Collision-Induced Dissociation of Electrosprayed Protein Complexes

May 24, 2016 - Collision-induced dissociation (CID) of these multiply protonated gaseous ions usually culminates in ejection of a single subunit with ...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/JPCB

Collision-Induced Dissociation of Electrosprayed Protein Complexes: An All-Atom Molecular Dynamics Model with Mobile Protons Vlad Popa,† Danielle A. Trecroce,‡ Robert G. McAllister,‡ and Lars Konermann*,†,‡ †

Department of Biochemistry, The University of Western Ontario, London, ON N6A 5C1, Canada Department of Chemistry, The University of Western Ontario, London, ON N6A 5B7, Canada

J. Phys. Chem. B 2016.120:5114-5124. Downloaded from pubs.acs.org by UNIV OF MASSACHUSETTS AMHERST on 09/25/18. For personal use only.

‡

S Supporting Information *

ABSTRACT: Electrospray ionization mass spectrometry (ESI-MS) has become an indispensable technique for examining noncovalent protein complexes. Collision-induced dissociation (CID) of these multiply protonated gaseous ions usually culminates in ejection of a single subunit with a disproportionately large amount of charge. Experiments suggest that this process involves subunit unfolding prior to separation from the residual complex, as well as H+ migration onto the unravelling chain. Molecular dynamics (MD) simulations are a promising avenue for gaining detailed insights into these CID events. Unfortunately, typical MD algorithms do not allow for mobile protons. Here we address this limitation by implementing a strategy that combines atomistic force fields (such as OPLS/AA and CHARMM36) with a proton hopping algorithm, focusing on the tetrameric complexes transthyretin and streptavidin. Protons are redistributed over all acidic and basic sites in 20 ps intervals, subject to an energy function that reflects electrostatic interactions and proton affinities. Our simulations predict that nativelike conformers at the onset of collisional heating contain multiple salt bridges. Collisional heating initially causes subtle structural changes that lead to a gradual decline of these zwitterionic patterns. Many of the MD runs show gradual unfolding of a single subunit in conjunction with H+ migration, culminating in subunit separation from the complex. However, there are also instances where two or more chains start to unfold simultaneously, giving rise to charge competition. The scission point where the “winning” subunit separates from the complex can be attained for different degrees of unfolding, giving rise to product ions in various charge states. The simulated product ion distributions are in close agreement with experimental CID data. Proton enrichment in the departing subunit is driven by charge−charge repulsion, but the combination of salt bridge depletion, charge migration, and proton affinity causes surprising compensation effects among the various energy terms. It appears that this work provides the most detailed account to date of the mechanism whereby noncovalent protein complexes disassemble during CID.

■

complex, [M4 + 14H]14+ transthyretin (TTR) releases M7+ or M8+, as well as the complementary charge-depleted trimers.28 Here, the monomeric CID products carry roughly half of the total charge but only a quarter of the total mass. The same behavior is seen when applying other slow heating methods.12 A framework has emerged that provides a phenomenological explanation for this “asymmetric charge partitioning” of collisionally activated complexes.12,13,22,24,26,29,30 The concept envisions that the analyte initially retains a compact structure where excess protons are spread over the surface. Collisional heating induces unfolding of a single subunit. Subsequently, this subunit is ejected from the complex. Unfolding is accompanied by proton migration toward the unravelling subunit. This intersubunit charge transfer takes place until the scission point is reached, i.e., until the departing subunit separates from the complex.12,13,22,24,26,29,30 The described framework is consistent with the known high mobility of

INTRODUCTION Many proteins in solution form noncovalent multisubunit complexes. Various methods are available for probing the structure and composition of these assemblies.1−4 One widely used approach is the transfer of protein complexes into the gas phase by electrospray ionization (ESI) for analysis by mass spectrometry (MS).5−10 Electrosprayed complexes can by further interrogated by various activation and fragmentation techniques11−13 and by ion mobility spectrometry (IMS).14−18 Most experiments are conducted in positive ion mode, where net charge on the analyte is imparted by excess protons.19,20 Of particular interest is the behavior of gaseous protein complexes during collision-induced dissociation (CID). CID is a slow heating process that involves thousands of collisions with background gas. 21−23 When applied to protein complexes, the CID settings are usually adjusted such that only noncovalent interactions are disrupted, while covalent bonds remain intact. Under these conditions most complexes eject a single subunit with a disproportionally large amount of charge.12,22,24−27 As a typical example of a homotetrameric © 2016 American Chemical Society

Received: March 23, 2016 Revised: April 30, 2016 Published: May 24, 2016 5114

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B protons in collisionally activated biomolecules,31−33 and it is supported by the observation of semiunfolded intermediates in IMS experiments.13,34 Many facets of the CID process nonetheless remain poorly understood. These include the interplay between unfolding and charge migration, as well as the structures of intermediate species.13,35 Alternative models that envision completely different mechanistic avenues have been put forward as well.36 Molecular dynamics (MD) simulations can provide detailed insights into the behavior of biomolecular systems.37,38 MD investigations on gaseous proteins are particularly attractive, because the absence of solvent implies considerable savings in computer time. The gas-phase protein community has made extensive use of MD techniques for exploring the structure and dynamics of desolvated proteins.39−45 A limitation of classical MD simulations is the fact they cannot account for H+ transfer events. This is a particular problem for studies on collisionally heated gaseous proteins, where protons are highly mobile.31−33 Realistic proton simulations call for quantum mechanical/molecular mechanical (QM/MM),46,47 ab initio MD,48−50 or density functional theory (DFT)/MD methods51 that are not practical for studying large proteins. Simplified proton-transfer schemes have been proposed, but most of these are geared toward solution-phase simulations52−54 or they remain limited to relatively small systems.47 Thachuk and Fegan30 recently developed a simple charge hopping strategy for modeling the CID process of protein complexes. Charge was redistributed over basic sites during the MD runs. The charge patterns were governed by the tendency of the system to minimize the overall Coulomb energy, with an additional term that accounts for the proton affinity (PA) of the various sites. Focusing on TTR, those simulations30 confirmed the expected behavior12,13,22,24,26,29 where thermal excitation triggered unfolding of a single subunit with concomitant charge accumulation, followed by ejection from the complex. Thachuk and Fegan’s charge hopping work30 represents an important advance. However, the method was implemented only for the coarse-grained Martini force field. Compared to all-atom models, coarse-grained force fields are greatly simplified because groups of atoms are mapped together into single beads.55 Hydrogens are not considered explicitly. Protonation and deprotonation are modeled by modifying charges on beads without removing or adding any atoms.30 Coarse-grained modeling is a powerful approach for studying very large systems in solution on very long time scales.55 However, the predictive power of coarse-grained methods tends to be lower than for all-atom models.56 The lack of solvent for gaseous proteins makes these systems readily accessible to all-atom simulations. Therefore, it is highly desirable to extend a charge hopping approach of the type described by Thachuk and Fegan30 to all-atom force fields. Another limitation of the method developed in ref 30 is the use a “positive-only” strategy that exclusively considers protonation changes of basic sites (NT0/+, Arg0/+, Lys0/+, and His0/+). Acidic groups were kept permanently protonated: Asp0, Glu0, and CT0 (NT represents N-termini; CT stands for C-termini). This approach reflects the view that the high gasphase proton affinity of R-COO− will favor the neutralization of carboxylates.39,42,57 However, there is strong evidence that electrosprayed biomolecules can have zwitterionic properties and salt bridges.36,58−63 Improved CID models should

therefore allow for protonation changes of basic and acidic sites. The present study extends the work of Thachuk and Fegan 30 by moving from coarse-grained to all-atom simulations of CID processes. We implement a proton hopping algorithm for redistributing charge in response to conformational changes. TTR and streptavidin serve as model systems. Both of these homotetrameric complexes form charge states around 14+ during native ESI, and their CID behavior has been well documented experimentally.28,64,65 The simulations are conducted by considering both positive-only and zwitterionic scenarios. Our data support the proposed unfolding/charge migration framework,12,13,22,24,26,29,30 while providing novel insights into the energetics and conformational changes of protein complexes during CID.

■

SIMULATION METHODS General Aspects. Protein simulations in a vacuum environment employed GROMACS 5.1 for leapfrog integration of Newton’s equations with a 1 fs time step.66,67 Bonds were constrained using LINCS.68 All runs were conducted on quad core desktop Linux workstations with GPU acceleration. Electrostatic and van der Waals interactions were modeled without cut-offs.69 Temperature control was achieved using the Nosé−Hoover thermostat70 with a coupling constant of 0.5 ps. Test runs employing a velocity rescaling thermostat71 resulted in artifactual accumulation of angular momentum. Resorting to the Nosé−Hoover scheme eliminated this problem. Center of mass translations were eliminated throughout the runs. Unless noted otherwise, the simulations employed the OPLS all-atom (OPLS/AA) force field.72 Additional runs were performed using the CHARMM36 allatom force field.73 TTR simulations used the X-ray structure 3GRG as starting point.74 All heteroatoms were removed from the pdb file prior to processing it for GROMACS using PDB2GMX. Due to chain disorder, the X-ray data do not show the positions of the first 9 N-terminal and the last 2 Cterminal residues in the TTR chains. These residues were added in extended conformations using the Pymol editor. Subsequent relaxation was performed for 100 ps at 1 K. Conformational dynamics of the termini during the early stages of the production runs (described below) quickly erased any memory of the initial extended conformations. To allow comparisons with experimental data28 TTR simulations were run on [M4 + 14H]14+ ions. An analogous strategy was used for streptavidin simulations, based on the X-ray structure 4Y5D.75 All runs were repeated at least five times with different starting conformations, as well as different initial atom velocities. Proton-Moving Algorithm. We implemented an algorithm for distributing H+ among possible protonation sites (NT0/+, Arg0/+, Lys0/+, His0/+, Asp0/−, Glu0/−, and CT0/−). Interactions of these charge sites give rise to a Coulomb energy, ECoul, that is given by ECoul =

1 4πε0κe

N

N

∑ ∑ i=1 j=i+1

qiqj rij

(1)

where qi and qj are the protonation-dependent charges (−e, 0, or e), with distances rij.30 Unless noted otherwise, a dielectric constant of κe = 1 was used.30,76 An additional term arises from the proton affinities of the various sites (PANT = 960 kJ mol−1, PAARG = 1029 kJ mol−1, PALYS = 937 kJ mol−1, 5115

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B

Figure 1. Illustration of the proton redistribution algorithm for gaseous TTR14+. Subunits are shown in different colors, with positive and negative charges as white and red spheres, respectively. (a) Manually selected, highly improbable protonation pattern in positive-only mode; (b) pattern after energy minimization. (c) Highly improbable pattern in zwitterionic mode; (d) optimized pattern after energy minimization. Tables display the charges on individual sites; the chemical nature of each site is denoted along the top, in order of sequence.

PAHIS = 958 kJ mol−1, PAASP‑ = 1450 kJ mol−1, PAGLU‑ = 1450 kJ mol−1, PACT‑ = 1450 kJ mol−1).30,77 These PA values give rise to the energy EPA

E PA = −∑ PA k k

protonation patterns involved a Monte Carlo strategy.30 However, data obtained in this way were virtually indistinguishable from those generated by always retaining the pattern with the lowest Etot. The simulations discussed below were thus generated without invoking Monte Carlo techniques. Similar to earlier work,76 our algorithm uses a point charge approach where protonation−deprotonation is attributed to the following sites (using OPLS/AA notation):72 N-terminal nitrogen for NT+, NH2 for Arg+, NZ for Lys+, CB for His+, C for CT−, OD2 for Asp−, and OE2 for Glu−. Neutral His was modeled in the Nε2-H form, rather than Nδ1-H, consistent with the prevalence of the former in X-ray structures.78 Mobile Proton MD Simulations. Simulations were run as a succession of brief MD segments, each of which used a static protonation pattern. These segments were interleaved with proton redistribution events at regular time intervals (Δtredist), using the algorithm outlined above. Each segment started with a steepest descent energy minimization for eliminating possible atom clashes that might arise from proton hopping. Subsequently, velocities were reassigned from a

(2)

where summation includes only protonated sites. These two terms result in a “total” energy30 Etot = E PA + ECoul

(3)

The premise of our model is that mobile protons will adopt a spatial pattern that minimizes Etot.29,30,39 A search algorithm was developed that numbers protonation sites according to sequence. The proton from the first occupied site is tentatively moved to each of the unoccupied sites. This is repeated for all occupied sites. The pattern producing the lowest Etot is retained, provided that it is lower than the previous minimum. Once all single proton moves have been tested, the process starts over. This sequence is repeated until the search algorithm fails to find a new Etot minimum. During the early stages of this project the acceptance of new 5116

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B

Figure 2. MD snapshots representing different time points during the collision-induced dissociation of TTR14+ in positive-only mode (a−d) and in zwitterionic mode (e−h).

Unfortunately, the astronomical number of combinations60 N!/((N − n)! × n!) in which n protons can be distributed over N possible protonation sites precludes such an analysis. The search algorithm used here typically converges after testing ∼2 × 105 patterns that can be computed within a few seconds. We first illustrate the algorithm for nativelike TTR14+, without MD simulations. Figure 1a represents a highly improbable positive-only distribution, with charges 14+/0/0/ 0 on subunits A/B/C/D. Symmetry considerations suggest that after optimization each chain should carry roughly the same charge. The output of the algorithm confirms this expectation, generating a 3+/3+/4+/4+ pattern that lowers the energy by |ΔEtot| = 4412 kJ mol−1 (Figure 1b). Similarly, a zwitterionic 17+/14+/7−/10− distribution is turned into 4+/ 3+/3+/4+, with |ΔEtot| = 27 500 kJ mol−1 (Figure 1c,d). Considering the vast parameter space, it is unlikely the algorithm will find the global Etot minimum. Nonetheless, Figure 1 illustrates that our steepest descent approach quickly converges on approximate solutions that represent physically reasonable protonation patterns. It is instructive to consider the zwitterionic scenario in more detail (Figure 1c,d). Charge redistribution causes neutralization of many carboxylates, consistent with their high proton affinity.77 However, the algorithm retains a considerable number of R-COO− moieties that are involved in salt-bridges such as +/− or +/−/+ (Figure 1d). The presence of such motifs is consistent with studies on various gaseous ions.36,58−61 For a closely spaced acceptor/donor pair a +/− salt-bridged structure is more favorable than a neutralized 0/0 arrangement. Similarly, a +/−/+ scenario has a lower Etot than +/0/0, provided that the titratable sites are in close proximity

Maxwell−Boltzmann distribution at the desired temperature. To mimic protein heating during CID21−23 the temperature was increased by 50 K after every 4 ns, starting at 350 K, until one of the subunits separated from the complex. For picking a suitable Δtredist one can consider that transition-state theory suggests a lower limit on the order of Δtredist ≈ (kBT/h)−1 ≈ 0.2 ps for an ideal barrier-free process.79 Solution-phase proton-transfer barriers are on the order of kBT, corresponding to hopping times on a picosecond time scale.48 The absence of solvent in gaseous proteins may cause barriers to be somewhat higher, with correspondingly longer hopping times. With this in mind, we conducted test runs with Δtredist values of 5, 20, and 100 ps. No discernible differences were found between these three conditions (data not shown). For the data discussed below we used Δtredist = 20 ps. To properly account for the high H+ mobility in thermally excited protein ions12,31−33 we did not impose a maximum proton-transfer distance. The underlying idea is that a single long-range H+ transfer from an occupied site H to a vacant site A is equivalent to a series of short-range hops taking place via a relay mechanism.80 For example, H-H-H-A → A-H-H-H is equivalent to H-H-H-A → H-H-A-H → H-A-H-H → A-HH-H.

■

RESULTS AND DISCUSSION Performance of the Proton Redistribution Algorithm. The simulations discussed below require an efficient algorithm for identifying energetically favorable protonation patterns within protein complexes of arbitrary structure. The conceptually simplest approach for determining the lowest possible Etot (eq 3) would be to test all possible patterns. 5117

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B

Figure 3. Time-dependent summary for typical TTR14+ CID simulations in positive-only mode (a−e) and in zwitterionic mode (f−j), referring to the data of Figure 2. Vertical dotted lines indicate the scission point, where a highly charged and extended subunit separates from the complex. Panels a and f: Net charge on individual subunits. The charge of the departing monomer at the scission point is indicated. Panels b and g: Radius of gyration (Rg) of individual subunits. Panels c and h: Total number of positive and negative charges in the complex. Panels d and i: Contributions to ECoul, separated according to the type of charge−charge interaction. Panels e and j: EPA and Etot.

accumulation on subunit B, culminating in 8+ at the scission point (see also Supporting Information, Movie 1). Similar events take place in zwitterionic mode, illustrated in Figure 2e−h, which shows ejection of a highly extended subunit C as an 8+ ion (Supporting Information, Movie 2). We found no evidence of preferential ejection for any particular subunit, consistent with the symmetric structure of native TTR. This is different from the simulations of ref 30 where subunit A was primed for ejection by making it the only chain with a chargeable N-terminus. Figure 3 presents a closer look at the CID events in positive-only and in zwitterionic mode. The two top panels of Figure 3 show the charge on each subunit, while the second row displays radii of gyration (Rg). Comparison of these

to one another (see Figure S1 for a detailed analysis). Clearly, eq 3 represents a relatively crude tool for predicting zwitterionic versus neutralized contacts. In-depth analyses would require DFT methods,61,63 but the computational cost of such endeavors precludes their use in the present case. The salt bridge patterns used here thus represent semiquantitative approximations. Despite these limitations, the ability to conduct simulations on zwitterionic proteins represents a step beyond earlier positive-only models.30,39,42 Collision-Induced Dissociation of TTR. Figure 2a−d illustrates a typical CID run in positive-only mode. After 12 ns, subunit B (blue) starts to protrude with its positively charged N-terminus. The subunit gradually unravels but remains associated with the complex until separation occurs at t ≈ 18 ns. Unraveling is accompanied by charge 5118

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B

Figure 4. Structures of subunit B in TTR14+ at 0 ns (a) and at 15.10 ns (b). Secondary structure (β strand) detection was performed using Pymol. Positive and negative charges are shown as white and red spheres, respectively.

Figure 5. Energetics of charge arrangements in a two-dimensional toy model. (a) Charges 1 and 2 are in contact with a stable salt bridge. (b) Charges 1 and 2 interact with an unstable salt bridge. (c) Same geometry as in panel B, but the salt bridge has been eliminated by proton transfer. (d) Same as panel C, except that the distance between charges 1 and 2 has increased. Scenarios a and d share the same Etot. Energies were calculated using eqs 1−3, with PA values for Lys and Glu−.

process does not cause major changes in EPA because the PA values of the various basic sites are similar to one another.30 Hence, the Etot profile is dominated by changes in ECoul (Figure 3e). The energetics of this positive-only scenario

profiles highlights the close correlation between subunit unfolding and charge accumulation. For the positive-only case, ECoul decreases significantly as protons spread into the unraveling subunit (Figure 3d). The 5119

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B confirm the expectation that H+ migration during CID is governed primarily by electrostatic repulsion.29,30,76 Surprisingly, under zwitterionic conditions, Etot remains almost constant throughout the CID process. This effect arises from the combination of an ECoul increase and an EPA decrease (Figure 3j). The origin of this behavior requires closer examination. As noted, the proximity of acidic and basic sites in nativelike conformations favors the formation of salt bridges (Figure 1c,d; see also Figure S1). Heating induces structural perturbations that render salt bridge formation less favorable. These events result in a gradual loss of zwitterionic character during heating, i.e., a steadily decreasing number of R-COO− sites. The decline of negative carboxylates is accompanied by a drop in the number of positive sites, keeping in mind that the overall TTR charge is fixed at 14+ (Figure 3h). Elimination of salt bridges (DH+ + A− → D + AH) lowers EPA, owing to the high PA of the acceptor sites Glu−, Asp−, and CT− (Figure 3j). At the same time, the decline of electrostatic attraction between opposite charges causes ECoul to increase. This interplay of EPA and ECoul results in Etot ≈ const (Figure 3i,j). The loss of native structure and zwitterionic character is highlighted in Figure 4; panel A shows subunit B at t = 0 with its native β sandwich and numerous salt bridges. After 15.1 ns, the structure is greatly perturbed, with retention of only one salt bridge (Figure 4b). The events seen in our MD simulations are consistent with IMS data, where structural changes are detectable long before subunit ejection takes place.64 For understanding why compact and extended structures can share the same Etot (Figure 3j), it is helpful to consider a minimalist model with four possible protonation sites and z = 2+. Figure 5a shows a compact “native state” where two positive charges (termed 1 and 2) interact with a salt bridge (Etot = −3295 kJ mol−1). In Figure 5b, a structural perturbation has repositioned the salt bridge and increased the donor−acceptor distance, causing Etot to increase. The perturbed system can lower Etot by salt bridge neutralization (Figure 5c); however, this new Etot remains less favorable than for the native state of Figure 5a. In Figure 5d, charges 1 and 2 are given the opportunity to separate from each other, e.g., as the result of subunit unfolding. The reduced electrostatic repulsion lowers Etot to −3295 kJ mol−1, which is identical to the value of Figure 5a. Overall, Figure 5a,d illustrates how a compact native state with salt bridges can have the same Etot as an extended state without salt bridges, reminiscent of the zwitterionic TTR scenario of Figure 3j. It is concluded that for zwitterionic TTR charge enrichment is driven by proton−proton repulsion, similar to the positiveonly scenario. However, the former case is complicated by the interplay of attractive−repulsive interactions and proton affinity effects. The overall electrostatics of the zwitterionic model are dominated by the loss of salt bridges during heating, masking the energetically favorable effects of charge accumulation in the unraveling subunit. Different Modes of Subunit Ejection. Not all simulation runs produced exactly the same CID behavior. This section illustrates the various types of events that were encountered in ten different OPLS/AA simulations on zwitterionic TTR14+. For most of these runs the initial step toward dissociation was the protrusion of a positively charged N-terminus. In addition, there were instances where a charged hairpin stuck out first. Roughly half of the runs followed a sequence similar to that depicted in Figure 2e−h, with

extensive unfolding of the departing chain prior to separation from the residual trimer. The other runs produced a range of scission point structures where the departing subunits retained more compact conformations. In one case subunit B started to unravel but then separated from the complex with very little further unfolding (Figure 6a). In another instance, subunit C showed extensive unfolding at the expense of contacts with chain A, such that the latter was the one to

Figure 6. (a−c) TTR structures just after subunit ejection from the 14+ tetramer. Each panel represents a different MD run conducted in zwitterionic mode. (d) Rg profiles for the competition scenario of panel C. (e) Charge of the ejected subunit vs degree of subunit unfolding, DU, for 9 TTR runs in zwitterionic mode. The solid line represents a linear regression (r2 = 0.93). 5120

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B

that proton redistribution was governed exclusively by electrostatic repulsion (Figure 7, first pair of bars). These conditions resulted in slightly higher z values, but the overall reaction sequence was similar to that of Figures 2a−d. The robustness of the TTR results for these different scenarios is reassuring. Also, the MD-generated monomer charge states are consistent with experiments, which show 7+ and 8+ monomers as the dominant CID products of TTR14+.28 To demonstrate the compatibility of our approach with different atomistic force fields we conducted additional TTR simulations in zwitterionic mode using CHARMM3673 (Figure 7, fourth set of bars). Compared to the OPLS/AA runs, the CHARMM36 data exhibited scission point structures that were more heterogeneous. Ejection of compact and extended chains took place with comparable probability (3/7 and 4/7 runs, respectively). The resulting average monomer charge is 5.5+ with a considerable standard deviation of ±2.5. Different structural propensities for CHARMM and OPLS/ AA have been noted previously.81 While CHARMM has been highly successful for modeling proteins in solution,37 the results of the current work suggest that OPLS/AA yields more consistent data for gas-phase CID simulations. After extensively characterizing the CID behavior of TTR with mobile protons, we also conducted OPLS/AA runs with a static 4+/3+/4+/3+ pattern, i.e., without charge hopping. This static pattern was generated in zwitterionic mode, analogously to the data of Figure 1d. Three out of five runs culminated in dimerization, with highly extended structures for both fragment ions. In two runs we observed separation of a compact monomer, with charge states 3+ and 4+ (Figure S2). Hence, ejection of a highly extended monomer is the dominant dissociation pathway only for mobile proton models. Also, simulations with static charges resulted in the highest TCID (averaged over all five runs, 690 K; Figure 7), suggesting that the presence of mobile protons lowers the activation barriers for dissociation. Coulombic interactions in a bulk medium are modulated by dielectric effects, reflecting orientational and electronic polarization phenomena. Charge interactions are much more pronounced in the gas phase (κe = 1) than in water (κe ≈ 80). There is some debate regarding the appropriate κe value for gas-phase proteins. As in previous work,30 we used κe = 1 for all preceding calculations, consistent with experimental data on gaseous peptides.82 Importantly, even with κe = 1, dielectric effects are explicitly considered in our model (albeit in a relatively crude way) because of the presence of zwitterionic/dipolar arrangements that interact with charge sites. Some earlier simulations42,83 used κe = 2, an estimate that originates from gas-phase basicity measurements on cytochrome c.57 For completeness, we conducted mobile proton OPLS/AA simulations on TTR14+ in zwitterionic mode with κe = 2. The reduction in electrostatic interactions destabilizes salt bridges (eq 1), causing the proton redistribution algorithm to eliminate all negative charges. Moreover, the diminished electrostatic repulsion lowers the propensity for subunit unfolding. As a result, all runs with κe = 2 culminated in ejection of compact monomers with an average charge of 4+ (Figure 7). The lack of unfolding and asymmetric charge partitioning under these conditions is not consistent with experimental observations.12,13,22,24,26 In other words, meaningful CID simulations for the model developed here require κe = 1.

separate (Figure 6b). Some of the runs displayed competition behavior, with two protruding chains that underwent unfolding in parallel (Figure 6c,d). The extent of charge accumulation on the ejected subunit is governed by its degree of unfolding, DU, relative to the residual trimer at the scission point. We define DU as Rg(departing chain)/ΣRg(all chains). A case where all four subunits are native (or equally unfolded) corresponds to DU = 0.25. Conversely, DU > 0.25 represents the case where the departing subunit is more unfolded than the three other chains. Figure 6e reveals a linear relationship between DU and the charge of the departing subunit. This correlation reflects the fact that extended chains have a larger capacity to accommodate excess protons. The observation of different DU values and various product charge states in our simulation is consistent with experiments, where CID always produces a range of z values rather than a single charge state.12,13,22,24,26,29,30 Versatility of the Mobile Proton Algorithm. Minor changes of the code allowed CID simulations to be conducted under a variety of conditions. Except for a single instance (one out of ten OPLS/AA runs on zwitterionic TTR), all mobile proton simulations showed the expected monomer− trimer separation. Figure 7 summarizes MD data for different scenarios by displaying the average charge on the ejected subunit, as well as the average temperature, TCID, at which separation took place.

Figure 7. TTR14+ and streptavidin14+ CID simulation results obtained under different conditions. Black bars represent the average charge state of the unfolded monomer at the scission point. Gray bars display the corresponding temperature. Only two of the five “TTR static charges” runs resulted in monomer ejection. Except for the “TTR zwitterionic CHARMM” data all results were generated with the OPLS/AA force field. Error bars represent standard deviations of 5−9 MD runs for each condition. Additional information is provided in the text.

The first three scenarios in Figure 7 refer to mobile proton TTR simulations using the OPLS/AA force field.72 The results obtained in positive-only mode and in zwitterionic mode are quite similar, with monomer charge states around 7+ and TCID between 500 and 550 K. A detailed discussion of the trajectories seen under these two conditions has been provided in the preceding sections. We also conducted positive-only runs that did not consider changes in EPA, such 5121

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B As a final step, we conducted mobile proton CID simulations on a noncovalent assembly other than TTR. Streptavidin represents one of the first complexes for which asymmetric charge partitioning was reported. Similar to TTR, experiments on streptavidin reveal the formation of 7+ and 8+ monomers as main CID products from 14+ tetramer ions.65 All five OPLS/AA runs of streptavidin in zwitterionic mode (with κe = 1) produced highly extended monomers. Subunit unraveling is favored by the presence of an N-terminal region that tends to carry two positive charges, owing to its high basicity with two Lys and the N-terminal amino group (Figure S3 and Supporting Information Movie 3). The simulated monomer charge states of 8 ± 1 agree well with experimental spectra65 (last data set Figure 7).

induced dissociation (SID)13,85 or mild solution-phase destabilization.86 In future work it will be interesting to extend simulations of the type described here to larger protein complexes that can undergo both collapse and unfolding.18,42 SID simulations will be another intriguing application,13 as well as studies on the mechanisms by which bound ligands and metal ions affect the dissociation behavior.28

■

ASSOCIATED CONTENT

S Supporting Information *

Supporting Figures and Movies as noted in the text. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.6b03035. Salt bridge energetics, TTR runs with static charges, and CID of streptavidin (PDF) CID of TTR14+ in positive-only mode (corresponds to Figure 2a−d) (mpg) CID of TTR14+ in zwitterionic mode (corresponds to Figure 2e−h) (mpg) CID of streptavidin14+ in zwitterionic mode (corresponds to Figure S2) (mpg)

■

CONCLUSIONS This study marks the first time that all-atom MD simulations were combined with a mobile proton model for examining the CID behavior of gaseous protein complexes. Our results are consistent with the view that asymmetric charge partitioning is caused by the interplay of subunit unfolding and charge migration, as previously suggested on the basis of experimental studies.12,13,22,24,26 Similar to earlier work,30 gas collisions were not explicitly modeled in our study; instead, temperature control was achieved via a thermostat. This approach greatly reduces the computational cost of the simulations.84 Experiments confirmed that asymmetric charge partitioning takes place for different slow heating methods (collisional excitation and blackbody radiation).12 This implies that the absence of an explicit collision gas is not a major shortcoming of the MD strategy used here. The product charge states generated in our simulations are in good agreement with experimental data. The results are also consistent with the coarse-grained MD simulations of Thachuk et al.30 We observed the ejection of unfolded monomeric chains in high charge states under various conditions, i.e., positive-only with and without accounting for proton affinity, as well as zwitterionic scenarios with two different force fields. It is reassuring that the observed behavior is largely model-independent. Nonetheless, zwitterionic OPLS/AA simulations appear most suitable for describing the CID process because they produced the most robust results while accounting for the known presence of salt bridges.36,58−63 Our simulations reveal that monomers can separate from the complex in various conformations, from relatively compact to highly extended. The corresponding DU values translate into different product charge states, consistent with experimental CID spectra that show a range of z values. A contributing factor to this charge state heterogeneity is the possibility that two or more chains can start to unfold simultaneously, causing competition for mobile protons. We are not aware of previous reports that have discussed such charge competition scenarios. For most MD runs the charge-depleted trimeric products adopt compact structures after monomer ejection. However, all subunits undergo extensive conformational changes during collisional heating. The interrogation of residual subcomplexes is therefore unlikely to provide information that is relevant to the solution structure of the protein. This last conclusion highlights the need to use methods other than CID for generating biologically relevant subcomplexes, such as surface-

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel: (519) 661-2111, ext. 86313. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Funding for this work was provided by the Natural Sciences and Engineering Research Council of Canada (Discovery Grant 217080-2013).

■

REFERENCES

(1) Alber, F.; Dokudovskaya, S.; Veenhoff, L. M.; Zhang, W. Z.; Kipper, J.; Devos, D.; Suprapto, A.; Karni-Schmidt, O.; Williams, R.; Chait, B. T.; Rout, M. P.; Sali, A. Determining the architectures of macromolecular assemblies. Nature 2007, 450, 683−694. (2) Kristensen, A. R.; Gsponer, J.; Foster, L. J. A high-throughput approach for measuring temporal changes in the interactome. Nat. Methods 2012, 9, 907−909. (3) Robinson, C. V.; Sali, A.; Baumeister, W. The molecular sociology of the cell. Nature 2007, 450, 973−982. (4) Bai, X.-C.; McMullan, G.; Scheres, S. H. W. How cryo-EM is revolutionizing structural biology. Trends Biochem. Sci. 2015, 40, 49− 57. (5) Dyachenko, A.; Gruber, R.; Shimon, L.; Horovitz, A.; Sharon, M. Allosteric mechanisms can be distinguished using structural mass spectrometry. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 7235−7239. (6) Uetrecht, C.; Heck, A. J. R. Modern Biomolecular Mass Spectrometry and its Role in Studying Virus Structure, Dynamics, and Assembly. Angew. Chem., Int. Ed. 2011, 50, 8248−8262. (7) Mehmood, S.; Allison, T. M.; Robinson, C. V. Mass Spectrometry of Protein Complexes: From Origins to Applications. Annu. Rev. Phys. Chem. 2015, 66, 453−474. (8) Kaddis, C. S.; Loo, J. A. Native Protein MS and Ion Mobility: Large Flying Proteins with ESI. Anal. Chem. 2007, 79, 1778−1784. (9) Benesch, J. L. P.; Ruotolo, B. T. Mass spectrometry: come of age for structural and dynamical biology. Curr. Opin. Struct. Biol. 2011, 21, 641−649. (10) Chen, F.; Gulbakan, B.; Weidmann, S.; Fagerer, S. R.; Ibanez, A. J.; Zenobi, R. Applying mass spectrometry to study non-covalent biomolecule complexes. Mass Spectrom. Rev. 2016, 35, 48−70.

5122

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B (11) Lermyte, F.; Konijnenberg, A.; Williams, J. P.; Brown, J. M.; Valkenborg, D.; Sobott, F. ETD Allows for Native Surface Mapping of a 150 kDa Noncovalent Complex on a Commercial Q-TWIMSTOF Instrument. J. Am. Soc. Mass Spectrom. 2014, 25, 343−350. (12) Felitsyn, N.; Kitova, E. N.; Klassen, J. S. Thermal Decomposition of a Gaseous Multiprotein Complex Studied by Blackbody Infrared Radiative Dissociation. Investigating the Origin of the Asymmetric Dissociation Behavior. Anal. Chem. 2001, 73, 4647− 4661. (13) Quintyn, R. S.; Zhou, M.; Yan, J.; Wysocki, V. H. SurfaceInduced Dissociation Mass Spectra as a Tool for Distinguishing Different Structural Forms of Gas-Phase Multimeric Protein Complexes. Anal. Chem. 2015, 87, 11879−11886. (14) Bernstein, S. L.; Dupuis, N. F.; Lazo, N. D.; Wyttenbach, T.; Condron, M. M.; Bitan, G.; Teplow, D. B.; Shea, J.-E.; Ruotolo, B. T.; Robinson, C. V.; Bowers, M. T. Amyloid-β protein oligomerization and the importance of tetramers and dodecamers in the aetiology of Alzheimer’s disease. Nat. Chem. 2009, 1, 326−331. (15) Ashcroft, A. E. Mass Spectrometry and the Amyloid Problem How Far Can We Go in the Gas Phase? J. Am. Soc. Mass Spectrom. 2010, 21, 1087−1096. (16) Servage, K. A.; Silveira, J. A.; Fort, K. L.; Clemmer, D. E.; Russell, D. H. Water-Mediated Dimerization of Ubiquitin Ions Captured by Cryogenic Ion Mobility-Mass Spectrometry. J. Phys. Chem. Lett. 2015, 6, 4947−4951. (17) Bush, M. F.; Hall, Z.; Giles, K.; Hoyes, J.; Robinson, C. V.; Ruotolo, B. T. Collision Cross Sections of Proteins and Their Complexes: A Calibration Framework and Database for Gas-Phase Structural Biology. Anal. Chem. 2010, 82, 9557−9565. (18) Hogan, C. J.; Ruotolo, B. T.; Robinson, C. V.; de la Mora, J. F. Tandem Differential Mobility Analysis-Mass Spectrometry Reveals Partial Gas-Phase Collapse of the GroEL Complex. J. Phys. Chem. B 2011, 115, 3614−3621. (19) Heck, A. J. R.; Van den Heuvel, R. H. H. Investigation of intact protein complexes by mass spectrometry. Mass Spectrom. Rev. 2004, 23, 368−389. (20) Allen, S. J.; Schwartz, A. M.; Bush, M. F. Effects of Polarity on the Structures and Charge States of Native-Like Proteins and Protein Complexes in the Gas Phase. Anal. Chem. 2013, 85, 12055−12061. (21) McLuckey, S. A.; Goeringer, D. E. Slow heating methods in tandem mass spectrometry. J. Mass Spectrom. 1997, 32, 461−474. (22) Benesch, J. L. P.; Ruotolo, B. T.; Simmons, D. A.; Robinson, C. V. Protein Complexes in the Gas Phase: Technology for Structural Genomics and Proteomics. Chem. Rev. 2007, 107, 3544− 3567. (23) Mayer, P. M.; Poon, C. The Mechanism of Collisional Activation of Ions in Mass Spectrometry. Mass Spectrom. Rev. 2009, 28, 608−639. (24) Light-Wahl, K. J.; Schwartz, B. L.; Smith, R. D. Observation of the Noncovalent Quaternary Association of Proteins by Electrospray Ionization Mass Spectrometry. J. Am. Chem. Soc. 1994, 116, 5271− 5278. (25) Jones, C. M.; Beardsley, R. L.; Galhena, A. S.; Dagan, S.; Cheng, G.; Wysocki, V. H. Symmetrical Gas-Phase Dissociation of Noncovalent Protein Complexes via Surface Collisions. J. Am. Chem. Soc. 2006, 128, 15044−15045. (26) Jurchen, J. C.; Williams, E. R. Origin of Asymmetric Charge Partitioning in the Dissociation of Gas-Phase Protein Homodimers. J. Am. Chem. Soc. 2003, 125, 2817−2826. (27) Abzalimov, R. R.; Frimpong, A. K.; Kaltashov, I. A. Gas-phase processes and measurements of macromolecular properties in solution: On the possibility of false positive and false negative signals of protein unfolding. Int. J. Mass Spectrom. 2006, 253, 207− 216. (28) Han, L.; Hyung, S.-J.; Ruotolo, B. T. Bound Cations Significantly Stabilize the Structure of Multiprotein Complexes in the Gas Phase. Angew. Chem., Int. Ed. 2012, 51, 5692−5695.

(29) Sciuto, S. V.; Liu, J.; Konermann, L. An Electrostatic Charge Partitioning Model for the Dissociation of Protein Complexes in the Gas Phase. J. Am. Soc. Mass Spectrom. 2011, 22, 1679−1689. (30) Fegan, S. K.; Thachuk, M. A Charge Moving Algorithm for Molecular Dynamics Simulations of Gas-Phase Proteins. J. Chem. Theory Comput. 2013, 9, 2531−2539. (31) Jørgensen, T. J. D.; Gårdsvoll, H.; Ploug, M.; Roepstorff, P. Intramolecular Migration of Amide Hydrogens in Protonated Peptides upon Collisional Activation. J. Am. Chem. Soc. 2005, 127, 2785−2793. (32) Dongré, A. R.; Jones, J. L.; Somogyi, Á .; Wysocki, V. H. Influence of Peptide Composition, Gas-Phase Basicity, and Chemical Modification on Fragmentation Efficiency: Evidence for the Mobile Proton Model. J. Am. Chem. Soc. 1996, 118, 8365−8374. (33) Boyd, R. K.; Somogyi, Á . The Mobile Proton Hypothesis in Fragmentation of Protonated Peptides: A Perspective. J. Am. Soc. Mass Spectrom. 2010, 21, 1275−1278. (34) Ruotolo, B. T.; Hyung, S.-J.; Robinson, P. M.; Giles, K.; Bateman, R. H.; Robinson, C. V. Ion Mobility−Mass Spectrometry Reveals Long-Lived, Unfolded Intermediates in the Dissociation of Protein Complexes. Angew. Chem., Int. Ed. 2007, 46, 8001−8004. (35) Benesch, J. L. P. Collisional Activation of Protein Complexes: Picking Up the Pieces. J. Am. Soc. Mass Spectrom. 2009, 20, 341−348. (36) Ogorzalek Loo, R. R.; Loo, J. A. Salt Bridge Rearrangement (SaBRe) Explains the Dissociation Behavior of Noncovalent Complexes. J. Am. Soc. Mass Spectrom. 2016, 27, 975−990. (37) Piana, S.; Lindorff-Larsen, K.; Shaw, D. E. Atomic-level description of ubiquitin folding. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 5915−5920. (38) Karplus, M. Development of Multiscale Models for Complex Chemical Systems: From H+H-2 to Biomolecules (Nobel Lecture). Angew. Chem., Int. Ed. 2014, 53, 9992−10005. (39) Mao, Y.; Woenckhaus, J.; Kolafa, J.; Ratner, M. A.; Jarrold, M. F. Thermal unfolding of unsolvated cytochrome c: Experiment and molecular dynamics simulations. J. Am. Chem. Soc. 1999, 121, 2712− 2721. (40) Chen, S. H.; Chen, L. X.; Russell, D. H. Metal-Induced Conformational Changes of Human Metallothionein-2A: A Combined Theoretical and Experimental Study of Metal-Free and Partially Metalated Intermediates. J. Am. Chem. Soc. 2014, 136, 9499−9508. (41) Khakinejad, M.; Kondalaji, S. G.; Maleki, H.; Arndt, J. R.; Donohoe, G. C.; Valentine, S. J. Combining Ion Mobility Spectrometry with Hydrogen-Deuterium Exchange and Top-Down MS for Peptide Ion Structure Analysis. J. Am. Soc. Mass Spectrom. 2014, 25, 2103−2115. (42) Hall, Z.; Politis, A.; Bush, M. F.; Smith, L. J.; Robinson, C. V. Charge-State Dependent Compaction and Dissociation of Protein Complexes: Insights from Ion Mobility and Molecular Dynamics. J. Am. Chem. Soc. 2012, 134, 3429−3438. (43) Wyttenbach, T.; Pierson, N. A.; Clemmer, D. E.; Bowers, M. T. Ion Mobility Analysis of Molecular Dynamics. Annu. Rev. Phys. Chem. 2014, 65, 175−196. (44) Patriksson, A.; Adams, C. M.; Kjeldsen, F.; Zubarev, R. A.; van der Spoel, D. A direct comparison of protein structure in the gas and solution phase: The TRP-cage. J. Phys. Chem. B 2007, 111, 13147− 13150. (45) Steinberg, M. Z.; Elber, R.; McLafferty, F. W.; Gerber, R. B.; Breuker, K. Early Structural Evolution of Native Cytochrome c After Solvent Removal. ChemBioChem 2008, 9, 2417−2423. (46) Goyal, P.; Qian, H. J.; Irle, S.; Lu, X. Y.; Roston, D.; Mori, T.; Elstner, M.; Cui, Q. Molecular Simulation of Water and Hydration Effects in Different Environments: Challenges and Developments for DFTB Based Models. J. Phys. Chem. B 2014, 118, 11007−11027. (47) Barnes, G. L.; Hase, W. L. Energy Transfer, Unfolding, and Fragmentation Dynamics in Collisions of N-Protonated Octaglycine with an H-SAM Surface. J. Am. Chem. Soc. 2009, 131, 17185−17193. 5123

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124

Article

The Journal of Physical Chemistry B (48) Marx, D.; Chandra, A.; Tuckerman, M. E. Aqueous Basic Solutions: Hydroxide Solvation, Structural Diffusion, and Comparison to the Hydrated Proton. Chem. Rev. 2010, 110, 2174−2216. (49) Iyengar, S. S.; Day, T. J. F.; Voth, G. A. On the amphiphilic behavior of the hydrated proton: an ab initio molecular dynamics study. Int. J. Mass Spectrom. 2005, 241, 197−204. (50) Burlet, O.; Orkiszewski, R. S.; Ballard, K. D.; Gaskell, S. J. Charge Promotion of Low-Energy Fragmentations of Peptide Ions. Rapid Commun. Mass Spectrom. 1992, 6, 658−662. (51) Nakai, H.; Sakti, A. W.; Nishimura, Y. Divide-and-ConquerType Density-Functional Tight-Binding Molecular Dynamics Simulations of Proton Diffusion in a Bulk Water System. J. Phys. Chem. B 2016, 120, 217−221. (52) Lill, M. A.; Helms, V. Molecular dynamics simulation of proton transport with quantum mechanically derived proton hopping rates (Q-HOP MD). J. Chem. Phys. 2001, 115, 7993−8005. (53) Donnini, S.; Tegeler, F.; Groenhof, G.; Grubmuller, H. Constant pH Molecular Dynamics in Explicit Solvent with lambdaDynamics. J. Chem. Theory Comput. 2011, 7, 1962−1978. (54) Peng, Y. X.; Swanson, J. M. J.; Kang, S. G.; Zhou, R. H.; Voth, G. A. Hydrated Excess Protons Can Create Their Own Water Wires. J. Phys. Chem. B 2015, 119, 9212−9218. (55) Marrink, S. J.; Tieleman, D. P. Perspective on the Martini model. Chem. Soc. Rev. 2013, 42, 6801−6822. (56) Tozzini, V. Coarse-grained models for proteins. Curr. Opin. Struct. Biol. 2005, 15, 144−150. (57) Schnier, P. D.; Gross, D. S.; Williams, E. R. Electrostatic Forces and Dielectric Polarizability of Multiply Protonated Gas-Phase Cytochrome c Ions Probed by Ion/Molecule Chemistry. J. Am. Chem. Soc. 1995, 117, 6747−6757. (58) Julian, R. R.; Hodyss, R.; Kinnear, B.; Jarrold, M. F.; Beauchamp, J. L. Nanocrystalline aggregation of serine detected by electrospray ionization mass spectrometry: Origin of the stable homochiral gas-phase serine octamer. J. Phys. Chem. B 2002, 106, 1219−1228. (59) Yoo, H. J.; Wang, N.; Zhuang, S. Y.; Song, H. T.; Hakansson, K. Negative-Ion Electron Capture Dissociation: Radical-Driven Fragmentation of Charge-Increased Gaseous Peptide Anions. J. Am. Chem. Soc. 2011, 133, 16790−16793. (60) Li, J.; Santambrogio, C.; Brocca, S.; Rossetti, G.; Carloni, P.; Grandori, R. Conformational Effects in Protein Electrospray Ionization Mass Spectrometry. Mass Spectrom. Rev. 2016, 35, 111− 122. (61) Forbes, M. W.; Bush, M. F.; Polfer, N. C.; Oomens, J.; Dunbar, R. C.; Williams, E. R.; Jockusch, R. A. Infrared spectroscopy of arginine cation complexes: Direct observation of gas-phase zwitterions. J. Phys. Chem. A 2007, 111, 11759−11770. (62) Zhang, Z.; Browne, S. J.; Vachet, R. W. Exploring Salt Bridge Structures of Gas-Phase Protein Ions using Multiple Stages of Electron Transfer and Collision Induced Dissociation. J. Am. Soc. Mass Spectrom. 2014, 25, 604−613. (63) Jockusch, R. A.; Price, W. D.; Williams, E. R. Structure of cationized arginine (Arg center dot M+, M = H, Li, Na, K, Rb, and Cs) in the gas phase: Further evidence for zwitterionic arginine. J. Phys. Chem. A 1999, 103, 9266−9274. (64) Pacholarz, K. J.; Barran, P. E. Distinguishing Loss of Structure from Subunit Dissociation for Protein Complexes with Variable Temperature Ion Mobility Mass Spectrometry. Anal. Chem. 2015, 87, 6271−6279. (65) Schwartz, B. L.; Bruce, J. E.; Anderson, G. A.; Hofstadler, S. A.; Rockwood, A. L.; Smith, R. D.; Chilkoti, A.; Stayton, P. S. Dissociation of Tetrameric Ions of Noncovalent Streptavidin Complexes Formed by Electrospray Ionization. J. Am. Soc. Mass Spectrom. 1995, 6, 459−465. (66) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. J. Chem. Theory Comput. 2008, 4, 435−447.

(67) Van der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A. E.; Berendsen, H. J. C. GROMACS: Fast, flexible, and free. J. Comput. Chem. 2005, 26, 1701−1718. (68) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. LINCS: A linear constraint solver for molecular simulations. J. Comput. Chem. 1997, 18, 1463−1472. (69) McAllister, R. G.; Metwally, H.; Sun, Y.; Konermann, L. Release of Native-Like Gaseous Proteins from Electrospray Droplets via The Charged Residue Mechanism: Insights from Molecular Dynamics Simulations. J. Am. Chem. Soc. 2015, 137, 12667−12676. (70) Hoover, W. G. Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A: At., Mol., Opt. Phys. 1985, 31, 1695−1697. (71) Bussi, G.; Donadio, D.; Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 2007, 126, 0141011. (72) Kaminski, G.; Duffy, E. M.; Matsui, T.; Jorgensen, W. L. Free Energies of Hydration and Pure Liquid Properties of Hydrocarbons from the OPLS All-Atom Model. J. Phys. Chem. 1994, 98, 13077− 13082. (73) Huang, J.; MacKerell, A. D. CHARMM36 all-atom additive protein force field: Validation based on comparison to NMR data. J. Comput. Chem. 2013, 34, 2135−2145. (74) Palmieri, L. D.; Lima, L.; Freire, J. B. B.; Bleicher, L.; Polikarpov, I.; Almeida, F. C. L.; Foguel, D. Novel Zn2+-binding Sites in Human Transthyretin: Implications for Amyloidogenesis and Retinol-Binding Protein Recognition. J. Biol. Chem. 2010, 285, 31731−31741. (75) Terai, T.; Kohno, M.; Boncompain, G.; Sugiyama, S.; Saito, N.; Fujikake, R.; Ueno, T.; Komatsu, T.; Hanaoka, K.; Okabe, T.; Urano, Y.; Perez, F.; Nagano, T. Artificial Ligands of Streptavidin (ALiS): Discovery, Characterization, and Application for Reversible Control of Intracellular Protein Transport. J. Am. Chem. Soc. 2015, 137, 10464−10467. (76) Sinelnikov, I.; Kitova, E. N.; Klassen, J. S. Influence of Coulombic Repulsion on the Dissociation Pathways and Energetics of Multiprotein Complexes in the Gas Phase. J. Am. Soc. Mass Spectrom. 2007, 18, 617−631. (77) Bienkowski, T.; Swider, P.; Blaziak, K.; Danikiewicz, W. Proton affinities of the anions of aromatic carboxylic acids measured by kinetic method. Int. J. Mass Spectrom. 2014, 357, 29−33. (78) Sudmeier, J. L.; Bradshaw, E. M.; Haddad, K. E. C.; Day, R. M.; Thalhauser, C. J.; Bullock, P. A.; Bachovchin, W. W. Identification of histidine tautomers in proteins by 2D H-1/C13(delta 2) one-bond correlated NMR. J. Am. Chem. Soc. 2003, 125, 8430−8431. (79) Hanggi, P.; Talkner, P.; Borkovec, M. Reaction-rate theory: fifty years after Kramers. Rev. Mod. Phys. 1990, 62, 251−342. (80) Cukierman, S. Et tu, Grotthuss! and other unfinished stories. Biochim. Biophys. Acta, Bioenerg. 2006, 1757, 876−885. (81) Cino, E. A.; Choy, W. Y.; Karttunen, M. Comparison of Secondary Structure Formation Using 10 Different Force Fields in Microsecond Molecular Dynamics Simulations. J. Chem. Theory Comput. 2012, 8, 2725−2740. (82) Gross, D. S.; Williams, E. R. Experimental Measurement of Coulomb Energy and Intrinsic Dielectric Polarizibility of a Multiply Protonated Peptide Ion Using Electrospray Ionization FourierTransform Mas Spectrometry. J. Am. Chem. Soc. 1995, 117, 883− 890. (83) Shelimov, K. B.; Clemmer, D. E.; Hudgins, R. R.; Jarrold, M. F. Protein Structure in Vacuo: The Gas-Phase Conformation of BPTI and Cytochrome c. J. Am. Chem. Soc. 1997, 119, 2240−2248. (84) Wedekind, J.; Reguera, D.; Strey, R. Influence of thermostats and carrier gas on simulations of nucleation. J. Chem. Phys. 2007, 127, 0645011. (85) Snijder, J.; Heck, A. J. R. Analytical Approaches for Size and Mass Analysis of Large Protein Assemblies. Annu. Rev. Anal. Chem. 2014, 7, 43−64. (86) Hall, Z.; Politis, A.; Robinson, C. V. Structural Modeling of Heteromeric Protein Complexes from Disassembly Pathways and Ion Mobility-Mass Spectrometry. Structure 2012, 20, 1596−1609. 5124

DOI: 10.1021/acs.jpcb.6b03035 J. Phys. Chem. B 2016, 120, 5114−5124