Neither Two-State nor Three-State: Dimerization of Lambda Cro

jz5b00524_si_001.pdf (1.69 MB) ... The authors acknowledge the Extreme Science and Engineering Discovery Environment (XSEDE) ...... Download Citation...
0 downloads 0 Views 2MB Size
Letter pubs.acs.org/JPCL

Neither Two-State nor Three-State: Dimerization of Lambda Cro Repressor John Yao† and Jin Wang*,†,‡,§,∥ †

Applied Mathematics and Statistics Department, ‡Chemistry Department, and §Physics Department, Stony Brook University, Stony Brook, New York 11794, United States ∥ State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, Jilin 130022, People’s Republic of China S Supporting Information *

ABSTRACT: Lambda Cro repressor is one of the best studied dimeric transcription factors. However, there has still been an unsettled debate for decades about whether it is a two-state dimer or three-state dimer. We provide a new mechanism model that can reconcile these seemingly conflicting (mutually exclusive) experimental results. From simulations with all-atom structure-based model, we observe that the dimerization process of Lambda Cro repressor starts from one folded monomer with one unfolded monomer. Intrasubunit folding and intersubunit binding are partially coupled, in a fly casting manner.

an additional monomeric or dimeric intermediate state (D ⇌ 2F ⇌ 2U or D ⇌ I2 ⇌ 2U; F is folded monomer, I2 is dimeric intermediate). In more recent computational simulation studies, thanks to the much finer spatial and temporal resolution in simulations, new terms like cooperative/coupled binding-folding, noncooperative/decoupled binding-folding, binding-prior-folding, or folding-prior-binding are used to characterize dimerization processes,17 emphasizing more on the kinetic part. These existing terms are more or less conceptually overlapping with each other, and in many cases, equivalencies and implications are made among them in discussing dimerization mechanisms. For example, it is intuitive to assume nonobligatory dimers (a subgroup of three-state dimers) have a decoupled binding-folding process, with an intermediate state of folded yet unbound monomers (2F). However, in this work, as we will see later, this implication does not hold for the Cro dimer. In our simulation, the folded yet unbound monomers are actually off-path intermediate states for dimerization process. Cro monomer folding can be decoupled from binding, but binding has to to be coupled with the folding of one of the two monomers in a dimer. Our molecular dynamic simulations are based on a newly developed all-atom structure-based model, which encodes every heavy atom in a protein and its native contacts.18−20 This model has been demonstrated to be able to reproduce the protein folding results achieved from coarse-grained model and

C

ro protein is a transcription repressor of bacteriophage Lambda. It is needed for the lambda phage to enter the lytic cycle, and competes with another repressor cI in determining the fate of host bacteria.1,2 Cro repressor binds to DNA operators as a homodimer,3−6 but its dimerization mechanism is not clear yet. In thermal denaturation studies7−9 and guanidine hydrochloride (GdnHCL) denaturation study,10 Cro dimer follows a one stage transition between only two significantly populated states, folded dimers (D) and unfolded monomers (2U). However, later evidence from urea induced denaturation suggests that the N-terminal part of a Cro monomer is partially structured.11 Additionally, linear extrapolation of GdnHCL denaturation results with a two-state dissociation model predicts a nanomolar dissociation constant at zero denaturant concentration, which is inconsistent with (unrealistically smaller than) the micromolar dissociation constant measured by DNA binding hydrodynamic experiments.10,12 These conflicts suggest the complexities involved in Cro dimerization,8,11−15 which cannot be explained within the conventional two-state or three-state classification of protein dimers. Protein homodimers are widely involved in many fundamental cellular functions. The current two most commonly used models in experimental researches to characterize dimerization mechanisms are the two-state model (also known as an obligatory dimer) and the three-state model (often known as a nonobligatory dimer).16 The two-state model involves only one transition between one native structured dimer and two unbound unstructured monomers (D ⇌ 2U); the three-state model includes two transitions with © 2015 American Chemical Society

Received: March 12, 2015 Accepted: May 15, 2015 Published: May 15, 2015 2022

DOI: 10.1021/acs.jpclett.5b00524 J. Phys. Chem. Lett. 2015, 6, 2022−2026

Letter

The Journal of Physical Chemistry Letters

Figure 1. Free energy profiles of Cro repressor at transition temperature.

empirical force-field MD simulations.18 Replica exchange molecular dynamics (REMD) is applied to enhance sampling. A flat-bottom potential well is used to constrain the center of mass of two monomers from getting too far apart. The reaction coordinate used to measure structural evolution is the number of native contacts formed. Qi is intersubunit contacts between two monomers; Qa(Qb) is intrasubunit contacts within monomer A(B). Atomic-level contacts are degenerated into residue-level contacts as the developers of all-atom model did,18 resulting in 140 contacts in monomer A, 140 in monomer B, and 57 in between. Free energy profiles are built from microcanonical density of states (DOS), which is derived by weighted histogram analysis method (WHAM). Free energy profiles from one to three dimensions are constructed (Figure 1). Together, they reveal the dimerization mechanism of lambda Cro repressor. At transition temperature, where dimer dissociation and monomer unfolding occur concurrently,6,7 there are four significantly populated states: one structured dimer (D), two unfolded and unbound monomers (2U), two folded but unbound monomers (2F), and two unbound monomers with one folded and one unfolded (FU). There is also a less populated intermediate state of one folded monomer bound with one unfolded monomer (F:U), which is largely merged with the D state, with a free energy barrier less than 0.5kT. The four well populated states (2U, 2F, FU, and D) are observed in simulations of three-state dimers.21−23 Because the FU state is often considered irrelevant and thus ignored in the discussion of dimerization mechanisms, people refer to this scenario as three-state dimerization, even thought the forth state (FU) clearly exists given the existence of 2F and 2U states. On a one-dimensional free energy profile, the only reaction coordinate is the total number of native contacts formed (Qa + Qb + Qi). Because the number of intersubunit contacts is very small compared with the intrasubunit contacts (Qi = 57, Qa = Qb = 140), we multiply Qi by 2 to push the D well further, and thus better reveal the 2F well (Figure 1a). Since the transition temperature is identified by the peak of heat capacity curve and there exist more than two states, there is no guarantee that these four wells will be of the same depth. Here, the states of 2U, FU, and D are predominant, having deeper wells than 2F state. Since the folding dimension and binding dimension are degenerated in the one-dimensional free energy curve, we do not recommend estimating transition barrier heights from it. On two-dimensional free energy landscape (Figure 1b), we use one dimension to capture folding (x-axis: Qa + Qb) and the

other dimension to describe binding (y-axis: Qi). Because the unbound states have 0 intersubunit contact, 2F, 2U, and FU are all located at the boundary of the figure (on the surface of Qi = 0, which gives the blue ribbon). We can see that the barrier of monomer folding is 1−2kT, while the barrier height between state D and state FU is much higher (6−7kT). These barrier heights suggest the folding and unfolding of one monomer can occur easily, while the binding transition between two monomers are hard to happen. Surprisingly, on this landscape, the lowest saddle point to reach the native dimer state (D) is from the state of one folded monomer with one unfolded monomer (FU). The barrier between 2F state and D state (9− 10kT) is higher than that between FU state and D state (6− 7kT). We do observe a dimerization tendency from the folded monomers (2F state), as demonstrated by the indent on the edge of the 2F well, which indicates the formation of a few intersubunit contacts. However, this indent does not reach the dimer well. This interesting feature suggests there may be steric clashes between two monomers when they try to bind each other in folded conformation. One possible source of this clash can come from PHE58 located on the extended third beta strand. Experimental studies suggest that this interface residue PHE58 in the Cro dimer may fold back to the same monomer and stabilize the monomer core.10,11 On three-dimensional free energy profile (Figure 1c), the transitions among these four states become even more clear. Cro repressor demonstrates a very special coupled bindingfolding pathway. Although it appears to be a nonobligatory dimer, also a three-state dimer, its binding process is not about two folded monomers coming together as intuition may tell us. Instead, in both our thermodynamic and kinetic simulations, the binding transition process avoids (does not start from) two well folded monomers (2F). The barrier height between the 2F state and the D state is bigger than that between the FU state and the D state. That is to say, the coexistence of two folded monomers (2F) is an off-pathway intermediate state for association process. The red line connecting red dots is the minimal free energy path (saddle line). We can see the transition barrier for individual monomer folding is relatively low (between 2U and FU, height 1−2 kT). While the coupled binding-folding transition from state FU to state D, through intermediate state F:U, has a higher barrier of 6−7kT, and thus is the rate limiting step in Cro dimerization. To illustrate the more detailed structural evolution, representative structures are picked, and the corresponding average contact formation probability maps are built (Figures 2023

DOI: 10.1021/acs.jpclett.5b00524 J. Phys. Chem. Lett. 2015, 6, 2022−2026

Letter

The Journal of Physical Chemistry Letters

Figure 2. φ-Values of the two transitions in Cro repressor dimerization. The light red colored square in the second map represents disappeared intramonomer φ-value pairs among the α helices of monomer B in the second transition.

dimer dissociation and monomer denaturation occur at the same time at 45 °C.6,7 Remarkably, this synchronous behavior is well observed in our simulation. The heat capacity curve of Cro repressor has only one peak, at which the free energy landscapes show two distinct transitions. On the other hand, PHE58 helps to initiate binding process, by being part of the fly casting arm of the unfolded monomer. Since we do observe the state of two folded yet unbound monomers (2F, Figure S1), Cro is thermodynamically a nonobligatory three-state dimer. However, from minimal free energy path and our constant temperature kinetic simulations, single unfolded monomer binding to its folded partner at the C terminus is widely observed (Figure S4). Based on these observations, we suggest that one unfolded monomer is required when binding occurs. That is to say, although independent folded monomers generally exist, they do not necessarily bind with each other in folded form, due to the harsh requirement of precise docking of well structured interfaces and smaller partner searching diameter. However, if one of the monomers gives up folded structure and becomes extended coil, fly casting can readily occur, with its PHE58 as a “sticky” arm targeting the partially exposed hydrophobic core of the other already folded monomer.27

S1 to S5) for the states on the minimal free energy path. To start binding, one of the monomers need to be completely unfolded while the other is relatively well folded. The relatively well folded monomer (called A) has a one-side exposed hydrophobic core, which serves as the target of the extended “sticky” fly casting arm (containing residue PHE58) from its unfolded partner (called B). When the second unfolded monomer is casted onto the first folded monomer, it completes the partially formed hydrophobic core of A, and the binding interface between the two β strand 3 develops. After that, with the interface relatively well formed and “sticky PHE58” arm from the first monomer (A) right around, the second monomer (B) readily gets folded. In agreement with this fly casting approach, we observe that the folding barrier of monomer B in the presence of folded template monomer A (