Mechanics of Protein Adaptation to High Temperatures - The Journal

Nov 20, 2017 - Inspired by Somero's corresponding state principle that relates protein enhanced thermal stability with mechanical rigidity, we deploye...
2 downloads 10 Views 4MB Size
Letter pubs.acs.org/JPCL

Cite This: J. Phys. Chem. Lett. 2017, 8, 5884-5890

Mechanics of Protein Adaptation to High Temperatures Guillaume Stirnemann* and Fabio Sterpone* CNRS Laboratoire de Biochimie Théorique, Institut de Biologie Physico-Chimique, Université Paris Denis Diderot, Sorbonne Paris Cité, PSL Research University, 13 rue Pierre et Marie Curie, 75005, Paris, France S Supporting Information *

ABSTRACT: Inspired by Somero’s corresponding state principle that relates protein enhanced thermal stability with mechanical rigidity, we deployed state of the art computational techniques (based on atomistic steered molecular dynamics and Hamiltonian-replica exchange simulations) to study the in silico realization of mechanical and thermal unfolding of two homologous Csp proteins that have evolved to thrive in different thermal environments. By complementing recent single-molecule experiments, we unambiguously show that, for these homologues whose structures are very similar, the increased thermal resistance of the thermophilic variant is not associated with an increased mechanical stability. Our approach provides microscopic insights that are otherwise inaccessible to experimental techniques, and explains why the protein weak spots for thermal and mechanical denaturation are distinct.

E

and the other one from the thermophilic Thermotoga maritima29 (Figure 1b). Our strategy uses Hamiltonian-replica exchange simulations based on the REST230,31 methodology that, for the first time, accesses the needed time scales to assess the thermal stability of a protein of that size, together with steered-molecular dynamics (SMD) simulations32 in order to inquire mechanical stability. The thermal stability of both homologues has been quantified for two different force fields. In the REST2 simulations, 16 replicas were propagated for 400 ns each. They all evolved at the same physical temperature, but they could exchange their Hamiltonian with a rescaled potential energy. When compared to standard Replica Exchange MD,37 REST2 can use a fewer number of replica and still achieve high exchange rates (close to 29−35% here). Indeed, the conformational sampling is sped up as only intraprotein and protein− water interactions are affected by the scaling procedure. However, recovering a physical temperature scale from REST2 sampling is not trivial as protein−protein and protein−water interactions bear different and nonlinear scalings. We have recently suggested a method to assign a corresponding effective temperature to any given replica evolving under a rescaled potential energy, which was successfully validated on simple protein systems.31 The idea is to exploit the corresponding state principle so that the density probability distribution ρ(β, Ũ λ) sampled via the rescaled energy potential Ũ λ at the reference temperature β matches the density probability distribution ρ(βeff, Ũ ) associated with the unscaled potential energy Ũ but at an

nzymes from thermophilic organisms are stable and functional at very high temperatures, where the vast majority of proteins would be unfolded, but they are in general marginally active under ambient conditions.1 Somero’s corresponding state principle2 explains this effect in terms of a more rigid protein matrix that only gets activated and functional at high temperatures.3 While seminal investigations have provided strong support of this picture4,5 with tentative links between protein conformational flexibility and enzymatic catalysis,6 other studies questioned its universal validity.7,8 Thus, the relationship between protein dynamics, conformational flexibility and melting is still an open debate.9,10 In particular, enhanced flexibility is a viable alternative route to gain stability at high temperature,11 and the role of mechanical stiffness must be nuanced in terms of the protein states (apo/ holo), functional modes,12 allosteric response13,14 and kinetic barrier for the enzymatic chemical step.15 Single-molecule force-spectroscopy techniques16 are natural tools to probe protein stiffness, and thus to study folding/ unfolding mechanisms and responseresistance and softnessto a directional pulling force. While mechanical and thermal unfolding are very different,17 and result in distinct unfolding state ensembles,18−20 force-spectroscopies are extensively used to study many aspects of protein folding/ unfolding (see, e.g., refs 21−25). Despite these known differences, it is unclear, and yet unexplored at a molecular level, whether thermophilic proteins could gain thermal resistance via a correlated enhanced mechanical stability when compared to their mesophilic homologues that are structurally and sequentially similar. The present work, inspired by recent force-spectroscopy experiments on homologous cold-shock proteins (Csp),26,27 fills the gap. To this purpose, we have deployed an advanced computational scheme to study two Csp proteins, one from the mesophilic Escherichia coli28 (Figure 1a), © XXXX American Chemical Society

Received: October 3, 2017 Accepted: November 17, 2017

5884

DOI: 10.1021/acs.jpclett.7b02611 J. Phys. Chem. Lett. 2017, 8, 5884−5890

Letter

The Journal of Physical Chemistry Letters

Figure 1. EcCsp and TmCsp are structurally very similar. Csp homologues from (a) EcCsp (PDB 1MJC) and (b) TmCsp (PDB 1G6P) fold into a compact β-barrel structure made of 5 β-strands numbered β1...β5 (yellow) starting from the N-terminal. These are connected with loops and turns (light blue and white), and EcCsp also exhibits a small α-helix (dark blue). (c) Side views of the structure alignment of EcCsp (blue) and TmCsp (red) showing that both homologues are very similar folds.33 However, members of the Csp family exhibit radically different thermal stabilities: their melting temperatures span a ≈35 K temperature range.33−36

Figure 2. EcCsp unfolds at lower temperatures as compared to TmCsp. (a) Distributions of the RMSD (with respect to the center of the most populated cluster around ambient temperature) for each replica effective temperature for EcCsp (top) and TmCsp (bottom), using the CHARMM36 force field. The color bar corresponds to the probability of finding a protein conformation within a bin of 0.4 Å. (b) Fraction of native contacts as a function of effective temperature for EcCsp (blue) and TmCsp (red). Standard deviations are shown as error bars. (c) Fraction of folded protein conformations, for which the RMSD of the protein rigid β-core (β1···5 strands), and of the small helix for EcCsp, is lower than 2 Å (blue dots: EcCsp and red dots: TmCsp). Dashed lines are the results of a fit using the stability formula (cf. Supporting Information), leading to the respective melting temperatures of 342 K for EcCsp, and 374 K for TmCsp. These values are shifted only by ≈12−15 K as compared to the experimental values, and the experimental 27 K stability shift is very well reproduced.

effective temperature βeff. In Ũ , the solvent−solvent interactions are excluded and considered to generate the same contribution for all the replicas31 (see Supporting Information). Although the simulation convergence for these “big” systems is difficult to assess as it would require checking the evolution of much longer time scales, the analysis of the last half of the simulations suggests that protein average relevant quantities do not significantly change on this time window (see Figure SI1), after an initial transition regime. The subsequent analysis is based on these data. As temperature increases, the protein explores conformations that are further away from the folded state. This is seen in the variations of the backbone Cα root-mean square displacement (RMSD), as illustrated in Figure 2a for the CHARMM36 force field (Figure SI2 for the Amber99sb* force field). For both homologues, the large values of the RMSD at high temperatures suggest that the protein visits largely unfolded conformations. However, the temperature where the transition between RMSD values typical of folded and unfolded conformations occurs is different for the two homologues, with TmCsp being the most stable upon thermal denaturation,

as expected. This is confirmed by examining the average smoothed fraction of native contacts Q (Figure 2b) as a function of temperature. TmCsp clearly denatures at higher temperatures as compared to EcCsp. The same trend is observed when using Amber99sb*, even if the transitions are shifted toward higher temperatures (Figure SI2). Protein thermal stability is more directly quantified by reconstructing the melting curves for the two proteins by 5885

DOI: 10.1021/acs.jpclett.7b02611 J. Phys. Chem. Lett. 2017, 8, 5884−5890

Letter

The Journal of Physical Chemistry Letters

Figure 3. Conformational clustering suggests that EcCsp and TmCsp exhibit analogous flexibilities at room temperature. Superposition of the centers of the three most populated clusters of (a) EcCsp (b) TmCsp, together with the corresponding network representations of the identified structural clusters. The position of the L3 loop is highlighted. In the network representation, the size of the circle depends on the occupation of the associated conformational state while the edges indicate the monitored transitions between states. (c) Residue root-mean-square fluctuations (RMSF) of EcCsp (blue) and TmCsp (red) along the aligned protein sequences. The secondary structure elements found in the protein crystal/ NMR structures are also shown below (the gray crosses before β1 and in β3 represent missing residues when performing the sequence alignment).

conditions. This is further confirmed by the molecular representations of the most populated states (Figure 3a,b). Finally, regardless of the Csp homologue or of the employed force field and despite the noise, the number of conformations exhibits almost the same scaling when plotted against the reduced temperature T/Tm (Figure SI5), suggesting that a common configurational flexibility is attained approaching the respective melting.9 In some sense the thermophilic Csp, which exhibits higher Tm, is more resilient to temperature increase, as suggested by earlier neutron scattering investigations.8 When focusing on local fragments of the protein matrix, we note that the L3 loop connecting β3 and β4 is significantly more mobile in TmCsp (Figure 3c and Figure SI6). This seems to confirm the idea that enhanced stability of TmCsp can be related to a higher entropy of the folded state mobile regions, as suggested by NMR.38 As we shall see below, the L3 loop is involved in the mechanical stability of Csp measured in the simulations. At the same time, the simulations also confirm the existence of a much greater number of salt-bridges between the β-sheets of TmCsp as compared to EcCsp, mostly involving ARG2, LYS6, ASP24, GLU33, GLU47 and LYS63. These residues form a dynamic network of ion-pairs cross-linking the protein core.29 Several experimental studies have highlighted the key role of charged residues on the protein surface in order to explain the different stabilities of cold-shock homologues.39−41 However, the creation of such favorable interactions may come at the cost of reduced conformational entropy, which would oppose to protein stabilization. Instead, TmCsp was found to be locally more flexible at ambient temperature, despite having many more ion pairs. To explore the mechanical stability of EcCsp and TmCsp, we have performed constant-force SMD simulations that mimic force-spectroscopy experiments. Experiments have shown that the mechanical resistance of a protein depends on the location of force application.42 We chose here to consider the obvious

considering the fraction of folded state taken by the proteins as a function of the temperature, and using the RMSD of the protein rigid core as the order parameter to distinguish folded and unfolded structures. The resulting curves are shown in Figure 2c and Figure SI3. The melting temperature Tm is defined as the temperature at which the fitted fractions of folded and unfolded proteins are equal. While the in silico Tm are slightly shifted as compared to experimental Texp m values for CHARMM36: Tm = 342 K (Texp m = 333 K) for EcCsp, and Tm = 374 K (Texp m = 360 K) for TmCspthe stability shift between EcCsp and TmCsp is very well reproduced by our simulations (32 K for CHARMM36, and 20 K for Amber99sb*; see Figure SI3). Having assessed the stability gap between the two homologues, we inquire next whether the thermal resistance of TmCsp correlates with a major stiffness of the protein matrix as suggested by the rigidity paradigm.2,4,5 The standard deviations of the protein total RMSD with respect to a reference structure can provide a first glimpse into Csps flexibilities at any temperature (see Figure 2a for CHARMM36 and Figure SI2 for Amber99sb*). However, this approach does not reveal the effect of conformational jumps on structurally close minima.11 The set of relevant protein conformations is instead revealed by a cluster analysis based on structural proximities. The number of populated states (18 and 15, respectively) and the associated time scales of conformational space explorations extracted from an exponential fit of the growth of the number of clusters, n(t) ≃ n∞(1 − e(−t/τc)), (τc being 56 and 40 ns, respectively) suggest that both homologues exhibit very similar conformational flexibilities at ambient temperature (Figure 3a,b, and Figure SI4 for Amber99sb*). Moreover, the network representation of the explored conformational landscape is similar between EcCsp and TmCsp, and its compact structure indicates that the protein fluctuates around a very limited number of basins at ambient 5886

DOI: 10.1021/acs.jpclett.7b02611 J. Phys. Chem. Lett. 2017, 8, 5884−5890

Letter

The Journal of Physical Chemistry Letters

Figure 4. Csps are not very resistant upon force, but EcCsp is more stable. (a) End-to-end distance as a function of time for EcCsp (25 traces, blue) and TmCsp (25 traces, red), under a load of 200 pN using CHARMM36. The proteins final extension is close to 21 nm, but the data is truncated at 18 nm. The histograms of rupture times, defined as the moment the end-to-end distance extends beyond a cutoff value of 4.5 nm, is shown on the top (full blue bars: EcCsp, and empty red bars: TmCsp; bin size of 2 ns). This cutoff value was selected because the first intermediates whose rupture triggers the further unfolding of the proteins are located below this threshold (Figure 3a). The distributions of end-to-end distances, averaged over all 25 trajectories for the time interval where the protein length is below 18 nm, is shown on the right. Small bars indicate the average extension of both proteins starting from the native state average length (red: ΔLTm for TmCsp, and ΔLEc for EcCsp). (b) Initial structure (left) and typical structure of the first intermediate (center) for EcCsp. (c) Further unfolding and unzipping of the β4−β5 motif, is only possible when the long L3 loop (light blue, indicated by a red arrow) opens (Figure SI10). The red circle indicates the position of a conserved glycine residue in EcCsp and TmCsp. (d) Similar data for TmCsp, where the intermediate is very short-lived and closer to the initial native state. (e) For TmCsp, unfolding of the β4−β5 and β1−β4 strands is not hindered by the L3 loop, which has a very different conformation as compared to EcCsp.

Constant-velocity experiments on TmCsp and another mesophilic Csp (from Bacillus subtilisis) evidenced that the rupture force of TmCsp was moderately higher than that of BsCsp at all investigated velocities.26 We have repeated our simulations on BsCsp (Figure SI9), and we found that it was slightly more stable than TmCsp, but less than EcCsp. While this seems at odds with the experimental results,26 we should keep in mind that simulations are performed in a higher forceregime, corresponding to a higher pulling-velocity in the framework of constant-speed AFM experiments. By extrapolating the velocity-dependence of the rupture forces of TmCsp and BsCsp measured in the experiments,26 it is actually expected that BsCsp would become stronger than TmCsp in a high velocity/force regime, which is validated by our simulations (Figure SI9). Overall, these results suggest that in the context of a pair of protein homologues with identical folds, protein thermal and mechanical stability are not correlated. Interestingly, recent experiments on the design of a more thermally stable mutant of BsCsp lead to a protein less mechanically resistant than both TmCsp and BsCsp.26 The unfolding traces also provide valuable insight into the unfolding mechanism of EcCsp and TmCsp. The observation of several steps and long-lived intermediates contrasts with the usual one-step event observed in the experiments,26,27 but is not surprising since the corresponding time scales are radically different. Interestingly, very recent experiments could detect the presence of intermediates for TmCsp using a force-ramp protocol.43 Here, these intermediates are evidenced by estimating the length distribution over the time interval between the initial application of the force and total protein

and natural scenario where force is applied to the N- and Ctermini of the protein. In the simulations, a terminal Cα is fixed, while a directional force is applied to the other terminal Cα. However, forces that result in protein unfolding on the second time scale in the experiments, do not lead to protein-unfolding in the much shorter sub-microsecond time scale of the simulations, where much larger forces thus need to be used. For mechanically resistant proteins such as the well-studied ubiquitin, the experimental rupture force is around 150 pN,25 but several hundreds of pN need to be applied to observe unfolding on the nanosecond time scale.32 Experimental results,26,27,43 as well as our own exploratory simulations, suggested that both TmCsp and EcCsp are not very mechanically stable: in experiments, the rupture force is in the 50−80 pN range, and in simulations, both Csps unfold on the nanosecond time scale at forces on the order of 200 pN (Figure 4). As protein unfolding is a stochastic event, we have repeated our simulations tens of times to accumulate reasonable statistics, both at 200 pN (Figure 4) and at a lower force of 120 pN (Figure SI7). It is quite remarkable that TmCsp, which is stable up to very high temperatures, readily unfolds at these low forces. This is because, despite the presence of many β-sheets in their structure, Csps lack a forceclamp element that would strongly resist to force thanks to cooperative hydrogen-bond (HB) patterns, as discussed later. Another significant observation is that TmCsp is not more mechanically stable than EcCsp at the two investigated forces and for both force fields (Figure 4, Figure SI7 and Figure SI8). In fact, the typical unfolding time is always longer for EcCsp as compared to TmCsp, suggesting it is slightly more resistant. 5887

DOI: 10.1021/acs.jpclett.7b02611 J. Phys. Chem. Lett. 2017, 8, 5884−5890

Letter

The Journal of Physical Chemistry Letters

in the protein native state would exist in any of the conformations visited by each replica (see Figures SI11 and SI12). In all cases, the loss of native contacts in the β-barrel structure between the edge of β3/L3 and β5 is observed at the protein melting temperature. These results are in line with NMR experiments on amide protons performed on EcCsp that showed that the β5 strand was the least protected from hydrogen exchange.46 Interestingly, the thermally weak β5/β3− L3 region is directly involved in the nucleic-acid binding function of Csp,47 suggesting that thermal denaturation of the protein matrix would immediately affect its functional elements. At higher temperatures, most of the EcCsp contacts are affected (β4−β5 and β1−β4 interactions), but in TmCsp, the β4−β5 interactions, enriched in ion-pairs, are still present. It is worth noting that salt bridges connecting β1, β4, and β5 in TmCsp, while absent in EcCsp, do not seem to provide any further stabilization against force unfolding. A BsCsp mutant with an increased number of ions pairs between these strands actually did not exhibit any increased mechanical stability.26 We have also repeated SMD simulations at a high temperature of 350 K (Figure SI13), where ion-pairing is stronger because of the reduced desolvation cost, but the respective order of mechanical stabilities between TmCsp and EcCsp remained unchanged. In conclusion, we have proved that by deploying a complementary simulation approach, it is possible to compare thermal and mechanical stability, to quantify conformational flexibility, and to provide microscopic insights that are otherwise inaccessible to experimental techniques. The REST2 method30,31 applied here for the first time to midsize proteins on time scales of 400 ns per replica, long enough to assess protein melting properties, allowed to quantify the thermal stability of the two Csp homologues by reproducing their stability curves, individuating their melting, and accessing to unprecedented molecular details at a reasonable computational cost. We have shown that the enhanced stability of the thermophilic Csp is not related to the mechanical resistance of the fold, or to the conformational rigidity sampled at ambient condition. Within our framework, it is now possible to explore the correlation between thermal and mechanical stabilities for other systems, and to investigate the design of mutants intended to suppress or enhance this correlation.

unfolding (Figure 4a), which shows intermediate peaks (i.e., states). As an effect of the initial transient diffusion away from the native state, EcCsp exhibits the presence of a shallow peak around 3.4 nm, whereas the native state end-to-end distance L0 is close to 1.8 nm. For TmCsp, because of a fast transition toward an initial intermediate, the peak is very narrow and it is centered around 2.0 nm (L0 = 1.4 nm). The disruption of the long-lived first intermediate state, which is very close to the native state for TmCsp (Figure 4d) but already slightly distorted for EcCsp (Figure 4b), triggers the subsequent total unfolding of the protein. The long-lived initial intermediates observed in EcCsp is related to the peculiar orientation of the loop L3 connecting the β3 and β4 strands. For TmCsp, β4 and β5 unzip from the Cterminal, breaking the HB between VAL45 and VAL65. This initial intermediate subsequently unfolds when the remaining HBs between β4 and β5, as well as that between β1 and β4, break. For EcCsp, the protein actually slowly extends from its N-terminal, by breaking some HBs between β1 and β4. As compared to TmCsp, the rupture of the β4−β5 interaction only occurs much later, and it triggers the rest of the protein unfolding. At a molecular level, the slow kinetics of β4−β5 separation is caused by the steric hindrance of the long L3 loop that actually blocks the unzipping of the two strands from the C-terminal (Figure 4c and Figure SI10). In TmCsp (Figure 4d), the L3 loop of similar length is present, but its geometry and position do not block the unzipping from the C-terminal (Figure 4e). We have further verified that BsCsp, whose stability is intermediate between that of EcCsp and that of TmCsp, exhibits an intermediate situation (Figure SI10). While it is not clear whether this role of the obstructing loop would manifest itself in AFM experiments performed on longer time scales and using lower forces, our results coincide with the recent observation that a single mutation increasing the flexibility of the TmCsp L3 loop resulted in a significant decrease of the mechanical stability.27 From the perspective of pulling experiments, the softness of a protein has been proposed to be related to the separation distance Δx, along the reaction coordinate (here, the end-toend distance), between the transition state for unfolding and the native state. By comparing the mechanical unfolding of TmCsp and BsCsp26 it was suggested that TmCsp was softer mechanically speaking26 as compared to its mesophilic homologue, thus showing a larger Δx. However, the approach to fit the parameter Δx on the basis of the Bell model,44,45 assumes that the position of the native and transition states are force-independent, and that this transition state is unique. Recent force-ramp experiments on TmCsp have evidenced the existence of multiple unfolding pathways,43 further questioning the relevance of the phenomenological Δx. Our results therefore suggest that the connection between Δx and mechanical softness is not obvious, as the softness is not uniquely defined, and may be measured differently in the simulations and in the experiments. Finally, we compare thermal and mechanical unfolding mechanisms. We first note that the average end-to-end distances close to the melting temperature (respectively equal to 1.5 nm for TmCsp and 1.8 nm for EcCsp) are radically different from that of the long-lived initial intermediates observed during force-unfolding. We have then localized the weak spots for thermal denaturation, and we have compared them to those of force induced unfolding, by computing the contact maps, i.e., the probability that a given contact that exists



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpclett.7b02611. Methods and supplementary figures cited in the main text (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Guillaume Stirnemann: 0000-0002-5631-5699 Fabio Sterpone: 0000-0003-0894-8069 Author Contributions

G.S. and F.S. designed the research, G.S. performed the simulations and the analysis, and G.S. and F.S. wrote the manuscript. 5888

DOI: 10.1021/acs.jpclett.7b02611 J. Phys. Chem. Lett. 2017, 8, 5884−5890

Letter

The Journal of Physical Chemistry Letters Notes

(17) Irbäck, A.; Mitternacht, S. Thermal Versus Mechanical Unfolding of Ubiquitin. Proteins: Struct., Funct., Genet. 2006, 65, 759−766. (18) Stirnemann, G.; Kang, S.-g.; Zhou, R.; Berne, B. J. How Force Unfolding Differs from Chemical Denaturation. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 3413−3418. (19) Dudko, O. K.; Graham, T. G. W.; Best, R. B. Locating the Barrier for Folding of Single Molecules Under an External Force. Phys. Rev. Lett. 2011, 107, 208301. (20) Sun, L.; Noel, J. K.; Sulkowska, J. I.; Levine, H.; Onuchic, J. N. Connecting Thermal and Mechanical Protein (Un)folding Landscapes. Biophys. J. 2014, 107, 2950−2961. (21) Beedle, A. E. M.; Lezamiz, A.; Stirnemann, G.; Garcia-Manyes, S. The Mechanochemistry of Copper Reports on the Directionality of Unfolding in Model Cupredoxin Proteins. Nat. Commun. 2015, 6, 7894. (22) Cao, Y.; Yoo, T.; Li, H. Single Molecule Force Spectroscopy Reveals Engineered Metal Chelation Is a General Approach to Enhance Mechanical Stability of Proteins. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 11152−11157. (23) Cecconi, C.; Shank, E. A.; Bustamante, C.; Marqusee, S. Direct Observation of the Three-State Folding of a Single Protein Molecule. Science 2005, 309, 2057−60. (24) Fernandez, J. M.; Li, H. Force-Clamp Spectroscopy Monitors the Folding Trajectory of a Single Protein. Science 2004, 303, 1674− 1678. (25) Schlierf, M.; Li, H.; Fernandez, J. M. the Unfolding Kinetics of Ubiquitin Captured with Single-Molecule Force-Clamp Techniques. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 7299−304. (26) Tych, K. M.; Batchelor, M.; Hoffmann, T.; Wilson, M. C.; Paci, E.; Brockwell, D. J.; Dougan, L. Tuning Protein Mechanics Through an Ionic Cluster Graft from an Extremophilic Protein. Soft Matter 2016, 12, 2688−99. (27) Tych, K. M.; Batchelor, M.; Hoffmann, T.; Wilson, M. C.; Hughes, M. L.; Paci, E.; Brockwell, D. J.; Dougan, L. Differential Effects of Hydrophobic Core Packing Residues for Thermodynamic and Mechanical Stability of a Hyperthermophilic Protein. Langmuir 2016, 32, 7392−7402. (28) Schindelin, H.; Jiang, W.; Inouye, M.; Heinemann, U. Crystal Structure of CspA, the Major Cold Shock Protein of Escherichia Coli. Proc. Natl. Acad. Sci. U. S. A. 1994, 91, 5119−5123. (29) Kremer, W.; Schuler, B.; Harrieder, S.; Geyer, M.; Gronwald, W.; Welker, C.; Jaenicke, R.; Kalbitzer, H. R. Solution NMR Structure of the Cold-Shock Protein from the Hyperthermophilic Bacterium Thermotoga Maritima. Eur. J. Biochem. 2001, 268, 2527−2539. (30) Wang, L.; Friesner, R. A.; Berne, B. J. Replica Exchange with Solute Scaling: A More Efficient Version of Replica Exchange with Solute Tempering (REST2). J. Phys. Chem. B 2011, 115, 9431−9438. (31) Stirnemann, G.; Sterpone, F. Recovering Protein Thermal Stability Using All-Atom Hamiltonian Replica-Exchange Simulations in Explicit Solvent. J. Chem. Theory Comput. 2015, 11, 5573−5577. (32) Lu, H.; Schulten, K. Steered Molecular Dynamics Simulations of Force-Induced Protein Domain Unfolding. Proteins: Struct., Funct., Genet. 1999, 35, 453−463. (33) Horn, G.; Hofweber, R.; Kremer, W.; Kalbitzer, H. R. Structure and Function of Bacterial Cold Shock Proteins. Cell. Mol. Life Sci. 2007, 64, 1457−1470. (34) Chatterjee, S.; Jiang, W.; Emerson, S. D.; Inouye, M. The Backbone Structure of the Major Cold-Shock Protein CS7.4 of Escherichia Coli in Solution Includes Extensive Beta-Sheet Structure. J. Biochem. 1993, 114, 663−669. (35) Makhatadze, G. I.; Marahiel, M. A. Effect of PH and Phosphate Ions on Self-Association Properties of the Major Cold-Shock Protein from Bacillus Subtilis. Protein Sci. 1994, 3, 2144−2147. (36) Perl, D.; Welker, C.; Schindler, T.; Schröder, K.; Marahiel, M. A.; Jaenicke, R.; Schmid, F. X. Conservation of Rapid Two-State Folding in Mesophilic, Thermophilic and Hyperthermophilic Cold Shock Proteins. Nat. Struct. Biol. 1998, 5, 229−235.

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The research leading to these results has received funding from the ERC (FP7/2007-2013) Grant Agreement No. 258748. Part of this work was performed using HPC resources from GENCI [CINES and TGCC] (Grant x201676818). This work was supported also by the “Initiative d’Excellence” program from the French State (Grant ”DYNAMO”, ANR-11-LABX-001101).



REFERENCES

(1) Vieille, C.; Zeikus, G. J. Hyperthermophilic Enzymes: Sources, Uses, and Molecular Mechanisms for Thermostability. Microbiol. Mol. Biol. Rev. 2001, 65, 1−43. (2) Somero, G. N. Temperature Adaptation of Enzymes. Annu. Rev. Ecol. Syst. 1978, 9, 1−29. (3) Sterpone, F.; Melchionna, S. Thermophilic Proteins: Insight and Perspective from in Silico Experiments. Chem. Soc. Rev. 2012, 41, 1665. (4) Závodszky, P.; Kardos, J.; Svingor, A.; Petsko, G. A. Adjustment of Conformational Flexibility Is a Key Event in the Thermal Adaptation of Proteins. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 7406−7411. (5) Jaenicke, R.; Böhm, G. The Stability of Proteins in Extreme Environments. Curr. Opin. Struct. Biol. 1998, 8, 738−748. (6) Wolf-Watz, M.; Thai, V.; Henzler-Wildman, K.; Hadjipavlou, G.; Eisenmesser, E.; Kern, D. Linkage Between Dynamics and Catalysis in a Thermophilic-Mesophilic Enzyme Pair. Nat. Struct. Mol. Biol. 2004, 11, 945−949. (7) Hernandez, G.; Jenney, F. E.; Adams, M. W. W.; LeMaster, D. M. Millisecond Time Scale Conformational Flexibility in a Hyperthermophile Protein at Ambient Temperature. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 3166−3170. (8) Tehei, M.; Madern, D.; Franzetti, B.; Zaccai, G. Neutron Scattering Reveals the Dynamic Basis of Protein Adaptation to Extreme Temperature. J. Biol. Chem. 2005, 280, 40974−40979. (9) Katava, M.; Stirnemann, G.; Zanatta, M.; Capaccioli, S.; Pachetti, M.; Ngai, K. L.; Sterpone, F.; Paciaroni, A. Critical Structural Fluctuations of Proteins upon Thermal Unfolding Challenge the Lindemann Criterion. Proc. Natl. Acad. Sci. U. S. A. 2017, 114, 9361− 9366. (10) Tehei, M.; Franzetti, B.; Madern, D.; Ginzburg, M.; Ginzburg, B. Z.; Giudici-Orticoni, M.-T.; Bruschi, M.; Zaccai, G. Adaptation to Extreme Environments: Macromolecular Dynamics in Bacteria Compared in Vivo by Neutron Scattering. EMBO Rep. 2004, 5, 66−70. (11) Kalimeri, M.; Rahaman, O.; Melchionna, S.; Sterpone, F. How Conformational Flexibility Stabilizes the Hyperthermophilic Elongation Factor G-Domain. J. Phys. Chem. B 2013, 117, 13775−13785. (12) Katava, M.; Kalimeri, M.; Stirnemann, G.; Sterpone, F. Stability and Function at High Temperature. What Makes a Thermophilic GTPase Different from Its Mesophilic Homologue. J. Phys. Chem. B 2016, 120, 2721−2730. (13) Kalimeri, M.; Girard, E.; Madern, D.; Sterpone, F. Interface Matters: The Stiffness Route to Stability of a Thermophilic Tetrameric Malate Dehydrogenase. PLoS One 2014, 9, e113895. (14) Katava, M.; Maccarini, M.; Villain, G.; Paciaroni, A.; Sztucki, M.; Ivanova, O.; Madern, D.; Sterpone, F. Thermal Activation of ‘Allosteric-Like’ Large-Scale Motions in a Eukaryotic Lactate Dehydrogenase. Sci. Rep. 2017, 7, 41092. (15) Roca, M.; Liu, H.; Messer, B.; Warshel, A. On the Relationship Between Thermal Stability and Catalytic Power of Enzymes. Biochemistry 2007, 46, 15076−15088. (16) Neuman, K. C.; Nagy, A. Single-Molecule Force Spectroscopy: Optical Tweezers, Magnetic Tweezers and Atomic Force Microscopy. Nat. Methods 2008, 5, 491−505. 5889

DOI: 10.1021/acs.jpclett.7b02611 J. Phys. Chem. Lett. 2017, 8, 5884−5890

Letter

The Journal of Physical Chemistry Letters (37) Sugita, Y.; Okamoto, Y. Replica-Exchange Molecular Dynamics Method for Protein Folding. Chem. Phys. Lett. 1999, 314, 141−151. (38) Schuler, B.; Kremer, W.; Kalbitzer, H. R.; Jaenicke, R. Role of Entropy in Protein Thermostability: Folding Kinetics of a Hyperthermophilic Cold Shock Protein at High Temperatures Using 19F NMR. Biochemistry 2002, 41, 11670−11680. (39) Perl, D.; Mueller, U.; Heinemann, U.; Schmid, F. X. Two Exposed Amino Acid Residues Confer Thermostability on a Cold Shock Protein. Nat. Struct. Biol. 2000, 7, 380−383. (40) Perl, D.; Schmid, F. X. Electrostatic Stabilization of a Thermophilic Cold Shock Protein. J. Mol. Biol. 2001, 313, 343−357. (41) Frankenberg, N.; Welker, C.; Jaenicke, R. Does the Elimination of Ion Pairs Affect the Thermal Stability of Cold Shock Protein from the Hyperthermophilic Bacterium Thermotoga Maritima? FEBS Lett. 1999, 454, 299−302. (42) Carrion-Vazquez, M.; Li, H.; Lu, H.; Marszalek, P. E.; Oberhauser, A. F.; Fernandez, J. M. The Mechanical Stability of Ubiquitin Is Linkage Dependent. Nat. Struct. Biol. 2003, 10, 738−743. (43) Schönfelder, J.; Perez-Jimenez, R.; Muñoz, V. A Simple TwoState Protein Unfolds Mechanically Via Multiple Heterogeneous Pathways at Single-Molecule Resolution. Nat. Commun. 2016, 7, 11777. (44) Bell, G. I. Models for the Specific Adhesion of Cells to Cells. Science 1978, 200, 618−627. (45) Schlierf, M.; Rief, M. Temperature Softening of a Protein in Single-Molecule Experiments. J. Mol. Biol. 2005, 354, 497−503. (46) Jaravine, V. A.; Rathgeb-Szabo, K.; Alexandrescu, A. T. Microscopic Stability of Cold Shock Protein a Examined by NMR Native State Hydrogen Exchange As a Function of Urea and Trimethylamine N-Oxide. Protein Sci. 2000, 9, 290−301. (47) Sachs, R.; Max, K. E. a.; Heinemann, U. D. O.; Balbach, J. RNA Single Strands Bind to a Conserved Surface of the Major Cold Shock Protein in Crystals and Solution. RNA 2012, 18, 65−76.

5890

DOI: 10.1021/acs.jpclett.7b02611 J. Phys. Chem. Lett. 2017, 8, 5884−5890