Characterization of Two Late-Stage Enzymes Involved in Fosfomycin Biosynthesis in Pseudomonads

Abstract

The broad-spectrum phosphonate antibiotic fosfomycin is currently in use for clinical treatment of infections caused by both Gram-positive and Gram-negative uropathogens. The antibiotic is biosynthesized by various streptomycetes, as well as by pseudomonads. Notably, the biosynthetic strategies used by the two genera share only two steps: the first step in which primary metabolite phosphoenolpyruvate (PEP) is converted to phosphonopyruvate (PnPy) and the terminal step in which 2-hydroxypropylphosphonate (2-HPP) is converted to fosfomycin. Otherwise, distinct enzymatic paths are employed. Here, we biochemically confirm the last two steps in the fosfomycin biosynthetic pathway of Pseudomonas syringae PB-5123, showing that Psf3 performs the reduction of 2-oxopropylphosphonate (2-OPP) to (S)-2-HPP, followed by the Psf4-catalyzed epoxidation of (S)-2-HPP to fosfomycin. Psf4 can also accept (R)-2-HPP as a substrate but instead performs an oxidation to make 2-OPP. We show that the combined activities of Psf3 and Psf4 can be used to convert racemic 2-HPP to fosfomycin in an enantioconvergent process. X-ray structures of each enzyme with bound substrates provide insights into the stereospecificity of each conversion. These studies shed light on the reaction mechanisms of the two terminal enzymes in a distinct pathway employed by pseudomonads for the production of a potent antimicrobial agent.

Figure 1

(A) Convergent biosynthetic pathways for the production of fosfomycin as found in different genera. For the pathway in pseudomonads (top), the reactions that have been confirmed by experimental data are shown as single transformations. For the pathway from streptomycetes (bottom), only the methyl transfer step has not been definitively confirmed and is shown with dashed lines. For the pathway in Pseudomonas, the epoxide oxygen in fosfomycin is hypothesized to originate from the β-carbonyl of PnPy (shown in red). This has been confirmed for the pathway in Streptomyces. (B-C) Reactions catalyzed by the orthologous enzymes Psf4 and HppE on different enantiomers of 2-HPP. The reactions are drawn using oxygen and an electron source a performed initially, but recent studies show that the reactions can also proceed using hydrogen peroxide as a co-substrate.

Figure 2

(A) Michaelis-Menten curve for the wild-type Psf3 reaction using 2-OPP as a substrate. Error bars represent the standard deviations calculated from measurements made in triplicate. (B) ³¹P NMR spectroscopic analysis of the reverse (oxidation) activity of Psf3 using the two enantiomers of 2-HPP. Reactions using (S)-2-HPP resulted in a new peak that is consistent with 2-OPP (as determined by spiking with authentic standards), while no new product could be observed with (R)-2-HPP.

Figure 3

(A) Overall structure of the Psf3 homodimer showing the orientation of the Rossmann fold (in cyan and black) and dimerization domains (in green). The second monomer is colored in gray. The location of bound ligands NADP⁺ (yellow) and 2-OPP (purple) are as shown, and Arg212 that is involved in crossover interactions between the two monomers is also indicated in brown sticks. (B) Simulated annealing difference Fourier maps (Fo-Fc) of Psf3 complexes contoured to 2.5σ (blue) showing the bound NADP⁺, and (C) 2-OPP. The coordinates for the ligand were omitted prior to map calculations. The final refined coordinates of the complexes are superimposed with important active site residues and ligands shown in stick representation. Hydrogen bond interactions are represented with dashed lines.

Figure 4

Structure-based multiple sequence alignment of Psf3 with other members of the 3-hydroxyacid dehydrogenase family. Secondary structural elements are shown above the alignment. Residues in Psf3 that interact with NADP⁺ are indicated with a blue triangle, and residues that contact 2-OPP are indicated with a yellow circle.

Figure 5

(A) Overall structure of Psf4 showing the cupin (purple) and dimerization (green) domains. A linker region (brown) completes the β-sheet cupin fold of an adjacent monomer (in gray) (B) Structure of HppE that contains a cupin domain (pink) that is structurally conserved with that in Psf4 and a dimerization domain (olive) that contains a separate protein fold. A linker region (tan) completes the β-sheet cupin fold of an adjacent monomer (in gray) (C) Structure of the Psf4 tetramer showing the crossover interactions that occur as residues Arg20 and Lys21 from one monomer are diverted into the active site of a different monomer. The catalytically requisite metal ion is shown as a gray sphere. (D) Close-up view of the active site of Psf4 in complex with (S)-2-HPP (brown) and (E) (R)-2-HPP (cyan). Important active site residues are shown as purple sticks. A simulated annealing difference Fourier maps (Fo-Fc), calculated with the coordinates of the bound substrate omitted, is shown at a contour level of 2.5σ (green) and 5σ (red).