J. Phys. Chem. B 2009, 113, 131–137
131
Key Role of Proximal Water in Regulating Thermostable Proteins Fabio Sterpone,*,† Claudia Bertonati,‡ Giuseppe Briganti,§ and Simone Melchionna§ Caspur, Via dei Tizii 6B, 00185, Rome, Italy, and Department of Biochemical Sciences “Rossi Fanelli”, SOFT-INFM-CNR and Department of Physics, UniVersity of Rome La Sapienza, Ple. Aldo Moro 2, 00185, Rome, Italy ReceiVed: June 13, 2008; ReVised Manuscript ReceiVed: September 22, 2008
Three homologous proteins with mesophilic, thermophilic and hyperthermophilic character have been studied via molecular dynamics simulations at four different temperatures in order to investigate how water controls thermostability. The water-exposed surface of the protein is shown to increase with the degree of thermophilicity, and the role of water in enhancing the protein internal flexibility and structural robustness is elucidated. The presence of water-water hydrogen bond clusters enveloping the macromolecule is shown to correlate with thermal robustness when going from the mesophilic to the hyperthermophilic variants. Our analysis indicates that essential contributions to thermostability stem from protein-water surface effects whereas the protein internal packing plays a minor role. I. Introduction One of the unsolved questions in biophysics concerns the understanding of protein stability under extreme temperature conditions. Mesophilic organisms optimally grow between 20 and 40 °C while thermophilic and hyperthermophilic ones thrive at higher temperature intervals, 50-80 and 80-110 °C, respectively. Proteins from thermophilic and hyperthermophilic organisms have been evolutionarily selected in order to reach their optimal activity at the respective living temperatures.1 The activity of these proteins is strongly reduced below 40 °C. The physical basis of thermostability has been investigated by several authors (see among others refs 2-6). Early investigations have focused on the hydrophilic/hydrophobic content of thermophilic species and their structural features. A high content of hydrophilic amino acids and the ensuing surplus of salt bridges, has been considered the key ingredient of thermostability.2 For instance, the importance of salt bridges was clearly established studying the unfolding kinetics of Rubredoxins, where the hyperthermophilic variant has an unfolding rate 2 orders of magnitude slower than the mesophilic counterpart, and this difference reduces on increasing pH.3 Comparative analysis among several different species has shown that thermostable proteins display increased packing of hydrophobic residues.4,5 However, no significant secondary and tertiary structural patterns are commonly associated to thermostability.6 The reduced biological activity of thermophilic proteins at mesophilic conditions suggested a kinetic origin of thermostability, possibly related to an increased rigidity of the backbone, but up to now a strong experimental evidence of such speculations has not yet been provided. Calorimetric studies, probing the free energy landscape as a function of temperature, have revealed that meso and (hyper-)thermophilic counterparts exhibit maximal stability at ap* To whom correspondence should be addressed. E-mail: f.sterpone@ caspur.it. † Caspur. ‡ Department of Biochemical Sciences “Rossi Fanelli” University of Rome La Sapienza. § SOFT-INFM-CNR and Department of Physics, University of Rome La Sapienza.
proximately room temperature with a down shift in free-energy by 1-2 kcal/mol in favor of the (hyper-)thermophiles and a systematic broader and flatter free-energy basin (see e.g., ref 7 and references therein). The data have been initially interpreted as due to the presence of a residual structure in the unfolded state that, by preventing complete hydration of apolar amino acids, would lower the heat capacity of the thermophiles.8 However, this explanation is questionable due to the additional hydrophilic content of thermophiles. In fact, a positive heat capacity is a signature of solvation of hydrophobic amino acids while negative contributions arise from solvation of hydrophilic ones.9 Thus, the stronger polar character of thermophiles should lower the heat capacity and the persistence of residual secondary structure does not seem to explain unequivocally calorimetric data. An alternative interpretation of the differences in heat capacity emerges from recent NMR relaxation studies that have estimated the contribution of native-state conformational entropy to stability.10 The main observation is that enhanced internal protein motions may increase stability by increasing the entropy of the native state and raise the melting temperature. This would also increase the heat capacity of the native state and decrease the heat capacity difference between the native and the denatured states. Clearly, this indication suggests that the higher microscopic packing of the thermophiles has a marginal importance on thermostability.11 Recent studies have focused on the behavior of the folded state and the role played by internal fluctuations in modulating the flexibility of thermophiles, questioning the paradigm that thermostability is due to an enhanced conformational rigidity of the macromolecules. Neutron spectroscopy experiments have shown that the short-time motion of the R-amylase thermophile presents increased structural flexibility.12 NMR experiments have probed the millisecond time scale dynamics of Rubredoxin and concluded that the protein’s conformational flexibility of the mesophilic and hyperthermophilic homologues are very similar, while on the picosecond time scale the thermophile has increased mobility.15 Simulation studies have reported similar flexibility between meso and thermophilic homologues or slightly larger fluctuations of the latter, or in some cases, larger fluctuations at low temperatures.13 In the case of a class of laboratory evolved
10.1021/jp805199c CCC: $40.75 2009 American Chemical Society Published on Web 12/15/2008
132 J. Phys. Chem. B, Vol. 113, No. 1, 2009 enzymes, fluctuations of the thermophilic species are smaller at ambient temperature than the mesophilic counterpart and larger at higher temperature.14 It is worth recalling that the lack of consensus in the literature on the flexibility of thermostable proteins may arise from two different sources. At first, different experimental techniques, typically probing local or nonlocal motion, may observe different aspects of internal or surface flexibility of proteins in which case, a clear-cut definition of flexibility has to be clearly established. Second, specific thermal behaviors could be related to the enzyme’s function, some thermophiles manifesting smaller fluctuations, as for example, in the case of the rather rigid Rubredoxins, and others showing larger fluctuation, for example, for enzymes whose function is related to large scale structural rearrangements.13,14 In order to address thermostable behavior in a representative class of proteins, we have recently undertaken a systematic study of the thermal behavior of a protein domain, the guanosine triphosphate binding region of the EF-Tu for EC (mesophilic) and T (moderately thermophilic) variants.18,19 In our previous studies, we have shown that the two biopolymers have similar macromolecular packing while the water-exposed surface area is larger for the thermophile, where enhanced protein fluctuations are related to the molecular micromorphology and interaction of the thermophile with water. However, the isolated G domain from the thermophilic variant was only marginally more thermostable (by about 10 °C) than the mesophilic homologue,19-21 and the results could not be taken as definitive indicators of thermostable behavior. In the present work, we present a study of the EF-Tu hyperthermophilic variant from Sulfolobus-Solfataricus (SS) and comparatively analyze the thermal behavior of the meso (M), thermophilic (T), and hyperthermophilic (H) counterparts. The isolated G domain from the hyperthermophilic EF-Tu presents maximal GDP binding activity at T ) 84 °C slightly decreased with respect to the whole protein (T ) 94 °C).22 Thus, the present study focuses in particular on the H protein and enables us to investigate the general principles of thermostability. Anticipating our main results, we stress that the most remarkable difference observed among the three proteins upon increasing the temperature concerns the behavior of hydration water and the protein surface exposed to the solvent II. Methods A. MD Simulations. The MD simulations have been carried out using the DLPROTEIN package.23 We employed the Charmm22 force field to model the interactions for protein and solvent atoms,24 including the TIP3P model for water.25 The M species consists of the amino acid stretch T8-E203 of the original sequence (crystallographic Pdbcode 1EFC), and the G domain then was solvated by 2929 water molecules. The G domain from the thermophilic organism (T), Pdbcode 1EXM, encompasses the stretch of residues T8-W200 and was solvated by 2906 water molecules. Similarly, the H protein (Pdbcode 1SKQ) has been cut at the level of residues K4-V229 and solvated with 2747 water molecules in a cubic box. The total charge of the protein is zero, and no counterions are included in the system. The crystallographic structure of the hyperthermophilic G-domain of Sulfolobus-Sulfataricus has a stretch of missing residues (R66-F76) that we rebuilt using the homology modeling software Modeler26 with a multi-templates procedure. Using the HHpred server,27 the structure of the Yeast Elongation Factor complex (Pdbcode 1F60)28 was selected as the best template for the stretch of missing residues (1SKQ was used for the rest of the model). The global sequence identity between the two
Sterpone et al. structures (1F60 and 1SKQ) is 54% and specifically for the rebuilt stretch is about 60%. The simulation protocols for the M and T proteins have been described elsewhere;18,19 we recall here that equilibrated production runs have been performed for 4 ns or slightly longer. For the H protein, at first we have simulated the protein in the isothermal-isobaric ensemble for an equilibration time of 500 ps at pressure of 1 atm and at each of the temperatures T ) 300, 330, 360, and 390 K. After the equilibration, the cubic cell has linear dimension of L ) 48.1, 48.5, 48.7, and 49.4 Å at T ) 300, 330, 360, and 390 K, respectively. Subsequently, at each temperature we have produced extended simulations in the isothermal-isobaric ensemble for a duration of 8-10 ns. The MD time step was 2 fs with statistics collected every 400 fs. Electrostatics was computed via the Ewald method using the SPME algorithm (see ref 19 for details). The cutoff for the short-range part of the nonbonded interactions has been set to rc ) 9 Å. In the simulations, all the chemical bonds have been constrained to their equilibrium values via the SHAKE algorithm.29 The NPT sampling ensemble was performed by using the algorithm developed by Melchionna et al.30 as implemented in DLPROTEIN. B. Structural Analysis. The secondary structure content has been computed using the package STRIDE.31 In structural terms, the H protein shows very specific features with respect to the M and T counterparts. Actually, the structural superposition reveals three major differences: the H specie has two helices insertion (R1 [E35-L48] and R2 [E51-E65]) at the N-terminus, and a small helix insertion (c′ [E121-M126]) and a loop of 10 residues (T201-G212, Pdbcode 1SKQ) at the C-terminus. The T homologue has a similar loop but its location is not identical (N180-N191, Pdbcode 1EXM). As a consequence of such main differences and other minors, the sequence identity between the H protein and the M and T is =34%, while it is =74% between the M and T homologues. The hydropathy value, measured via the Kyte-Doolittle scale of amino acidic composition,32 gives values of -0.16, -0.32, and -0.43 for the M, T, and H proteins, respectively. Thus, not only does the H species have structural specificity but it also owns an enhanced hydrophilic character. C. Hydrogen Bond. The hydrogen bond (HB) is defined via the following geometrical criterion: in the case of two water molecules the distance between the two oxygens (donor and acceptor) is less than 3.5 Å and the oxygen-hydrogen-oxygen angle ranges in the [150-210°] interval. The criterion is extended to hydrogen bonds between protein-protein and protein-water partners. The number of protein-protein HB, HBPP, and protein-water HB, HBPW, are used to define the density of HB per protein unit volume and per protein unit surface, FVHB ) (HBPP + HBPW)/V, FSHB ) HBPW/S. The volume and surface per atom are evaluated employing the Voronoi construction. The space is partitioned into atom-based polyhedra, providing an accurate measure of the protein-occupied volume. On the other hand, the atom-based surface areas are evaluated by computing the contacts between protein-protein and protein-water atoms defined by the frontier planes separating neighboring pairs of atoms via the Voronoi tessellation. The volumetric data are obtained as averages over the last 2 ns of the trajectories upon verifying the stationarity of these quantities in this time window, as illustrated in Figure 1. Similarly H-bond counts have been averaged over the same segment of the trajectories. III. Results A. Volumetric Data. The average volume per atom of the three proteins and the relative temperature dependence is
Key Role of Proximal Water in Regulating Thermostable Proteins
Figure 1. Time evolution of the protein volume during the last 2 ns of the equilibrated trajectories at T ) 300 (solid line), 330 (dashed line), and 390 K (dot-dashed line). Horizonatal lines denote time averages. Inset: evolution during the full production trajectory.
Figure 2. Protein specific volume (upper panel), protein-protein exposed specific surface (mid panel) and protein-water exposed specific surface (lower panel) vs temperature for the mesophilic (open circles), thermophilic (filled squares) and hyperthermophilic homologues (gray diamonds). The adjective specific indicates that quantities were divided by the number of protein atoms.
reported in Figure 2, showing linear thermal expansion, with slope of 0.0040-0.0044 Å3 K-1 for all species. The relative values follow the ordering VM >VT > VH, in agreement with the accepted view of a closer packing of the thermophilic proteins, but the percentage variation going from M to H is rather small (sT >sH , whereas a reversed order is observed for the protein-water (PW) contact PW PW PW
