Pyranodipyran Derivatives with Tyrosyl DNA Phosphodiesterase 1 Inhibitory Activities and Fluorescent Properties from Aspergillus sp. EGF 15-0-3

Four new benzodipyran racemates, namely (±)-aspergiletals A–D (3–6), representing a rare pyrano[4,3-h]chromene scaffold were isolated together with eurotiumide G (1) and eurotiumide F (2) from the soft-coral-derived fungus Aspergillus sp. EGF 15-0-3. All the corresponding optically pure enantiomers were successfully separated by a chiral HPLC column. The structures and configurations of all the compounds were elucidated based on the combination of NMR and HRESIMS data, chiral separation, single-crystal X-ray diffraction, quantum chemical 13C NMR, and electronic circular dichroism calculations. Meanwhile, the structure of eurotiumide G was also revised. The TDP1 inhibitor activities and photophysical properties of the obtained compounds were evaluated. In the TDP1 inhibition assay, as a result of synergy between (+)-6 and (−)-6, (±)-6 displayed strong inhibitory activity to TDP1 with IC50 values of 6.50 ± 0.73 μM. All compounds had a large Stokes shift and could be utilized for elucidating the mode of bioactivities by fluorescence imaging.

Benzodipyran is a small but unique class that mostly comes from chemical synthesis [11]. The biological activity of benzopyran scaffolds has gained considerable attention over the last few decades [12]. However, benzodipyran bearing a pyrano [4,3-h]chromene unit in its backbone is rare in nature, with only two examples reported to date [13].

Results
Previously, the metabolite capability of the title strain was preliminarily accessed by integration of the OSMAC strategy and LC-MS/MS analyses. As a result, the chemical profiles of benzodipyran were identified in four different media ( Figure S1), indicating that benzodipyran was produced in 4 out of 18 tested media. The rice medium was selected for further scale fermentation, and the EtOAc crude extract was subjected to silica gel, Sephadex LH-20, and semipreparative HPLC to obtain six benzodipyran racemates.

Results
Previously, the metabolite capability of the title strain was preliminarily accessed by integration of the OSMAC strategy and LC-MS/MS analyses [16]. As a result, the chemical profiles of benzodipyran were identified in four different media ( Figure S1), indicating that benzodipyran was produced in 4 out of 18 tested media. The rice medium was selected for further scale fermentation, and the EtOAc crude extract was subjected to silica gel, Sephadex LH-20, and semipreparative HPLC to obtain six benzodipyran racemates.
Compound 1 was isolated as yellow oil with a molecular formula of C 21 28.0). The aforementioned spectroscopic data were highly similar to those of eurotiumide G, a natural racemate isolated from the gorgonian-derived fungus Eurotium sp. XS-200900E6 [13], suggesting 1 might be the same compound as eurotiumide G. Further interpretation of the 1 H− 1 H COSY and HMBC spectra of 1 (Figure 2) confirmed the assumption. 53.9), four methylenes (δH 1.79 (m, H2-1′), 1.66 (m, H2-2′), 1.61 (m, H2-3′), and 1.51 (m, H2 4′); δC 30. 4, 25.9, 32.0, and 22.8), and three methyls (δH 0.93 (t, J = 6.4, H3-5′), 1.41 (s, H3-4″) and 1.41 (s, H3-5″); δC 14.2, 27.4, and 28.0). The aforementioned spectroscopic data were highly similar to those of eurotiumide G, a natural racemate isolated from the gorgonian derived fungus Eurotium sp. XS-200900E6 [13], suggesting 1 might be the same compound as eurotiumide G. Further interpretation of the 1 H− 1 H COSY and HMBC spectra of 1 ( Figure 2) confirmed the assumption. Interestingly, when we attempted to evaluate the relative configuration of 1, we found that although the NMR data of 1 were consistent with eurotiumide G reported by Wang [13] et al. as a natural racemate from the gorgonian-derived fungus Eurotium sp XS-200900E6 and by Namba [19] et al. as a synthesis through asymmetric total synthesis their relative configurations were in conflict. Although the structure of origina eurotiumide G were claimed to be revised by synthesis, small molecules with differen stereo configurations may have highly similar spectra. Meanwhile, according to the key NOE correlations described in the originally reported literature, it seems that the structure of eurotiumide G they proposed is correct. This interesting phenomenon aroused ou curiosity and we wondered whether these two molecules happened to have very simila NMR spectra as to be indistinguishable or the original proposed structure of eurotiumide G was incorrect. Therefore, GIAO-based quantum chemical 13 C NMR calculations a mPW1PW91/6-31+G(d,p) level using chloroform as solution were performed to clarify the relative configuration of 1 (eurotiumide G) unambiguously. There were four possible relative configurations that needed to be taken into account: 1R*,3R*,4R*-1 (1A) 1R*,3S*,4S*-1 (1B); 1S*,3S*,4R*-1 (1C); and 1R*,3S*,4R*-1 (1D). As a result, the calculated carbon chemical shifts of 1B were in excellent agreement with the experimental data, with coefficient value (R 2 ) of 0.9973, while the predicted data for the rest of the configuration did not match the experimental ones as indicated by R 2 of 0.9947, 0.9930, and 0.9949 respectively. Furthermore, DP4+ probability of 100% in favor of 1B suggested the relative configuration of the synthesis by Namba et al. was indeed correct, and the relative configuration of 1 was unambiguously concluded as 1R*,3S*,4S* ( Figure 3). Interestingly, when we attempted to evaluate the relative configuration of 1, we found that although the NMR data of 1 were consistent with eurotiumide G reported by Wang [13] et al. as a natural racemate from the gorgonian-derived fungus Eurotium sp. XS-200900E6, and by Namba [19] et al. as a synthesis through asymmetric total synthesis, their relative configurations were in conflict. Although the structure of original eurotiumide G were claimed to be revised by synthesis, small molecules with different stereo configurations may have highly similar spectra. Meanwhile, according to the key NOE correlations described in the originally reported literature, it seems that the structure of eurotiumide G they proposed is correct. This interesting phenomenon aroused our curiosity and we wondered whether these two molecules happened to have very similar NMR spectra as to be indistinguishable or the original proposed structure of eurotiumide G was incorrect. Therefore, GIAO-based quantum chemical 13 C NMR calculations at mPW1PW91/6-31+G(d,p) level using chloroform as solution were performed to clarify the relative configuration of 1 (eurotiumide G) unambiguously. There were four possible relative configurations that needed to be taken into account: 1R*,3R*,4R*-1 (1A); 1R*,3S*,4S*-1 (1B); 1S*,3S*,4R*-1 (1C); and 1R*,3S*,4R*-1 (1D). As a result, the calculated carbon chemical shifts of 1B were in excellent agreement with the experimental data, with coefficient value (R 2 ) of 0.9973, while the predicted data for the rest of the configurations did not match the experimental ones as indicated by R 2 of 0.9947, 0.9930, and 0.9949, respectively. Furthermore, DP4+ probability of 100% in favor of 1B suggested the relative configuration of the synthesis by Namba et al. was indeed correct, and the relative configuration of 1 was unambiguously concluded as 1R*,3S*,4S* ( Figure 3). The absolute configurations of (+)-1 and (−)-1 were further determined by quantum chemical ECD calculation ( Figure 4) and Cu Kα radiation of X-ray crystallographic analysis with quality flack parameter of 0.11(16) ( Figure 5). Hence, based on the combination of the above evidence, the structure including absolute configurations of 1 was unambiguously established, and the (+) eurotiumide G and (−) eurotiumide G were revised correctly as 1S,3R,4R and 1R,3S,4S, respectively.   However, as for the absolute configuration of eurotiumide G enantiomers, although the absolute configurations were previously established using enantioselective synthetic routes [19], the method they used to determine the structure was based on a Mo Kα radiation of X-ray crystallographic analysis with poor flack parameter of 0.3 (4), which may be unreliable. With the optical rotation value close to zero as well as almost no signal in its ECD spectrum, 1 was assumed to be a racemic mixture and subsequently separated into a pair of enantiomers, (+)-1 and (−)-1, with a ratio of 1:1 via a chiral HPLC column. The absolute configurations of (+)-1 and (−)-1 were further determined by quantum chemical ECD calculation ( Figure 4) and Cu Kα radiation of X-ray crystallographic analysis with quality flack parameter of 0.11 (16) (Figure 5). Hence, based on the combination of the above evidence, the structure including absolute configurations of 1 was unambiguously established, and the (+) eurotiumide G and (−) eurotiumide G were revised correctly as 1S,3R,4R and 1R,3S,4S, respectively.   Interpretation of the 1D and 2D NMR of 2 indicated that it was eurotiumide F, which was also obtained as a racemate from the gorgonian-derived fungus Eurotium sp. XS-200900E6 [13]. Similar to eurotiumide G, the absolute configuration of eurotiumide F was not solved thoroughly. In our experiments, the chiral resolution of 2 was successfully carried out, and the absolute configuration of (+)-2 and (−)-2 were determined as 1S,3R,4S and 1R,3S,4R by quantum chemical ECD calculations ( Figure 4). In addition, considering the experimental and calculated ECD spectra of compound 2 was complex and did not seem to be perfectly matched in the region between 200 and 220 nm, GIAO-based quantum chemical 13 C NMR calculations were performed to further confirm correctness of the relative configuration of 2. As shown in Figure S1, the calculated carbon chemical shifts of 1R*,3S*,4R*-2 (2D) were in excellent agreement with the experimental data with coefficient value (R 2 ) of 0.999 and DP4+ probability of 100% ( Figure S1), indicating the correct stereochemical assignments of both enantiomers.
Aspergiletal A (3) was isolated as yellow oil with a molecular formula of C22H32O6 by its HR-ESI-MS m/z 391.2120 [M-H] − (calculated for 391.2121), indicating the presence of seven degrees of unsaturation. Comprehensive analyses of the 1D NMR (Table 1) spectra of 3 with those of 1 suggested 3 was highly similar to 1, and the major differences were the appearance of an additional methoxyl group (δH 3.15 (s, OCH3-2″); δC 51.3) and an olefinic proton (δH 5.68 (d, H-2″)) of 1 being absent in 3. Interpretation of the HMBC spectrum from OCH3-2″ to C-2′′ located the methoxyl group at C-2″ ( Figure 2). The coupling constant of H-3 and H-4 and key NOE correlations of 3 were similar to 1, so the    Interpretation of the 1D and 2D NMR of 2 indicated that it was eurotiumide F, which was also obtained as a racemate from the gorgonian-derived fungus Eurotium sp. XS-200900E6 [13]. Similar to eurotiumide G, the absolute configuration of eurotiumide F was not solved thoroughly. In our experiments, the chiral resolution of 2 was successfully carried out, and the absolute configuration of (+)-2 and (−)-2 were determined as 1S,3R,4S and 1R,3S,4R by quantum chemical ECD calculations ( Figure 4). In addition, considering the experimental and calculated ECD spectra of compound 2 was complex and did not seem to be perfectly matched in the region between 200 and 220 nm, GIAO-based quantum chemical 13 C NMR calculations were performed to further confirm correctness of the relative configuration of 2. As shown in Figure S1, the calculated carbon chemical shifts of 1R*,3S*,4R*-2 (2D) were in excellent agreement with the experimental data with coefficient value (R 2 ) of 0.999 and DP4+ probability of 100% ( Figure S1), indicating the correct stereochemical assignments of both enantiomers.
Aspergiletal A (3) was isolated as yellow oil with a molecular formula of C22H32O6 by its HR-ESI-MS m/z 391.2120 [M-H] − (calculated for 391.2121), indicating the presence of seven degrees of unsaturation. Comprehensive analyses of the 1D NMR (Table 1) spectra of 3 with those of 1 suggested 3 was highly similar to 1, and the major differences were the appearance of an additional methoxyl group (δH 3.15 (s, OCH3-2″); δC 51.3) and an olefinic proton (δH 5.68 (d, H-2″)) of 1 being absent in 3. Interpretation of the HMBC spectrum from OCH3-2″ to C-2′′ located the methoxyl group at C-2″ ( Figure 2). The coupling constant of H-3 and H-4 and key NOE correlations of 3 were similar to 1, so the Interpretation of the 1D and 2D NMR of 2 indicated that it was eurotiumide F, which was also obtained as a racemate from the gorgonian-derived fungus Eurotium sp. XS-200900E6 [13]. Similar to eurotiumide G, the absolute configuration of eurotiumide F was not solved thoroughly. In our experiments, the chiral resolution of 2 was successfully carried out, and the absolute configuration of (+)-2 and (−)-2 were determined as 1S,3R,4S and 1R,3S,4R by quantum chemical ECD calculations ( Figure 4). In addition, considering the experimental and calculated ECD spectra of compound 2 was complex and did not seem to be perfectly matched in the region between 200 and 220 nm, GIAO-based quantum chemical 13 C NMR calculations were performed to further confirm correctness of the relative configuration of 2. As shown in Figure S2, the calculated carbon chemical shifts of 1R*,3S*,4R*-2 (2D) were in excellent agreement with the experimental data with coefficient value (R 2 ) of 0.999 and DP4+ probability of 100% ( Figure S2 (Table 1) spectra of 3 with those of 1 suggested 3 was highly similar to 1, and the major differences were the appearance of an additional methoxyl group (δ H 3.15 (s, OCH 3 -2"); δ C 51.3) and an olefinic proton (δ H 5.68 (d, H-2")) of 1 being absent in 3. Interpretation of the HMBC spectrum from OCH 3 -2" to C-2 located the methoxyl group at C-2" (Figure 2). The coupling constant of H-3 and H-4 and key NOE correlations of 3 were similar to 1, so the relative structure of 3 was determined as 1S*,3R*,4R*. Moreover, 3 was a racemic mixture and successfully separated into a pair of enantiomers (+)-3 and (−)-3 by a chiral HPLC column. The absolute configuration of (+)-3 and (−)-3 were established as 1S,3R,4R and 1R,3S,4S, respectively, using the quantum chemical ECD calculation method ( Figure 4).
Aspergiletal B (4) was isolated as yellow oil with a molecular formula of C 22 H 34 O 6 by its HR-ESI-MS m/z 393.2298 [M-H] − , (calculated for 393.2277). Interpretation of the NMR spectroscopic data as well as the molecular formula of 4 indicated that it was a dihydro analogue of 3. The main difference was the appearance of one methylene (δ H 1.81 (m, H-1 ); δ C 22.8] and one oxygenated methine (δ H 4.74 (dt, J = 6.8, 3.2, H-2 ); δ C 88.4) in 4 instead of two olefin carbons (δ H 6.57 (s, H-1"), δ C 103.8 and 161.5) in 3. This conclusion was confirmed by further interpretation of the 1 H− 1 H COSY and HMBC spectra of 4 ( Figure 2). As the structure of 4 contained two isolated stereoclusters, C-1−C-4 in part A and C-1"−C-2"in part B, separated by a pentasubstituted benzene ring, the relative configuration of isolated stereocluster was independently investigated. Based on the coupling constants of the observed protons and key NOESY correlations, the relative configuration of part A was concluded as 1S*,3R*,4R*. Due to the lack of confident NOE correlations, it was insufficient to make conclusions on the relative configurations of two individual stereoclusters using NMR methods alone. The optical values of 4 in MeOH solvent was close to zero, indicating 4 was a racemate. In order to establish the relative and absolute configurations of 4, it was subsequently separated by a chiral HPLC column, and quantum chemical ECD calculation was carried out. In terms of the whole-compound stereochemistry, four possible absolute configurations needed to be considered: 1S,3R,4R,2"S-4 (4A); 1R,3S,4S,2"R-4 (4B); 1S,3R,4R,2"R-4 (4C); and 1R,3S,4S,2"S-4 (4D). As shown in Figure 4, the comparison of predicted ECD curves of 1S,3R,4R,2"S-4 (4A) and 1R,3S,4S,2"R-4 (4B) with the experimental ones of (+)-4 and (−)-4 showed quite good agreement (Figure 4), which allowed the absolute configurations of (+)-4 and (−)-4 to be established. Similar to 2, as the experimental and calculated ECD spectra of compound 4 did not seem to be perfectly matched in the region between 230 and 280 nm, GIAO-based quantum chemical 13 C NMR calculations were also performed, and the result showed that the stereochemistry of compound 4 that we had derived was correct ( Figure S3).
Aspergiletal C (5) was isolated as yellow oil with a molecular formula of C 21 H 32 O 6 by its HR-ESI-MS m/z 379.2123 [M−H] − , (calculated for 379.2121). The 1D NMR data of 5 was structurally similar to those of 4, except for the difference of the resonance assigned to C-1" and absence of the signal corresponding to C-2" methoxyl (δ H 3.17 (s); δ C 50.2), suggesting the presence of C-2" hydroxyl in 5. Similar to 4, 5 was also a racemate and contained two isolated stereoclusters, and the absolute configurations of (+)-5 and (−)-5 were independently assigned as 1S,3R,4R,2"S-5 and 1R,3S,4S,2"R-5 ( Figure 4). Certainly, the absolute stereochemical configuration at C-2" of 5 could be determined by Mosher's method. However, unfortunately, 5 was only obtained with total amount of 1.0 mg. When 5 was chiral separated by HPLC, only 0.5 mg each of (+)-5 and (−)-5 were obtained, making it difficult to perform Mosher's method. Considering there were two isolated stereoclusters in 5, the stereochemistry of compound 5 was also confirmed using GIAO-based quantum chemical 13 C NMR calculations ( Figure S4).
Aspergiletal D (6) was isolated as yellow oil with a molecular formula of C 21 H 32 O 7 by its HR-ESI-MS m/z 395.2076 [M−H] − , (calculated for 395.2070). The 1D NMR data highly resembled those of 5, suggesting 6 had the same scaffold as 5, except for the appearance of the oxygenated methine (δ H 5.31 (d, J = 4.8, H-1 ); δ C 74.0) instead of methylene (δ H 3.18 (m, H-1 a) and 3.14 (d, J = 8.4, H-1 b); δ C 31.0). This suggested the presence of C-1 hydroxyl in 6. Similarly, 6 was a racemate and contained two isolated stereoclusters, and the relative configuration of part A in 6 was established as 1S*,3R*,4R*. Meanwhile, the key NOE correlations of H-1 /H-2 and H 3 -4 as well as small coupling constant (J 1 ,2 = 4.8 Hz) for part B in 6 indicated that H-1 and H-2 were in cis-diaxial ( Figure 6A). The absolute configuration of the 1 ,2 -diol unit in part B of 6 could be assigned by measuring the Mo 2 (OAc) 4 induced circular dichroism spectrum (ICD) of Mo complex of (+)-6. The ICD spectrum showed a positive sign of Cotton effect (CE) at 303 nm ( Figure 6B), allowing the assignment of C-1 and C-2 as 1 S,2 S based on the empir-ical rule [20,21]. Finally, the absolute configurations of (+)-6 and (−)-6 were independently assigned as 1S,3R,4R,1 S,2 S-6 and 1R,3S, 4S,1 R,2 R-6 by quantum chemical ECD calculation ( Figure 4). Similar to 5, the stereochemistry of 6 was also confirmed using GIAO-based quantum chemical 13 C NMR calculations ( Figure S5). ICD spectrum showed a positive sign of Cotton effect (CE) at 303 nm ( Figure 6B), allowing the assignment of C-1″ and C-2″ as 1″S,2″S based on the empirical rule [20,21]. Finally, the absolute configurations of (+)-6 and (−)-6 were independently assigned as 1S,3R,4R,1″S,2″S-6 and 1R,3S, 4S,1″R,2″R-6 by quantum chemical ECD calculation ( Figure  4). Similar to 5, the stereochemistry of 6 was also confirmed using GIAO-based quantum chemical 13 C NMR calculations ( Figure S4). Based on the coisolation of the precursor aspergin (7), a plausible biosynthetic pathway for compounds 1-6 is proposed in Scheme 1. Epoxidation of the unsaturated C7alkyl chain in 7 was followed by isomerization of the double bond and epoxidation to obtain i and ii. Subsequently, selective epoxide cleavage of i and ii were used to obtain the key intermediates iii and iv, respectively. Then, intermediates iii and iv were transformed to v and vi, respectively, under nucleophilic addition. Next, v and vi were subjected to methylation and intramolecular electrophilic addition-elimination to produce 1 and 2, respectively. The subsequent sequence of oxidation, methylation, reduction, and oxidation of 1 led to the furnishment of 3-6, respectively.
Considering the chiral properties of compounds will affect their activity, all of the obtained compounds, including racemates and enantiomers, were screened for TDP1 inhibition through a fluorescence assay using a quenched fluorescent single-stranded oligonucleotide (5′-FAM-AGGATCTAAAAGACTT-BHQ-3′) as substrate. Indeed, (+)-6, (−)-6, and (±)-6 exhibited different inhibitory activity against the TDP1 enzyme with IC50 values of 27.76 ± 1.73, 37.31 ± 3.63, and 6.50 ± 0.73 μM, respectively (Table 2). To further investigate the mechanism of the TDP1 inhibitory activity of (+)-6 and (−)-6, a hypothetical binding mode was built using in silico docking from the X-ray crystal structure of humanized TDP1-inhibitor (PDB: 6DJD) [22] to evaluate the mode of interactions of (+)-6 and (−)-6 with TDP1. The inhibitor was docked into the binding site. The hypothetical structural position of the top-ranked (+)-6 is shown in Figure 7A. The tricyclic core of (+)-6 binds to the DNA binding groove in the proximity of the catalytically important residue His263, indicating a potential π-π stacking interaction stabilizing the inhibitor TDP1 complex. Three hypothetical hydrogen bonds were observed between the phenol hydroxy group and the amide group of residue Asn516 (2.9 Å), the 1″-hydroxy group and Ser518 residue (3.0 Å), and the pyran O atom and Ser400 residue (3.5 Å), which might contribute to TDP1 inhibition. Similarly, the polycyclic core scaffold of (−)-6 also lied along the DNA binding groove (Figure 7B), and the H-bonds (3.1 Å) between the phenol hydroxy group and the amide group of residue Asn516 was also observed. In addition, two H-bonds formed between the 2″-hydroxyl group and Ser400 residue (2.5 Å) and the pyran oxygen atom and phenol hydroxy group of Tyr204 residue (3.1 Å). As shown in Figure 7C, the configuration of (+)-6 was more geometrically suitable for the narrow catalytic pocket than Based on the coisolation of the precursor aspergin (7), a plausible biosynthetic pathway for compounds 1-6 is proposed in Scheme 1. Epoxidation of the unsaturated C 7 -alkyl chain in 7 was followed by isomerization of the double bond and epoxidation to obtain i and ii. Subsequently, selective epoxide cleavage of i and ii were used to obtain the key intermediates iii and iv, respectively. Then, intermediates iii and iv were transformed to v and vi, respectively, under nucleophilic addition. Next, v and vi were subjected to methylation and intramolecular electrophilic addition-elimination to produce 1 and 2, respectively. The subsequent sequence of oxidation, methylation, reduction, and oxidation of 1 led to the furnishment of 3-6, respectively.

Compound TDP1 Inhibition (μM) Compound TDP1 Inhibition (μM)
a IC50 value, defined as the concentration of the compound to inhibit 50% of TDP1 activity. Every Considering the chiral properties of compounds will affect their activity, all of the obtained compounds, including racemates and enantiomers, were screened for TDP1 inhibition through a fluorescence assay using a quenched fluorescent single-stranded oligonucleotide (5 -FAM-AGGATCTAAAAGACTT-BHQ-3 ) as substrate. Indeed, (+)-6, (−)-6, and (±)-6 exhibited different inhibitory activity against the TDP1 enzyme with IC 50 values of 27.76 ± 1.73, 37.31 ± 3.63, and 6.50 ± 0.73 µM, respectively (Table 2). To further investigate the mechanism of the TDP1 inhibitory activity of (+)-6 and (−)-6, a hypothetical binding mode was built using in silico docking from the X-ray crystal structure of humanized TDP1-inhibitor (PDB: 6DJD) [22] to evaluate the mode of interactions of (+)-6 and (−)-6 with TDP1. The inhibitor was docked into the binding site. The hypothetical structural position of the top-ranked (+)-6 is shown in Figure 7A. The tricyclic core of (+)-6 binds to the DNA binding groove in the proximity of the catalytically important residue His263, indicating a potential π-π stacking interaction stabilizing the inhibitor TDP1 complex. Three hypothetical hydrogen bonds were observed between the phenol hydroxy group and the amide group of residue Asn516 (2.9 Å), the 1"-hydroxy group and Ser518 residue (3.0 Å), and the pyran O atom and Ser400 residue (3.5 Å), which might contribute to TDP1 inhibition. Similarly, the polycyclic core scaffold of (−)-6 also lied along the DNA binding groove ( Figure 7B), and the H-bonds (3.1 Å) between the phenol hydroxy group and the amide group of residue Asn516 was also observed. In addition, two H-bonds formed between the 2"-hydroxyl group and Ser400 residue (2.5 Å) and the pyran oxygen atom and phenol hydroxy group of Tyr204 residue (3.1 Å). As shown in Figure 7C, the configuration of (+)-6 was more geometrically suitable for the narrow catalytic pocket than (−)-6, leading to closer proximity to the catalytic domain. The gold score values of (+)-6 and (−)-6 were 55.62 and 44.90, respectively. Interestingly, although the mode of interactions of (+)-6 and (−)-6 with TDP1 were similar, with the racemate (±)-6 exhibiting stronger inhibitory activity against the TDP1 enzyme with IC 50 value of 6.50 ± 0.73 µM, which may be a result of synergy between (+)-6 and (−)-6. The mechanism of the TDP1 inhibitory activity of (±)-6 remains to be validated by biophysical assays in further research. Table 2. TDP1 enzyme inhibition of the isolated compounds 1-6 (IC 50 ± SD in µM) a .   In addition, the fluorescent properties of 1-6 were also evaluated. Interestingly, all the probes emitted detectable fluorescence ( Figure 8). Fluorescence measurements revealed that all the compounds had a large Stokes shift (up to 266 nm, Table 3), suggesting they were capable of the resistance of fluorescence quenching. Surprisingly, the fluorescence of 1 increased substantially with time, indicating a structural change to a more fluorescent, undetermined species upon 365 nm irradiation. Compound 2 responded similarly to 1 but with a smaller magnitude improvement in fluorescence. In contrast, 4 appeared to be inert to photobleaching compared to 3, 5, and 6, all of which showed significant loss of fluorescence with time ( Figure 9). Meanwhile, compounds 1-6 with the concentration of 100 µM/mL were exposed to 485 nm UV light, and their fluorescence emission spectra were recorded ( Figure 10) as a control experiment because the assay used to assess TDP1 inhibition was a fluorescence assay. As the figure shows, 1-6 did not exhibit fluorescence emission spectra at 510 nm when exposed to 485 nm UV light. In addition, during the TDP1 inhibition assay, the tested compound solutions were read by a Flash multimode reader at Ex485/Em510 nm to identify false-positive compounds that had autofluorescence. The above results indicate that the fluorescence of the compounds did not interfere with the inhibition results. (C) Overlay of the binding mode of (+)-6 or (−)-6 with TDP1. The TDP1 inhibitor (+)-6 is represented as a green stick model, (−)-6 is represented as a purple stick model, and the protein is represented as a cartoon model (A, B) and as a surface model (C).

General Experimental Procedures
Details of the instruments and materials used in this work are included in the Supporting Information.

Fungal Strain and Fermentation
The fungal strain EGF 15-0-3 was isolated from a soft coral collected from the South China Sea and identified as Aspergillus sp. [23]. A voucher sample of EGF15-0-3 was preserved at the Lab of Marine Natural Medicine, Guangzhou University of Chinese Medicine, P. R. China.
The fungus was cultured on a plate containing PDB medium at 28 °C for 3 days. A single colony was transferred aseptically to an Erlenmeyer flask containing PDB medium and incubated on a rotatory shaker at 165 rpm at 28 °C for 3 days to obtain the seed culture. Subsequently, the seed culture was subcultured in polished glutinous rice medium (autoclaved sterilization with rice and artificial seawater) and incubating at 28 °C for 35 days under static conditions.

General Experimental Procedures
Details of the instruments and materials used in this work are included in the Supporting Information.

Fungal Strain and Fermentation
The fungal strain EGF 15-0-3 was isolated from a soft coral collected from the South China Sea and identified as Aspergillus sp. [23]. A voucher sample of EGF15-0-3 was preserved at the Lab of Marine Natural Medicine, Guangzhou University of Chinese Medicine, P. R. China.
The fungus was cultured on a plate containing PDB medium at 28 • C for 3 days. A single colony was transferred aseptically to an Erlenmeyer flask containing PDB medium and incubated on a rotatory shaker at 165 rpm at 28 • C for 3 days to obtain the seed culture. Subsequently, the seed culture was subcultured in polished glutinous rice medium (autoclaved sterilization with rice and artificial seawater) and incubating at 28 • C for 35 days under static conditions.

Extraction and Isolation
After 35 days of cultivation, the rice medium (10 kg

Extraction and Isolation
After 35 days of cultivation, the rice medium (10 kg) of EGF 15-0-3 was extracted with ethyl acetate three times (8 h per time) on a rotatory shaker at 165 rpm and concentrated under reduced pressure at 40 °C to yield 50.0 g of the EtOAc residue. The residue was then subjected to silica gel column eluted with a gradient mixture of petroleum ether/ethyl acetate (100:0 to 0:100, v/v) to obtain 10 13 C NMR (100 MHz) data in CDCl3, see Table 1.  13 C NMR (100 MHz) data in CDCl3, see Table 1.  13 C NMR (100 MHz) data in CDCl3, see Table 1.

Extraction and Isolation
After 35 days of cultivation, the rice medium (10 kg) of EGF 15-0-3 was extracted with ethyl acetate three times (8 h per time) on a rotatory shaker at 165 rpm and concentrated under reduced pressure at 40 °C to yield 50.0 g of the EtOAc residue. The residue was then subjected to silica gel column eluted with a gradient mixture of petroleum ether/ethyl acetate (100:0 to 0:100, v/v) to obtain 10 13 C NMR (100 MHz) data in CDCl3, see Table 1.  13 C NMR (100 MHz) data in CDCl3, see Table 1.  13 C NMR (100 MHz) data in CDCl3, see Table 1.

Extraction and Isolation
After 35 days of cultivation, the rice medium (10 kg) of EGF 15-0-3 was extracted with ethyl acetate three times (8 h per time) on a rotatory shaker at 165 rpm and concentrated under reduced pressure at 40 °C to yield 50.0 g of the EtOAc residue. The residue was then subjected to silica gel column eluted with a gradient mixture of petroleum ether/ethyl acetate (100:0 to 0:100, v/v) to obtain 10 major fractions (Fr. A−Fr. J) by TLC analysis. Fr. J  13 C NMR (100 MHz) data in CDCl3, see Table 1.  13 C NMR (100 MHz) data in CDCl3, see Table 1.  13 C NMR (100 MHz) data in CDCl3, see Table 1.

Extraction and Isolation
After 35 days of cultivation, the rice medium (10 kg 13 C NMR (100 MHz) data in CDCl3, see Table 1.   13 C NMR (100 MHz) data in CDCl3, see Table 1.

Extraction and Isolation
After 35 days of cultivation, the rice medium (10 kg 13 C NMR (100 MHz) data in CDCl3, see Table 1.  13 C NMR (100 MHz) data in CDCl3, see Table 1.

Chiral Separation of 1-6
Compounds 1-6 were subsequently performed on a semipreparative HPLC with a chiral chromatographic column (FLM Chiral ND (2)) (Guangzhou FLM Scientific Instrument Co., Ltd., Guangzhou, China) and successfully separated into the optical pure enantiomers. More detailed information are included in the Supporting Information.

Quantum Chemical Calculations
Firstly, random conformational searches were performed by SYBYL X 2.1.1 program (Tripos) using MMFF94s molecular force field, with an energy cutoff of 10 kcal mol −1 to the global minima to obtain the conformers, and the obtained conformers were subsequently subjected to structural optimization using Gaussion 09 software at the B3LYP/6-31+G(d) level. The optimized stable conformers were performed at mPW1PW91/6-311+G(d,p) level with chloroform or methanol as the solvent for quantum chemical NMR calculations, and ECD calculations were performed at the B3LYP/6-31+G* level in the gas phase using Gaussion 09 software package [24] according to previous reports [25]. Boltzmann distributions of the overall NMR and ECD data were estimated and subsequently compared with the experimental ones. The DP4+ analysis was performed according to the published protocol [26]. The ECD spectra were produced by SpecDis 1.70.1 software (University of Wuerzburg, Wuerzburg, Germany) [27].

X-ray Crystallographic Analysis
Crystal (−)-1 was obtained using the solvent vapor diffusion method. The intensity data of (−)-1 was recorded on a Rigaku Oxford Diffraction Supernova diffractometer (Agilent SuperNova, America) with Cu Kα radiation (λ = 1.54184 Å). The crystal structure was solved and refined with a similar method reported previously [28]. For the structural refinements, nonhydrogen atoms were refined anisotropically and the hydrogen atom positions were geometrically idealized and allowed to ride on their parent atoms. Crystal data of (−)-1 in the standard CIF format were deposited with the Cambridge Crystallographic Data Centre with CCDC number 1981300. The crystallographic data of (−)-1 are shown in Table S1.

TDP1 Inhibition Assay
The TDP1 inhibition was tested through a fluorescence assay according to a previously reported method [29]. Detailed procedures are included in the Supporting Information.

Molecular Modeling
Molecular Modeling was according to a previously reported method [8]. Details are included in the Supporting Information.

Measurement of Fluorescent Properties
Each probe was prepared at 300 µg/mL in MeOH. The absorption spectra were recorded from 200 to 400 nm using a Chirascan circular dichroism spectrometer (Jasco, Tokyo, Japan). The fluorescence spectra were recorded from 385 to 750 nm with an excitation wavelength of 365 nm using Edinburgh Instruments FLS1000. For UV-irradiated probes, each probe was exposed to 365 nm UV light from a handheld UV lamp for a certain time, and the fluorescence spectra were then recorded using the same method mentioned above.

Conclusions
Four new benzodipyran racemates representing a rare pyrano[4,3-h]chromene scaffold, possibly originating from a cascade sequence of epoxidation, intramolecular electrophilic addition-elimination, and oxidation from the aspergin side chain, were isolated from the soft-coral-derived fungus Aspergillus sp. EGF 15-0-3. They were successfully separated by a chiral HPLC, and their absolute stereochemistry were determined according to XRD (Cu Kα), Snatzke's method, and ECD calculations. Meanwhile, the structure of eurotiumide G was also revised. The enzyme assay indicated 6 to be a novel TDP1 inhibitor, showing strong TDP1 inhibition ability with IC 50 value of 6.50 ± 0.73 µM. Preliminary results of molecular docking showed that different configurations of (+)-6 and (−)-6 would bring different hydrogen bond interactions and that (+)-6 was more geometrically suitable for the narrow catalytic pocket than (−)-6, leading to closer proximity to the catalytic domain.
However, the strong inhibitory activity against the TDP1 enzyme by (±)-6 indicated that the combination of the enantiomers provided inhibitory synergy. The mechanism of the significant TDP1 inhibitory activity of (±)-6 remains to be validated by biophysical assays in further research. In addition, the superior fluorescent properties of benzodipyran derivatives could be utilized for elucidating the mode of bioactivities by fluorescence imaging.  Table S1: Crystal data and structure refinement for (−)-1; Tables S2-S6: Experimental chemical shifts and calculated unscaled shifts for PD4+ probability analysis for 1, 2, 4, 5, and 6; Tables S7-S15: Details of ECD calculations of 1-6; Figures S12-S17: Key molecular orbitals 1-6; Figures S18-S67: the UV, IR, HRESIMS, 1D NMR, and 2D NMR spectra of compounds 1-6.