*2.1. Epigenetic Manipulation*

The epigenetic manipulation of *Phomopsis asparagi* DHS-48 was conducted in both liquid medium and solid medium by using the DNMT inhibitor 5-aza, the HDAC inhibitor sodium butyrate, and the combination of these inhibitors at different concentrations (0, 10, 50, 100 μM). Cultivation without these epigenetic modifiers was used as a control. By comparing the colony growth on PDA (Figure 2a) and dry biomass (calibration graph Figure 2b) in PDA (Figure 2c) and PDB (Figure 2d), we found that the DNMT and HDAC inhibitors produced inconsistent results, and 50 μM sodium butyrate solid fermentation was preferable to induce more remarkable chemical diversity of the secondary metabolites. The HPLC analyses of the EtOAc extracts of *Phomopsis asparagi* DHS-48 cultivated in the presence of different epigenetic agents in all the cases further confirmed our deduction (Figure S44). Consequently, a scaled-up fermentation with 50 μM sodium butyrate was carried out.

**Figure 2.** *Cont*.

**Figure 2.** Comparison of colony growth, dry biomass (in PDA and PDB) of *Phomopsis asparagi* DHS-48 in the presence of different concentrations (0, 10, 50, 100 μM) of 5-aza, sodium butyrate, and the combination of these inhibitors**:** (**a**) colony growth. In (A), 5-aza was added; in (B), sodium butyrate was added; in (C), the combination of these inhibitors was added; (**b**) calibration graph for calculating dry biomass based on nucleic acid contents with different epigenetic doses in PDA; (**c**) the dry biomass of fungi cultivated in PDA with different epigenetic doses; (**d**) the dry biomass of fungi cultivated in PDB with different epigenetic doses.

The EtOAc extracts of the mycelia and solid rice medium incubated with 50 μM sodium butyrate were subjected to HPLC analyses. By comparing with the blank control (Figures 3 and S45), the production levels of the known metabolites **6**–**8**, **10**, and **11** were considerably enhanced in the sodium-butyrate-inhibited fermentation at the same injection concentration. In addition, certain peaks of **1**–**5** and **9** appear to be present in the chromatograms from the 50 μM HDAC inhibitor that are absent in the control group. Continuously, these differences were also supported by the fact that the 1H NMR metabolic profile (Figure 4) of EtOAc extracts showed several additional significant hydrogen resonances between 5.5 and 8.0 ppm, compared with the control group.

**Figure 3.** HPLC profiles of fungal EtOAc extracts (**a**) obtained from rice solid-substrate medium after 50 μM sodium butyrate treatment and (**b**) obtained from rice solid-substrate medium without epigenetic inhibitor treatment. HPLC chromatograms: C18 column (Agilent Technologies 10 mm × 250 mm). Solvents: A, H2O; B, MeOH. Linear gradient: 0 min, 60% B; 40 min, 100% B.Temperature 25 ◦C. Flow rate 2 mL/min. UV detection at λ = 210 nm. Peaks **1**–**11** represent the isolated metabolites. - Compounds **1** and **5** in (**a**) represent the new compounds stimulated by epigenetic manipulation.

#### *2.2. Structure Elucidation of the New Compounds*

Phaseolorin J (**1**) was isolated as a light yellow amorphous powder. Its molecular formula was determined as C15H16O7 on the basis of HRESIMS data (*m*/*z* 331.0781 [M + Na] +, calcd for C15H16O7 Na 331.0788), which clearly indicated the presence of eight indices of unsaturation. The 1H and 13C NMR data of **1** (Table 1) and its 1H-1H COSY and HSQC spectra showed the presence of a series of characteristic signals for a 1,2,3-trisubstituted benzene ring (*δ*H 6.43 (d, *J* = 8.2 Hz), *δ*C 109.3, d, CH-2; *δ*H 7.38 (t, *J* = 8.2 Hz), *δ*C 138.9, d, CH-3; *δ*H 6.52 (d, *J* = 8.2 Hz), *δ*C 108.7, d, CH-4), one olefinic methine of a trisubstituted double bond (*δ*H 5.61 (dq, *J* = 4.8, 1.7 Hz); *δ*C 122.3, d, CH-7), a tertiary methyl (*δ*H 1.86, 3H, d, 1.7; *δ*C 19.2, q, CH3–11), an oxygenated methylene (*δ*H 4.12, (d, *J* = 13.2 Hz), *δ*H 4.03 (d, *J* = 13.2 Hz); *δ*C 64.1, t, CH2-12), and two oxygenated methines (*δ*H 4.69, 1H, br s, *δ*C 74.5, d, CH-5; 4.55, 1H, d, *J* = 4.8 Hz, *δ*C 67.9, d, CH-8). The magnitude of the 1H-1H COSY spectrum led to the observation of long-range correlations, including the assignments of vicinal coupling with H-5 and proton H-7 on the *cis*-substituted double bond, as well as homoallylic couplings with H-8 and CH3-11. A comparison of the 1H and 13C NMR data of **1** with those of phaseolorin D (**2**) [52] revealed that both had the same chromone core, except for the presence of one trisubstituted double bond at C-6 (*δ* 140.5) and C-7 (*δ* 122.3) instead of sp<sup>3</sup> methine (C-6) and sp<sup>3</sup> methylene (C-7) in **2**. Confirming evidence was obtained from the 1H-1H COSY correlation from olefinic proton (*δ*H 5.61, H-7) to the oxygenated methine (*δ*H 4.69, H-8) and HMBC correlations from H3-11 to C-5, C-6 and C-7 (Figure 5).

**Figure 5.** Key COSY and HMBC correlations of Compounds **1** and **5**.


**Table 1.** 1H (400 MHz) and 13C (100 MHz) NMR data of 1 and 2 in CD3OD.

In the NOESY experiment of **1** (Figure 6), the correlations of H2-12/H-5 indicated the same spatial orientation. Biogenetically, the configuration of **1** was deduced to be the same as that of **2**, and the calculated ECD spectrum method can be used to predict the absolute configuration of C-8a and C-10a, respectively (Figure 7). Consequently, the absolute configuration of C-5 was assigned to be *S*. However, neither the lack of NOE between H-5/H-8 nor the adjacent coupling constant of *J*7,8eq = 4.8 Hz between H-7 and H-8 supported the relative configuration between H-5 and H-8. To solve this problem, the *δ*C values of two plausible epimers, namely 5*S*,5a*S*,8*S*,8a *R*-**1** and 5*S*,5a*S*,8 *R*,8a *R*-**1** (8-*epi*-**1**), were performed after the optimization of the selected conformers at the B3LYP/6-31G(d) level. The results showed that the calculated 13C NMR spectrum of the truncated model 5*S*,5a*S*,8*S*,8a *R* -**1** perfectly matched with the experimental one (Figure 8). Therefore, the configuration of **1** was conclusively assigned and given the tentative name phaseolorin J.

**Figure 6.** Key NOESY correlations of compounds **1** and **5**.

**Figure 7.** Experimental and calculated electronic circular dichroism (ECD) spectra of **1** and **5**.


**Figure 8.** 13C NMR calculation results of two plausible epimers (**1** and 8-*epi*-**1**) at the B3LYP/6-31G(d) level: (**a**) linear correlation plots of calculated and experimental 13C values; (**b**) DP4+ probability of 13C values of **1**.

Phomoparagin D (**5**) was obtained as a colorless amorphous powder. The molecular formula of **5** was established as C28H37NO5 from its HRESIMS (*m*/*z* 506.2304 [M + K] +, calcd for C28H37NO5K 506.2303. The 1H NMR spectrum (Table 2) showed proton signals for a mono substituted phenyl at *δ*H (7.22−7.31, 5H), a tertiary methyl at *δ*H (0.92, 3H, s, H3-23), two secondary methyl groups at *δ*H (0.80, 3H, d, *J* = 6.7 Hz, H3-11; 1.02, 3H, d, *J* = 6.8 Hz, H3-22), an exocyclic methylene group at *δ*H (5.22 and 5.01, 2H, both s, H2-12), four oxygenated methine groups at *δ*H (3.82, 1H, d, *J* = 10.8 Hz, H-7; 3.21, 1H, d, *J* = 2.4 Hz, H-19; 2.99, 1H, m, H-20; 3.43, 1H, s, H-21), and two olefinic methine groups at *δ*H (5.72, 1H, d, *J* = 15.5 Hz, 9.5 Hz, H-13; 5.54, 1H, m, H-14). The 13C NMR and DEPT spectra (Table 1) of compound **5** displayed 28 carbons, including 3 sp<sup>3</sup> methyls, 3 sp<sup>3</sup> methylenes, 9 sp<sup>3</sup> methines, 2 sp<sup>3</sup> quaternary carbons, 1 sp<sup>2</sup> exocyclic methylene, 7 sp<sup>2</sup> olefinic methines, and 3 sp<sup>2</sup> quaternary carbons (2 olefinic carbon and 1 amide carbonyl). The carbon profile and characteristic 1H NMR signals, as well as the 2D NMR spectra of **5** revealed that it has a similar indole-based cytochalasin skeleton as that of cytochalasin J (**6**), which was first reported in 1981 as deacetylcytochalasin H from the same *Phomopsis* sp. [53]. The main difference between the two compounds is the lack of the typical C19-C20 double bond (*δ*H 5.76, *δ*C 129.3, d, CH-19; *δ*H 5.85, *δ*C 137.2, d, CH-20) in the macrocycle ring of the latter that was replaced by a 19, 20-epoxide ring (*δ*H 3.21 (d, *J* =2.4 Hz), *δ*C 63.1, CH-19; *δ*H 2.99, m, *δ*C 57.7, CH-20) in **5**. The existence of the epoxide ring was deduced by the analysis of its HRESIMS data, and the molecular weight of **5** was 16 mass units larger than that of **6**. This finding was supported by the 1H-1H COSY correlations of H-7/H-8/H-13/ H-14/ H-15/ H-16/ H-17/ H-18/ H-19/H-20/H-21, along with the HMBC correlations from H3-23 (*δ*H 0.92, 3H, s) to C-17, C-18 and C-19, and H-21 (*δ*H 3.43, 1H, s) to C-8, C-9, C-19

and C-20 (Figure 5). The diagnostic ROESY correlations (Figure 6) positioned H-3, H3-11, H-7, H3-22, H3-23, H-20, and H-21 on the α-face and H-4, H-5, H-8, H-14, H-16, and H-19 on the β-face of **5**, whereas the absolute configuration was assigned by a comparison of the experimental and simulated electronic circular dichroism (ECD) spectra generated by the time-dependent density functional theory (TDDFT) calculations at the B3LYP/6-31+G(d,p) level using the Gaussian 09 program. The experimental ECD spectrum (CH3OH) for 3*S*, 4*R*, 5*S*, 7*S*, 8*R*, 9*R*, 16*R*, 16*R*, 19*R*, 20*S*, and 21*R* -**5** matched well with the calculated spectrum (Figure 7), which confirmed the unambiguous assignment of the absolute configuration of **5**, and the trivial name phomoparagin D was assigned. The possible biogenetic pathway of phomoparagin D (**5**) was postulated (Scheme 1), which might arise from cytochalasin J (**6**) by a different set of catalyzed reactions.


**Table 2.** 1H (400 MHz) and 13C (100 MHz) NMR data of **5** and **6** in CD3OD.

A plausible biosynthesis of compounds **1**–**11** was proposed, as shown in Schemes 1 and 2. More than 4000 chromones have been isolated and structurally elucidated from natural origin until now, and they are biosynthesized by the type III polyketide synthases (PKSs) [54]. Compounds **1** and **2** isolated from *P. asparagi* DHS-48 are assumed to be derived from one acetyl-CoA starter and seven molecules of malonyl-CoA extender units to form an octaketide that undergoes Claisen condensation and cyclization to yield anthraquinone precursors such as emodin, even though it was not isolated in this study. Oxidative cleavage, cyclization via epoxidation, and nucleophilic attack by a hydroxyl group to give the ringclosed dihydroxanthone involved the epimerization of C-10a. The subsequent keto–enol equilibrium and redox would provide compounds **1** and **2**, referring to the reports made

by Rönsberg et al. [55]. Previous feeding experiments with sodium 13C-labeled acetate by Lösgen et al. [56] in 2007 revealed that a heptaketide precursor is involved in the biosynthesis of **3** and **4**, which are analogues to phomochromenones D-G isolated in our previous study [46], implying some cryptic post-synthesis modification genes were stimulated by the currently adopted epigenetic manipulation for the production of those metabolites previously unobserved or merely increased sufficiently under epigenetic control to be detected. Cytochalasins **5–11** might rationally share a common biosynthetic precursor as we previously described via polyketide synthase (PKS)/nonribosomal peptide synthetase (NRPS) hybrid machinery [38]. The stimulated metabolite **5** was is likely to be also derived from **6** by epoxidation, meanwhile **9** feasibly converted through catalytic dehydration.

**Scheme 1.** Proposed biosynthetic pathway for compounds **5** and **9** from **6**.

**Scheme 2.** Proposed biosynthetic pathway for compounds **1**–**4**.

#### *2.3. Biological Activity of Compounds*

The immunosuppressive assay showed that compounds **1** and **8** exhibited moderateto-weak inhibitory activity against ConA-induced T and LPS-induced B murine splenic lymphocytes in vitro, with the IC50 values of 42 and 88 μM and 15 and 110 μM (Table 3), respectively, whereas the other investigated compounds showed no apparent inhibitory effect. Additionally, compound **5** showed significant in vitro cytotoxicity against human cancer cell lines Hela, with an IC50 value of 5.8 μM, and showed moderately significant in vitro cytotoxicity against human cancer cell lines HepG2, with an IC50 value of 59 μM (Table 4), respectively, which was comparable with the positive controls adriamycin and fluorouracil. These results suggested that the 19,20-epoxide ring in compound 5 is essential for its inhibition of tumor cell proliferation compared with compounds **6**–**11**.

**Table 3.** Immunosuppressive activities of tested compounds.


a Data are presented as mean ± SD from three separate experiments. b Positive control. '-' stands for no inhibitory effect at 200 μM.


a Data are presented as mean ± SD from three separate experiments. b Hela cell positive control. c Hepg2 cell positive control. '-' stands for no inhibitory effect at 200 μM.

#### **3. Materials and Methods**
