Three New and Eleven Known Unusual C25 Steroids: Activated Production of Silent Metabolites in a Marine-Derived Fungus by Chemical Mutagenesis Strategy using Diethyl Sulphate

Three new (1–3) and 11 known (4–14) C25 steroids with an unusual bicyclo[4.4.1]A/B ring system were isolated by tracing newly produced metabolites in the EtOAc extract of an antitumor mutant AD-1-2 obtained by the diethyl sulphate (DES) mutagenesis of a marine-derived Penicillium purpurogenum G59. HPLC-PDAD-UV and HPLC-ESI-MS analyses indicated that the G59 strain did not produce these metabolites and the production of 1–14 in the mutant AD-1-2 extract was caused by the activation of silent metabolites in the original G59 strain by DES mutagenesis. The structures of the new compounds, named antineocyclocitrinols A (1) and B (2) and 23-O-methylantineocyclocitrinol (3), including their absolute configurations were determined by various spectroscopic methods, especially the NMR and Mo2-induced CD analyses. Compounds 1–3 provide the first examples of the C25 bicyclo[4.4.1]A/B ring steroids with the Z-configuration of 20,22-double bond. All of 1–14 weakly inhibited several human cancer cell lines to varying extents. These results provided additional examples for the successful application of the chemical mutagenesis strategy using DES to discover new compounds by activating silent metabolites in fungal isolates and supported also the effectiveness and usefulness of this new strategy.


Introduction
The C25 steroids with bicyclo [4.4.1]A/B rings are a family of unusual steroids with 16 known family members [1]. Their structures differed mainly in the side chains at C-17 and could be classified into four subtypes: the cyclocitrinol, isocyclocitrinol, neocyclocitrinol, and precyclocitrinol subtypes. The first member of the family cyclocitrinol was isolated from a terrestrial Penicillium citrinum and reported as a new sesterpene [2]. However, the structure of cyclocitrinol was later revised to a bicyclo [4.4.1]A/B ring steroid with a trans-22,23-double bond, when it was re-isolated from a sponge-derived Penicillium citrinum [3]. The cyclocitrinol subtype steroids include cyclocitrinol, 12-hydroxycyclocitrinol, 20-O-methylcyclocitrinol, 24-epi-cyclocitrinol, 20-O-methyl-24-epi-cyclo citrinol, and 24-oxiocyclocitrinol. The isocyclocitrinol subtype steroids carry a trans-23,24-double bond and consist of isocyclocitrinols A and B, and 22-acetylisocyclocitrinol. Neocyclocitrinol was initially obtained as a mixture of 23,24-epimers with a 20,22-double bond, and the configuration of the 20,22-double bond was not determined [4]. All of four 23,24-diastereomers were isolated later in pure forms, named neocyclocitrinols A-D, respectively, and their absolute configurations were determined, including the E-configurations of the 20,22-double bonds [1]. The neocyclocitrinol subtype steroids involve also two methyl derivatives, erythroand threo-23-O-methylneocyclocitrinols, in addition to neocyclocitrinols A-D. The precyclocitrinol subtype steroid has been known only precyclocitrinol B, which possesses a 20,22-epoxy unit differing from the structures of other subtypes in this class.
Fungal metabolites, especially those derived from marine fungi, are particularly important as rich sources of new drug candidates [5,6] and have continuously attracted considerable attention [5][6][7][8]. As many fungal biosynthetic pathways are silenced in standard culture conditions, activation of the silent pathways may enable access to new metabolites. Indeed, various strategies have been developed to activate silent pathways and elicit metabolite production from fungal isolates [9]. The one strain-many compounds (OSMAC) strategy [10] and chemical epigenetics methodology [11] have been widely applied by microbial chemists to access cryptic secondary metabolites [12], because the culture-based, simple procedures outlined by the strategies are suitable for use by microbial chemists. The ribosome engineering strategy [13] provided also additional simple way to activate silent pathways by introducing drug-resistance mutations in bacteria to discover new antibiotics [14]. We have previously reported a new approach that extended the ribosome engineering strategy to fungi to activate silent metabolites by introducing drug-resistance in Penicillium purpurogenum G59 using dimethyl sulfoxide (DMSO) as accessorial agent [15,16]. Recently, we developed another new, practical strategy for activating silent fungal metabolites using P. purpurogenum G59 and an old, modified method of diethyl sulphate (DES) mutagenesis [17][18][19]. P. purpurogenum G59 is a marine-derived wild type fugal strain initially isolated by our group [20] and originally did not produce any metabolites with antitumor activities in repeated MTT assays using K562 cells [15][16][17][18][19][20]. Using the chemical mutagenesis strategy, we have previously discovered several new antitumor metabolites including some with novel structures by activating silent secondary metabolites in P. purpurogenum G59 [17][18][19].
In a continuation of the above mentioned research work, we have undertaken investigations on the bioactive metabolites from an antitumor mutant AD-1-2 that was obtained by the DES mutagenesis of P. purpurogenum G59 [17]. Both bioassays and HPLC-photodiode array detector (PDAD)-UV and HPLC-electron spray ionization (ESI)-MS analyses indicated that some kinds of secondary metabolites exist in the EtOAc extract of the mutant AD-1-2 but not in the EtOAc extract of the original G59 strain. This has encouraged us to investigate these secondary metabolites. Indeed, we have obtained different types of compounds by tracing the newly produced metabolites in the mutant AD-1-2 extract. In this paper, we report our research results on the C25 steroids with unusual bicyclo[4.4.1]A/B ring system, including the discovery of three new compounds 1-3 in this class and the identification of 11 known steroids 4-14 in the same class.

Fermentation, Isolation of 1-14, and Identification of Known Steroids 4-14
Large-scale fermentation and extraction of the mutant AD-1-2 produced an EtOAc extract that inhibited K562 cells with an inhibition rate (IR%) of 58.6% at 100 µg/mL. However, the EtOAc extract of the control G59 strain, obtained by fermentation at the same time and same conditions, did not inhibit the K562 cells (an IR% of 5.6% at 100 µg/mL). HPLC-PDAD-UV and HPLC-ESI-MS analyses of the mutant AD-1-2 and the control G59 extracts showed that newly produced C25 steroids existed in the mutant extract and were detected within 41-52 min retention times (see Figures S1 and S2 in the Supplementary Information). The separation of a small amount of the mutant AD-1-2 extract by thin layer chromatography (TLC) followed by the HPLC-PDAD-UV analysis of the obtained TLC fractions indicated that the C25 steroids were enriched in two fractions, B3 and B4 (Figures S3 and S4 in the Supplementary Information). Thus, the separation of the mutant EtOAc extract was performed tracing the newly produced C25 steroids under guidance of bioassays and TLC analyses.
A vacuum liquid chromatography (VLC) separation of the mutant extract on a silica gel column by monitoring the C25 steroids gave a fraction containing these C25 steroids. Further separation of the fraction by repeated column chromatography on Sephadex LH-20 and ODS gave two crude C25 steroid fractions ( Figures S5 and S6 in the Supplementary Information). The two fractions were separated by preparative HPLC and semi-preparative HPLC to obtain 1-14 ( Figure 1).
Among the compounds obtained, structures of three new compounds 1-3, including their absolute configurations, were determined by spectroscopic methods and named antineocyclocitrinols A (1) and B (2) and 23-O-methylantineocyclocitrinol (3). On the other hand, 11 known C25 steroids were identified as neocyclocitrinols A (5), B (4), C (7), and D (6), threo-23-O-methylneocyclocitrinol (8), erythro-23-O-methylneocyclocitrinol (9), 24-epi-cyclocitrinol (10), cyclocitrinol (11), 20-O-methyl-24-epi-cyclocitrinol (12), 20-O-methylcyclocitrinol (13), and isocyclocitrinol B (14), respectively, by comparison of their physicochemical and spectroscopic data with those reported in the literature [1].   [4]. In the 1 H and 13 C NMR spectra (the NMR spectra in the Supplementary Information), 1 and 2 showed 1 H and 13 C NMR signals that closely resembled the signals from the known steroids 4-7, except several proton and carbon signals from the structural parts around the side chains at C-17 in 1 and 2 differed slightly (Tables 1 and 2). These NMR data indicated that the planar structures of 1 and 2 are the same as 4-7, but the stereo structures of the side chain moieties are slightly different. This was confirmed by detailed analyses of their DEPT, 1 H-1 H COSY, HMQC, and HMBC spectra (see Table S1 for 1 and Table S2 for 2 in the Supplementary Information) to establish their planar structures ( Figure 2). The stereochemistry of 1 and 2 including their absolute configurations were determined as follows.     (3) was obtained as a mixture of the 23-O-methyl derivatives of 1 (24R) and 2 (24S) in an approximate 4:1 ratio of the 24R and 24S forms.

Structure Determination of New Compounds 1-3
Signals of the C-20, 21, 23 and 25 carbons appeared as a pair of separated signals in the 13 C NMR spectrum and their data are given as the former/latter signals for the 24R and 24S forms, respectively.
The NOEs between pairs of the protons, H 3 -21/H 3 -19, H 3 -19/Hα-18 and H-9/H-14, indicated the orientations of C-20, C-19 and C-18 on the same side of the A-D ring system and H-9, H-14 and H-17 on the other side of the ring system in 1 and 2. Inspection of the HGS molecular modeling indicated that H 3 -19 and Hα-18 are in an interatomic distance to show NOE interactions, while Hβ-18 points away from H 3 -19. So, the NOE interaction from one of the H 2 -18 protons with the H 3 -19 protons could be ascribed to Hα-18, though the signals from Hα-18 and Hβ-18 are overlapped. Further, the 1 H and 13 C NMR signals of the A/B ring moieties in 1 and 2 are almost identical with the signals from the same moieties in 4-7 (Tables 1 and 2), indicating the 3β-OH in 1 and 2 are the same as 4-7. Absolute configurations of the A-D ring moieties in 1 and 2 were determined to be the same as 4 by their circular dichroism (CD) spectra that closely resembled the CD spectrum of 4 as shown in Figure 3  The 1 H and 13 C NMR data of 1 and 2 in DMSO-d 6 (Tables 1 and 2) are almost identical except for the signals from CH-23 and CH-24. These NMR data indicated that 1 and 2 should be stereoisomers at C-23 and/or C-24. Further, the couplings of H-23/H-24 (6.0 Hz for 1 and 4.1 Hz for 2) indicated that 1 was a 23,24-threo-isomer and 2 was a 23,24-erythro-isomer. The H-23/H-24 couplings were larger than 6 Hz in threo-isomers but smaller than 5 Hz in erythro-isomers [1,21]. Further support was also from the resonances of H-23 and H-24 in 1 which was upfield than 2. The same protons in the threo-isomers had chemical shifts upfield than in the erythro-isomers [1]. The same configurations at C-23 in 1 and 2 were evident because the chemical shifts of C-21 and C-23 in 1 and 2 were consistent [1]. On the other hand, the 23,24-threo-diol steroids 4 and 5 gave their C-25 signals at around δ C 19.0, but the 23,24-erythro-diol steroids 6 and 7 at around δ C 18.4 (Table 2). By comparison of the 13 C signals of 4-7 (Table 2) between threo/erythro pairs of C-24 epimers, it appears that the difference of C-25 δ values (∆δ C25 ) between 4/6 (∆δ C25 = 19. 23,24-threo-diols 4 (23S,24S; δ C25 19.1) and 5 (23R,24R; δ C25 18.9) or of the 23,24-erythro-diol 2 (23S,24R; δ C25 18.5) with 23,24-erythro-diols 6 (23S,24R; δ C25 18.3) and 7 (23R,24S; δ C25 18.5) also supported the same conclusion, the 23S,24S for 1 and the 23S,24R for 2. The absolute configurations 23S,24S for 1 and 23S,24R for 2 could be also assigned by the dimolybdenum induced CD (ICD) analysis. In the ICD measurements by the Snatzke's method using dimolybdenum tetraacetate [Mo 2 (OAc) 4 ] [22,23], the Mo 2 -complex of 1 in DMSO gave positive Cotton effects around 310 nm (band IV) and near 400 nm (band II), while the Mo 2 -complex of 2 gave negative bands II and IV ( Figure 4). By the helicity rule of the Snatzke's method, the sign of the torsional angle determines the signs of particular Cotton effects [22,23], that is, the negative signs of the bands II and IV were determined by the    [1]. The IR spectrum also showed the absorptions due to OH (3374 cm −1 ) and CH 3 /CH 2 (2946, 2910, 2878 and 1378 cm −1 ) groups. The 1 H and 13 C NMR spectra of 3 in DMSO-d 6 gave 1 H and 13 C signals (Tables 2 and 3 (Tables 1-3). The absolute configuration of the A-D ring system in 3 was established to be the same as 1-2 and 4 by the closely resembled CD spectra of 1-4 as shown in Figure 3. The couplings of H-23 and H-24 in 3, 5.9 Hz for major isomer and 4.0 Hz for minor isomer, indicated that the major isomer in 3 was the 23,24-erythro and the minor isomer was the 23,24-thero [1,21]. This was supported also by the downfield resonances of H-23 and H-24 in the major erythro isomer than in the minor thero isomer (see Table 3) [1]. Further, the chemical shifts of C-21 and C-23 in the major and minor isomers were consistent ( Table 2) and indicated the same configuration at C-23 [1]. Thus, 3 was the 4:1 mixture of 23-methoxylated 2 and 1.

Absolute Configuration Assignment of vic-Diols in 4-7 and 14 by ICD Analysis
Absolute configuration assignments of the acyclic and thus flexible 1,2-diols have now been able to use the dimolybdenum tetracetate-induced CD method for various kinds of 1,2-diol groups [23,24].  [23]. This thus enabled extending the empirical -helicity rule‖ to the erythro-1,2-diols. In other words, the largest R or R' group determines the preferred conformation in the formed Mo 2 -complex and does itself points away from the rest of the complex to avoid the steric interaction with the remaining acetate ligands in the stock complex [23]. As a matter of fact, the steric interaction of bulky group plays a primary role and dominates the formation of preferred conformation. Therefore, discrimination of the stronger steric interaction from R and R' in erythro-1,2-diols is most important for the conformational analysis prior to the use of the empirical -helicity rule‖. However, the stronger steric interaction from two groups was not always easily discriminated only according to the bulkiness of two groups. Sometimes, the difference in the bulkiness of the two groups is too small and additional lager group also affected the steric interaction of the two groups. The same is true of the case of erythro-23,24-diols 2, 5 and 6. In this case, the C-25 methyl group is spatially bulkier to the 23,24-diols than the C-22 sp 2 methine, however, a larger A-D ring moiety linked to C-22 has made it difficult to estimate which is stronger in steric interaction. This has brought about doubts in the application of the -helicity rule‖ to these erythro-23,24-diols. This problem was solved by following ICD investigations on 4-7 and 14 using the Mo 2 (OAc) 4 -induced CD method.

Inhibitory Effects of 1-14 on Several Human Cancer Cell Lines
Antitumor activities of 1-14 were preliminarily tested by the MTT method using the human cancer K562, HL-60, HeLa, and BGC-823 cell lines. All of 1-14 were assayed for their effects on each of the four human cancer cell lines.

Discussions
By tracing newly produced metabolites in the mutant AD-1-2 extract, we have isolated 14 C25 steroids with unusual bicyclo[4.4.1]A/B rings, including three new (1-3) and 11 known ones (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14). HPLC-PDAD-UV and HPLC-ESI-MS analyses indicated that the original G59 strain did not produce these metabolites and the production of 1-14 in the mutant AD-1-2 extract was caused by the activation of silent metabolites in original G59 strain by DES mutagenesis.  [17,19], and supported also the effectiveness and usefulness of the new strategy [17]. Relating to the absolute configuration assignments of new erythro-23,24-diol 2 and threo-23,24-diol 1, we investigated the ICDs of the known erythro-23,24-diols 5 and 6 and threo-vic-diols 4, 7 and 14 by the Mo 2 -induced CD method [23,24]. The present results indicated that the empirical -helicity rule‖ could be straightforwardly applied to the threo-vic-diols 1, 4, 7 and 14 to unambiguously assign their absolute configurations as reported [23]. Although intensities of the band II and IV Cotton effects were rather lower in the ICD of 14, it could be explained by the strong steric effects of three bulky groups around the vic-diol group, leading to the decrease of the Mo 2 -complex formation. On the other hand, the ICD investigations on the erythro-vic-diols 2, 5 and 6 showed that two possible conformations with an antiperiplanar O-C 23  In present MTT assay, 1, 3, 5, 10 and 12 inhibited the K562, HL-60, and/or HeLa cells with the IR% values over 31.4% at 100 µg/mL. However, as a whole, 1-14 showed very weak effects on the tested four human cancer cell lines. This coincides with the previous report on the isocyclocitrinol A and 22-acetylisocyclocitrinol A, which recorded that both compounds were inactive against murine and human tumor cell lines but showed weak antibacterial activity on Staphylococcus epidermidis and Enterococcus durans [3]. Similar antibacterial activity was also recorded for cyclocitrinol (11) in the literature [2] (the structure reported for the cyclocitrinol in this paper [2] was later revised to 11 [3]). On the other hand, 4, 7, 8, 10, 11 have been shown to induce the receptor-dependent production of cAMP in GPR12-transfected CHO cells at 10 µM [1]. However, the 16 known C25 steroids in this class of compounds have not been identified for their more promising bioactivities so far. Nevertheless, the very unusual skeletal structures of these steroids warrant further evaluation of biological activity.

MTT Assay
The EtOAc extracts, fractions, pure compounds and DOC in MeOH at 10 mg/mL were used in MTT assays. DOC was used as positive control and MeOH was used as blank control.
Exponentially growing K562, HL-60, HeLa and BGC-823 cells were suspended in fresh RPMI-1640 medium containing 10% fetal bovine serum and 100 μg/mL penicillin and streptomycin at the cell density of 5 × 10 4 cells/mL and seeded into 96-well plates each 200 μL/well. The suspension cells, K562 and HL-60, were incubated at 37 °C for 2 h, whereas the adherent cells, HeLa and BGC-823, were incubated at 37 °C for 12 h. Then, 2 μL of MeOH or DMSO for control and the test sample solutions was added to each well, and the cells were cultured at 37 °C for 24 h. After morphological examination of the cells under an inverted microscope, MTT (20 μL; 5 mg/mL in PBS) was added into each well, incubated at 37 °C for 4 h, and centrifuged at 2000 rpm, 4 °C for 20 min. After removal of the supernatant by aspirating, 150 μL DMSO was added into each well, and shaken for 5 min to dissolve formazan crystals. The optical density (OD) value at 570 nm was read for each well using the VERSAmax-BN03152 plate reader (MD, San Francisco, CA, USA). Each three wells were set for control and test groups, respectively, and the inhibition rate (IR%) was calculated using OD mean values according to the formula, IR% = (OD control − OD sample )/OD control × 100%.

Initial Fungal Strain and its Mutant the 1-14 Producing Strain
The initial fungal strain Penicillium purpurogenum G59 that was used as control strain in the present study was isolated from a soil sample collected at the tideland of Bohai Bay around Lüjühe in Tanggu district of Tianjin, China, in September 2004 [20]. P. purpurogenum G59 was identified by Professor Dr. Liangdong Guo, Institute of Microbiology, Chinese Academy of Sciences, China. This strain has been deposited at the China General Microbiological Culture Collection Center under the accession number CGMCC No.3560. The strain P. purpurogenum G59 did not produce any secondary metabolites with antitumor activities in the repeated MTT assays using K562 cells at the high concentrations of 100 and 1000 µg/mL [15][16][17][18][19][20].
The producing strain used for 1-14 production in the present study is a bioactive mutant AD-1-2 that was obtained by DES mutagenesis of the strain G59 [18]. Fresh G59 spores in 50% (v/v) DMSO were treated with 0.5% (v/v) DES at 4 °C for 1 day, and then single colony isolation on the treated spores, followed by bioassays, afforded the mutant AD-1-2. The EtOAc extract of the mutant AD-1-2 cultures inhibited K562 cells (an IR% of 74.5% at 100 µg/mL), whereas the EtOAc extract of the G59 strain by the fermentation at the same time and same conditions did not inhibit the K562 cells (an IR% of 4.8% at 100 µg/mL) [18]. The mutant AD-1-2 has been deposited at the China General Microbiological Culture Collection Center under the accession number CGMCC No.8634.

Preparation of Spore Suspensions
The mutant AD-1-2 was inoculated onto potato dextrose agar (PDA) plates from a PDA slant stock stored at 4 °C and incubated at 28 °C for 4 days. Fresh spores formed on the PDA plates were harvested and suspended in 60 mL of sterilized, distilled water with several glass beads in a 100 mL cone-shaped flask and scattered well by shaken enough to prepare a crude spore suspension. A 100 μL portion of this crude spore suspension was added into a well of 96-well plates, diluted with water with its OD at 600 nm measured using a VERSAmax-BN03152 plate reader, and the dilution ratio was recorded when the OD value reached 0.35. Then, the remaining whole crude spore suspension was diluted with sterilized, distilled water in the same proportion to obtain a mutant AD-1-2 spore suspension. This mutant spore suspension was used for the producing fermentation in the following experiments.
Similarly, the G59 spore suspension was also prepared in the same manner as mentioned above using fresh spores formed by cultivation of the G59 strain at 28 °C for 3 days on PDA plates. This G59 spore suspension was used as control strain in the following experiments.

Fermentation and Extraction
Aliquot (200 µL) of the mutant AD-1-2 spore suspension was inoculated into 240 cone-shaped 500 mL flasks containing 300 mL of the liquid medium (glucose 2%, maltose 1%, mannitol 2%, glutamic acid 1%, peptone 0.5% and yeast extract 0.3% in distilled water, adjusted to pH 6.0 prior to sterilization) and fermented at 28 °C for 12 days on rotary shakers at 200 rpm to obtain approximate 72 L of the fermentation broth. The whole broth (72 L) was filtered to separate into a filtrate (70 L) and a mycelial cake. The filtrate was extracted three times with same volumes of EtOAc to give an EtOAc extract (13.3 g) that inhibited K562 cells with an IR% value of 47.9% at the 100 µg/mL. The mycelial cake was extracted three times each time with 10 L acetone-water (2:1) by ultrasonication for 2 h. The aqueous acetone solution obtained by filtration was evaporated under reduced pressure to remove acetone. Then, the remaining water layer was extracted three times with same volumes of EtOAc to give another EtOAc extract (24.0 g) that inhibited the K562 cells with an IR% value of 59.0% at 100 µg /mL. The EtOAc extracts both from the filtrate and mycelia gave same TLC spot profiles and thus were combined to afford total 37.3 g EtOAc extract. This EtOAc extract showed an inhibitory effect on the K562 cells with the IR% value of 58.6% at 100 µg/mL, which was used for the isolation of 1-14 in the following experiments.
On the other hand, each 200 µL of the G59 spore suspension was inoculated into three cone-shaped 500 mL flasks containing 300 mL of the same liquid medium and fermented at the same time and same conditions to obtain 900 mL of the fermentation broth. The whole broth was extracted as described for mutant AD-1-2 to afford an EtOAc extract (610 mg), which did not show inhibitory effect on the K562 cells (an IR% value of 5.6% at 100 µg/mL). This G59 extract was used in the MTT assays and TLC analyses as negative control and also in the HPLC-PDAD-UV and HPLC-ESI-MS analyses for detecting 1-14 in the following experiments.
The data for 5-14 except for the 1 H and 13 C NMR data (Tables 1-3) are in the Supplementary Information.

ICD Measurements for 1, 2, 4-7 and 14 Using Dimolybdenum Tetracetate
The CD data were acquired at room temperature with a 1-mm cell, and step scans were collected at 0.5 nm per step with an integration time of 0.5 seconds over the range 200-500 nm. The ICD spectra were measured for 1, 2, 4-7 and 14 (the molecular weight were all 400 daltons) as follows. First, the 175 µg compound was dissolved in 350 µL DMSO (0.5 µg/µL, 0.00125 M/L) and the CD data were recorded immediately. Second, 350 µg dimolybdenum tetracetate (Mo 2 (OAc) 4 , molecular weight 428 daltons) was dissolved in 350 µL DMSO containing 0.00125 M/L compound (the molar concentration of Mo 2 (OAc) 4 was 0.00234 M/L and the molar ratio of the compound to the Mo 2 (OAc) 4 was 1:1.87). The first ICD data was recorded immediately after dissolving the Mo 2 (OAc) 4 , its time evolution was monitored with a rate of one spectrum every 3 min, until a stationary ICD was reached about 10 min after dissolving the Mo 2 (OAc) 4 , and then the stationary ICD data were taken. Third, the inherent CD data of the compound were subtracted from the stationary ICD data to obtain the ICD spectrum of Mo-compound complex. The ICD spectrum was normalized to the molar concentration of the compound and is presented as the Δε′ values [22,23]. These Δε′ values are calculated by Δε′ = ΔA/c × d, where c is the molar concentration of the compound, assuming 100% complexation (A = the difference in absorption of left and right circularly polarized light; d = path length of the cell). Δε′ is expressed in [M −1 × cm −1 ] units.

HPLC-PDAD-UV and HPLC-ESI-MS Analyses
The HPLC-PDAD-UV analysis was carried out on a Venusil MP C18 column (5 μm, 100 Å, 4.6 mm × 250 mm; Agela Technologies, Tianjin, China) using the Waters HPLC equipment mentioned above. Each 5 μL of the sample MeOH solutions at 10 mg/mL was injected into the column and the elution was performed using MeOH-H 2 O in linear gradient (20% MeOH at initial time 0 min → 100% MeOH at 60 min → 100% MeOH at 90 min; flow rate, 1 mL/min). The acquired photodiode array (PDA) data were processed using the Empower PDA software to obtain targeted HPLC-PDAD-UV data. By analyzing both HPLC profiles at different wave lengths and UV absorption curves of related peaks, the peaks of C25 steroids in the range of 40-52 min retention times were authenticated to be recently appeared in the mutant AD-1-2 extract compared to the control G59 extract ( Figure S1 in the Supplementary Information).
The HPLC-ESI-MS analysis was performed on an LC-MS equipment equipped with Agilent 1100 HPLC system, AB Sciex API 3000 LC-MS/MS system and AB Sciex Analyst 1.4 software. HPLC was carried out on the Venusil MP C18 column (5 μm, 100 Å, 4.6 mm × 250 mm; Agela Technologies, Tianjin, China) at the same conditions mentioned for the HPLC-PDAD-UV analysis. The mass detector was set to scan a range from m/z 150-1500 both in the positive and negative modes. The acquired data were processed by Analyst 1.4 software to obtain targeted HPLC-ESI-MS data. The and negative ion peaks of the C25 steroids appeared with the retention times shortened approximately 3 min than in HPLC-PDAD-UV analysis because of the shortened flow length from the outlet of the HPLC column to the inlet of MS, and further supported the results of the HPLC-PDAD-UV analysis. Several data from the HPLC-ESI-MS analysis is given in Figure S2 in the Supplementary Information.

Conclusions
Three new (1-3) and 11 known (4-14) unusual C25 steroids with bicyclo[4.4.1]A/B rings were isolated by tracing newly produced metabolites in the EtOAc extract of an antitumor mutant AD-1-2 that was obtained by the DES mutagenesis of a marine-derived fungus Penicillium purpurogenum G59. HPLC-PDAD-UV and HPLC-ESI-MS analyses showed that the original G59 strain did not produce these metabolites and demonstrated that the production of 1-14 in the mutant AD-1-2 extract was caused by the activation of silent metabolites in original G59 strain by DES mutagenesis. Structures of the new compounds including their absolute configurations were determined by spectroscopic methods, especially the NMR and Mo 2 -induced CD spectra analyses.  [17,19] and supported also the effectiveness and usefulness of this new strategy [17].