Radical-Mediated Nucleophilic Peptide Cross-Linking in Dynobactin Biosynthesis

Abstract

Dynobactins are recently discovered ribosomally synthesized and post-translationally modified peptide (RiPP) antibiotics that selectively kill Gram-negative pathogens by inhibiting the β-barrel assembly machinery (Bam) located on their outer membranes. Such activity of dynobactins derives from their unique cross-links between Trp1-Asn4 and His6-Tyr8. In particular, the His6-Tyr8 cross-link is formed between N_τ of His6 and C_β of Tyr8, an unprecedented type of cross-link in RiPP natural products. The mechanism of the C-N cross-link formation remains elusive. In this work, using in vitro characterizations, we demonstrate that both cross-links in dynobactins are biosynthesized by the radical S-adenosylmethionine (SAM) enzyme DynA. Subsequent mechanistic studies using deuterium-labeled DynB precursor peptides suggested that the C-N cross-linking proceeds through the Tyr8-H_β atom abstraction by 5'-deoxyadenosyl radical. The absence of solvent exchange of Tyr8-H_α suggested that the mechanism unlikely involves α,β-desaturation of Tyr8. Furthermore, DynA catalyzed covalent modification of Tyr8 of H6A-DynB with small-molecule nucleophiles, suggesting the presence of a highly electrophilic Tyr-derived intermediate. Based on all these observations, we propose that DynA catalyzes Tyr8-H_β atom abstraction to generate Tyr8-C_β radical followed by its oxidation to a p-quinone methide intermediate, to which His6-N_τ attacks to form the C-N cross-link. This quinone methide-dependent mechanism of RiPPs cross-linking is distinct from the previously reported RiPPs cross-linking mechanisms and represents a novel mechanism in RiPPs biosynthesis. We will also discuss the functional, mechanistic, and evolutional relationships of DynA with other peptide-modifying radical SAM enzymes.

Figure 1.

Biosynthesis of daro- and dynobactin A. (a) Biosynthesis of darobactin A. (b) Proposed biosynthesis of dynobactin A (1).

Figure 2.

Functional characterization of DynA. (a,b) Extracted ion chromatograms (EICs) at m/z 829.8210 ± 0.0083 (z = +8) (a) and mass spectra (b) of the product of DynA reactions under the complete conditions (i) and controls lacking one of the reaction components (ii−v). Panel (b) also shows the mass spectrum of purified and unmodified DynB (bottom). (c,d) EIC at m/z 653.292 ± 0.007 (z = +2) (c) and mass spectra (d) of the proteinase K digest of 2 and dynobactin A (1).

Figure 3.

H-abstraction of C−N cross-linking in dynobactin A biosynthesis. (a) Proposed Tyr8-H_β abstraction by 5′-dA^•. (b) Mass spectra of 5′-dA formed in DynA assays with [β-²H₂]Y8-DynB (i), 5′-dA formed in DynA assays with DynB with natural isotope abundance (ii), and 5′-dA authentic standard (iii). Shown are z = +1 ions. Calculated m/z for [5′-²H]5′-dA = 253.1154; observed m/z 253.1145 (−3.6 ppm). Calculated m/z for 5′-dA = 252.1097; observed m/z 252.1091 (−2.4 ppm). (c) Mass spectra of the modified DynB produced in the DynA assay with [β-²H₂]Y8-DynB (i) and [β-²H₂]Y8-DynB substrate (ii).

Figure 4.

Evidences against the the α,β-desaturation mechanism. (a) Deconvoluted mass spectra of DynB after the DynA assays in H₂O or ²H₂O. (b) The DynA-catalyzed C−N cross-linking of [α-²H]Y8-DynB. (c) Deconvoluted mass spectra of [α-²H]Y8-DynB (i) and [α-²H]Y8-DynB after incubation with DynA and SAM (ii). Mass spectra in (a,c) were extracted from the retention time of 15−19 min, which covers both unmodified and cross-linked DynB.

Figure 5.

Characterization of H6A-DynB adducts with imidazole and imidazole-like compounds. (a) Extracted ion chromatograms (EICs) of H6A-DynB reactions with imidazole, substituted imidazoles, and other heterocycles. Blue traces are EIC at m/z 822.0719 ± 0.0082 (z = +8) for detection of 3 and unmodified H6A-DynB. Red traces are EIC at m/z 830.4550 ± 0.0083 (z = +8, trace iii), m/z 832.0766 ± 0.0083 (z = +8, traces iv and v), m/z 832.4448 ± 0.0083 (z = +8, trace vi), m/z EIC at 830.4476 ± 0.0083 (z = +8, trace vii); m/z 830.2002 ± 0.0083 (z = +8, trace viii); m/z 830.7042 ± 0.0083 (z = +8, trace ix); m/z 836.5766 ± 0.0083 (z = +8, trace x); m/z 836.4522 ± 0.0083 (z = +8, trace xi) for detection of corresponding nucleophile adducts. (b) Deconvoluted mass spectra of H6A-DynB (retention time 19.9−21.5 min) and 3 (retention time 16.1−17.5 min). (c, d) Deconvoluted mass spectra of 4 (retention time 14.1−16 min) and 5 (retention time 17.9−20 min). (e) Time-course analysis of DynA assays with H6A-DynB in the presence of imidazole. Filled circles represent the relative abundance of unmodified H6A-DynB (red), 3 (orange), 4 (green), and 5 (blue). The solid lines represent the result of kinetic simulation using the catalytic model and rate constants in the scheme.

Figure 6.

Reactivity of H6A-DynB variant with nucleophiles. Deconvoluted mass spectra of H6A-DynB and its nucleophile adducts. The mass spectra are extracted from the retention time of 17.1−23.1 min, where all the modified and unmodified H6A-DynB peptides elute.

Figure 7.

Characterization of DynA reaction’s intermediate. (a) Time course analysis of DynA assay. Shown are extracted ion chromatograms at m/z 830.4550 ± 0.0083 (z = +8) of the DynA assay quenched at 15, 30, 60, 120, 180, or 240 min. (b) Deconvoluted mass spectra of unmodified DynB (top) and the reaction intermediate 9 (bottom). (c) MS quantitation of the time course analysis in panel (a) for DynB (circle and blue line), 9 (triangle and yellow line), and 2 (square and red line). Each data point represents an average of three repeats, and the error bars represent one standard deviation. The solid lines represent the result of kinetic simulation using the model and rate constants in panel (d). (d) Proposed order of the DynA-catalyzed DynB cross-linking reaction with the rate constants determined in panel (c). (e) DynA stepwise assay. DynA was first incubated with DynB, SAM, and reductant for 60 min and the reaction was quenched with FA. After removal of precipitated DynA and FA, the sample containing 9 was incubated with DynA (ii), or without DynA (i) for 3 h and analyzed by LCMS. Shown are EICs at m/z 830.4550 ± 0.0083 (z = +8). (f) Deconvoluted mass spectra of unmodified [α-²H]Y8-DynB (top) and the reaction intermediate ([α-²H]Y8–9) from the DynA reaction with [α-²H]Y8-DynB as substrate (bottom).

Figure 8.

Evolutionary relationships between DynA, DarE, and anSME. Phylogenetic tree showing the evolutionary relationship between DynA, DarE, and anSME generated by Clustal Omega. Reaction schemes indicate functions of corresponding enzymes. Outgroups include selected members of SPASM/Twitch rSAM enzyme subfamily.

Scheme 1.

Examples of Macrocyclization Mechanisms in RiPP Biosynthesis; (a) Nucleophilic Crosslinking in Lanthipeptides Biosynthesis; (b,c) Radical Crosslinking in Subtilosin Biosynthesis (b) and Streptide Biosynthesis (c)

Scheme 2.

Proposed Mechanisms of C−N Crosslinking in Dynobactin A Biosynthesis; Proposed C−N Crosslinking Mechanism through α,β-Desaturation (a), and p-Quinone Methide Intermediate (b)