Rare Chromones from a Fungal Mutant of the Marine-Derived Penicillium purpurogenum G59

Three new and rare chromones, named epiremisporine B (2), epiremisporine B1 (3) and isoconiochaetone C (4), along with three known remisporine B (1), coniochaetone A (5) and methyl 8-hydroxy-6-methyl-9-oxo-9H-xanthene-1-carboxylate (6) were isolated from a mutant from the diethyl sulfate (DES) mutagenesis of a marine-derived Penicillium purpurogenum G59. The structures of 2–4 including the absolute configurations were determined by spectroscopic methods, especially by NMR analysis and electronic circular dichroism (ECD) experiments in conjunction with calculations. The absolute configuration of the known remisporine B (1) was determined for the first time. Compounds 2 and 3 have a rare feature that has only been reported in one example so far. The compounds 1–6 were evaluated for their cytotoxicity against several human cancer cell lines. The present work explored the great potential of our previous DES mutagenesis strategy for activating silent fungal pathways, which has accelerated the discovery of new bioactive compounds.

AD-1-2 is a mutant strain from the DES mutagenesis [28] of a marine-derived fungus, Penicillium purpurogenum G59 [32]. Previously, we reported three new unusual C25 steroids from AD-1-2 [30]. Here, we report three new chromone derivatives, epiremisporine B (2), epiremisporine B1 (3) and isoconiochaetone C (4), together with the three known ones, remisporine B [10] (1), coniochaetone A (5) [1,2] and methyl 8-hydroxy-6-methyl-9-oxo-9H-xanthene-1-carboxylate (6) [33] (Figure 1), from the same mutant AD-1-2. All of the six compounds were not presented in the parent G59 extract. The absolute configuration of 1 was determined by electronic circular dichroism (ECD) experiments in conjunction with calculations, for the first time in the present study. Compounds 2 and 3 are new CPC dimers with a rare feature that has only been reported in one case (1) as far as we know.   13 C NMR spectra of 1 in DMSO-d6 showed signals arising from nearly a single major isomer (Table 1), which were confirmed to be identical with those of remisporine B [10]. Eventually, 1 was identified as remisporine B by the NOEs detected between H-4/H-3, H-3/Hα-4′, Hα-4′/H-3′, and H-3′/HO-2′ by the 1D ROESY experiments in DMSO-d6 and the CD spectrum ( Figure 2) which is identical to that of remisporine B [10]. The 1 H NMR spectrum of 1 in CDCl3 showed two sets of 1 H signals from the isomers 1a and 1b in 1:0.8 ratio (Table 2), which were analyzed by comparison with the data in DMSO-d6 (Table 1). Significant differences in the 1 H NMR data (Table 2) between 1a and 1b appeared mainly on the protons around C-2′ on the aliphatic rings, H-3, H-3′, H-4, H-4′, and HO-2′, indicating that they are epimers at C-2′. Careful examination of the 1 H NMR spectrum of 1 in DMSO-d6 (in the Supplementary Information) has shown that except for the major set of 1 H signals from 1a (Table 1), additional very weak signals from H-4, HO-2′, HO-11, and H-11′ of the minor isomer 1b were also clearly recognized in the spectrum, although the other proton signals for 1b were hardly assigned because of the signal broadening or overlapping. The ratio of 1a to 1b was determined approximately to be 1:0.2 by the standard integrals of related proton signals. Similarly, in CD3OD, 1 also existed as a pair of 1a and 1b in the same 1:0.2 ratio. These observations indicated that 1 exists in a dynamic isomerism between 1a and 1b in the solutions.

Identification and Absolute Configuration Determination of Remisporine B (1)
For the major isomer 1a which should dominate the CD curve of 1, there are two possible absolute configurations, 2S3R4S2′S3′R (1a) and its enantiomer 2R3S4R2′R3′S (ent-1a). We performed ECD calculations on 1a and ent-1a. Time dependent density functional theory (TDDFT) ECD calculations performed at the B3LYP/6-31+G(d) level were used to generate ECD spectra for a set of five lowest-energy conformers for each of 1a and ent-1a. The resulting ECD spectra were combined by Boltzmann weighting to give a composite spectrum. By contrast with the enantiomer ent-1a, the calculated ECD spectrum of the enantiomer 1a matched well the experimental data of 1 ( Figure 2). This thus enabled us to assign the 2S3R4S2′S3′R absolute configuration to the major isomer 1a. Accordingly, the minor isomer 1b was reasonably assigned the 2S3R4S2′R3′R absolute configuration.    Remisporine B (1) existed in dynamic isomerism between major isomer 1a and minor isomer 2b in CDCl 3 , and the signals of 1a and 1b were assigned by comparison with the data of 1a in DMSO, given in Table 1. The ratio of 1a and 1b was determined approximately to be 1:0.8 by the standard integrals of their H-4 signals.

Structure Determination of New Chromones 2-4
Epiremisporine B (2) had the molecular formula C30H24O12 by HRESIMS, the same as 1, and the IR absorptions indicated the ester (1742 cm −1 ) and conjugated (1655 cm −1 ) carbonyls in 2. The similar UV and 1D NMR data (Table 1) of 2 and 1 indicated their similar structures. Differing from 1, however, 2 gave both 1 H and 13 C NMR signals (Table 1) as pairs in a ratio of 1:0.8 in DMSO-d6, indicating that 2 is a stereoisomer of 1 and exists in a dynamic isomerism between major (2a) and minor (2b) isomers in DMSO-d6. The planar structure of 2, the same as 1, was deduced by 1D (Table 1) and 2D NMR data (Table S1). Finally, 2 was determined to be an epimer of 1 at C-3′ by the 1D GOESY experiments. The orientation of H-3′, H-3 and H-4 in the same spatial direction on the same side of the ring system in 2 was established by the NOEs detected between H-3′/H-3, H-3′/H-4 and H-3/H-4 of 2a and 2b by the 1D GOESY experiments. Further in the 1D GOESY experiments, a more remarkable NOE was detected on 2′-OH in 2b than in 2a by irradiating H-3′ in 2a and 2b, respectively. This thus evidenced that 2b is an isomer with cis H-3′/2′-OH and 2a is another one with trans H-3′/2′-OH.
Epiremisporine B1 (3) was assigned the molecular formula C31H26O12 by HRESIMS, which had one more CH2 than 2. It showed UV and IR absorptions similar to 2, and the 1 H and 13 C NMR spectra of 3 in DMSO-d6 and CDCl3 gave both pairs of signals in a ratio of 1:0.8, indicating the presence of dynamic isomerism in the solutions between major 3a and minor 3b isomers, like 2. The 1 H and 13 C NMR data of 3a and 3b in DMSO-d6 (Table 1) and CDCl3 (Table 3) are almost identical with those of 2a and 2b in DMSO-d6 (Table 1) and CDCl3 (Table 3), respectively, except for the signals from an ethoxy group in 3a and 3b instead of the signals of the 16′ methoxy group in 2a and 2b. Thus, 3 must be a derivative of 2, with an ethoxy group instead of the 16′ methoxy group in 2. The NOEs detected between H-4/H-3′, H-3′/H-3 and H-3/H-4 in the NOESY spectrum of 3 in DMSO-d6, especially the NOE between H-4/H-3′, further evidenced the relative stereochemistry of 3, the same as 2. The absolute configuration of 2 and 3 was determined on the basis of experimental and theoretically ECD data. Because 2a/2b and 3a/3b exist in nearly equal proportion (1:0.8) in the solutions of 2 and 3, there are four questionable absolute configurations 2S3R4S2′S3′S (2a/3a) and 2S3R4S2′R3′S (2b/3b) and their enantiomers 2R3S4R2′R3′R (ent-2a/ent-3a) and 2R3S4R2′S3′R (ent-2b/ent-3b) for 2 and 3. We performed theoretical ECD calculations on all four questionable absolute configurations for 2 and 3 because it was unknown which one of them mainly dominates the CD of 2 and 3, although the chirality of C-2′ would little affect the CD in view of its location far away from the aromatic ring systems. TDDFT calculations performed at the B3LYP/6-31+G(d) level for 2 and the B3LYP/6-31G(d) level for 3 were used to generate ECD spectra for a set of lowest-energy conformers for each absolute configuration, and the resulting spectra were combined by Boltzmann weighting to give a composite spectrum. The calculated ECD spectra of the enantiomers 2a and 2b for 2 and 3a and 3b for 3 properly reproduced the experimental data. Especially the calculated ECD spectra of the enantiomers 2b and 3b agreed well with the experimental data of 2 and 3 (Figure 3), indicating that the enantiomers 2b and 3b likely dominated the CD of 2 and 3, respectively. Thus, the 2S3R4S2′S3′S and 2S3R4S2′R3′S absolute configurations were assigned to the isomers 2a/3a and 2b/3b, respectively.
Isoconiochaetone C (4), colorless needles (MeOH), m.p. 99-100 °C, [α] 25 D +76.7 (c 0.16, MeOH), was assigned the molecular formula C14H14O4 by HRESIMS. The molecular formula and the typical UV absorptions indicated that 4 is a monomeric chromone. The 1 H and 13 C NMR data of 4 in CDCl3 (Table 2) matched well the data of coniochaetone C in CDCl3 [3] although splitting patters of H-1 and H2-2 signals in the literature [3] could not be fully analyzed probably because of the limitation in the spectral resolution, indicating the same planar structure of both compounds. Coniochaetone

Detection of 1-6 in the Mutant AD-1-2 Extract by HPLC-PDAD-UV/HPLC-ESI-MS Analyses
Each of the EtOAc extracts of mutant AD-1-2 and the parent G59 strain was analyzed by HPLC-photodiode array detector (PDAD)-UV and HPLC-electron spray ionization (ESI)-MS. In the HPLC-PDAD-UV analysis, 1-6 were detected in the mutant but not in the G59 extract, which was confirmed by both retention times and UV spectra in comparison to those of standard compounds 1-6 ( Figure S1). In the HPLC-PDAD-UV profile, compound 3 was only detected in a trace at 242 nm ( Figure S1), by contrast, it was clearly detected by HPLC-SEI-MS. The other five compounds were also detected in the mutant extract by selective pseudo-molecular ion monitoring with both extracted ion chromatograms and MS spectra, but not in the G59 extract ( Figure S2).

Mechanism of 1-3 Formation
According to the mechanism of 1 formation from remisporine A (7) [10], 2 and 3 were expected to be formed by dimerization of 7 and 8 ( Figure 4). Indeed, both 7 and 8 were presumably detected in the mutant but not the G59 extract in the HPLC-PDAD-UV and HPLC-ESI-MS analyses (Figures S1 and S2). However, 7 and 8 could not be obtained in pure form because of their instability [10]. Thus, we proposed the mechanism for the 1-3 formation, as shown in Figure 4, which involved the formation of C-3′ isomers, extending the original proposal of Kong et al. [10]. The spontaneous Diels-Alder reaction of two molecules of 7 or of combination of 7 and 8 followed by the retro-Aldol condensation and aromatization could produce I. The intermediate I would further undergo keto-enol tautomerism with II as an intermediate to give III bearing an inverted absolute configuration of C-3′. The intramolecular cyclization of I via attacking the keto carbonyl at C-2′ by the hydroxyl at C-2 could produce 1 and X, existing as interconvertible isomers in solutions. A similar process could be applied to III to generate 2 and 3. Incidentally, according to the mechanism that requires the endo mode of the Diels-Alder reaction of 7 and 8, adopting the less hindered orientation with hydroxyl groups facing each other rather than the bulky methoxy carbonyls [10], the absolute configuration of 7 and 8 both should be S.

Discussion
In a continuation of our previous work on the C25 steroids newly produced in mutant AD-1-2 [30], further chromatography of the AD-1-2 extract, tracing newly produced chromones by bioassays and chemical analyses, resulted in the isolation of six choromone derivatives 1-6, including the three new and rare chromones 2-4. Two rare CPC monomers 7 and 8 were also found both to be produced in the mutant AD-1-2 by HPLC-PDAD-UV and HPLC-ESI-MS analyses. In contrast, none of 1-8 was detected in the parent G59 extract by the HPLC-PDAD-UV/HPLC-ESI-MS analyses. The proposed mechanism (Figure 4) according to the present results, extending the original proposal of Kong et al. [10], explored well the formation of 1-3 from 7 and 8, including the conversion of the absolute configuration at C-3′ and the isomerism of 1-3 between the isomers at C-2′. Compounds 7 and 8 were detected in the ethyl acetate (EtOAc) extract of the mutant AD-1-2. We used MeOH and EtOH in the separation, but not in the extraction of AD-1-2 cultures to obtain the EtOAc extract. This thus excluded the doubt whether the methyl or ethyl ester in 7 and 8 were formed by esterification during the extraction. These results proved that the production of 4-8 in mutant AD-1-2 was caused by the activation of silent pathways in G59 strain by DES mutagenesis, although biological details for regulating the activation are still unknown, while 1-3 were formed by subsequently occurred spontaneous dimerization of the produced 7 and 8 via Diels-Alder reaction (Figure 4).
During NMR experiments in the present study, approximately 70% of 1 was converted into 2 after seven days at room temperature in DMSO-d6, evidenced by the appearance of a pair of the 1 H NMR signals (2a/2b in 1:0.8 ratio), but no more than 70% of 1 was further converted into 2 even after 15 days. The conversion of 2 into 1 was not observed in DMSO-d6 solution. In the NMR experiments for crude X, which showed UV absorptions similar to 1-3 and had the molecular size (m/z 591 [M + H] + in ESI-MS) the same as 3, over 90% of X was converted into 3 within two days in DMSO-d6 at room temperature, and further purification of the samples provided an additional amount of 3 but X could not be obtained. Further, in the 1 H NMR spectra of 3 in DMSO-d6 and CDCl3, very weak signals from X were detected together with some other also very weak signals from III (R=CH2CH3, in Figure 4), as seen in the 1 H NMR spectra in the Supplementary Information. These weak signals always coexisted with the signals of 3 even after the samples for the NMR experiment were repeatedly purified by HPLC, indicating the presence of the conversion between X and 3 in the solutions. These observations further supported the mechanism in Figure 4, which involved the keto-enol tautomerism that enables both conversions of 1/2 and 3/X.

MTT Assay
EtOAc extracts and fractions were dissolved in MeOH at 10 mg/mL, and the MeOH solutions were used in MTT assays. Compounds 1-6 and DOC were dissolved in MeOH to prepare 10.0 mg/mL stock solutions, respectively, and serial dilutions for compounds 1-3 were made for MTT assay. DOC was used as positive control, and MeOH was used as blank control.

Parent Fungal Strain and Its Mutant AD-1-2 the 1-8 Producing Strain
The parent strain Penicillium purpurogenum G59 was isolated from a soil sample collected at the tideland of Bohai Bay around Lüjühe in Tanggu district of Tianjin, China, in September 2004 [32], and was identified by Liang-Dong Guo, the Institute of Microbiology, Chinese Academy of Sciences, China. This strain was deposited at the China General Microbiological Culture Collection Center under the accession number CGMCC No. 9721.
AD-1-2 is a bioactive mutant obtained by DES mutagenesis of the G59 strain through treatment of G59 spores with 0.5% (v/v) DES in 50% (v/v) DMSO at 4 °C for 1 day [28]. The mutant AD-1-2 was deposited at the China General Microbiological Culture Collection Center under the accession number CGMCC No.8634.

Fermentation, Extraction, and Preparation of Targeted Bioactive Fraction
By a large-scale fermentation (72 L) and extraction, we had previously obtained an EtOAc extract (37.3 g) of the mutant AD-1-2, showing inhibitory effect on K562 cells with an IR% of 58.6% at 100 μg/mL [30], and a vacuum liquid chromatography of the EtOAc extract (37.2 g) on a silica gel column (bed 7.5 cm × 20 cm, silica gel 300 g) eluted by b.p. 60-90 °C petroleum ether→dichloromethane (D)-methanol (M) 100:0→0:100 had afforded a fraction Fr-3 (7.0 g, eluted by DM 98:2 → 96:4) [30]. Fr-3 contained targeted chromone derivatives and inhibited K562 cells with an IR% of 53.1% at 100 μg/mL. Thus, Fr-3 was further separated in the present study to isolate 1-6, and the EtOAc extract of the mutant AD-1-2 was used in HPLC-PDAD-UV and HPLC-ESI-MS analyses to detect 1-8. By fermentation and extraction of the parent G59 strain at the same time with the same conditions of the mutant AD-1-2, we had also previously obtained an EtOAc extract (610 mg) of the G59 strain, which did not inhibit the K562 cells (an IR% of 5.6% at 100 μg/mL) [30]. This extract was used for tracing newly produced 1-6 in the mutant AD-1-2 extract in the separation of Fr-3 and also in the HPLC-PDAD-UV and HPLC-ESI-MS analyses for detecting 1-8, all as negative controls, in the following experiment.

Physicochemical and Spectroscopic
(for 3b and ent-3b) of the weights for 3, respectively. The calculated spectrum was blue-shifted by 5 nm for 1 and red-shifted by 6 nm for 2 and 3 to facilitate comparison with the experimental data.

Conclusions
A chemical investigation on a mutant AD-1-2 of marine-derived fungus P. purpurogenum G59 resulted in the discovery of three new chromones 2-4 and the isolation of three known chromones 1, 5, and 6. Two rare CPCs 7 and 8 were also found to be produced in the mutant AD-1-2, but none of 1-8 were detected in the G59 extract. The present work further exemplified the effectiveness of our previous DES mutagenesis strategy [28][29][30][31] for activating silent fungal pathways to discover new bioactive compounds.