Key Residues that Play a Critical Role in Urea-Induced Lysozyme

Nov 5, 2010 - IBM Thomas J. Watson Research Center. J. Phys. Chem. B 2010, 114, 15687–15693. 15687. 10.1021/jp1052453 2010 American Chemical ...
0 downloads 0 Views 1MB Size
Download PDF
J. Phys. Chem. B 2010, 114, 15687–15693

15687

Key Residues that Play a Critical Role in Urea-Induced Lysozyme Unfolding Meng Gao,† Zhen-Su She,†,‡ and Ruhong Zhou*,§,⊥ College of Engineering, Center for Theoretical Biology, and State Key Laboratory for Turbulence and Complex Systems, Peking UniVersity, Beijing 100871, China, Department of Chemistry, Columbia UniVersity, New York, New York 10027, United States, and Computational Biology Center, IBM Thomas J. Watson Research Center, 1101 Kitchawan Road, Yorktown Heights, New York 10598, United States ReceiVed: June 8, 2010; ReVised Manuscript ReceiVed: October 1, 2010

In this paper, we have developed a simple sensitivity score, based on the relative population of solvent molecules near each residue, to analyze the detailed motions of both urea and water around the hen egg-white lysozyme protein (W62G mutant) during its early stage of urea-induced unfolding for a better understanding of the atomic picture of the chemical denaturation process. Our simulation and analysis show that some hydrophobic core residues can keep dry from water for tens of nanoseconds in 8 M urea, while their contacts with urea increase significantly at the same time, forming a molten dry-globule-like state. Also, different from previously proposed actions that urea molecules preferentially absorb onto charged residues, our analysis shows that the noncharged residues, rather than the charged ones, attract more urea molecules in their surroundings (acting as attractants for urea), which is consistent with our earlier findings that urea molecules preferentially bind to protein through their stronger dispersion interactions than water. Once the initial adsorption surrounding the protein surface is accomplished, the further intrusion is found to be facilitated by a group of key residues, including Leu8, Met12, Val29, and Ala95, which play a critical role in the formation of the dry-globule structure. These hydrophobic dry residues form a local contact map which excludes the intrusion of water but accommodates the presence of urea due to their stronger binding to protein during this swelling process, thus maintaining an interesting transient dry-globule state. Introduction The characteristic native topologies of proteins result from the balance between the protein-water and protein-protein interactions under the normal physiological conditions, which has been well-illustrated by the hydrophobic-force-driven folding mechanism.1 When the environment changes, for example, changing to a different temperature or adding other chemicals such as urea,2-5 the native structure may no longer be stable, and the protein may start to unfold. To explain how urea breaks the protein-water/ protein-protein interaction balance and induces the protein unfolding, two mechanisms, “indirect” and “direct” models, have been proposed based on many experimental2-7 and theoretical8-18 studies. The indirect model pays more attention to the interaction between urea and water,3,5,8 suggesting that urea acts as a structure breaker for water so as to reduce the hydrophobic force and facilitate the solvation of hydrocarbons by water. The direct model, on the other hand, focuses on the interaction between urea and protein, either through stronger electrostatic interactions4,7,14,15 or nonpolar interactions2,12,15,17,19 with protein than those with water. Though more and more results support the direct interaction model through the urea preferential binding to the protein backbone or side chains,15,20,21 some recent studies indicate that the indirect mechanism also plays a role in the urea-induced protein denaturation.13,16 Meanwhile, there is still * To whom correspondence should be addressed. E-mail: ruhongz@ us.ibm.com. † College of Engineering and Center for Theoretical Biology, Peking University. ‡ State Key Laboratory for Turbulence and Complex Systems, Peking University. § Columbia University. ⊥ IBM Thomas J. Watson Research Center.

a lack of a detailed atomic picture on how exactly protein residues interact with urea and water molecules during the unfolding dynamical process, despite the extensive studies from both experiment and theory in last several decades. Given the space limit, only a very brief description of the previous efforts on this detailed atomic picture (mainly for the direct mechanism) is provided in the following, and interested readers can refer to refs 11, 14, 15, 17, and 22-25 for a more complete overview. Wallqvist et al.22 studied two methane molecules in aqueous urea using molecular dynamics simulations and proposed an “outside-in” mechanism to describe the chemical denaturation of globular proteins. They found that urea molecules preferentially absorb onto the charged hydrophilic residues on the surface, which gives rise to the swelling of the protein and subsequently the exposure of hydrophobic residues.22 Later studies by Grubmu¨ller and co-workers12,24 as we well as those of Zhou and co-workers15 showed that charged residues still prefer to be solvated by water molecules in 8 M urea solution. Furthermore, several recent studies indicate that urea molecules preferentially bind to proteins, especially hydrophobic residues24 through stronger dispersion interactions, and a transient dry-globule-like structure, where urea molecules intrude into the protein interior ahead of water molecules, can form in the early stage of unfolding.11,15 Meanwhile, both simulation and experimental studies have shown that there exist key residues in protein folding which often form critical contact networks in their folding transition states26,27 and consequently facilitate the folding process.28 Are there similar key residues (whether the same ones as the folding process or not) in the urea-induced protein unfolding process? Stumpe and Grubmu¨ller have recently studied the binding preference of each amino acid with urea or water12,24 and concluded that charged residues favor

10.1021/jp1052453  2010 American Chemical Society Published on Web 11/05/2010

15688

J. Phys. Chem. B, Vol. 114, No. 47, 2010

Gao et al.

i Figure 1. (a) RFSS, the ratio of Nw and Nu in the FSS of protein during the initial 100 ns MD simulation. (b) RFSS , ratio of 〈Nw〉 and 〈Nu〉 for each i means greater urea binding preferences. The results show a much preferred urea binding residue, averaged during the initial 100 ns. Small RFSS around many residues as compared to the background urea/water ratio (4.3) and mean RFSS (2.1).

water and nonpolar residues favor urea. Bennion and Daggett also examined certain key hydrophobic residues (such as Ile20, Leu49, Ile57) in CI213 denaturing in aqueous urea. However, many questions still remain, for example, which residues, if any, play a critical role in the initiation of the unfolding process, and which residues play a role in the formation of the dryglobule? How do they contribute to the intrusion of the urea into the hydrophobic core? In this article, we address these questions by carefully analyzing the changes of water and urea molecules in the first solvation shell (FSS) of each residue during the initial stage of unfolding (first 100 ns of the MD unfolding trajectory) of the W62G single mutant hen egg-white lysozyme in 8 M urea as a case study, which is part of our extensive multimicroseconds trajectories generated in a previous work.29 We find that when the protein starts to unfold from a well-packed structure, some surface residues act as attractants for urea molecules, and then, urea molecules intrude the protein hydrophobic core, which facilitate the formation of the molten dry-globule through some critical key residues. After that transient period (which lasts 20-50 ns), the hydrophobic core is exposed to water, and the protein further unfolds into extended forms. Our results provide a detailed atomic picture on this process and identify some key residues that initiate the protein unfolding process. Methods Molecular Dynamics (MD) Simulations. The analysis presented in this paper is based on molecular dynamics simulations15,30-35 of a mutant hen lysozyme (W62G), which has been reported in detail elsewhere.29 Therefore, only a brief description of the simulation setup will be presented here. Simulation was performed by the NAMD2 MD program,36,37 with the all-atom CHARMM (c32b1) force field38 for lysozyme and solvent urea. A slightly simplified TIP3P water model39 was used for water. Lysozyme was put in a 8 M urea box (73.1 Å × 73.1 Å × 73.1 Å). The entire system included 1 lysozyme protein (129 residues), 7799 water molecules, 1811 urea molecules, and 8 Cl- counterions. Only the initial 100 ns data from the five separate 1 µs long trajectories were used here to analyze the initial stage of lysozyme unfolding. Sensitivity Score (SS) for Each Residue. Before defining the SS for each residue in protein lysozyme, we define the first solvation shell (FSS) first. Similar to previous studies,12,15 FSS is defined as a 4.0 Å thick solvation shell from any heavy atom (non-hydrogen atom) of a single residue (or a group of residues). In other words, a water or urea molecule is considered to be inside of the FSS of a particular residue when the minimum

distance between any heavy atom of the residue and the solvent (urea or water) is no greater than 4.0 Å. The number of water and urea molecules in the FSS, Nw and Nu, are then calculated for that particular residue (or a group of residues) in lysozyme during the protein unfolding process. For any given residue (or a group of residues) R, the SS, Snx(R), is defined as

Snix(R) )

〈Nx(R)〉Ci 〈Nx(R)〉C*

where Nx refers to water (Nw) or urea (Nu) molecules around R, Ci refers to some particular ensemble of interest, and C* refers to the reference ensemble (more later). For any given R, ResSnix(R) is denoted as the SS for a single residue, and Grp-Snix(R) is denoted for a group of residues. They measure whether the particular residue (or a group of residues) is sensitive to solvent exposures and motions of the protein from the reference ensemble state C* to the current ensemble state Ci. Many different ensemble states might be used to indicate the progress of the urea-induced unfolding process. Results Water/Urea Ratio Indicating a Preferential Binding to Urea. Figure 1 shows the ratio of the number of water molecules to that of urea near the protein surface for one representative trajectory (other trajectories show similar behavior). In the bulk solution, this ratio is approximately 4.3 (7799/1811). However, at the protein surface, this ratio RFSS drops to ∼2.7 in a very short period of time (∼1 ns, during the equilibration) and then further drops to ∼2.0 after about 40 ns, indicating a strong binding preference of protein to urea. During this initial ∼40 ns of MD simulation, RFSS decreases quickly, as shown in Figure 1a, which is found to be mainly caused by the replacement of water by urea around the protein. With the further progress of unfolding, both Nu and Nw increase, but the relative enrichment of urea near the protein surface stays, with the mean ratio of Nw and Nu in the FSS of protein maintaining a roughly constant value at ∼2.0 from 40 to 100 ns. In order to further identify the specific preferential binding i for each residue in sites for urea, we calculated the ratio RFSS i 〉 is ∼1.61, the initial 100 ns. Averaged over 129 residues, 〈RFSS showing strong urea binding preference with the protein. It should be noted that there are much more shared urea molecules than water among residues (as indicated by our previous radial distribution analyses15); therefore, the mean ratio of single

Key Residues in Urea-Induced Lysozyme Unfolding

J. Phys. Chem. B, Vol. 114, No. 47, 2010 15689

Figure 2. (a) Free-energy landscape with CR-Rg as a reaction coordinate for the initial 50 ns MD data. Here, the maximum free energy was set to be 0kBT. The C1, C2, and C3 ensembles are subdivided from configurations in 1-50 ns by their CR-Rg, and C* includes configurations in the first 1 ns as the reference ensemble (only the first 50 ns of data are used here to catch the initial kinetics of the dry-globule formation; after 50 ns, water molecules also rush into the hydrophobic core, as shown in Figure 6 below). (b) The group SS Grp-Sn for water (black solid line) or urea (red dash line) of the core (O) and surface (0) groups, respectively. (c) The number of water molecules 〈Nw〉 near sensitive inner core residues (Res-Sn > 2 for water). (d) The number of urea molecules 〈Nu〉 near sensitive inner core residues (Res-Sn > 2 for urea). In (c, d), some of these sensitive residues are dry residues (see the text) and are plotted by red solid lines, whereas the others are plotted by black dash lines. i residues 〈RFSS 〉, averaged over 129 residues (∼1.61), is smaller than that of the entire protein (〈RFSS〉 ≈ 2.1). Interestingly, quite a few residues, Met12, Val29, Cys30, Val32, Cys94, and Ala95, are found to display the strongest preference to urea binding, i less than 1.0 (the smaller the ratio, the stronger with their RFSS the preference; see Figure 1b). A quick structural analysis shows that these residues are buried residues and are part of the protein hydrophobic core and thus have fewer contacts with water or urea in the first 1.0 ns (see Supporting Information Figure S1). On the other hand, most surface residues, both hydrophobic and hydrophilic, tend to have a larger water/urea ratio, even though they are still remarkably smaller than the bulk ratio (see Supporting Information Figure S1). Residues showing less urea i > 2.5) are all charged residues binding preference (with RFSS (Lys1, Glu7, Asp18, Glu35, Asp87, Lys96, Lys97, Asp101, and Asp119) and are highly exposed to solvent in the native structure (Supporting Information Figure S1). However, not all charged residues show less urea binding preference. Some of them, such as Arg21, Arg61, Asp66, and Arg125, show slight urea binding i as small as 1.8, but the majority of preference, with their RFSS charged residues do show less urea binding preference than most noncharged residues (Supporting Information Figure S1). Our previous radial distribution analyses have also shown that these charged residues solvate better in water than in urea (see Supporting Information of ref 15). Important Residues in the Early Urea Intrusion. A quick view of the MD trajectory movies show that most abrupt changes happen in the first 50-100 ns (different trajectories show slightly different kinetics, but the overall trends are the same). After the initial 40-50 ns, the water/urea ratio RFSS

reaches a plateau value (Figure 1a), even though the radius of gyration (CR-Rg) keeps increasing, along with the total number of FSS urea and water numbers. Therefore, we focus on the initial 50 ns data first. If we calculate the free-energy landscape using the radius of gyration as the reaction coordinate for the first 50 ns (Figure 2a), the most stable state is the native-like but somewhat swelled globule structures with a slightly larger Rg (even though this is not in equilibrium, but we use the freeenergy landscape from histograming40 anyway to indicate potentially different “ensemble sets”). This was also found by Stumpe and Grubmu¨ller on another protein, the cold shock protein, where a urea-induced equilibrium shift was reported toward the unfolded state.21 To explore the key residues involved in the so-called urea outside-in action,22 we employed a strategy similar to the one used in our previous work to identify the key residues in the Trp-cage folding process.18 First, a reference set C* was defined, which included all conformations in the first 1 ns. These were protein structures similar to the native structure (with CR-rmsd ) 0.26 ( 0.03 nm and CR-Rg ) 1.41 ( 0.01 nm) but with urea in the solvent. Then, the remaining conformations (from 1 to 50 ns in this representative trajectory) were classified into three subsets subdivided by their Rg (Figure 2a). These three subsets are named as C1, C2, and C3, with C1 being the most collapsed state, C3 being the most extended one, and C2 in-between. Following the above definition in the Methods section, for any given residue or a group of residues R, a SS, Snx(R), can be calculated as Snix(R) ) (〈Nx(R)〉Ci)/(〈Nx(R)〉C*) for all three subsets. This is to measure if the residue (or a group of residues) is sensitive to solvent exposures and motions of the protein from the reference subset C* to the current subset

15690

J. Phys. Chem. B, Vol. 114, No. 47, 2010

Gao et al.

Figure 3. (a) Number of dry residues (Ndry) in the initial 100 ns unfolding process. (b) Total dry time, tdry, for each residue in the initial 100 ns unfolding process (results do not change much even if we use only the first 50 ns data for most of the dry time comes from the first 50 ns, as shown in Figure 3a). A total of 13 residues (Leu8, Ala9, Ala11, Met12, Gly26, Trp28, Val29, Cys30, Val32, Val92, Cys94, Ala95, and Ile98) are found to show a tdry g 15 ns. These residues are found to be distributed mainly in three groups, the three out of four helices in the lysozyme.

Ci. Separately, according to their very initial exposure, all residues in the protein lysozyme are classified into two groups, the inner group, which includes 39 residues with Nw e 1 or Nu e 1 in C*, and the surface group, which includes the remaining 90 residues. Note that this subdivision is based on the reference initial exposure only (to water or urea) and is somewhat arbitrary, but it should give us a reasonable approximation to the initial spatial distribution of these residues. Notably, both inner and surface groups show larger Grp-Sn for urea than for water, indicating an overall stronger intrusion of urea than water. Furthermore, the inner group had much greater Grp-Sn than the surface group, particularly for urea (see Figure 2(b)), which means a remarkable increase of urea exposure for the inner group residues during the initial unfolding process. Moreover, the increase of urea Grp-Snu for the inner residues from ensemble (or subset) C* to C1, and C2, confirms the trend that hydrophobic core residues have a significant preference of urea binding over water during the initial unfolding process. The SSs of individual residues, Res-Sn, in both the inner and surface groups are also calculated. All 39 inner residues show a score for urea larger than 2 (Res-Snu > 2). To avoid large fluctuations due to the initial close-to-zero solvent exposure for some of the inner residues, we have also plotted their 〈Nw〉 and 〈Nu〉 directly (Figure 2c,d). As one can see from Figure 2c,d, 〈Nw〉 stays flat or even decreases for most of the inner residues from the C* state to the C1 state (for those red-colored “dry residues”, 〈Nw〉 stays flat throughout the C2 and C3 states as well), while 〈Nu〉 constantly increases from the C* state to the C1 state and then to the C2 and C3 states. On the other hand, for surface residues, although most of them show relatively small sensitivity scores, some of them, such as Lys1, Tyr20, Asn59, Thr69, Cys80, and Ile124 (total of 12 sensitive surface residues; see Figure 4 and Supporting Information Figure S2), they do attract many more urea molecules when the protein starts to unfold, with their Res-Snu1 > 2 at C1 (Supporting Information Figure S2), and except for Lys1, all of these residues are noncharged residues. Detailed analysis on protein-urea interactions shows that, in the initial 20 ns, urea molecules within the FSS of sensitive surface residues have both lower van der Walls and electrostatic interaction energies with the lysozyme than urea around nonsensitive residues (see Supporting Information Figure S4). In other words, the sensitive residues provide a more favorable FSS for urea molecules to approach the protein surface. These sensitive surface residues, accompanying with inner residues, play important roles in the initial urea intrusion process (Figure 4).

Previous studies have shown that during the initial stage of urea-induced lysozyme unfolding, there is a molten globulelike state, termed dry-globule (ensemble of native-like but somewhat swelled intermediates).11,15 Interestingly, direct experimental evidence has been reported very recently for this dry molten globule using two-site FRET probes and far-UV CD spectra in the chemical denaturation of a small plant protein, single-chain monellin.6 Here, we find that Grp-Snu (for urea) increases much faster than Grp-Snw (for water) from C* to C1 and C2, illustrating clearly that urea intrudes both the surface and inner core of the protein ahead of water. These data provide further evidence of the dry-globule formation. When time moves on, this dry molten globule is broken by the further intrusion of both water and urea from the C2 subset to C3. In this particular case, the dry-globule exists about 20 ns of time, and other trajectories show similar results that this dry molten globule can exist for approximately 20-50 ns for lysozyme unfolding in 8 M urea. Obviously, this duration will depend on the specific protein and the urea concentration; for lower urea concentration, it should be longer. Key “Dry” Residues in the Dry Globule. To further explore the atomic details of the dry-globule, residues with few water molecules around (Nw e 1) are recorded in each frame from the initial 100 ns simulation, and their total numbers were counted as Ndry. During the first 20 ns, even though the protein swells to some extent and the number of urea molecules increases significantly near the FSS, there are still more than 15 residues dry from water (Figure 3a). As the lysozyme unfolding proceeds, Ndry starts to decrease in general. After ∼60 ns, almost all residues are hydrated by water. Separately, for each residue, the total time of being dry (Nw e 1), tdry, can be recorded during this same time period. Especially, the following four hydrophobic residues, Leu8, Met12, Val29, and Ala95, show a drying time tdry of more than 40 ns, indicating that they might play a critical role in the formation of the dry-globule structure. If we classify those residues with tdry g 15 ns to be dry residues, a total of 13 such residues (Leu8, Ala9, Ala11, Met12, Gly26, Trp28, Val29, Cys30, Val32, Val92, Cys94, Ala95, and Ile98) are identified (Figure 3b). One thing interesting to note is that these dry residues are mainly distributed in three groups, belonging to three out of the four helices in lysozyme, Leu8, Ala9, Ala11, and Met12 on helix A, Gly26, Trp28, Cys30, and Val32 on helix B, and Val92, Cys94, Ala95, and Ile98 on helix C (Figures 3b and 4). This is consistent with our previous findings that the alpha-domain of the protein lysozyme unfolds after the beta-domain.29,41 Interestingly, these dry residues also belong to three out six hydrophobic clusters

Key Residues in Urea-Induced Lysozyme Unfolding

J. Phys. Chem. B, Vol. 114, No. 47, 2010 15691

Figure 4. Snapshots from the first 50 ns trajectory, (a) 0, (b) 10, (c) 20, (d) 30, (e) 40, and (f) 50 ns. The 13 dry residues are shown by white van der Waals (vdW) balls (Leu8, Ala9, Ala11, Met12, Gly26, Trp28, Val29, Cys30, Val32, Val92, Cys94, Ala95, and Ile98), and the 12 sensitive surface residues (with significant increasing Nu in C1) are shown by blue vdW balls (Lys1, His15, Tyr20, Phe38, Tyr53, Asn59, Thr69, Cys80, Leu83, Ala107, Val109, and Ile124; all labeled).

Figure 5. (a) The CR-Rg of 13 dry residues in lysozyme during the initial 100 ns unfolding process. These dry residues pack closely until the dry-globule is destroyed (after 50 ns). (b) The total local contact number among those 13 dry residues. These dry residues can maintain their local contact network for more than 20 ns. The local contacts of each dry residue also show a similar trend. (c) The total contact number among all 129 residues in the lysozyme, which shows a constant decay even in the first 20 ns. Here, data have been smoothed with a 1 ns sliding window average.

that have been identified by previous experimental NMR measurements.42 Moreover, these three groups of dry residues show the strongest urea-binding preference (Figure 1b), indicating that there are stronger tendencies for urea molecules to gather around these dry residues (Figure 2c,d). Therefore, these key dry residues are important components of the dry-globule. It should be noted that these results do not depend on the exact definition of the FSS. We have tried another two criteria to define the FSS, that is, distances from CR centers only or with a radius of 5.0 instead of 4.0 Å, but the conclusions obtained are basically the same as the above ones. Both the inner core dry residues and surface residues with significant increase in urea exposure (high sensitivity score) at the C1 stage are shown in Figure 4, with snapshots from 0 to 50 ns. When many of the surface residues are loosened from their initial compact structure quickly in the first 20 ns, the dry residues are closely packed together with little structural disruption in the same period (Figure 4a-c). Moreover, CR-Rg of these 13 dry residues maintains a near-native state for more than 40 ns (Figure 5a). In order to measure the packing of these dry residues quantitatively, the number of local contacts Ncontact, is calculated among these 13 dry residues. Similar to the definition in previous works,29,43 two residues i, and i + n (n g 3) are said to be in

contact, if the distance between their backbone CR-CR atoms is no greater than 10 Å. As shown in Figure 5b, Ncontact among these dry residues keeps roughly a constant number (around 22) until urea intrudes after 20 ns; however, they are still close to each other in the following time (20-40 ns) for their CR-Rg remains unchanged for ∼40 ns. As illustrated above, water penetrates after urea; therefore, the dry-globule is not destroyed during this period (Figure 2c,d). After the dry-globule dissolves, the local contact network is damaged quickly (Figure 5b). These results are also consistent with previous findings using our wavelet analysis43 and experimental NMR measurements27 (see more results in Supporting Information Figure S5, where the local contact numbers are also calculated for some representative residues from the six hydrophobic clusters identified in NMR experiments). We then plotted the Nw and Nu with time for both a representative dry residue (Met12) and a representative surface residue (Ile124) for further comparison in Figure 6 (see more in Supporting Information Figure S6). On the protein surface, both urea and water bind to the surface hydrophobic residue Ile24, but urea shows a more favorable binding than water as the water/urea ratio drops to almost 1.0 (Figure 6b; see more results in Supporting Information Figure S6). On the other hand, the dry inner core residue Met12 can stay dry for 40 ns, with

15692

J. Phys. Chem. B, Vol. 114, No. 47, 2010

Gao et al.

Figure 6. Nw (black solid line) and Nu (red dash line) around (a) a dry inner core residue Met12 and (b) a surface hydrophobic residue Ile124. Here, Nw and Nu are smoothed with with a 1 ns sliding window average. More figures of dry residues and surface residues with significant urea exposure increase are shown in the Supporting Information (Figure S6).

essentially no water molecules nearby, but meanwhile, 2-3 urea molecules approach it during this same 40 ns period. When protein continues to unfold, dry residues are then exposed to urea, but only after 40 ns, in this case, do their contacts with water increase, and the dry molten globule dissolves. This disappearance of the dry-globule then causes the further unfolding of the protein (see Figure 4e,f). Though there are much more water molecules than urea in the FSS of the protein, the urea molecules play a dominant role in the initial intrusion of the hydrophobic core during the early stage of unfolding. Discussion As mentioned above, our recent large-scale MD simulations support the direct mechanism in urea-induced lysozyme unfolding.15,29 These studies show that urea denaturizes protein though a two-stage kinetic process, first by forming a dry molten globule and then unfolding with both urea and water exposure. In addition, other case studies using both larger and smaller proteins, such as γ-crystallin, CI2, and ubiquitin, [Zhou, R.; Li, J.; Zhang, Z.; Bruce, B. J. J. Phys. Chem. B, under review], also show the same two-stage kinetic process during ureainduced protein denaturation. The dominant driving force is through the preferential binding of the protein to urea, with a stronger van der Waals dispersion interaction with urea than that with water.15 Here, by further analysis of the same simulation data set, we focus on contacts between each residue and its surrounding urea and water molecules and have obtained both structural and kinetic information on the protein denaturation process. During the initial unfolding process, some key hydrophobic inner residues keep dry from water (dry residues) for quite some time (20-50 ns). When they are exposed to solvent, they preferentially bind to urea molecules (during the initial 0-50 ns MD), which is the origin of dry-globule proposed in the previous studies.11,15 The well-packed dry residues possess a very stable local contact map which remains intact until urea molecules intrude the protein interior within ∼50 ns. Only after the local contact network is damaged does the dry-globule start to dissolve. As we know, during the protein folding process, packing of key residues at the hydrophobic core is an important step;18,26-28 here, similarly for the protein unfolding process, we find that there exist some key dry residues, such as Leu8, Met12, Val29, and Ala95, which play an important role in the formation of the dry-globule. The outside-in action was previously proposed based on MD simulations of a pair of methane molecules in urea.22 This mechanism suggests that urea is absorbed to charged residues on the protein surface, causing a swelling of the protein and

exposure of the hydrophobic core.11,15,22 Here, we find that some surface residues show significant increases of urea exposure (contacts with urea) in the early stage of unfolding (Figure 4 and Supporting Information Figure S2), and Nu increases around these surface residues ahead of the inner core residues (Figure 6 and Supporting Information Figure S6). However, different from the previous results by Wallqvist et al.,22 these key surface residues are mostly hydrophobic residues or noncharged hydrophilic residues. Our results are consistent with Stumpe et al.’s recent findings.12,21 A kinetic picture is then proposed to provide deeper insight into the outside-in mechanism with a two-stage process of urea-induced protein unfolding. Initially, urea molecules intrude the protein surface through accumulation around the hydrophobic and noncharged hydrophilic residues in the early stage of urea-induced protein unfolding. Gradually, core residues start to contact with urea but still keep dry from water. Such a period (20-50 ns for lysozyme in 8 M urea), during which urea starts intruding the protein but water is away from core residues, is the lifetime of the dry-globule. After this period, water enters the hydrophobic core following urea, which causes further unfolding of the protein. It should be noted that some previous studies have shown a water intrusion ahead of urea. Bennion and Dagget found that water rather than urea first solvates the hydrophobic core of chymotrypsin inhibitor 2 (CI2).13 One possible explanation is that CI2 is a smaller protein (64 residues) and also that its shape is more like a stretched ellipse; therefore, its hydrophobic core is more readily to be exposed to water. Obviously, it requires a certain core size to form a dry-globule with urea molecules captured inside due to the larger molecular size of urea. In contrast, protein lysozyme is larger (129 residues) and has many hydrophobic core residues attracting urea molecules, which can keep the core in a dry state for quite some time (Figure 3) until water intrudes into the protein interior eventually (Figure 6). A recent study44 shows that instead of strong interaction with hydrophobic groups, urea significantly alters the peptide secondary structure preferences by directly interacting with the peptide backbone and disrupting the secondary structure of the backbone and causing the swelling of the molecule. Whereas in our previous study15 protein backbones contribute about 0.9 kcal/ mol to the interaction energy gain for a urea molecule moving from the bulk to the protein surface, protein side chains contribute about 1.2 kcal/mol, slightly more than the backbones. Further understanding of the interaction between urea and the protein will benefit from the comparison of experimental and computational results, particularly the dynamical process of ureainduced protein denaturation.

Key Residues in Urea-Induced Lysozyme Unfolding Conclusions In this study, we have developed a simple sensitivity score (SS) based on the relative solvent exposure near each residue to examine and identify key residues involved in the early outside-in action, as well as the consequential dry-globule formation, during the initial stage of urea-induced lysozyme denaturation. Both urea and water in the first solvation shell (FSS) of each residue are analyzed in great detail, and a group of important inner dry residues (total 13) and surface-sensitive residues (total 12) are identified in protein lysozyme. These dry residues are found to be well-packed during the initial stage of unfolding, and only after their local networked contacts are interrupted by urea are they exposed to and solvated by water. This indicates that urea molecules intrude into the protein hydrophobic core ahead of water, during which a dry-globule6,15 is formed and lasts for nearly 50 ns. Also different from the previously proposed outside-in process,22 we find that urea molecules actually prefer to gather around noncharged residues instead of charged ones. These findings provide new insights into the molecular picture of the protein chemical denaturing process, and we believe our SS analysis should be applicable to other protein systems with or without chemical denaturants. Acknowledgment. The authors would like to thank Yongqi Huang for critical reading of the draft and Daqi Yu for technical suggestions. We would also like to thank Zhirong Liu, Payel Das, and Bruce Berne for helpful discussions. R.Z. acknowledges the support from the IBM BlueGene Science Program. Supporting Information Available: Additional experimental results. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Dill, K. A. Biochemistry 1990, 29, 7133. (2) Nozaki, Y.; Tanford, C. J. Biol. Chem. 1963, 238, 4074. (3) Wetlaufer, D. B.; Malik, S. K.; Stoller, L.; Coffin, R. L. J. Am. Chem. Soc. 1964, 86, 508. (4) Robinson, D. R.; Jencks, W. P. J. Am. Chem. Soc. 1965, 87, 2462. (5) Finer, E. G.; Franks, F.; Tait, M. J. J. Am. Chem. Soc. 1972, 94, 4424. (6) Jha, S. K.; Udgaonkar, J. B. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 12289. (7) Lim, W. K.; Rosgen, J.; Englander, S. W. Proc. Natl. Acad. Sci U.S.A. 2009, 106, 2595. (8) Frank, H. S.; Franks, F. J. Chem. Phys. 1968, 48, 4746. (9) Tirado-Rives, J.; Orozco, M.; Jorgensen, W. L. Biochemistry 1997, 36, 7313. (10) Caflisch, A.; Karplus, M. Struct. Fold. Des. 1999, 7, 477. (11) Mountain, R. D.; Thirumalai, D. J. Am. Chem. Soc. 2003, 125, 1950.

J. Phys. Chem. B, Vol. 114, No. 47, 2010 15693 (12) Stumpe, M. C.; Grubmu¨ller, H. PLoS Comput. Biol. 2008, 4, e1000221. (13) Bennion, B. J.; Daggett, V. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 5142. (14) Tobi, D.; Elber, R.; Thirumalai, D. Biopolymers 2003, 68, 359. (15) Hua, L.; Zhou, R.; Thirumalai, D.; Berne, B. J. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 16928. (16) Das, A.; Mukhopadhyay, C. J. Phys. Chem. B 2009, 113, 12816. (17) Das, P.; Zhou, R. J. Phys. Chem. B 2010, 114, 5427. (18) Gao, M.; Zhu, H. Q.; Yao, X. Q.; She, Z. S. Biochem. Biophys. Res. Commun. 2010, 392, 95. (19) Zangi, R.; Zhou, R.; Berne, B. J. J. Am. Chem. Soc. 2009, 131, 1535. (20) O’Brien, E. P.; Dima, R. I.; Brooks, B.; Thirumalai, D. J. Am. Chem. Soc. 2007, 129, 7346. (21) Stumpe, M. C.; Grubmu¨ller, H. Biophys. J. 2009, 96, 3744. (22) Wallqvist, A.; Covell, D. G.; Thirumalai, D. J. Am. Chem. Soc. 1998, 120, 427. (23) Stumpe, M. C.; Grubmu¨ller, H. J. Phys. Chem. B 2007, 111, 6220. (24) Stumpe, M. C.; Grubmu¨ller, H. J. Am. Chem. Soc. 2007, 129, 16126. (25) Finkelstein, A. V.; Shakhnovich, E. I. Biopolymers 1989, 28, 1681. (26) Fersht, A. R. Curr. Opin. Struct. Biol. 1997, 7, 3. (27) Vendruscolo, M.; Paci, E.; Dobson, C. M.; Karplus, M. Nature 2001, 409, 641. (28) Yao, X. Q.; She, Z. S. Biochem. Biophys. Res. Commun. 2008, 373, 64. (29) Zhou, R.; Eleftheriou, M.; Royyuru, A. K.; Berne, B. J. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 5824. (30) Zhou, R. H. J. Mol. Graphics Modell. 2004, 22, 451. (31) Liu, P.; Huang, X. H.; Zhou, R. H.; Berne, B. J. J. Phys. Chem. B 2006, 110, 19018. (32) Kang, S. G.; Saven, J. G. Curr. Opin. Chem. Biol. 2007, 11, 329. (33) Krone, M. G.; Hua, L.; Soto, P.; Zhou, R.; Berne, B. J.; Shea, J. E. J. Am. Chem. Soc. 2008, 130, 11066. (34) Zheng, L.; Chen, M.; Yang, W. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 20227. (35) Kamberaj, H.; van der Vaart, A. J. Chem. Phys. 2009, 130, 074906. (36) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kale´, L.; Schulten, K. J. Comput. Chem. 2005, 26, 1781. (37) Kumar, S.; Huang, C.; Zheng, G.; Bohm, E.; Bhatele, A.; Phillips, J. C.; Yu, H.; Kale´, L. V. IBM J. Res. DeV. 2008, 52, 177. (38) MacKerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; JosephMcCarthy, D.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, W. E.; Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.; Wio´rkiewicz-Kuczera, J.; Yin, D.; Karplus, M. J. Phys. Chem. B 1998, 102, 3586. (39) Neria, E.; Fischer, S.; Karplus, M. J. Chem. Phys. 1996, 105, 1902. (40) Zhou, R.; Berne, B. J.; Germain, R. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 14931. (41) Eleftheriou, M.; Germain, R. S.; Royyuru, A. K.; Zhou, R. H. J. Am. Chem. Soc. 2006, 128, 13388. (42) Klein-Seetharaman, J.; Oikawa, M.; Grimshaw, S. B.; Wirmer, J.; Duchardt, E.; Ueda, T.; Imoto, T.; Smith, L. J.; Dobson, C. M.; Schwalbe, H. Science 2002, 295, 1719. (43) Ye, L.; Wu, Z.; Eleftheriou, M.; Zhou, R. Biochem. Soc. Trans. 2007, 35, 1551. (44) Holthauzen, L. M. F.; o¨sgen, J.; Bolen, D. W. Biochemistry 2010, 49, 1310.

JP1052453