PcxL and HpxL are flavin-dependent, oxime-forming N-oxidases in phosphonocystoximic acid biosynthesis in Streptomyces

Abstract

Several oxime-containing small molecules have useful properties, including antimicrobial, insecticidal, anticancer, and immunosuppressive activities. Phosphonocystoximate and its hydroxylated congener, hydroxyphosphonocystoximate, are recently discovered oxime-containing natural products produced by Streptomyces sp. NRRL S-481 and Streptomyces regensis NRRL WC-3744, respectively. The biosynthetic pathways for these two compounds are proposed to diverge at an early step in which 2-aminoethylphosphonate (2AEPn) is converted to (S)-1-hydroxy-2-aminoethylphosphonate ((S)-1H2AEPn) in S. regensis but not in Streptomyces sp. NRRL S-481). Subsequent installation of the oxime moiety into either 2AEPn or (S)-1H2AEPn is predicted to be catalyzed by PcxL or HpxL from Streptomyces sp. NRRL S-481 and S. regensis NRRL WC-3744, respectively, whose sequence and predicted structural characteristics suggest they are unusual N-oxidases. Here, we show that recombinant PcxL and HpxL catalyze the FAD- and NADPH-dependent oxidation of 2AEPn and 1H2AEPn, producing a mixture of the respective aldoximes and nitrosylated phosphonic acid products. Measurements of catalytic efficiency indicated that PcxL has almost an equal preference for 2AEPn and (R)-1H2AEPn. 2AEPn was turned over at a 10-fold higher rate than (R)-1H2AEPn under saturating conditions, resulting in a similar but slightly lower k_cat/K_m We observed that (S)-1H2AEPn is a relatively poor substrate for PcxL but is clearly the preferred substrate for HpxL, consistent with the proposed biosynthetic pathway in S. regensis. HpxL also used both 2AEPn and (R)-1H2AEPn, with the latter inhibiting HpxL at high concentrations. Bioinformatic analysis indicated that PcxL and HpxL are members of a new class of oxime-forming N-oxidases that are broadly dispersed among bacteria.

Keywords: enzyme kinetics; flavin adenine dinucleotide (FAD); natural product biosynthesis; nicotinamide; oxidase; oxime; phosphonates; phosphonic acid; substrate specificity.

Figure 1.

Oxime-containing natural products.

Figure 2.

The proposed biosynthetic route of oxime formation in phosphonocystoximates. HpxV is a 2-oxoglutarate- and ferrous iron-dependent 2AEPn dioxygenase that converts 2AEPn to (S)-1H2AEPn. PcxL and HpxL are FAD- and NADPH-dependent amine oxidases that catalyze 2AEPn or 1H2AEPn oxidation to yield 2-hydroxyiminoethylphosphonic acid (1) and 1-hydroxy-2-hydroxyiminoethylphosphonic acid (3), respectively. Multiple steps are required to convert these oxime-containing intermediates to the final products (dashed-line arrows).

Figure 3.

Phosphorus NMR spectroscopy analysis of N-oxidase activity in the presence of 2AEPn or (S)-1H2AEPn. A, HpxL and PcxL form three major products in vitro when incubated for 16 h with NADPH, FAD, and 2AEPn as the substrate, whereas no product formation is observed in the absence of enzyme. B, these products have been assigned as the (E)- and (Z)-isomers of 2-iminoethylphosphonic acid (1) and 2-nitroethylphosphonic acid (2) based on the ¹H-³¹P HMBC analysis (Figs. S2 and S3). Note that the (E)- and (Z)-isomers of 1 have identical ³¹P chemical shifts and thus display only a single peak in the one-dimensional ³¹P spectrum shown in A. C, HpxL and PcxL both generate two products in vitro using (S)-1H2AEPn as the substrate. D, these products were assigned as (Z)-1-hydroxy-2-hydroxyiminoethylphosphonic acid (3) and 1-hydroxy-2-nitroethylphosphonic acid (4) based on the ¹H-³¹P HMBC analysis (Fig. S4 and S5). Chemical-mixing experiments with the control reaction were performed to confirm that substrate had been completely consumed. Substrates and cofactors were added at the following concentrations: 100 μm FAD, 500 μm NADPH, 3 mm 2AEPn or HpxV-generated (S)-1H2AEPn, 25 μm PcxL or HpxL, 25 mm phosphite, and 10 μm PTDH17x.

Figure 4.

Kinetic curves for PcxL and HpxL in the presence of different substrates. PcxL (A) and HpxL (B) with 2AEPn as the substrate were fit to a standard Michaelis-Menten model. PcxL (C) and HpxL (D) with (S)-1H2AEPn were fit to a standard Michaelis-Menten model. PcxL (E) with (R)-1H2AEPn as a substrate was fit to a standard Michaelis-Menten model, and HpxL (F) with (R)-1H2AEPn as substrate was fit to the substrate inhibition model. Extracted kinetic constants are shown in Table 4.

Figure 5.

Cladograms showing the phylogenetic distribution of putative N-oxidases related to HpxL and PcxL. A, phylogeny of the 500 closest HpxL homologs. hpxL-containing gene clusters that also contain pepM are highlighted in orange. B, phylogeny of actinobacterial PepM sequences. pepM-containing gene clusters that contain an HpxL homolog are highlighted in orange. HpxL is denoted with a blue asterisk, PcxL is denoted with a purple asterisk, and FzmM is denoted with a pink asterisk. Gene clusters were arbitrarily defined to include the 15 genes upstream or downstream of the N-oxidase.

Figure 6.

Proposed stepwise oxidation of 2AEPn to yield 1 or 2. This may explain the 1:2.5 substrate:NADPH ratio observed for oxime formation in vitro. We propose that each N-oxidation would require 1 molar equivalent of NADPH and molecular oxygen. This would require two N-oxidations and a dehydration to yield 2-nitrosoethylphosphonic acid, which could tautomerize to form 2-hydroxyiminoethylphosphonic acid. An enzyme-catalyzed overoxidation of either 2-nitrosoethylphosphonic acid or 2-hydroxyiminoethylphosphonic acid could potentially yield 2-nitroethylphosphonic acid (shown by gray arrows). Compounds shown in purple are all products observed in the reactions catalyzed by HpxL and PcxL using 2AEPn as the substrate (Fig. 3).