Broadband Depolarized Light Scattering Study of Diluted Protein Aqueous Solutions

Stefania Perticaroli,† Lucia Comez,‡,§ Marco Paolantoni,*,† Paola Sassi,† Laura Lupi,§

Daniele Fioretto,§ Alessandro Paciaroni,§ and Assunta Morresi†

Dipartimento di Chimica, UniVersita degli Studi di Perugia, Via Elce di Sotto 8, IOM-CNR, c/o Dipartimentodi Fisica, UniVersita degli Studi di Perugia, Via A. Pascoli 1, and Dipartimento di Fisica, UniVersita degliStudi di Perugia, Via A. Pascoli 1, I-06123 Perugia, Italy

ReceiVed: March 2, 2010; ReVised Manuscript ReceiVed: May 4, 2010

A broadband depolarized light scattering (DLS) study is performed on diluted lysozyme aqueous solutions asa function of temperature and concentration. The dynamical susceptibility, obtained in a wide spectral range(0.6-36000 GHz) through the coupled use of interferometric and dispersive devices, is interpreted and comparedwith neutron scattering and Raman-induced optical Kerr-effect literature data, thus giving a general pictureof relaxation phenomena. We show that the proposed approach represents a suitable tool for investigating thehydration dynamics of protein-water solutions. A detailed analysis of the quasi-elastic scattering regionevidences the existence of two distinct relaxational processes at picosecond time scales. The fast process(fractions of picosecond) is attributed to bulk water dynamics, while the slow one (few picoseconds) is attributedto dynamical rearrangements of water molecules strongly influenced by the protein (hydration water). Theretardation effect here estimated of about 6-7 can be regarded as a direct measure of the increasedprotein-water and water-water hydrogen bond stability of the water molecules within the protein hydrationshell. Interestingly, a similar effect was previously observed on small hydrophilic sugar molecules. Moreover,backbone and side chains torsional motions of the protein in the 600-5300 GHz frequency range are foundto be insensitive to thermal variations and to eventual changes occurring in the premelting zone.

1. Introduction

Dynamical features of water molecules localized in thehydration layer of a protein heavily affect its functionality; thisis the reason why various biological processes, such as enzymeactivity and protein denaturation, strongly depend on theprotein-water interface.1-7 The internal dynamics of the proteinitself is strongly influenced by the interfacial water.8 Recentneutron scattering experiments revealed that the local diffusionof side chains in hydrated powders is activated by one waterlayer and that increasing the hydration level up to dilutedsolutions enhances rate and amplitude of the picosecondmotions.9,10 Thus, a deep understanding of the protein-waterinterface requires knowledge of the temporal scales characteristicof both local protein rearrangements and water dynamics withinthe hydration layer.

From an experimental point of view obtaining informationon dynamical properties of hydration water is not an easy task,especially in the case of diluted systems. In fact, (i) the dynamicsin protein-water solutions, involving solute and solvent relax-ation processes, usually cover large frequency ranges that cannotbe explored by a single technique. Moreover, it may bechallenging to discern between protein and water contributions11

also considering that protein relaxation processes involve widetemporal scales9,10,12 and that the hydration may activate someparticular protein dynamics.9,10 Eventual dynamical cross cor-relations should be also taken into account.12 (ii) The hydrationshell response cannot be easily distinguished from the bulk oneif the characteristic time scale of the probing technique is not

fast enough compared to the residence time of the watermolecules within the shell (slow exchange condition). (iii) Thehydration shell itself is heterogeneous at a molecular scale eitherdue to the topological disorder of the protein surface or to thechemical heterogeneity of the solvent-exposed groups1,13-15

leading to a distribution of typical relaxation times.4,6,16 (iv)Finally, the fact that different experimental techniques probedifferent physical quantities further complicates the overallcomparison and interpretation of the results.11,12,17

Experiments and molecular dynamics (MD) simulations haverevealed that the dynamics of interfacial water is retarded withrespect to the bulk, yet the magnitude of this slowing down isstill questioned.1-4,6,11,12,15,17-22 According to NMR experiments,4,6

the water rotation near a biomolecular surface is 2-3-fold slowerthan in the bulk in agreement with new interpretations ofdielectric experiments.12,22 Similar perturbation effects were alsodetected for small solutes.6,23 By contrast, time-resolved fluo-rescence spectroscopy2,17,19-21 and other dielectric studies24

observe slow relaxation contributions attributed to surface waterup to orders of magnitude slowed down with respect to the bulk.

Since the motion of water molecules is intimately connectedto the process of hydrogen bond (H-bond) forming and breaking,a deep knowledge of the H-bond kinetics in these complexsystems is essential to elucidate overall dynamical character-istics. In this respect, while several MD simulation studies havepointed out the role played by water-water and protein-waterH-bond relaxation in the interface,1,3,7,15,18,25-29 it is verychallenging to achieve direct experimental estimates of theseprocesses in the hydration shell.30

Depolarized light scattering (DLS) and its time-domain coun-terpart, the time-resolved optical Kerr effect (OKE), are powerfultechniques for the study of complex aqueous systems, sensitive todynamical events of both solute and solvent in temporal scales

* Corresponding author. E-mail: marcopa@unipg.it.† Dipartimento di Chimica.‡ IOM-CNR, c/o Dipartimento di Fisica.§ Dipartimento di Fisica.

J. Phys. Chem. B 2010, 114, 8262–82698262

10.1021/jp101896f 2010 American Chemical SocietyPublished on Web 05/28/2010

ranging from fractions to hundreds of picoseconds.31-37 Addition-ally, the dominant relaxation component of the DLS spectrumin pure water is closely related to the fast (picoseconds) H-bondrearrangement dynamics as computed in MD studies38-40 andmay be referred to as H-bond lifetime.34,37 Thus, in principle,these experiments represent ideal tools to directly probe H-bondkinetics in biorelevant systems.

Recent efforts have been devoted to the study of protein-waterinteractions in solid samples at low water content by using DLSexperiments41 also in connection with neutron scattering anddielectric investigations.41,42 However, the properties of theinterfacial water in these hydrated powders are expected to bedifferent from those measured at physiological conditions, owingto the strong confinement and the absence of multiple hydrationlayers and the bulk component.4,6,43 Diluted aqueous systemshave been the object of very recent DLS and OKE studies31-33,44,45

indicating that relevant insights can be extracted from theterahertz spectral region of the Raman spectrum. Nevertheless,the interpretation of this frequency range is still incomplete,especially in the very low frequency region (<300 GHz) whereliterature data are lacking. Therefore, high-resolution DLSstudies should be helpful to draw a complete picture on thepicosecond dynamics of both protein and water in solutionconditions.

Recently, we proved the uniqueness of the combined use ofinterferometric and dispersive DLS devices for the dynamicalcharacterization of aqueous solutions of carbohydrates.35-37 Inthis way we evidenced the presence of two distinct relaxationprocesses at picosecond time scales assigned to hydrating andbulk water and a relative retardation factor of about 5-6.37 Inthe present work we apply this experimental approach to thestudy of diluted solutions of the model protein lysozyme(25-150 mg protein/mL water) covering the wide frequencyrange going from 0.6 to 36000 GHz. In particular, the 100 mg/mL solution has been deeply investigated as a function oftemperature between 25 and 70 °C. The highest temperaturesconsidered relate to the premelting region (50-70 °C) oflysozyme in which changes of the tertiary structure weresuggested.44

Adding the protein to water leads to the appearance of newvibrational and relaxational spectral features. The comparisonwith Raman-induced Kerr effect spectroscopy (OHD-RIKES)literature data32 and neutron scattering (INS) profiles is usedfor interpretation. In this way a consistent picture of the wholespectral distribution has been derived and the dynamicalproperties of hydration water have been estimated.

2. Experimental Technique

Lysozyme from chicken egg white lyophilized powder waspurchased from Sigma-Aldrich and used without further puri-fication. The protein was dissolved in doubly distilled anddeionized water to prepare solutions in the 25-150 mg/mLconcentration range (ca. 1.7-9.1 mM). For a different samplea pH of ∼4 was measured with a FiveEasy pH meter FE20Mettler Toledo. All solutions were conditioned at 40 °C for ca.1 h to achieve a complete dissolution maintaining the nativestructure of the protein that denatures at higher temperatures.Dust-free solutions were obtained by filtering the samples withMillipore filters of 0.20 µm diameter pores. Then, the sampleswere placed in a 10 mm path optical quartz cell to be analyzedat room temperature. Measurements were also performed as afunction of temperature in the 25-70 °C range for pure waterand for a selected concentration (100 mg/mL).

DLS spectra were acquired as described before.35-37 Inparticular, a Coherent-Innova 90 Ar+ laser operating on a single

mode at λ0 ) 514.5 nm line (typical power of 300 mW) wasemployed as a light source. In order to obtain the wide spectralrange from 0.6 to 36000 GHz, horizontally polarized light wasanalyzed by means of two different spectrometers. The low-frequency region from ∼0.6 to 140 GHz was explored by aSandercock-type (3 + 3)-pass tandem Fabry-Perot interfer-ometer, characterized by a finesse of about 100 and a contrastratio greater than 5 × 1010. To cover the whole frequency rangethree different mirror separations (d ) 13-14 mm, d ) 4-3.5mm and d ) 1 mm), depending on temperature and concentra-tion, were used. The cell was placed into a thermo-controlledcopper holder and the temperature was settled by a conventionaldc power supply, keeping temperature fluctuations within 0.1°C during the measurements. An ISA Jobin-Yvon model U1000double monochromator having 1 m focal length with holo-graphic gratings was used to record the spectra in the 30-36000GHz frequency range. The scattered light was detected using athermoelectrically cooled (-30 °C) Hamamatsu model 943XXphotomultiplier that is computer-controlled by the ISA Jobin-Yvon SpectraMax package. The spectra were measured between30 and 1800 GHz with a resolution of 15 GHz and from 120 to36000 GHz with a resolution of 30 GHz. The temperature wascontrolled by circulating water from a Haake F6 ultrathermostatwith a precision of 0.1 °C. Some parts of the same solutionwere scrupulously subjected to measurements of differentlengths of time to control the reproducibility of experimentaldata at all temperatures. Raman spectra of the hydrated powderof lysozyme (h ) 0.33; h ) grams of water/grams of dryprotein) were recorded with the dispersive apparatus at roomtemperature between 240 and 15000 GHz (30 GHz resolution).

After subtraction of the dark count contribution, low and highfrequency spectra were spliced, exploiting an overlap of abouthalf a decade in frequency. Moreover, the imaginary part ofthe dynamic susceptibility was calculated according to therelation �DLS′′ (ν) ∝ I(ν)/[nB(ν) + 1], where I(ν) is the measuredintensity and nB(ν) ) 1/[exp(hν/kT) - 1] is the Bose-Einsteinoccupation number.46

3. Results

3.1. DLS Spectra. �DLS′′ (ν) spectra of pure water andlysozyme aqueous solutions obtained in the frequency rangegoing from 0.6 to 36000 GHz at 25 °C are depicted in Figure1. All the spectra were normalized to the intensity of the10500-36000 GHz distribution, which is ascribed to hinderedrotations (librations) of water molecules.34,46,47 This band wasselected owing to its poor dependence on temperature andconcentration changes. The spectral profile of pure water exhibitstwo main peaks at 5100 and 1500 GHz assigned to the H-bond(O · · ·O unit) stretching and H-bond (O · · ·O · · ·O unit) bendingvibrations, respectively. These intermolecular modes correspondto restricted translations of water molecules in a parallel orperpendicular direction with respect to the H-bond.34,46 At lowerfrequencies (ν < 500 GHz) the water spectrum shows apicosecond relaxation process that may be related to the fastH-bond rearrangement dynamics (i.e., H-bond lifetime).34,36-40,48-50

The spectra of Figure 1 evidence two gradual intensity increaseson increasing the protein content: one in the low-frequencyregion of the spectrum between 0.6 and 100 GHz and anotherin the 600-5300 GHz range. These additional features mainlyaccount for the direct contribution of the protein to the signal.

3.2. Comparison with OHD-RIKES Spectra. To rationalizethe intensity enhancement localized between 600 and 5300 GHz,reference is made to the treatment executed by Giraud et al.32

We then proceeded with the removal of the solvent contribution

DLS of Protein Aqueous Solutions J. Phys. Chem. B, Vol. 114, No. 24, 2010 8263

subtracting the water spectrum after rescaling on the librationalregion. The resulting profile (Figures 2a and 2b), hereafter calledsolvent free (SF) spectrum, compares to the equivalent distribu-tion obtained by OHD-RIKES experiments which was reliabledown to ∼270 GHz.32 We remark that the SF distribution, apartfrom protein contributions, may also contain solute-inducedsolvent changes. The analysis in terms of the multiple Brownianoscillators performed between 270 and 5300 GHz (Figure 2)supplies parameters similar to those reported in ref 32 (Table1),suggesting that the two approaches give consistent results. Thesame procedure has been applied for all the considered tem-peratures, and it is found that above 270 GHz the SF distributionscarcely depends on thermal variations. In fact very similarfitting parameters have been obtained between 25 and 70 °C,strongly supporting the assignment of the isolated spectralfeature as a genuine vibrational contribution of the lysozyme.

Concerning the slower dynamics, our high resolution DLSmeasurements give evidence of additional features below ∼300GHz related either to other protein contributions or to the waterrelaxation term. To facilitate the analysis of this spectral region,the lysozyme vibrational contribution previously calculated atany given T was subtracted from the corresponding experimentalsusceptibilities.51 The resulting profiles, hereafter called differ-ence spectra (DS), are represented in Figure 3. As can beobserved, the intensity of the low-frequency tail (which can beattributed to slow protein relaxation processes) increases withtemperature. This is a genuine effect considering that below 10GHz the intensity of pure water spectrum is very weak (seeFigure 2b) and that it further decreases at higher T owing tothe high frequency shift of the water relaxation mode34,36 (seeinset of Figure 3). Moreover, between ∼20 GHz and ∼300 GHzthe presence of an additional component can be envisaged. Inorder to single out the different contributions, a comparison withinelastic neutron scattering (INS) data has been made.

3.3. Comparison to Neutron Scattering Data. The INStechnique is mainly sensitive to the motions of the hydrogenatoms of the sample, because of their dominating cross section.In this respect, the protein dynamics can be convenientlyinvestigated by monitoring these atoms, which are largely

distributed within the whole molecule.9,10,41,42 In these experi-ments the interference of the solvent can be eliminated byworking in deuterated water. Since solute-solvent relativeactivities are expected to be similar in INS and DLS spectra,41

their mutual comparison may be used to recognize the differentcontributions. To this end, the INS profiles of partially deuteratedlysozyme/D2O samples10 have been compared to the DLSspectra taking advantage of the use of the susceptibilityformalism.41 The dynamic susceptibility is related to thedynamical structure factor S(Q,ν) obtained in INS experimentsthrough the relation �INS′′ (Q,ν) ∝ S(Q,ν)/[nB(ν) + 1].52 We notethat in the case of hydrated powders with h ) 0.33 thecontribution of the first hydration shell to the signal can beneglected.10,42 On the contrary, to completely isolate the proteinsignal in the solution spectra, the subtraction of the D2O solventsignal is needed.10 The resulting INS solvent free spectrum isrelated to the single particle dynamics of the protein and mainlyarises from its internal motions.

In Figure 4 is shown a comparison between DLS (SFDLS; blueline) and INS (SFINS; red line) solvent free spectra obtained at25 °C. The SFINS spectrum has been rescaled to the intensity ofthe SFDLS one within the 600-1200 GHz frequency range. Herethe susceptibility relates to the vibrational contribution of the

Figure 1. Imaginary part of the susceptibility of pure water andlysozyme-water solutions at different protein concentration (T ) 25°C). The spectra are rescaled to the librational band (frequencies higherthan 104 GHz). The arrow indicates the direction of spectral variationsobserved at increasing lysozyme concentration. The inset depicts anexample of combination between interferometric (FPI; blue) anddispersive (U1000; green) spectra.

Figure 2. DLS solvent free spectrum (SFDLS, blue) obtained bysubtracting the bulk solvent susceptibility (gray) from that of thesolution (orange). The total curve and the single components (pink)obtained by fitting the SFDLS spectrum with three Brownian oscillatorsare also shown. The profiles are depicted both in linear (a) andlogarithmic (b) frequency scales.

8264 J. Phys. Chem. B, Vol. 114, No. 24, 2010 Perticaroli et al.

protein (boson peak).10,32 This assignment is further supportedconsidering that in this region all the profiles are similar,included both DLS and INS signals of hydrated powders.

Looking at Figure 4, two relevant observations can be made:(1) the SFDLS spectrum shows a clear intensity excess at around40 GHz that can be connected to the changes induced by theprotein on the solvent and, more specifically, assigned tohydrating water; (2) the low-frequency tail of the measured DLSspectrum well extrapolates to the SFINS one through a commonν-0.3 power law.

3.4. DLS Data Analysis. Based on the previous findings, toattain a more quantitative description, each DS spectrum ofFigure 3 has been modeled as the sum of five components: apower law (R ) 0.3) to reproduce the low-frequency tail, twoCole-Davidson (CD) functions to account for the relaxationterms, and two damped harmonic oscillator (DHO) forms forthe water resonant modes,

where A is the amplitude of the power law; τ and � are thecharacteristic time and stretching parameters of the CD func-tions; ωb, ωs, and Γb, Γs, are the characteristic frequency anddamping of the water bending and stretching resonant modes;∆ is the relative amplitude of each term. The two DHO functionsare used to model bending and stretching intermolecularvibrations in accord to previous investigations,35-37 while thechoice of the CD functions relates to the nonexponentialbehavior of the water relaxation process.36,37,53,54 We remark that,with respect to pure water, the modeling of the solutions’susceptibilities requires an additional CD function at around40-50 GHz and that a single relaxation term, even with avariable stretching parameter, does not adequately reproducethe data. To reduce the number of free fitting parameters, �slow

and �fast have been fixed to 0.6 as obtained in pure water.36,37

In any case we verified that the results do not strongly dependon these values. As an example Figure 5 depicts the best-fitcurve together with the calculated five components. Thecorresponding CD average relaxation times ⟨τ⟩ ) τ × � arereported in Figure 6. The temperature behavior of the fastrelaxation process (fraction of picoseconds) closely resemblesthat of pure water and is on average ca. 7-fold faster (inset ofFigure 6) than the slower one (few picoseconds). The ∆slow/(∆slow+∆slow) ratio divided by the protein/water molar ratio, flys,is reported in Figure 7. This quantity represents the averagenumber of water molecules affected by one lysozyme molecule,as long as the protein dynamics do not contribute to ∆slow. Weare confident on the robustness of the adopted parametrization,since slightly different fitting procedures were attempted (i.e.,fixing ωb, ωs, Γb, Γs as in pure water and/or leaving Runconstrained) and similar two-relaxation patterns obtained.

4. Discussion

The DLS solvent free susceptibility obtained in the 270-5300GHz region (Figure 2) is very similar to that derived in theOHD-RIKES study of Giraud et al.32 confirming the generalpicture presented there. This is not a trivial result, consideringthat the two techniques are based on very different experimentalapproaches and data treatments. In Table 1 the parametersobtained as best-fitting result using three Brownian oscillatorsare reported in comparison to the findings of ref 32. Only thevalues of the first component show meaningful discrepancies,evidencing in our case a much stronger overdamping characterof the mode. This difference is mainly related to the fact that,differently from ref 32, we adopted a total free fitting procedure.From a physical point of view, the larger damping rate of thefirst oscillator in comparison with the two high frequency onesmay reflect the different nature of the motion described by theformer component. In particular, the lowest frequency over-damped mode was associated to the boson peak visible in theINS profile55 and ascribed to librational motions of solvent-exposed side chains,32 while the two additional Brownianoscillators, located at ca. 2490 and 3360 GHz, were related tobackbone torsions.32

Hedoux et al.,44 by investigating the amide I Raman band,clearly evidenced that the thermal denaturation of lysozyme ataround 77 °C, due to the unfolding of the secondary structure,follows a premelting phenomenon within the 57-72 °C rangeassigned to a change of the tertiary structure. In this contextour findings suggest that this structural modification does notaffect the low-frequency vibrational pattern of lysozyme,indicating that side-chain librations and backbone torsions donot experience strong inter- and intramolecular energy changes.This is in agreement with the fact that the low-frequency features(<6000 GHz) of several globular proteins with different second-ary structures look very similar to each other.32

While, for the 270-5300 GHz region, our results agree withthe previous findings,32,44,45,56 remarkable differences weredetected at both higher and lower frequencies. In particular,Figure 2 shows an accurate overlapping between water andsolution spectra at ν > 6000 GHz, making doubtful the presenceof a distinct component at ∼9000 GHz, previously attributedto specific intermolecular interactions.45,56 The weakness of thisfeature and its temperature independence seem to indicate itsorigin as an artifact of the subtraction, as already suggested.32

More interestingly, our high resolution data can extend downto fractions of GHz, highlighting the complex nature of thespectral distribution below 270 GHz, which directly probesprotein-induced solvent perturbations (see Figure 2).

Figure 3. Difference spectra obtained by subtracting from thesusceptibility of the lysozyme solution (100 mg/mL) the vibrationalcontribution of the protein extended on the whole frequency range.This latter has been estimated as best-fitting result (see Figure 2 andref 51). The arrow indicates the direction of spectral variations observedat increasing temperatures. Inset: Imaginary part of the susceptibilityof pure water at different temperatures.

�′′(ω) ) Aω-R + Tm{- ∆slow

[1 + iωτslow]�slow-

∆fast

[1 + iωτfast]�fast

+

∆bωb2

ω2 - ωb2 - iωΓb

+∆sωs

2

ω2 - ωs2 - iωΓs

} (1)

DLS of Protein Aqueous Solutions J. Phys. Chem. B, Vol. 114, No. 24, 2010 8265

In view of the similarity of the fast relaxation time (⟨τfast⟩ ≈0.5 ps at 25 °C) with that of pure water, this process can bereasonably attributed to bulk water molecules not muchinfluenced by the solute. In pure water this relaxation is relatedto the fast rearrangements of local structures and its activity in

DLS experiments, which depends on the modulation of theanisotropic polarizability of the system, ascribed to intermo-lecular interaction-induced effects (collision contributions).34,54

Even if the molecular nature of this process is still not clarified,its time scale is that of the fast H-bonding dynamics (i.e., H-bond

Figure 4. Colored blue area represents the intensity excess obtained comparing DLS solvent free susceptibility (blue line) of the lysozyme-H2Osolution (100 mg/mL; 25 °C) with the equivalent INS solvent free spectrum (red line) of the lysozyme-D2O solution (100 mg/mL; 25 °C). DLS(green) and INS (orange) profiles of hydrated powders (h ≈ 0.33) are also shown. All these spectra are rescaled to the signal at frequencies higherthan 600 GHz due to the vibrational protein contribution. The low-frequency tail of the corresponding DLS difference spectrum (black) can bereproduced by a ν-0.3 power law which extrapolates to the INS solvent free spectrum.

Figure 5. DLS difference spectrum (black) of the lysozyme-watersolution (100 mg/mL; 25 °C) shown together with the total curve (red)and the different components resulting from the fitting procedure. Fromleft to right: power law (green), two Cole-Davidson functional formsfor the slow (cyan) and the fast (blue) relaxation terms, and two dampedharmonic oscillators reproducing the water bending (magenta) andstretching (orange) resonant modes (see eq 1).

Figure 6. Average relaxation time of the hydrating ⟨τslow⟩ (cyan circles)and bulk ⟨τfast⟩ (blue circles) water relaxation processes obtained bythe adopted fitting procedure for the 100 mg/mL lysozyme solution atdifferent temperatures. The relaxation time of pure water is also reported(gray squares) for comparison. Error bars also include uncertaintiesobtained by different curve fitting procedures (see text for details). Theinset shows the calculated retardation factor � ) ⟨τslow⟩/⟨τfast⟩ at differenttemperatures.

8266 J. Phys. Chem. B, Vol. 114, No. 24, 2010 Perticaroli et al.

lifetime).34,38-40,57 In fact ⟨τfast⟩ is comparable to the fastrelaxation time τHB of the continuous H-bond time correlationfunction computed in MD simulations.38-40,48-50 This shortrelaxation time τHB measures the temporal scale of thermallyactivated local restructuring process.38,39,50

Recently, insights into the H-bond rearrangement process inwater have been obtained by studying the OH stretchingvibration through ultrafast two-dimensional infrared (2D-IR)spectroscopy.58-61 From these experiments the frequency-frequency correlation function (FFCF), whose decay is directlyrelated to the H-bonding dynamics, can be obtained. After anultrafast component, connected with the OH frequency modula-tion mainly induced by intermolecular stretching motions, theFFCF shows a biexponential relaxation with time constants of0.4 and 1.7 ps.62 The slowest time is thought to represent thetime scale of the global H-bond reorganization involvingbreaking and re-forming of H-bonds together with reorientingand switching processes.62 The faster one has been related tomore local H-bond fluctuations. The two relaxation times mayalso describe a single restructuring process with an intrinsicnonexponential nature.62 Thus, the use of a CD function for theDLS relaxation process of water with an average relaxation time⟨τ⟩ ≈ 0.5 ps appears consistent with these findings. The overallagreement is especially remarkable taking into account the verydifferent nature of the experimental approaches together withthe complex character of the H-bond relaxation itself, which

involves translational and reorientational motions spanning widetemporal ranges.63 Likely, ⟨τfast⟩ derived in DLS experimentsof pure water mainly accounts for faster (and more local) H-bonddynamical events that precede a more global restructuring. Thiscan be explained considering that the dominating contributionto the DLS of water arises from translational interaction-inducedmechanism.34,54

In analogy with the recent results obtained for carbohydratesaqueous solutions,37 the other picosecond relaxation process(⟨τslow⟩ ∼ 3 ps) is attributed to water molecules directlyinfluenced by the protein (i.e., hydration water). Comparisonwith the solvent free INS susceptibility of Figure 4 supportsthis interpretation. We note that none of the relaxation processesdetected in hydrated powders and assigned to internal lysozymemotions is distinctly peaked in the frequency range of 40-50GHz corresponding to ⟨τslow⟩.41,42 This reinforces the idea that⟨τslow⟩ relates to the H-bond dynamics of the water moleculesin close proximity of the protein surface and that � ) ⟨τslow⟩/⟨τfast⟩ ∼ 6-7 at 25 °C essentially represents the dynamicretardation factor of this hydration water as compared to thebulk. This dynamic perturbation factor slightly increases athigher temperatures, even if detected variations do not exceedthe experimental errors (inset of Figure 6).

Several experimental investigations revealed the slowingdown of the rotational dynamics of water. Nevertheless, whilevery large perturbation effects going from tens to thousandswere derived by dielectric and fluorescence spectroscopy,2,17,19-21,24

smallervaluesof someunitscomeout fromNMRinvestigations.4,6

This more contained effect is confirmed by MD simulations16,64

and by other interpretations of dielectric12,22 and fluorescence11

data. The � estimated in the present work is reasonablycompatible with these latter evidence. It also emphasizes howthe mechanism involved in the rotational motion water mol-ecules strongly relates to the fast H-bond rearrangementdynamics.18,27,57,65 We want to remark that the dynamical processobserved in DLS experiments is considerably faster than thatdetected by other techniques. For instance, in pure water thefaster relaxation time measured in dielectric studies (∼8 ps),12,22

sensitive to collective dipolar orientations, is more than ten timeslarger than the ⟨τfast⟩ of the present work because of the differentphysical origin of the mechanism probed by the experiments.A dielectric relaxation time near 20-60 ps is commonlyassigned to hydration water12,22 corresponding to � ∼ 3-8.Interestingly, despite the different physical nature of the involvedprocesses, similar retardation factors were derived.

Comparison with previous DLS results on aqueous solutionsof carbohydrate37 indicates that � is similar for both sugars andlysozyme, suggesting (i) that the fast (and local) rearrangementdynamics at picosecond time scales is not dramatically influ-enced by the size of the solute, as already deduced by NMRfindings;4,6,23 and (ii) that, likely, hydrophilic solute-solventH-bonding contributions must play a relevant role in influencingthe detected process.28,57 MD simulations show that H-bondcorrelation times slow down, going from the bulk to the proteinhydration shell,18,28,29 and that within the shell both water-waterand water-protein H-bonds stabilize.28,29 It may be interestingto remark that the calculated inverse of the forward rate constantof the water-protein breaking process is around 1-4 ps, inclose correspondence with the detected ⟨τslow⟩.28,29

Figure 6 shows that the observed ⟨τslow⟩ is practicallytemperature independent, with a minor decrease detected athigher temperatures. The trend should be regarded with prudenceconsidering that the accuracy in estimating ⟨τslow⟩ slightlydecreases at higher temperatures when the low-frequency protein

Figure 7. Temperature dependence of the average number of watermolecules dynamically perturbed by one lysozyme molecule derivedconsidering the amplitude of the hydrating (∆slow) and bulk (∆fast) waterrelaxation processes as resulting from adopted fitting procedure.

TABLE 1: Frequency of the Maximum of the Peak (νmax),Amplitude %, Characteristic Frequency (ω) and Damping(γ) Parameters Listed with Their Statistical UncertaintiesObtained As Multiple Brownian Oscillator Best-Fit of theSolvent Free (SFDLS) Spectruma

νmax (GHz) amplitude % ω (GHz) γ (GHz)

LS1150 ( 50 36 ( 15 1710 ( 180 2940 ( 4502360 ( 50 29 ( 6 2490 ( 30 1560 ( 903271 ( 50 35 ( 5 3360 ( 30 1680 ( 60

OHD-RIKESb

1075 ( 50 9 ( 1 1350 ( 60 1740 ( 1502304 ( 50 46 ( 5 2460 ( 60 1740 ( 1503244 ( 50 44 ( 4 3360 ( 60 1740 ( 150

a SFDLS spectrum obtained subtracting the water spectrum fromthat of a 100 mg/mL solution (25 °C). b Parameters found by Giraudet al.32 from OHD-RIKES data relative to a 125 mg/mL solution.

DLS of Protein Aqueous Solutions J. Phys. Chem. B, Vol. 114, No. 24, 2010 8267

tail intensifies. In any case a reduced thermal effect on ⟨τslow⟩seems consistent with MD calculations on hydrated powderswhich indicate that the fast protein-water H-bond relaxation(τHB) is scarcely affected by temperature variations.25

Other useful information can be derived considering therelative amplitude between slow and fast water relaxationprocesses.37 In this regard, if the slow component only accountsfor the dynamics of the protein-perturbed water, then it can beestimated that ca. 2300 water molecules per protein aredynamically influenced by the solute (see Figure 7). This numberis practically temperature independent until 70 °C when itreduces considerably. This evaluation implies that the proteineffect extends for more then a single hydration layer suggesting,in contrast with the NMR findings,4,6,23 a long-range perturbationeffect.66 On the other hand, it must be noticed that the presenceof some additional contribution in the 40-50 GHz spectralregion arising from side-chain diffusive motions cannot be, inprinciple, excluded.9,10,67,68 Such a contribution would hardlyshift the frequency location of the bump in the solvent freesusceptibility spectrum (blue line in Figure 4) associated withhydration water. This guarantees the safe evaluation of dynamicperturbation factors. Conversely, the area of the peak (that is,the hydration number evaluation) can be significantly affectedby this possible contribution. To achieve more quantitativeinformation on this point, high resolution inelastic neutronscattering experiments are planned.

Finally, with reference to Figure 5, the tail observed in thelowest frequency region is interpreted in terms of proteinrelaxation processes extending outside the accessed spectralwindow. These dynamics are slower compared to the structuralrelaxation of the protein detected in hydrated powders (30-80ps)41,42 and may involve the global rotation (tumbling) oflysozyme in the aqueous medium. In this regard NMR and lightscattering findings indicate that the corresponding second rankcorrelation time is around 5-10 ns,69,70 in agreement withdielectric results.12 Thus, the lysozyme rotation, which isexpected to be DLS active, should be peaked at around tens ofMHz and may partially enter into our spectral window. Thetemperature-induced intensity increase (Figure 3) observed atlow frequencies is compatible with the progressive accelerationof this process. Nevertheless, the considerable stretching of theexponent (R ) 0.3) may put forward the presence of additionalprotein contributions.

5. Conclusions

A broadband depolarized light scattering study has beenperformed on diluted aqueous solutions of lysozyme at differenttemperatures. For the first time the dynamical susceptibility ofa protein aqueous solution has been measured in a wide spectralrange (0.6-36000 GHz) by the combined use of two differentinstruments. Our experimental method allowed us to investigatedisparate dynamical events and to analyze contributions arisingfrom both protein and water.

A two-step global fitting procedure has been applied in orderto rationalize the complex spectral distributions achieving aconsistent modeling of the whole spectrum. Above ∼300 GHzthe water contribution is superimposed to a distinct bandascribed to the protein internal vibrations which is not affectedby temperature changes. Therefore, the extensive spectralvariations observed from 25 to 70 °C are basically due to thewater signals that essentially behave as in pure water (Figure3). In this respect any eventual increased protein flexibility andstructural rearrangement in the premelting region44 does notinfluence (or is not influenced by) the corresponding internal

dynamics. Our findings also indicate that the spectral contribu-tion detected in the solvent free spectrum at around 9000GHz32,45,56 is very weak and does not show any remarkabletemperature dependence. This small feature, which was previ-ously considered as a signature of strong protein-solventinteractions,45 is probably caused by the lack of a perfect spectralcompensation in the normalization procedure and can hardlybe representative of a distinct component.

Below ∼300 GHz we evidenced two relaxation processesrelated to the solvent dynamics: a fast component (fraction ofpicoseconds) related to water molecules not perturbed by thesolute (bulk water) and a slow component (few picoseconds)attributed to proximal water molecules (hydrating water). Asimilar picture was recently drawn for trehalose aqueoussolutions.37 Even if the contribution of other fast protein motionscannot be totally ignored, the comparison performed with theneutron scattering spectrum supports the idea that this slowspectral component is mainly due to water molecules localizedaround the protein. Following this assignment it is thenestablished that the fast rearrangement dynamics of the hydratingwater at 25 °C is 6-7 times slower compared to the bulk; thisquantitative experimental indication broadly confirms previousNMR estimates.4,6,23 On a molecular basis the involved relax-ation process is directly connected with the H-bond lifetime.In this respect the measured retardation factor is similar to thatobtained for the aqueous solutions of sugars and in line withmolecular dynamics simulations. Overall findings suggest thefollowing: (1) besides slower processes widely explored by otherexperimental techniques, the protein induces a considerableretardation degree even for fast water rearrangements atpicosecond time scales, singularly revealed by DLS experiments;(2) within the lysozyme surface hydrophilic (H-bond) groupsshould play a major role in determining these effects; (3) withinthe protein hydration shell the stability of both protein-waterand water-water H-bond increases; (4) this dynamical behaviorseems to have a local character being not strongly dependenton the solute size.

Finally, the lowest frequency spectral region, related to slowprotein motions, was modeled by a power law which nicelyextrapolates to the neutron scattering profile. This occurrencemust be verified by means of a more quantitative comparisonof INS and DLS spectra; this is mandatory in order to reachquantitative information on the spatial extent of the affectedhydration shell.4,6,23,60,71 High resolution neutron scatteringmeasurements have been scheduled to this end.

References and Notes

(1) Bizzarri, A. R.; Cannistraro, S. J. Phys. Chem. B 2002, 106, 6617–6633.

(2) Pal, K. S.; Peon, J.; Bagchi, B.; Zewail, A. H. J. Phys. Chem. B2002, 106, 12376–12395.

(3) Bagchi, B. Chem. ReV. 2005, 105, 3197–3219.(4) Mattea, C.; Qvist, J.; Halle, B. Biophys. J. 2008, 95, 2951–2963.(5) Mattos, C.; Clark, A. C. Arch. Biochem. Biophys. 2008, 496, 118–

131.(6) Qvist, J.; Persson, E.; Mattea, C.; Halle, B. Faraday Discuss. 2009,

141, 131–144.(7) Zhang, Y.; Lagi, M.; Liu, D.; Mallamace, F.; Fratini, E.; Baglioni,

P.; Mamontov, E.; Hagen, M.; Chen, S.-H. J. Chem. Phys. 2009, 130,135101/1–8.

(8) Fenimore, P. W.; Frauenfelder, H.; McMahon, B. H.; Young, R. D.Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 14408–14413.

(9) Perez, J.; Zanotti, J. M.; Durand, D. Biophys. J. 1999, 77, 454–469.

(10) Marconi, M.; Cornicchi, E.; Onori, G.; Paciaroni, A. Chem. Phys.2008, 345, 224–229.

(11) Halle, B.; Nilsson, L. J. Phys. Chem. B 2009, 113, 8210–8213.(12) Oleinikova, A.; Smolin, N.; Brovchenko, I. J. Phys. Chem. B 2004,

108, 8467–8474.

8268 J. Phys. Chem. B, Vol. 114, No. 24, 2010 Perticaroli et al.

(13) Cheng, Y. K.; Rossky, P. J. Nature 1998, 392, 696–699.(14) Johnson, M. E.; Malardier-Jugroot, C.; Murarka, R. K.; Head-

Gordon, T. J. Phys. Chem. B 2009, 113, 4082–4092.(15) Pizzitutti, F.; Marchi, M.; Sterpone, F.; Rossky, P. J. J. Phys. Chem.

B 2007, 111, 7584–7590.(16) Henchman, R.; McCammon, J. A. Protein Sci. 2002, 11, 2080–

2090.(17) Pal, S. K.; Peon, J.; Zewail, A. H. Proc. Natl. Acad. Sci. U.S.A.

2002, 99, 1763–1768.(18) Xu, H.; Berne, B. J. J. Phys. Chem. B 2001, 48, 11929–11932.(19) Zhang, L.; Wang, L.; Kao, Y.-T.; Qiu, W.; Yang, Y.; Okobiah,

O.; Zhong, D. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 18461–18466.(20) Li, T.; Hassanali, A. A.; Zhang, L.; Kao, Y.-T.; Zhong, D.; Singer,

S. J. J. Am. Chem. Soc. 2007, 129, 3376–3382.(21) Zhang, L.; Yang, Y.; Kao, Y.-T.; Wang, L.; Zhong, D. J. Am. Chem.

Soc. 2009, 131, 10677–10691.(22) Sasisanker, P.; Weingartner, H. Chem. Phys. Chem. 2008, 9, 2802–

2808.(23) Qvist, J.; Halle, B. J. Am. Chem. Soc. 2008, 130, 10345–10353.(24) Mashimo, S.; Kuwabara, S.; Yagihara, S.; Higasi, K. J. Phys. Chem.

1987, 91, 6337–6338.(25) Tarek, M.; Tobias, D. J. Phys. ReV. Lett. 2002, 88, 138101/1–4.(26) Wood, K.; Frolich, A.; Paciaroni, A.; Moulin, M.; Hartlein, M.;

Zaccai, G.; Tobias, D. J.; Weik, M. J. Am. Chem. Soc. 2008, 130, 4586–4587.

(27) Jana, B.; Pal, S.; Bagchi, B. J. Phys. Chem. B 2008, 112, 9112–9117.

(28) Bandyopadhyay, S.; Chakraborty, S.; Bagchi, B. J. Am. Chem. Soc.2005, 127, 16660–16667.

(29) Chakraborty, S.; Bandyopadhyay, S. J. Chem. Phys. B 2008, 112,6500–6507.

(30) Kropman, M. F.; Bakker, H. J. Science 2001, 291, 2118–2120.(31) Giraud, G.; Wynne, K. J. Am. Chem. Soc. 2002, 124, 12110–12111.(32) Giraud, G.; Karolin, J.; Wynne, K. Biophys. J. 2003, 85, 1903–

1913.(33) Hunt, N. T.; Kattner, L.; Shanks, R. P.; Wynne, K. J. Am. Chem.

Soc. 2007, 129, 3168–3172.(34) Paolantoni, M.; Sassi, P.; Morresi, A.; Santini, S. J. Chem. Phys.

2007, 127, 24504/1–9.(35) Fioretto, D.; Comez, L.; Gallina, M. E.; Morresi, A.; Palmieri, L.;

Paolantoni, M.; Sassi, P.; Scarponi, F. Chem. Phys. Lett. 2007, 441, 232–236.

(36) Paolantoni, M.; Comez, L.; Fioretto, D.; Gallina, M. E.; Sassi, P.;Morresi, A.; Scarponi, F. J. Raman Spectrosc. 2008, 39, 238–243.

(37) Paolantoni, M.; Comez, L.; Gallina, M. E.; Sassi, P.; Scarponi, F.;Fioretto, D.; Morresi, A. J. Phys. Chem. B 2009, 113, 7874–7878.

(38) Starr, F. W.; Nielsen, J. K.; Stanley, H. E. Phys. ReV. Lett. 1999,82, 2294–2297.

(39) Teixeira, J.; Luzar, A.; Longeville, S. J. Phys.: Condens. Matter2006, 18, S2353–S2362.

(40) Han, S.; Kumar, P.; Stanley, H. E. Phys. ReV. E 2009, 79, 041202/1-5.

(41) Roh, J. H.; Curtis, J. E.; Azzam, S.; Novikov, V. N.; Peral, I.;Chowdhuri, Z.; Gregory, R. B.; Sokolov, A. P. Biophys. J. 2006, 91, 2573–2588.

(42) Khodadadi, S.; Pawlus, S.; Sokolov, A. P. J. Phys. Chem. B 2008,112, 14273–14280.

(43) Persson, E.; Halle, B. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 6266–6271.

(44) Hedoux, A.; Ionov, R.; Willart, J.-F.; Lebret, A.; Affouard, F.;Guinet, Y.; Descamps, M.; Prevost, D.; Paccou, L.; Danede, F. J. Chem.Phys. 2006, 124, 014703/1-7.

(45) Hedoux, A.; Willart, J.-F.; Paccou, L.; Guinet, Y.; Affouard, F.;Lebret, A.; Descamps, M. J. Phys. Chem. B 2009, 113, 6119–6126.

(46) Nielsen, O. F. Annu. Rep. Prog. Chem., Sect. C: Phys. Chem. 1997,93, 57–99.

(47) Carey, D. M.; Korenowski, G. M. J. Chem. Phys. 1998, 108, 2669–2675.

(48) Chandra, A. Phys. ReV. Lett. 2000, 85, 768–771.(49) Elola, M. D.; Ladanyi, B. M. J. Chem. Phys. 2006, 125, 184506/

1–13.(50) Chowdhuri, S.; Chandra, A. J. Phys. Chem. B 2006, 110, 9674–

9680.(51) Prior to the subtraction the best-fitting curve related to the

vibrational profile has been extrapolated up to few GHz (Figure 2). Wenote that the shape of the quasi-elastic region (ν < 270 GHz) is not stronglyaffected by the subtraction of this tail that contributes for less than 7% ofthe total intensity.

(52) Lovesey, S. W. Theory of Neutron Scattering from CondensedMatter; Oxford Science Publications: Oxford, 1984.

(53) Sokolov, A. P.; Hurst, J.; Quitmann, D. Phys. ReV. B 1995, 51,12865–12868.

(54) Torre, R.; Bartolini, P.; Rhighini, R. Nature (London) 2004, 428,296–299.

(55) Orecchini, A.; Paciaroni, A.; Bizzarri, A. R.; Cannistraro, S. J. Phys.Chem. B. 2001, 105, 2150–12156.

(56) Colaianni, S. E. M.; Nielsen, O. F. J. Mol. Liq. 1995, 347, 267–284.

(57) Russo, D.; Ollivier, J.; Teixeira, J. Phys. Chem. Chem. Phys. 2008,10, 4968–4974.

(58) Asbury, J. B.; Steinel, T.; Kwak, K.; Corcelli, S. A.; Lawrence,C. P.; Skinner, J. L.; Fayer, M. D. J. Chem. Phys. 2004, 121, 12431–12446.

(59) Roberts, S. T.; Loparo, J. J.; Tokmakoff, A. J. Chem. Phys. 2006,125, 084502/1-8.

(60) Loparo, J. J.; Roberts, S. T.; Tokmakoff, A. J. Chem. Phys. 2006,125, 194522/1–12.

(61) Kumar, R.; Schmidt, J. R.; Skinner, J. L. J. Chem. Phys. 2007,126, 204107/1–11.

(62) Park, S.; Fayer, M. D. Proc. Natl. Acad. Sci. U.S.A. 2007, 104,16731–16738.

(63) Tay, K. A.; Bresme, F. Phys. Chem. Chem. Phys. 2009, 11, 409–415.

(64) Marchi, M.; Sterpone, F.; Ceccarelli, M. J. Am. Chem. Soc. 2002,124, 6787–6791.

(65) Laage, D.; Hynes, J. T. Science 2006, 311, 832–835.(66) Ebbinghaus, S.; Kim, S. J.; Heyden, M.; Yu, X.; Heugen, U.;

Gruebele, M.; Leitner, D. M.; Havenith, M. Proc. Natl. Acad. Sci. U.S.A.2007, 104, 20749–20952.

(67) Ishikawa, H.; Finkelstein, I. J.; Kim, S.; Kwak, K.; Chung, J. K.;Wakasugi, K.; Massari, A. M.; Fayer, M. D. Proc. Natl. Acad. Sci. U.S.A.2007, 104, 16116–16121.

(68) Gaspar, A. M.; Appavou, M.-S.; Busch, S.; Unruh, T.; Doster, W.Eur. Biophys. J. 2008, 37, 573–582.

(69) Dubin, S. B.; Clark, N. A.; Benedek, G. B. J. Chem. Phys. 1971,54, 5158–5164.

(70) Halle, B.; Davidovic, M. Proc. Natl. Acad. Sci. U.S.A. 2003, 100,12135–12140.

(71) Matyushov, D. V. Phys. ReV. E 2010, 81, 021914/1-11.

JP101896F

DLS of Protein Aqueous Solutions J. Phys. Chem. B, Vol. 114, No. 24, 2010 8269