DFG Dynamics

View Article Online
View Journal
PCCP
Physical Chemistry Chemical Physics
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: Q. Shao, Z. Xu, J.
Wang, J. Shi and W. Zhu, Phys. Chem. Chem. Phys., 2016, DOI: 10.1039/C6CP06624K.
Volume 18 Number 1 7 January 2016 Pages 1–636 This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
PCCP
Physical Chemistry Chemical Physics Accepted Manuscripts are published online shortly after
www.rsc.org/pccp
acceptance, before technical editing, formatting and proof reading.

Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the

author guidelines.
Please note that technical editing may introduce minor changes

to the text and/or graphics, which may alter content. The journal’s
ISSN 1463-9076 standard Terms & Conditions and the ethical guidelines, outlined
PERSPECTIVE
in our author and reviewer resource centre, still apply. In no
Darya Radziuk and Helmuth Möhwald
Ultrasonically treated liquid interfaces for progress in cleaning and
separation processes
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
rsc.li/pccp
Page 1 of 28 Physical Chemistry Chemical Physics
View Article Online
DOI: 10.1039/C6CP06624K
1 Energetics and Structural Characterization of the

“DFG-Flip” Conformational Transition of B-RAF
Physical Chemistry Chemical Physics Accepted Manuscript

2
Published on 02 December 2016. Downloaded by UNIVERSITY OF NEW ORLEANS on 08/12/2016 15:39:49.
3 Kinase: A SITS Molecular Dynamics Study

4 Qiang Shao1*, Zhijian Xu1, Jinan Wang1, Jiye Shi2*, Weiliang Zhu1
1
5 Drug Discovery and Design Center, Key Laboratory of Receptor Research, Shanghai Institute of
6 Materia Medica, Chinese Academy of Sciences, 555 Zuchongzhi Road, Shanghai, 201203, China
2
7 UCB Biopharma SPRL, Chemin du Foriest, Braine-l’Alleud, Belgium
8 *To whom correspondence should be addressed. Qiang Shao, Tel: +86 21 50806600-1304, E-mail:
9 qshao@mail.shcnc.ac.cn; Jiye Shi, Jiye.Shi@ucb.com.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
1
Physical Chemistry Chemical Physics Page 2 of 28
View Article Online
DOI: 10.1039/C6CP06624K
1 Abstract B-RAF protein kinase is a promising target to treat malignant melanoma. The kinase
2 activity of B-RAF is regulated by a “DFG-flip” conformational transition between functional DFG-in

3 and DFG-out states. The difficulty in resolving the activation loop in crystal structures and the even
4 greater difficulty in experimentally capturing high-energy-level transient structures renders elusive
5 the molecular mechanism of B-RAF functional conformational transition. Here, homology modeling
6 technique and enhanced sampling molecular dynamics simulation were used to identify and
7 energetically characterize the conformational transition pathway of B-RAF on multi-dimensional
8 free-energy landscapes. The results reveal that the conformational transition is a two-state transition,
9 with the evaluated free-energy barrier comparable to those of other kinds of kinases as reported in
10 previous literatures. Hydrophobic interactions between activation loop and neighboring segments are
11 suggested to dominate the conformational transition and determine the free-energy barrier. The
12 detailed analysis of hydrophobic interactions involved in the conformational transition may show
13 suitable pathway for the development of B-RAF inhibitor.
14
15
16
17
18
19
20
21 Keywords: B-RAF kinase, DFG-flip conformational transition, homology modeling technique,
22 enhanced sampling molecular simulation, hydrophobic interactions
23 2
View Article Online
DOI: 10.1039/C6CP06624K
1 Introduction
2 It is well-known that proteins usually undergo sophisticated conformational transitions while

3 carrying out their biological functions.1-3 Such conformational transitions are essentially the
4 collective movements of protein atoms, occurring on the time-scale of µs~s. Experiments have
5 confirmed that specific proteins can spontaneously sample a variety of conformational sub-states
6 between functional structures in equilibrium without any external perturbation such as ligand binding,
7 implying that these proteins are intrinsically flexible that is essential to their functionally correlated
8 conformational transitions.4-10 As drug molecule binds to a target protein, the distribution of protein
9 conformation sampling can be changed and the conformation is forced to convert to the drug-binding
10 preferred one that plays an important role in medical therapy of disease. Therefore, learning how a
11 protein interconverts from one functional structure to another is of fundamental importance to
12 elucidating its biological function, which could provide novel insights into the structure-based drug
13 design. Accordingly, it is now well understood that not only the static structures but also the
14 structural dynamics of proteins are indispensable for computer-aided drug design.11-15
15 RAF kinase is one of the dominant structures in highly conserved mitogen-activated protein kinase
16 (MAPK) signaling pathway (RAS-RAF-MEK-ERK) which transduces signals from membrane-based
17 receptor to the nucleus so as to mediate cell proliferation, differentiation, and survival.16 Numerous
18 cancers are related to the constitutive activation of the MAPK signaling pathway. The RAF kinase
19 family consists of three isoforms (A-, B-, and C-RAF), of particular importance is B-RAF which is
20 most frequently mutated in a range of human cancers.17, 18 Specifically, B-RAFV600E accounts for
21 over 90% of the mutations and has been found to be ~500 times more active than the wild-type
22 protein in vivo.19 The kinase activity of B-RAF is regulated by an equilibrium between inactive and
3
View Article Online
DOI: 10.1039/C6CP06624K
1 active conformations of the enzyme. X-ray crystallography experiments have indicated that the
2 structure difference between the two functional conformations mainly focuses on the activation loop

3 (Asp594-Glu623 of the C-terminal lobe, hereinafter referred to as A loop) which contains a largely
4 conserved DFG motif (Asp594-Phe595-Gly596) in kinase protein superfamily.19-29 As shown in Fig.
5 1, the DFG motif can adopt either “DFG-in” (catalytically active) state21-25 in which the A loop is
6 open allowing ATP or substrate binding, or “DFG-out” (catalytically inactive) state19, 26-29 in which
7 the Phe595 side-chain is flipped and occupies the ATP-binding site while the Asp594 points out of
8 the active site and breaks a key salt-bridge of Asp594-Lys483. It is noteworthy that whilst the crystal
9 structures of wild-type B-RAF and various mutants are amply demonstrated, none of the structures
10 contains a fully resolved A loop. Thus the crystal structures reported so far are unable to provide a
11 comprehensive picture of how B-RAF kinase interconverts between its DFG-in and DFG-out states
12 (“DFG-flip” transition).
13
14 Figure 1. Crystallographic structure of B-RAF kinase with a detailed view of the orientation of DFG
15 motif in DFG-in (PDB code: 3C4C) and DFG-out (PDB code: 1UWH) conformations. The missing
16 part of A loop is represented by dash line. The neighboring segments around A loop are marked with
17 various colors.
4
View Article Online
DOI: 10.1039/C6CP06624K
1 As the DFG-flip conformational transition is crucial to the regulation of kinases but the crystal
2 structures can only provide snapshots of low-resolution structures, a methodology examining the

3 structural dynamics of B-RAF kinase is necessary. Molecular dynamics (MD) simulation could
4 essentially provide time-resolved trajectory at atomistic resolution and thus becomes an attractive
5 approach for the studies of protein dynamics.30-32 To make a sufficient sampling of large-scale,
6 long-time conformational transition of protein, both a priori knowledge of high-resolution structure
7 and a great amount of computational resource are required.33 On one hand, for B-RAF kinase which
8 lacks experimentally identified full-length A loop configuration, homology modeling methodologies
9 are needed to rebuild the missing structural element but the quality highly relies on the sequence
10 identity between the “target” protein and “template”.34 On the other hand, the low efficiency of
11 conventional MD simulation in sampling high-dimensional and rugged configuration space of
12 protein makes the computation very time-consuming. To reach the global equilibrium of simulated
13 system, the simulation should be much longer than the longest relaxation time of the system. As a
14 result, using conventional MD simulation to measure the dynamic properties of B-RAF kinase
15 constructed by homology modeling may have high risk.
16 Enhanced conformational sampling methods have been developed to improve the sampling
17 efficiency of MD simulation.35-38 By smoothing potential energy surface, MD simulation
18 implemented with specific enhanced sampling methods (and of cause accurate force field) is capable
19 to effectively search configuration space and directly observe, characterize functionally
20 indispensable states in protein motion. The improved sampling ability of MD may thus to some
21 extent impair the influence from the use of low-quality structural model in the simulation. The
22 representative enhanced sampling methods include but are not limited to accelerated MD,39
5
View Article Online
DOI: 10.1039/C6CP06624K
1 metadynamics,40 replica exchange MD (REMD)41 and so on. These methods either require some
2 knowledge of the slowest motions to be explored to bias dynamics in a focused way (collective

3 variables) which results in the inconvenient setup of enhanced MD explorations, or require access to
4 massive computational resources. Recently, Yang and Gao proposed a selective
5 integrated-tempering-sampling (SITS) method in which the potential energy of interested system can
6 be efficiently broadened and the sampling on potential energy surface is largely enhanced by
7 introducing a sum-over-temperature non-Bolzmann distribution factor in molecular simulation (see
8 the detailed description in Refs. 42, 43 and the Computational Methods section hereinafter). Multiple
9 tests on small systems (e.g., alanine dipeptide and methyl maltoside)42, 44, 45 have indicated that the
10 SITS method can accurately measure thermodynamic quantities using economized computational
11 resources.42 The validity of such method in sampling collective movement of large-size biomolecule
12 system and evaluating the associated thermodynamics, however, has not been displayed yet.
13 In the present study, we attempted to use explicit-solvent all-atom MD simulations implemented
14 with SITS method (SITS-MD) to investigate the DFG-flip conformational transition of wild-type
15 B-RAF kinase (DFG-in <―> DFG-out). Starting from a crystallographic DFG-in structure, the
16 Modeller program was used to construct the missing A loop and the SITS-MD simulation was
17 performed on the full-atomistic protein structure to sample the configuration space of B-RAF kinase.
18 The DFG-out structure was searched out without any a prior knowledge of the conformational
19 transition process. More importantly, the transition between DFG-in and DFG-out structures was
20 sampled back and forth in single trajectories of SITS-MD. Such sufficient sampling allowed the
21 computation of detailed map of the free-energy landscape, providing a proper estimate of the relative
22 statistical weights of all functionally indispensable states involved in the conformational transition.
6
View Article Online
DOI: 10.1039/C6CP06624K
1 The favorable configurations of A loop in all states were quantitatively identified. The free-energy
2 difference of DFG-in and DFG-out functional states and the free-energy barrier associated with the

3 conformational transition of B-RAF were evaluated, which are comparable to the reported
4 counterparts of kinase homologs. The structural properties of the conformational transition pathway
5 as well as the crucial intra-protein interactions were analyzed, providing detailed insights into the
6 molecular mechanism of B-RAF conformational transition and giving useful information of suitable
7 pathway for inhibitor development.
8 Computational Methods
9 B-RAF simulation system preparation. The atomic coordinates of B-RAF were retrieved from
10 protein data bank (PDB code: 3C4C22) with the crystal water maintained but the crystal inhibitor
11 PLX4720 removed. The missing residues 597-613 (17 residues) within A loop was modeled using
12 the loop building routine of Modeller.46 The modeled protein structure can be seen in Fig. S1 in
13 Supporting Information. AMBER 11 suite of program47 was employed to run all simulations of
14 which the force fields are FF99 force field for protein48 and SPC/E model for water.49 In the
15 constructed SITS-MD simulation system, the DFG-in structure of B-RAF was solvated in a cubic
16 box with the size of 75.0×86.8×79.0 Å3. Multiple Cl- anions were added to neutralize the charges on
17 protein and a total of 12364 water molecules were subsequently added to fill the empty position of
18 cubic box.
19 SITS-MD simulation details. The constructed system of B-RAF in aqueous solution was initially
20 minimized for 2500 steps with the protein being fixed using harmonic restraints (force constant =
21 500.0 kcal mol-1 Å-2). Then the system was heated upon 300 K and equilibrated for 10~20 ns with
22 harmonic restrains applied on protein backbone atoms (force constant = 10.0 kcal mol-1 Å-2). The
7
View Article Online
DOI: 10.1039/C6CP06624K
1 equilibrated systems with varied coordinates and velocities (mainly on water molecules) were used
2 for multiple independent relatively long-time production runs generating simulation data for

3 statistical analysis. Production runs were performed at constant temperature of 300 K and constant
4 pressure of 1 atm (NPT ensemble), but with potential energy modified via SITS method to encourage
5 thorough sampling over a large potential energy range. The SHAKE algorithm50 was used to fix all
6 covalent bonds involving hydrogen atoms and periodic boundary conditions were used to avoid edge
7 effects. The Particle Mesh Ewald method was applied to treat long-range electrostatic interactions51
8 and the cutoff distance for long-range terms (electrostatic and van der Waals energies) was set as
9 10.0 Å. The simulation data in each product run was collected every 2.0 ps.
10 The details of SITS method have been described in previous literatures.42, 43 Generally, SITS
11 method broadens the potential energy range of simulation system through the introduction of a
12 sum-over-temperature non-Boltzmann distribution factor. The potential energy distribution
13 corresponding to the distribution factor is achieved through either a Monte Carlo directly or using the
14 effective force corresponding to the following effective potential energy:
1 − βkU
15 U eff = − ln ∑k n k e (1)
β0
16 where U is the original potential energy of the system, β 0 = 1 / k BT0 , (kB is the Boltzmann constant
17 and T0 the temperature of the system), ߚ௞ is a series of temperatures that cover both low and high
18 temperatures around T0, ݊௞ ’s are parameters with their values being determined in a preliminary
19 iteration process.42 As a result of the proper adjustment of the values of ݊௞ ’s, the potential energy
20 can be sampled evenly in a large desired range corresponding to the integration over a large
21 temperature range {β k } given in Eq. 1, which allows a sufficient sampling on potential energy
22 surface and conformation collection.42, 43

8
View Article Online
DOI: 10.1039/C6CP06624K
1 Since the conformational transition of B-RAF kinase mainly focuses on its A loop, in the
2 simulation system of B-RAF solvated in water solution, we attempt to only enhance sampling the A

3 loop while simulating the remaining part of simulation system in a conventional MD manner. Such
4 selective sampling enhancement on specific position can avoid unnecessary oversampling of
5 unconcerned configurations and thus improve the computational efficiency. Here, the overall
6 potential energy of the simulation system can be divided into the energy of A loop (interested part,
7 EA), the energy of the remaining part of protein (EP(-A)), the energy of water solvent (EW), as well as
8 the interaction energies between the abovementioned components (A loop-remaining part of protein:
9 EA-P(-A), A loop-water: EA-W, remaining part of protein-water: EP(-A)-W). That is to say:
10 U=EA+EP(-A)+EW+EA-P(-A)+EA-W+EP(-A)-W. The effective potential can be rewritten as the following
11 form:
1 − βk (E A + E A − W + E A − P ( − A ))
12 U eff = E P ( − A ) + E W + E P ( −A )−W − ln ∑k n k e (2)
β0
13 The A loop relevant components of the potential energy can be sampled in enhanced manner
14 intentionally while leaving other components sampled with original forms. In the present SITS-MD
15 simulation, a total of 250 discrete temperatures at evenly spaced intervals in the range of 280K-370
16 K were utilized in Eq. 2.
17 It is noteworthy that a converged calculation of SITS-MD simulation yields a biased distribution
− β 0U eff ( r )
18 in the configuration space: ρeff (r ) ∝ e . The unbiased distribution function at any
19 temperature T0 can be easily recovered as:
20 ρ(r ) = ρeff (r )e − β (U (r )−U eff (r ))

0
(3)
21 where ρeff (r ) is measured in SITS-MD simulation. Thermodynamic quantities can be calculated
22 once ρ(r ) is known.

9
View Article Online
DOI: 10.1039/C6CP06624K
1 Conventional MD simulation details. To evaluate the structural stability of the DFG-out
2 structure of B-RAF kinase captured by SITS-MD simulation, conventional MD simulation was run

3 on such DFG-out structure in aqueous solution. The construction of the simulation system, the
4 following energy minimization, and the heating-up procedure were similar to those in SITS-MD. The
5 equilibrium simulation was then performed for ~200 ns at 300 K without either any constrains
6 artificially added on protein or the perturbation from SITS method.
7 MM-PB/GBSA calculation details. To indicate the inter-residue interactions which experience
8 changes during the DFG-flip transition, we calculated the binding interaction energies between A
9 loop and the neighboring segments of protein in various states involved in transition using the
10 standard approach of MM/GBSA.52 Multiple representative structures of DFG-in, DFG-out, and
11 transition states identified from SITS-MD simulation were solvated in cubic boxes to run
12 explicit-solvent MD simulations. After the energy minimization and heating-up procedures (similar
13 to those in SITS-MD), the subsequent equilibrium simulations were performed for ~2 ns at 300 K.
14 For each system, the MM/GBSA energy was calculated using MMPBSA.py.MPI program in
15 AMBER software53 on 200 snapshots collected from the equilibrium simulation. The van deer Waals
16 (vdW) and electrostatic interaction energies between A loop residues and the residues from other
17 segments of protein were extracted for data analysis.
18 Results
19 Reversible active-inactive transition visiting in individual trajectories. A total of 9
20 independent trajectories with varied initial coordinates and velocities were run in the present
21 SITS-MD simulation on B-RAF kinase. In individual trajectories, SITS method was utilized to
22 selectively enhance the dynamics of A loop (the main contributor for DFG-flip conformational
10
View Article Online
DOI: 10.1039/C6CP06624K
1 transition) and the simulation lasted for ~120 ns. The accumulated simulation time of SITS-MD was
2 thus ~1 µs. Figure S2 shows the time series of the potential energy components in a representative

3 trajectory. In comparison to conventional MD, the potential energy contributed by A loop (EA
4 +EA-P(-A)+EA-W) can be expanded in a large range and the sampling is frequent in both low and high
5 energy regions in SITS-MD simulation whereas the remaining part of potential energy
6 (EP(-A)+EW+EA-P(-A)+EP(-A)-W) is maintained in similar range as that in conventional MD simulation.
7 As a result, the transition between the DFG-in and DFG-out states of BRAF kinase can be promoted
8 in SITS-MD simulation.
9 There are seven major protein segments around A loop including P loop (Ile463-Tyr472), αC helix
10 (Gln493-Lys507), αE helix (Met550-Lys570), catalytic loop (Ile572-Asn581), F loop (Ile178-Ser187)
11 between A loop and αF helix (Phe635-Met650), and αG helix (Asn660-Gly672, Fig. 1). As shown in
12 Fig. 2A, the side-chain distance between Phe595 in DFG motif and Leu567 in αE helix and that
13 between Phe595 and Val471 in P loop are two suitable reaction coordinates to monitor the DFG-flip
14 conformational transition. We checked 14 crystal structures of B-RAF kinases deposited in PDB and
15 found that the average Phe595-Leu567 side-chain distance is 6.52±0.78 Å or 13.30±0.65 Å, and the
16 average Phe595-Val471 side-chain distance is 14.03±0.41 Å or 9.06±0.32 Å in DFG-in or DFG-out
17 structures, respectively (the detailed side-chain distances in individual crystal structures can be seen
18 in Table 1). One can see from Fig. 2B that the conformational transition can be captured repeatedly
19 in a single trajectory of SITS-MD simulation. A total of ~18 transition events have been recorded in
20 9 trajectories eventually, which provide quantitative data for the thermodynamics evaluation for the
21 conformational transition of B-RAF.
11
View Article Online
DOI: 10.1039/C6CP06624K

1
2 Figure 2. (A) A schematic diagram indicating the side-chain distances of the residue pairs of
3 Phe595-Leu567 and Phe595-Val471. (B) Time series of the side-chain distances of Phe595-Leu567
4 (black) and Phe595-Val471 (red) in a representative trajectory of B-RAF simulated by SITS-MD.
5 Purple colored arrows indicate the occurrence of DFG-flip conformational transition.
6
DFG-in DFG-out
Phe595-Leu567 Phe595-Val471 Phe595-Leu567 Phe595-Val471

PDB ID PDB ID
(Å) (Å) (Å) (Å)
2FB8 6.96 13.67 1UWH 13.35 8.22
3C4C 7.79 14.56 1UWJ 12.97 8.45
3P4Q 6.17 13.76 4FC0 13.38 8.54
3PPJ 6.54 13.70 4G9C 13.00 8.16
3PRI 6.43 13.85 4G9R 11.96 8.38
3TV6 5.20 14.65 4JVG 13.47 8.77
4KSP 14.19 7.87
4KSQ 14.01 7.71
7 Table 1. Side-chain distances of the hydrophobic pairs of Phe595-Leu567 and Phe595-Val471 in

8 representative crystal DFG-in and DFG-out structures of B-RAF.
9
10 Thermodynamics estimate for the conformational transition of B-RAF. To measure the
11 thermodynamics of the conformational transition of B-RAF, we calculated a one-dimensional
12
View Article Online
DOI: 10.1039/C6CP06624K
1 free-energy profile as a function of Phe595-Leu567 side-chain distance (dF595-L567) (Fig. 3). Two
2 distinct local minima are located at 4.5 Å≤dF595-L567≤5.5 Å and 12.5 Å≤dF595-L567≤13.5 Å. To identify

3 the corresponding states, hierarchical clustering analyses were performed using the kclust algorithm
4 available in MMTSB Toolset54 on snapshots selected from the simulation trajectories which satisfy
5 the abovementioned criteria, respectively. The results indicate that it is the DFG-in and DFG-out
6 states of B-RAF that correspond to the two abovementioned local minima (see their representative
7 conformations in Fig. 4B).
9 Figure 3. One-dimensional free-energy profile as the function of Phe595-Leu567 side-chain distance

10 for B-RAF. The profile is calculated at 300 K.
11
12 The DFG-in and DFG-out states of B-RAF can be also clearly seen in a two-dimensional
13 free-energy landscape in the plane spanned by the side-chain distances of Phe595-Leu567 and
14 Phe595-Val471 (Fig. 4A). The DFG-in state is more stable than DFG-out state with the free-energy
15 difference (∆G) of -2.03 kBT and the free-energy barrier (∆G*) which needs to be overcome for the
16 transition from the former state to the latter is 5.10 kBT. To the best of our knowledge, no direct
17 measurement of the conformational transition thermodynamics is available for B-RAF from either
18 experiments or theoretical simulation studies. However, consistencies can be seen indeed between
13
View Article Online
DOI: 10.1039/C6CP06624K
1 the free-energy barrier measured here for B-RAF and those of other kinds of kinases in previous
2 simulation studies with either extensive non-enhanced MD55 or enhanced sampling MD.56

3 Specifically, Pande and co-workers employed massive MD simulations (550 µs), Markov state
4 models, and novel adaptive sampling algorithms to map the free-energy landscape of the
5 conformational transition of c-Src tyrosine kinase, and the free-energy barrier was evaluated as 6.4
6 kBT (3.8 kcal/mol) between the DFG-in and DFG-out states.55 In addition, the combined string
7 method with swarms-of-trajectories (determining the reaction pathway) and umbrella sampling
8 simulation (determining the conformational transition thermodynamics) by Roux and co-workers
9 gave the free-energy barriers of ~11.7 kBT for c-Abl kinase and 3.4 kBT for c-Src tyrosine kinase.56
10
11 Figure 4. (A) Two-dimensional free-energy landscape as the function of the side-chain distances of
12 Phe595-Leu567 and Phe595-Val471 for B-RAF at 300 K (the contours are spaced at intervals of 0.5
13 kBT ). (B) Representative conformations of DFG-in, transition (TS), and DFG-out states indicated in
14 the free-energy landscapes of Figs. 3 and 4A. The side-chains of Asp594 and Phe595 are shown with
15 licorice representation and colored in green.
16
17 The well-populated DFG-in and DFG-out structures identified in the present simulation are
18 consistent with the corresponding crystal structures (see the superposition of the representative
19 structures of the simulated DFG-in and DFG-out states with the corresponding crystal structures in
20 Fig. 5). For instance, the representative structure of DFG-in state from the simulation (green color)
21 has the root-mean-square deviation (RMSD) value of 1.14 Å with respect to the PDB structure of
14
View Article Online
DOI: 10.1039/C6CP06624K
1 3TV625 which contains the largest fraction of A loop among experimentally measured DFG-in
2 conformations. In addition, the representative structure of DFG-out state from the simulation has the

3 RMSD value of 2.57 Å with respect to the PDB structure of 4G9C27 which contains the largest
4 fraction of A loop among DFG-out conformations. To further measure the stability of the SITS-MD
5 simulated DFG-out structure of B-RAF kinase, a conventional MD simulation was run for such
6 structure in aqueous solution. One can see from Fig. S3 that within the entire simulation time (~200
7 ns), the overall DFG-out structure of the kinase keeps stable with the RMSD of the entire kinase with
8 respect to the initial structure only fluctuating within small values. In addition, the side-chain of
9 Phe595 dominantly stays at its DFG-out position, as revealed by the small fluctuation in
10 Phe595-Leu567 and Phe595-Val471 side-chain distances. Therefore, given one endpoint structure,
11 the SITS-MD simulation can search for the other functional structure of protein.
12
13 Figure 5. (A) The superposition of the representative structure of DFG-in state (green color)
14 identified by SITS-MD with multiple crystal structures adopting DFG-in conformations (cyan: 2FB8,
15 magenta: 3C4C, pink: 3D4Q, silver: 3PPJ, purple: 3PRI, orange: 3TV6). (B) The superposition of
16 the representative structure of DFG-out state (green color) identified by SITS-MD with multiple
17 crystal structures adopting DFG-out conformations (cyan: 1UWH, magenta: 1UWJ, yellow: 4FC0,
18 pink: 4G9C, silver: 4G9R, purple: 4JVG, orange: 4KSP). The expanded view of DFG motif
15
View Article Online
DOI: 10.1039/C6CP06624K
1 orientation in the superposition is given in the middle panel.

2
Both one- and two-dimensional free-energy profiles indicate that the conformational transition of

3
4 B-RAF is a two-state transition. Snapshot conformations were extracted to visualize the transition
5 (TS) state. As shown in the representative conformation (Fig. 4B) and the expanded view of the DFG
6 motif orientation (Fig. 6), in the TS state, the A loop (particularly the region adjacent to DFG motif)
7 translocates away from the DFG-in configuration to a position in the middle of the pathway. As a
8 result, the side-chain of Phe595 slightly shifts away from the DFG-in orientation but is not flipped
9 yet in the TS state. Such side-chain can be fully flipped unless the entire structure is converted to
10 DFG-out state.
11
12 Figure 6. Expanded view of the comparison of DFG motif orientations in DFG-in (blue), TS (green),
13 and DFG-out (red) states of B-RAF. Dashed arrows indicate the transition tendency of DFG motif.
14
15 It is noteworthy that the curvature of the free-energy surface may depend on the selection of
16 reaction coordinates, which consequently affects the assessment of the detailed thermodynamic
17 quantities and the configuration of transition pathway. To indicate the suitability of the reaction
16
View Article Online
DOI: 10.1039/C6CP06624K
1 coordinates used here (Phe595-Leu567 and Phe595-Val471 side-chain distances) in constructing the
2 free-energy surface of B-RAF, we ran a series of conventional MD simulations starting from the

3 structure ensembles of TS state to investigate the transition pathways from TS to DFG-in and
4 DFG-out states, respectively. We took 6 representative structures of TS state and performed 3
5 simulations each ~20 ns in length from each structure for a total of 18 20-ns simulations. The
6 simulation trajectories were depicted in Figs. S4-S6. As shown in these figures, the DFG motif of
7 B-RAF can spontaneously convert from the TS state to either DFG-in or DFG-out state in short
8 simulation time, and the transition events from TS to DFG-in are much more than the events from TS
9 to DFG-out. Figure 7 shows the representative TS → DFG-in (blue colored) and TS → DFG-out (red
10 colored) transition paths depicted in a plane spanned by the side-chain distances of Phe595-Leu567
11 and Phe595-Val471. The two paths roughly follow the minimum free-energy pathway connecting
12 DFG-in and DFG-out states in the free-energy landscape given by SITS-MD simulation (Figs. 4A
13 and 7). Thus, we may say that the Phe595-Leu567 and Phe595-Val471 distances are suitable reaction
14 coordinates for the DFG-flip transition of B-RAF.
15
16 Figure 7. Representative transition paths from TS to DFG-in (blue colored) and from TS to DFG-out
17 (red colored). Underlying SITS-MD free-energy contours are smoothened and plotted in gray.
17
View Article Online
DOI: 10.1039/C6CP06624K
1 Detailed molecular interactions involved in conformational transition of B-RAF. The crucial
2 residue-residue interactions in the three states were analyzed to figure out the molecular mechanism

3 underlying DFG-flip conformational transition of B-RAF. As shown in Fig. S7A, the protein
4 segments around A loop contain a bunch of hydrophobic as well as multiple charged amino-acid
5 residues. Accordingly, it is hydrophobic interactions and salt-bridges that are mainly formed between
6 A loop and other protein segments in various states. The detailed interactions are categorized in Table
7 2. In addition, the interactions with the N-terminal residues of A loop involved are particularly
8 depicted in Fig. 8 for an intuitive representation.
Residue Sites DFG-in TS DFG-out
Hydrophobic Interactions
A Loop-A Loop L597- A598, L597-
L597- V600, L613-I617 L597-V600, F610-L613
V600, F610-L613
F595-I463, F595-V471,
A Loop-P loop V600-F468 A598-F468
V600-V471
L597-F498,
A Loop-αC Helix ------ L597-L505
V600-A497, V600-F498
A Loop-αE Helix F595-L567 ------ ------
A Loop-Catalytic
F595-I572 ------ ------
Loop
A Loop-F Loop W604-Y633 ------ ------
A Loop-αG-Helix L613-I666 ------ ------
A Loop-Other F595-I513, L597-I527, L597-I527, L597-L485, F595-A481, F595-L514,
L597-L525, F595-W531, F595-F583,
Sites† V600-L485 A598-L485, V600-L485 L597- I527
Salt-Bridge Interactions
A Loop-A Loop ------ ------ D594-R603
A Loop-αC Helix R603-E501 R603- E501 ------
A Loop-Catalytic
------ E611-R575 E611-K578
Loop
A Loop-Other
D594-K483 D594-K483 ------
Sites†
9 Table 2. Crucial interactions between A loop and neighboring protein segments in DFG-in, TS, and
18
View Article Online
DOI: 10.1039/C6CP06624K
1 DFG-out states of B-RAF. †Other sites mainly refer to the anti-parallel β-strands nearby β-hairpin
2 containing P loop (green-colored region in Fig. S7B).

3
4 In DFG-in state, the N-terminal segment of A loop (adjacent to the DFG motif) forms a β-sheet
5 interaction with the β6 strand. In such configuration, DFG motif forms multiple hydrophobic
6 interactions with surrounding αE helix and catalytic loop. Meanwhile, the C-terminal segment of A
7 loop can also form hydrophobic interactions with residues from F loop and αG helix. All of these
8 hydrophobic contacts are, however, eliminated in TS state due to the translocation of A loop.
9 Although new hydrophobic interactions are formed between A loop and other sites (e.g., the
10 anti-parallel β-strands nearby P loop), the total number of hydrophobic interactions with A loop
11 involved is actually decreased in TS state, which accounts for the increased free-energy level of such
12 state. As the DFG motif reorients to the DFG-out position, more hydrophobic interactions are formed
13 between DFG motif and P loop as well as the nearby anti-parallel β-strands. As a result, the total
14 number of hydrophobic interactions is increased which stabilizes the DFG-out state. This observation
15 is in agreement with the simulation results in Ref. 56 which suggested that P loop stabilizes the
16 DFG-out state mainly through vdW interactions in both c-Abl and c-Src kinases. As hydrophobic
17 interactions play an important role in DFG-flip conformational transition, charge-charge interactions
18 also participate into such conformational transition. For instance, the salt-bridges Asp594-Lys483
19 and Arg603- Glu501 characterized in DFG-in state can survive in TS state but are eventually broken
20 in DFG-out state. Meanwhile, a new salt-bridge is formed between A loop and catalytic loop
21 (E611-R575) in TS state. This salt-bridge along with another one within the A loop (D594-R603)
22 contribute to the stabilization of DFG-out state.
19
View Article Online
DOI: 10.1039/C6CP06624K

2 Figure 8. (A-C) Hydrophobic interactions and (D-F) salt-bridges formed between A loop residues
3 (adjacent to DFG motif) and residues out of A loop in DFG-in, TS, and DFG-out states of B-RAF,
4 respectively.
5
6 While Table 2 and Fig. 8 qualitatively point out the favorite interactions involved in DFG-flip
7 conformational transition of B-RAF kinase, we attempted to analyze in a quantitative manner the
8 interactions which experience changes in the DFG-flip transition from energetic perspective. We
9 calculated the vdW and electrostatic energies of A loop binding to other protein segments in DFG-in,
10 TS, and DFG-out states, respectively, using MM/GBSA approach (see the details in the
11 Computational Methods section). One can see from Fig. 9 that the vdW interaction between A loop
12 and the neighboring protein segments is the strongest in DFG-in state (as shown by the lowest vdW
13 energy), which is weakened in TS state and reincreased in DFG-out state. On the other hand, the
14 strength of the electrostatic interaction is at the maximum in TS state, in the middle in DFG-in state,
20
View Article Online
DOI: 10.1039/C6CP06624K
1 and at the lowest level in DFG-out state. Therefore, the changing tendencies of vdW and electrostatic
2 interaction strengths are certainly consistent with the changing tendencies of the numbers of

3 hydrophobic contacts and salt-bridges along the conformational transition (Table 2), respectively.
4 The combination of Table 2 and Fig. 9 clearly demonstrates the importance of hydrophobic
5 interactions (as well as charge-charge interactions) in DFG-flip conformational transition.
7 Figure 9. The vdW and electrostatic energies of the interactions between A loop and the neighboring
8 segments of B-RAF in the structure ensembles of DFG-in, DFG-out, and TS states.
9
10 We also calculated the distribution of the number of favorite contacts in the three states along
11 conformational transition. The calculation was focused on hydrophobic interactions between A loop
12 and P loop which display the most dramatic change from DFG-in to DFG-out states as shown in
13 Table 2. The residue-residue contacts were considered as formed as the minimal distance of heavy
14 atoms between the pair of residue side-chains was less than 6.0 Å. One can see from Fig. 10 that both
15 the total contacts and the hydrophobic contacts between A and P loops keep increasing in the
16 transition from DFG-in to DFG-out states. In addition, the hydrophobic interactions dominate the
17 interactions between A and P loops in this transition.
21
View Article Online
DOI: 10.1039/C6CP06624K

1
2 Figure 10. The distribution of (A) total side-chain contacts and (B) hydrophobic side-chain contacts
3 formed between A loop and P loop in DFG-in, TS, and DFG-out states of B-RAF, respectively.
4
5 Discussion and Conclusions
6 Kinase proteins play key roles in signaling pathways whose activities are regulated by a controlled
7 conformational transition between catalytically active and inactive states.19, 57, 58 The dysfunction of
8 kinases might lead to a number of diseases, most notably cancer, e.g., the shifting of the
9 active-inactive equilibrium toward the active state results into enhanced signaling and probably
10 uncontrolled cell growth.59, 60 Knowing what are the crucial interactions in the transition could be
11 helpful for not only the understanding of the biological function of kinase but also the inhibitor
12 development.
13 In this study, all-atom MD simulation was run to investigate the DFG-flip conformational
14 transition of B-RAF kinase. It is well-known that the validity of a molecular simulation highly
15 depends on the accuracy of molecular force field and the effectiveness of conformational sampling.
16 Here, a sophisticated force field (AMBBER FF9948) which has been widely used for measuring
17 biomolecular dynamic motions and predicting relevant free-energy changes,61-67 was used to model
22
View Article Online
DOI: 10.1039/C6CP06624K
1 the molecular interactions of B-RAF. Meanwhile, an enhanced sampling method, namely selective
2 integrated tempering sampling (SITS),42-45 was utilized to improve the sampling ability of MD

3 simulation to bridge the time-scale gap between simulation and experiment. Intriguingly, within the
4 simulation time of ~120 ns, both functional DFG-in and DFG-out structures of B-RAF could be
5 sampled back and forth in individual trajectories. The performance of multiple independent
6 trajectories recorded plenty events of DFG-flip transitions, accumulating enough data for a proper
7 estimate of the relative statistical weights of various conformations in free-energy landscape.
8 The detailed free-energy landscape analysis indicates that the conformational transition of B-RAF
9 is a two-state transition with the DFG-in state more thermodynamically favorable than DFG-out state.
10 The structure analysis provides quantitative information for the configurations of activation loop (A
11 loop) favorably adopted in both DFG-in and DFG-out structures, which are missing in the
12 corresponding crystal structures. The free-energy barrier associated with the minimum free-energy
13 transition pathway of B-RAF is similar to those of other kinds of kinases such as c-Src tyrosine
14 kinase as reported in previous literatures,55, 56 which to some extent indicates the validity of the
15 present simulation in conformational sampling and free-energy evaluation.
16 It is interesting to observe that hydrophobic interactions between A loop and neighboring protein
17 segments play a crucial role in the DFG-flip transition of B-RAF kinase. The reduction in
18 hydrophobic interactions between A loop and the segments of αC helix, αE helix, catalytic loop, F
19 loop and αG helix responds to the free-energy barrier arisen in the transition from DFG-in to
20 transition (TS) states. The formation of the hydrophobic cluster between the N-terminal residues of A
21 loop including DFG motif and the residues from P loop and neighboring anti-parallel β-strands
22 promotes the subsequent transition from TS to DFG-out and stabilizes the latter functional state.
23
View Article Online
DOI: 10.1039/C6CP06624K
1 Salt-bridge interactions, on the other hand, also participate into the DFG-flip conformational
2 transition. In comparison to the dominating effects of hydrophobic interactions in determining the

3 thermodynamics of B-RAF transition, the presence of multiple salt-bridges in all three states of the
4 conformational transition may marginally contribute to the structural stabilities of these states.
5 In summary, the SITS induced enhanced sampling MD simulation of B-RAF kinase reported in
6 this study has provided novel insights into the molecular mechanism of DFG-flip conformational
7 transition. However, there are several caveats worth noting. First, it is likely that the DFG-flip
8 transition of kinases may be influenced by the protonation of DFG motif.68, 69 Our recent simulation
9 studies on p38α kinase indicated that the protonation of D404 residue of the kinase changes the
10 free-energy difference between DFG-in and DFG-out states whereas has insignificant effects on the
11 detailed transition pathway.70 Second, B-RAF protein kinase is mutated at a high frequency in human
12 cancer and a single V600E mutation is the widespread oncogenic mutation in melanoma. Therefore,
13 the investigation of how the V600E mutation affects the dynamic properties of B-RAF should be
14 necessary for the further study. The present study reveals the important roles of detailed hydrophobic
15 interactions (Table 2) in DFG-flip transition, which could provide useful information for the design
16 of B-RAF inhibitor. For instance, naturally hydrophobic inhibitors are supposed to interfere the
17 hydrophobic interactions between A loop and surrounding segments so as to exert a direct influence
18 on the DFG-flip transition. As an example, the extensive MD simulation in Ref. 55 observed that an
19 8-anilino-1-naphthalene sulfonate (ANS) molecule can form hydrophobic clusters with the residues
20 around binding pocket of c-Src tyrosine kinase, which contribute to the stabilization of the
21 intermediate state and thus block the activation process of this kinase.
22 Acknowledgements
24
View Article Online
DOI: 10.1039/C6CP06624K
1 We thank the National Natural Science Foundation of China (Grant No. 21373258), National
2 Basic Research Program (Grant No. 2014CB910400), and Special Program for Applied Research on

3 Super Computation of the NSFC-Guangdong Joint Fund (the second phase) for financial support.
4 The simulations were run at TianHe 1 supercomputer in Tianjin and TianHe II supercomputer in
5 Guangzhou, China.
7 References
8 1. A. Bakan and I. Bahar, Proc. Natl. Acad. Sci. USA, 2009, 106, 14349-14354.
9 2. R. P. Bywater, J. Biomol. Struct. Dyn., 2013, 31, 351-362.
10 3. S. Hayward and H. J. C. Berendsen, Proteins, 1998, 30, 144-154.
11 4. K. A. Henzler-Wildman, V. Thai, M. Lei, M. Ott, M. Wolf-Watz, T. Fenn, E. Pozharski, M. A.
12 Wilson, G. A. Petsko, M. Karplus, C. G. Hubner and D. Kern, Nature, 2007, 450, 838-844.
13 5. J. A. Hanson, K. Duderstadt, L. P. Watkins, S. Bhattacharyya, J. Brokaw, J. W. Chu and H.
14 Yang, Proc. Natl. Acad. Sci. USA, 2007, 104, 18055-18060.
15 6. E. Z. Eisenmesser, O. Millet, W. Labeikovsky, D. M. Korzhnev, M. Wolf-Watz, D. A. Bosco,
16 J. J. Skalicky, L. E. Kay and D. Kern, Nature, 2005, 438, 117-121.
17 7. H. Beach, R. Cole, M. L. Gill and J. P. Loria, J. Am. Chem. Soc., 2005, 127, 9167-9176.
18 8. N. M. Antikainen, R. D. Smiley, S. J. Benkovic and G. G. Hammes, Biochemistry, 2005, 44,
19 16835-16843.
20 9. J. Gsponer, J. Christodoulou, A. Cavalli, J. M. Bui, B. Richter, C. M. Dobson and M.
21 Vendruscolo, Structure, 2008, 16, 736-746.
22 10. D. D. Boehr, R. Nussinov and P. E. Wright, Nat. Chem. Biol., 2009, 5, 789-796.
23 11. H. A. Carlson and J. A. McCammon, Mol. Pharmacol., 2000, 57, 213-218.
24 12. H. A. Carlson, Curr. Opin. Chem. Biol., 2002, 6, 447-452.
25 13. R. E. Amaro, R. Baron and J. A. McCammon, J. Comput.-Aid. Mol. Des., 2008, 22, 693-705.
26 14. J. D. Durrant and J. A. McCammon, Curr. Opin. Pharmacol., 2010, 10, 770-774.
27 15. W. Sinko, S. Lindert and J. A. McCammon, Chem. Biol. Drug. Des., 2013, 81, 41-49.
28 16. C. Peyssonnaux and A. Eychene, Biol. Cell, 2001, 93, 53-62.
29 17. M. J. Garnett and R. Marais, Cancer Cell, 2004, 6, 313-319.
30 18. H. Davies, G. R. Bignell, C. Cox, P. Stephens, S. Edkins, S. Clegg, J. Teague, H. Woffendin,
31 M. J. Garnett, W. Bottomley, N. Davis, N. Dicks, R. Ewing, Y. Floyd, K. Gray, S. Hall, R.
32 Hawes, J. Hughes, V. Kosmidou, A. Menzies, C. Mould, A. Parker, C. Stevens, S. Watt, S.
33 Hooper, R. Wilson, H. Jayatilake, B. A. Gusterson, C. Cooper, J. Shipley, D. Hargrave, K.
34 Pritchard-Jones, N. Maitland, G. Chenevix-Trench, G. J. Riggins, D. D. Bigner, G. Palmieri,
35 A. Cossu, A. Flanagan, A. Nicholson, J. W. C. Ho, S. Y. Leung, S. T. Yuen, B. L. Weber, H. F.
36 Siegler, T. L. Darrow, H. Paterson, R. Marais, C. J. Marshall, R. Wooster, M. R. Stratton and
37 P. A. Futreal, Nature, 2002, 417, 949-954.
38 19. P. T. C. Wan, M. J. Garnett, S. M. Roe, S. Lee, D. Niculescu-Duvaz, V. M. Good, C. M. Jones,
25
View Article Online
DOI: 10.1039/C6CP06624K
1 C. J. Marshall, C. J. Springer, D. Barford, R. Marais and C. G. Project, Cell, 2004, 116,

2 855-867.
3 20. R. Roskoski, Biochem. Bioph. Res. Co., 2010, 399, 313-317.

4 21. A. J. King, D. R. Patrick, R. S. Batorsky, M. L. Ho, H. T. Do, S. Y. Zhang, R. Kumar, D. W.
5 Rusnak, A. K. Takle, D. M. Wilson, E. Hugger, L. F. Wang, F. Karreth, J. C. Lougheed, J. Lee,
6 D. Chau, T. J. Stout, E. W. May, C. M. Rominger, M. D. Schaber, L. S. Luo, A. S. Lakdawala,

7 J. L. Adams, R. G. Contractor, K. S. M. Smalley, M. Herlyn, M. M. Morrissey, D. A. Tuveson
8 and P. S. Huang, Cancer Res., 2006, 66, 11100-11105.
9 22. J. Tsai, J. T. Lee, W. Wang, J. Zhang, H. Cho, S. Mamo, R. Bremer, S. Gillette, J. Kong, N. K.
10 Haass, K. Sproesser, L. Li, K. S. M. Smalley, D. Fong, Y. L. Zhu, A. Marimuthu, H. Nguyen,
11 B. Lam, J. Liu, I. Cheung, J. Rice, Y. Suzuki, C. Luu, C. Settachatgul, R. Shellooe, J.
12 Cantwell, S. H. Kim, J. Schlessinger, K. Y. J. Zhang, B. L. West, B. Powell, G. Habets, C.
13 Zhang, P. N. Ibrahim, P. Hirth, D. R. Artis, M. Herlyn and G. Bollag, Proc. Natl. Acad. Sci.
14 USA, 2008, 105, 3041-3046.
15 23. J. D. Hansen, J. Grina, B. Newhouse, M. Welch, G. Topalov, N. Littman, M. Callejo, S. Gloor,
16 M. Martinson, E. Laird, B. J. Brandhuber, G. Vigers, T. Morales, R. Woessner, N. Randolph, J.
17 Lyssikatos and A. Olivero, Bioorgan. Med. Chem. Lett., 2008, 18, 4692-4695.
18 24. L. Ren, S. Wenglowsky, G. Miknis, B. Rast, A. J. Buckmelter, R. J. Ely, S. Schlachter, E. R.
19 Laird, N. Randolph, M. Callejo, M. Martinson, S. Galbraith, B. J. Brandhuber, G. Vigers, T.
20 Morales, W. C. Voegtli and J. Lyssikatos, Bioorgan. Med. Chem. Lett., 2011, 21, 1243-1247.
21 25. S. Wenglowsky, L. Ren, K. A. Ahrendt, E. R. Laird, I. Aliagas, B. Alicke, A. J. Buckmelter, E.
22 F. Choo, V. Dinkel, B. N. A. Feng, S. L. Gloor, S. E. Gould, S. Gross, J. Gunzner-Toste, J. D.
23 Hansen, G. Hatzivassiliou, B. N. Liu, K. Malesky, S. Mathieu, B. Newhouse, N. J. Raddatz, Y.
24 Q. Ran, S. Rana, N. Randolph, T. Risom, J. Rudolph, S. Savage, L. T. Selby, M. Shrag, K.
25 Song, H. L. Sturgis, W. C. Voegtli, Z. Y. Wen, B. S. Willis, R. D. Woessner, W. I. Wu, W. B.
26 Young and J. Grina, ACS Med. Chem. Lett., 2011, 2, 342-347.
27 26. M. Hirose, M. Okaniwa, T. Miyazaki, T. Imada, T. Ohashi, Y. Tanaka, T. Arita, M. Yabuki, T.
28 Kawamoto, S. Tsutsumi, A. Sumita, T. Takagi, B. C. Sang, J. Yano, K. Aertgeerts, S. Yoshida
29 and T. Ishikawa, Bioorgan. Med. Chem., 2012, 20, 5600-5615.
30 27. S. Wenglowsky, D. Moreno, E. R. Laird, S. L. Gloor, L. Ren, T. Risom, J. Rudolph, H. L.
31 Sturgis and W. C. Voegtli, Bioorgan. Med. Chem. Lett., 2012, 22, 6237-6241.
32 28. H. Lavoie, N. Thevakumaran, G. Gavory, J. J. Li, A. Padeganeh, S. Guiral, J. Duchaine, D. Y.
33 L. Mao, M. Bouvier, F. Sicheri and M. Therrien, Nat. Chem. Biol., 2013, 9, 428-436.
34 29. M. Okaniwa, M. Hirose, T. Arita, M. Yabuki, A. Nakamura, T. Takagi, T. Kawamoto, N.
35 Uchiyama, A. Sumita, S. Tsutsumi, T. Tottori, Y. Inui, B. C. Sang, J. Yano, K. Aertgeerts, S.
36 Yoshida and T. Ishikawa, J. Med. Chem., 2013, 56, 6478-6494.
37 30. M. Karplus and J. A. McCammon, Nat. Struct. Mol. Biol., 2002, 9, 646-652.
38 31. T. Hansson, C. Oostenbrink and W. van Gunsteren, Curr. Opin. Struct. Biol., 2002, 12,
39 190-196.
40 32. J. L. Klepeis, K. Lindorff-Larsen, R. O. Dror and D. E. Shaw, Curr. Opin. Struct. Biol., 2009,
41 19, 120-127.
42 33. R. O. Dror, R. M. Dirks, J. P. Grossman, H. F. Xu and D. E. Shaw, Annu. Rev. Biophys., 2012,
43 41, 429-452.
44 34. C. Venclovas and M. Margelevicius, Proteins, 2005, 61, 99-105.
26
View Article Online
DOI: 10.1039/C6CP06624K
1 35. C. Abrams and G. Bussi, Entropy, 2014, 16, 163-199.

2 36. A. Liwo, C. Czaplewski, S. Oldziej and H. A. Scheraga, Curr. Opin. Struct. Biol., 2008, 18,
3 134-139.

4 37. D. M. Zuckerman, Annu. Rev. Biophys., 2011, 40, 41-62.
5 38. S. A. Adcock and J. A. McCammon, Chem. Rev., 2006, 106, 1589-1615.
6 39. D. Hamelberg, J. Mongan and J. A. McCammon, J. Chem. Phys., 2004, 120, 11919-11929.
7 40. G. Bussi, A. Laio and M. Parrinello, Phys. Rev. Lett., 2006, 96, 090601.
8 41. U. H. E. Hansmann, Chem. Phys. Lett., 1997, 281, 140-150.
9 42. L. Yang and Y. Q. Gao, J. Chem. Phys., 2009, 131, 214109.
10 43. L. Yang, C. W. Liu, Q. Shao, J. Zhang and Y. Q. Gao, Acc. Chem. Res., 2015, 8, 947-955.
11 44. C. W. Liu, F. Wang, L. Yang, X. Z. Li, W. J. Zheng and Y. Q. Gao, J. Phys. Chem. B, 2014,
12 118, 743-751.
13 45. J. Zhang, Y. I. Yang, L. J. Yang and Y. Q. Gao, J. Phys. Chem. B, 2015, 119, 5518-5530.
14 46. A. Fiser and A. Sali, Macromolecular Crystallography, Pt D, 2003, 374, 461-491.
15 47. D. A. Case, T. A. Darden, T. E. III Cheatham, C. L.Simmerling, J. Wang, R. E. Duke, R. Luo,
16 R. C. Walker, W. Zhang, K. M. Merz, B. Roberts, S. Hayik, A. Roitberg, G. Seabra, J. Swails,
17 A. W. Gotz, I. Kolossvary, K. F. Wong, F. Paesani, J. Vanicek, R. M. Wolf, J. Liu, X. Wu, S.
18 R. Brozell, T. Steinbrecher, H. Gohlke, Q. Cai, X. Ye, J. Wang, M.-J. Hsieh, G. Cui, D. R.
19 Roe, D. H. Mathews, M. G. Seetin, R. Salomon-Ferrer, C. Sagui, V. Babin, T. Luchko, S.
20 Gusarov, A. Kovalenko, and P. A. Kollman, AMBER 11, University of California, San
21 Francisco, 2011.
22 48. J. M. Wang, P. Cieplak and P. A. Kollman, J. Comput. Chem., 2000, 21, 1049-1074.
23 49. H. J. C. Berendsen, J. R. Grigera and T. P. Straatsma, J. Phys. Chem., 1987, 91, 6269-6271.
24 50. J. P. Ryckaert, G. Ciccotti and H. J. C. Berendsen, J. Comput. Phys., 1977, 23, 327-341.
25 51. T. Darden, D. York and L. Pedersen, J. Chem. Phys., 1993, 98, 10089-10092.
26 52. P. A. Kollman, I. Massova, C. Reyes, B. Kuhn, S. H. Huo, L. Chong, M. Lee, T. Lee, Y. Duan,
27 W. Wang, O. Donini, P. Cieplak, J. Srinivasan, D. A. Case and T. E. Cheatham, Acc. Chem.
28 Res., 2000, 33, 889-897.
29 53. B. R. Miller, T. D. McGee, J. M. Swails, N. Homeyer, H. Gohlke and A. E. Roitberg, J. Chem.
30 Theory Comput., 2012, 8, 3314-3321.
31 54. M. Feig, J. Karanicolas and C. L. Brooks, J. Mol. Graph. Modell., 2004, 22, 377-395.
32 55. D. Shukla, Y. L. Meng, B. Roux and V. S. Pande, Nat. Commun., 2014, 5, 3397.
33 56. Y. Meng, Y. L. Lin and B. Roux, J. Phys. Chem. B, 2015, 119, 1443-1456.
34 57. B. Nolen, S. Taylor and G. Ghosh, Mol. Cell, 2004, 15, 661-675.
35 58. M. Tong and M. A. Seeliger, ACS Chem. Biol., 2015, 10, 190-200.
36 59. F. Zuccotto, E. Ardini, E. Casale and M. Angiolini, J. Med. Chem., 2010, 53, 2681-2694.
37 60. G. Kar, O. Keskin, A. Gursoy and R. Nussinov, Curr. Opin. Pharmacol., 2010, 10, 715-722.
38 61. A. Spasic, J. Serafini and D. H. Mathews, J. Chem. Theory Comput., 2012, 8, 2497-2505.
39 62. M. Purmonen, J. Valjakka, K. Takkinen, T. Laitinen and J. Rouvinen, Protein Eng. Des. Sel.,
40 2007, 20, 551-559.
41 63. D. W. Li and R. Bruschweiler, J. Phys. Chem. Lett., 2010, 1, 246-248.
42 64. A. Onufriev, D. Bashford and D. A. Case, Proteins, 2004, 55, 383-394.
43 65. K. A. Beauchamp, Y. S. Lin, R. Das and V. S. Pande, J. Chem. Theory Comput., 2012, 8,
44 1409-1414.
27
View Article Online
DOI: 10.1039/C6CP06624K
1 66. V. Hornak, A. Okur, R. C. Rizzo and C. Simmerling, Proc. Natl. Acad. Sci. USA, 2006, 103,
2 915-920.
3 67. A. Weis, K. Katebzadeh, P. Soderhjelm, I. Nilsson and U. Ryde, J. Med. Chem., 2006, 49,

4 6596-6606.
5 68. Y. B. Shan, M. A. Seeliger, M. P. Eastwood, F. Frank, H. F. Xu, M. O. Jensen, R. O. Dror, J.
6 Kuriyan and D. E. Shaw, Proc. Natl. Acad. Sci. USA, 2009, 106, 139-144.
7 69. S. Lovera, L. Sutto, R. Boubeva, L. Scapozza, N. Dolker and F. L. Gervasio, J. Am. Chem.
8 Soc., 2012, 134, 2496-2499.
9 70. J. Wang, Q. Shao, Z. Xu, Y. Liu, Z. Yang, B. P. Cossins, H. Jiang, K. Chen, J. Shi and W. Zhu,
10 J. Phys. Chem. B, 2014, 118, 134-143.
11
12
13
14
15
16
17 Table of Contents Entry
18
19 A combination of a homology modeling technique and an enhanced sampling molecular dynamics

20 simulation implemented with SITS method is performed to compute the detailed map of free-energy
21 landscape and explore the conformational transition pathway of B-RAF kinase.
28

DFG Dynamics

Uploaded by

Copyright:

Available Formats

DFG Dynamics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DFG Dynamics

Uploaded by

Copyright:

Available Formats

View Article Online

acceptance, before technical editing, formatting and proof reading.

You can find more information about Accepted Manuscripts in the

Please note that technical editing may introduce minor changes

1 Energetics and Structural Characterization of the

Physical Chemistry Chemical Physics Accepted Manuscript

3 Kinase: A SITS Molecular Dynamics Study

9 qshao@mail.shcnc.ac.cn; Jiye Shi, Jiye.Shi@ucb.com.

2 activity of B-RAF is regulated by a “DFG-flip” conformational transition between functional DFG-in

Physical Chemistry Chemical Physics Accepted Manuscript

4 greater difficulty in experimentally capturing high-energy-level transient structures renders elusive

7 energetically characterize the conformational transition pathway of B-RAF on multi-dimensional

13 suitable pathway for the development of B-RAF inhibitor.

21 Keywords: B-RAF kinase, DFG-flip conformational transition, homology modeling technique,

22 enhanced sampling molecular simulation, hydrophobic interactions

2 It is well-known that proteins usually undergo sophisticated conformational transitions while

Physical Chemistry Chemical Physics Accepted Manuscript

11 protein interconverts from one functional structure to another is of fundamental importance to

14 structural dynamics of proteins are indispensable for computer-aided drug design.11-15

16 (MAPK) signaling pathway (RAS-RAF-MEK-ERK) which transduces signals from membrane-based

Physical Chemistry Chemical Physics Accepted Manuscript

4 conserved DFG motif (Asp594-Phe595-Gly596) in kinase protein superfamily.19-29 As shown in Fig.

Physical Chemistry Chemical Physics Accepted Manuscript

6 long-time conformational transition of protein, both a priori knowledge of high-resolution structure

8 lacks experimentally identified full-length A loop configuration, homology modeling methodologies

11 conventional MD simulation in sampling high-dimensional and rugged configuration space of

15 constructed by homology modeling may have high risk.

17 efficiency of MD simulation.35-38 By smoothing potential energy surface, MD simulation

19 to effectively search configuration space and directly observe, characterize functionally

Physical Chemistry Chemical Physics Accepted Manuscript

4 massive computational resources. Recently, Yang and Gao proposed a selective

7 introducing a sum-over-temperature non-Bolzmann distribution factor in molecular simulation (see

13 In the present study, we attempted to use explicit-solvent all-atom MD simulations implemented

Physical Chemistry Chemical Physics Accepted Manuscript

7 pathway for inhibitor development.

Physical Chemistry Chemical Physics Accepted Manuscript

12 sum-over-temperature non-Boltzmann distribution factor. The potential energy distribution

14 effective force corresponding to the following effective potential energy:

22 surface and conformation collection.42, 43

Physical Chemistry Chemical Physics Accepted Manuscript

4 selective sampling enhancement on specific position can avoid unnecessary oversampling of

9 EA-P(-A), A loop-water: EA-W, remaining part of protein-water: EP(-A)-W). That is to say:

10 U=EA+EP(-A)+EW+EA-P(-A)+EA-W+EP(-A)-W. The effective potential can be rewritten as the following

16 K were utilized in Eq. 2.

17 It is noteworthy that a converged calculation of SITS-MD simulation yields a biased distribution

19 temperature T0 can be easily recovered as:

20 ρ(r ) = ρeff (r )e − β (U (r )−U eff (r ))

21 where ρeff (r ) is measured in SITS-MD simulation. Thermodynamic quantities can be calculated

22 once ρ(r ) is known.

1 Conventional MD simulation details. To evaluate the structural stability of the DFG-out

Physical Chemistry Chemical Physics Accepted Manuscript

6 artificially added on protein or the perturbation from SITS method.

7 MM-PB/GBSA calculation details. To indicate the inter-residue interactions which experience

10 standard approach of MM/GBSA.52 Multiple representative structures of DFG-in, DFG-out, and

17 segments of protein were extracted for data analysis.

19 Reversible active-inactive transition visiting in individual trajectories. A total of 9

Physical Chemistry Chemical Physics Accepted Manuscript

6 (EP(-A)+EW+EA-P(-A)+EP(-A)-W) is maintained in similar range as that in conventional MD simulation.

10 (Gln493-Lys507), αE helix (Met550-Lys570), catalytic loop (Ile572-Asn581), F loop (Ile178-Ser187)