Elsevier

Bioorganic & Medicinal Chemistry

Volume 18, Issue 18, 15 September 2010, Pages 6805-6812
Bioorganic & Medicinal Chemistry

Insights into MAPK p38α DFG flip mechanism by accelerated molecular dynamics

https://doi.org/10.1016/j.bmc.2010.07.047Get rights and content

Abstract

The DFG motif at the beginning of the activation loop of the MAPK p38α undergoes a local structural reorganization upon binding of allosteric type-II and type-III inhibitors, which causes the residue F169 to move from a buried conformation (defined as DFG-in) to a solvent exposed conformation (defined as DFG-out). Although both experimental and computer simulation studies had been performed with the aim of unveiling the details of the DFG-in to DFG-out transition, the molecular mechanism is still far from being unequivocally depicted.

Here, the accelerated molecular dynamics (AMD) technique has been applied to model the active loop flexibility of p38α and sample special protein conformations which can be accessible only in some conditions or time periods. Starting from the assumption of an experimentally known initial and final state of the protein, the study allowed the description of the interaction network and the structural intermediates which lead the protein to change its loop conformation and active site accessibility. Besides a few important hydrogen bond interactions, a primary role seems to be played by cation–π interactions, involving the DFG-loop residue F169, which participate in the stabilization of an intermediate conformation and in its consequent transition to the DFG-out conformation. From this study, insights which may prove useful for inhibitor design and/or site directed mutagenesis studies are derived.

Introduction

The p38α proteins belong to the mitogen-activated protein kinases (MAPKs), a super-family of enzymes that is involved in many critical cellular processes such as proliferation, apoptosis and differentiation.1 Their activation mechanism is based on a dual phosphorylation reaction taking place on specific threonine and tyrosine residues belonging to the flexible activation loop responsible for the opening and the closing of the binding cavity,2, 3 where one ATP molecule binds during the activation process (see Fig. 1a).

The activation loop carries a DFG (Asp-Phe-Gly) motif, largely conserved among most of the proteins belonging to the kinase protein super-family.4 The conformations assumed by this flexible loop (or DFG-loop) are mainly characterized by the shift of the F side-chain (F169 in the p38α protein) from its usual buried conformation (DFG-in) (see Fig. 1b) to a conformation that sterically interferes with ATP binding (DFG-out) (see Fig. 1c and d).

To date, a large number of completely resolved DFG-in three-dimensional structures, both in their apo and ligand-complexed forms, are collected in the RSCB Protein Data Bank,5 thus simplifying the development of new drugs designed to compete with ATP binding (type-I inhibitors6, 7, 8) (Fig. 1b). On the contrary, the number of published DFG-out three-dimensional structures with the activation loop completely resolved is not as significant. Moreover, until now, this latter conformation of the loop has been experimentally observed only in the presence of particular inhibitors based on a modified urea or amide scaffold capable to bind meanwhile to the ATP site and to a nearby allosteric hydrophobic pocket which becomes accessible by the flip of the DFG-loop (type-II inhibitors9) or exclusively within the allosteric pocket (type-III inhibitors10) (Fig. 1c and d). The urea/amide moiety of the allosteric inhibitors is considered to be fundamental for the DFG flip11 triggered by the establishment of two specific H-bonds with residues D168 (belonging to the activation loop) and residue E71 (belonging to Helix-C).12, 13

Whereas the hypothesis of a DFG-out conformation induced by allosteric inhibitors is generally supported by the results of X-ray experiments and by their slow binding kinetics experimentally observed,14, 12 the existence of a dynamic equilibrium between the DFG-in and DFG-out conformations of p38α in its apo and complexed form has been first advanced on the basis of the results of NMR studies.15 The authors hypothesize that the DFG-out state is sampled less frequently than the DFG-in conformation; type-II inhibitors bind to the DFG-out form, freezing the protein in this conformation and suppressing the conformational exchange, while the binding of traditional inhibitors (ATP-site binders) affects the DFG equilibrium to a lesser extent. Besides NMR, Molecular Dynamics (MD) simulations are also being used extensively to sample the accessible protein conformations and build adequate pharmacophoric models. In particular, the dynamic behaviour of the DFG-loop was taken into account in a recent paper by Fremberg-Kesner and Elcock,16 who applied a high temperature (HT) molecular dynamics simulation to fasten the DFG-in to DFG-out transition and back, combined with a docking protocol to analyze the ligand binding properties of the different conformations. They described the presence of two transitional conformations defined as pseudo-DFG-in and pseudo-DFG-out, due to their respective similarity to the DFG-in and DFG-out states, and demonstrated that type-II inhibitors are able to bind not only to the p38α DFG-out state but also to its DFG-in state. Thus, this observation supports the hypothesis that the pre-existence of the DFG-out conformation is not a prerequisite for inhibitor binding to p38α, rather the DFG-out conformation might be triggered by the initial binding of type-II-inhibitor into the DFG-in conformation.

Recently, a detailed interpretation of the DFG flip in the Abl kinase has been given in the paper by Shana et al.17 Here, a combined experimental and theoretical approach, was applied to demonstrate that the DFG flip is regulated by the protonation of the DFG-loop aspartate residue; this allows the formation of a H-bond between the aspartate side-chain and the backbone carbonyl of a nearby valine, which promotes the DFG-loop conformational switch. Also in this work, evidence is provided for the existence of a stable intermediate conformation, which allows binding of a specific DFG-out binder inhibitor.

This thesis finds support in other computational and experimental studies recently performed. In particular, Kufareva and Abagyan,9 by applying a structure-based approach to design and evaluate type-II kinase inhibitors, demonstrated that the determinants for allosteric binding are preserved in most DFG-in structures. In 2007, an experimental analysis of the DFG conformational transition in p38α has been attempted by Bukhtiyarova et al.18 They classified five p38α DFG-motif mutants (F169G, F169R, F169Y, F169A and D166G) according to their local DFG conformation and, beside pointing out the determinant role played by residue F169 in the structural dynamics of the activation loop, they suggested the existence in the free protein of a peculiar spatial organization, called α-DFG-out, which could be intended as an intermediate between the DFG-in and the DFG-out conformations and which could provide a transient root of entry for type-II inhibitors. Similar intermediate conformations, also named DFG-‘in between’, were found depending on the structure of the ligands co-crystallized with the p38α proteins manipulated in positions different from the DFG-motive and under different crystallization conditions.19, 20, 21, 10 In summary, two mechanisms have been up to now envisaged in the literature for allosteric inhibitors: (1) exclusively binding to proteins in the DFG-out conformation, which, although sampled less frequently, exists in equilibrium with the DFG-in conformation; (2) initial binding to proteins in the DFG-in conformation with subsequent promotion of the DFG-in to -out transition. Besides the recent evidences brought to support this second thesis, no description of the mechanism at the atomic level has yet produced.

In this work, a computational procedure, based on an Accelerated Molecular Dynamics (AMD) technique, will be applied to the human p38α protein, in order to highlight the main interactions which take part in the DFG-in to DFG-out conformational change triggered by the binding of allosteric inhibitors. A viable transition mechanism will be proposed and implications to drug design will be discussed.

Section snippets

Materials and methods

The accelerated molecular dynamics technique, as implemented in GROMACS 3.3.3,22, 23, 24 has been applied to neglect non-energetically convenient pathways by driving the system to a pre-defined direction.

The X-ray structure of p38α from mouse (PDB ID: 1P38) was chosen as the starting conformation since the whole structure is resolved. First, this structure was changed to the human sequence by computationally mutating H48 and A263 into L48 and T263, respectively, with the software QUANTA

Results and discussion

The results of the AMD simulations performed in this work are analyzed to obtain a picture of the interaction network and the structural intermediates which lead the protein to change its active site conformation. The study is based on the assumption of fan-shaped motion of F169 during the DFG transition between the initial DFG-in and final DFG-out states determined by X-ray crystallography.

The inter-residue interactions involved in the DFG flip

The analysis of the interaction energies (IE) between residues in the binding site region, together with detailed visual inspections of the protein structure conformations along the whole simulations, allow the determination of the main interactions responsible for the DFG-in/out transition.

The initial DFG-in conformation is characterized by the presence of two H-bonds between two residues of the DFG loop (L167 and G170) and one residue (H145) belonging to the C-terminal domain. By analyzing

Conclusion

This work is based on the assumption that the conformational change of the activation loop is induced by the presence of allosteric inhibitors in the binding cavity in agreement with recently published findings which support the idea that it is more probable for allosteric inhibitors to bind first a DFG-in protein and then turn it into a DFG-out state than to encounter a free protein in the DFG-out conformation and directly bind to it.

The DFG-in to DFG-out mechanism proposed here describes at

References and notes (32)

  • J. Adams et al.

    Prog. Med. Chem.

    (2001)
  • F. Bernstein et al.

    Arch. Biochem. Biophys.

    (1978)
  • K. Wilson et al.

    Chem. Biol.

    (1997)
  • B. Okram et al.

    Chem. Biol.

    (2006)
  • R. Angell et al.

    Bioorg. Med. Chem. Lett.

    (2008)
  • T. Frembgen-Kesner et al.

    J. Mol. Biol.

    (2006)
  • H. Berendsen et al.

    Comp. Phys. Commun.

    (1995)
  • D. Dougherty

    J. Nutr.

    (2007)
  • W. Humphrey et al.

    J. Mol. Graphics

    (1996)
  • G. Pearson et al.

    Endocr. Rev.

    (2001)
  • J. Subramanian et al.

    ChemMedChem

    (2008)
  • A. Kornev et al.

    Proc. Natl. Acad. Sci. U.S.A.

    (2006)
  • A. Gill et al.

    J. Med. Chem.

    (2005)
  • F. Zuccotto et al.

    J. Med. Chem.

    (2010)
  • I. Kufareva et al.

    J. Med. Chem.

    (2008)
  • J. Simard et al.

    J. Am. Chem. Soc.

    (2010)
  • Cited by (36)

    • Multiscale computational study of ligand binding pathways: Case of p38 MAP kinase and its inhibitors

      2021, Biophysical Journal
      Citation Excerpt :

      Because the structure of apo DFG-out p38 is still not available by crystallography, one outstanding question is whether the presence of the DFG-out conformation is induced by ligand binding. Early accelerated MD simulations (57) proposed an induced-fit hypothesis, whereas both NMR (55) and free energy studies (18) suggested a dynamic equilibrium model in which the two DFG positions have an equal population in the apo state. In our ∼270 simulations of ligand association, the ligands did not show any tendency to change the DFG position in either the diffusion or binding steps (Fig. S7), illustrating that the DFG conformation remained the same during all simulations.

    • Discovery of potent p38α MAPK inhibitors through a funnel like workflow combining in silico screening and in vitro validation

      2019, European Journal of Medicinal Chemistry
      Citation Excerpt :

      As a continuation of our previous efforts in the p38α MAPK field [10,11,14,19], we herein report the development and application of a workflow (Fig. 2) for the rational identification of novel scaffolds potentially acting as p38α MAPK TI½-Is. Available co-crystal structures of p38α MAPK with TI½-Is [11] show that, similarly to other kinases, the ATP-binding site of p38α MAPK has a high degree of dynamism, as different structural elements surrounding this pocket, i.e. the glycine-rich loop (residues 31–37) [20], activation loop (residues 171–192) [21], DFG motif (residues 168–170) [22], αC-helix (residues 62–77) [23] and the hinge region, can be found in different conformations, influencing in turn the size, shape and chemical properties of the active site. For these reasons, the effect of pocket plasticity on pose prediction and virtual screening performance was evaluated through: (i) cross-docking experiments, in which each target model was analyzed for its ability to both reproduce the crystallographic position of known TI½-Is and retrieve the higher number of correctly predicted poses; (ii) assessment of the agreement between the experimental activity and the estimated HYDE ligand affinity [24] for the analyzed inhibitors.

    • Phenylindenone isomers as divergent modulators of p38α MAP kinase

      2016, Bioorganic and Medicinal Chemistry Letters
    • Drugging MYCN through an Allosteric Transition in Aurora Kinase A

      2014, Cancer Cell
      Citation Excerpt :

      To identify such CD inhibitors we synthesized a set of compounds with either diaminopyrimidine (VX-680-like) or pyrazolopyrimidine (PP-1-like) scaffolds (Figure 1A) predicted to induce a large structural shift in Aurora A. Derivatives of each of these scaffolds were known to bind to Aurora A. Structural data were available on both scaffolds bound to related kinases, and the routes to their synthesis were tractable. To these ATP-competitive cores, we fused biphenyl urea and amide moieties predicted to stabilize distinct conformations of Aurora A (Dietrich et al., 2010; Filomia et al., 2010). To test whether this panel of 32 putative CD inhibitors would destabilize MYCN, we initially treated Kelly MYCN-amplified neuroblastoma cells with these compounds and measured MYCN protein using western blot.

    • Insights from free-energy calculations: Protein conformational equilibrium, driving forces, and ligand-binding modes

      2012, Biophysical Journal
      Citation Excerpt :

      Computational methods can help shed light on these issues. Molecular-dynamics and Monte Carlo methods with explicit waters are alternative methods to study protein dynamics (6–10). Such methods can also be used to study free-energy pathways, such as free-energy perturbation (FEP) and thermodynamic integration (TI), to compute the absolute or relative work of binding ligands to a protein (11–18).

    View all citing articles on Scopus
    View full text