OX4 is an NADPH-dependent Dehydrogenase Catalyzing an Extended Michael Addition Reaction to Form the Six-membered Ring in the Antifungal HSAF

Abstract

The polycyclic tetramate macrolactam HSAF is an antifungal natural product isolated from Lysobacter enzymogenes. HSAF and analogs have a distinct chemical structure and new mode of antifungal action. The mechanism by which the 5/5/6 tricycle of HSAF is formed from the polyene precursor is not totally clear. Here, we used purified OX4, a homologous enzyme of alcohol dehydrogenase/Zn-binding proteins, to show the enzymatic mechanism for the six-membered ring formation. The results from the deuterium isotope incorporation demonstrated that OX4 selectively transfers the pro-R hydride of NADPH to C21 and one proton from water to C10 of 3-deOH alteramide C (1), resulting in 3-deOH HSAF (2) through a reductive cyclization of the polyene precursor by a mechanism consistent with an extended 1,6-Michael addition reaction. The regioselective incorporation of the NADPH hydride into C21 of 1 is also stereoselective, leading to the 21S configuration of 2. This work represents the first characterization of the activity and selectivity of the enzyme for the six-membered ring formation in a group of distinct antifungal polycyclic tetramate macrolactams.

Graphical Abstract

The antifungal natural product HSAF (Heat-Stable Antifungal Factor) is a polycyclic tetramate macrolactam (PoTeM) that has been isolated from diverse sources.¹ The structure and mode of action of HSAF and analogs are distinct from the existing antifungal drugs and fungicides.^2–4 PoTeMs share the same basic scaffold but differ by the cyclic systems, which can vary from 5/5, 5/5/6, 5/4/6, 5/6/5, to 5/5/5/8, carrying different stereochemistry.^5–8 HSAF’s precursors, 3-deOH alteramide C (1) and 3-deOH HSAF (2), contain a 5/5 and 5/5/6 cyclic system respectively (Figure 1). Our previous studies showed that four redox enzymes (OX1–4) are responsible for the cyclic system of HSAF and analogues.^5–7 However, mechanistic details for the tricycle formation has not been fully elucidated. In this work, we performed deuterium isotope incorporation and NMR analyses of the labeled products resulted from the OX4-catalyzed reaction, which forms the six-membered ring in 2 from 1 (Figure 1).

Figure 1.

LC-HRMS analysis of the reaction products of OX4 with 3-deOH Alteramide C (1) as substrate. (I) standard 3-deOH Alteramide C (1); (II) OX4 in H₂O containing D-[1-²H]glucose, NADP⁺, BmGDH, and 1; (III) OX4 in H₂O containing D-[1-²H]glucose, NADP⁺, TaGDH, and 1; (IV) OX4 in heavy water (D₂O) containing NADPH, and 1; (V) OX4 in heavy water (D₂O) containing D-[1-²H]glucose, NADP⁺, TaGDH, and 1. Chemical structure and calculated mass of 1 (the observed MS of 1 taken from IV) and 3-deOH HSAF products (2, 2a, 2b, 2c) are included. A similar set of reactions were also carried out using 3-deOH Alteramide B (3) as substrate, which produced 3-deOH Alteramide D products (4, 4a, 4b, 4c, also see Figure S3).

We previously showed that 1 is the precursor of 2, and OX4, a homolog of alcohol dehydrogenase/Zn-binding protein, is responsible for this conversion.⁷ The conversion of 1 into 2 is a two-electron reduction whereby the two double bonds at C21-C22 and C10-C11 are replaced by a new single bond between C11 and C22, and isomerization of the trans double bond at C8-C9 to cis. The overall reaction is a reductive cyclization with hydrogen incorporation at C10 and C21. In theory, the hydride could attack either C21 (then C10 would be protonated) or C10 (then C21 would be protonated). To find out, we expressed the OX4 gene in E. coli and purified OX4 (Figure S1). The enzyme was incubated with 1, in the presence of NADPH, (R)-[4-²H]NADPH, or (S)-[4-²H]NADPH, and the isotope incorporation in products were analyzed by LC-HRMS (Figure 1). (R)-[4-²H]NADPH was generated from D-[1-²H]glucose by the glucose dehydrogenases (GDH) from Thermoplasma acidophilum ATCC 25905, and (S)-[4-²H]NADPH was generated from D-[1-²H]glucose by the GDH from Bacillus megaterium DSM 2894 (Figure S1).^9–11 When OX4 was incubated with 1 in the presence of (S)-[4-²H]NADPH in H₂O, the LC-HRMS gave m/z 497.2966 for [M + H]⁺ of 2, showing that no deuterium was incorporated in the product and OX4 does not use the pro-S hydride of NADPH for the reduction reaction (Figure 1). When OX4 was incubated with 1 in the presence of (R)-[4-²H]NADPH in H₂O, the LC-HRMS gave m/z 498.3038 for [M + H]⁺ for 2a. The increase of one mass unit showed that OX4 specifically uses the pro-R hydride of NADPH for the reduction reaction. When OX4 was incubated with 1 in the presence of NADPH in heavy water (D₂O), the LC-HRMS gave m/z 498.2988 for [M + H]⁺ of 2b. The one mass unit increase in the product supported that water is the proton source for the reduction reaction. Finally, when OX4 was incubated with 1 in the presence of (R)-[4-²H]NADPH in heavy water (D₂O), the LC-HRMS gave m/z 499.3170 for [M + H]⁺ of 2c, demonstrating that the reductive cyclization involves the incorporation of the pro-R hydride of NADPH and one proton from the medium.

To further test the selectivity of OX4, we carried out the similar set of reactions using 3-deOH Alteramide B (3) as substrate (Figure 1 and Figure S3). Our previous result showed that OX4 could reduce the diene at C21-C22 and C23-C24 of 3, to produce Alteramide D (4) which contains a C22-C23 double bond.⁷ However, it was not clear which hydride of NADPH was used in the reduction. As shown in Figure S3, the result clearly demonstrated that OX4 incorporates the pro-R hydride of NADPH and one proton from water into 4 in this non-cyclized, reduction reaction.

Next, we investigated the regioselectivity and stereoselectivity of OX4 toward the hydride and the proton, through NMR analyses of the deuterium-labeled products. To obtain a sufficient amount of the substrate 1, we set out a large scale of solid culture (8 liter LB) of the mutant strain C3△OX14⁷ and obtained 28 mg 1 after a series of purification steps (Supporting Information). We then performed a 40 mL enzymatic preparation of 2a, using OX4, 1 and (R)-[4-²H]NADPH generated in situ from D-[1-²H]glucose by TaGDH. After purification, we obtained approximately 5 mg of the deuterated product 2a from the enzymatic reaction. 2a was analyzed by ¹H-NMR, ¹H-¹H COSY and NOESY (Table S1, Figure S4–S7). Comparison of ¹H-¹H COSY spectra of 2 and 2a clearly showed that the correlations of H-21b/H-20 and H-21a/H-21b in 2 was not observed in 2a, while the correlations of H-21a/H-20 and H-10a/H-10b were observed in both 2 and 2a (Figure 2). These observations confirmed the deuterium incorporation at C21 in 2a. Thus, OX4 regioselectively transfers the pro-R hydride of NADPH to C21 and the proton of water to C10 during the six-membered ring formation.

Figure 2.

Comparison of the ¹H-¹H COSY spectra of 2 (A) and 2a (B) in DMSO-d₆. The blue arrows in (B) are pointing to the lack of correlation for H-21a/H-21b and H-21b/H-20 in 2a. The results show that NADPH hydride is incorporated into C21 at the pro-S position (H-21b) (Figure S6–7).

A further comparison of the ¹H-NMR spectra showed that the H-8/H-9 coupling constant of 2a (J ~ 11.5) and 1 (J ~ 15.0) was consistent with the 8E configuration in 1 and 8Z configuration in 2a (Figure S4). The result confirmed the trans to cis isomerization of the C8-C9 double bond during the conversion from 1 to 2a. It also supports the reductive cyclization by OX4 goes via an extended Michael addition reaction, in which C8-C9 can transiently exist as a single bond in the intermediate that undergoes a tautomerization to afford 2 (or 2a, 2b, 2c) (Figure 3). Finally, a careful analysis of the NOESY data revealed a correlation between H-21a and H-19, which makes 21S configuration for C21 of 2a (Figure S7)

Figure 3.

A proposed mechanism for OX4-catalyzed reductive cyclization to form the six-membered ring in 3-deOH HSAF.

Together, our data revealed the regioselectivity and stereoselectivity of the OX4-catalyzed reductive cyclization that forms the six-membered ring of HSAF (Figure 3). The results support a mechanism consistent with a 1,6-Michael addition reaction that not only forms a new ring, but also leads to the unusual isomerization at the C8-C9 double bond, which is a challenging transformation considering the relatively rigid, “closed” macrolactam system of the PoTeM. It should be pointed out that D₂O could exchange with an acidic group in the enzyme, which then protonates C10, although we proposed D₂O directly protonate C10.

Zhang et al. first demonstrated the enzymatic mechanism for the IkaC-catalyzed inner five-membered ring in ikarugamycin.⁹ IkaC and OX4 share a 62.3% sequence identity (Figure S2). Their data showed that IkaC selectively transferred the deuterated pro-R hydride from (R)-[4-²H]NADPH to C15, resulting in a 15R configuration in ikarugamycin. This group also investigated the PtmC-catalyzed six-membered ring formation in pactamides.¹² LC-MS data showed that the enzyme was selective for the pro-R hydride of NADPH. However, the regioselectivity and stereoselectivity for the hydride transfer in the PtmC-catalyzed six-membered ring formation were not determined. Here, our data show that OX4 is also selective for the pro-R hydride of NADPH. We further demonstrate that OX4 selectively transfers the pro-R hydride to C21, resulting in a 21S configuration in HSAF during the six-membered ring formation. This work represents the first characterization of both the regioselectivity and stereoselectivity of the enzyme for the six-membered ring formation in a group of distinct antifungal PoTeMs.

ABBREVIATIONS

PoTeM

polycyclic tetramate macrolactam

GDH

glucose dehydrogenases