Skip to main content
NCBI home page
Biophysical Journal logoLink to Biophysical Journal
. 2013 Aug 6;105(3):732–744. doi: 10.1016/j.bpj.2013.06.028

Flexibility of the Bacterial Chaperone Trigger Factor in Microsecond-Timescale Molecular Dynamics Simulations

Andrew S Thomas 1, Suifang Mao 1, Adrian H Elcock 1,∗
PMCID: PMC3736688  PMID: 23931321

Abstract

The bacterial chaperone trigger factor (TF) is the first chaperone to be encountered by a nascent protein chain as it emerges from the ribosome exit tunnel. Experimental results suggest that TF possesses considerable conformational flexibility, and in an attempt to provide an atomic-level view of this flexibility, we have performed independent 1.5-μs molecular dynamics simulations of TF in explicit solvent using two different simulation force fields (OPLS-AA/L and AMBER ff99SB-ILDN). Both simulations indicate that TF possesses tremendous flexibility, with huge excursions from the crystallographic conformation caused by reorientations of the protein’s constituent domains; both simulations also predict the formation of extensive contacts between TF’s PPIase domain and the Arm 1 domain that is involved in nascent-chain binding. In the OPLS simulation, however, TF rapidly settles into a very compact conformation that persists for at least 1 μs, whereas in the AMBER simulation, it remains highly dynamic; additional simulations in which the two force fields were swapped suggest that these differences are at least partly attributable to sampling issues. The simulation results provide potential rationalizations of a number of experimental observations regarding TF’s conformational behavior and have implications for using simulations to model TF’s function on translating ribosomes.

Introduction

The prokaryotic chaperone trigger factor (TF) sits at the exit of the 70S ribosome’s protein exit tunnel and is therefore the first point of contact for emerging nascent chains with the outside world (see Hoffmann et al. (1) for an excellent, comprehensive review of all aspects of TF function). TF has been shown to directly modulate the folding pathways of model proteins, apparently delaying folding to occur posttranslationally (2), and a number of experimental studies have characterized its interactions with nascent chains as a function of nascent chain length (3–7). A recent coarse-grained molecular simulation study has provided interesting views of how TF might affect the folding behavior of nascent chains (8), and intriguing experimental evidence indicating that TF might also have an unfoldase activity has recently been reported (9). Additionally, there is crystallographic evidence that TF can function in a posttranslational mode by helping the assembly of oligomeric proteins (10). Because TF has been shown to interact with a large number of Escherichia coli proteins, although by no means all (11), it is clear that any attempt to model the de novo folding of bacterial proteins as it is likely to occur in vivo must ultimately include a role for TF.

TF has a modular structure constituted of the following domains (Fig. 1):

  • 1.

    An N-terminal domain (residues 1–110; colored ice-blue) responsible for ribosome binding;

  • 2.

    A PPIase domain (residues 151–243; yellow); and

  • 3.

    A C-terminal domain (residues 251–432), which is often considered as being made up of two subdomains: Arm 1 (251–360; red) and Arm 2 (361–432; pink).

Figure 1.

Figure 1

Open in a new tab

Domain structure of trigger factor. (Ice blue) N-terminal domain responsible for ribosome binding (residues 1–110); (yellow) PPIase domain (residues 151–243); (red) Arm 1 domain (residues 251–360); and (pink) Arm 2 domain (residues 361–432). This and all other structural figures were produced with the software VMD (47).

The structure is unusual in that the N-terminal and PPIase domains, while being close to each other in sequence, are very distantly separated in structure: a long linker (residues 111–150) connects the two domains. A variety of experimental results indicate that TF has considerable conformational flexibility (1). Significant differences in the disposition of its domains are seen in different crystal forms; for example, superposition of the N-terminal domains of the two molecules in the asymmetric unit of full-length E. coli TF (12) results in an 11 Å displacement of the PPIase domains (6). Substantial differences exist between the crystallographic conformations of Thermotoga maritima TF in its apo state and in the dimeric complex that it forms with the ribosomal protein S7, and even greater differences are apparent when the apo TF structures of T. maritima and E. coli are compared (10). Additional evidence for flexibility in TF comes from a NMR study on a construct containing essentially only the C-terminal domain (13), showing that it exhibits flexibility over a range of timescales; unfortunately, NMR studies of full-length TF appear unlikely, owing to aggregation issues (13).

Structural data also indicate that conformational changes occur within TF when it binds to the ribosome: changes in the region of the protein responsible for ribosome binding have been inferred from studies of the N-terminal domain bound to the ribosome (12,14–16), and cryoelectron microscopy studies of TF bound to a ribosome-nascent-chain complex required that the PPIase domain be rotated by ∼24° toward the Arm domains to fit the observed electron density (6). Finally, indications of TF’s conformational flexibility have also come from ensemble Förster resonance energy transfer (FRET) measurements of constructs with donor and acceptor chromophores placed at residue positions 14, 150, 326, and 376 (17). These data have suggested that TF can assume at least two generic conformational states: a compact conformation characteristic of the free (ribosome-unbound) state, and an extended conformation that appears to be induced by binding to the ribosome. These conformational states have not thus far been characterized at higher resolution (17).

Given that more detailed studies of full-length TF’s conformational behavior may not be forthcoming from experimental studies, we set out to conduct a series of long timescale molecular dynamics (MD) simulations of TF in explicitly modeled aqueous solution. The extent of the conformational flexibility exhibited by TF in these simulations is remarkable; it provides possible structural explanations for the compact and extended conformations described above, and suggests that a full understanding of TF’s function will almost certainly need to explicitly account for its extreme conformational plasticity.

Materials and Methods

All MD simulations were performed with the GROMACS 4.5 software (18) using two different but widely used simulation force fields: the AMBER ff99SB-ILDN force field (19,20) combined with the TIP4P-Ew water model (21), and the OPLS-AA/L force field (22) combined with the TIP4P water model (23). A full description of the protocols used to perform and analyze the simulations is provided in the Supporting Material.

Results

As noted in Materials and Methods, we performed independent 1.5-μs MD simulations of TF in explicit solvent with two different but widely used simulation force fields: AMBER ff99SB-ILDN (19,20) and OPLS-AA/L (22). The conformational flexibility exhibited by TF with both force fields is tremendous (see Movie S1 and Movie S2 in the Supporting Material). In Fig. 2 A we plot the root-mean-square deviations (RMSDs) of TF’s backbone atoms from their crystallographic positions as a function of time for the duration of the simulations. The RMSD obtained using the AMBER ff99SB-ILDN force field (red line) quickly rises to 10 Å and then fluctuates dramatically throughout the simulation, falling as low as 5 Å and reaching as high as 18 Å. With the OPLS-AA/L force field (blue line), on the other hand, the computed RMSD rises even higher, approaching 20 Å within 200 ns, before reaching a plateau at 22 Å. To provide a point of reference for these extraordinarily large changes, we compare in Fig. 2 B the RMSDs obtained during the first 100 ns of the TF simulations with that obtained from a corresponding 100-ns MD simulation of threonine synthase, a globular protein with similar size and charge properties to TF. With threonine synthase (green line), the backbone RMSD value hovers at a value ∼2–3 Å, which is typical of stable MD simulations reported in the literature; in contrast, during the same time period, the two TF simulations reach RMSDs of ∼7.5–12.5 Å.

Figure 2.

Figure 2

Open in a new tab

Conformational flexibility of trigger factor. (A) Root-mean-square deviation (RMSD) of backbone atoms from their positions in the PDB:1W26 crystal structure (12), plotted as a function of the simulation time for simulations performed with the AMBER ff99SB-ILDN (red) and OPLS-AA/L (blue) force fields. (B) Same as panel A, but comparing behavior obtained over 100 ns with the RMSD computed for threonine synthase (green) simulated with the ff99SB-ILDN force field.

One possible but essentially trivial explanation for the above RMSDs could be that—with both force fields—TF is simply unfolding in the simulations. If so, then we should expect to see that RMSDs computed separately for each of TF’s individual domains should also reach very high values. To show that this is not the case, we plot in Fig. 3 the RMSDs computed for the N-terminal, PPIase, Arm 1, and Arm 2 domains as a function of time. These RMSDs show that the individual domains do not undergo any drastic deviations from the conformations found in the crystal structure. For the AMBER simulation (Fig. 3 A), the N-terminal and PPIase domains both show conformational stabilities expected of independently folded domains even though very large fluctuations are seen in the RMSD computed for the entire TF molecule (Fig. 2 A); the Arm 1 domain appears to be somewhat more variable. For the OPLS simulation (Fig. 3 B), the conformations of individual domains again appear to be comparatively stable—at least in comparison to the conformation of the entire molecule—although the N-terminal domain shows more significant changes than are seen with the corresponding AMBER simulation.

Figure 3.

Figure 3

Open in a new tab

Conformational flexibility of individual domains. Plots of the root mean square deviations (RMSD) of backbone atoms from the crystallographic conformation computed for individual domains. Results are shown for (A) the AMBER ff99SB-ILDN and (B) OPLS-AA/L simulations. For calculations of the N-terminal domain RMSD, the highly mobile residues implicated in ribosome binding (approximated here as residues 40–60) were omitted from the calculations. For calculations of the Arm 1 domain RMSD, the highly flexible loop residues 321–333 were omitted from the calculations.

Although the individual domains of the protein basically preserve their crystallographic conformations, both simulations identify two regions of significant conformational variability within domains that may be of functional importance. The first is within the N-terminal domain and involves residues ∼40–60; notably, this region encompasses the 43GFRxGxxP50 peptide considered to be the signature motif (12) responsible for binding to the ribosome and where conformational changes in the Deinococcus radiodurans TF concomitant with ribosome binding have been inferred from crystallography (15,16). The top panel of Fig. 4 shows superimposed snapshots of the N-terminal domain sampled at 100-ns intervals, and compares the motion seen in residues 43–50 with the conformational variability seen in five E. coli TF crystal structure conformations. Structural differences in this region of the protein are also apparent when N-terminal domains of TFs from different organisms are compared (24). Clearly, both simulation force fields identify this region as one capable of undergoing significant conformational fluctuations.

Figure 4.

Figure 4

Open in a new tab

Comparison of flexibility in the N-terminal and Arm 1 domains. (Top panel) Crystal shows five TF N-terminal domain chains extracted from two crystal structures: chains A and B of PDB:1W26 (full-length TF), and chains A, B, and C of PDB:1OMS (crystal structure of the N-terminal domain alone). (Red) 43GFRxGxxP50 peptide considered to be the signature motif (12). AMBER and OPLS show snapshots taken at 100-ns intervals. For all three pictures, structures were aligned using backbone atoms of residues 1–39 and 61–110 (aligned to the PDB:1W26 chain-A structure). (Bottom panel) Crystal shows the two TF Arm 1 domains obtained from chains A and B of PDB:1W26 (full-length TF). AMBER and OPLS show snapshots taken at 100-ns intervals. (Green) The highly mobile residues 321–333. For all three pictures, structures were aligned using backbone atoms of residues 251–320 and 334–360.

Another part of the TF molecule where substantial conformational flexibility is observed is in the loop at the tip of Arm 1. The bottom panel of Fig. 4 compares superimposed snapshots obtained from the MD simulations with the two available conformations from the PDB:1W26 crystal structure; in this case, residues 321–333 are highlighted in green. Experimental studies in which chromophores have been placed at residues 320 and 326 both indicate that this region of the protein engages in significant interactions with nascent chains (4,7). In addition, this region of TF (together with the entire PPIase domain; see below) has been noted to have weaker density in a cryoelectron microscopy study of TF bound to a ribosome-nascent-chain complex (6). For reasons noted below, much of the conformational variability seen at this region in the MD simulations may be a consequence of interactions with the PPIase domain.

The fact that the individual domains are comparatively stable in the simulations strongly suggests that the very high RMSD values obtained when calculations are performed on the entire TF molecule are primarily due to large changes in the relative orientations of TF’s constituent domains. This view is supported when we examine representative structural snapshots from the simulations (Fig. 5). The top panel of Fig. 5 shows snapshots of TF taken from the simulation performed with the AMBER force field (see also Movie S1); in this case, all snapshots have been aligned using the N-terminal (ribosome-binding) domain for reference, allowing us to visualize—to a first approximation—the extent to which TF might undergo conformational fluctuations while bound to the ribosome. Clearly, in this simulation TF adopts an extraordinarily wide range of conformations, twisting and bending repeatedly, and at one point (∼900 ns) even appearing to double-over. Much of the flexibility seen in these snapshots is attributable to hinge-like movements at the junction of the N-terminal domain (ice blue) and the Arm domains (red and pink), i.e., at residues ∼110–115. The middle panel of Fig. 5 shows the same TF conformations aligned instead using the PPIase domain as a point of reference; from this it can be seen that additional flexibility within the molecule is also to be found at the junction between the PPIase domain (yellow) and the Arm domains, i.e., at residues ∼150 and ∼245.

Figure 5.

Figure 5

Open in a new tab

Snapshots from the 1.5-μs simulations. Snapshots are shown extracted at 100-ns time intervals and aligned to provide a common reference. (Top panel) Snapshots from the AMBER ff99SB-ILDN simulation aligned using the N terminal domain. (Middle panel) Snapshots from the AMBER ff99SB-ILDN simulation aligned using the PPIase domain. (Bottom panel) Snapshots from the OPLS-AA/L simulation aligned using the N-terminal domain. The coloring scheme is the same as used in Fig. 1.

In contrast to the continued sampling of widely different conformations that occurs in the AMBER simulation, in the simulation performed with the OPLS-AA/L force field the TF molecule adopts only a few distinct structures before settling into an extremely compact conformation (bottom panel of Fig. 5; see also Movie S2); this structure remains largely unchanged over the course of at least 1 μs. The highly compact nature of this structure is indicated by the fact that the radius of gyration of the entire TF molecule computed over the last microsecond of the OPLS simulation is 25.6 ± 0.5 Å. For comparison, the radius of gyration sampled over the same timescale for the AMBER simulation is 32.7 ± 3.0 Å, and that of the PDB:1W26 crystal structure is 35.0 Å. Plots of the radius of gyration values sampled during both simulations are shown versus time and in histogram form in Fig. S1 (see the Supporting Material) from which it can be seen that the two simulations produce very different views of TF’s conformational behavior: OPLS predicts a compact, relatively static structure while AMBER predicts an extended, highly dynamic structure. The substantial interdomain motions indicated by the snapshots in Fig. 5—and the rapid timescale over which such changes can occur—can be seen in plots of the domain-domain distances versus simulation time (see Fig. S2).

An important consequence of the huge conformational changes that occur in both TF simulations is the formation of a large number of new atomic contacts that are not seen in the crystal structure. Fig. 6 shows residue-residue contact maps constructed from the simulations and compares them with a corresponding map constructed from the TF crystal structure. In comparison with the crystal structure, both simulations show new contacts between the PPIase domain and the two Arm domains, and between Arms 1 and 2; in Fig. S3 we show that these additional contacts seen in the MD simulations cannot be rationalized in terms of localized fluctuations of the crystal structure conformation.

Figure 6.

Figure 6

Open in a new tab

Residue-residue contact maps. Maps are shown computed from (A) the crystal structure, (B) the AMBER ff99SB-ILDN simulation, and (C) the OPLS-AA/L simulation. In the case of the simulation contact maps, both the symbol size and the symbol color reflect the frequency with which a contact was observed. Contacts were defined using a heavy atom distance cutoff of 5 Å; contacts with frequencies of occurrence <0.001 are omitted. (Colors are expressed in a linear scale from blue-to-green-to-red with contacts of frequency 1 colored blue, and contacts with frequency 0.001 colored red.) Symbol sizes are computed using 3∗(log10(frequency)+3)/100.

The contacts formed between the PPIase domain and Arm 1 are especially intriguing. Although the specific residues involved in the contacts differ somewhat between the AMBER and OPLS simulations, they both involve residues ∼185–195 and ∼220–225 of the PPIase domain interacting with the very flexible loop residues ∼320–345 of Arm 1 (see above). In the AMBER simulation, the R193 side chain of the PPIase domain—which is solvent-exposed in the crystal structure—becomes buried within a bowl formed by the flexible loop of Arm 1 (e.g., see the snapshot taken at 1.5 μs in Fig. 7 a); here, contacts with E326 form, break, and reform on a number of occasions. A series of snapshots showing the repeated incursions of R193 into the bowl of Arm 1 is shown in Fig. S4. In the OPLS simulation, much more extensive contacts occur between the PPIase and Arm 1 domains, and the interface is characterized by an almost continuous surface of five aromatic side chains contributed by the PPIase domain; again, however, R193 is also present at the interface (Fig. 7 b).

Figure 7.

Figure 7

Open in a new tab

Close-up views of interdomain interfaces. Structures are shown as sampled at a point 1.5 μs into the simulations; the coloring scheme for the cartoon representations is the same as in Fig. 1. Selected protein side chains are labeled.

Intriguingly, in the OPLS simulation, contacts also develop between the flexible (ribosome-binding) loop of the N-terminal domain and the flexible loop of Arm 1 (residues ∼320–345) (Fig. 7 c); the contacts between these two loops—which are ∼46 Å apart at the beginning of the simulation—develop between 100 and 150 ns into the simulation (see Fig. S5 for snapshots). Perhaps even more surprisingly, in the OPLS simulation, contacts are subsequently observed between the N-terminal domain and the PPIase domain: this happens at ∼600 ns and is again notable given that in the crystal structure, the shortest distance between any pair of atoms in these two domains is ∼41 Å. A close-up view of the contacts formed between the N-terminal and PPIase domains at the end of the simulation is shown in Fig. 7 d. These interactions appear to be primarily electrostatic in nature: there are two simple salt bridges (K37:E178 and D42:K181) and a tripartite salt bridge (D167:K46:E210). Interestingly, all three of the N-terminal domain residues that participate in the interaction with the PPIase domain are known to be involved in ribosome binding (12). The ribosome-binding loop of the N-terminal domain therefore simultaneously forms contacts with both Arm 1 and the PPIase domain (see Discussion).

Whereas there are similarities between the two simulations, therefore, it is also clear that they predict quite different overall behaviors. Given that a complete sampling of TF’s conformational behavior would almost certainly require many microseconds to achieve, it is important to ask whether the differences that we see are real—i.e., true differences between the two force fields—or whether they may instead just be consequences of using what are still comparatively short simulation times. One way to answer this question is to take a snapshot obtained from a simulation performed with one force field and use it as a starting point for a new simulation performed with the other force field. As detailed in the Supporting Material, we have performed this force-field swapping in both directions using snapshots taken at a point 1 μs into the original simulations.

Probably the most interesting question to address is what happens when the highly compact TF conformation seen in the OPLS-AA/L simulation is used to seed a new simulation that uses the AMBER ff99SB-ILDN force field. Interestingly, as indicated by the series of snapshots shown in the top panel of Fig. 8 and in Movie S3, a highly compact structure is maintained, although the details of the interdomain contacts change somewhat; the computed radius of gyration is 26.0 ± 0.6 Å (see Fig. S1). Importantly, the PPIase-Arm 1 interface remains essentially unchanged over the course of the 500-ns simulation: a comparison of the interface seen in the OPLS and the AMBER-starting-from-OPLS simulations is shown in Fig. S6. More variation is observed at the interfaces formed by the N-terminal domain with the Arm 1 and the PPIase domains: Fig. S7 shows that the contacts between the N-terminal loop and the Arm 1 loop break before forming again at 500 ns. Fig. S8, on the other hand, shows that the salt-bridge contacts between the N-terminal domain and the PPIase domain also break and reform and that the relative orientations of the two domains shift gradually during the course of the simulation. Although much longer simulations would be required to determine the relative stabilities of the various domain-domain interfaces seen with the two force fields, the observed behavior suggests that the highly compact conformation for TF is at least metastable with both force fields.

Figure 8.

Figure 8

Open in a new tab

Snapshots from the 0.5-μs force-field swap simulations. Snapshots are shown extracted at 100-ns time intervals and aligned using the N-terminal domain to provide a common reference. (Top panel) Snapshots from the AMBER ff99SB-ILDN simulation that started from a conformation sampled at 1 μs in the OPLS-AA/L simulation. (Bottom panel) Snapshots from the OPLS-AA/L simulation that started from a conformation sampled at 1 μs in the AMBER ff99SB-ILDN simulation. The coloring scheme is the same as used in Fig. 1.

When the force-field swap is carried out in the other direction, i.e., when a more extended conformation seen in the AMBER ff99SB-ILDN simulation is used as the starting point for a simulation performed with the OPLS-AA/L force field, we see a continuation of the highly dynamic behavior seen in the original AMBER simulation (bottom panel of Fig. 8 and see Movie S4). We do not see—at least over the course of 500 ns—a recurrence of the long-lived, highly compact conformation seen in the original OPLS-AA/L simulation: the computed radius of gyration during this new simulation is 30.2 ± 1.6 Å (see Fig. S1). Interestingly, however, we again see the formation of new contacts between the PPIase and Arm 1 domains. In this case, in comparison with the behavior seen in the original OPLS simulation, the contacts are less extensive, but R193 of the PPIase domain is involved yet again, this time forming contacts initially with E331 before moving over to E335 as R163 takes its place by forming a salt bridge to E331 (see Fig. S9 for a sequence of snapshots). Overall, therefore, this simulation appears to indicate that both force fields are capable of producing quite dynamic behavior in TF, and that the prediction of contacts between the PPIase and Arm 1 domains emerges as a consistent feature of all four of the simulations that we have performed.

Finally, we return to an analysis of the 1.5-μs simulations in an attempt to make connections with FRET measurements reported by Kaiser et al. (17). In that study, Kaiser et al. performed equilibrium FRET measurements on TF constructs with chromophores placed at residue positions 14, 150, 326, and 376 and concluded that TF appears to adopt a compact form when in its ribosome-unbound state (17). Specifically, in the absence of ribosomes—and at a concentration at which TF is expected to be primarily in a monomeric state (as here)—the FRET efficiencies, Et, for the 14–150, 14–326, and 14–376 intramolecular chromophore pairs were found to be 0.86, 0.88, and 0.82, respectively (reading from Fig. 1b of Kaiser et al. (17)). Because the chromophores were not explicitly included in our MD simulations, we have had to resort to computing FRET efficiencies by retroactively adding the chromophores (in a variety of conformations) to the MD-sampled snapshots and directly computing Et for each combination (see Materials and Methods). The Et values that we obtain from the OPLS simulation for the 14–150, 14–326, and 14–376 pairs are 0.43 ± 0.06, 0.68 ± 0.15, and 0.18 ± 0.22, respectively, which qualitatively reproduce the results obtained by Kaiser et al. Interestingly, although these results are not in especially good quantitative agreement with experiment, they are a clear improvement over what is obtained when we compute FRET efficiencies from the PDB:1W26 crystal structure: for this situation we obtain Et values of 0.24 ± 0.22, 0.83 ± 0.24, and 0.40 ± 0.30 for the 14–150, 14–326, and 14–376 pairs, respectively. It is impossible, therefore, to rationalize even qualitatively the experimental measurements based on the PDB:1W26 crystal structure. That the experimental data can, however, be qualitatively reproduced by FRET calculations performed on the OPLS simulation snapshots suggests that there may be a connection between the very compact conformation obtained in that simulation and the compact state reported by Kaiser et al. (17).

Kaiser et al. have also shown that the FRET signals for the chromophore pairs described above all decrease when TF associates with translating ribosomes, with the 14–150 and 14–376 signals showing especially large changes (17). These changes have been proposed to reflect the formation of a more extended conformation for TF, and it is obviously tempting to ask, therefore, whether the FRET signals for that state might be consistent with the more extended conformations seen in the AMBER simulation reported here. The Et values that we obtain from the snapshots sampled from the AMBER simulation for the 14–150, 14–326, and 14–376 pairs are 0.14 ± 0.10, 0.39 ± 0.17, and 0.26 ± 0.10, respectively; these are to be compared with the corresponding experimental values of 0.64, 0.81, and 0.63 (again, reading from Fig. 1b of Kaiser et al. (17)). The AMBER simulation results are, therefore, only partly consistent with the FRET data, which, it should be remembered, are for TF in complex with ribosomes. On the one hand, the decrease in FRET efficiency for the 14–150 pair that occurs experimentally (0.87 in the absence, 0.64 in the presence of ribosomes, respectively) is nicely mirrored in a comparison of the two simulations (0.43 and 0.14 in the OPLS and AMBER simulations, respectively). On the other hand, the large decrease in FRET efficiency for the 14–376 pair that occurs experimentally (0.82 in the absence, 0.63 in the presence of ribosomes, respectively) is not captured at all by a comparison of the two simulations: the FRET efficiency for this pair actually increases from 0.18 with OPLS to 0.26 with AMBER. It is clearly not possible, therefore, to easily reconcile all of the experimental data reported by Kaiser et al. based on the simulations described here.

Discussion

The conformational flexibility exhibited by TF in the MD simulations reported here is tremendous: very large-scale conformational changes occur, and do so on a timescale of only a few tens of nanoseconds. Given this very high degree of flexibility, it is important to consider the credibility of the simulation force fields that we have used before discussing the potential implications of the observed behavior. We note that the threonine synthase simulation shows that there is at least nothing obviously broken with the AMBER force field: on the timescale simulated here, it is quite capable of producing a stable MD simulation of a protein with similar size and charge characteristics. In recent work testing a variety of simulation force fields in long MD simulations (25) both of the force fields used here were shown to maintain the proteins ubiquitin and GB3 close to their native state structures on a timescale of 10 μs; of the two, however, the AMBER force field performed considerably better at reproducing NMR observables (25). Given that both force fields can maintain globular proteins in their native states on long timescales, we consider it unlikely that the huge deviations from the crystal structure conformation that we have seen here for TF are due to obvious problems with either of them.

It appears that residues that form the junction between the N-terminal domain and the body of TF, and between the PPIase domain and the body of TF, are primarily responsible for the molecule’s extreme flexibility (Fig. 5). Interestingly, these regions are predicted to be potential sites of hinge motions by analysis of the full-length TF crystal structure using the HingeProt server ((26); see the Supporting Material). The fact that much of the flexibility appears to reside at hinges means that very large changes in the RMSD of the full-length TF molecule can be obtained even though the individual domains are themselves of considerable stability. The possibility of rotational motion of the PPIase domain toward the Arms—which is evident especially in Fig. 5 b—is consistent with the 24° rotation required to fit TF into cryoelectron microscopy data (6) and with differences seen in the disposition of this domain in the E. coli and T. maritima apo TF crystal structures (10). Interestingly, this rotation motion is especially apparent when a principal component analysis (27) is performed on the initial 100 ns of each of the 1.5-μs MD simulations described here: with both force fields, the dominant mode of motion is a clear closing of the gap between the PPIase and Arm1 domains (see Movie S5 and Movie S6). Within individual domains, however, both of the 1.5-μs MD simulations have also identified two regions where conformational flexibility is pronounced (Fig. 4), and both of these appear to be credible findings. The observed flexibility in the ribosome binding region of the N-terminal domain (top panel of Fig. 4) is consistent with structural differences seen in the various crystal structures of the N-terminal domain (24), and with the report that conformational changes result in this region when trigger factor binds to the ribosome (12,14–16). The observed flexibility in the loop at the tip of Arm 1 (bottom panel of Fig. 4), on the other hand, appears reasonable in light of the fact that this region of the protein has been strongly implicated in binding to nascent chains, where conformational plasticity is likely, therefore, to be functionally beneficial: as shown by Lakshmipathy et al. (4,7), for example, all tested constructs of nascent luciferase could be cross-linked to a probe placed at residue 320 in TF (4), and similar interactions with this region were also demonstrated by time-resolved fluorescence studies with a chromophore placed at residue 326 (7). Conformational flexibility in this region of the protein can also be discerned from a Translation/Libration/Screw analysis (28) of the PDB:1W26 crystal structure (see the Supporting Material).

In addition to the fact that both of the 1.5-μs MD simulations agree that there should be substantial flexibility in both the ribosome-binding loop of the N-terminal domain and the loop of the Arm 1 domain, there is one other important point of correspondence between the two simulations, which is the repeated prediction of the formation of contacts between residues of the PPIase domain and Arm 1. These two domains are separated by ∼30 Å in the PDB:1W26 crystal structure of full-length E. coli TF. Because it is a common feature of all of the simulations, it is reasonable to ask, why is this interaction not observed crystallographically? One strong possibility is suggested by the comment made by Martinez-Hackert and Hendrickson (10) that the available apo TF crystal structures are “in a sense, not strictly substrate free”: in both crystal structures, the C-terminal domain Arms and the PPIase domains are separated from one another by the intervention of N-terminal domains of neighboring molecules in the crystal lattice (see Figs. 3E and 3F of Martinez-Hackert and Hendrickson (10)). Indirect experimental evidence against the idea that there might be an interaction between the PPIase domain and Arm 1, however, comes from the observation that the two-dimensional NMR spectrum for a C-terminal domain construct that omits the PPIase domain (residues 113–432Δ150–246) matches well with that of a construct that retains it (residues 113–432) (13). It is worth noting, however, that assignment of resonances in the construct that contains the PPIase domain has not been carried out, and that a number of the residues in the flexible loop of Arm 1 were not able to be assigned in the 113–432Δ150–246 construct (e.g., residues 314–321 and 327–333). Ideally, further NMR experiments would explicitly confirm or refute the possibility of the PPIase-Arm 1 domain interactions predicted here.

If the PPIase-Arm 1 interaction turns out to be a correct prediction of the simulations, it may have interesting implications. Because the flexible loop at the tip of Arm 1 is known to strongly interact with nascent chains (4,7), the interaction seen here suggests that the PPIase domain might hinder an emerging nascent chain’s attempts to gain access to the binding surface of Arm 1. Despite many years of study, the exact functional role played by the PPIase domain remains unclear (1). It has been shown, for example, to be nonessential for TF function in vivo (e.g., (29,30)) and has very recently been shown not to play a particularly important role in engaging with nascent peptides (9). In fact, it has even been shown that removal of the PPIase domain can actually lead to an increased efficiency of de novo folding and that this effect is connected with a decreased residence time of TF on the nascent chain (31). Examining how the presence of the putative PPIase:Arm 1 interface might affect a nascent chain’s interaction with TF, therefore, could constitute an interesting avenue for future research.

A key difference between the two 1.5-μs simulations is that OPLS predicts the formation of a compact conformation, while AMBER predicts more extended conformations: these differences can be nicely visualized in the histograms of radius of gyration values (see Fig. S1). The force-field swapping simulations that we have performed appear to indicate that these apparent differences between the two force fields may in fact be due primarily to sampling issues. In particular, the OPLS-AA/L force field appears quite capable of giving dynamic behavior and the AMBER ff99SB-ILDN force field appears capable of predicting a stable, compact conformation: notably, although the N-terminal-PPIase interaction appears to be clearly more dynamic when the simulation is restarted with AMBER, the PPIase-Arm1 interaction undergoes essentially no change at all. Although the simulations reported here are very long for systems of this size (376,000 atoms), much longer simulations are likely to be required to precisely determine the relative populations of the many various conformations that TF may adopt in solution. In particular, it is quite possible that there are still other energetically favorable conformations that TF might adopt that have not been sampled in any of the 4 μs of simulation time recorded here.

Despite this word of caution, we think that the fact that two quite different conformational states are observed at all merits further discussion. In particular, it is interesting to speculate whether there may be a connection between the very compact conformation seen in the OPLS simulation and that reported by Kaiser et al. as “the compact form” of TF that predominates when TF is in its ribosome-unbound state (17). The comparison of the computed and experimental FRET data shows that there is at least a qualitative agreement between the relative Et values of the different chromophore pairs, and that such an agreement cannot be obtained from an analysis of the PDB:1W26 crystal structure. Obviously, however, a more direct comparison with experiment would require that the chromophores be explicitly modeled in the MD simulations rather than added, as here, after the fact; a number of simulation studies have already been reported that have explicitly computed FRET efficiencies for other proteins in this manner (32–35).

One interesting characteristic feature of the highly compact conformation seen in the OPLS simulation is the formation of an interface between the N-terminal domain and the flexible loop at the tip of Arm1 (Fig. 7 c). Although again somewhat speculative, it is worth noting that this provides a potential rationalization for the observation by Kaiser et al. that “binding to the ribosome conformationally activates TF for nascent-chain association” (17), by inducing it to switch from a compact conformation to an extended conformation. In particular, it does not seem unreasonable to imagine that engagement of the ribosome-binding loop with the ribosome, and the concomitant conformational rearrangement of the loop (15,16), might disrupt the putative interface between the loop and Arm 1, releasing the latter to interact with an emerging nascent chain.

Complicating somewhat what might otherwise be quite a neat story is the fact that the highly compact conformation seen in the OPLS simulation is also characterized by an interaction between the N-terminal and the PPIase domains (Fig. 7 d). The existence of this interface is, at first sight, much more difficult to square with the study of Kaiser et al. as those authors reported that the ribosome binding kinetics of a TF construct from which the PPIase domain had been deleted were very similar to those of the full-length TF, although data were not explicitly shown (17). It may, therefore, be important to note that even though the ribosome-binding loop is involved simultaneously in interactions with both Arm 1 and the PPIase domain in the OPLS simulation, it nevertheless retains exposed basic residues that could serve as an initial recognition site by the ribosome. In Fig. S10 we show a comparison of the electrostatic potentials around TF when in its initial conformation and when in the conformation found at the end of the 1.5-μs OPLS simulation. Importantly, despite the enormous difference in the overall conformation of TF, the base of the N-terminal domain that is responsible for binding to the ribosome retains a significant positive electrostatic potential. In terms of TF’s electrostatic interactions, therefore, it does not appear that an interaction between the N-terminal domain and the PPIase domain would preclude the former from undergoing an initial recognition event by the ribosome; subsequent transformation into a more tightly bound configuration would, however, certainly require disruption of the putative N-terminal:PPIase domain interaction.

In any case, however, it appears that formation of the putative N-terminal:PPIase domain interface is not absolutely necessary for formation of the highly compact conformation seen in the OPLS simulation. Fig. S11 illustrates how the abrupt decrease in the distance between residues 14 and 150 correlates with the formation of the various putative interfaces that emerge during the simulation. From this it can be seen that the first new interface to form is that between the PPIase domain and the residues at the tip of Arm 1 and that this is accompanied by a large decrease in the distance between residues 14 and 150. The next interface to form is that between the N-terminal domain and the residues at the tip of Arm 1; this is accompanied by a further, but much less pronounced decrease in the 14–150 distance. The final interface to form is that between the N-terminal domain ribosome-binding loop and the PPIase domain; formation of this interface, however, is uncorrelated with any further change in the 14–150 distance. It appears, therefore, that the interaction between the N-terminal and PPIase domains follows opportunistically from Arm 1’s interaction with the PPIase domain and from the N-terminal domain’s interaction with Arm 1.

Interestingly, the view of TF’s conformational flexibility that is obtained here appears to be largely consistent with that reported in another MD simulation study of TF’s conformational dynamics that was published while this article was under revision (Singhal et al. (36)). In their study, Singhal et al. carried out 12 independent 250-ns MD simulations of TF: eight using the AMBER03 force field (37), and four using the same OPLS-AA/L force field employed here. They reported seeing a range of behaviors, with six out of 12 simulations resulting in a fully collapsed state, five resulting in a semicollapsed state, and one final simulation resulting in a collapsed and deformed state. Importantly, as was found to be the case here, the formation of contacts between the PPIase domain and Arm 1 was a consistent feature of all of their simulations. Interestingly, however, it was in the AMBER simulations, not the OPLS simulations, that the fully collapsed conformation was formed, with a feature of this state being the repeated formation of contacts between the N-terminal domain and the Arm 2 domain. The N-terminal domain residues involved in those contacts appear to be distinct from those involved in the interaction with Arm 1 that is observed in the (longer) OPLS simulation reported here. Although the details of the two studies differ somewhat, therefore, the same overall conclusions are obtained—namely, that TF is capable of very large-scale changes in conformation due to hinge-like motions of the residues connecting its structured domains.

A full exploration of any relationship between the putative interfaces observed by us and by Singhal et al. (36) with TF’s function would require that corresponding simulations be performed of TF in complex with a ribosome-nascent-chain complex. Such studies would in particular provide an important opportunity to rationalize the interesting decreases in FRET efficiencies that occur when TF engages with translating ribosomes and that have been interpreted as indicating that it forms a more extended conformation when ribosome-bound (17). Our attempts to connect FRET efficiencies calculated from the AMBER simulation of TF alone with those measured for TF in complex with ribosomes have met, perhaps not surprisingly, without much success. Given that simulations of TF bound to a ribosome-nascent-chain complex are already underway in another laboratory (Profs. Klaus Schulten, University of Illinois at Urbana-Champaign and Roland Beckmann, Ludwig Maximilian University of Munich, 2013; personal communication), it may soon be possible to see a more direct comparison with the experimental FRET measurements of TF in its ribosome-bound state.

Another area where further MD simulations of TF might be helpful would be to examine the conformational dynamics of a TF dimer in its apo state, because it has been shown that TF exists in a weak monomer-dimer equilibrium with an apparent Kd of ∼1–2 μM (38,39). Unfortunately, while there is a 3.5 Å resolution crystal structure of dimeric T. maritima TF in complex with the ribosomal protein S7 (10), this structure is unlikely to represent a substrate-free TF dimer as it is clearly stabilized by extensive interactions with S7. A low-resolution putative model of the TF dimer has been constructed by Kaiser et al. (17) based on intermolecular FRET measurements, but a high-resolution structure suitable as a starting point for MD simulation has yet to appear.

One final implication of both the results reported here and those very recently reported by Singhal et al. (36) concerns developing a mechanistic understanding of TF’s functional behavior using molecular simulation techniques. Of special interest is likely to be modeling TF’s effects on the structure and dynamics of emerging nascent chains. A number of coarse-grained simulation studies of cotranslational folding events on the ribosome have already been reported (8,40–42), with the most recent being an interesting attempt to explicitly model the potential effects of TF in modulating the folding behavior of nascent chains while still attached to the ribosome (8); evidence in support of some of the findings of the latter study has come from recent experimental work (10). The flexibility exhibited by TF in the coarse-grained simulations appears to have been much more modest than that observed here. In the future, given that a very high degree of interdomain flexibility such as that seen in the AMBER simulation is likely to be functionally beneficial to TF in enabling it to engage a wide variety of substrates (11), e.g., by altering the disposition of its Arm domains, it may well be worth extending such coarse-grained simulation studies to incorporate information obtained from more structurally resolved all-atom MD simulations. In particular, while the sampling issues encountered here suggest that a fully converged view of TF’s flexibility is likely to be challenging to obtain, it appears that this might be an important feature to include when modeling the dynamics of TF-substrate interactions.

Acknowledgments

The authors thank the anonymous reviewers for insightful suggestions for further analysis.

This work was made possible by the support of National Institutes of Health grant No. R01 GM099865 (to A.H.E.). This work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant No. OCI-1053575; generous allocations of time on the supercomputers Kraken and Ranger were awarded to A.H.E. under project No. TG-MCB090131.

Footnotes

Andrew S. Thomas’s present address is Department of Chemistry, Morningside College, Sioux City, IA 51105.

Supporting Material

Document S1. Additional sections with Materials and Methods as well as Protocols, thirteen figures, and references (43–60)
mmc1.pdf (1.3MB, pdf)
Movie S1. 1.5 microsecond MD simulation of trigger factor with the Amber ff99SB-ILDN force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (4.8MB, avi)
Movie S2. 1.5 microsecond MD simulation of trigger factor with the OPLS-AA/L force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (4.1MB, avi)
Movie S3. 0.5 microsecond MD simulation of trigger factor restarted with the Amber ff99SB-ILDN force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (1.3MB, avi)
Movie S4. 0.5 microsecond MD simulation of trigger factor restarted with the OPLS-AA/L force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (1.6MB, avi)
Movie S5. Projection of the first 100 ns of the MD simulation performed with the Amber ff99SB-ILDN force field on to the dominant mode of motion
Download video file (3.2MB, avi)
Movie S6. Projection of the first 100 ns of the MD simulation performed with the OPLS-AA/L force field on to the dominant mode of motion
Download video file (3.2MB, avi)
Document S2. Article plus Supporting Material
mmc8.pdf (4.3MB, pdf)

References

  • 1.Hoffmann A., Bukau B., Kramer G. Structure and function of the molecular chaperone Trigger Factor. Biochim. Biophys. Acta. 2010;1803:650–661. doi: 10.1016/j.bbamcr.2010.01.017. [DOI] [PubMed] [Google Scholar]
  • 2.Agashe V.R., Guha S., Barral J.M. Function of trigger factor and DnaK in multidomain protein folding: increase in yield at the expense of folding speed. Cell. 2004;117:199–209. doi: 10.1016/s0092-8674(04)00299-5. [DOI] [PubMed] [Google Scholar]
  • 3.Raine A., Lovmar M., Ehrenberg M. Trigger factor binding to ribosomes with nascent peptide chains of varying lengths and sequences. J. Biol. Chem. 2006;281:28033–28038. doi: 10.1074/jbc.M605753200. [DOI] [PubMed] [Google Scholar]
  • 4.Lakshmipathy S.K., Tomic S., Etchells S.A. Identification of nascent chain interaction sites on trigger factor. J. Biol. Chem. 2007;282:12186–12193. doi: 10.1074/jbc.M609871200. [DOI] [PubMed] [Google Scholar]
  • 5.Rutkowska A., Mayer M.P., Bukau B. Dynamics of trigger factor interaction with translating ribosomes. J. Biol. Chem. 2008;283:4124–4132. doi: 10.1074/jbc.M708294200. [DOI] [PubMed] [Google Scholar]
  • 6.Merz F., Boehringer D., Deuerling E. Molecular mechanism and structure of Trigger Factor bound to the translating ribosome. EMBO J. 2008;27:1622–1632. doi: 10.1038/emboj.2008.89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Lakshmipathy S.K., Gupta R., Hartl F.-U. Versatility of trigger factor interactions with ribosome-nascent chain complexes. J. Biol. Chem. 2010;285:27911–27923. doi: 10.1074/jbc.M110.134163. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.O’Brien E.P., Christodoulou J., Dobson C.M. Trigger factor slows co-translational folding through kinetic trapping while sterically protecting the nascent chain from aberrant cytosolic interactions. J. Am. Chem. Soc. 2012;134:10920–10932. doi: 10.1021/ja302305u. [DOI] [PubMed] [Google Scholar]
  • 9.Hoffmann A., Becker A.H., Kramer G. Concerted action of the ribosome and the associated chaperone trigger factor confines nascent polypeptide folding. Mol. Cell. 2012;48:63–74. doi: 10.1016/j.molcel.2012.07.018. [DOI] [PubMed] [Google Scholar]
  • 10.Martinez-Hackert E., Hendrickson W.A. Promiscuous substrate recognition in folding and assembly activities of the trigger factor chaperone. Cell. 2009;138:923–934. doi: 10.1016/j.cell.2009.07.044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Niwa T., Kanamori T., Taguchi H. Global analysis of chaperone effects using a reconstituted cell-free translation system. Proc. Natl. Acad. Sci. USA. 2012;109:8937–8942. doi: 10.1073/pnas.1201380109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Ferbitz L., Maier T., Ban N. Trigger factor in complex with the ribosome forms a molecular cradle for nascent proteins. Nature. 2004;431:590–596. doi: 10.1038/nature02899. [DOI] [PubMed] [Google Scholar]
  • 13.Yao Y., Bhabha G., Dyson H.J. Structure discrimination for the C-terminal domain of Escherichia coli trigger factor in solution. J. Biomol. NMR. 2008;40:23–30. doi: 10.1007/s10858-007-9207-1. [DOI] [PubMed] [Google Scholar]
  • 14.Kristensen O., Gajhede M. Chaperone binding at the ribosomal exit tunnel. Structure. 2003;11:1547–1556. doi: 10.1016/j.str.2003.11.003. [DOI] [PubMed] [Google Scholar]
  • 15.Baram D., Pyetan E., Yonath A. Structure of trigger factor binding domain in biologically homologous complex with eubacterial ribosome reveals its chaperone action. Proc. Natl. Acad. Sci. USA. 2005;102:12017–12022. doi: 10.1073/pnas.0505581102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Schlünzen F., Wilson D.N., Fucini P. The binding mode of the trigger factor on the ribosome: implications for protein folding and SRP interaction. Structure. 2005;13:1685–1694. doi: 10.1016/j.str.2005.08.007. [DOI] [PubMed] [Google Scholar]
  • 17.Kaiser C.M., Chang H.-C., Barral J.M. Real-time observation of trigger factor function on translating ribosomes. Nature. 2006;444:455–460. doi: 10.1038/nature05225. [DOI] [PubMed] [Google Scholar]
  • 18.Hess B., Kutzner C., Lindahl E. GROMACS 4: algorithms for highly efficient, load-balanced, and scalable molecular simulation. J. Chem. Theory Comput. 2008;92:435–447. doi: 10.1021/ct700301q. [DOI] [PubMed] [Google Scholar]
  • 19.Hornak V., Abel R., Simmerling C. Comparison of multiple AMBER force fields and development of improved protein backbone parameters. Proteins. 2006;65:712–725. doi: 10.1002/prot.21123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Lindorff-Larsen K., Piana S., Shaw D.E. Improved side-chain torsion potentials for the AMBER ff99SB protein force field. Proteins. 2010;78:1950–1958. doi: 10.1002/prot.22711. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Horn H.W., Swope W.C., Head-Gordon T. Development of an improved four-site water model for biomolecular simulations: TIP4P-Ew. J. Chem. Phys. 2004;120:9665–9678. doi: 10.1063/1.1683075. [DOI] [PubMed] [Google Scholar]
  • 22.Kaminski G.A., Friesner R.A., Jorgensen W.L. Evaluation and reparameterization of the OPLS-AA force field for proteins via comparison with accurate quantum chemical calculations on peptides. J. Phys. Chem. B. 2001;105:6474–6487. [Google Scholar]
  • 23.Jorgensen W.L., Chandrasekhar J., Klein M.L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 1983;79:926–935. [Google Scholar]
  • 24.Martinez-Hackert E., Hendrickson W.A. Structures of and interactions between domains of trigger factor from Thermotoga maritima. Acta Crystallogr. D Biol. Crystallogr. 2007;63:536–547. doi: 10.1107/S090744490700964X. [DOI] [PubMed] [Google Scholar]
  • 25.Lindorff-Larsen K., Maragakis P., Shaw D.E. Systematic validation of protein force fields against experimental data. PLoS ONE. 2012;7:e32131. doi: 10.1371/journal.pone.0032131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Emekli U., Schneidman-Duhovny D., Haliloglu T. HingeProt: automated prediction of hinges in protein structures. Proteins. 2008;70:1219–1227. doi: 10.1002/prot.21613. [DOI] [PubMed] [Google Scholar]
  • 27.Amadei A., Linssen A.B.M., Berendsen H.J.C. Essential dynamics of proteins. Proteins. 1993;17:412–425. doi: 10.1002/prot.340170408. [DOI] [PubMed] [Google Scholar]
  • 28.Painter J., Merritt E.A. Optimal description of a protein structure in terms of multiple groups undergoing TLS motion. Acta Crystallogr. D Biol. Crystallogr. 2006;62:439–450. doi: 10.1107/S0907444906005270. [DOI] [PubMed] [Google Scholar]
  • 29.Genevaux P., Keppel F., Georgopoulos C. In vivo analysis of the overlapping functions of DnaK and trigger factor. EMBO Rep. 2004;5:195–200. doi: 10.1038/sj.embor.7400067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Kramer G., Patzelt H., Deuerling E. Trigger factor peptidyl-prolyl cis/trans isomerase activity is not essential for the folding of cytosolic proteins in Escherichia coli. J. Biol. Chem. 2004;279:14165–14170. doi: 10.1074/jbc.M313635200. [DOI] [PubMed] [Google Scholar]
  • 31.Gupta R., Lakshmipathy S.K., Hartl F.-U. Trigger factor lacking the PPIase domain can enhance the folding of eukaryotic multi-domain proteins in Escherichia coli. FEBS Lett. 2010;584:3620–3624. doi: 10.1016/j.febslet.2010.07.036. [DOI] [PubMed] [Google Scholar]
  • 32.VanBeek D.B., Zwier M.C., Krueger B.P. Fretting about FRET: correlation between κ and R. Biophys. J. 2007;92:4168–4178. doi: 10.1529/biophysj.106.092650. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Allen L.R., Paci E. Orientational averaging of dye molecules attached to proteins in Förster resonance energy transfer measurements: insights from a simulation study. J. Chem. Phys. 2009;131:065101. doi: 10.1063/1.3193724. [DOI] [PubMed] [Google Scholar]
  • 34.Deplazes E., Jayatilaka D., Corry B. Testing the use of molecular dynamics to simulate fluorophore motions and FRET. Phys. Chem. Chem. Phys. 2011;13:11045–11054. doi: 10.1039/c1cp20447e. [DOI] [PubMed] [Google Scholar]
  • 35.Speelman A.L., Muñoz-Losa A., Krueger B.P. Using molecular dynamics and quantum mechanics calculations to model fluorescence observables. J. Phys. Chem. A. 2011;115:3997–4008. doi: 10.1021/jp1095344. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Singhal K., Vreede J., Bolhuis P.G. Hydrophobic collapse of trigger factor monomer in solution. PLoS ONE. 2013;8:e59683. doi: 10.1371/journal.pone.0059683. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Duan Y., Wu C., Kollman P.A. A point-charge force field for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. J. Comput. Chem. 2003;24:1999–2012. doi: 10.1002/jcc.10349. [DOI] [PubMed] [Google Scholar]
  • 38.Patzelt H., Kramer G., Deuerling E. Three-state equilibrium of Escherichia coli trigger factor. Biol. Chem. 2002;383:1611–1619. doi: 10.1515/BC.2002.182. [DOI] [PubMed] [Google Scholar]
  • 39.Maier R., Eckert B., Schmid F.-X. Interaction of trigger factor with the ribosome. J. Mol. Biol. 2003;326:585–592. doi: 10.1016/s0022-2836(02)01427-4. [DOI] [PubMed] [Google Scholar]
  • 40.Elcock A.H. Molecular simulations of cotranslational protein folding: fragment stabilities, folding cooperativity, and trapping in the ribosome. PLOS Comput. Biol. 2006;2:e98. doi: 10.1371/journal.pcbi.0020098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.O’Brien E.P., Hsu S.-T.D., Dobson C.M. Transient tertiary structure formation within the ribosome exit port. J. Am. Chem. Soc. 2010;132:16928–16937. doi: 10.1021/ja106530y. [DOI] [PubMed] [Google Scholar]
  • 42.O’Brien E.P., Christodoulou J., Dobson C.M. New scenarios of protein folding can occur on the ribosome. J. Am. Chem. Soc. 2011;133:513–526. doi: 10.1021/ja107863z. [DOI] [PubMed] [Google Scholar]
  • 43.Joung I.S., Cheatham T.E., 3rd Determination of alkali and halide monovalent ion parameters for use in explicitly solvated biomolecular simulations. J. Phys. Chem. B. 2008;112:9020–9041. doi: 10.1021/jp8001614. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Åqvist J. Ion water interaction potentials derived from free-energy perturbation simulations. J. Phys. Chem. 1990;94:8021–8024. [Google Scholar]
  • 45.Chandrasekhar J., Spellmeyer D.C., Jorgensen W.L. Energy component analysis for dilute aqueous solutions of Li+, Na+, F−, and Cl− ions. J. Am. Chem. Soc. 1984;106:903–910. [Google Scholar]
  • 46.Essmann U., Perera L., Pederson L.G. A smooth particle mesh Ewald method. J. Chem. Phys. 1995;103:8577–8593. [Google Scholar]
  • 47.Nosé S. A unified formulation of the constant temperature molecular-dynamics methods. J. Chem. Phys. 1984;81:511–519. [Google Scholar]
  • 48.Hoover W.G. Canonical dynamics: equilibrium phase-space distributions. Phys. Rev. A. 1985;31:1695–1697. doi: 10.1103/physreva.31.1695. [DOI] [PubMed] [Google Scholar]
  • 49.Parrinello M., Rahman A. Polymorphic transitions in single-crystals—a new molecular-dynamics method. J. Appl. Phys. 1981;52:7182–7190. [Google Scholar]
  • 50.Hess B., Bekker H., Fraaije J.G.E.M. LINCS: a linear constraint solver for molecular simulations. J. Comput. Chem. 1997;18:1463–1472. [Google Scholar]
  • 51.Gasteiger E., Hoogland C., Bairoch A. Protein identification and analysis tools on the ExPASy Server. In: Walker J.M., editor. The Proteomics Protocols Handbook. Humana Press; New York: 2005. [Google Scholar]
  • 52.Baker N.A., Sept D., McCammon J.A. Electrostatics of nanosystems: application to microtubules and the ribosome. Proc. Natl. Acad. Sci. USA. 2001;98:10037–10041. doi: 10.1073/pnas.181342398. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Sitkoff D., Sharp K.A., Honig B. Accurate calculation of hydration free energies using macroscopic solvent models. J. Phys. Chem. 1994;98:1978–1988. [Google Scholar]
  • 54.Humphrey W., Dalke A., Schulten K. VMD: visual molecular dynamics. J. Mol. Graph. 1996;14:33–38. doi: 10.1016/0263-7855(96)00018-5. 27–28. [DOI] [PubMed] [Google Scholar]
  • 55.Schüttelkopf A.W., van Aalten D.M.F. PRODRG: a tool for high-throughput crystallography of protein-ligand complexes. Acta Crystallogr. D Biol. Crystallogr. 2004;60:1355–1363. doi: 10.1107/S0907444904011679. [DOI] [PubMed] [Google Scholar]
  • 56.Förster T. Intermolecular energy migration and fluorescence [Zwischenmolekulare energiewanderung und fluoreszenz] Ann. Phys. 1948;2:55–75. [Google Scholar]
  • 57.Odom O.W., Deng H.-Y., Hardesty B. Binding of S21 to the 50S subunit and the effect of the 50S subunit on nonradiative energy transfer between the 3′ end of 16S RNA and S21. Biochemistry. 1984;23:5069–5076. doi: 10.1021/bi00316a037. [DOI] [PubMed] [Google Scholar]
  • 58.van der Meer B.W. Kappa-squared: from nuisance to new sense. J. Biotechnol. 2002;82:181–196. doi: 10.1016/s1389-0352(01)00037-x. [DOI] [PubMed] [Google Scholar]
  • 59.Bahar I., Atilgan A.R., Erman B. Direct evaluation of thermal fluctuations in proteins using a single-parameter harmonic potential. Fold. Des. 1997;2:173–181. doi: 10.1016/S1359-0278(97)00024-2. [DOI] [PubMed] [Google Scholar]
  • 60.Painter J., Merritt E.A. TLSMD web server for the generation of multi-group TLS models. J. Appl. Cryst. 2006;39:109–111. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Additional sections with Materials and Methods as well as Protocols, thirteen figures, and references (43–60)
mmc1.pdf (1.3MB, pdf)
Movie S1. 1.5 microsecond MD simulation of trigger factor with the Amber ff99SB-ILDN force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (4.8MB, avi)
Movie S2. 1.5 microsecond MD simulation of trigger factor with the OPLS-AA/L force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (4.1MB, avi)
Movie S3. 0.5 microsecond MD simulation of trigger factor restarted with the Amber ff99SB-ILDN force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (1.3MB, avi)
Movie S4. 0.5 microsecond MD simulation of trigger factor restarted with the OPLS-AA/L force field. Coloring scheme is as used in the main text. All snapshots have been aligned using the N-terminal domain backbone atoms (residues 1–110)
Download video file (1.6MB, avi)
Movie S5. Projection of the first 100 ns of the MD simulation performed with the Amber ff99SB-ILDN force field on to the dominant mode of motion
Download video file (3.2MB, avi)
Movie S6. Projection of the first 100 ns of the MD simulation performed with the OPLS-AA/L force field on to the dominant mode of motion
Download video file (3.2MB, avi)
Document S2. Article plus Supporting Material
mmc8.pdf (4.3MB, pdf)

Articles from Biophysical Journal are provided here courtesy of The Biophysical Society

RESOURCES