Delineating Monascus azaphilone pigment biosynthesis: oxidoreductive modifications determine the ring cyclization pattern in azaphilone biosynthesis

Abstract

The product profiles of mppF, mppA, and mppC mutants substantiate that MppA-mediated ω-2 ketoreduction is a prerequisite for the synthesis of the pyranoquinone bicyclic core of the Monascus azaphilone pigment and that MppC activity determines the regioselectivity of the spontaneous Knoevenagel condensation.

---

Fungal polyketide synthases (fPKSs) are iteratively acting multi-domain polypeptides and generally classified as non-reducing- (NR-), partially reducing-, or highly reducing-types, depending on the extent of reductive modifications that occur during the assembly of acetate units through a decarboxylative Claisen condensation.¹ Azaphiloid is a class of polyketides that bears a highly oxygenated pyranoquinone bicyclic core and is generally known of fungal origin.² The azaphilone polyketide is synthesized by an NR-fPKS with a reductive release domain (-R), and the pyran ring cyclization follows a hydroxylation-mediated dearomatization of a benzaldehyde intermediate, as shown in azanigerone biosynthesis (Scheme 1A).³ Azanigerone NR-fPKS-R incorporates one acetyl-CoA, five malonyl-CoAs, and a methyl moiety from S-adenosyl methionine; the resulting polyketide intermediate undergoes a ketoreduction at the C-11 position, generating FK17-P2a (1), but the timing of this ketoreduction is unresolved. A biochemical study clearly demonstrated that a monooxygenase AzaH introduces a hydroxyl group at the C-4 position of 1 to generate azanigerone E (2) (Scheme 1A).³ This conversion was proposed to involve 3 and 4. We maintain the position numbering pattern used in the structures of 1 and 2 throughout this report in order to emphasize the biosynthetic relatedness of the compounds described.

Scheme 1 (A) AzaH-mediated pyran ring cyclization mechanism. (B) The proposed biosynthetic origin of MAzP with the role of the acyltransferase MppB and the MAzP fatty acid synthase MpFAS2. Bold lines and black circles denote acetate units and S-adenosyl methionine-derived carbons.

Monascus species including M. pilosus, M. purpureus, and M. ruber are known to produce Monascus azaphilone pigments (MAzPs), which are the active ingredients of the traditional food colorant derived from the fermentation of Monascus.⁴ It is also known that some MAzPs display diverse biological activities that include anti-diabetic, anti-inflammatory, anti-atherosclerotic, and anti-cancer activities.^5–8 MAzPs include ankaflavin (5), monascin (6), rubropunctatin (7), and monascorubrin (8) (Scheme 1B). The MAzP biosynthetic gene cluster was previously described, and a genetic knockout of the MAzP PKS gene (MpPKS5) abolished MAzP production in M. purpureus.⁹ MpPKS5 belongs to NR-fPKS-R. MpFAS2, the canonical fatty acid synthase encoded in the gene cluster, was also shown to be essential for MAzP biosynthesis.¹⁰ MpFAS2 is proposed to synthesize short-chain 3-oxo-fatty acyl thioesters for MAzP biosynthesis. The MAzP biosynthetic gene cluster encodes an acyltransferase MppB, which is assigned as the catalyst that mediates the installation of the MpFAS2 products at the tertiary alcoholic oxygen at C-4 (Scheme 1B).

We also confirmed that the mppB-knockout mutant is incapable of producing MAzP (data not shown). Knoevenagel condensation between the α carbon of the 3-oxo-fatty acyl moiety and the C-5 carbonyl group may generate a 2-furanone moiety, leading to the production of 5–8 (Scheme 1B). The biosynthetic study of chaetoviridin in Chaetomium globosum demonstrated that the 2-furanone-forming cyclization is non-enzymatic,¹¹ but the nature of this reaction for 5–8 is yet to be unveiled because the ring cyclization geometry for 5–8 differs from that for chaetoviridin. The pyranoquinone bicyclic structure of 7 and 8 is identical to 2, but the exocyclic 2-hydroxypropane moiety of 2 is replaced with a propene moiety in 7 and 8. This suggests two plausible scenarios for MpPKS5 catalysis: crotonyl-CoA serves as the starting unit to generate 9 or MpPKS5 generates 2 through the same synthetic route as azanigerone (Scheme 1B). The latter scenario requires a dehydration step to yield MAzP. With the presumed intermediacy of 2 or 9 in MAzP biosynthesis, it has been tempting to think of 7/8 as the precursor to 5/6. It was previously shown that an mpp7 knockout resulted in accumulation of monasfluol A (10) and B (11) without 5–8 (Scheme 2).¹² Mpp7 was proposed to control the regioselectivity of the Knoevenagel aldol condensation. Although the catalytic role of Mpp7 was not defined, accumulation of 10 and 11 supports the idea that 2 is the true intermediate of MAzP biosynthesis. In order to assess the polyketide intermediate of MAzP biosynthesis, the homolog of azaH (mppF) was inactivated in the MAzP biosynthetic gene cluster (see ESI† for gene inactivation), which resulted in a high accumulation of 1 (Fig. 1B; see ESI† for NMR spectra). In this report, the LC-MS result of the culture supernatant extract is provided (see ESI† for LC-MS result of the culture mycelium extract).

Scheme 2 Structure of 12–15 and the proposed biosynthetic rout highlighting the catalytic role of MppA.

Fig. 1 LC-MS analysis of the culture supernatant extracts of M. purpureus WT (A) and the mutants of mppF (B), mppA (C), and mppC (D).

Administration of 1 to the MpPKS5 knockout mutant restored the production of 5–8 (data not shown). This experiment established that the early biosynthetic route of MAzP is common to that of azanigerone. In the azanigerone biosynthetic gene cluster, two oxidoreductase genes azaE and azaJ were identified.³ AzaE was proposed to mediate the ketoreduction that is required for the generation of 1, but there is no experimental data supporting this functional assignment. The MAzP biosynthetic gene cluster encodes three reductase candidates, MppA, MppC, and MppE, but none of them bears a significant homology with AzaE. Instead, AzaJ has a significant homology with MppE (47% identity/62% similarity). To access the roles of the oxidoreductive modifications in the MAzP biosynthesis, we generated genetic knock-out mutants of mppA and mppC in M. purpureus (see ESI† for gene inactivation). Inactivation of mppE could not be achieved for unidentified reasons; the role of MppE in the MAzP biosynthesis is yet to be unveiled. An mppC-knockout mutant of Monascus ruber was reported to accumulate yellow pigments other than 5–8.¹³ However, the identities of these pigments were not determined. LC-MS analysis indicated that MAzPs are below the detection level in the mppA mutant. Instead, this mutant accumulated four compounds (12–15), which eluted substantially faster than 5–8 (Fig. 1C). MS analysis identified the molecular ions of [M + H]⁺ 251, 249, 233, and 247, for 12, 13, 14, and 15, respectively, substantiating that 12–15 were devoid of the short fatty acyl chain. The mppC mutant produced four compounds, which possessed elution times similar to those of 5–8 (Fig. 1D). These yellow pigments have molecular masses comparable to those of 5–8, with molecular ion peaks of [M + H]⁺ at m/z 357, 355, 385, and 383 (16–19), respectively.

We purified 12–15, which have the names of MA-1 to -4 in our lab, respectively, and determined their structures (Scheme 2; see ESI† for structural determination). 12 and 14 appear structurally related. It is thus tempting to suggest that 14 is converted into 12 by the MppF-mediated hydroxylation and then subsequent reduction. 12 is the previously reported 1-tetralone compound (monaspurpurone) from M. purpureus.¹⁴ Although the monaspurpurone isolate was found to be racemic,¹⁴ we assume here that the C-4 of 12 retains the same R-configuration as 5–8. 14 has not previously been identified to the best of our knowledge. 13 and 15 were 1,4-naphtoquinone derivatives known to be produced by a A. nidulans transformant of a NR-fPKS-R gene (ATEG_03432) of A. terreus origin.¹⁵

The notable structural feature of 12–15 is their C10 bicyclic core, suggesting that they are generated by a Knoevenagel aldol condensation of the presumed intermediate, 20. It seems plausible that MppA is involved in the formation of 21 (Scheme 2). The timing of the MppA-mediated ketoreduction, whether it occurs on the very early intermediate (acetoacetyl-thioester tethered on the acyl carrier domain) or it precedes the reductive release from MpPKS5, could not be determined. We here adopt the latter scenario due to the considerable yield of 12–15: their isolation yields from a 1.5 liter culture were 37, 14, 29, and 8 mg, respectively. In the former scenario, acetoacetyl-thioester tethered on MpPKS5 will be generated in the mppA mutant. MpPKS5 is probably incapable of processing efficiently this aberrant intermediate, giving rise to a low level of 12–15, even if they are produced. The biosynthetic proposal for 12–15 is that 22 is released from MpPKS5, and the resulting compound 20 is converted into the C10 bicyclic structure. We also propose that the MppA-mediated ketoreduction suppresses the spontaneous Knoevenagel aldol condensation that leads to 12–15 by lowering acidity of the C-10 position, paving the way to a pyranoquinone bicyclic structure.

MppA is 297 amino acids long and belongs to the Rossmann-fold NAD(P)⁺-binding protein family (cd05233, COG1028). In terms of polyketide biosynthesis, MppA can be classified as a member of the ketoreductase family. MppA is homologous to A. terreus ATEG_03438 (59% identity/77% similarity) and mppA is neighbored by the aforementioned NR-fPKS-R gene, ATEG_03432 that generated 13 and 15 in an ectopic expression.¹⁵ The azaH (the C-4 hydroxylase gene) homolog (ATEG_03443) can also be found nearby ATEG_03432. It is thus tempting to suggest that the gene cluster containing ATEG_03432 encodes the biosynthesis of a yet-unknown azaphilone compound. The azanigerone gene cluster does not harbor an mppA homolog, while 2 is predicted to be the common intermediate in both of the azanigerone and MAzP pathways. There are two possibilities. One is that MppA and its counterpart in the azanigerone pathway are not similar due to their convergent evolution. The other is that the timing of this ketoreduction differs in two pathways, and thus, the two ketoreductases differ intrinsically.

The four compounds in the mppC mutant are identified as MAzP derivatives (16–19, named MC-1 to -4 in our experiments), possessing the lactone (2-furanone for 17 and 19) ring geometry of 10 and 11 (Scheme 3).¹² LC-MS analysis indicated that small amounts of 16 and 18 were also detectible in WT (Fig. 1A). It was thus possible that trace amounts of 17 and 19 also existed in the WT extract, but this LC-MS analysis could not identify them due to their overlapping with 7 and 8 in the elution. The structural determination of 16–19 was straightforward due to their resemblance to 10 and 11 (see ESI† for NMR spectra). 16 and 18 are notably different from 10 and 11 in possessing two allylic proton signals at δ_H (ppm) 6.54 (10-H, doublet of quartet) and 5.99 (11-H, doublet). 16 and 18 were found to be identical to monasfluore A and B, respectively, which were reported from a Monascus species (Scheme 3).¹⁶ 17 and 19 are novel MAzP derivatives possessing a double bond between C-3 and C-15 (Scheme 3). A reduction may convert 17 and 19 into 16 and 18, respectively, but the reduction catalyst is unknown, as in the case for 10 and 11.¹² In the mppC mutant, another compound (23) was accumulated (Fig. 1D). The elution time of 23 was similar to those of 1 and 13, but the mass analysis (m/z 237 and 259 for [M + H]⁺ and [M + Na]⁺, respectively) indicated that 23 is distinct from them. We completed structural determination of 23 (see ESI† for NMR spectra and the structure), identifying it as FK17-P2b1, which was previously reported from Aspergillus sp.¹⁷ and later from a yellow mutant of Monascus kaoliang.¹⁸ 23 is closely related to monascusone A¹⁸ that was previously found to accumulate in the MpfasB mutant.¹⁰

Scheme 3 Structure of 16–19 and the proposed biosynthetic rout featuring the role of MppC controlling the regioselectivity of the intramolecular Knoevenagel condensation. R is –CH₃ or –(CH₂)₂CH₃ in the cases of 24–27.

Accumulation of 10 and 11 in the mpp7 mutant had previously led us to propose that Mpp7 controls regioselective Knoevenagel condensation during the 2-furanone ring formation.¹² The ring cyclization geometry for 10 and 11 is the same as that for chaetoviridin.¹¹ It was thus proposed that the ring cyclization for 5–8 demands Mpp7, while the reaction for 10 and 11 is non-enzymatic as shown in chaetoviridin biosynthesis.^11,12 The biochemical role of Mpp7 was veiled because it bears no significant sequence homology to other proteins with known biochemical functions. However, it was found that the mppC mutant (mpp7⁺/mppC⁻) accumulated 16–19 that have the ring geometry of 10 and 11 (Fig. 1D). This result indicates that Mpp7 alone is incapable of completing the regioselective Knoevenagel condensation for 5–8. It seems more likely that an Mpp7-mediated modification is a factor affecting the regioselectivity of the aldol condensation. The intermediacy of 1 (Fig. 1A) strongly supports the involvement of a dehydration step in MAzP biosynthesis. We here propose that Mpp7 is the dehydratase. In the proposed biosynthetic pathway, 2 is acylated by MppB into 24, which is then converted into 25 by Mpp7 (Scheme 3). In the absence of Mpp7, 24 is converted to 10/11 through a Knoevenagel aldol condensation/cyclization and then subsequent reduction. The reduction step is efficient enough to drive the quantitative conversion of 24 into 10/11: the mpp7 mutant yielded a large accumulation of 10 and 11 with no other notable pigment product.¹² In the case of 17/19, the reductive conversion into 16/18 is relatively inefficient, allowing a comparable balance between 17/19 and 25. This situation resulted in the hydrolysis of 25 into 9, which is then converted into 23 (see ESI† for proposed biosynthetic route of 23). As demonstrated in the chaetoviridin biosynthesis,¹¹ together with the occurrence in 10–11 and 16–19, the nucleophilic attack of C-15 occurs to C-3 carbonyl but not to C-5, as for 24 and 25. This regioselectivity may operate because C-3 is a better electron acceptor than C-5, which belongs to an extended π-conjugated system of the 4H-pyran-4-ylidene moiety. It is thus reasonable to envision that MppC induces a structural change in the pyranoquinone structure of 25. We propose here that MppC reduces 25 into 26, making C-5 a better electron acceptor by eliminating the π-conjugated system of 25 (Scheme 3). In this proposal, the resulting compound 27 undergoes two parallel reactions each of which leads to 5/6 and 7/8.

Conclusions

The present study advances our understanding of the chemical strategy used in the azaphilone biosynthesis. This study defines NR-fPKS-R, MppA, and MppF as a catalytic tool-kit for azaphilone biogenesis. The study of MppC provides a lesson on how the divergence of the azaphilone pathway evolved: the coordination of oxidoreductive catalyst in altering the geometric specificity of the spontaneous Knoevenagel aldol condensation.

Acknowledgements

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2013R1A1A2059458).

Notes and references

1. R. J. Cox, Org. Biomol. Chem., 2007, 5, 2010–2026.
2. J. M. Gao, S. X. Yang and J. C. Qin, Chem. Rev., 2013, 113, 4755–4811.
3. A. O. Zabala, W. Xu, Y. H. Chooi and Y. Tang, Chem. Biol., 2012, 19, 1049–1059.
4. Y. Feng, Y. Shao and F. Chen, Appl. Microbiol. Biotechnol., 2012, 96, 1421–1440.
5. L. C. Hsu, Y. W. Hsu, Y. H. Liang, C. C. Liaw, Y. H. Kuo and T. M. Pan, Molecules, 2012, 17, 664–673.
6. C. L. Lee, Y. P. Hung, Y. W. Hsu and T. M. Pan, J. Agric. Food Chem., 2013, 61, 143–150.
7. C. L. Lee, J. Y. Wen, Y. W. Hsu and T. M. Pan, J. Agric. Food Chem., 2013, 61, 1493–1500.
8. L. C. Hsu, Y. H. Liang, Y. W. Hsu, Y. H. Kuo and T. M. Pan, J. Agric. Food Chem., 2013, 61, 2796–2802.
9. B. Balakrishnan, S. Karki, S. H. Chiu, H. J. Kim, J. W. Suh, B. Nam, Y. M. Yoon, C. C. Chen and H. J. Kwon, Appl. Microbiol. Biotechnol., 2013, 97, 6337–6345.
10. B. Balakrishnan, H. J. Kim, J. W. Suh, C. C. Chen, K. H. Liu, S. H. Park and H. J. Kwon, J. Korean Soc. Appl. Biol. Chem., 2014, 57, 191–196.
11. J. M. Winter, M. Sato, S. Sugimoto, G. Chiou, N. K. Garg, Y. Tang and K. Watanabe, J. Am. Chem. Soc., 2012, 134, 17900–17903.
12. B. Balakrishnan, C. C. Chen, T. M. Pan and H. J. Kwon, Tetrahedron Lett., 2014, 55, 1640–1643.
13. Q. Liu, N. Xie, Y. He, L. Wang, Y. Shao, H. Zhao and F. Chen, Appl. Microbiol. Biotechnol., 2014, 98, 285–296.
14. M. J. Cheng, M. D. Wu, I. S. Chen, C. Y. Chen, W. L. Lo and G. F. Yuan, Nat. Prod. Res., 2010, 24, 1719–1725.
15. Y. M. Chiang, C. E. Oakley, M. Ahuja, R. Entwistle, A. Schultz, S. L. Chang, C. T. Sung, C. C. Wang and B. R. Oakley, J. Am. Chem. Soc., 2013, 135, 7720–7731.
16. Z. Huang, Y. Xu, L. Li and Y. Li, J. Agric. Food Chem., 2007, 56, 112–118.
17. T. Arai and H. Sano, Patent No. J.P. 6,329,576, 29 November 1997.
18. S. Jongrungruangchok, P. Kittakoop, B. Yongsmith, R. Bavovada, S. Tanasupawat, N. Lartpornmatulee and Y. Thebtaranonth, Phytochemistry, 2004, 65, 2569–2575.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details, extra liquid chromatograms, NMR spectra, and UV-Vis absorption spectra. See DOI: 10.1039/c4ra11713a

---