THE JOURNAL OF CHEMICAL PHYSICS 142, 055102 (2015)

Microscopic dynamics of water around unfolded structures of barstar at room temperature Somedatta Pal, Kaushik Chakraborty, Prabir Khatua, and Sanjoy Bandyopadhyaya) Molecular Modeling Laboratory, Department of Chemistry, Indian Institute of Technology, Kharagpur 721302, India

(Received 9 August 2014; accepted 16 January 2015; published online 5 February 2015) The breaking of the native structure of a protein and its influences on the dynamic response of the surrounding solvent is an important issue in protein folding. In this work, we have carried out atomistic molecular dynamics simulations to unfold the protein barstar at two different temperatures (400 K and 450 K). The two unfolded forms obtained at such high temperatures are further studied at room temperature to explore the effects of nonuniform unfolding of the protein secondary structures along two different pathways on the microscopic dynamical properties of the surface water molecules. It is demonstrated that though the structural transition of the protein in general results in less restricted water motions around its segments, but there are evidences of formation of new conformational motifs upon unfolding with increasingly confined environment around them, thereby resulting in further restricted water mobility in their hydration layers. Moreover, it is noticed that the effects of nonuniform unfolding of the protein segments on the relaxation times of the protein–water (PW) and the water–water (WW) hydrogen bonds are correlated with hindered hydration water motions. However, the kinetics of breaking and reformation of such hydrogen bonds are found to be influenced differently at the interface. It is observed that while the effects of unfolding on the PW hydrogen bond kinetics seem to be minimum, but the kinetics involving the WW hydrogen bonds around the protein segments exhibit noticeably heterogeneous characteristics. We believe that this is an important observation, which can provide valuable insights on the origin of heterogeneous influence of unfolding of a protein on the microscopic properties of its hydration water. C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4907007]

I. INTRODUCTION

Protein folding is a spontaneous process of assembling a polypeptide chain into a distinct three-dimensional structure.1,2 On the other hand, unfolding is the competing reverse process that normally leads to denaturation of the protein with a consequent loss of function.1 It is believed that a protein folds by exploring pathways in a multidimensional free energy landscape.3–6 The wide opening of the landscape represents an ideal state of the protein characterized by high degree of conformational entropy and relatively high free energy. The landscape can be rugged in nature with several local minima separated by saddle points. There is a free energy gradient that directs the protein toward the global minimum (lowest point) that corresponds to its native state. The local minima corresponds to different intermediates along the pathway that can slow down the folding process by trapping the polypeptide molecule in those minima. Such intermediates may have long lifetimes with a significant fraction of native contacts and secondary structures, and are often referred to as the molten globule (MG) state.7 It is believed that trapping of the protein in such misfolded states is responsible for diseases like cystic fibrosis and mad cow.1

a)Author to whom correspondence should be addressed. Electronic mail:

[email protected] 0021-9606/2015/142(5)/055102/14/$30.00

Considering the importance of protein folding-unfolding and its biological significance, it has been one of the most active areas of research for many years. Several interesting aspects of the problem have been understood in recent years from experimental, analytical, and computer simulation studies. However, the role played by water in controlling the structure and dynamics of the intermediates and hence, the folding pathways have not been explored much. It is essential to study protein-water interactions and the effects of protein conformations on the microscopic properties of the surrounding solvent for complete understanding of protein folding. Here, we restrict our discussion on the limited attempts made in the literature in this direction. Nuclear magnetic resonance (NMR) spectroscopy is one of the most powerful tools to study the structural details of partially folded or unfolded proteins, as well as the dynamical environment of the solvent around those.8–10 For example, in a recent study, Halle and co-workers11 used 1H and 17O magnetic relaxation data and showed that the hydration water around the unfolded forms of halophilic proteins suffers weaker dynamical perturbation as compared to the folded forms. Timeresolved fluorescence spectroscopy has been extensively used over past few years to explore the solvation dynamics (SD) of partially folded or unfolded proteins.12–14 It is demonstrated that different partially unfolded states of a protein can exhibit widely different solvation characteristics.12 Such studies showed that the SD of the unfolded state of a protein can

142, 055102-1

© 2015 AIP Publishing LLC

055102-2

Pal et al.

be faster or slower than that of the native state depending on the nature of the probe and its location.13 Using tryptophan as an intrinsic probe, Zhong and co-workers15 showed the hydration dynamics of partially unfolded apoMb protein to be faster compared to the native form. Since the seminal work by Duan and Kollman,16 different aspects of protein folding-unfolding have been probed from computer simulation studies. Specific attempts have been made to explore the role of water in guiding protein folding from minimalistic model-based calculations and MD simulations. Harano and Kinoshita17 showed from theoretical calculations that the translational entropy gained by water during its expulsion from the core can be a powerful driving force in protein folding. The importance of water-mediated long-range interactions in protein folding has been demonstrated from MD simulations.18 The correlation between the differential unfolding pattern of a protein and the dynamic behavior of the surrounding solvent has been revealed from simulation studies.19 In an important contribution, the hydration dynamics of a partially denatured form of the protein α-lactalbumin have been probed by Tobias and co-workers.20 Relatively faster dynamics of the surface water molecules around the unfolded protein were noticed. Marchi and co-workers21 recently investigated the effect of water content on unfolding of cytochrome c in a confined environment. They observed that unfolding of the protein is favored by low confinement and high water content. Udgaonkar and co-workers22–24 carried out detailed experiments with barstar to explore different intermediate molten globule structures of the protein along multiple unfolding pathways using different denaturants and also by varying temperature. They showed that the degree of expansion of the protein depends on the denaturant used and its concentration. Structure and dynamics of barstar have also been studied from MD simulations. Existence of a fluidlike hydrophobic core of the protein was revealed from an early simulation study by Wong and Daggett.25 Recently, we carried out extensive MD simulations to probe the influence of the structural and energetic heterogeneites at the surface of barstar on the local structure, ordering, and dynamics of the nearby water molecules.26–28 Besides, we have also studied unfolding of barstar at high temperatures (400 K and 450 K).29 The unfolded structures obtained at such high temperatures were studied around room temperature to probe the effects of heterogeneous unfolding of the protein segments on microscopic structure and ordering of the surrounding solvent. In this work, we study room temperature microscopic dynamics and hydrogen bond properties of water molecules hydrating different segments of the protein in the unfolded forms. To examine how the unfolding of the protein along two different pathways affect the hydration water dynamics, the results are compared with that corresponding to the folded native barstar under similar conditions. The rest of the article is organized as follows. In Sec. II, we provide a brief description of the setup of the systems and the simulation methods employed. The results obtained from our investigations are presented and discussed in Sec. III. In Sec. IV, we summarize in brief the important findings and the conclusions reached from this study.

J. Chem. Phys. 142, 055102 (2015)

II. SYSTEM SETUP AND SIMULATION METHODS

Barstar formed by the bacterium bacillus amyloliquefacien is an intracellular inhibitor of the ribonuclease barnase.30,31 It contains 89 amino acid residues with secondary structures consisting of three parallel α-helices and a threestranded parallel β-sheet. Additionally, there is another αhelix that connects the second central β-strand and the third parallel α-helix. The amino acid sequence of the protein is K(1) KAVINGEQIRSISDLHQTLKKELALPEYYGENLDALWD CLTGWVEYPLVLEWRQFEQSKQLTENGAESVLQVFRE AKAEGCDITIILS(89), with the N-terminus residue K(1) and the C-terminus residue S(89). For convenience of our discussion, we identify these secondary structures as helix-1 (Ser-14 to Ala-25), helix-2 (Asn-33 to Gly-43), helix-3 (Phe-56 to Thr63), helix-4 (Glu-68 to Gly-81), and β-sheet (Lys-1 to Asn6, Leu-49 to Arg-54, and Asp-83 to Ser-89). The loops that interconnect these segments are denoted as loop-1 (Gly-7 to Ile-13), loop-2 (Leu-26 to Glu-32), loop-3 (Trp-44 to Pro-48), and loop-4 (Glu-64 to Ala-67). We have carried out three different simulations with the protein barstar in aqueous medium using the NAMD code.32 The folded native protein around room temperature was studied in simulation S1, while the unfolding studies were carried out in simulations U1 and U2 at two different temperatures. The initial coordinates of the protein for simulation S1 were taken from the NMR structure (PDB code: 1BTA).30 Further details of the setup of the solvated S1 system and the simulation protocols followed have been mentioned in our earlier work.26 The protein conformation as obtained after energy minimization of system S1 was taken to initiate the temperatureinduced unfolding simulations, U1 and U2. NPT simulations with isotropic cell fluctuations were carried out for 150 ns each at 400 K (U1) and 450 K (U2), respectively. The protein molecule unfolded during this period in both cases. Next, the two systems were cooled down to room temperature, and 75 ns simulations were carried out for each following the same protocols as described in detail earlier.29 We have employed the all-atom CHARMM22 force field with CMAP corrections for proteins33,34 and TIP3P model35 for water in the calculations. The integration time step used was 1 fs and the trajectories were stored at 500 fs time intervals. The “minimum image convention”36 was used and the short-range Lennard-Jones interactions were calculated with a spherical cutoff distance of 12 Å and a switch distance of 10 Å. The particle-mesh Ewald (PME) method37 was used to calculate the long-range electrostatic interactions.

III. RESULTS AND DISCUSSION

Recently, we studied in detail the unfolding of the protein barstar at high temperatures of 400 K and 450 K in simulations U1 and U2, respectively.29 The calculations at 450 K revealed transformation of the whole protein into a flexible expanded conformation due to breaking of almost all the secondary structures, whereas partial unfolding of only two α-helices (helix-1 and helix-2) and their interconnecting loop occurred at 400 K. In Fig. 1, we show representative configurations of the protein in the folded native (S1) and in the two unfolded forms (U1 and

055102-3

Pal et al.

J. Chem. Phys. 142, 055102 (2015)

FIG. 1. Representative configurations of the protein in its folded native and the two unfolded forms as obtained from simulations (a) S1, (b) U1, and (c) U2. The native form of the protein contains four α-helices (drawn in red) packed onto a three-stranded parallel β-sheet (drawn in blue). The interconnecting loops are shown in green. For clarity, the water molecules that are bound to the surface of the unfolded form U2 by hydrogen bonds are shown in the figure.

U2) as obtained from our simulations. For easy reference, the locations of different secondary structural segments (α-helices, β-sheet, and loops) are marked in the native configuration. For clarity, we have shown the water molecules that are bound to the surface of the unfolded form in U2 by hydrogen bonds. We will discuss further the properties of such bound water molecules later. The two unfolded forms of the protein obtained under two different conditions were then studied around room temperature to probe the effects of heterogeneous unfolding of the protein segments on microscopic structure and ordering of the surrounding solvent. It was shown that the room temperature distributions of water molecules present in the hydration layers around the two unfolded forms are different from that around the native structure. In general, unfolding of the protein segments was found to alter the water distribution and hence the local environment around the segments with increasingly hydrated relatively more ordered interface. The heterogeneous unfolding patterns of the protein in simulations U1 and U2 are expected to influence the dynamical nature of the water molecules present in the hydration layer spanning the protein surface. In this work, we study the room temperature microscopic dynamics and hydrogen bond properties of water molecules hydrating different segments of the protein in the unfolded forms. To examine how the unfolding of the protein along two different pathways affect the hydration water dynamics, the results are compared with that corresponding to the folded native barstar under similar conditions. The calculations are based on the last 50 ns of the NVE trajectories obtained from the three simulations under similar conditions around room temperature (see Sec. II). Further, in consistent with our earlier study,27 only those water molecules that are present within a distance of 5 Å from the protein surface are included in the calculations. This essentially corresponds to the water molecules present in the first hydration layer whose dynamical properties are primarily influenced by a protein.38 It may be noted here that the re-entries of the tagged water molecules

within the first hydration layer of the protein over times have been taken into account in all the calculations. In other words, only the durations of the trajectory of a tagged water molecule when it is found to reside within 5 Å of a protein atom are included in the calculations. The errors involved in the results presented have been estimated from the standard deviations of the room temperature data as obtained by splitting each of the three 50 ns long equilibrated trajectories into 20 blocks, where each block is of 2 ns duration with a separation of 500 ps between two successive blocks. A. Water translational motion

We first probe the room temperature translational motions of water molecules hydrating different segments of barstar as defined in Sec. II with respect to its native structure (αhelices, β-sheet, and interconnecting loops) in the two unfolded forms as obtained from simulations U1 and U2. This is done by measuring their mean square displacements (MSD), ⟨∆r 2⟩, with respect to time. MSD is defined as ⟨∆r 2⟩ = ⟨|ri (t) − ri (0)|2⟩,

(1)

where, ri (t) and ri (0) are the position vectors of the ith water oxygen atom at time t and at t = 0, respectively. For each of the protein segments, the calculations are carried out by averaging over all the tagged water molecules and over different time origins. The results are shown in Fig. 2. The corresponding results around the secondary structures of the native form as obtained from simulation S1 under identical conditions are also shown in the figure. As a reference, MSD of pure bulk water is included in the figure. The bulk water data have been obtained from a separate MD simulation of TIP3P water following the same protocol as described in Sec. II under NPT ensemble conditions at room temperature (300 K). First, the results show that irrespective of the conformational state of the protein,

055102-4

Pal et al.

J. Chem. Phys. 142, 055102 (2015)

FIG. 2. MSD of water molecules present in the first hydration layers of different segments of barstar around room temperature for (a) the folded native form (simulation S1) and ((b) and (c)) the two unfolded structures as obtained from the last 50 ns of the simulations U1 and U2. As a reference, the MSD of pure bulk water at room temperature is included in the figure.

the hydration water molecules exhibit retarded translational motions as compared to pure bulk water. Such restricted water motions at protein surfaces agree well with earlier simulation studies.39–41 However, the present result shows interesting dynamical behavior of water around the unfolded protein segments. In general, due to increased flexibility of the unfolded structures and their greater exposures to solvent,29 one would expect water molecules hydrating the protein to exhibit relatively faster and perhaps more homogeneous dynamics in systems U1 and U2 as compared to that in system S1. However, a closer examination of Fig. 2 reveals that it is not necessarily the case around all the segments. In addition, we notice that heterogeneous restrictions in water translations around the protein segments exist even in the unfolded forms, though the relative degree of heterogeneity often differs. A comparison between Figs. 2(a) and 2(b) shows that with respect to the native structure, water molecules around most of the protein segments (except for loop-3) in the unfolded form in U1 exhibit either similar or to some extent faster mobility. However, the water molecules around the segments of the unfolded protein in U2 (Fig. 2(c)) exhibit interesting features. It is observed that the water molecules around the unfolded segments that correspond to helix-1 and the β-strands are noticeably more mobile than the water molecules around those in S1 and U1. As already mentioned, such relatively faster water motions around these segments correlate well with their increased degree of unfolding and greater solvent exposure.29 In contrast, relatively slower water motions have been observed around the other unfolded helical segments in U2 with respect to the water molecules hydrating the corresponding segments in U1. Water mobility around the protein loops in U2 is even more intriguing. It can be seen that while unfolding of loops 1 and 3 results in noticeably faster water translations around those in U2 as compared to the native and partially unfolded protein structures in S1 and U1, but water motions reduce drastically around loop-4. We have calculated the statistical errors from the standard deviations of the data corresponding to different segments of the protein in its folded and unfolded forms as obtained from the simulated trajectories. It is found that the estimated errors are rather

small and vary within 1%–2%. This suggests that the results reported on water motions around the protein segments are reliable. The hydrophilicity (or hydrophobicity) of a protein segment is expected to play an important role in governing the dynamical behavior of the nearby water molecules. To explore the origin of the counterintuitive water dynamics around loop4 in the unfolded form U2, we have calculated the average per residue hydropathy (H R ) values for different segments using the amino acid residue hydropathy scale.42 The values are listed in Table I. It can be seen that the segments corresponding to the helices 1, 3, and 4, and the four loops are marginally hydrophilic with their H R values varying between −0.36 (helix-4) and −1.4 (loop-4). On the other hand, the remaining segments comprising of the helix-2 and the three-stranded βsheet are found to be marginally hydrophobic with H R values of 0.29 (helix-2) and 0.76 ( β-sheet), respectively. Thus, the H R values do not reveal significant difference between the hydrophilic nature of loop-4 and that of the other segments. This shows that the highly retarded loop-4 hydration water motion in U2 not necessarily arises due to stronger interaction between its residues and water. Rather, it is apparent that such behavior originates from increased geometrical constraints around the loop-4 segment in the unfolded forms. This is TABLE I. The average per residue hydropathy (H R ) values for different segments of barstar as obtained using the amino acid residue hydropathy scale.42 Segment

HR

Helix-1 Helix-2 Helix-3 Helix-4 β-sheet Loop-1 Loop-2 Loop-3 Loop-4

−0.82 0.29 −1.16 −0.36 0.76 −0.53 −1.11 −0.62 −1.40

055102-5

Pal et al.

J. Chem. Phys. 142, 055102 (2015)

consistent with our earlier findings of reduced exposure of loop-4 residues to water in the unfolded form U2.29 We have calculated the diffusion coefficients (D E ) of water molecules hydrating the protein segments in the native (S1) and the two unfolded forms (U1 and U2) as obtained from the slopes of the corresponding MSD curves using the Einstein’s formulation36 ⟨∆r 2⟩ , ∆t→ ∞ 2d∆t

D E = lim

(2)

where d is the dimensionality of the system. The results are listed in Table II along with that for pure bulk water for comparison. It can be seen that compared to water in bulk state, the diffusion coefficients of water molecules around the native form in S1 are 4–7 times lower, whereas the corresponding values are 4–5 and 3–5 times lower around most of the segments in the two unfolded forms U1 and U2 (except around loop-4 in U2), respectively. Note that the diffusion time scale of water molecules confined around loop-4 in U2 is an order slower than the bulk water diffusion time. Besides, comparison of the D E values around the native and the two unfolded forms reveals increase in diffusivities of hydration water molecules with unfolding around most of the protein segments. However, reverse is the case for water molecules around loop-3 in U1 and around loop-4 in U2. In addition, comparing the data between the two unfolded forms (U1 and U2), we find that the increasing extent of unfolding in U2 has a non-uniform effect on the surrounding water molecules. Instead of expected increase in hydration water diffusivities with degree of unfolding, water molecules around several segments (helices 2 and 3, and loop4) exhibit noticeably lower D E values. These are interesting observations which demonstrate that the unfolding of a protein instead of relaxing the degree of confinement around its segments may actually increase it due to local conformational modifications of the protein. One should note at this stage that water diffusivity at the surface of a complex biomolecule like protein in general is sublinear.39 Therefore, the relative differences between the D E values as discussed here are more meaningful than the corresponding absolute values. We have TABLE II. The translational diffusion coefficients (D E ) and average reorientational times (⟨τ µ ⟩) for water molecules present in the first hydration layers around different segments of barstar in the native (S1) and in the two unfolded forms (U1 and U2) around room temperature. The corresponding values for water in pure bulk state are listed for comparison. D E (10−5 cm2 s−1)

⟨τ µ ⟩ (ps)

Segment

S1

U1

U2

S1

U1

U2

Helix-1 Helix-2 Helix-3 Helix-4 β-sheet Loop-1 Loop-2 Loop-3 Loop-4

0.77 0.83 1.14 1.11 0.96 0.72 1.05 1.13 0.80

0.96 1.20 1.28 1.15 0.95 1.15 1.16 0.98 1.18

1.03 0.90 1.16 1.25 1.30 1.48 1.15 1.95 0.40

12.90 10.68 12.78 10.92 14.75 16.16 8.64 12.24 14.44

14.28 10.05 10.56 10.10 14.05 10.36 11.79 13.49 11.00

12.04 14.01 12.10 11.82 11.11 8.12 12.45 7.57 32.74

Bulk water

4.8

2.03

probed the extent of such sublinearity by fitting the MSD curves for the three systems with a power law as ⟨∆r 2⟩ ∼ t α ,

(3)

where the exponent α is expected to be less than unity for sublinear diffusion. The average α values for the hydration water molecules around the protein segments in the native and the two unfolded forms are found to vary within 0.5 to 0.7. This demonstrates that the degree of anomaly in water diffusion can be nonuniform around different segments of a protein irrespective of its conformational state. To further understand the heterogeneous influence of unfolding on water translational motions, we have probed the space-time correlation of water molecules around the protein segments in the native and the unfolded forms. This is done by calculating the time evolutions of the van Hove autocorrelation function G s (r,t) for the water molecules that were present initially (t = 0) in the first hydration layers of the protein segments. G s (r,t) for a N particle system is defined as43 N

G s (r,t) =

1 ⟨δ[r + ri (0) − ri (t)]⟩, N i=1

(4)

where ri (0) and ri (t) denote ith particle positions at times t = 0 and t, respectively. According to the definition, G s (r,t) provides an estimate of the probability density of finding a particle at time t, given that the particle was at the origin at t = 0. We have calculated the probability of a water molecule present in the first hydration layer at time t = 0 to have moved a distance r after 5, 20, and 50 ps, respectively. The results as obtained for the three systems are shown in Fig. 3. As a reference, the corresponding data for pure bulk water are shown in the inset of the figure. Restricted dynamic environment around the protein conformations at short times is once again apparent from the figure. Note that the time evolution of the distribution for water molecules around loop-4 in U2 is strikingly different from that around the other segments. Significantly narrower widths and peak heights for water near the loop-4 segment in U2 are signatures of highly constrained hydration environment around the conformational motif formed around it during unfolding transition in U2. B. Water rotational motion

In this section, we study the rotational motions of hydration water molecules around different segments (within 5 Å) of barstar in the two unfolded forms (U1 and U2), and compare the results with that obtained for the native protein (S1). The calculations are carried out by measuring the reorientational dynamics of water dipole, ⃗µ, defined as the vector connecting the oxygen atom of the tagged water to the center of the line joining its two hydrogens. The time evolution of ⃗µ is then monitored by calculating the dipole–dipole time correlation function (TCF), Cµ (t), defined as Cµ (t) =

⟨ µˆ i (t). µˆ i (0)⟩ . ⟨ µˆ i (0). µˆ i (0)⟩

(5)

Here, µˆ i (t) is the unit dipole moment vector of the ith tagged water at time t, and the angular brackets denote that the

055102-6

Pal et al.

J. Chem. Phys. 142, 055102 (2015)

FIG. 3. van Hove autocorrelation function (G s (r, t)) for water molecules present in the first hydration layers of different segments of barstar at t = 0 around room temperature for (a) the folded native form (simulation S1) and ((b) and (c)) the two unfolded structures as obtained from the last 50 ns of the simulations U1 and U2. The evolution of the distribution at t = 5 ps (solid line), 20 ps (dotted line), and 50 ps (dashed line) is shown in each panel. As a reference, the corresponding data for pure bulk water are included in the inset.

calculations are averaged over the tagged water molecules at different reference initial times. The results for the three systems are displayed in Fig. 4. Once again, the result for pure bulk water is included in the figure for comparison. Significantly, slower relaxations of Cµ (t) for water molecules surrounding the protein segments as compared to pure bulk water indicate highly restricted water reorientations near the protein surface. Consistent with water translational motions, their restricted rotations too are independent of whether the protein is in its native form or in the unfolded forms. Comparing the results with that shown in Fig. 2, it is apparent that for each of the three systems, the relative heterogeneity in water rotations around the protein segments is quite similar to that observed for the corresponding translational motions. This shows that the translational and rotational motions of water molecules present close to the protein surface are in general uniformly influenced by the unfolding of the protein. In particular, note the drastically restricted water rotations around the unfolded loop-4 in U2. Along with our earlier study,29 it is now clear that unfolding of the protein in U2 resulted in formation of a few local conformational motifs with curved interfaces that resulted in the corresponding hydration

water molecules trapped in confined environments and thereby exhibiting highly restricted local motions. Once again, the errors involved in the results shown in Fig. 4 are found to be insignificant (

Microscopic dynamics of water around unfolded structures of barstar at room temperature.

The breaking of the native structure of a protein and its influences on the dynamic response of the surrounding solvent is an important issue in prote...
2MB Sizes 0 Downloads 4 Views