Absolute configuration of azaphilones from Monascus kaoliang KB9 and solvent effects on their keto and enol forms

Abstract Monascus fermented rice, also known as red yeast rice, exhibits a broad spectrum of biological activities due to its chemical constituents, such as monacolins and azaphilone pigments. Here, we cultured Monascus kaoliang KB9 in a liquid malt medium instead of on rice as a carbon source. Eleven known compounds (1–11) containing azaphilones and their early intermediate were isolated and identified. However, this was the first time that angular tricyclic azaphilones, monasfluols A (4) and B (7), acetyl-monasfluol A (5) and monasfluore A (6), were isolated from this species. Interestingly, all isolated tricyclic azaphilones existed exclusively in enol form in CD3OD, as evidenced by NMR spectroscopy. The absolute configuration of compounds 4–7 was also first experimentally identified based on ECD spectroscopy combined with conformational analyses using computational techniques. The assigned stereochemistry of Monascus azaphilones in this work provides essential structural information that will benefit future biological and pharmaceutical investigations. Graphical Abstract


Introduction
Rice fermented with Monascus has been employed in food products such as beverages, food colourants and supplements consumed in many Asian countries (Wang and Lin 2007). It typically has a distinctive red colour and is known as red yeast rice (RYR). Major bioactive compounds found in RYR are monacolins and azaphilone pigments. To date, only a few strains of M. kaoliang have been explored, including M. kaoliang KB9 and its colour mutant derivatives (Yongsmith et al. 1994;2000). These Monascus strains were grown by solid-state fermentation on rice medium for high-level pigment production (Chayawat et al. 2009;Songjanthuek et al. 2021). However, the growth of M. kaoliang KB9 by submerged culture could lead to the discovery of novel bioactive compounds.
Here, we report metabolites isolated from M. kaoliang KB9 cultured in a malt liquid medium, including their antibacterial and antifungal activities. In addition, we investigated solvent effects on the distribution of keto and enol forms of the isolated azaphilones. The absolute configurations of some azaphilones, which have not been previously reported, were also determined based on NMR spectroscopy and experimental and calculated electronic circular dichroism (ECD). The latter has proved to be successful in determination of absolute configuration of bioactive azaphilones (Cao et al. 2019;2020).

Results and discussion
2.1. Extraction and isolation of azaphilones and their early intermediate M. kaoliang KB9 was first screened for metabolite production in four different liquid media: potato dextrose medium, rice medium, sucrose-yeast extract medium and malt medium. Malt medium was then used for large-scale production due to the number of metabolites produced ( Figure S1). A large amount of compound 1 (Figure 1) was isolated and identified as (þ)-FK17-P2a (Table S1 and Figure S2). Compound 1 also showed a positive optical rotation in methanol, similar to that of the previous report in which the absolute configuration was also confirmed by Mosher's method and established as 10 R (Stierle et al. 2012). Compound 1 is also a methylated hexaketide intermediate for the Monascus azaphilones (Bijinu et al. 2014), of which ten (2-11) were identified in this work ( Figure 1). They are lunatinin (2), FK17-P2b1 (3), monasfluol A (4), acetyl-monasfluol A (5), monasfluore A (6), monasfluol B (7), acetyl-monasfluol B (8), monasfluore B (9), monascin (10) and ankaflavin (11). Compound 2 was isolated in a small amount with some contamination ( Figure S3), so its stereochemistry was not determined. Compound 3 was also possible to be identified as FK17-P2b1 despite the impurity of the sample ( Figure S4). Both compounds 3 and 10 were previously isolated from the yellow mutant strain of M. kaoliang KB9 (Jongrungruangchok et al. 2004).
In addition to the typical bicyclic azaphilones, isolation and structural elucidation of compounds 4-7 by NMR and HRMS techniques were based on a tricyclic azaphilone framework with an angular lactone structure (Tables S4-S7 and Figures S5-S28). Compounds 8 and 9 were identified solely by LC-MS and UV absorption profile comparison with compounds 5 and 6, respectively ( Figures S29-S30). The structural frameworks of the angular tricyclic azaphilones differ from those of classical pigments (compounds 10 and 11) (Tables S8-S9 and Figures S31-S41) due to a different pattern of Knoevenagel cyclization between the a-carbon (C14) of the b-ketoacid ester and the carbonyl group on the bicyclic azaphilone core at C6 for the linear and C8 for the angular tricyclic frameworks (Chen et al. 2017). Compounds 4-6, containing the same number of carbons on the acyl chain (R ¼ C 5 H 11 ), are shunt products from the biosynthesis of compound 10 due to C8-C14 Knoevenagel condensation. Similarly, compounds 7-9, containing a longer acyl chain (R ¼ C 7 H 15 ), are shunt products of compound 11 biosynthesis. To date, no biological activity has been reported for both compounds 4 and 7. Compound 5, the acetylated derivative of compound 4 with moderate nitric oxide (NO) inhibitory activity, was first isolated from Monascus purpureus BCRC 38108 and named monascusazaphilone C (Wu et al. 2013). Similar to compound 5, compound 8 is the acetylated form of compound 7. Both compounds 5 and 8 are products arising from the disruption of deacetylase in Monascus azaphilone pigment biosynthesis (Chen et al. 2017). Elimination of acetic acid to generate the propenyl group at C3, followed by C8-C14 Knoevenagel condensation, results in compounds 6 and 9, named monasfluores A and B, respectively. Both compounds were first isolated from Monascus AS3.4444, and they exhibit intense blue fluorescence (Huang et al. 2008) and an inhibitory effect on NO production (Hsu et al. 2010).

Solvent effects on keto and enol forms of azaphilones
NMR data of isolated azaphilones (compounds 4-7 and 10-11) were first collected in CD 3 OD (Table S4-9). Based on the 1 H NMR spectra, the 1,3-dicarbonyl group of all compounds appeared to be in the enol form. The signals of the H8 of compounds 4-7 and the H6 of compounds 10-11 appeared to be singlet ( Figure S42 and Tables S4-S7) and doublet of doublets ( Figure S43 and Tables S8-S9). This is due to the lack of a neighbouring proton at C14, which results from enol formation. The relative stereochemistry between the H8 and the methyl H9 of compounds 4-7 was also determined by a 1 D selective gradient NOESY experiment (Figures S9, S15, S21 and S27), and the results showed that both the H8 and the methyl group are in the cis position. The absolute configuration of these compounds was assumed to be 7S, and 8S, based on their biosynthetic origin and results from NOE correlations (Balakrishnan et al. 2014;Bijinu et al. 2014). For compounds 10 and 11, no methyl signal of H9 was observed in 1 D selective gradient NOESY spectra when the H6 was irradiated and vice versa ( Figures S35 and S41). This indicates that both protons are in the trans position, and their stereochemistry was assumed to be 6 R and 7 R based on the previous report (Chen et al. 1971). However, the absolute configuration of all isolated compounds was further investigated using experimental and computational ECD analysis, as described in the next section.
The effect of different deuterated solvents towards the NMR chemical shifts has been previously reported (Venditti et al. 2019) therefore compounds 4-7 and 10 were further investigated using an aprotic polar solvent, DMSO-d 6 . The doublet signal, resulting from the coupling between H8 and H14 of compounds 4-7 (Figures S10, S16, S22 and S28), and the doublet signal of H14 arising from coupling with H6 of compounds 10 ( Figure S36), could be observed. The doublet signals of H14 from the vicinal coupling with H8 in compounds 4-7 and H6 in compound 10 indicate the keto form of the 1,3-dicarbonyl group. The 13 C and 2 D NMR data of compound 7 in DMSO-d 6 were collected ( Figures S44-S47). The 13 C chemical shifts in both CD 3 OD and DMSO-d 6 were also compared, and most of the carbon signals showed similar chemical shifts ( Figure S44). However, in CD 3 OD, the signal of C14 could only be observed from the correlation with H8 in the HMBC spectrum ( Figure S26) but not directly from the 13 C experiment ( Figure S23). Only the C14 signal of monascin (10) could be observed in the 13 C NMR spectrum ( Figure S31). On the contrary, in DMSO-d 6 , the C14 signal appeared as a methine carbon based on the HSQC spectrum ( Figure S46) and could be observed in the 13 C experiment due to the NOE enhancement ( Figure S44). The unusual chemical shift of C14 in CD 3 OD at around 57 ppm is still unknown, but the assignment based on the HMBC spectrum and indirect evidence from the 13 C experiment confirmed the assignment of this carbon. These results strengthen the existence of the 1,3-dicarbonyl group of compound 7 in DMSO-d 6 . The results from the 1 D selective gradient NOESY experiment also revealed that the vicinal protons, H8 and H14, in compounds 4-7 (Figures S10, S16, S22, and S28) and H6 and H14 in compound 10 ( Figure S36) are in cis and trans configurations, respectively. The same splitting pattern could also be observed when CDCl 3 was used as an NMR solvent (Akihisa et al. 2005;Huang et al. 2008;Wu et al. 2013;Balakrishnan et al. 2014). Although the chemical shifts of compounds 4-11 were previously reported in CD 3 OD (Mart ınkov a et al. 1999; Chen et al. 2017), the observed signals appeared to be the same as our results. The structure of the 1,3-dicarbonyl group was reported as the keto form.
Based on our results, it is feasible that the protic solvent could promote the enol form by providing a proton source to the oxygen atom of the carbonyl group, unlike DMSO or CHCl 3 . Interestingly, no signal of the keto form was observed in CD 3 OD, indicating only the enol form. Furthermore, two possible carbons of the 1,3-dicarbonyl groups could be involved in keto-enol tautomerization that is, the carbonyl ester (C13) and the carbonyl ketone (C15). This led us to perform simulations of the enol forms at different carbonyl carbons using compound 7 to represent the azaphilones isolated from this work. The results showed that the enol form of the C15 is more favoured than that of the C13 based on its lower electronic energy structure. This may be explained by the high electron density around the carbonyl ester. The solvent effect on the tricyclic azaphilone structures observed in this work was first investigated and revealed the solvent-dependent keto-enol formation of the tricyclic azaphilones.

Stereochemistry of azaphilones in an enol form
Despite a large number of reports on azaphilones, the absolute configuration of some azaphilone derivatives is still unidentified, as in compounds 5 and 8 (Wu et al. 2013;Chen et al. 2017), or only inferred based on the biosynthetic origin and NOE correlations, as in compounds 4, 6, 7, and 9 (Balakrishnan et al. 2014;Bijinu et al. 2014). Also, the enol forms of these azaphilones in solution have never been investigated. Identification of their solution structures is crucial as it could impact their physicochemical properties and biological activities (Katritzky et al. 2010). Therefore, this work determined the absolute configuration of all isolated azaphilones in methanol based on NMR and ECD data combined with computational methods. Although the enol form of the carbonyl ketone (C15) of compound 7 was proposed to be more favoured, there are still two possible geometrical isomers, 14E and 14Z, that could be formed. Initially, conformational analysis of compound 7 was performed using the molecular mechanics method implemented in Avogadro. The stereochemistry of the initial structure was based on the 1 D selective gradient NOESY data, and the geometry of the enol form was then optimised. The lowest energy structure was the 14E isomer. Therefore, this isomer was further optimised using the time-dependent density functional theory (TD-DFT) with B3LYP hybrid functional and 6-31 G (d,p) basis set implemented in Gaussian09W to simulate the theoretical ECD spectrum. The results showed that the calculated ECD spectrum of 7S, 8S, 11 R, 14E-7 displayed positive and negative cotton effects similar to the experimental spectrum with a slight red-shift of the peaks ( Figure S48D). The TD-DFT calculations were also performed to generate the theoretical ECD spectra of the other isolated compounds (Figures S48-S49). The combined results from simulated ECD spectra and the 1 D selective gradient NOESY experiments in methanol also led us to propose the absolute configuration of compounds 4-6 as 7S, 8S, and 14E with an additional stereocentre at 11 R for compounds 4-5 and 7, and 6 R, 7 R and 14E for compounds 10-11. The more favoured 14E isomer of azaphilones observed in this work could be explained by the reduced steric interaction between the long alkyl chain (-C 5 H 11 or -C 7 H 15 ) and the azaphilone core structure compared to the Z isomer ( Figure S50). In addition, the 1 D selective gradient NOESY data in DMSOd 6 were also collected, allowing us to identify the absolute configuration of isolated azaphilones as 7S, 8S, and 14 R for compounds 4-7 with the additional stereochemistry of 11 R for compounds 4, 5, and 7 and 6 R, 7 R, and 14S for compound 10.
This was the first time that the absolute configuration of compounds 4-7 was determined. Moreover, the assignment was in agreement with the previous studies in which the stereochemistry was based on the biosynthetic pathway and NOE correlations (Balakrishnan et al. 2014;Bijinu et al. 2014). Collectively, our results provide absolute configuration, one of the most crucial pieces of structural information, of tricyclic azaphilones.

Antibacterial and antifungal activities
Among several reports on compounds isolated and identified from M. kaoliang KB9, there is no report on antibacterial and antifungal activities of compounds 1 and 4-7.
To the best of our knowledge, there are only a few studies on antimicrobial activities of compounds 10 and 11 (Mart ınkov a et al. 1999;Cheng et al. 2011). In this work, all isolated compounds (1, 4-7, and 10-11) were evaluated for their antimicrobial activities against gram-negative bacteria (Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853), gram-positive bacteria (Staphylococcus aureus ATCC 25923, Bacillus subtilis ATCC 6633 and Micrococcus luteus ATCC 9341), and a fungus Candida albicans ATCC 10231 using the disc diffusion method. After incubation, no inhibition zones could be observed on the plates containing tested compounds at a concentration of 1 mg/mL. Our test results for compounds 10 and 11 against B. subtilis and C. albicans are different from previously observed, possibly due to differences in the strains used for testing. Based on antimicrobial activity results, both angular tricyclic azaphilone and yellow pigments lack the potential to be antimicrobial agents compared to the orange pigments, which exhibit antimicrobial activities against several species of bacteria, yeasts, and filamentous fungi (Mart ınkov a et al. 1995).

Experimental
Experimental information is provided in Supplementary Material.

Conclusions
Several azaphilones (compounds 2-11) and their early intermediate (compound 1) were isolated and identified from M. kaoliang KB9. The effect of solvents on the azaphilone structures was explored, and their absolute configurations were determined. The tricyclic azaphilones studied in this work existed in the enol form in methanol, and the absolute configuration of compounds 4-7 was identified for the first time. In addition, compounds 1, 4-7, and 10-11 were also tested for their antimicrobial activities. None of them were active against gram-negative and gram-positive bacteria, as well as a fungus, indicating their lack of potential to be antimicrobial agents. Our findings provide essential structural information on azaphilones that could be further used to study their functions and biological activities.

Disclosure statement
No potential conflict of interest was reported by the authors.