Conserved Mechanism of 2′-Phosphorylation-Aided Amide Ligation in Peptidyl Nucleoside Biosynthesis

Abstract

Peptidyl nucleoside antifungals, represented by nikkomycins and polyoxins, consist of an unusual six-carbon nucleoside [aminohexuronic acid (AHA)] ligated to a nonproteinogenic amino acid via an amide bond. A recent study suggested that AHA is biosynthesized through cryptic phosphorylation, where a 2′-phosphate is introduced early in the pathway and required to form AHA. However, whether 2′-phosphorylation is necessary for the last step of biosynthesis, the formation of the amide bond between AHA and nonproteinogenic amino acids, remains ambiguous. Here, we address this question with comprehensive in vitro and in vivo characterizations of PolG and NikS, which together provide strong evidence that amide ligation proceeds with 2′-phosphorylated substrates in both pathways. Our results suggest that 2′-phosphorylation is retained for the entirety of both nikkomycin and polyoxin biosynthesis, providing important insights into how cryptic phosphorylation assists with nucleoside natural product biosynthesis.

Peptidyl nucleosides (PNs), such as nikkomycins and polyoxins, exhibit selective antifungal activities with no known side effects to higher eukaryotes. Consequently, polyoxin D [1 (Figure 1A)] is a safely used fungicide in agriculture,¹ and nikkomycin Z has been clinically investigated for the treatment of coccidioidomycosis (valley fever).^2,3 However, their limited activities against more clinically prevalent pathogens, such as Candida albicans, have hindered their development as antifungal drugs. PNs target chitin synthases, and resistant pathogens carry chitin synthases that are not inhibited by existing PNs.⁴ Because PNs structurally mimic UDP-GlcNAc, a substrate common to all chitin synthases, the difference in drug sensitivities likely derives from structural differences within each isozyme’s active site.⁵ Thus, the development of structurally novel PNs is an important step toward discovering PN derivatives active against PN-resistant chitin synthases.

Figure 1.

Amide ligation in PN biosynthesis. (A and B) Reported functions of amide ligases PolG and NikS in the polyoxin and nikkomycin biosynthetic pathways, respectively. (C) Possible orders of amide ligation and dephosphorylation in nikkomycin and polyoxin pathways.

Polyoxins and nikkomycins are structurally characterized by unusual nucleosides with a six-carbon sugar, such as 5-carboxy aminohexuronic acid [5cAHA (2)] or aminohexuronic acid [AHA (3)], ligated to a nonproteogenic amino acid, such as carbamoyl polyoxamic acid [CPOAA (4)] for polyoxins or 4′-hydroxy-2′-pyridinyl homothreonine [HPHT (5)] for nikkomycins. While the mechanism of AHA biosynthesis had remained unclear for more than two decades, a recent study revealed that it is biosynthesized through a cryptic phosphorylation mechanism.⁶ The proposed mechanism involves phosphorylation of 2′-OH of a precursor nucleotide early in the pathway, which is essential for the activity of all subsequent AHA biosynthetic enzymes. Analogous cryptic phosphorylation has also been found in other nucleoside^7–9 and aminoglycoside¹⁰ natural products, suggesting the generality of this biosynthetic mechanism.

Although the number of pathways with the cryptic phosphorylation mechanism is increasing, the duration of phosphorylation in most pathways remains elusive. In the PN biosynthesis, dephosphorylation timing remains ambiguous due to conflicting observations about the pathway’s last step, formation of the amide bond between AHA and the N-terminal amino acid. While the early characterization of PolG demonstrated its in vitro activity to ligate nonphosphorylated AHA and CPOAA in an ATP-dependent manner (Figure 1A),¹¹ recent characterization of NikS demonstrated its activity with AHA 2′-phosphate [AHAP (6) (Figure 1B)].⁶ These PolG and NikS characterizations were qualitative without kinetic studies, and the activity of PolG has never been studied with 2′-phosphorylated nucleosides. Consequently, significant ambiguity remained whether AHAP/5cAHAP is the physiological substrate of the amide ligase (path A in Figure 1C) or dephosphorylation precedes amide ligation (path B in Figure 1C).

To distinguish these possibilities, we investigated the substrate specificity of PolG in vitro. Initially, we qualitatively assessed PolG’s specificity among 5cAHA, AHA, and AHAP. PolG was expressed and purified as an N-terminally His_6-tagged protein as described previously.¹¹ All PolG assays were performed using CPOAA as the carboxylate substrate. CPOAA was prepared by acid hydrolysis of polyoxin D followed by HPLC purification. The concentration of CPOAA was determined on the basis of its conversion to polyoxin D by PolG in the presence of a large excess of 5cAHA. The error in the CPOAA concentrations determined by this approach was ~6%. 5cAHA and AHA were prepared by hydrolysis of polyoxin D and nikkomycin Z, respectively, and AHAP was prepared enzymatically as previously reported.⁶

When PolG was assayed with either AHAP or 5cAHA, PolG ligated both 5cAHA and AHAP with CPOAA and produced polyoxin D and polyoxin L 2′-phosphate, respectively, in an ATP-dependent fashion (Figure 2A). The observed activity of PolG with 5cAHA is consistent with the previous report.¹¹ However, when the PolG assay was performed with stoichiometric amounts (100 μM) of AHAP, 5cAHA, and CPOAA, AHAP was completely consumed and converted into polyoxin L 2′-phosphate, while 5cAHA remained mostly unreacted with little to no formation of polyoxin D (Figure 2A, trace iii). The identities of polyoxin L 2′-phosphate and polyoxin D were unambiguously confirmed by liquid chromatography-high-resolution mass spectrometry (LC-HRMS) analysis (Figures S1 and S2). In similar competition assays between AHA and 5cAHA in the presence of CPOAA, PolG accepted both nucleosides with seemingly comparable efficiencies with a slight preference for 5cAHA (Figure 2B). These results strongly suggested that 2′-phosphorylated nucleosides are the preferred substrates of PolG.

Figure 2.

Comparison of PolG’s specificity toward (A) 5cAHA vs AHAP and (B) 5cAHA vs AHA. PolG (25 μM) was incubated at 21 °C for 15 min with 5cAHA, AHAP, or AHA (100 μM) as an amino substrate in the presence of CPOAA (100 μM for panel A and 150 μM for panel B), ATP (1 mM), and MgCl₂ (3 mM). Shown are UV absorption chromatograms at 260 nm of assays under the conditions indicated above each trace.

To provide quantitative insights into the PolG’s substrate specificity, we kinetically characterized PolG under pseudo-first-order and steady-state conditions, with saturating concentrations of ATP (1 mM) and MgCl₂ (3 mM). Because CPOAA displayed substrate inhibition with PolG at concentrations of >200 μM (Figure S3), all of the subsequent kinetic analyses were performed with CPOAA at a concentration with minimal inhibition (200 μM). For the nucleoside and nucleotide substrates, we focused on 5cAHA and AHA and AHAP to compare the effects of 2′-phosphorylation and 5-carboxylation.

Consistent with our qualitative characterization, PolG displayed robust activity with AHAP and CPOAA as substrates: K_m = 9.1 ± 0.4 μM, k_cat = 3.5 ± 0.2 min⁻¹, and k_cat/K_m = 0.38 ± 0.03 μM⁻¹ min⁻¹ (Figure 3A). These kinetic parameters are comparable to those reported for other steps in the PN pathways: PolH radical SAM cyclase¹² (K_m = 17 ± 3 μM, and k_cat = 0.35 ± 0.03 min⁻¹), PolQ2 kinase⁶ (turnover rate of 0.1–0.21 min⁻¹), and NikO¹³ (k_cat = 6.8 min⁻¹). In contrast, PolG exhibited poor activity with nonphosphorylated AHA and CPOAA as substrates: K_m = 931 ± 122 μM, k_cat = 0.67 ± 0.04 min⁻¹, and k_cat/K_m = 0.00072 ± 0.00010 μM⁻¹ min⁻¹ (Figure 3B; see Figure S4 for a typical HPAEC chromatogram used for this analysis). On the basis of the k_cat/K_m values, PolG preferentially accepts AHAP over AHA with a >500-fold greater specificity, suggesting the significant role of 2′-phosphate in PolG’s substrate recognition. Assays with 5cAHA revealed kinetic parameters comparable to those for AHA: K_m = 334 ± 68 μM, k_cat = 1.4 ± 0.4 min⁻¹, and k_cat/K_m = 0.0042 ± 0.0010 μM⁻¹ min⁻¹ (Figure 3C). These observations suggest that the 5-carboxylic acid functional group has only weak influences (5–6-fold difference in k_cat/K_m) on the substrate specificity of PolG.

Figure 3.

Steady-state kinetic analyses of reactions of PolG with (A) AHAP, (B) AHA, and (C) 5cAHA and reactions of NikS with (D) AHAP, (E) AHA, and (F) 5cAHA. CPOAA was used in all PolG assays, and HPHT was used in all NikS assays. No detectable amount of amide ligation product (<1 μM) was observed in the NikS assays with AHA⁶ or 5cAHA [100 μM (see Figure S5)]. Each point represents an average of two or three replicates, and error bars represent standard errors. Solid lines in panels A–C represent the results of the nonlinear curve fitting to the Michaelis–Menten equation with the parameters shown in each panel. The solid line in panel D is a fit to the substrate inhibition kinetic model.¹⁴ Less than 10% of CPOAA or HPHT was consumed in each assay to maintain the pseudo-first-order and steady-state kinetic conditions.

We also characterized NikS-catalyzed amide ligation. Because NikS did not accept AHA⁶ (Figure 3E) or 5cAHA (Figure 3F and Figure S5), we focused on AHAP for kinetic characterizations. Steady-state kinetic analysis with a substrate inhibition kinetic model¹⁴ yielded the following values: K_m = 10.9 ± 2.6 μM, k_cat = 1.7 ± 0.3 min⁻¹, and k_cat/K_m = 0.15 ± 0.05 μM⁻¹ min⁻¹ (Figure 3D). These parameters are very similar to those for PolG with AHAP. Because other homologous enzymes in the nikkomycin and polyoxin pathways also exhibit similar kinetic parameters,¹² the kinetic resemblance of NikS and PolG with AHAP provides strong evidence that the 2′-phosphorylated nucleotides are likely the physiological substrates of PolG and NikS.

A noticeable difference between NikS and PolG was substrate inhibition. NikS was inhibited by high concentrations of AHAP [K_i = 111 ± 38 μM (Figure 3D)], while PolG was not inhibited by any of the nucleoside/nucleotide substrates. PolG, on the contrary, was inhibited by CPOAA at concentrations of >200 μM (Figure S3). This difference may indicate the difference in the substrate specificity of NikS and PolG and/or different mechanisms of metabolic flux regulation between the two pathways.

To test the physiological relevance of these in vitro characterizations, we prepared an in-frame polG gene deletion mutant strain of Streptomyces cacaoi (ΔpolG) and characterized the metabolites in wild-type (wt) and ΔpolG strains. In the wt strain, we observed accumulation of AHA at a cellular concentration of 1.2 ± 0.6 μM (Figure 4A) while 5cAHA 2′-phosphate, AHAP, and 5cAHA were not detectable within our detection limit [2 μM (Figure S6)]. Due to the high K_m value of PolG toward AHA (931 ± 122 μM), the low micromolar concentration of AHA should be consumed very slowly and thus AHA accumulates in cells. Additionally, the absence of phosphorylated intermediates likely suggests efficient consumption by PolG.

Figure 4.

LCMS analyses of (A and B) AHA, (C) polyoxin D, and (D) polyoxin L in S. cacaoi wt (trace i), ΔpolG (trace ii), ΔpolG complemented with the wt polG gene (ΔpolG + polG) (trace iii), and the AHA (10 μM) or polyoxin D (10 μM) standard (trace iv). Shown are extracted ion chromatograms for AHA (m/z 288.0826), polyoxin D (m/z 522.1314), and polyoxin L (m/z 478.1416) of mycelial extraction or culture supernatant of each strain. The cellular concentration of AHA was estimated on the basis of the concentration in mycelial extraction and estimated cell volume.¹⁵ See Figures S6 and S7 for the other polyoxins and biosynthetic intermediates.

In contrast, we did not observe polyoxins, biosynthetic intermediates, or shunt metabolites in either the mycelium or supernatant of the ΔpolG strain (Figure 4A,B and Figures S6 and S7). A ΔpolG strain complemented with the wt polG gene (ΔpolG + polG) successfully restored polyoxin production at concentrations comparable to those in the wt strain (Figure 4C,D and Figure S7), suggesting that the expression of the other pol genes was not affected by the deletion of the polG gene. The absence of intermediate accumulation in the ΔpolG strain recapitulates our previous characterization of an S. cacaoi ΔpolQ2 strain.⁶ However, gene disruption mutants of the nikkomycin producer, Streptomyces tendae, accumulated detectable amounts of 2′-phosphorylated intermediates.⁶ Consequently, the polyoxin pathway may have a mechanism for tightly regulating the metabolic flux and minimizing the accumulation of biosynthetic intermediates. Together, these in vivo observations suggest that the steady-state concentrations of polyoxin biosynthetic intermediates are likely less than 2 μM. When combined with our in vitro kinetic characterizations, the low concentration of biosynthetic intermediates suggests that PolG’s physiological substrates likely possess a 2′-phosphate (e.g., AHAP and 5cAHAP).

The results described above suggest that the amide ligation step in the nikkomycin and polyoxin pathways requires 2′-phosphorylation. While the identity of the enzyme responsible for the dephosphorylation of nikkomycin or polyoxin 2′-phosphate is currently unknown, a recent study of apramycin biosynthesis¹⁰ identified a promiscuous phosphatase (AprZ) as being responsible for dephosphorylating apramycin phosphate after the metabolite’s translocation to the periplasm. Although no such phosphatase is encoded in the nikkomycin or polyoxin pathway, it is possible that a similar phosphatase outside the biosynthetic gene cluster promiscuously dephosphorylates 2′-phosphorylated PNs once exported to the periplasm.

In conclusion, this study demonstrates that cryptic phosphorylation is retained until the last step of the nikkomycin and polyoxin biosynthetic pathways. Elucidation of the mechanism of PN dephosphorylation is ongoing.