Genome- and metabolome-guided discovery of marine BamA inhibitors revealed a dedicated darobactin halogenase

Abstract

Darobactins represent a class of ribosomally synthesized and post-translationally modified peptide (RiPP) antibiotics featuring a rare bicyclic structure. They target the Bam-complex of Gram-negative bacteria and exhibit in vivo activity against drug-resistant pathogens. First isolated from Photorhabdus species, the corresponding biosynthetic gene clusters (BGCs) are widespread among γ-proteobacteria, including the genera Vibrio, Yersinia, and Pseudoalteromonas (P.). While the organization of the BGC core is highly conserved, a small subset of Pseudoalteromonas carries an extended BGC with additional genes. Here, we report the identification of brominated and dehydrated darobactin derivatives from P. luteoviolacea strains. The marine derivatives are active against multidrug-resistant (MDR) Gram-negative bacteria and showed solubility and plasma protein binding ability different from darobactin A, rendering it more active than darobactin A. The halogenation reaction is catalyzed by DarH, a new class of flavin-dependent halogenases with a novel fold.

Keywords: antibiotic; bromination; darobactin; genome-mining; gram-negative; natural products; protein structure modeling.

Figure 1.. Conserved organization of the darobactin BGC in genomes deposited in NCBI

Blue: darA encoding the precursor peptide, yellow: radical SAM enzyme, green: halogenase darH, and gray: others (e.g., transport-related and peptidase). The BGCs found in Vibrio and Pseudoalteromonas strains harbor additional open reading frames. darH is present in the strains P. luteoviolaceae H33, H33S, JG1, and sp. NC201.

Figure 2.. UPLC-HRMS analysis of a P. luteoviolaceae extract

Extracted ion chromatograms of 1 (C₄₇H₅₅N₁₁O₁₂ [M+2H]²⁺ ±0.05, red) and 2 (C₄₇H₅₄N₁₁O₁₂Br [M+2H]²⁺ ±0.05, black) as well as the respective mass spectra of 1 (red peak 1), 3 (red peak 2), 2 (black peak 1), and 4 (black peak 2). All Δ m/z ≤ 1.15 ppm.

Figure 3.. Chemical structures and NMR correlations of darobactin derivatives

Top: Molecular structure of naturally occurring darobactin derivatives. Darobactin A as well as halogenated and dehydrogenated variants are displayed. Bottom: Key correlations for the structure elucidation of bromodarobactin A. The most important correlations with respect to the position of the bromine atom are highlighted in red.

Figure 4.. HPLC chromatograms of the DarH assays

DarH_Pse (100 μM) was incubated with darobactin A (100 μM) sodium bromide (25 μM), NAD(P)H (1.0 mM), FAD (10 μM), and flavin reductase (10 μM). Shown are HPLC UV chromatograms of the darobactin A standard, the complete condition (Full) and controls without flavin reductase, NAD(P)H/FAD, DarH, or darobactin A. * marks an unrelated peak from assay components.

Figure 5.. Schematic biosynthetic hypothesis for the formation of halogenated darobactin A

The linear precursor peptide DarA is modified by DarE, thereby installing the two intramolecular rings. Subsequently, the heptapeptide is cleaved off the leader and follower peptide by a not yet identified protease(s), yielding darobactin A (1). The latter is halogenated by DarH. The timing of the formal dehydrogenation remains elusive until now.

Figure 6.. Phylogenetic and in silico structural analysis of DarH. Sequence-based maximum likelihood (ML) phylogenetic tree

(A) showing the three clusters formed by representative homolog sequences deposited in the PDB. The pdb codes of each model are shown on the right. The central panel (B) shows the structural alignment of the DarH_Pse AlphaFold model (green) with the Trp halogenase MibH from Microbiospora sp. (pdb code 5uao, orange) and the right panel (C) the superimposition with the disulfide reductase YpdA from S. aureus (pdb code 7a7b, gray). DarH_Pse contains both an FAD and NAD(P) binding domains and a C-terminal domain without structural homologs in the PDB. (D) SSN analysis of PF13738 shows DarH_Pse (green) and direct neighbor nodes (pale green) placed in Cluster 3 (green). Nodes represented in orange correspond to YpdA homologs. A zoom in of Cluster 3 is shown to highlight the position of DarH.

Figure 7.. Identification of putative binding sites in DarH in silico model

(A) The YpdA structural model (C chain, pdb id 7a7b) was superimposed to the DarH_Pse AF one. Only the FAD and NAD of YpdA are shown as sticks with C, N, O, and P atoms colored in green, blue, red, and purple, respectively. The whole DarH_Pse AF structure is displayed as gray ribbons. The first putative binding site we identified (see supplemental information), binding site 1, where the red, blue, and yellow surfaces correspond to hydrogen-bond donors, hydrogen-bond acceptors, and hydrophobic regions, respectively, is displayed. A zoom of binding site 1 is shown in Figure S12. (B) Binding site 3 (Data S16, color code as in A) is displayed in the DarH_Pse AF model, (gray ribbons). Part of this cavity lies about 14 Å (shortest distance) from FAD.