Secondary Metabolites with α-Glucosidase Inhibitory Activity from Mangrove Endophytic Fungus Talaromyces sp. CY-3

Eight new compounds, including two sambutoxin derivatives (1–2), two highly oxygenated cyclopentenones (7–8), four highly oxygenated cyclohexenones (9–12), together with four known sambutoxin derivatives (3–6), were isolated from semimangrove endophytic fungus Talaromyces sp. CY-3, under the guidance of molecular networking. The structures of new isolates were elucidated by analysis of detailed spectroscopic data, ECD spectra, chemical hydrolysis, 13C NMR calculation, and DP4+ analysis. In bioassays, compounds 1–5 displayed better α-glucosidase inhibitory activity than the positive control 1-deoxynojirimycin (IC50 = 80.8 ± 0.3 μM), and the IC50 value was in the range of 12.6 ± 0.9 to 57.3 ± 1.3 μM.


Introduction
According to the WHO forecast, the number of diabetes patients will reach 693 million in 2045. Type II diabetes accounts for 90%, and α-glucosidase inhibitors originating from natural products, such as acarbose, miglitol, and voglibose, are used to treat type II diabetes [1]. However, most clinical antidiabetic drugs cause side effects [2]. Therefore, there is an urgent need to find and discover new antidiabetic drugs.
Recently, the advent of visual molecular network technology has led to a new perspective in the research of natural products [15]. Global Natural Product Social (GNPS) can establish a molecular network to classify compounds with the same LC-MS/MS ion fragments into similar clusters. Moreover, it can rapidly discover novel compounds through accurate MS data and database comparison [16].

Structure Identification
Sambutoxin A (1), obtained as a light-yellow oil, was displayed to have a molecular formula of C28H37NO4 with 10

Structure Identification
Sambutoxin A (1), obtained as a light-yellow oil, was displayed to have a molecular formula of C 28  HMBC from H-6 to C-2, C-4, and C-5 and from H-23 to C-2 and C-6, constructed a ring B moiety. 1 H-1 H COSY H-2 /H-3 /H-4 /H-5 /H-6 , together with HMBC from H-2 to C-1 and from H-6 to C-1 , constructed ring A. Finally, rings A, B, C, and side chain D were connected by the HMBCs from H-6' to C-5, H-7 to C-2, C-3, and C-4, and H-21 to C-11. The 1D and 2D NMR data were similar to those of 3 ( Table 1). The only difference between them was that the 4 -OH (δ C 156.5) is reduced to a hydrogen atom (δ H 7.29-7.34, δ C 127.5). Thus, the planar structure of 1 was shown ( Figure 3).  The relative configuration of ring C and the double bond between C-12 and C-13 of compound 1 were defined by the NOESY spectrum. The correlations of H-22/H-7/H-11 and H-14/H-21 were also observed in the NOESY spectrum ( Figure 4), which means H-22, H-7, and H-11 were on the same side, as were H-14 and H-21. Therefore, the relative configuration of C-7, C-10, and C-11 in ring C was deduced to be (7S*, 10R*, 11R*), and the double bond between C-12 and C-13 was assigned to be (E). In the present work, assigning the configurations at C-14 and C-16 in the aliphatic side chains of 1 was a challenging task due to the high conformational flexibility of fatty chains. As described in the literature, the absolute side-chain configurations for C-14 and C-16 were also determined by 13 C NMR calculation and DP4+ analysis [17,18]. (14R, 16S)-1 was assigned with a 100% probability ( Figures S68 and S69). Consequently, the side-chain configurations of C-14 and C-16 of 1 were assigned to be 14R, 16S. The absolute configuration of 1 was determined by comparing the calculated ECD spectra (7S, 10R, 11R, 14R, 16S)-1 and (7R, 10S, 11S, 14R, 16S)-1 with the experimental one. The calculated ECD curves (7S, 10R, 11R, 14R, 16S)-1 showed better agreement with the experimental one ( Figure 5A). Thus, the absolute configuration of 1 was assigned to be 7S, 10R, 11R, 13E, 14R, 16S. Compounds 1 and 3 share a common biosynthetic pathway for sambutoxin derivatives, which is consistent with the configuration reported in the literature [19,20].
Sambutoxin B (2), also isolated as a light-yellow oil, displayed a molecular formula of C28H37NO4 with 11 degrees of unsaturation at m/z 452.2794 [M+H] + (calcd. 452.2795) by positive HR-ESI-MS spectrum. Its 1D and 2D NMR were similar to 3. The only difference between them was that the single bond between C-9 (δH 1.44, 1.97, δC 32.2) and C-10 (δH 1.64-1.70, δC 32.1) was converted to a double bond (δH 5.72, δC 121.9, 132.9). Thus, the planar structure of 2 was as shown in Figure 3. The correlations of H-7/H-11 and H-14/H-21 were observed in the NOESY spectrum ( Figure 4). Thus, the double bond between C-12 and C-13 was assigned to be (E), and the relative configuration was deduced to be (7S*, 11R*). The calculated ECD spectra of (7S, 11R)-2 and (7R, 11S)-2 were compared to the measured one, and the calculated ECD curve of (7S, 11R)-2 was showed a good agreement with the experimental one ( Figure 5B). The stereochemistry of C-14 and C-16 was biogenetically established as 14R, 16S. Moreover, they were also verified by 13 C NMR calculation and DP4+ probability ( Figures S70 and S71). Thus, the absolute configuration of 2 was determined as 7S, 11R, 13E, 14R, 16S. The relative configuration of ring C and the double bond between C-12 and C-13 of compound 1 were defined by the NOESY spectrum. The correlations of H-22/H-7/H-11 and H-14/H-21 were also observed in the NOESY spectrum ( Figure 4), which means H-22, H-7, and H-11 were on the same side, as were H-14 and H-21. Therefore, the relative configuration of C-7, C-10, and C-11 in ring C was deduced to be (7S*, 10R*, 11R*), and the double bond between C-12 and C-13 was assigned to be (E). In the present work, assigning the configurations at C-14 and C-16 in the aliphatic side chains of 1 was a challenging task due to the high conformational flexibility of fatty chains. As described in the literature, the absolute side-chain configurations for C-14 and C-16 were also determined by 13 C NMR calculation and DP4+ analysis [17,18]. (14R, 16S)-1 was assigned with a 100% probability ( Figures S68 and S69). Consequently, the side-chain configurations of C-14 and C-16 of 1 were assigned to be 14R, 16S. The absolute configuration of 1 was determined by comparing the calculated ECD spectra (7S, 10R, 11R, 14R, 16S)-1 and (7R, 10S, 11S, 14R, 16S)-1 with the experimental one. The calculated ECD curves (7S, 10R, 11R, 14R, 16S)-1 showed better agreement with the experimental one ( Figure 5A). Thus, the absolute configuration of 1 was assigned to be 7S, 10R, 11R, 13E, 14R, 16S. Compounds 1 and 3 share a common biosynthetic pathway for sambutoxin derivatives, which is consistent with the configuration reported in the literature [19,20].
Talaketides C (9) was isolated as a yellow oil, and its molecular formula was determined as C 15 H 24 O 5 by the HR-ESI-MS data at m/z 307.1517 [M+Na] + (calcd. 307.1516). The 1D and 2D NMR were also similar to 7, and the only difference between them was compound 9 had one more methoxy group (δ H 3.16, δ c 51.8) and one more methylene group (δ H 2.82, 3.03, δ c 31.2) than compound 7. According to the HMBCs, the methylene group was connected oxymethine (δ H 5.10, δ c 71.7) and a quaternary carbon (δ c 188.2), and methoxy was connected to oxymethine (δ H 5.10, δ c 71.7), thus the planar structure of 9 was established ( Figure 3). The NOESY correlation of H-5 to H-12 was observed (Figure 4), and the ECD curves of (2R, 3R) and (2S, 3S) were compared to the measured one. The calculated ECD of (2R, 3R) showed a good agreement with the experimental one ( Figure 5E), and the side-chain stereostructure of C-10 was also determined as 10S through chemical hydrolysis. (Figure 6 and Figure S72). Therefore, the absolute configuration of 9 was determined as 2R, 3R, 10S.
Talaketides D (10) was isolated as a light-yellow oil, and its molecular formula was determined as C 9 H 14 O 4 by the HR-ESI-MS data at m/z 209.0783 [M+Na] + (calcd. 209.0784). Analysis of the 1D and 2D NMR spectrum found compound 10 to be similar to the known compound phomaligol D [21]. The only difference between them was the departure of hydroxyl at C-4 (δ c 73.1 changed to δ H 2.55, δ c 41.5). Thus, the planar structure of 10 was determined (Figure 3), which was dehydroxylated phomaligol D. H-3/H-8 and H-4/H-7 correlations were observed in the NOESY spectrum, which indicated that the relative configuration of 10 was (2S*, 3R*, 4R*) ( Figure 4). In order to determine its absolution configuration, the ECD spectra of (2S, 3R, 4R)-10 and (2R, 3S, 4S)-10 were compared with the measured one. The calculated CD curve of (2S, 3R, 4R) showed a good agreement with the experimental one ( Figure 5F). Therefore, the absolution configuration of 10 was deduced to be (2S, 3R, 4R) and named Talaketides D. Talaketides A(7) was isolated as a yellow oil, and its molecular formula was determined as C13H20O5 by the HR-ESI-MS data at m/z 279.1197 [M+Na] + (calcd. 279.1203). The 1 H NMR of 7 displayed four methyls, one methylene, two methines, and one methoxy (Table 2). Its 13 C NMR showed a total of 13 carbons resonances, including four methyls, one methylene, two methines, one methoxy, one ester carboxyl, one carbonyl, two     6. Confirmation for C-10 of compounds 7 and 9 through chemical hydrolysis.

Proposed Biosynthesis Pathway
A hypothetical biosynthetic pathway for compounds 1-2 and 7-12 was proposed ( Figure 7) [25,26]. Compounds 1-2 are the PKS-NRPS biosynthetic pathway. Starting from one L-Phe molecule, one acetyl-CoA molecule, six malonyl-CoA molecules, and four SAM molecules through the PKS pathway formed intermediate i. Then, the formation of 1 was constructed by rearranging and reducing. The formation of 2 was similar to 1, and the only difference between them was that L-Phe was replaced by L-Tyr ( Figure 7A). ECD calculation and comparison, and the absolution configuration of 11 and 12 was confirmed as (2R, 3R, 4R) and (2S, 3S, 4R), respectively ( Figure 5G).

Proposed Biosynthesis Pathway
A hypothetical biosynthetic pathway for compounds 1-2 and 7-12 was proposed ( Figure 7) [25,26]. Compounds 1-2 are the PKS-NRPS biosynthetic pathway. Starting from one L-Phe molecule, one acetyl-CoA molecule, six malonyl-CoA molecules, and four SAM molecules through the PKS pathway formed intermediate i. Then, the formation of 1 was constructed by rearranging and reducing. The formation of 2 was similar to 1, and the only difference between them was that L-Phe was replaced by L-Tyr ( Figure 7A).
The remaining compounds 7-12 are considered to be the origin of biosynthetic polyketides. The key intermediate vi was obtained through the PKS pathway, oxidation, and formed 10-12 by electron transfer, methylation, and epimerization. Further electron transfer, methylation, and esterification to form 9. 7 was formed through electron transfer, methylation, epimerization, methyltransferase, and esterification. Together, compound 8 started from two SAM molecules and three acetyl-CoA molecules through the PKS pathway and further rearranging and methylating ( Figure 7B).

α-Glucosidase Inhibitory Activity
Compounds 1−12 were tested for their α-glucosidase inhibitory activity (Table 4). Compounds 1−5 displayed better α-glucosidase inhibitory activity with an IC50 value in the range of 12.6 ± 0.9 to 57.3 ± 1.3 μM compared to the positive control 1-deoxynojirimycin (IC50 = 80.8 ± 0.3 μM). The IC50 value of compound 2 is 37.4 ± 1.4 μM, lower than that of compound 1 (12.6 ± 0.9 μM) and compound 3 (16.9 ± 0.6 μM), which illustrates that the double bond formed between C-9 and C-10 reduced the α-glucosidase inhibitory activity. Different from compounds 1−3, compound 5 presented a much bigger IC50 value of 57.3 ± 1.3 μM, and the IC50 of compound 6 was even bigger than 100 μM. Therefore, it can be consid- The remaining compounds 7-12 are considered to be the origin of biosynthetic polyketides. The key intermediate vi was obtained through the PKS pathway, oxidation, and formed 10-12 by electron transfer, methylation, and epimerization. Further electron transfer, methylation, and esterification to form 9. 7 was formed through electron transfer, methylation, epimerization, methyltransferase, and esterification. Together, compound 8 started from two SAM molecules and three acetyl-CoA molecules through the PKS pathway and further rearranging and methylating ( Figure 7B).
The IC 50 value of compound 2 is 37.4 ± 1.4 µM, lower than that of compound 1 (12.6 ± 0.9 µM) and compound 3 (16.9 ± 0.6 µM), which illustrates that the double bond formed between C-9 and C-10 reduced the α-glucosidase inhibitory activity. Different from compounds 1-3, compound 5 presented a much bigger IC 50 value of 57.3 ± 1.3 µM, and the IC 50 of compound 6 was even bigger than 100 µM. Therefore, it can be considered that the branch chain attached to ring C contributes a lot to the inhibitory activity. Compound 4 showed relatively higher inhibitory activity (IC 50 = 16.5 ± 0.7 µM), which may be due to its different structure from the other five.

Molecular Docking Study
To explain the difference in inhibitory activity of compounds 1-5 to α-glucosidase, molecular docking between them and α-glucosidase was carried out using Autodock. The interaction energies of compounds 1-5 with α-glucosidase were 8. 25, 7.99, 8.85, 9.19, and 7.96 kcal/mol, respectively, which is consistent with the change in IC 50 value. Compound 1 mainly formed a hydrogen bond with Glu-411 with a bond length of 2.85 Å ( Figure 8A

General Experimental Procedures
The 1D and 2D NMR were recorded on a Bruker Avance 400 MHz spectrometer (Karlsruhe, Germany) at room temperature. HR-ESI-MS spectra of all test compounds were acquired on a ThermoFisher LTQ-Orbitrap-LC-MS spectrometer (Palo Alto, CA, USA). UV-vis spectra were measured on a Shimadzu UV-2600 spectrophotometer

General Experimental Procedures
The 1D and 2D NMR were recorded on a Bruker Avance 400 MHz spectrometer (Karlsruhe, Germany) at room temperature. HR-ESI-MS spectra of all test compounds were acquired on a ThermoFisher LTQ-Orbitrap-LC-MS spectrometer (Palo Alto, CA, USA). UV-vis spectra were measured on a Shimadzu UV-2600 spectrophotometer (Kyoto, Japan). Optical rotations were acquired on an Anton-Paar MCP500 automatic polarimeter at 25°C (Graz, Austria). CD curves were recorded on an Applied Photophysics Chirascan spectropolarimeter (Surrey, UK). All spectrophotometric measurements used a 96-well Bio-Rad microplate reader (Hercules, CA, USA). Solvent was removed by a Heidolph rotavapor with a vacuum pump. Semipreparative HPLC chromatography was used on a U3000 separation module coupled with a DAD detector manufactured by ThermoFisher and a chiral semipreparative column (Nu-Analytical Solutions Co., LTD-packed chiral INB, 5 µm, 4.6 × 250 mm) was used for separation. Column chromatography (CC) used silica gel (200-300 mesh (Qingdao Marine Chemical Factory)) and Sephadex LH-20 (Amersham Pharmacia, Stockholm, Sweden). Precoated silica gel plates (Qingdao Huang Hai Chemical Group Co., G60, F-254) were used for TLC analysis. LC-MS analysis was performed on a Q-TOF manufactured by Waters and a Waters Acquity UPLC BEH C18 column (1.7 µm, 2.1 × 100 mm) was used for analysis.

Fungal Material
Fungus CY-3 was isolated from the fresh leaves of the semimangrove Hibiscus tiliaceus (collected in June 2020 from Zhanjiang Mangrove National Nature Reserve in Guangdong Province, China). It was identified as Talaromyces sp. using ITS gene sequencing. The ITS rDNA gene sequence data of the fungi were deposited to GenBank (Accession No. MZ614621), and fungus CY-3 was deposited at Sun Yat-Sen University, China.

Fermentation
CY-3 was activated on a potato dextrose agar (PDA) Petri dish at 28°C, then cultured in potato dextrose broth (PDB) in 6 × 500 mL Erlenmeyer flasks at 28°C for 3 days in a shaker to obtain spore inoculum. The routine-scale fermentation was performed in 60 × 1 L Erlenmeyer flasks, each containing 50 mL of 2% saline and 50 g of rice. The Erlenmeyer flask containing the culture medium was autoclaved at 121°C for 25 min. After cooling to room temperature, 10 mL of CY-3 inoculum was inoculated in each bottle and incubated at room temperature for 30 days.

Molecular Networking
The crude extract of CY-3 was analyzed by LC-MS/MS (LTQ Velos Pro-Orbitrap, Waltham, MA, USA) and a C 18 column (Thermo Fisher Scientific-packed Hypersil GOLD, 1.9 µm, 2.1 × 100 mm). Samples were dissolved in MeCN at 1 mg/mL. A  and with an automated fully dependent MS/MS scan enabled. The molecular networking were made as described previously [27,28].

ECD and 13 C NMR Calculations
ECD calculations and 13C NMR calculations were performed by the Gaussian 09 program and Spartan'14. The conformation with a Boltzmann population greater than 5% was selected for optimization and calculation in methanol at B3LYP/6-31+G (d, p). The ECD spectra were generated by the program SpecDis 1.6 (University of Würzburg, Würzburg, Germany) and drawn by OriginPro 8.0 (OriginLab, Ltd., Northampton, MA, USA) from dipole-length rotational strengths by applying Gaussian band shapes with sigma = 0.30 eV [29,30].

Bioassay
Compounds 1-12 were evaluated for α-glucosidase inhibitory activity, as described previously [31]. Three parallel concentrations of 1-deoxynojirimycin were taken as positive controls. DMSO was used as blank controls.