Tersone A-G, New Pyridone Alkaloids from the Deep-Sea Fungus Phomopsis tersa

Four phenylfuropyridone racemates, (±)-tersones A-C and E (1–3, 5), one phenylpyridone racemate, (±)-tersone D (4), one new pyridine alkaloid, tersone F (6), single new phenylfuropyridone, tersone G (7) and two known analogs 8 and 9 were isolated from the deep-sea fungus Phomopsis tersa. Their structures and absolute configurations were characterized on the basis of comprehensive spectroscopic analyses, single-crystal X-ray diffraction experiments, and electronic circular dichroism (ECD) calculations. Moreover, compounds 1–9 were evaluated for in vitro antimicrobial and cytotoxic activity. Compounds 5b and 8b exhibited antibacterial activity against S. aureus with the MIC value of 31.5 μg/mL, while compound 5b showed cytoxic activities against SF-268, MCF-7, HepG-2 and A549 cell lines with IC50 values of 32.0, 29.5, 39.5 and 33.2 μM, respectively.


Introduction
Pyridone analogues, a family of naturally occurring alkaloids, discovered from a variety of fungal sources have displayed a wide spectrum of fascinating biological activities, including cytotoxic [1,2], antibacterial [3][4][5] and antifungal [6,7] pharmacological actions, which has attracted broad attention from synthetic chemists and pharmacological communities in their search for discovering pharmaceutical drugs [8][9][10][11][12][13]. Regarding the known pyridine alkaloids, citridone A was first discovered as a potentiator for miconazole against Candida albicans with a unique 6-6/5/5-ring system containing a rare phenyl-R-furopyridone skeleton [6]. In recent years, citridone A was chemically synthesized by synthetic chemists [14,15], and these citridone derivatives were also discovered as a kind of novel bacterial-infection blocking agent to inhibit yellow pigment production in methicillin-resistant Staphylococcus aureus (MRSA) [15]. Hence, these new isolates can serve as promising lead compounds or clinical strategies for developing new anti-infective drugs for combating MRSA.
In our continuing efforts to discover structurally unique and biologically meaningful metabolites from fungal resources [16][17][18][19], the strain FS441, which was isolated from the deep-sea sediment sample, showed a remarkable diversity of secondary metabolites in the preliminary HPLC screening, and it was therefore subjected to detailed chemical analysis. As a result, seven new pyridine alkaloids tersones A (1)-G (7), along with two known derivatives citridone A [20] (8a), ent-citridone A [15] (8b) Figure 1. The structures of compounds 1-9.

Structure Elucidation
Tersone A was isolated as white needle crystals. Its molecular formula of C 19 H 19 NO 3 was deduced from its HRESIMS m/z 310.1438 [M + H] + , indicating eleven degrees of unsaturation. The 1 H NMR spectroscopic data together with the HSQC spectra of 1 indicated the presence of the critical phenylpyridone moiety based on the characteristic signals at δ H 7. 34 (1H,s,7.32 (1H,d,J = 8.7 Hz,7.32 (1H,d,J = 8.7 Hz, H-13), 6.79 (1H, d, J = 8.6 Hz, H-10), 6.79 (1H,d,J = 8.6 Hz,, and 5.40 (1H, s, H-1 ); two methines at δ H 3.22 (2H, d, J = 1.8 Hz, H-3) and 2.80 (2H,q,J = 7.4 Hz,, as well as three methyl functional groups at δ H 1.74 (3H, s, H 3 -5 ), 1.63 (3H, s, H 3 -6 ), and 1.27 (3H,d,J = 7.2 Hz,. The 13 C NMR of 1 displayed 19 carbon resonances that were ascribed to an amide carbonyl functionality, twelve olefinic carbons including six quaternary carbons and six methines, one oxygen quaternary carbon, two methine and three methyl moieties. A close comparison of the NMR data of 1 ( Table 1) with those of the co-isolate citridone A [20] tentatively revealed that compound 1 should also share the same phenylfuropyridone carbon skeleton with rare naturally occurring 6-6/5/5 fused ring system. The planar structure of 1 was elucidated on the basis of 1 H-1 H COSY, HSQC, and HMBC experiments ( Figure 2). The following three spin systems, a (C-9/C-10), b (C-12/C-13), and c (C-3/C-3 /C-4 ), were successfully established and attributed to the key 1 H-1 H COSY correlations. Moreover, the HMBC correlations from H-9 to C-11 and C-7, H-6 to C-8, C-7a, and C-4, coupled with the fragments a and b could readily confirm the structure of rings A and B, whereas the rings C and D were rationally established by the independent spin system c in conjunction with the HMBC correlations from H-3 to C-7a and C-3a, H-1 to C-3 and C-3 , as well as H-4 to C-2 . Furthermore, the connections of C-6 to C-2 and C-5 to C-2 were strongly suggested by the HMBC correlations of H 3 -6 to C-2, C-3, C-1 and H 3 -5 to C-1 , C-2 , which were absolutely responsive for the location of the C-6 and C-5 methyl functionalities, respectively. Thus, the gross structure of 1 was unambiguously established.
Mar. Drugs 2019, 17, x 3 of 13 compound 1 should also share the same phenylfuropyridone carbon skeleton with rare naturally occurring 6-6/5/5 fused ring system. The planar structure of 1 was elucidated on the basis of 1 H− 1 H COSY, HSQC, and HMBC experiments ( Figure 2). The following three spin systems, a (C-9/C-10), b (C-12/C-13), and c (C-3/C-3′/C-4′), were successfully established and attributed to the key 1 H− 1 H COSY correlations. Moreover, the HMBC correlations from H-9 to C-11 and C-7, H-6 to C-8, C-7a, and C-4, coupled with the fragments a and b could readily confirm the structure of rings A and B, whereas the rings C and D were rationally established by the independent spin system c in conjunction with the HMBC correlations from H-3 to C-7a and C-3a, H-1′ to C-3 and C-3′, as well as H-4′ to C-2′. Furthermore, the connections of C-6′ to C-2 and C-5′ to C-2′ were strongly suggested by the HMBC correlations of H3-6′ to C-2, C-3, C-1′ and H3-5′ to C-1′, C-2′, which were absolutely responsive for the location of the C-6′ and C-5′ methyl functionalities, respectively. Thus, the gross structure of 1 was unambiguously established. Compound 1 was isolated as a scalemic mixture, and the further chiral resolution by the semi-preparative HPLC fortunately separated them to afford the anticipated enantiomers with a Compound 1 was isolated as a scalemic mixture, and the further chiral resolution by the semi-preparative HPLC fortunately separated them to afford the anticipated enantiomers with a ratio of 1.5:1, which were found to exhibit opposite optical rotations and CD cotton effects. The absolute configuration for the enantiomer 1a was absolutely assigned by the single-crystal X-ray diffraction experiment ( Figure 3). A single crystal was amazingly achieved in methanol/water solvent system, which beautifully set the stage for an efficacious X-ray diffraction analysis with a Flack parameter of 0.09 (10). Then, the further X-ray diffraction experiment completely determined the whole structure and absolute configuration of 1a as 2S,3S,3 R. Therefore, the absolute configuration of 1b was finally assigned to be 2R,3R,3 S.
Mar. Drugs 2019, 17, x 4 of 13 ratio of 1.5:1, which were found to exhibit opposite optical rotations and CD cotton effects. The absolute configuration for the enantiomer 1a was absolutely assigned by the single-crystal X-ray diffraction experiment ( Figure 3). A single crystal was amazingly achieved in methanol/water solvent system, which beautifully set the stage for an efficacious X-ray diffraction analysis with a Flack parameter of 0.09 (10). Then, the further X-ray diffraction experiment completely determined the whole structure and absolute configuration of 1a as 2S,3S,3′R. Therefore, the absolute configuration of 1b was finally assigned to be 2R,3R,3′S. Compound 2 was isolated as white powders and assigned a positive HRESIMS ion peak at m/z 310.1446 ([M + H] + , calcd for C19H20NO3 + ), which was well matched to a molecular formula of C19H19NO3 with eleven degrees of unsaturation in the molecule. The comparison of the 1D NMR spectra of 2 ( Table 1) with those of 1 revealed that they shared the same furopyridone carbon skeleton. One of the two obvious differences was that the methyl group at C-2′ position in 1 was oxidized to be a hydroxy one in 2. This conclusion could be further confirmed by the HMBC correlations from H2-5′ to C-1′, C-2′, and C-3′. However, the other distinctive difference was that the hydroxyl group in the benzene ring A in 1 was replaced by a hydrogen atom in 2, which was further verified by the existence of the crucial 1 H− 1 H COSY spin system a (H-9/H-10/H-11/H-12/H-13). Therefore, the planar structure of 2 was established.
Likewise, compound 2 was also disclosed as a racemate. Based on the previously established methodology, a similar chiral HPLC resolution with YMC-pack Cellulose-SB column (250 × 10 mm × 5 μm) was carried out, which resulted in the successful resolution for the racemates of (±)-2 to afford a pair of enantiomers (−)-tersone B (2a) and (+)-tersone B (2b) with a ratio of 1:1. With reference to the same 1D NMR spectral data, HRMS data, optical rotation values and relative configurations between 2a and the synthesized one, the absolute configuration of compound 2a was considered to be the same as that of (5aS,8R,8aS)-7-(hydroxymethyl)-5a,8-dimethyl-4-phenyl-2,5a,8,8a-tetrahydro-1H-cyclopenta [4,5]furo [3,2-c]pyridin-1-one, which was previously reported as synthesized cirtidone derivative by Tomoda's group [15]. Then, the configuration of 2b was elucidated as 2R,3R,3′S, which could be also corroborated by the CD spectrum with positive cotton effects obviously observed at 210 and 240 nm, and a negative cotton effect at 224 nm. Therefore, the absolute configuration of 2 was also logically established.
Tersone C was obtained as white cubic crystals, and its molecular formula was assigned as C19H19NO3 based on the HRESIMS at m/z 310.1435 [M + H] + (calcd C19H20NO3 + ). The 1 H and 13 C NMR data of 3 (Table 2) closely resembled those of 2, except for the oxidation of the methyl group into a hydroxymethyl one at C-2 position in 3 and the replacement of the hydroxymethyl functionality at C-2′ position in 2 by a methyl one in 3. The aforementioned deduction could be further strengthened by the HMBC correlations from H2-6′ to C-1′, C-2, and C-3, as well as H3-5′ to C-1′, C-2′ and C-3′.
The relative configuration of 3 was determined by a single-crystal X-ray diffraction experiment using CuKα radiation ( Figure 4). Further efforts were performed to determine the absolute configuration of compound 3. Apparently, compound 3 might also be an enantiomeric racemate as it was found to lack the essential optical rotation and circular dichroism effect. Subsequent chiral separation through the semi-preparative HPLC using a YMC-pack Cellulose-SB column (250 × 10 mm × 5 μm) to afford a pair of optically pure enantiomers (−)-tersone C (3a) and (+)-tersone C (3b) with a ratio of approximately 1:1. The absolute configurations for the enantiomers 3a and 3b were  Table 1) with those of 1 revealed that they shared the same furopyridone carbon skeleton. One of the two obvious differences was that the methyl group at C-2 position in 1 was oxidized to be a hydroxy one in 2. This conclusion could be further confirmed by the HMBC correlations from H 2 -5 to C-1 , C-2 , and C-3 . However, the other distinctive difference was that the hydroxyl group in the benzene ring A in 1 was replaced by a hydrogen atom in 2, which was further verified by the existence of the crucial 1 H-1 H COSY spin system a (H-9/H-10/H-11/H-12/H-13). Therefore, the planar structure of 2 was established. Likewise, compound 2 was also disclosed as a racemate. Based on the previously established methodology, a similar chiral HPLC resolution with YMC-pack Cellulose-SB column (250 × 10 mm × 5 µm) was carried out, which resulted in the successful resolution for the racemates of (±)-2 to afford a pair of enantiomers (−)-tersone B (2a) and (+)-tersone B (2b) with a ratio of 1:1. With reference to the same 1D NMR spectral data, HRMS data, optical rotation values and relative configurations between 2a and the synthesized one, the absolute configuration of compound 2a was considered to be the same as that of (5aS,8R,8aS)-7-(hydroxymethyl)-5a,8-dimethyl-4-phenyl-2,5a,8,8a-tetrahydro-1H-cyclopenta [4,5]furo [3,2-c]pyridin-1-one, which was previously reported as synthesized cirtidone derivative by Tomoda's group [15]. Then, the configuration of 2b was elucidated as 2R,3R,3 S, which could be also corroborated by the CD spectrum with positive cotton effects obviously observed at 210 and 240 nm, and a negative cotton effect at 224 nm. Therefore, the absolute configuration of 2 was also logically established.
Tersone C was obtained as white cubic crystals, and its molecular formula was assigned as C 19 (Table 2) closely resembled those of 2, except for the oxidation of the methyl group into a hydroxymethyl one at C-2 position in 3 and the replacement of the hydroxymethyl functionality at C-2 position in 2 by a methyl one in 3. The aforementioned deduction could be further strengthened by the HMBC correlations from H 2 -6 to C-1 , C-2, and C-3, as well as H 3 -5 to C-1 , C-2 and C-3 . The relative configuration of 3 was determined by a single-crystal X-ray diffraction experiment using CuKα radiation ( Figure 4). Further efforts were performed to determine the absolute configuration of compound 3. Apparently, compound 3 might also be an enantiomeric racemate as it was found to lack the essential optical rotation and circular dichroism effect. Subsequent chiral separation through the semi-preparative HPLC using a YMC-pack Cellulose-SB column (250 × 10 mm × 5 µm) to afford a pair of optically pure enantiomers (−)-tersone C (3a) and (    Tersone D was also purified as white powders, and its molecular formula was determined as C 19 .2), as well as seven quaternary carbons, including one ketone carbonyl (δ C 162.5), and six olefinic ones (δ C 129. 7, 108.3, 116.6, 160.8, 134.9 and 133.9). The HMBC correlations from H-10 to C-8, H-7 to C-9, C-8a and C-5 along with the 1 H-1 H COSY fragment of H-10/H-11/H-12/H-13/H-14 collectively pointed to the existence of the key phenylpyridone core. The comparison of the NMR spectroscopic profiles of compound 4 with those of citridone E [5], which was previously confirmed by single crystal X-ray diffraction, revealed close structure similarities with regard to the 6-6/6 ring systems and substituent patterns. The only difference was attributable to the trans double bond at C-3 and C-4 positions in 4 which was replaced by a carboxyl functional moiety in citridone E. This deduction was further confirmed by the chemical shifts of C-2 [δ C 133.9] and C-3 [δ C 126.2, δ H 5.61 (1H, q, J = 6.2 Hz,)] in conjunction with key HMBC correlations from the H 3 -1 to C-2 and C-3 , H 3 -4 to C-2 and C-3 . Moreover, the chemical shifts of C-3 [δ C 129.7] and C-4 [δ C 115.1, δ H 6.50 (1H, s)] as well as the HMBC correlations from H 3 -5 to C-2, C-3, C-4 and H-2 to C-4 further consolidated this conclusion.
In analogy with compounds 1-3, the small specific rotation value and extraordinarily weak CD cotton effect of 4 implied that it was also a racemic mixture. Interestingly, subsequent chiral-phase HPLC separation successfully accessed the two anticipated enantiomers (+)-tersone D (4a) and (−)-tersone D (4b), which were stable at room temperature and showed opposite specific rotation values and electronic circular dichrosim (ECD) curves mutually. However, the potential spontaneous retro 6π electrocyclization/6π electrocyclization cascaded reaction tended to make them as an unseparated enantiomers (as shown in Scheme 1). By comparing the 13 C NMR data of citridone E and citridone F in the literature [5], it was confirmed that the double bond at positions C-2 of compound 4 was the E configuration. In order to unravel their absolute structures, efforts toward conclusive evidence by comparison of the experimental and calculated circular dichroism (CD) spectra using the time-dependent DFT were performed. As depicted in Figure 5, the calculated ECD spectra closely matched the experimental ones, which set the stage for their further absolute structure determinations. Therefore, the absolute configurations of 4a and 4b were clarified as 2R and 2S respectively, and the compound 4 was given the trivial name, tersone D. In analogy with compounds 1-3, the small specific rotation value and extraordinarily weak CD cotton effect of 4 implied that it was also a racemic mixture. Interestingly, subsequent chiral-phase HPLC separation successfully accessed the two anticipated enantiomers (+)-tersone D (4a) and (−)-tersone D (4b), which were stable at room temperature and showed opposite specific rotation values and electronic circular dichrosim (ECD) curves mutually. However, the potential spontaneous retro 6π electrocyclization/6π electrocyclization cascaded reaction tended to make them as an unseparated enantiomers (as shown in Scheme 1). By comparing the 13 C NMR data of citridone E and citridone F in the literature [5], it was confirmed that the double bond at positions C-2′ of compound 4 was the E configuration. In order to unravel their absolute structures, efforts toward conclusive evidence by comparison of the experimental and calculated circular dichroism (CD) spectra using the time-dependent DFT were performed. As depicted in Figure 5, the calculated ECD spectra closely matched the experimental ones, which set the stage for their further absolute structure determinations. Therefore, the absolute configurations of 4a and 4b were clarified as 2R and 2S respectively, and the compound 4 was given the trivial name, tersone D. Calced R Calced S 4a 4b Figure 5. The experimental ECD curves of 4a (solid black line) and 4b (solid red line).
Compound 5 was a yellow powder, its molecular formula was assigned as C19H19NO2 with eleven indices of hydrogen deficiency by the positive HRESIMS ion at m/z 294.1497 [M + H] + (calcd 294.1489). The planar structure of 5 was ultimately determined by analysis of its NMR spectra (Table  3)  In analogy with compounds 1-3, the small specific rotation value and extraordinarily weak CD cotton effect of 4 implied that it was also a racemic mixture. Interestingly, subsequent chiral-phase HPLC separation successfully accessed the two anticipated enantiomers (+)-tersone D (4a) and (−)-tersone D (4b), which were stable at room temperature and showed opposite specific rotation values and electronic circular dichrosim (ECD) curves mutually. However, the potential spontaneous retro 6π electrocyclization/6π electrocyclization cascaded reaction tended to make them as an unseparated enantiomers (as shown in Scheme 1). By comparing the 13 C NMR data of citridone E and citridone F in the literature [5], it was confirmed that the double bond at positions C-2′ of compound 4 was the E configuration. In order to unravel their absolute structures, efforts toward conclusive evidence by comparison of the experimental and calculated circular dichroism (CD) spectra using the time-dependent DFT were performed. As depicted in Figure 5, the calculated ECD spectra closely matched the experimental ones, which set the stage for their further absolute structure determinations. Therefore, the absolute configurations of 4a and 4b were clarified as 2R and 2S respectively, and the compound 4 was given the trivial name, tersone D. Calced R Calced S 4a 4b Figure 5. The experimental ECD curves of 4a (solid black line) and 4b (solid red line).
Compound 5 was a yellow powder, its molecular formula was assigned as C19H19NO2 with eleven indices of hydrogen deficiency by the positive HRESIMS ion at m/z 294.1497 [M + H] + (calcd 294.1489). The planar structure of 5 was ultimately determined by analysis of its NMR spectra (Table Compound 5 was a yellow powder, its molecular formula was assigned as C 19 H 19 NO 2 with eleven indices of hydrogen deficiency by the positive HRESIMS ion at m/z 294.1497 [M + H] + (calcd 294.1489). The planar structure of 5 was ultimately determined by analysis of its NMR spectra (Table 3) involving HSQC, 1 H-1 H COSY, and HMBC correlations. The 1 H-1 H COSY spectrum of 5 displayed two proton spin−spin systems in the core structure affiliated to a (H-9/H-10/H-11/H-12/H-13) and b (H-3/H-1 /H 3 -6 ) as depicted with bold lines in Figure 2. Moreover, the HMBC correlations from H-6 to C-4, C-7a, and C-8 and H-9 to C-5 coupled with the fragment a successfully which helped to reveal the presence of the 3-phenylpyridin-4(1H)-one scaffold. Futhermore, the fragment b together with the HMBC correlations from H-1 to C-2, C-3 and C-3a, from H 3 -6 to C-3 and C-2 , from H-4 to C-2, C-3 and C-3 and from H-5 to C-1 and C-2 , unambiguously constructed the cyclopentadienyl furan ring system. Then, the planar structure of 5 was established as shown in Figure 1. The relative configuration of 5 was established by the NOESY experiment, wherein the NOE correlations between H 3 -4 to H-3 and H 3 -6 to H-3 indicated that the protons of H-3, H 3 -4 , and H 3 -6 ought to be oriented in the same direction ( Figure 6). Furthermore, the chiral-phase HPLC analysis of 5 revealed it to be also a racemate, and the further semi-preparative chiral HPLC separation successfully resulted in the resolution of the two corresponding enantiomers (−)-tersone E (5a) and (+)-tersone E (5b). The compound 5b was previously synthesized by Tomoda's group [15], which was confirmed by the comparison of 1D NMR spectral data, HRMS data and the optical rotation values with those of the synthesized one. Finally, the absolute configuration of 5b was established to be the same as (4bS,5R,7aS)-5,6,7a-trimethyl-3-phenyl-4b,5-dihydro-1H-cyclopenta [4,5]furo [2,3-b]pyridin-4(7aH)one [15], which was defined as 2S,3S,3 R in this study. Therefore, the absolute stereochemistry of 5a could be further determined as 2R,3R,3 S. was confirmed by the comparison of 1D NMR spectral data, HRMS data and the optical rotation values with those of the synthesized one. Finally, the absolute configuration of 5b was established to be the same as (4bS,5R,7aS)- 5,6,7a-trimethyl-3-phenyl-4b,5-dihydro-1H-cyclopenta[4,5]furo [2,3-b]pyridin-4(7aH)-o ne [15], which was defined as 2S,3S,3′R in this study. Therefore, the absolute stereochemistry of 5a could be further determined as 2R,3R,3′S. Tersone F was isolated as white powders. Its HRESIMS data was disclosed by the obvious positive ion peak at m/z 344.1494 [M + H] + in the HRESIMS spectrum, and it corresponded to a Tersone F was isolated as white powders. Its HRESIMS data was disclosed by the obvious positive ion peak at m/z 344.1494 [M + H] + in the HRESIMS spectrum, and it corresponded to a molecular formula of C 19 H 21 NO 5 . The 1 H NMR data of 6 (  4.13 (1H, d, J = 2.9 Hz, H-9) and 5.14 (1H, d, J = 2.9 Hz, H-10)], and five olefinic proton signals at [δ H 7.39 (1H,d,J = 6.9 Hz,7.37 (1H,t,J = 6.9 Hz,7.28 (1H,t,J = 6.9 Hz,7.37 (1H,t,J = 6.9 Hz,and 7.39 (1H,d,J = 6.9 Hz,]. The planar structure of 6 was defined by the analysis of its 2D NMR spectra. Firstly, the COSY and HSQC correlations revealed the presence of the two structural fragments, a (C-12/C-13/C-14/C-15/C- 16) and b (C-9/C-10). Secondly, the HMBC correlations from H-10 to C-8, C-11, and C-16 as well as H-9 to C-6, C-8 and C-11, coupling with the fragments a and b, led to tentative establishment of the connection between the mono-substituted benzene ring and the pyrrolidine-2,4-dione ring. Moreover, the cyclopent-2-en-1-one subunit was constructed by the cross peaks from H 2 -4 to C-2, C-5 and C-7, H 3 -17 to C-1 and C-2, H 3 -18 to C-2, C-3 and C-4, as well as H 3 -19 to C-5. These aforementioned informative results collectively indicated that compound 6 was structurally related to citridone C [20], excepting the obvious chemical shift changes at C-7 and C-8 positions. With a careful inspection and interpretation of their NMR data, the major difference between them was that the quaternary enol carbons at C-7 and C-8 positions in citridone C were replaced by two quaternary carbons bearing hydroxyl and carbonyl functionalities, respectively. The deduction was evidenced by the chemical shift of C-7 [δ C 76.1], C-8 [δ C 208.9] as well as the HMBC correlations from H-9 to C-8, H-10 to C-8, and H 3 -19 to C-7. Therefore, the planar structure of 6 was completely interpreted as shown in Figure 1. The relative configuration of 6 was not determined because the active protons of OH and NH had a lack of signals in the 1 H NMR and NOESY spectra.
Compound 7 was obtained as a yellow powder, and its molecular formula was established as C 14 H 11 NO 2 with nine indices of hydrogen deficiency based on the HRESIMS with a positive deprotonated molecular ion [M + H]  The 1D NMR (Table 4) and HSQC spectroscopic data of 7 revealed the existence of fourteen carbon resonances ascribable to one methyl (δ C 13.6), seven olefinic methines (δ C 103. 3, 128.6, 128.6, 129.9, 128.9, 129.9, and 128.6), and six quaternary carbons (δ C 156.4, 118.4, 113.2, 161.6, 133.9 and 159.6). The above-mentioned information indicated that compound 7 might be also a phenylpyridone derivative. The HMBC correlations from H-6 to C-4, C-7a and C-8, H-9 to C-7 coupling with the critical COSY fragment H-9/H-10/H-11/H-12/H-13, confirmed the construction of the 5-phenylpyridone moiety. Additionally, the HMBC correlations from H 3 -1 to C-2, C-3 and C-3a, and H-3 to C-3a and C-7a suggested the presence of the 2-methylfuran moiety. Therefore, the structure of 7 was finally established, which represented the first member of 5-phenylpyridone derivatives with an unprecedented furo [3,2-c]pyridin-4(5H)-one skeleton as shown in Figure 1, and given the trivial name tersone G. In addition, the known phenylpyridone analogues were elucidated as citridone A [20] (8a), ent-citridone A [15] (8b) and 196 [3] (9) by comparison of their NMR spectra, MS data, and specific rotation values with those of the reported literature.

Biological Activity
The minimum inhibitory concentration (MIC) for compounds 1-9 against bacteria Staphylococcus aureus ATCC 29,213 and Escherichia coli ATCC 8739 was evaluated by means of the microdilution susceptibility test, and performed in Mueller−Hinton broth (MHB) following CLSI guidelines by using the vancomycin as the positive control. Compounds 5b and 8b exhibited antibacterial activity against S. aureus with the MIC value of 31.5 µg/mL. However, compounds 1-3, 6-7, and 9 were found to be devoid of significant activity (Table 5). Additionally, compounds 1-9 were also tested for their cytotoxic activities against four human tumor cell lines (SF-268, MCF-7, HepG-2 and A549). As a result, 5b showed cytotoxic activities against SF-268, MCF-7, HepG-2 and A549 cell lines with IC 50 values of 32.0, 29.5, 39.5 and 33.2 µM, respectively, while the IC 50 values of the positive control drug cisplatin against the four cell lines were 3.0, 2.6, 2.4, 1.9 µM, respectively. The other compounds were inactive with IC 50 values more than 100 µM.

Fungal Material
The fungal strain FS441 used in this work was isolated from a sediment sample, which was collected from the depth of 3000 m in the Indian Ocean (88 • 58 640" E, 0 • 00 307" S), in April 2016. The isolated strain was deposited in the Guangdong Provincial Key Laboratory of Microbial Culture Collection and Application, Guangdong Institute of Microbiology. The strain was identified by using BLAST (nucleotide sequence comparison program) to search the GenBank database, which had 98.9% similarity to Phomopsis tersa SYJM09 (GenBank Accession No. JF923840). Hence, the strain was identified as Phomopsis tersa (GenBank Accession No. MK592793).

Fermentation and Extraction
The marine fungus P. tersa FS441 was cultured on potato dextrose agar (PDA) at 28 • C for 7 days to prepare the seed culture, and then inoculated into flasks (3 L) containing 9 g sea salts, 250 g of rices and 300 mL of waters. After that, all flasks were incubated at 28 • C for one month and the fermented rice substrate was extracted repeatedly with EtOAc. After the evaporation of the solvent, a dark brown solid (177.9 g) was obtained. The crude extract was fractionated by silica gel column chromatography (100-200 mesh) with two gradient systems of increasing polarity (petroleum ether/EtOAc, 10:1→1:1; CH 2 Cl 2 /CH 3 OH, 10:1→0:1) to furnish seven fractions (A-G).