English

• The diversity of chemical structures of active natural products, which are the focus of innovative drug research [1], have spurred interest in nature's chemical strategy to produce complexity [2-4]. The biosynthetic pathways of natural products employ powerful chemical concepts to more concise and efficient [2], including convergent synthesis [5], oxidative cyclization [6], and cascade reactions [7]. Thus, uncovering the enzymatic basis of biosynthesis could improve complexity generation strategies and increases the toolbox of biocatalysts used to generate structural diversity.

  Pyrrolobenzoxazines are a hybrid terpene-amino acid natural product family [8-13], in which a rare pyrrolobenzoxazine core is esterified to a heavily oxidized juniper-type sesquiterpene. This characteristic of fusion of different fragment families is exemplary of the complexity generation strategy used by nature. The pyrrolobenzoxazines are however rare, with eight examples (Fig. 1) [8-13], and the majority of these products have been shown to have potent biological activities [8-10,13-17]. For example, paeciloxazine (1), which was first isolated from the fungus Paecilomyces BAUA3058, was reported to show significant insecticidal activity [8]. CJ-12662 and CJ-12663 which was isolated from Aspergillus fischeri var. thermomutatus ATCC 18618, shows remarkable antiparasitic activity [10].

  Figure 1

  

  Figure 1.  The family of pyrrolobenzoxazine natural products.

  DownLoad: Full-Size Img PowerPoint

  The unique scaffold of pyrrolobenzoxazines has led to several studies of total synthesis [11,18-20]. While the biosynthesis of pyrrolobenzoxazines is rarely reported, in our previous study, we identified the biosynthetic gene cluster of 1 from Penicillium janthinellum, and elucidated the mechanism that flavin dependent monooxygenase PaxA catalyzed the construction of 1, 2-oxazine in 1 via Meisenheimer rearrangement [12]. Here, we continue to explore the full biosynthetic pathway of 1 and peniciloxazine A (2).

  The pax cluster, which is responsible for the biosynthetic of 1, contains 9 genes (paxA–I) coding for several functional proteins, including two P450 monooxygenases (PaxD and PaxH), two flavin dependent monooxygenases (PaxA and PaxF), a nonheme, iron and α-ketoglutarate dependent dioxygenase (PaxC), a methyltransferase (PaxB), an acyltransferase (PaxE), a mono-modular NRPS (PaxG), and a terpene cyclase (PaxI) (Fig. 2A, Fig. S1 in Supporting information).

  Figure 2

  

  Figure 2.  Biosynthesis of sesquiterpene moiety in 2. (A) The pax gene cluster from P. janthinellum. (B) LC-MS analysis of metabolites produced by A. nidulans transformed with different combinations of pax genes. (C) Biosynthetic pathway from farnesyl diphosphate (FPP) to 15.

  DownLoad: Full-Size Img PowerPoint

  Based on the structural features of 1 and 2, we proposed that the terpene and the alkaloid portions are heavily oxidized by redox enzymes, and the sesquiterpene is a triply-oxygenated derivative of amorphadiene, which is the precursor of antimalarial artemisinin (Fig. S2 in Supporting information) [21]. Bioinformatic analysis showed PaxD, PaxH and PaxC encoded by the pax cluster share moderate similarity to sesquiterpene oxidase AneD (36%), AneG (41%), and AneA (53%), respectively (Fig. S3 in Supporting information) [12,22]. This indicates that PaxD, PaxC, and PaxH possibly function at the sesquiterpene moiety in 2.

  To elucidate the biosynthetic pathway of the sesquiterpene moiety in 1/2, we constructed recombinant strains of A. nidulans expressing different combinations of genes paxC/D/E/H/I. When paxC/I or paxD/I were co-expressed in A. nidulans, no new metabolite was detected (Fig. 2B, traces iii and iv). While co-expression of paxH/I led to genenation of two additional metabolites 9 and 10 (Fig. 2B, trace ii). NMR characterization revealed the oxidation at C-2 and epoxidation at C-10/11 in the structures of 9 and 10. We proposed that PaxH catalyzed the transformation of amorphadiene into 9, which dehydrogenated to form 10 by the host. To confirm the function of PaxH, we fed amorphadiene into Saccharomyces cerevisiae expressing paxH (SC-paxH) and observed the generation of 9 (Fig. S4 in Supporting information). These results verified that P450 monooxygenase PaxH catalyze two-step oxidations, hydroxylation and epoxidation, at amorphadiene to form 9.

  With paxH/I, further expression of paxC did not lead to new metabolite, while adding the paxD led to formation of 11 (Fig. 2B, traces v–vi). The additional hydroxy group was assigned at C1 based on the HMBC correlations, which suggested PaxD was responsible for catalyzing hydroxylation at C1. In addition, when co-expressed paxE/H/I in A. nidulans, we observed the formation of 12, which was confirmed to be 2-O-acetyl-9 (Fig. 2B, trace vii). We proposed that acyltransferase PaxE catalyzed the conversion of 9 to 12. To confirm the function of PaxE, we expressed and purified PaxE as a recombinant protein from Escherichia coli BL21 Transetta and directly performed in vitro assays in the presence of 2.3 µmol/L PaxE, 0.1 mol/L of acetyl-CoA and 0.1 mmol/L of substrates (9 or 10). Liquid chromatograph-mass spectrometer (LC-MS) analysis showed that the purified PaxE converted 9 and 11 to 12 and 14, respectively (Fig. S5 in Supporting information). In order to further confirm the natural substrate of PaxE, we performed in vitro reaction using low concertation of PaxE (0.1 µmol/L) with 9 and 11 at a ratio of 1:1 as substrate, which showed that only 9 was converted to 12 (Fig. S6 in Supporting information). This result highly suggested 9 as the preferred substrate of PaxE. Then we separately fed amorphadiene, 9, 12, and 13 into S. cerevisiae expressing paxD (SC-paxD) to identify the role of PaxD. Although feeding amorphadiene did not lead to new metabolite, the transformation of 9, 12, and 13 into 11, 14, and 16, respectively, was observed (Fig. S7 in Supporting information), which indicated that PaxD is the enzyme required for hydroxylation at C1 and the formation of C1-OH happened after oxidation at C-2. Furthermore, we performed in vitro assays using reactions containing PaxD microsomes, 0.1 mol/L of NADPH, and 0.2 mmol/L of the substrate (9/12/13) and observed that only 12 was converted to 14 (Fig. S8 in Supporting information). These highly suggested that the natural substrate of PaxD was 12. We speculated that activity of PaxD in microsome was weak in the in vivo assay, resulting in inability of conversion of non-natural substrates.

  We proposed the next step to form the terpene fragment is hydroxylation at C-8a. Co-expression of paxC/D/E/H/I in A. nidulans led to the generation of 15, suggesting that PaxC was responsible for the hydroxylation at C-8a (Figs. 2B and C). Furthermore, we also detected the formation of 16 and 17, which were derived from the hydration and acetylation of 14, respectively. To confirm the function of PaxC, we then expressed and purified the recombinant PaxC from E. coli BL21, and initially performed in vitro assays using reactions containing 100 µmol/L of purified PaxC, 1 mmol/L FeSO₄, 10 mmol/L α-ketoglutarate (α-KG) and 0.1 mmol/L of the substrate (amorphadiene, 9, 11, 12, or 14, separately). The transformation from 12 and 14 to 12a and 15, respectively, was observed (Fig. S9 in Supporting information), which indicates that hydroxylation at C-8a was catalyzed by PaxC. While the failure of PaxD in conversion of 12a to 15, highly suggested that 12a was a shunt product (Figs. S7 and S8). In addition, the incomplete consumption of 14 at the high concentration of PaxC, suggested that 14 may not be the natural substrate of PaxC.

  Having established the pathway of the sesquiterpene fragments in 2, we next tried to identify the pathways leading to the pyrrolobenzoxazine core. To identify the role of NRPS PaxG, compounds 11, 14, and 15 were separately fed to A. nidulans expressing paxG (AN-paxG) (Fig. 3A, Fig. S10 in Supporting information). While feeding 11 (or 15) did not led to any new metabolites, feeding 14 led to the formation of 18, an adduct of 14 and l-tryptophan (Fig. 3A, Fig. S10). These supported that PaxG function before PaxC which catalyze the hydroxylation at C-8a.

  Figure 3

  

  Figure 3.  Biosynthesis of pyrrolobenzoxazine core of 2. (A) LC-MS analysis of chemical remediation experiment of 14. (B) LC traces from in vitro assays of PaxC activity. Reaction conditions: 0.1 mmol/L substrate, 0.1 µmol/L PaxC, 1 mmol/L FeSO₄, 10 mmol/L α-KG and 50 mmol/L Tris–HCl buffer (pH 7.9), 20 ℃ for 5 min. (C) Biosynthetic pathway from 14 to 1 and 2. Blue marks: terpene moiety; black marks: alkaloid portions.

  DownLoad: Full-Size Img PowerPoint

  With 18 in hand, we next investigated the formation of the N-methylpyrroloindoline scaffold. Feeding of 14 to A. nidulans expressing paxG/B did not lead to any methylated indole products (Fig. 3A, trace iii), which suggested that the function of PaxB probably required the presence of pyrroleindole ring in the substrate. The remain oxidases, PaxA and PaxF, were clustered with the indole hydroxylase AspB [23] in phylogenetic analysis (Fig. S11 in Supporting information) and our previous study has identified PaxA as the enzyme responsible for the formation of 1,2-oxazine core [12], which suggested that PaxF could be a candidate enzyme to install the pyrroleindole core. Indeed, when 14 was fed to A. nidulans expressing paxG/F (AN-paxG/F), the generation of a new product 19 was observed (Fig. 3A, trace iv). NMR characterization revealed a pyrroleindole core in the structure of 19. These results highly suggested PaxF responsible for installation of pyrroleindole moiety. Base on the enzymatic mechanism of homologues of PaxF, such as AspB [23], NotB [24], and CtdE [25] (Fig. S12 in Supporting information), a mechanism of PaxF catalyzing the formation of pyrroleindole core could be proposed. A first formation of 2,3-β-face epoxidation to facilitate α-amino attack at electrophilic C2 of the epoxyindole ring (Fig. 3C). Next, we fed 14 into the AN-paxG/F/B to identify the role of PaxB, and N-methylpyrroleindole product 20 was detected (Fig. 3A, trace v). These results indicated that the formation of pyrroleindole ring happened before N-methylation catalyzed by PaxB.

  Previously, we have identified 21 as the natural substrate of PaxA, which suggested that hydroxylation at C-8′a occurred before the formation of 1,2 oxazine core [12]. To identify the natural substrate of PaxC, the in vitro reactions using low concertation (0.1 µmol/L) of PaxC with substrates (14/18/19/20) at 20 ℃ for 5 min were performed (Fig. S13 in Supporting information). When co-incubated with 14 and PaxC (0.1 µmol/L), the formation of C8′a-OH product 15 was not detected, which supported the above result of feeding 14 to AN-paxG. Then the compounds 18, 19, and 20 were separately add to in vitro assays of PaxC, and only the corresponding C-8′a hydroxylation products of 19 and 20 were detected. Furthermore, when a mixture of 19 and 20 (1:1 ratio) was used in the in vitro assay, 20 was firstly converted to 21 (Fig. 3B). Therefore, 20 was suggested as the natural substrate of PaxC.

  Based on the results of heterologous expression, feeding experiment and in vitro biochemical assays, we established the complete biosynthetic pathway of paeciloxazine (1) and peniciloxazine A (2). Firstly, the terpene cyclase PaxI converts FPP to amorphadiene, which is converted to 9 catalyzed by the P450 monooxygenase PaxH through two-step oxidations, hydroxylation at C-2 and epoxidation at C-9/10. Next, 9 is transform into 12 catalyzed by acyltransferase PaxE, followed by the formation of C1-OH to obtain 14 catalyzed by PaxD. Then the esterification of l-tryptophan with 14 occurs in the presence of PaxG to form 18, which is catalyzed by a FMO PaxF and methyltransferase PaxB in turn to form N-methypyrroleindole 20. Then, 20 is converted to paeciloxazine (1) catalyzed by PaxA. Alternatively, 20 can also be converted to 21 by PaxC, following by the formation of 2 catalyzed by PaxA (Fig. S14 in Supporting information).

  In all, the full biosynthetic pathway of 1 and 2 was established and 15 compounds were obtained by heterologous expression. In order to explore the potential applications of these on-pathway intermediates, 1, 2, 9–17, 20 and 21 were evaluated for their inhibitory effect on epileptiform activity of 4-AP-induced synchronized calcium oscillations (SCOs), with the NMDA receptor inhibitor MK801 [26] as a positive control. Compounds 13, 17, and 20 showed significant activities of inhibiting calcium oscillation (Fig. 4), which suggested their potential antiepileptic activity.

  Figure 4

  

  Figure 4.  Representative curves of inhibition of SCOs by the positive drugs MK801 and compounds 13, 17, and 20. The arrows: vehicle control.

  DownLoad: Full-Size Img PowerPoint

  During our investigation of the biosynthesis of 1, Tang group reported the biosynthetic pathway of another pyrrolobenzoxazine compound CJ-12662 (Fig. S15 in Supporting information) [27]. Analysis of the biosynthetic pathways of CJ-12662 and peniciloxazine A (2) showcased the different strategies nature employs to generate similar nature product (Fig. S16 in Supporting information). First, the hydroxylation happened at C2, C1, and C8a in turn in the biosynthesis of paeciloxazine A (2), while happened at C1, C2, and C8a in turn in CJ-12662. Contrast with the P450 monooxygenase ThmG in the biosynthesis of CJ-12662, which is responsible for C8a hydroxylation of dihydroxyl sesquiterpene, the α-KG dependent dioxygenase PaxC in biosynthesis of 2 catalyzes the C8a hydroxylation of an alkaloid-terpene substrate 20. It is notable that different strategies were employed for acetylation during the biosynthesis of these two compounds. The acetylation requires the presence of polyketide synthase (PKS) ThmK and acyltransferase ThmJ in the biosynthesis of CJ-12662, whereas the acetylation is accomplished by acetyltransferase PaxE alone in the case of 2. Different from ThmG that catalyzes the hydroxylation at C2, PaxH is responsible for the epoxidation at C9/10 besides the formation of C2-OH.

  In summary, we fully elucidated the biosynthetic pathway of paeciloxazine (1), and revealed the diversity and complexity of constructing natural products by organisms. In addition, a total of 15 compounds were obtained by heterologous expression, of which compounds 13, 17, and 20 showed potential antiepileptic activities.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  CRediT authorship contribution statement

  Kunya Wang: Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – original draft. Bingyu Liu: Formal analysis, Funding acquisition, Investigation, Methodology. Daojiang Yan: Investigation, Methodology. Jian Bai: Methodology, Writing – review & editing. Haibo Yu: Investigation, Validation. Youcai Hu: Data curation, Funding acquisition, Project administration, Supervision, Validation, Writing – original draft, Writing – review & editing.

  Acknowledgments

  This work was supported financially by the National Natural Science Foundation of China (Nos. 22107122 and 82225042) and the CAMS Innovation Fund for Medical Sciences (CIFMS, No. 2021-I2M-1–029).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.109811.

  
  
• 1. [1]

     D.J. Newman, M.C. Gordon, J. Nat. Prod. 83 (2020) 770–803. doi: 10.1021/acs.jnatprod.9b01285

  2. [2]

     C.T. Walsh, Y. Tang, Natural Product Biosynthesis: Chemical Logic and Enzymatic Machinery, 1st. ed., The Royal Society of Chemistry, Croydon, 2017.

  3. [3]

     W. Yuan, F. Li, Z. Chen, et al., Chin. Chem. Lett. 35 (2024) 108788. doi: 10.1016/j.cclet.2023.108788

  4. [4]

     F. Jiang, A. Liu, Q. Wei, Y. Hu, Chin. Chem. Lett. 35 (2024) 109504. doi: 10.1016/j.cclet.2024.109504

  5. [5]

     T. Dairi, T. Kuzuyama, M. Nishiyama, I. Fujii, Nat. Prod. Rep. 28 (2011) 1054–1086. doi: 10.1039/c0np00047g

  6. [6]

     M.C. Tang, Y. Zou, K. Watanabe, et al., Chem. Rev. 117 (2016) 5226–5333.

  7. [7]

     C.T. Walsh, B.S. Moore, Angew. Chem. Int. Ed. 58 (2019) 6846–6879. doi: 10.1002/anie.201807844

  8. [8]

     K. Yoshinori, F. Tsukasa, S. Kochi, et al., J. Antibiot. 57 (2004) 24–28. doi: 10.7164/antibiotics.57.24

  9. [9]

     D.A. Perry, H. Maede, J. Tone, Patent, GB2240100 A, 1991.

  10. [10]

      Y. Kojima, H. Kaminawa, N. Aichiken, et al., Patent, WO 95/19363, 1995.

  11. [11]

      C. Didier, D.J. Critcher, N.D. Walshe, et al., J. Org. Chem. 69 (2004) 7875–7879. doi: 10.1021/jo048711t

  12. [12]

      D. Yan, K. Wang, S. Bai, et al., J. Am. Chem. Soc. 144 (2022) 4269–4276. doi: 10.1021/jacs.2c00881

  13. [13]

      B. Chaiyosang, K. Kanokmedhakul, K. Soytong, et al., Planta Med. 87 (2021) 600–610. doi: 10.1055/a-1392-1038

  14. [14]

      K. Kanokmedhakul, S. Kanokmedhakul, R. Suwannatrai, et al., Tetrahedron 67 (2011) 5461–5468. doi: 10.1016/j.tet.2011.05.066

  15. [15]

      J. Paluka, K. Kanokmedhakul, M. Soytong, et al., Fitoterapia 137 (2019) 104257. doi: 10.1016/j.fitote.2019.104257

  16. [16]

      N.A.L. Chubb, D.J. Critcher, J.J. Eshelby, et al., Patent, WO 2004/072086, 2004.

  17. [17]

      M. Masi, A. Andolfi, V. Mathieu, et al., Tetrahedron 69 (2013) 7466–7470. doi: 10.1016/j.tet.2013.06.031

  18. [18]

      C. Commandeur, M. Commandeur, K. Bathany, et al., Tetrahedron 67 (2011) 9899–9908. doi: 10.1016/j.tet.2011.09.112

  19. [19]

      D. Schwaebisch, K. Tchabanenko, R.M. Adlington, et al., Chem. Commun. 36 (2004) 2552–2553.

  20. [20]

      B.K. Das, E. Tokunaga, K. Harada, et al., Org. Chem. Front. 4 (2017) 1726–1730. doi: 10.1039/C7QO00234C

  21. [21]

      C.M. Bertea, J.R. Freije, H. van der Woude, et al., Planta Med. 71 (2005) 40–47. doi: 10.1055/s-2005-837749

  22. [22]

      C.F. Lee, L.X. Chen, C.Y. Chiang, et al., Angew. Chem. Int. Ed. 58 (2019) 18414–18418. doi: 10.1002/anie.201910200

  23. [23]

      S.W. Haynes, X. Gao, Y. Tang, C.T. Walsh, J. Am. Chem. Soc. 134 (2012) 17444–17447. doi: 10.1021/ja308371z

  24. [24]

      S. Li, J.M. Finefield, J.D. Sunderhaus, et al., J. Am. Chem. Soc. 134 (2012) 788–791. doi: 10.1021/ja2093212

  25. [25]

      Z. Liu, F. Zhao, B. Zhao, et al., Nat. Commun. 12 (2021) 4158. doi: 10.1038/s41467-021-24421-0

  26. [26]

      P. Kovacic, R. Somanathan, Oxid. Med. Cell. Longev. 3 (2010) 13–22. doi: 10.4161/oxim.3.1.10028

  27. [27]

      W. Cheng, M. Chen, M. Ohashi, Y. Tang, Angew. Chem. Int. Ed. 61 (2022) e202116928. doi: 10.1002/anie.202116928

• 

  Figure 1  The family of pyrrolobenzoxazine natural products.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  Biosynthesis of sesquiterpene moiety in 2. (A) The pax gene cluster from P. janthinellum. (B) LC-MS analysis of metabolites produced by A. nidulans transformed with different combinations of pax genes. (C) Biosynthetic pathway from farnesyl diphosphate (FPP) to 15.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  Biosynthesis of pyrrolobenzoxazine core of 2. (A) LC-MS analysis of chemical remediation experiment of 14. (B) LC traces from in vitro assays of PaxC activity. Reaction conditions: 0.1 mmol/L substrate, 0.1 µmol/L PaxC, 1 mmol/L FeSO₄, 10 mmol/L α-KG and 50 mmol/L Tris–HCl buffer (pH 7.9), 20 ℃ for 5 min. (C) Biosynthetic pathway from 14 to 1 and 2. Blue marks: terpene moiety; black marks: alkaloid portions.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  Representative curves of inhibition of SCOs by the positive drugs MK801 and compounds 13, 17, and 20. The arrows: vehicle control.

  下载: 全尺寸图片 幻灯片
•