Urea-Mediated Protein Denaturation: A Consensus View

Atanu Das and Chaitali Mukhopadhyay*Department of Chemistry, UniVersity of Calcutta, 92, A. P. C. Road, Kolkata - 700 009, India

ReceiVed: July 6, 2009

We have performed all-atom molecular dynamics simulations of three structurally similar small globularproteins in 8 M urea and compared the results with pure aqueous simulations. Protein denaturation is precededby an initial loss of water from the first solvation shell and consequent in-flow of urea toward the protein.Urea reaches the first solvation shell of the protein mainly due to electrostatic interaction with a considerablecontribution coming from the dispersion interaction. Urea shifts the equilibrium from the native to denaturedensemble by making the protein-protein contact less stable than protein-urea contact, which is just thereverse of the condition in pure water, where protein-protein contact is more stable than protein-watercontact. We have also seen that water follows urea and reaches the protein interior at later stages of denaturation,while urea preferentially and efficiently solvates different parts of the protein. Solvation of the protein backbonevia hydrogen bonding, favorable electrostatic interaction with hydrophilic residues, and dispersion interactionwith hydrophobic residues are the key steps through which urea intrudes the core of the protein and denaturesit. Why urea is preferred over water for binding to the protein backbone and how urea orients itself towardthe protein backbone have been identified comprehensively. All the key components of intermolecular forcesare found to play a significant part in urea-induced protein denaturation and also toward the stability of thedenatured state ensemble. Changes in water network/structure and dynamical properties and higher degree ofsolvation of the hydrophobic residues validate the presence of “indirect mechanism” along with the “directmechanism” and reinforce the effect of urea on protein.

I. Introduction

Denaturing osmolyte such as urea is used extensively to assessprotein stability, the effects of mutations on stability, and proteinunfolding.1 Despite its widespread use, the molecular mechanismof urea-induced denaturation of proteins is poorly understood.Urea may exert its effect directly, by binding to the protein,2-7

and compete with native interactions, thereby actively partici-pating in the unfolding process. Most versions of the directinteraction model posit that urea binds to, and stabilizes, thedenatured state (D), thereby favoring unfolding. However, thisinterpretation does not explain how the protein itself surmountsthe kinetic barrier leading to unfolding. Given the fact that thereis no built-in binding site for osmolytes and the universal natureof their actions, a common mechanism via protein backbone islikely to be method of choice for the “direct” interaction.8

Alternatively, it has been proposed that urea acts indirectly byaltering the solvent environment,9-11 thereby mitigating thehydrophobic effect and facilitating the exposure of residues ofthe hydrophobic core. In a recent work by Hua et al., it wasreported that urea is attracted toward the protein surface due todispersion.7 The direct dispersion interaction between urea andthe protein backbone and side chains is stronger than for water,which gives rise to the intrusion of urea into the protein interiorand also to urea’s preferential binding to all regions of theprotein.7 The noninvolvement of electrostatics comes as asurprise as urea has sufficient polarity (dipole moment 4.56 D),and there had been numerous reports on the importance ofelectrostatics in protein-urea interaction.12-14 In addition, ac-cording to a recent report, protein denatured state stability ismainly governed by electrostatic interaction.15 Can the observa-

tion of Hua et al.7 be protein dependent? Even accepting thefact that urea is preferred over water in the vicinity of theprotein, the question why protein-protein contacts are brokenin the presence of urea still remains unanswered.

Recently, we have reported the early events of urea-inducedprotein denaturation for three proteins, Ubiquitin, G311, andGB1, which share a similar secondary structural arrangement(�-grasp fold). The sequence similarity between G311 and GB1is higher (>80% similarity), but the ubiquitin sequence is quitedifferent (∼30% similarity). It was shown that protein-ureacontacts replace both protein-protein and protein-water con-tacts, and both hydrophobic and hydrophilic residues contributefavorably to the unfolding process.16 We have extended theunfolding trajectory lengths here until the global unfoldingnature of the simulation ceases to minimal fluctuation. A totalof 100 ns data for ubiquitin and 75 ns each for the other twoproteins have been collected. Our current study mainly aims toelucidate and quantify: (1) mode of protein-urea (P-U)interaction, i.e., whether the driving force is electrostatic or vander Waals (VDW), (2) favorable binding site of urea-proteinbackbone or side chain, and (3) intrinsic mechanism behind urea-induced protein unfoldings “direct mechanism” or “indirectmechanism”.

II. Methods

NVT simulations were performed with CHARMM packageusing CHARMM22 force field and parameters. The ureaparameters were taken from ref 17. SHAKE was used tomaintain the bond lengths and angles of urea and water. Anonbonded cutoff of 8 Å was used. Periodic boundary conditionsand minimum image were used to reduce edge effects. PMEwas applied to deal with the long-range electrostatic interactions

* Corresponding author. E-mail address: chaitalicu@yahoo.com;cmchem@caluniv.ac.in.

J. Phys. Chem. B 2009, 113, 12816–1282412816

10.1021/jp906350s CCC: $40.75 2009 American Chemical SocietyPublished on Web 08/26/2009

with a 9 Å cutoff. Three proteins were simulated in 8 M aqueousurea solutionssUbiquitin (1UBQ.PDB), G311 (1ZXH.PDB),and GB1 (1GB1.PDB). The 8 M aqueous urea system ofUbiquitin contained 438 urea molecules and 1739 watermolecules. For G311, 1793 water molecules and 460 ureamolecules were present in the 8 M aqueous urea solution. Inthe case of GB1, 464 urea molecules and 1784 water moleculeswere present in the final arrangement. A set of three independentsimulations was performed for all the protein systems. ForUbiquitin, each simulation was truncated after 33.4 ns and datawere reported up to 33 ns, and for G311 and GB1, simulationswere truncated after 25 ns. Resulting water boxes had 2265,2348, and 2321 water molecules for Ubiquitin, G311, and GB1,respectively, in 0 M solutions. For each of the three proteins, aset of three independent simulation trajectories, each of 15 nslength, was obtained for pure aqueous solutions (0 M) of theproteins. Details of the simulation protocol have already beenreported in our earlier paper16 except that we have extendedthe simulation times here (Supporting Information) to get acomplete picture of the denaturation process.

III. Results

In all the simulations, almost completely denatured states ofthe proteins with residual secondary structures (central R-helix)were obtained. These proteins are structurally similar in theirnative states, and their urea-denatured states also resemblestructurally to a high extent (Figure S.1, Supporting Informa-tion). The existence of residual helical structure for severalglobular proteins [R- + �-type] in the urea-denatured states hasbeen reported earlier.7,18,19

Dynamics of P-U Contact Formation. In the aqueousenvironment, protein remains in its folded conformation bymaintaining a careful balance between its native contacts andthe contacts with neighboring water. Addition of urea introducesanother contact option, and protein negotiates its contactpreferences and thereby changes its conformation to achievethermodynamic stability in the altered environment.

We have previously shown16 that loss of protein-protein(P-P) and protein-water (P-W) contacts occurs in tandemwith the formation of P-U contacts. The time course of theexclusion of water from the first solvation shell (FSS) to bulkis further shown by plotting the number of water (nw) and ureamolecules (nu) in the FSS as a function of time (Figure 1). Itcan be seen clearly from the figure that there is a rapid loss ofsurface water within first 3-5 ns for all three proteins. Thiscan be related to the popular “dry globule” formation concept7

where the P-W contacts break ahead of P-U contact formation.The ratio nw/nu in both FSS and bulk is calculated. The ratio

decreases from 4.03 to 1.68 in FSS, within a fairly short period(3-5 ns), and fluctuates around the value at longer times,whereas for bulk it is near 4.15 in all three protein systems. Itis seen that for a loss of ∼150 water molecules from FSS (Figure1) the increase in nu is ∼70. The molecular volume of urea isnearly three times that of water, which implies that a loss of150 water molecules can at best make space for 50 ureamolecules. The additional urea can be accommodated withinthe FSS, if only the size of the FSS increases, i.e., if the proteinexpands. This is actually observed in the first few nanosecondswhere the Rg of protein increased significantly (SupportingInformation of ref 16). Snapshots of gradual urea accumulationaround the protein are shown in Figure S.2 (SupportingInformation). It is interesting to see that though the number ofwater molecules in FSS decreases initially after ∼15 ns thereis a sharp increase, indicating that at later stages both waterand urea solvate the denatured protein. This fact is furtherdiscussed later. However, the initial loss of water from FSS orfinal inclusion of water to the FSS is not observed for 0 Msimulations (Figure S.3, Supporting Information).

The accumulation of urea can be justified from the stabilityof the contact formation between different atoms pairs, e.g.,backbone-backbone (CR-CR), backbone-water (CR-WO), andbackbone-urea (CR-UO). The contacts are counted if therespective pairs come within a distance of 4 Å. The contactsurvival time correlation function (CST) [expression in Sup-porting Information] (Figure 2) shows that the CR-WO cor-relation function has a faster decay than CR-CR and CR-UO

correlation. The faster decay of the CR-WO correlation explainsthe mechanism of “dry globule” formation. In pure water, thescenario is quite different, and both the CR-WO and CR-CR

CSTs show much slower decay indicating stability of the formedcontacts, i.e., no unfolding of the protein.

Energetics of P-U Interaction. The reason for the increasein nu in the FSS relative to bulk can be explained from anenergetic perspective. Following the recent analysis of Hua etal.,7 we have computed the electrostatic and VDW interactionenergies of each urea/water molecule in the FSS (within 5 Åfrom any protein atoms) and in bulk (defined as more than 5 Åaway from any protein atoms) with the rest of the system. Theelectrostatic energy distribution for the urea/water molecule inthe FSS and bulk (Figure 3, for Ubiqutin, for the other twoproteins see Figure S.4a,4b, Supporting Information) shows thatwhen a urea molecule moves from the bulk (∼-7.78 kcal/mol)to the FSS (∼-19.87 kcal/mol) there is significant stabilization(∼-12.09 kcal/mol), whereas for water it is only -0.25 kcal/mol. Urea in the FSS of protein has a distribution of VDWenergy with a sharper peak at lower energy (at ∼-2.6 kcal/

Figure 1. Exclusion/inclusion of water and urea in the FSS of the protein.

Molecular Basis of Urea Denaturation J. Phys. Chem. B, Vol. 113, No. 38, 2009 12817

mol) than that for urea in the bulk (at ∼-0.6 kcal/mol) (Figure3). In contrast, there is relatively little change in the VDWenergy distribution of water in the FSS and bulk in both thepeak magnitude and position. For all the three protein systems,

the difference in the averaged VDW energy for urea in the FSSand in the bulk is within -2.0 kcal/mol, whereas for water it isonly -0.14 kcal/mol. Thus, accumulation of urea in the FSS isfavored mainly due to electrostatic interaction. VDW interaction

Figure 2. Contact pair survival time correlation function of CR-CR, CR-UO, and CR-WO in 0 M (right) and 8 M (left) solutions for all threeproteins.

Figure 3. Interaction energy distribution. The probability distribution function of VDW (right) and electrostatic (left) energy of urea (top) andwater (bottom) in the first solvation shell of protein and in the bulk region with the rest of system. The first solvation shell (FSS) is defined aswithin 5.0 Å of protein, and the bulk region is defined as not within 6.0 Å of protein (Ubiquitin).

12818 J. Phys. Chem. B, Vol. 113, No. 38, 2009 Das and Mukhopadhyay

has a smaller effect in guiding urea to the FSS of the protein.This observation is quite opposite to that proposed by Hua etal.,7 and the probable reason might be the inherent differencesof the nature of the protein surfaces under study. This has beendiscussed later. For 0 M solution, water has no energeticadvantage between FSS and bulk, and hence, neither initial waterexclusion nor final water inclusion was observed in thesesimulations (Figure S.5, Supporting Information).

Factors Responsible for Binding of Urea to Protein. Afterthe initial accumulation of urea near protein, the next step ofdenaturation will include binding of urea to protein atoms andbreaking the P-P contacts. To evaluate the strength of variouscontacts, the distribution of interaction energy (normalized overresidue number, number of urea, and number of water mol-ecules) between atoms of (1) P-P, (2) P-W, and (3) P-U iscalculated (Figure 4) for 8 and 0 M urea solutions. In all theprotein systems, the figure shows that in 8 M urea solution theP-P interaction is unfavorable compared to P-W and P-U.Thus, P-P contacts can be replaced by either P-U or P-Wcontactssthe former is more energetically favorable than thelatter. Interestingly, for 0 M solutions, normalized P-P interac-tion energy is ∼-25 cal/mol, and P-W interaction energy is∼-13.5 cal/mol. The P-P interaction is therefore morestabilized than P-W, and thus protein retains its native contactsin water.

Favorable Binding Site of Urea: Backbone vs Side Chain.As the side chain atoms of protein are usually more exposedthan the backbone atoms, it is expected that urea will prefer-entially bind to the side chain in the initial stages. We havecalculated the number of the heavy atom contacts betweenbackbone/side chain atoms and urea/water, separately. However,as the side chain atoms are much more exposed than main chainand as backbone contributes 25% and side chain contributes75% to the newly exposed surface area,16 to remove that bias,the numbers were normalized by calculating the ratio of changein the number of contacts to the change in the solvent accessible

surface area (SASA) (Figure 5). We see that the ratio for thebackbone-urea contacts increases from 0.05 to 0.28 withoccasionally reaching a value more than 0.4, whereas the sidechain-urea contacts increase from 0.06 to 0.12 indicating thaturea preferentially binds with the backbone. The ratio forbackbone-water contacts decreases from a value of 0.15 to 0.02,and that for side chain-water contacts decreases from 0.2 to0.05. For comparison, P-W heavy atom contacts in 0 Msolutions, both for backbone and side chain, are shown in thefigure. The values remain almost constant with a small degreeof fluctuation. Here, side chain has a slight preference over thebackbone for water.

Contribution of Hydrophilic and Hydrophobic Contactsto Denaturation. We have previously shown that both hydro-philic and hydrophobic residues contribute to urea-mediatedprotein unfolding.16 In aqueous (0 M urea) solution, a proteinremains in its native fold with most of the hydrophobic residuesburied in the core and not exposed to the solvent. However, inconcentrated aqueous urea solution (8 M), the protein unfoldswith a consequent exposure of the hydrophobic residues towardthe solvent. This is well evidenced in the contact distributionplot of total contacts, hydrophilic-hydrophilic contacts, andhydrophobic-hydrophobic contacts of all the three proteins in0 and 8 M aqueous urea solutions (Figure S.6, SupportingInformation). For Ubiquitin, there is a shift in the peak of thenumber of total contacts from 290 to 180 on going from 0 to 8M urea solution. The consequent change in the hydrophilic-hydrophilic contact peak is from 140 to 125, and that of thehydrophobic-hydrophobic contact peak is from 85 to 40. ForG311, the total contact peak shifts from 215 to 100; thehydrophilic-hydrophilic contact peak shifts from 90 to 65; andthe hydrophobic-hydrophobic contact peak shifts from 65 to17. In the case of GB1, the shifts are from 217 to 108 for thetotal contacts, from 88 to 62 for the hydrophilic-hydrophiliccontacts, and from 63 to 19 for the hydrophobic-hydrophobiccontacts. Hence, the decrease in the total number of contacts is

Figure 4. (a) Distribution of P-P, P-W, and P-U interaction energy terms normalized over protein residue number, number of water molecules,and number of urea moleculesswater and urea molecules within 5 Å from the protein (i.e., in the FSS) were considered (8 M)sP-P (left), P-W(middle), and P-U (right), and (b) same as previous for 0 M solutionssP-P (left) and P-W (right).

Molecular Basis of Urea Denaturation J. Phys. Chem. B, Vol. 113, No. 38, 2009 12819

mainly contributed by the loss in the hydrophobic-hydrophobiccontacts in the denatured state of the protein. Hydrophilic-hydrophilic contacts also decrease, but the extent is smaller.Urea forms more contacts with protein hydrophobic residuesthan water leading to more exposure of the core hydrophobicresidues and a concomitant decrease in the intraproteinhydrophobic-hydrophobic contacts.

Interaction of Different Segments of Protein with Urea.To compare how the different parts of the denatured proteinsare efficiently solvated by water and urea, the normalizedinteraction energy terms (electrostatic and VDW) of backbone,nonpolar side chain, charged polar side chain, and polar sidechain with urea and water have been calculated (Figure S.7a-7c,Supporting Information). It is seen from the figures that proteinbackbone is stabilized by both electrostatic (-0.3 to -2.0 kcal/mol) and VDW (-0.06 to -0.15 kcal/mol) interactions withurea. Water also stabilizes backbone by electrostatic interaction,

though by a much smaller amount (-0.1 to -0.5 kcal/mol).Nonpolar side chains are stabilized by VDW dispersion of urea(-0.05 kcal/mol), and the corresponding terms in water are alldestabilizing. Charged polar side chains are solvated favorablyby electrostatics of urea (-0.4 to -0.8 kcal/mol) and water(-0.5 to -1.3 kcal/mol); the latter contributes more. Polar sidechains interact almost comparably with urea (-0.1 to -0.4 kcal/mol) and water (-0.1 to -0.2 kcal/mol) in terms of electrostat-ics, but the VDW interaction is more stabilized in urea (-0.02to -0.08 kcal/mol) than in water (-0.005 to 0.01 kcal/mol).These facts can be confirmed from the radial distribution plotsof urea and water atoms around different protein atoms (FigureS.8a-8c, Supporting Information). It is seen that backbone andhydrophobic residues are preferably solvated by urea, whereascharged hydrophilic residues are preferentially solvated by water.Propensity of urea around the charged polar residues alsoincreases during the course of denaturation, but it was less than

Figure 5. Time evolution of ∆Contact/∆SASA for backbone and side chainsboth with water and urea. Preference of urea toward backbone (left).No distinct change for the values from 0 M simulations (right).

12820 J. Phys. Chem. B, Vol. 113, No. 38, 2009 Das and Mukhopadhyay

the contribution from water. Corresponding interaction energyterms (Figure S.9, Supporting Information) and radial distribu-tion functions (Figure S.10, Supporting Information) have alsobeen calculated for water simulations for comparison.

Preferable Orientation of Urea toward Protein. Both theenergetics as well as contact evolution analysis suggest that ureapreferentially binds to the backbone. Urea can bind to the proteinbackbone (via H-bond) by two different binding modessviaits amide hydrogens that bind to the protein backbone carbonyloxygen and via the oxygen atom that binds to the protein amidehydrogens. To investigate the orientation preference of urea,we have calculated the N (urea)-O (peptide carbonyl) and O(urea)-N (backbone amide) contact distributions. As urea hastwo N-atoms and one O-atom, it is obvious that the averagevalue of the distribution will be higher for the first one. Hence,the number of N (urea)-O (amide) contacts was divided by 2(Figure 6). In all three cases, we see that the N-atom of ureahas a preference over the O-atom toward binding to thebackbone. The preference is 28%, 30%, and 36% for Ubiquitin,G311, and GB1, respectively. So urea preferentially binds tothe backbone and via the nitrogen end mainly.

Evidence for Indirect Mechanism. i. Change in WaterStructure and SolWent Dynamical Properties. Addition of ureato the aqueous solution of the protein causes retardation of waterrotational reorientation, slower diffusion of water molecules ofthe FSS, and increase in water-water H-bond survival times.Detailed descriptions of the results of solvent dynamicalproperties along with relevant tables (S.I.-Tables 1-3) and

figures (Figure S.11-S.13) have been given in the SupportingInformation. The observed changes in the water dynamicalproperties in the presence of 8 M urea signify the change inwater characteristics and hence water structure from that of purewater. In our previous study, we had shown from the radialdistribution function of water that there is no significantdeformation of the water structure.16 But according to Vanzi etal.,20 examination of the radial distribution function alone is notsufficient to conclude that the oxygen atom of a urea moleculealways fits into the network of water molecules. One has tocheck hydrogen bond angle, which is a very sensitive measureof hydrogen bond characteristics. The changes in the water-waterhydrogen bond angle (decreases) and distance (increases) in 8M aqueous urea solution (shown in Figure S.14, SupportingInformation) are consequently indicative of the breaking of waterstructure and so the presence of “indirect” mechanism in urea-mediated protein denaturation.

ii. Formation of Urea Cluster. We have calculated theurea-urea radial distribution function (rdf) (Figure 7). Thesignature of urea cluster formation is evident from the rdf valuesobtained at a later stage of unfolding (shown for last 1 ns) whichis absent initially (shown for first 1 ns). As the simulationprogresses, urea not only binds to the protein but also forms anetwork structure. Previously, it was proposed that water canaccommodate urea within its network due to the high H-bondforming propensity of urea, and thus the water structure remainsunaltered after addition of urea.21 In our simulations, urea formsdynamic clusters (Figure 7) (three urea molecules in a cluster),

Figure 6. Contact probability distribution of two different binding modes between the protein backbone and urea. Urea nitrogen has a slightlyhigher preference for backbone amide oxygen than the complementary mode.

Figure 7. Radial distribution function between center of masses of urea at first 1 ns and last 1 ns of simulations for the three proteins, clearevidence of cluster of formation.

Molecular Basis of Urea Denaturation J. Phys. Chem. B, Vol. 113, No. 38, 2009 12821

and consequently water may no longer accommodate urea dueto the larger size of the urea cluster compared to the watertetrahedral cavity. This is evident from the triplet angledistribution (see Supporting Information for definition). Thesharp diminution of the broad feature near cos(θ) )-0.2 (whichin pure water is indicative of the tetrahedral H-bond networkbeing present at short distances) and marked growth aroundcos(θ) ) 0.5 (corresponding to near neighbor interstitialmolecules) indicate a distorted local water structure in the 8 Maqueous urea solution (Figure S.15, Supporting Information).However, we have not observed any segregation of urea due toclustering in any simulation trajectory. Hence, the presence of8 M urea alters the water network, which further provides prooffor the evidence of indirect mechanism. The fact that urea formsclusters in concentrated solution and consequently breaks thewater structure is in agreement with previous reports.22,23

IV. Discussion

The results obtained from molecular dynamics simulationsof the three proteins in the presence of 8 M urea lead us to geta lucid visualization of the stepwise mechanism of urea-mediatedprotein denaturation. We have done control simulations of allthe three proteins at 0 M urea (pure aqueous environment). Theformation of a “dry globule”7 is convincingly shown from thetime series of nu and nw in the FSS (Figure 1), which shows arapid initial loss of water from the vicinity of the protein. Asthe size of urea is nearly 3 times that of water, the loss in nw islarger than the gain in nu at the initial stages, making theapproach of urea to protein surface entropically favorable. Thisis also evident from the faster decay of P-W CST than P-P/P-U CSTs (Figure 2). The time evolution of the fraction ofthese contacts (Figure S.16, Supporting Information) alsosupports this observation. Thus, the mechanism of urea-mediateddenaturation proceeds at the first step by forming P-U contactsat the cost of P-W and/or P-P contacts.

The claim made by Hua et al.7 that urea accumulation at FSStakes place due to strong dispersion interaction with proteinsmay not be a general one, as we have observed that it isadvantageous for urea with both the electrostatic as well asdispersion components (Figure 3). This observation that elec-trostatic also plays a crucial role in P-U interaction has beensupported by O’Brien et al.12 and Tobi et al.13 where theelectrostatics was shown to be the major interaction. We suggestthat this might be protein dependent. The electrostatic surfaceof all three proteins of the present study is highly polar withconsiderable presence of positive and negative potentials (FigureS.17, Supporting Information). Rossky14 has recently argued thatcollective evidence shows that all the key components of theintermolecular forces (van der Waals attraction, electrostatics,hydrogen bonding, and hydrophobic interaction) play a signifi-cant part in creating the relative affinities. We speculate that itis electrostatic interaction that makes urea approach closer tothe protein surface at the initial stage than dispersion interaction(which is inherently weaker and short range).

Once urea reaches the protein surface, it can replace bothP-P and P-W contacts by P-U contacts, as the pairwiseinteraction energy of P-U contacts is more favorable (Figure4a). It has been recently pointed out by Rossky14 that themarginal stability of native fold of globular proteins is restrictedto a relatively narrow window of thermodynamic and solutioncomposition conditions. In aqueous solutions, to maintain thenative structure, the protein must have more propensities to formP-P contacts than P-W contacts (Figure 4b). On the contrary,in the presence of denaturing agents, this propensity should get

reversed. This is actually what has been observed from ourresults. This might be the reason urea molecules can shift theequilibrium from native to denatured state. This is furthervalidated by the gradual increase in the number of P-U contactsand corresponding decrease in the P-P and P-W contacts(Figure S.16, Supporting Information). Strikingly, there is anincrease in the P-W contacts at the later stage of denaturation(Figure S.16, Supporting Information), which indicates that asthe protein opens up both urea and water can stabilize the newlyexposed regions providing the stabilization of the denaturedstate.

Preference of urea for the backbone is clearly shown fromthe time evolution of contact formation (Figure 5), interactionenergy terms (Figure S.7a-7c, Supporting Information), radialdistribution functions (Figure S.8a-S.8c, Supporting Informa-tion), and H-bond geometry (angle and distance distributions)(Figure S.18a-S.18b, Supporting Information). It might be dueto the smaller sacrifice that a protein has to make to replace abackbone-backbone H-bond by a backbone-urea H-bondrather than a backbone-water H-bond. This fact is supportedby the observation that the geometry of P-U H-bonds (FigureS.18a-S.18b, Supporting Information) is very similar to P-PH-bonds. Also, urea stays for a longer time than water at thevicinity of the backbone, which is evident from the longersurvival time of backbone-urea H-bonds than backbone-waterH-bonds (Figure S.18c, Supporting Information), and thiscorrelates with one earlier finding of Dotsch et al.24 Thisbackbone-based theory of urea-protein interaction corroborateswith many earlier findings.7,8,12,25-28 It has been argued by Roseet al.8 that the universal nature of osmolyte effect throughoutall three kingdoms of life suggests that their primary effect ison backbone, not side chains. Urea preferentially orients itsnitrogen sites toward the backbone rather than the oxygen site(Figure 6), and this supports one earlier result of Smith et al.29

though differs from that proposed by Hua et al.7 According toChen et al.,30 excess positive or negative charge on the proteinsurface dictates orientation of urea. Our protein systems have alarge number of exposed oxygen atoms on their surfaces(indicated by red regions in Figure S.17, Supporting Informa-tion). Therefore, urea orients its nitrogen toward the protein inthe present study and oxygen toward the protein in the case oflysozyme7 (exposed nitrogen atoms indicated by blue regionsin Figure S.17, Supporting Information) for favorable bindingwith the exposed counterparts.

Various regions of protein contribute differently to thedenatured state stability. Backbone-urea interaction is primarilyelectrostatic in nature, and it is more favorable than backbone-water interaction (Figure S.7a-7c, Supporting Information).Nonpolar side chain-urea interaction is stabilized by dispersion,and it has no stabilization with water. Polar side chains havealmost equal preference for urea and water in terms ofelectrostatic, but the dispersion part is more stabilized in urea.Similar conclusions have been reported earlier.7,16 Our findingsslightly differ from Hua et al.7 where polar side chains havestronger dispersion interaction with urea than nonpolar sidechains; we have gotten the reverse order. This might be due tothe fact that we have subdivided the polar residues in chargedand uncharged components. Loss in number of contacts isdistinctly high within the hydrophobic residues, and that hasthe major contribution to the overall loss in number ofintraprotein contacts (Figure S.6, Supporting Information). Theextent of solvation of different parts of protein in concentratedaqueous urea solution obtained from the radial distributionfunctions (Figure S.8, Supporting Information) also supports

12822 J. Phys. Chem. B, Vol. 113, No. 38, 2009 Das and Mukhopadhyay

the above findings where protein is preferentially solvated byurea; only a departure is observed for the charged polar sidechains which prefer water over urea.7,31,32 In our simulations,urea solvates both positive and negatively charged side chainssignificantly, having larger interaction with the later part,compared to that of Hua et al.,7 where urea solvates only thepositively charged side chains and to a very small extent. Finally,we can say that preferential solvation of the hydrophobicresidues and binding of urea to the protein backbone are themain features of urea-mediated protein denaturation which mightbe due to the reason that preferential binding of urea to thebackbone and hydrophobic residues with displacement of wateris favorable both entropically and enthalpically: first, due totranslational and rotational entropy gain as one urea replacesthree water molecules, and second because these water mol-ecules can form strong water-water H-bonds.31 Urea’s abilityto substitute the intraprotein H-bonds and stronger dispersioninteraction with nonpolar residues allows urea’s intrusion intothe core of globular proteins and provides a plausible explanationfor urea’s strong denaturing power.7,33

All the above arguments are in support of the “directinteraction model” of urea-induced protein denaturation. But,simultaneous presence of an “indirect mechanism” is alsoevident from our extensive analysis of solvent network anddynamical properties. The mean square displacements, rotationalorientations, and survival time correlation functions of thesolvent molecules clearly indicate a distinct effect of urea onwater stucture and dynamics.

The interaction of urea with the protein is preceded with adeparture of water molecules from the FSS to the bulk leadingto formation of a dry globule. This indicates a change in solventenvironment (contact arrangement) and simultaneously assistsurea to reach the FSS by providing space, and consequentlyurea forms contacts with backbone and hydrophobic residuesand denatures protein. This is called the “chaotropic effect” ofurea.18,34,35 Urea disrupts the formation of water structure bybreaking down the hydrogen-bonded network via creatingweaker hydrogen bonding in water which is evident from anincrease in water-water H-bond length (Figure S.14, SupportingInformation). Urea is pushed away from the bulk water to thesolute particles, i.e., preferentially excluded from the solution,an effect known as the “preferential binding” to the solute.36

Thus, when urea molecules are less than optimally hydrated,urea hydrogen bonds to itself and the protein (significantlyinvolving the peptide links) in the absence of sufficient water,so becoming more hydrophobic and thereby contributing tosolvation of hydrophobic parts, and hence is more able to interactwith further sites on the protein, leading to localized dehydra-tion-led denaturation. In our simulations, there is enhancedinteraction between the chaotrope (urea) and the solute (protein),and thus the effect is much more enhanced. In summary,“indirect mechanism” paves and smoothens the way of the“direct mechanism”.

V. Conclusions

The stability of the native folds of globular proteins is ratherremarkable, in that this stability is marginal and restricted to arelatively narrow window of thermodynamic and solutioncomposition conditions.14,37 What urea does is that it shifts theequilibrium from the native conformation toward the denaturedensemble by altering the contact network of the protein behavingitself as a soluble amide.38 In this report, we have drawndistinctive conclusions about the initial events of urea-mediatedprotein unfolding as well as the stabilization of the denatured

state, which has been attracting attention for a long time.39 Forcethat increases urea population near the protein surface is mainlyelectrostatic in nature, similar to that observed by O’Brien etal.12 and Tobi et al.,13 along with a relatively low contributionfrom the VDW dispersion, as observed by Hua et al.7 Onecrucial observation is that the protein-protein interactions,which were favorable in water, become unfavorable in thepresence of urea, which effectively tilts the balance from nativetoward the denatured states. The ability of urea to interactfavorably with the backbone as well as with both the polar andnonpolar side chains plays a crucial role in the mechanism.These observations clearly point toward a “direct” interactionphenomenon. On the other hand, from the solvent structure anddynamical property analysis, we see that contact networkchanges as unfolding progresses. This indicates the existenceof an “indirect” mechanism in the unfolding process. Bettersolubilization of hydrophobic residues in the urea/water mixturestrengthens the concept of “indirect mechanism”. There are somereports on urea association40 and simultaneous presence of directand indirect mechanisms,18,40-43 which corroborate nicely withour results. So, the sequence of events in denaturation are: waterleaves the FSS, urea reaches protein surface, solubilizes theexposed charged side chains, binds to the H-bonded backboneO and NH groups with alternative partners, and interacts withthe nonpolar residues by VDW dispersion, and finally waterfollows urea and reaches the protein and successfully solvatesthe newly exposed protein atoms.

Acknowledgment. We are thankful to the Department ofChemistry, University of Calcutta, for the computational facili-ties under UPE scheme. A.D. is thankful to Council of Scientificand Industrial Research, India, for the Senior Research Fellow-ship through CSIR-NET.

Supporting Information Available: Additional informationon the simulation details, denatured state conformations, urea’sgradual accumulation around protein, preferential solvation byradial distribution functions, protein-urea contact formation,solvent structural and dynamical properties, i.e., rotationalreorientation, diffusion coefficient, survival time correlationfunction, H-bond structures of water-water, protein-water, andprotein-urea, and additional information of control simulations(aqueous urea solution). This material is available free of chargevia the Internet at http://pubs.acs.org.

References and Notes

(1) Pace, C. N. Methods Enzymol. 1986, 134, 266–280.(2) Astrand, P.-O.; Wallqvist, A.; Karlstroem, G. J. Phys. Chem. 1994,

98, 8224–8233.(3) Tirado-Rives, J.; Orozco, M.; Jorgensen, W. L. Biochemistry 1997,

36, 7313>7329.(4) Grdadolnik, J.; Marechal, Y. J. Mol. Struct. 2002, 615, 177–189.(5) Mountain, R. D.; Thirumalai, D. J. Am. Chem. Soc. 2003, 125,

1950–1957.(6) Klimov, D. K.; Straub, J. E.; Thirumalai, D. Proc. Natl. Acad. Sci.

U.S.A. 2004, 101, 14760–14765.(7) Hua, L.; Zhou, R.; Thirumalai, D.; Berne, B. J. Proc. Natl. Acad.

Sci. U.S.A. 2008, 105, 16928–16933.(8) Rose, G. D.; Fleming, P. J.; Banavar, J. R.; Maritan, A. Proc. Natl.

Acad. Sci. U.S.A. 2006, 103, 16623–16633.(9) Frank, H. S.; Franks, F. J. Chem. Phys. 1968, 48, 4746–4757.

(10) Finer, E. G.; Franks, F.; Tait, M. J. J. Am. Chem. Soc. 1972, 94,4424–4429.

(11) Hoccart, X.; Turrell, G. J. Chem. Phys. 1993, 99, 8498–8503.(12) O’Brien, E. P.; Dima, R. I.; Brooks, B.; Thirumalai, D. J. Am. Chem.

Soc. 2007, 129, 7346–7353.(13) Tobi, D.; Elber, R.; Thirumali, D. Biopolymers 2003, 68, 359–

369.(14) Rossky, P. J. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 16825–16826.

Molecular Basis of Urea Denaturation J. Phys. Chem. B, Vol. 113, No. 38, 2009 12823

(15) Weinkman, R.; Pletneva, E. V.; Gray, H. B.; Winkler, J. R.;Wolynes, P. G. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 1796–1801.

(16) Das, A.; Mukhopadhyay, C. J. Phys. Chem. B 2008, 112, 7903–7908.

(17) Caflisch, A.; Karplus, M. Structure 1999, 7, 477–488.(18) Bennion, B. J.; Daggett, V. Proc. Natl. Acad. Sci. U.S.A 2003, 100,

5142–5147.(19) Meier, S.; Strohmeier, M.; Blackledge, M.; Grzesiek, S. J. Am.

Chem. Soc. 2007, 129, 754–755.(20) Vanzi, S.; Madan, B.; Sharp, K. J. Am. Chem. Soc. 1998, 120,

10748–10753.(21) Wallqvist, A.; Covell, D. G.; Thirumalai, D. J. Am. Chem. Soc.

1998, 120, 427–428.(22) Soper, A. K.; Castner, E. W.; Luzar, A. Biophys. Chem. 2003, 105,

649–666.(23) Hernandez-Cobos, J.; Ortega-Blake, I.; Bonilla-Marin, M.; Moreno-

Bello, M. J. Chem. Phys. 1993, 99, 9122.(24) Dotsch, V.; Wider, G.; Siegal, G.; Wuthrich, K. FEBS Lett. 1995,

366, 6–10.(25) Caballero-Herrera, A.; Nordstrand, K.; Berndt, K. D.; Nilsson, L.

Biophys. J. 2005, 89, 842–857.(26) Auton, M.; Holthauzen, L. M. F.; Bolen, D. W. Proc. Natl. Acad.

Sci. U.S.A. 2007, 104, 15317–15322.(27) Meier, S.; Grzesiek, S.; Blackledge, M. J. Am. Chem. Soc. 2007,

129, 9799–9807.(28) Timasheff, S. N.; Xie, G. Biophys. Chem. 2003, 105, 421–448.(29) Smith, L. J.; Jones, R. M.; van Gunsteren, W. F. Proteins 2005,

58, 439–449.

(30) Chen, X.; Sagle, L. B.; Cremer, P. S. J. Am. Chem. Soc. 2007,129, 15104–15105.

(31) Stumpe, M. C.; Grubmuller, H. J. Am. Chem. Soc. 2007, 129,16126–16131.

(32) Stumpe, M. C.; Grubmuller, H. PloS Comput. Biol. 2008, 4,e1000221.

(33) Zangi, R.; Zhou, R.; Berne, B. J. J. Am. Chem. Soc. 2009, 131,1535–1541.

(34) Walrafen, G. E. J. Chem. Phys. 1966, 44, 3726.(35) Roseman, M.; Jencks, W. P. J. Am. Chem. Soc. 1975, 97, 754–

755.(36) Moelbert, S.; Normand, B.; Rios, P. D. L. Biophys. Chem. 2004,

112, 45–57.(37) Zhang, J.; Peng, X.; Jonas, A.; Jonas, J. Biochemistry 1995, 34,

8631–8641.(38) Zou, Q.; Habermann-Rottinghaus, S. M.; Murphy, K. P. Proteins

1998, 31, 107–115.(39) Stumpe, M. C.; Grubmuller, H. Biophys. J. 2009, 96, 3744–3752.(40) Czarnik-Matusewicz, B.; Kim, S. B.; Jung, Y. M. J. Phys. Chem.

B 2009, 113, 559–566.(41) Lee, M.-E.; van der Vegt, N. F. A. J. Am. Chem. Soc. 2006, 128,

4948–4949.(42) Rocco, A. G.; Mollica, L.; Ricchiuto, P.; Baptista, A. M.; Gianazza,

E.; Eberini, I. Biophys. J. 2008, 94, 2241–2251.(43) Moglich, A.; Krieger, F.; Kiefhaber, T. J. Mol. Biol. 2005, 345,

153–162.

JP906350S

12824 J. Phys. Chem. B, Vol. 113, No. 38, 2009 Das and Mukhopadhyay