypothesized 12 M interfacial density of urea. The red and the black dotted lines correspond to the GCS model (employing the Stern layer and bulk dielectric effects) with estimated osmolyte-induced modifications of both silanol deprotonation (pKa) and cation binding (pKC) constants by urea and by glycerol, respectively. (B) Estimated silanol deprotonation (pKa) and cation binding (pKC) constants vs osmolyte concentration.

different osmolyte concentrations. The surface charge at three pH values, 4, 5.7, and 8.5, and the corresponding prediction of the GCS model are added for reference. The effect of osmolytes on bulk screening is straightforward. Addition of glycerol reduces the dielectric constant to εb = 71.3 at 3.1 M.41 Conversely, addition of urea increases the bulk dielectric constant, to εb = 85.0 at 2.4 M.42 Substituting these values into the GCS model we find that the modulation of εb by urea and glycerol affects the surface charge in the right direction but the total change is small, 3.7 out of 21 mCm−2 (black and red dashed lines in Figure 6A). Two unknowns complicate the analysis of Stern capacitance modulation by added osmolytes. First, the dielectric constant of 15018


Article


denaturation by urea at similar concentrations.4 With η used as a measure for urea accumulation near the silica surface, Figure 3B disclosed surface concentrations in the range of ∼12 M for bulk concentrations of ≥0.4 M.61,62 Conversely, the surface viscosity for 2.5 M glycerol matched the bulk value at 0.5 M, which is only slightly higher than that of pure water. The deduced accumulation of urea and depletion of glycerol near the silica surface are also consistent with excess enthalpy of dilution data.71 The negative value for urea indicates unfavorable urea−water interaction compared with water− water. Urea still dissolves sparingly in water due to excess entropy, but enthalpy prefers phase separation. This trend is opposite that with glycerol. The excess enthalpy of dilution is positive in this case, indicating that enthalpy acts in concert with entropy to keep a mixed phase. In the presence of silica, urea apparently tends to separate from water and accumulate near the surface, while glycerol prefers an aqueous environment. The same conclusions can be drawn from measurements of water activity in the presence of these osmolytes.72 The changes in σ and η are clearly correlated. The steep rise in surface viscosity with urea below 0.4 M (Figure 3B) overlaps with the steep enhancement of σ disclosed by Figure 1B. Similarly, the growth of η with glycerol concentration up to ∼1.5 M correlates with the gradual suppression of σ (Figure 2B). Moreover, the reduced η above 1.5 M glycerol is accompanied by the saturation of σ. As seen in the previous subsection, mere variation of the dielectric constants is too small to account for the full variation of σ. We were therefore led to conclude that the main effect was incurred by modulation of the surface reaction constants by accumulated urea and depleted glycerol. Beyond the GCS Model. The GCS model with two or three pKs is widely used for calculating the charge of protic surfaces in contact with pure saline solutions. In this treatment, the chemical components of the reaction constants are kept fixed and the degree of ionization is calculated self-consistently by solving for the local electrostatic potential. In the first subsection of the Discussion we computed the effect of dielectric constant modification by added osmolytes and found it to be small compared with the experimentally measured variation of σ. In the subsection that followed, we analyzed the surface viscosity and found that urea accumulated near silica while glycerol depleted at the interface. These findings reveal that in the balance between full hydration and interaction with silica, urea prefers the surface, while glycerol prefers bulk water. It takes a small step from there to conclude that urea competes with protons for surface sites better than water, hence shifting the effective deprotonation constant to lower values. Stronger binding of urea to charged oxygen (carboxylate) than to neutral oxygen (hydroxyl) was previously reported,4,24 suggesting that urea competes for deprotonated silanols, the target of protons. The resulting shift in pK value for silanol deprotonation is consistent with the enhanced surface charge. Using the GCS model we found that the change in σ is faithfully reproduced if the modification of bulk and surface dielectric constants is supplemented with a urea-induced decrease in silanol deprotonation and in cation binding constants (Figure 6A, red dotted line). The extracted modifications of pKa and pKC are plotted by a red line in Figure 6B vs urea concentration. In their extraction, we assumed that the relative change in pK is similar for protons and cations.39,40 The overall change in both pK values is 30% or less.

interfacial water, even in the absence of osmolytes, is markedly different from its bulk value and notoriously difficult to estimate. Second, the interfacial concentration of urea or glycerol is unknown, let alone their effect on the Stern capacitance. For the assumed layer thickness, 0.5 nm, the Stern capacitance (0.8 Fm−2) gave in saline an interfacial dielectric constant of εW i ≈ 45. Its reduced value compared with bulk water, εW b ≈ 78.5, indicates a restricted orientational freedom of interfacial waters due to their bonding to the silica surface and to adsorbed ions. Interestingly, a similar value for εW i was found to explain measured pKa changes at the surface of the acidic protein RNase Sa.70 To estimate the effect of εi on σ, we ε therefore made the plausible assumption εi = εiW ε Wb . As seen b

in Figure 6A, inclusion of the effect of εi on σ enhances the total variation in surface charge to 5 mCm−2 (black and red solid lines), still small compared with the total change, 21 mCm−2. As discussed in the context of Figure 3, the enhanced surface viscosity is consistent with a 12 M urea concentration near the surface. For the dielectric constant corresponding to such a concentration,42 εb = 102, the GCS model gives the dashdotted turquoise line of Figure 6A. This is better, though still in insufficient agreement with the measured σ. Analyzed differently, to account for the experimental σ in the presence of 3.1 M glycerol or 2.4 M urea, εi had to be set at 10 and 125, respectively. The value in the presence of glycerol is unrealistically small, especially as viscosity measurements show no accumulation of glycerol near the surface (Figure 4B). Similarly, the value needed for explaining the results in the presence of urea was unrealistically high. Altogether, the GCS analysis led to the following conclusions: (a) As expected, higher dielectric constants promote silanol ionization, while lower ones suppress it. (b) The effect of osmolytes on surface charge due to modulation of the bulk dielectric constant is small compared with the experimental change. (c) Adding the effect of osmolytes on Stern capacitance increases the theoretical change in σ to 5 mCm−2, but the overall effect is still too small to account for the measured variation of σ. (d) The effect of osmolytes on surface charge must therefore include additional ingredients beyond variation of the dielectric constant. Accumulation and Depletion of Osmolytes near the Silica Surface. Figures 1B, 2B, 3B, and 4B disclose substantially different behavior for glycerol−water and for urea−water solutions. In the case of urea, concentrations as low as 50 mM sufficed to induce a sizable increase in σ and η, while for glycerol, no change was discerned at such concentrations. Furthermore, while added urea enhanced surface viscosity to more than twice its bulk value, the surface viscosity with glycerol followed the bulk value up to ∼1.5 M concentration, above which it dropped below the bulk value (Figure 4B). Urea was hence found to accumulate near the silica surface already at low concentrations, while glycerol seemed to maintain its bulk concentration near the surface, up to 1.5 M, and then deplete at higher concentrations. This latter observation concurs with Gekko and Timasheff,64 who observed glycerol effects only at high (1−4 M) glycerol concentrations. The accumulation of urea and the depletion of glycerol are consistent with the general findings with proteins. Stabilizing osmolytes, such as glycerol, are depleted at the protein surface, while the destabilizing osmolyte urea accumulates there. The onset of urea accumulation at concentrations as low as 50 mM concurs with previous studies showing protein 15019


Article

Journal of the American Chemical Society Thermodynamic and bulk measurements found that urea also interacts favorably with Cl− and unfavorably with Na+.24 Significant accumulation of urea near the surface is hence predicted to encourage transfer of Na+ ions from the Helmholtz plane (Figure 5) to the bulk or transfer of Cl− ions in the opposite direction. While only modulation of Na+ binding is included in the GCS model (Figure 6B, pKC), both types of displacement result in enhanced negative surface charge and may account, in part, for the dramatic change in σ. The situation with glycerol is opposite that with urea, indicating that the two reaction constants are reduced in the presence of glycerol. Indeed, a 25% increase in pKa and pKC on top of the dielectric constant’s modification by added glycerol reduces the calculated surface charge to its measured value (Figure 6A, black dotted line). The extracted pK values vs glycerol concentration are depicted by a black line in Figure 6B.

effect against the adverse effect of salt is well established in the literature.1,12,20,64,76

CONCLUSION Simultaneous measurements of the conservative and dissipative components of the force acting between two silica surfaces in the presence of different concentrations of urea and glycerol shed new light on the way by which neutral biological molecules control the charge of protic surfaces. The extracted interfacial viscosity reveals extensive accumulation of urea near the silica surface, already at bulk concentrations as low as 50 mM. The conservative component of the force shows that this accumulation leads to simultaneous enhancement of surface charge by more than 50%, largely due to a shift of the silanol deprotonation, pKa, and cation binding, pKC, to lower values. In glycerol on the other hand, these binding constants increase and the negative surface charge density is neutralized by adsorption of protons and monovalent cations. As a result, the intersurface repulsion is diminished. Assuming the aforementioned shift in pKa takes place also with acidic residues of proteins, which are negatively charged under physiological pH, and considering it is the balance between net protein charge and hydrophobicity that governs protein stability, we hypothesize that regulation of surface charge by osmolytes contributes to osmolyte-induced protein stabilization/destabilization. The data suggest that urea may destabilize proteins by increasing their negative charge, while glycerol may stabilize them by neutralization.45 This conjecture is supported by the fact that the effect of urea and polyol osmolytes on protein stability is maximal31,33,34,73 when the pH level is set close to pKa74 of carboxylic residues, namely, when small changes in pKa affect the charge most efficiently. The revealed modulation of pKa by osmolytes is linked this way to osmolyte function. The destabilizing/stabilizing effect of urea/ glycerol takes here an unexpected turn, where non-Coulombic accumulation/depletion of neutral osmolytes near protic surfaces leads to surface charging/discharging and modulation of intersurface force. Screening by added sodium and potassium chloride promotes dissociation of cation−anion pairs and protonated silanols.59 In enzymes utilizing charged substrates, charge screening often suppress activity by disrupting substrate binding to active enzymatic sites. This nonspecific damage (with regard to Na+ and K+) manifests itself equally in isozymes from different organisms, typically below 0.5 M salt.75 The data in Figure 2B show that cation−anion dissociation is recovered by addition of glycerol, indicating that the latter osmolyte may act as an antidote to deleterious effects of monovalent salts when those stem from charge screening. Such a protective

ORCID

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b07036. SEM tip image, extracted screening length and slip length data, and a detailed description of the Gouy−Chapman− Stern surface model. (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected]

■

Roy Govrin: 0000-0002-4112-9340 Itai Schlesinger: 0000-0003-4039-7885 Uri Sivan: 0000-0002-9807-2656 Author Contributions ‡

R. Govrin and I. Schlesinger contributed equally.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was supported by the Israeli Science Foundation through grant number 10/1051 and the Single Molecule ICore Center of Excellence, grant number 1902/12. We are grateful to Prof. Daniel Harries for several useful discussions.

■

REFERENCES

(1) Yancey, P. H.; Clark, M. E.; Hand, S. C.; Bowlus, R. D.; Somero, G. N. Science 1982, 217, 1214−1222. (2) Hochachka, P. W.; Somero, G. N. Biochemical Adaptation: Mechanism and Process in Physiological Evolution; Oxford University Press: New York, 2002. (3) Auton, M.; Rösgen, J.; Sinev, M.; Holthauzen, L. M. F.; Bolen, D. W. Biophys. Chem. 2011, 159, 90−99. (4) Heyda, J.; Kožíšek, M.; Bednárova, L.; Thompson, G.; Konvalinka, J.; Vondrásě k, J.; Jungwirth, P. J. Phys. Chem. B 2011, 115, 8910−8924. (5) Harries, D.; Rösgen, J. Methods Cell Biol. 2008, 84, 679−735. (6) Schneck, E.; Horinek, D.; Netz, R. R. J. Phys. Chem. B 2013, 117, 8310−8321. (7) Garcia-Manyes, S.; Dougan, L.; Fernandez, J. M. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 10540−10545. (8) Politi, R.; Harries, D. Chem. Commun. (Cambridge, U. K.) 2010, 46, 6449−6451. (9) Markarian, M. Z.; Schlenoff, J. B. J. Phys. Chem. B 2010, 114, 10620−10627. (10) Nordstrom, L. J.; Clark, C. A.; Andersen, B.; Champlin, S. M.; Schwinefus, J. J. Biochemistry 2006, 45, 9604−9614. (11) Buche, A.; Colson, P.; Houssier, C. J. Biomol. Struct. Dyn. 1993, 11, 95−119. (12) Pollard, A.; Wyn Jones, R. G. Planta 1979, 144, 291−298. (13) Costanzo, J. P. J. Exp. Biol. 2005, 208, 4079−4089. (14) Kukal, O.; Serianni, A. S.; Duman, J. G. J. Comp. Physiol., B 1988, 158, 175−183. (15) Sakamoto, A.; Murata, N. Plant, Cell Environ. 2002, 25, 163− 171. (16) Dashnau, J. L.; Nucci, N. V.; Sharp, K. A.; Vanderkooi, J. M. J. Phys. Chem. B 2006, 110, 13670−13677. (17) Voelkel, S. A.; Hu, Y. X. Theriogenology 1992, 37, 687−697. 15020


Article

Journal of the American Chemical Society (18) Asenjo, J. A.; Andrews, B. A. J. Chromatogr. A 2011, 1218, 8826−8835. (19) Angell, M.; Pan, C.-J.; Rong, Y.; Yuan, C.; Lin, M.-C.; Hwang, B.-J.; Dai, H. Proc. Natl. Acad. Sci. U. S. A. 2017, 114, 834−839. (20) Munns, R.; Tester, M. Annu. Rev. Plant Biol. 2008, 59, 651−681. (21) Zhang, Y.; Huang, Y.; Li, S. AAPS PharmSciTech 2014, 15, 862− 871. (22) Cheah, W.-K.; Sim, Y.-L.; Yeoh, F.-Y. Mater. Chem. Phys. 2016, 175, 151−157. (23) Horinek, D.; Netz, R. R. J. Phys. Chem. A 2011, 115, 6125− 6136. (24) Guinn, E. J.; Pegram, L. M.; Capp, M. W.; Pollock, M. N.; Record, M. T. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 16932−16937. (25) Gilman-Politi, R.; Harries, D. J. Chem. Theory Comput. 2011, 7, 3816−3828. (26) Knowles, D. B.; Shkel, I. A.; Phan, N. M.; Sternke, M.; Lingeman, E.; Cheng, X.; Cheng, L.; O’Connor, K.; Record, M. T. Biochemistry 2015, 54, 3528−3542. (27) Ben-Naim, A. Molecular Theory of Solutions; Oxford University Press: Oxford, 2006. (28) Street, T. O.; Bolen, D. W.; Rose, G. D. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 13997−14002. (29) Roberts, M. F. Saline Syst. 2005, 1, 5. (30) Nevoigt, E.; Stahl, U. FEMS Microbiol. Rev. 1997, 21, 231−241. (31) Pace, C. N.; Laurents, D. V.; Thomson, J. A. Biochemistry 1990, 29, 2564−2572. (32) Xie, G.; Timasheff, S. N. Protein Sci. 1997, 6, 211−221. (33) Haque, I.; Singh, R.; Moosavi-Movahedi, A. A.; Ahmad, F. Biophys. Chem. 2005, 117, 1−12. (34) Hand, S. C.; Somero, G. N. J. Biol. Chem. 1982, 257, 734−741. (35) Bartlett, G. J.; Porter, C. T.; Borkakoti, N.; Thornton, J. M. J. Mol. Biol. 2002, 324, 105−121. (36) Jungwirth, P.; Winter, B. Annu. Rev. Phys. Chem. 2008, 59, 343− 366. (37) O’Brien, E. P.; Brooks, B. R.; Thirumalai, D. J. Am. Chem. Soc. 2012, 134, 979−987. (38) Cox, B. G. Acids and Bases, Solvent Effects on Acid-Base Strength; Oxford University Press: Oxford, 2013. (39) Bjerrum, N. Kgl. Danske Vidensk. Selsk., Mater.-fys. Medd. 1926, 7, 1. (40) Fuoss, R. M. J. Am. Chem. Soc. 1958, 80, 5059−5061. (41) Physical Properties of Glycerine and its Solutions; Glycerine Producers’ Association: New York, 1963. (42) Wyman, J. J. Am. Chem. Soc. 1933, 55, 4116−4121. (43) Bonthuis, D. J.; Gekle, S.; Netz, R. R. Phys. Rev. Lett. 2011, 107, 166102. (44) Requião, R. D.; Fernandes, L.; de Souza, H. J. A.; Rossetto, S.; Domitrovic, T.; Palhano, F. L. PLoS Comput. Biol. 2017, 13, e1005549. (45) Uversky, V. N.; Gillespie, J. R.; Fink, A. L. Proteins: Struct., Funct., Genet. 2000, 41, 415−427. (46) Zhao, C.; Ebeling, D.; Siretanu, I.; van den Ende, D.; Mugele, F. Nanoscale 2015, 7, 16298−16311. (47) Morag, J.; Dishon, M.; Sivan, U. Langmuir 2013, 29, 6317− 6322. (48) Zohar, O.; Leizerson, I.; Sivan, U. Phys. Rev. Lett. 2006, 96, 177802. (49) Wang, L.; Siretanu, I.; Duits, M. H. G.; Stuart, M. A. C.; Mugele, F. Colloids Surf., A 2016, 494, 30−38. (50) Mcguiggant, P. M.; Pashley, R. M. J. Phys. Chem. 1988, 92, 1235−1239. (51) Kanda, Y.; Nakamura, T.; Higashitani, K. Colloids Surf., A 1998, 139, 55−62. (52) Christenson, H. K.; Horn, R. G. J. Colloid Interface Sci. 1985, 103, 50−55. (53) Schlesinger, I.; Kuchuk, K.; Sivan, U. Rev. Sci. Instrum. 2015, 86, 83705. (54) Israelachvili, J. Intermolecular and Surface Forces, 3rd ed.; Academic Press: Burlington, MA, 2011.

(55) Brown, M. A.; Goel, A.; Abbas, Z. Angew. Chem., Int. Ed. 2016, 55, 3790−3794. (56) Brenner, H. Chem. Eng. Sci. 1961, 16, 242−251. (57) Sader, J. E.; Jarvis, S. P. Appl. Phys. Lett. 2004, 84, 1801. (58) Schlesinger, I.; Sivan, U. Langmuir 2017, 33, 2485−2496. (59) Dishon, M.; Zohar, O.; Sivan, U. Langmuir 2009, 25, 2831− 2836. (60) Kilpatrick, J. I.; Loh, S.-H.; Jarvis, S. P. J. Am. Chem. Soc. 2013, 135, 2628−2634. (61) Halonen, S.; Kangas, T.; Haataja, M.; Lassi, U. Emiss. Control Sci. Technol. 2017, 3, 161−170. (62) Kawahara, K.; Tanford, C. J. Biol. Chem. 1966, 241, 3228−3232. (63) Cheng, N.-S. Ind. Eng. Chem. Res. 2008, 47, 3285−3288. (64) Gekko, K.; Timasheff, S. N. Biochemistry 1981, 20, 4667−4676. (65) Nightingale, E. R. J. Phys. Chem. 1959, 63, 1381−1387. (66) Wen, N.; Brooker, M. H. J. Phys. Chem. 1993, 97, 8608−8616. (67) Woolley, E. M.; Tomkins, J.; Hepler, L. G. J. Solution Chem. 1972, 1, 341−351. (68) Aschenbrenner, O.; Styring, P. Energy Environ. Sci. 2010, 3, 1106. (69) Akers, D.; Goldberg, R. N. Math. J. 2001, 8. (70) Pace, C. N.; Grimsley, G. R.; Scholtz, J. M. J. Biol. Chem. 2009, 284, 13285−13289. (71) Savage, J. J.; Wood, R. H. J. Solution Chem. 1976, 5, 733−750. (72) Miyawaki, O.; Saito, A.; Matsuo, T.; Nakamura, K. Biosci., Biotechnol., Biochem. 1997, 61, 466−469. (73) Haque, I.; Islam, A.; Singh, R.; Moosavi-Movahedi, A. A.; Ahmad, F. Biophys. Chem. 2006, 119, 224−233. (74) Grimsley, G. R.; Scholtz, J. M.; Pace, C. N. Protein Sci. 2009, 18, 247−251. (75) Greenway, H.; Osmond, C. B.Plant Physiol. 1972, 49, 256− 25910.1104/pp.49.2.256. (76) Incharoensakdi, A.; Takabe, T.; Akazawa, T. Plant Physiol. 1986, 81, 1044−1049.

15021


Regulation of Surface Charge by Biological Osmolytes - Journal of the

Recommend Documents