Flavoenzyme-catalyzed atropo-selective N,C-bipyrrole homocoupling in marinopyrrole biosynthesis

Abstract

Axially chiral biaryl compounds are frequently encountered in nature where they exhibit diverse biological properties. Many are biphenols that have C–C or C–O linkages installed by cytochrome P450 oxygenases that control the regio- and stereoselectivity of the intermolecular coupling reaction. In contrast, bipyrrole-coupling enzymology has not been observed. Marinopyrroles, produced by a marine- derived streptomycete, are the first 1,3′-bipyrrole natural products. On the basis of marinopyrrole’s unusual bipyrrole structure, we explored its atropo-selective biosynthesis in Streptomyces sp. CNQ-418 in order to elucidate the N,C-bipyrrole homocoupling enzymology. Through a series of genetic experiments involving the discovery and heterologous expression of marinopyrrole biosynthesis genes, we report that two flavin-dependent halogenases catalyze the unprecedented homocoupling reaction.

Crosscoupling and homocoupling reactions are important organic transformations that conjoin two hydrocarbon fragments through the aid of a catalyst. While a myriad of synthetic approaches have been developed to construct axially chiral biaryl compounds inspired by naturally occurring organic molecules,¹ our mechanistic understanding of enzyme-catalyzed oxidative coupling reactions is limited largely to cytochrome P450 oxygenases, laccases and peroxidases that biosynthesize biphenols, alkaloids, and plant-derived aromatics.² Nature’s capacity to biosynthesize biaryl compounds is far greater and includes pyrrole-containing aromatics such as hexabromo-2,2′-bipyrrole,³ pentabromopseudiline⁴ and marinopyrrole⁵ for which pyrrole coupling enzymology has not been reported. Herein, we describe the discovery of a pair of FADH₂-dependent halogenases involved in marinopyrrole biosynthesis that catalyze an atropo-selective N,C-biaryl homocoupling reaction.

The densely halogenated bipyrrole natural products marinopyrroles A–F (1–6, Figure 1) were recently reported from the bacterium Streptomyces sp. CNQ-418 isolated from ocean sediments.^5,6 These marine microbial metabolites are the first natural 1,3′-bipyrroles and possess potent antibiotic activity against methicillin-resistant Staphylococcus aureus.^6,7 Because of their axially chiral structures, the biosynthesis of the marinopyrroles may proceed through an unprecedented bipyrrole homocoupling reaction operating in an atropo-selective manner. The fact that monomeric monodeoxypyoluteorin (7) is co-produced by strain CNQ-418⁵ strongly suggests its intermediacy in the biosynthesis of the dimeric marinopyrroles. Such a strategy was recently executed in the biomimetic total synthesis of racemic, marinopyrrole A via an intermolecular Ullman coupling reaction.⁸ We thus set out to explore marinopyrrole biosynthesis in S. sp. CNQ-418 in order to elucidate how nature has solved this atropo-selective, bipyrrole homocoupling reaction that could influence the future development of new biaryl coupling reagents.

Figure 1.

Structures of marinopyrroles A–F (1–6) and related compounds.

The chemical structure of the monomer 7 is closely related to that of two phenylpyrrole antibiotics, pyoluteorin (plt, 8) and pyrrolomycin D (pyr, 9), produced by Pseudomonas fluorescens Pf-5⁹ and Streptomyces vitaminophilum ATCC31673 (formerly Actinosporangium vitaminophilum),¹⁰ respectively. Based on the biosynthetic studies of 8 and 9, ^9,10 we were able to clone and sequence the 36-kb putative marinopyrrole biosynthetic gene (mpy) cluster spanning two cosmid clones (Figure 2a, Table S3). Sequence analysis revealed 20 open reading frames (ORFs) that showed clear relationships with most ORFs associated with 8 and 9 biosynthesis, thereby suggesting that 7 is biosynthesized from L-proline and three malonate molecules via a modular polyketide synthase (PKS) (Figure 2b). We, however, were not able to probe marinopyrrole biosynthesis in the native strain CNQ-418 since it did not survive long-term storage. Therefore, to connect the mpy locus with marinopyrrole biosynthesis, we assembled the entire mpy cluster from two overlapping cosmid clones (A3 and B6) into a single cosmid (mpyKY6) by λ-RED recombination (Figure S1)¹² in order to express the complete cluster in the host Streptomyces coelicolor M512. The heterologous expression of the reconstituted mpy cluster resulted in the production of dimeric marinopyrroles 1–4 (Figure 3a) with productivity of the major metabolite 1 at 10–20 mg/L, thus providing us clear evidence that all genes required for marinopyrrole biosynthesis were present on the reconstituted cluster. Furthermore, chirality analysis of heterologously biosynthesized 1 confirmed that it maintained its M-configuration as in the natural product. We thus employed this heterologous expression system in all later experiments to probe the in vivo function of the mpy genes.

Figure 2.

(a) Gene organization of marinopyrrole biosynthetic gene (mpy) cluster in Streptomyces sp. CNQ-418 and schematic representation of cosmids used in this study. Each arrow represents the direction of transcription of an ORF. See the SI for the deduced functions of the ORFs. (b) Proposed biosynthetic pathway of monomeric monodeoxypyoluteorin (7). Abbreviations: A, adenylation domain; ACP, acyl carrier protein; KS, ketosynthase; AT, acyltransferase; KR, ketoreductase; R, thioester reductase.

Figure 3.

Comparative HPLC analysis of heterologously produced marinopyrroles extracted from (a) S. coelicolor M512/mpyKY8 (reconstituted); (b) S. coelicolor M512/mpyKY33 (mpy1-16); (c) S. coelicolor M512/mpyKY15 (◿mpy1); (d) S. coelicolor M512/mpyKY28 (◿ mpy10); and (e) S. coelicolor M512/mpyKY29 (◿mpy11). HPLC was monitored at 320 nm.

We first deleted the mpy6 gene encoding the putative offloading PKS module-3. As expected, this mutant completely lost the ability to produce monomeric and dimeric pyrroles (Figure S2a), thereby providing an opportunity to explore the intermediacy of monomer 7 in marinopyrrole biosynthesis. Chemical complementation of the mutant with 7 restored dimeric marinopyrrole synthesis (1–4) (Figure S2b), which confirmed it as a precursor and allowed for the concerted search for the desired marinopyrrole coupling enzyme gene. We next constructed several minimized mpy gene clusters by eliminating peripheral genes from the reconstituted cluster in order to determine the exact boundaries of the marinopyrrole locus (Table S4). While we originally considered that the hypothetical protein encoded by orf4 participated in the coupling reaction, the heterologous expression of mpy1–16 resulted in the production of 1–4 (Figure 3b), demonstrating that all genes involved in mpy biosynthesis are harbored on the minimized cosmid mpyKY33 (Figure 2a). Among the sixteen genes, most have homologues in the plt cluster¹³ except for the mpy1 flavin reductase-encoding gene and the mpy4 putative membrane protein-encoding gene. While elimination of the mpy4 gene resulted in unaltered marinopyrrole production, deletion of the mpy1 gene dramatically reduced production of dimeric 1–4 with concomitant production of monomeric 7 (Figure 3c), thereby clearly suggesting that the desired coupling enzyme is a flavoprotein.

Four flavoprotein-encoding genes (mpy5, mpy10, mpy11, and mpy16) are encoded in the minimized mpy cluster, and all four annotate as FADH₂-dependent halogenases by BLAST analysis (Figure 2a, Table S3). Among them, Mpy5 lacks the motifs GXGXXG and WXWXI associated with other characterized FADH₂-dependent halogenases,¹³ suggesting that Mpy5 may not be a functional flavoprotein. We therefore generated three halogenase gene deletion mutants in order to individually interrogate their in vivo functions. As anticipated, the mpy16 gene deletion completely abolished production of the marinopyrroles, supporting our biosynthetic proposal that Mpy16 catalyzes the initiating dichlorination of pyrrolyl-S-carrier protein as previously demonstrated for the homologous PltA in pyoluteorin biosynthesis (Figure 1b).¹¹ Deletion of mpy10, on the other hand, completely abolished production of dimeric 1–4 with alternative overproduction of 7 (Figure 3d). We similarly observed the same outcome for the deletion of mpy11 (Figure 3e), suggesting that both Mpy10 and Mpy11 halogenases are responsible for the coupling reaction. To exclude the possibility of unintended polar effects on gene expression in these mutants, we reintroduced the deleted genes back to the coupling-deficient mutants through conjugative integration and under control of the strong constitutive ermE* promoter (Figure S3). As expected, dimeric 1–4 formation was restored only when the mutants had both the mpy10 and mpy11 genes (Figure 4), clearly demonstrating that two FADH₂-dependent halogenases Mpy10 and Mpy11 are involved in the atropo-selective N,C-bipyrrole homocoupling in the presence of the pathway specific flavin reductase Mpy1.

Figure 4.

Comparative HPLC analysis of heterologously produced marinopyrroles extracted from (a) S. coelicolor M512/mpyKY28 (◿mpy10)/pKY01-mpy10; (b) S. coelicolor M512/mpyKY28 (◿mpy10)/pKY01-mpy11; (c) S. coelicolor M512/mpyKY29 (◿mpy11)/pKY01-mpy10; and (d) S. coelicolor M512/mpyKY29 (◿mpy11)/pKY01-mpy11. HPLC was monitored at 320 nm.

We further corroborated this observation by constructing a new expression system harboring only the three genes mpy1, mpy10 and mpy11 (Figure S4). To our surprise, however, we did not observe dimerization of exogenously added 7 by the engineered S. coelicolor mutant. Dimerization was only achieved upon coexpression of the adjacently transcribed ABC transporter gene cassette mpy12–14 that presumably facilitates the transmembrane transport of the marinopyrroles in a similar manner to the plt ABC transporters.¹⁴ In fact, deletion of the transporter gene cassette from the minimized cluster resulted in no secretion of marinopyrroles, thereby supporting this observation (data not shown). Once again, both mpy10 and mpy11 were required for monomer homocoupling as elimination of either gene prevented dimerization.

The distinctive gene organization of two tandem FADH₂-dependent halogenase genes showing significant similarity to mpy10 and mpy11 was previously observed in the gene cluster of the structurally related antibiotic pyrrolomycin D (9) where their protein products are presumed to chlorinate the phenol and pyrrole residues (Table S5).¹⁰ In the case of marinopyrrole biosynthesis, we speculate that C-3 halogenation of 7 directs its enzymatic coupling with second molecule of 7 (Figure 5a). This mechanistic scenario parallels the approach employed in the Kanakis and Sarli chemical synthesis of racemic 1.⁸ Alternatively, halogenation of the pyrrole nitrogen of 7 could in turn facilitate an electrophilic aromatic substitution reaction with a second equivalent of monomer 7 (Figure 5b). Furthermore, a single electron transfer mechanism to give the N-radical cation is also plausible. Since 3-halo-7 and N-halo-7 were not reported from the native strain S. sp. CNQ-418⁵ nor did we observe their production in any of our mpy heterologous S. coelicolor M512 constructs, consequently, we hypothesize that a tri-halogenated derivative of 7 may be formed in trans and not released by a functional complex of Mpy10-Mpy11 prior to atropo-selective dimerization with unmodified 7 (Figure 5). Additional halogenation of 1 to the chlorinated and brominated 2–5 may take place immediately following the homocoupling reaction. An alternative possibility of unmodified 7 serving as the directing coupling agent in which H-3 is lost as hydride upon homocoupling is compatible for a flavoenzyme acting as a dehydrogenase, yet is not consistent with the requirement of the flavin reductase Mpy1.

Figure 5.

Proposed N,C-bipyrrole homocoupling mechanisms via (a) C-3 halogenation and (b) N-halogenation of 7.

In conclusion, we have successfully characterized the molecular basis of marinopyrrole biosynthesis that includes an unprecedented N,C-bipyrrole homocoupling reaction. Our work supports the hypothesis that the novel chiral coupling reaction of two 7 molecules is catalyzed by the two FADH₂-dependent halogenases Mpy10 and Mpy11 in the presence of the pathway specific flavin reductase Mpy1 in which the 46% identical (78% similar) halogenases may operate as a functional complex. Such a cryptic biohalogenation strategy has only been previously observed in the biosynthesis of cyclopropyl rings.¹⁵ Since the involvement of FADH₂-dependent halogenases in biaryl formation has never been observed, our discovery lays the foundation for future in vitro studies with recombinant proteins to explore the mechanistic details of this novel N,C-biaryl homocoupling enzymology.