In vivo and in vitro Reconstitution of Biosynthesis of N‐Prenylated Phenazines Revealing Diverse Phenazine‐Modifying Enzymes

Abstract

Phenazine natural products play various roles such as signal molecules, antibiotics, or electron carriers in their producer strains. Among these products, phenazinomycin and lavanducyanin, which are produced by Streptomyces species, are characterized by an N‐alkyl modification. Herein, we established the biosynthetic pathways for these two phenazine natural products. Gene‐disruption experiments and in vitro reconstitution of the phenazine‐tailoring pathway revealed the late steps of the biosynthetic pathway of the phenazines. The class II terpene cyclase homolog Pzm1 catalyzes the cyclization reaction of farnesyl diphosphate to form monocyclic farnesyl diphosphate. Additionally, the prenyltransferase homolog PzmP functions as the N‐prenyltransferase of 5,10‐dihydrophenazine‐1‐carboxylic acid. The flavin monooxygenase homolog PzmS catalyzes the oxidative decarboxylation of prenylated 5,10‐dihydrophenazine‐1‐carboxylic acid to yield phenazinomycin. This study highlights unprecedented modification enzymes for phenazine natural products.

This study highlights unprecedented modification enzymes for natural phenazines. The class II terpene cyclase Pzm1 catalyzes the cyclization of farnesyl diphosphate (FPP) to form monocyclic FPP. The prenyltransferases PzmP and LavP function as the N‐prenyltransferases of the phenazine‐backbone, followed by the oxidative decarboxylation catalyzed by the flavin monooxygenases PzmS and LavS to yield phenazinomycin and lavanducyanin, respectively.

Introduction

Natural products exhibit diverse activities such as antibacterial, antiviral, and herbicidal activities, which are largely dependent on their diverse structures. Highly sophisticated enzymes often catalyze the modification reactions that lead to this structural diversity. These enzymes are encoded in the genome of producing strains, and such genes are clustered in bacterial genomes as biosynthetic gene clusters (BGCs).[ ¹ , ² ] Thus, with the increasing ease of genome sequencing, identifying BGCs is the initial step toward discovering new modification enzymes.

Phenazine natural products (PNPs) are a class of nitrogen‐containing heterocyclic compounds produced by pseudomonads, streptomycetes, and some other bacterial species. They exhibit many biological activities, such as antioxidant, antiviral, antitumor, and antimalarial activities. PNPs are derived from phenazine‐1‐carboxylic acid (PCA) or phenazine‐1,6‐dicarboxylic acid, which are converted into various phenazine derivatives by diverse modification reactions.[ ³ , ⁴ ] Phenazinomycin (1) and lavanducyanin (2), which are produced by streptomyces, are members of PNPs with an N‐prenyl group in their structure (Figure 1).[ ⁵ , ⁶ ] Prenylation is a ubiquitous modification reaction in the biosynthesis of natural products; however, N‐prenylation modifications are very rare.

Figure 1.

Structures of natural products containing the phenazine skeleton.

Several BGCs of PNPs and some enzymes involved in their biosynthesis have been identified. ^[7] Among them, the prenyltransferases EpzP and PpzP have been shown to catalyze the prenylation reaction toward the C9 carbon atom of 5,10‐dihydrophenazine‐1‐carboxylic acid (dhPCA) during endophenazine biosynthesis (Figure 1).[ ⁸ , ⁹ ] However, there have been no reports on the prenylation modification involving the nitrogen atom of the phenazine skeleton as observed in 1 and 2. These two PNPs have a unique cyclic prenyl group named a monocyclic farnesyl group and a cyclolavandulyl group, respectively. We have previously reported cyclolavandulyl diphosphate synthase (CLDS), which catalyzes the condensation reaction of two molecules of dimethylallyl diphosphate to form cyclolavandulyl diphosphate (CLDP) in the 2 producing Streptomyces sp. CL190. ^[10] However, the BGC for 2 has not been identified yet. Although a BGC for 1 (phm) has been reported, the detailed biosynthetic pathway, including the modification reaction of the phenazine backbone, has been elusive. ^[11] Herein, we elucidate the vital enzymatic reactions in the biosynthesis of 1 and 2, including terpene cyclization, N‐prenylation, and oxidative decarboxylation reactions, to highlight the strategy for the structural diversification of PNPs.

Results and Discussion

Heterologous Expression of the Phenazinomycin Biosynthetic Gene Cluster

Previously, the BGC for 1 (phm cluster) was identified in the genome of Streptomyces iakyrus DSM 41873. ^[11] To verify whether this gene cluster is also conserved in the genome of another phenazinomycin‐producing strain, Streptomyces sp. WK‐2057, we sequenced its genomic DNA. The sequence comparison revealed the region containing several homologs of the phm cluster in the genome of Streptomyces sp. WK‐2057 and designated it as the pzm cluster (Figure 2A and Table S1). To verify the involvement of this gene cluster in the biosynthesis of 1, we constructed a cosmid library for Streptomyces sp. WK‐2057 (Tables S2 and S3). Then, one cosmid (cos510), including pzm1–24, was selected from the resultant approximately 2,000 cosmid clones, and the insert DNA fragment was transferred to the heterologous expression vector pKU465cos. The resultant plasmid was then introduced into the heterologous expression host S. avermitilis SUKA17 to construct the SUKA17::cos510 transformant. Liquid chromatography–mass spectrometry (LC–MS) analysis of the extract of the transformant revealed that the transformant produced 1, while the transformant harboring the empty vector did not (Figure 2B). This indicates that SUKA17::cos510 contains all genes responsible for 1 biosynthesis. To gain further insight into the function of each gene, we analyzed each gene encoded in this cluster in detail. In silico analyses suggested that pzm15–pzm21 encode the mevalonate pathway enzymes and the pzm4‐encoded protein possesses the polyprenyltransferase and LytB motifs in its N and C termini, respectively. LytB catalyzes the last step of the methylerythritol phosphate pathway for terpenoid precursors. In addition, pzm3 encodes 1‐deoxy‐d‐xylulose 5‐phosphate synthase, an enzyme associated with the methylerythritol phosphate pathway. ^[12] Therefore, these enzymes are presumably involved in providing a precursor of farnesyl diphosphate (FPP). PzmF to PzmA and PzmB to PzmE show high similarity to those of pyocyanin BGC (phz cluster), indicating that they are responsible for the biosynthesis of the phenazine backbone (Figures 1, 2A, and S1).[ ¹³ , ¹⁴ ] PzmS shows high sequence similarity to PhzS, which is proposed to catalyze oxidative decarboxylation during the biosynthesis of pyocyanin. ^[15] Therefore, PzmS may play a similar role to that of PhzS in pyocyanin biosynthesis during the biosynthesis of 1. PzmR and PzmT exhibit sequence similarities to a transcriptional activator and a transporter, respectively. Thus, we suspect that PzmR and PzmT are responsible for regulating the expression of this BGC and the localization of the product or precursors, respectively. PzmP exhibits high sequence similarity to the phenazine C‐prenyltransferases PpzP and EpzP. As PpzP and EpzP catalyze C‐prenylation of dhPCA during the biosynthesis of endophenazine,[ ⁸ , ⁹ ] we hypothesized that PzmP catalyzes N‐prenylation of dhPCA during the biosynthesis of 1.

Figure 2.

(A) Organization of biosynthetic gene clusters for pyocyanin (phz), phenazinomycin (phm and pzm), and lavanducyanin (lav). (B) Extracted ion count chromatograms for 1 (m/z 401.259 [M+H]⁺) obtained using liquid chromatography–mass spectrometry analysis of the culture extracts from (i) Streptomyces sp. WK2057; (ii), S. avemitilus SUKA17::cos510; (iii), S. avemitilus SUKA17::pKU465cos.

Identification and Heterologous Expression of the Lavanducyanin Biosynthetic Genes

We aimed to identify the BGCs for 2 in its producer, Streptomyces sp. CL190. As mentioned above, 1 and 2 share the structure, except for their prenyl side chain moiety. Thus, we speculated that the gene clusters of 1 and 2 are highly similar, except for their terpene cyclization enzymes. In a previous study, we have reported the clds gene encoding the CLDS enzyme, which catalyzes the formation of CLDP. ^[10] Sequence analysis of the CL190 genome revealed that the clds gene is located at the locus far from the BGC containing the phenazine biosynthetic genes (lav cluster), which lacks the pzmS gene homolog.

Further genome analysis identified the homologs of pzmP and pzmS (designated lavP and lavS, respectively) in different loci from that of the lav cluster in the genome of Streptomyces sp. CL190 (Figure 2A and Table S4). To clarify the roles of lavP and lavS in the biosynthesis of lavanducyainin, we constructed their disrupted strains ΔlavP and ΔlavS, respectively (Figure S2). To construct these disruptants, we used Streptomyces sp. CL190dORF2‐8 as a parent strain because the disruptant produced more 2 than Streptomyces sp. CL190 (wild type; see the Supplementary Note for more information). ^[16] LC–MS analysis of the culture extracts of ΔlavP and ΔlavS revealed that both disruptants abolished 2 production (Figure 3A). Furthermore, as expected, PCA accumulated in the extract of ΔlavP (Figure 3A). An unknown metabolite (3) accumulated in the extract of ΔlavS (Figure 3A). High‐resolution mass spectrometry (HRMS) and NMR spectral analyses (Table S5 and Figures S3–S7) revealed that 3 is PCA with an N5‐cyclolavandylyl group and an N10‐methyl group. These results strongly suggested that LavP catalyzes the prenylation reaction at the N5 nitrogen atom of dhPCA, followed by a LavS‐catalyzed oxidative decarboxylation reaction to form 2 (Figure 3B). We suspect that unknown promiscuous methyltransferase catalyzes the addition of a methyl group to the N10 nitrogen atom to form 3 in the ΔlavS strain.

Figure 3.

(A) Extracted ion count (EIC) chromatograms for (left) 2 (m/z 333.196 [M+H]⁺); (center) PCA (m/z 225.066 [M+H]⁺); (right) compound 3 (m/z 377.222 [M+H]⁺), obtained using liquid chromatography–mass spectrometry (LC–MS) analysis of the culture extracts of (i) ΔlavS, (ii) ΔlavP, and parent strain Streptomyces sp. CL190dORF2‐8. The asterisk indicates the MS fragment ion derived from 3 during the analysis. (B) Proposed biosynthetic pathway for 2, PCA, and 3. (C) EIC chromatograms for 2 (m/z 333.196 [M+H]⁺) obtained using LC–MS analysis of the culture extracts from (i) Streptomyces sp. CL190dORF2‐8; (ii), S. lividans TK23::pKU490; (iii), S.lividans TK23::pKULAV.

Next, to reconstitute the in vivo production of 2, we introduced the lav cluster along with clds, lavP, and lavS to a 2 nonproducing strain. We cloned these genes into the heterologous expression vector pKU490 (Figure S8) and then introduced the resultant vector pKULAV to the heterologous expression host S. lividans TK23. LC–MS analysis of the extract from the resultant transformant revealed that it produces 2 albeit in small amounts, while the strain harboring the empty vector does not (Figure 3C). These results indicate that the three genes clds, lavP, and lavS, scattered loci different from that of the lav cluster on the genome of CL190, are responsible for the production of 2. Moreover, we successfully constructed the artificial BGC for 2 in a heterologous host.

Functional Analysis of Class II Terpene Cyclase Homolog Pzm1

The only difference in the structures of 1 and 2 is their prenyl side chains (Figure 1). 1 has a monocyclic farnesyl group, whereas 2 has a rare monoterpene skeleton, cyclolavandulyl group. We have identified the clds gene encoding the CLDS enzyme in the Streptomyces sp. CL190 genome. ^[10] Although the clds homolog was not found in the pzm cluster, we focused on Pzm1 containing a putative terpene cyclase/PT domain (IPR008930), which was found in class II terpene cyclases. ^[17] We hypothesized that Pzm1 catalyzes the cyclization of FPP to form (S)‐trans‐γ‐monocyclofarnesyl diphosphate, corresponding to the side chain of 1. To test this hypothesis, we prepared a recombinant enzyme of Pzm1 and subjected it to a reaction with FPP in vitro (Figure S9). After dephosphorylation of the reaction product with alkaline phosphatase to remove a pyrophosphate group, GC–MS analysis showed the enzyme‐dependent formation of a new compound (4) (Figure 4A). The reaction product 4 was isolated, and its structure was determined as monocyclofarnesol using NMR spectral analyses (Figures 4B and S10–S14 and Table S6). These results indicate that Pzm1 catalyzes the cyclization of FPP to form monocyclofarnesyl diphosphate presumably through a protonation‐type cyclization (class II mechanism). ^[18]

Figure 4.

(A) Total ion chromatograms obtained using gas chromatography–mass spectrometry analysis of the extract of the reaction mixture i) with Pzm1 and ii) without Pzm1. (B) Structure of 4 and the proposed reaction mechanism of Pzm1. (C) High‐performance liquid chromatography chromatograms (λ=250 nm) of the extract from the reaction mixture i) with PzmS and PzmP, ii) without reductant, iii) with PzmP alone, and iv) with PzmS alone. (D) Proposed structure of 5 and the proposed biosynthetic pathway for 5 and 6 catalyzed by PzmP and PzmS. See Figure S17 for a more detailed mechanism.

Pzm1 exhibits sequence similarity to a class II terpene cyclase with βγ didomain, which is generally involved in the biosynthesis of diterpenes and triterpenes. SsDMS and ScDMS are the only reported bacterial sesquiterpene cyclases in this class of terpene cyclases. ^[19] SsDMS and ScDMS catalyze the formation of bicyclic drimenyl diphosphate from FPP. In contrast, Pzm1 catalyzes the formation of the monocyclic sesquiterpene (S)‐trans‐γ‐monocyclofarnesyl diphosphate and shows 34 % and 46 % sequence similarity to SsDMS and ScDSM, respectively. Only Pzm1 and its homologs including Phm1 likely have a DXDC motif, which is an alternative to the canonical DXDD motif conserved in class II terpene cyclases (Figure S15). The DXDC motif may be characteristic of the monocyclization reaction. However, further biochemical analyses of Pzm1 are needed to clarify its characteristic function in the monocyclization reaction.

In vitro Functional Analyses of PzmP and PzmS

As mentioned above, LavP and PzmP showed high sequence similarity to the phenazine prenyltransferase EpzP and PpzP, which catalyze the regioselective C‐prenylation of dhPCA to yield 5,10‐dihydroendophenazine A.[ ⁸ , ⁹ ] Thus, we hypothesized that LavP and PzmP catalyze the regioselective N‐prenylation against dhPCA. To test this hypothesis, we constructed recombinant PzmP using Escherichia coli as a host and conducted in vitro reaction experiments (Figure S16). In these experiments, we used NADH as a reductant to convert PCA to dhPCA, following a previous report. ^[8] Additionally, we used FPP instead of monocyclofarnesyl diphosphate because it was difficult to prepare amount of monocyclofarnesyl diphosphate. When PzmP was incubated with FPP, PCA, and a reductant (NADH), we observed the production of 5 using HPLC analysis whereas nothing was observed without the reductant (Figure 4C). HRMS spectra supported that 5 is likely farnesyl‐dhPCA (Figure S16). This result indicated that PzmP catalyzes the N‐prenyltransfer reaction using dhPCA, not PCA, as a prenyl acceptor and also accepts FPP as a prenyl donor (Figures 4D and S17). To further validate the biosynthetic pathway for 1, we added recombinant PzmS enzyme into the previously mentioned reaction (Figure S16). Using HPLC and HRMS analyses of the extract of the reaction mixture, the formation of compound 6, whose proposed molecular formula was concomitant with those of 5‐N‐farnesyl‐1‐phenazinone, was detected (Figures 4C and S16). This result suggested that, as we hypothesized, PzmS catalyzed the oxidative decarboxylation reaction of 5 (Figures 4D and S17). Additionally, we conducted in vitro analyses of LavP and LavS using geranyl pyrophosphate as a prenyl donor, and the results were consistent with those of PzmP and PzmS, except for the length of the prenyl group (Figure S18).

Conclusions

We established the biosynthetic pathway for the rare N‐prenylated phenazines 1 and 2. We identified two noteworthy enzymes; one is Pzm1, a class II sesquiterpene synthase that catalyzes the formation of monocyclic sesquiterpenes. This is the third example of the class II sesquiterpene synthase discovered from bacteria, following SsDMS and ScDMS, and the first discovered monocyclic sesquiterpene synthase. The other is PzmP, which is a prenyltransferase that catalyzes N‐specific prenylation. Although PzmP shares high homology with the previously characterized phenazine C‐prenyltransferases PpzP and EpzP, PzmP catalyzes the prenylation of the N5 nitrogen of phenazine rather than the C10 carbon of phenazine. Further structural analyses and comparison of both enzymes in detail will enable us to elucidate the mechanism that determines the regioselectivity of prenylation, providing insights that will help expand the diversity of natural phenazines.

Supporting Information Summary

The data that support the findings of this study are available in the Supporting Information of this article. The authors have cited additional references within the Supporting Information.[ ²⁰ , ²¹ , ²² , ²³ , ²⁴ ]

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Kato T., Xia D., Ozaki T., Nakao T., Zhao P., Nishiyama M., Shiraishi T., Kuzuyama T., ChemBioChem 2025, 26, e202400723. 10.1002/cbic.202400723

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.