Inhibition of 3-Hydroxykynurenine Transaminase from Aedes aegypti and Anopheles gambiae: A Mosquito-Specific Target to Combat the Transmission of Arboviruses

Abstract

Arboviral infections such as Zika, chikungunya, dengue, and yellow fever pose significant health problems globally. The population at risk is expanding with the geographical distribution of the main transmission vector of these viruses, the Aedes aegypti mosquito. The global spreading of this mosquito is driven by human migration, urbanization, climate change, and the ecological plasticity of the species. Currently, there are no specific treatments for Aedes-borne infections. One strategy to combat different mosquito-borne arboviruses is to design molecules that can specifically inhibit a critical host protein. We obtained the crystal structure of 3-hydroxykynurenine transaminase (AeHKT) from A. aegypti, an essential detoxification enzyme of the tryptophan metabolism pathway. Since AeHKT is found exclusively in mosquitoes, it provides the ideal molecular target for the development of inhibitors. Therefore, we determined and compared the free binding energy of the inhibitors 4-(2-aminophenyl)-4-oxobutyric acid (4OB) and sodium 4-(3-phenyl-1,2,4-oxadiazol-5-yl)butanoate (OXA) to AeHKT and AgHKT from Anopheles gambiae, the only crystal structure of this enzyme previously known. The cocrystallized inhibitor 4OB binds to AgHKT with K _i of 300 μM. We showed that OXA binds to both AeHKT and AgHKT enzymes with binding energies 2-fold more favorable than the crystallographic inhibitor 4OB and displayed a 2-fold greater residence time τ upon binding to AeHKT than 4OB. These findings indicate that the 1,2,4-oxadiazole derivatives are inhibitors of the HKT enzyme not only from A. aegypti but also from A. gambiae.

Figure 1

HKT overall structure. (A) The asymmetric unit of AeHKT is composed of four monomers (chain A in purple, chain B in yellow, chain C in blue, and chain D in green), related to A–A′ (homodimer) and C–C′ axes (asymmetric unit). (B) Detail of the protomer A fold, containing the N-terminal region in green, Rossmann-like motif in blue, and C-terminal in orange. The pyridoxal-5′-phosphate (PLP) cofactor is located at the intermediate region of the protomer, in which the active site is composed by the interface between two monomers of one homodimer. (C) Close-up of the active site at the homodimer interface (chain C in blue and chain D in green) and most important residues interacting with PLP (orange). PLP of chain C with an electronic density map is shown.

Figure 2

Comparison of AeHKT and homologues. (A) Sequence alignment for HKT from A. aegypti (AeHKT^6MFB), A. gambiae (AgHKT^2CH1), and AGT from A. aegypti (AeAGT^2HUF). Alignment colors indicate red (conserved residues), yellow (conservative mutation), and white nonconservative mutations. Correspondent 6MFB secondary structures are indicated above the alignment using α (helix), ß (sheet), and η (turn). Evidence suggests that AeHKT/AgHKT are significantly closer together than AeAGT/AeHKT. (B) Structural superposition of AB homodimers of AeHKT, AgHKT, and AeAGT X-ray structures. Conserved and dissonant residues are highlighted in a ramp color between blue and red, respectively. Nonconservative mutations are seen in areas exposed to solvent (red). (C) Detail of the identical active site residues for superimposed X-ray structures of B (chain A). A common ancestor between AGT and HKT was duplicated to create HKT, as indicated by Chen et al. Consequently, the functional variations between 6MFB and 2HUF, where the only active site mismatch is between positions A78 and G79, seem to be due to substrate recognition rather than active site polymorphism. Alignment and a portion of the analysis had been completed with the ENDscript webserver.

Figure 3

Time-series properties and conformations obtained from MD simulations. (A) Violin plot of average RMSD between the α carbons from the simulated structural ensemble and the crystallographic structure or lowest energy docking structure. (B) RMSF per residue of the α carbons calculated for the last 50 ns of simulation. (C) Open and closed conformations of AeHKT-OXA upon OXA binding. The residues A-Gln²⁵³-A-Arg²⁵⁵ in the loop connecting helices α8 and α9 are shown as solid pink surfaces. The OXA ligand is shown in van der Waals radii in element-color code. Red arrows show the position of the active-site embedded ligands.

Figure 4

Initial and final conformations of four systems from MD simulations. AeHKT (green/blue) and AgHKT (pink/blue) bound to 4OB (yellow) and OXA (orange). The PLP cofactor and closest neighboring residues are also shown. Chemical structures of the inhibitors 4-(2-aminophenyl)-4-oxobutyric acid (4OB) and sodium 4-(3-phenyl-1,2,4-oxadiazol-5-yl)butanoate (OXA) are also shown.

Figure 5

Three-dimensional free energy landscape for the dissociation of the ligands from the HKT enzymes as a function of the chosen CVs. The landscape is color-coded from red (high regions on the configurational space) to blue (low energy basins).

Figure 6

(A) Two-dimensional free energy surface landscapes for the dissociation of the ligands from the HKT enzymes depicted through the potential of mean force as a function of CV1. (B) Binding free energies calculated by meta-MD simulations, inferred from the free energy landscape for the dissociation of the ligands from the active site of the HKT enzymes. The latter values are shown for the dissociation of the ligands using one (top) or two (bottom) CV for the dissociation of the respective ligands.

Figure 7

Relative residence time between the ligands and the HKT enzymes averaged over 6 τRAMD simulation replicas applying a random force of 20 kJ mol^–1 Å^–1 magnitude. The values of τ were averaged over six replicas, each one with 15 trajectories. The statistics to assess the convergence for each replica are shown in Figures S2–S5 from the electronic supplemental information (ESI).