Remarkable similarity in Plasmodium falciparum and Plasmodium vivax geranylgeranyl diphosphate synthase dynamics and its implication for antimalarial drug design

Abstract

Malaria, mainly caused by Plasmodium falciparum and Plasmodium vivax, has been a growing cause of morbidity and mortality. P. falciparum is more lethal than is P. vivax, but there is a vital need for effective drugs against both species. Geranylgeranyl diphosphate synthase (GGPPS) is an enzyme involved in the biosynthesis of quinones and in protein prenylation and has been proposed to be a malaria drug target. However, the structure of P. falciparumGGPPS (PfGGPPS) has not been determined, due to difficulties in crystallization. Here, we created a PfGGPPS model using the homologous P.vivaxGGPPS X-ray structure as a template. We simulated the modeled PfGGPPS as well as PvGGPPS using conventional and Gaussian accelerated molecular dynamics in both apo- and GGPP-bound states. The MD simulations revealed a striking similarity in the dynamics of both enzymes with loop 9-10 controlling access to the active site. We also found that GGPP stabilizes PfGGPPS and PvGGPPS into closed conformations and via similar mechanisms. Shape-based analysis of the binding sites throughout the simulations suggests that the two enzymes will be readily targeted by the same inhibitors. Finally, we produced three MD-validated conformations of PfGGPPS to be used in future virtual screenings for potential new antimalarial drugs acting on both PvGGPPS and PfGGPPS.

Keywords: Plasmodium falciparum; Plasmodium vivax; GGPPS; Gaussian accelerated molecular dynamics simulations; Malaria; homology model; molecular dynamics simulations.

Figure 1.

A) Sequence alignment between Pv and PfGGPPS, with different residues highlighted in pale green. B) Localization of non-identical residues in the template structure.

Figure 2.

QMEAN density plots for the modeled PfGGPPS structures in apo (A) or GGPP-bound (B) states, and also for the corresponding crystallographic (template) structures of PvGGPPS in apo (C) or GGPP-bound states (D). Red vertical lines indicate the QMEAN of our structures as compared to a distribution of QMEAN values obtained for a reference set of proteins of similar size. The resulting (QMEAN) Z-scores are indicated in black.

Figure 3.

α-carbon RMSD of GGPPS monomers in 100 ns of conventional (A) or Gaussian accelerated (B) MD simulations. Light grey lines represent the RMSD of GGPPS monomers in each MD replica, dark grey lines represent the corresponding running average over 100 steps, and colored lines indicate the average RMSD.

Figure 4.

RMSF of GGPPS residues in conventional (A) or Gaussian accelerated (B) MD simulations. The RMSF was calculated for the α-carbons, with respect to the average structure The colored lines represent the averaged RMSF for all monomer replicas in the same system: Apo PvGGPPS (blue), Apo PfGGPPS (red), GGPP-bound PvGGPPS (cyan), or GGPP-bound PfGGPPS (pink).

Figure 5.

Conformational dynamics of Plasmodium GGPPS. A) ‘Sausage’ plot of PfGGPPS colored according to the root mean square fluctuation (RMSF; in Angstrom), showing that most of the enzyme’s flexibility is concentrated in the L9–10 loop, which forms the ‘lid’ to the active site. B) Principal Component Analysis of PfGGPPS and PvGGPPS simulations in apo- and GGPP-bound states, colored according to the distance between loop 2–3 and loop 9–10. The histograms show that apo-systems can sample conformations with larger PC1 values, which correspond to ‘open’ conformations. Inset: the distance, D, refers to the separation between residue 301 (L9–10) and residue 81 (L2–3). C) MD snapshots illustrate how the main motion (PC1) controls the accessibility to the binding pocket.

Figure 6.

A) Binding site accessibility as measured by the distances between K81 and K301, and K136 and K301. The ligand (GGPP) is shown in purple as a reference. B) Normalized distance histograms for conventional (upper panels) or Gaussian accelerated (bottom panels) MDs.

Figure 7.

Salt bridges between GGPP and PfGGPPS (pink) or PvGGPPS (blue). In both cMD (A) and GaMD (B) simulations salt bridges are formed between the ligand, GGPP, and positively charged residues located at L2–3 (K81), L4–5 (R135, R136) and L9–10 (K301).

Figure 8.

Pocket flexibility in PfGGPPS and PvGGPPS. A) Three binding pocket conformations sampled in the MD simulations and their correspondent volume histograms. The wireframe maps correspond to regions that are empty for at least 50% of the simulation time. For clarity, the GGPP side-chain is shown in yellow. Clustering of pocket shapes was performed with POVME 3.0^[32] using a Tanimoto-metric algorithm. B) Frequency of each pocket conformation throughout the MD simulations. PfGGPPS and PvGGPPS apo- states visit conformations 1 and 3, while GGPP-bound states visit conformations 1 and 2.

Figure 9.

Residues controlling the (A) polar or (B) apolar sub-pockets of PfGGPPS and PvGGPPS in clusters 1 (cyan), 2 (magenta), and 3 (purple). Representative occupancies (> 15% of simulation time) of key side-chains are illustrated by meshes. To provide a reference of the binding pockets, the structures were superimposed with zoledronate (Zol), an inhibitor that mimics the natural allylic substrate, and/or with the product, GGPP (in yellow).