New Dihydroisocoumarin Root Growth Inhibitors From the Sponge-Derived Fungus Aspergillus sp. NBUF87

Six new dihydroisocoumarins, aspergimarins A−F (1−6), were discovered together with five known analogs (7−11) from a monoculture of the sponge-derived fungus Aspergillus sp. NBUF87. The structures of these compounds were elucidated through comprehensive spectroscopic methods, and absolute configurations were assigned after X-ray crystallography, use of the modified Mosher’s method, and comparison of electronic circular dichroism (ECD) data with literature values for previously reported analogs. Compounds 1−11 were evaluated in a variety of bioassays, and at 100 μM, both 1 and 5 showed significant inhibitory effects on the lateral root growth of Arabidopsis thaliana Columbia-0 (Col-0). Moreover, at 100 μM, 5 also possessed notable inhibition against the primary root growth of Col-0. Meanwhile, 1−11 were all found to be inactive in vitro against acetylcholinesterase (AChE) (IC50 > 100 μM), four different types of human-derived cancer cell lines (IC50 > 50 μM), as well as methicillin-resistant Staphylococcus aureus and Escherichia coli (MIC > 50 μg/mL), and Plasmodium falciparum W2 (EC50 > 100 μg/mL), in phenotypic tests.


INTRODUCTION
Sponges are among the most primitive multicellular invertebrates and harbor vast microbial populations, owing largely to the unique filter-feeding physiology that is full of pores and channels (Hentschel et al., 2006;Webster and Taylor, 2012). As a result of the long-standing interaction and coevolution with sponges, marine symbiotic microorganisms have differentiated from those of terrestrial origins in terms of their biosynthetic pathways that lead to the production of structurally interesting and biologically active compounds (Thomas et al., 2010;Fan et al., 2012). Therefore, sponge-associated microbes have become an exciting area of drug discovery research (Thomas et al., 2010;Pita et al., 2016). Endophytic fungi that are associated with sponges, especially members of the genus Aspergillus, have been recognized as a source of structurally diverse natural products with biological activities that provide value for drug discovery (Blunt et al., 2018;Zhang et al., 2018). In the past decade, the secondary metabolites from sponge-derived Aspergillus fungi have been reported from many classes, including polyketides (Wang et al., 2014;Kong et al., 2015), terpenoids (Liu et al., 2009;Li D. et al., 2012), alkaloids (Zhou et al., 2013(Zhou et al., , 2014, diketopiperazines (Ahmed et al., 2017), and peptides (Lee et al., 2011). Many of these metabolites have been shown to exhibit strong antitumor, antibacterial, antiviral, and other bioactivities.
As part of a continuing research program investigating the biologically active secondary metabolites from sponge-derived fungi (Ding et al., 2018;Huang et al., 2019;Li W. et al., 2019), a detailed chemical investigation was initiated on the culture of fungus Aspergillus sp. NBUF87. The fungus was isolated from a South China Sea marine sponge of the genus Hymeniacidon. The EtOAc extract of the culture of fungus Aspergillus sp. NBUF87 exhibited inhibitory effects on the root growth of Arabidopsis thaliana Columbia-0 (Col-0), a typical model organism for studying plant growth and development. The separation and purification of the bioactive extract led to the discovery of six new dihydroisocoumarin compounds (1−6) and five known analogs (7−11) (Figure 1). Herein, the detailed isolation and structure elucidation of these dihydroisocoumarin derivatives, together with the evaluation of their inhibitory effects against the root growth of Col-0 and a preliminary broader biological activity screening are described.

General Experimental Procedures
Optical rotation measurements were conducted with a JASCO P-2000 digital polarimeter. IR and UV spectra were obtained with a Thermo Scientific Nicolet iS5 FT-IR spectrometer and a Thermo Scientific Evolution 201 spectrophotometer, respectively. Electronic circular dichroism (ECD) spectra were collected on a JASCO J-1500 spectrophotometer. 1D and 2D NMR spectra were recorded in DMSO-d 6 (or CDCl 3 ) with a Palo Alto Varian 600 MHz spectrometer, using standard pulse sequences. HRESIMS data were collected on an Agilent Technologies 6224 TOF MS. X-ray single-crystal diffraction data were acquired using an Agilent Gemini Ultra diffractometer with Cu Kα radiation (λ = 1.54178 Å). Medium-pressure liquid chromatography (MPLC) was performed using a Bonna-Agela FLEXA purification instrument. Column chromatography (CC) was carried out with silica gel (200−300 mesh, Qingdao) and Amersham Biosciences Sephadex LH-20. Reversed-phase HPLC (RP-HPLC) was conducted using a Waters 1525 binary HPLC pump equipped with a Waters 2996 photodiode array detector and a YMC-Pack C18 column (YMC, 20 × 250 mm, 5 µm).

Fungal Material
The fungus Aspergillus sp. NBUF87 was isolated from the sponge Hymeniacidon sp. obtained from the Paracel Islands in the South China Sea, and was determined as being Aspergillus sp. by its morphology and gene sequence (ITS rDNA region) analyses (GenBank accession no. MH595747.1). The strain specimen was deposited in PDB medium to the repository conserved at the College of Food and Pharmaceutical Sciences, Ningbo University, China.

Preparation of MTPA Esters of 3−5 for Modified Mosher's Analysis
Under an atmosphere of nitrogen, pyridine-d 5 (500 µL) and (R)-MTPA-Cl (8 µL) was sequentially added to an EP tube containing compounds 3, 4, or 5 (1.0 mg), separately. The mixture was shaken at 28 • C for 12 h and then purified by RP-HPLC to obtain the (S)-MTPA esters 3a, 4a, and 5a. By the same procedure, the (R)-MTPA esters 3b, 4b, and 5b were obtained using (S)-MTPA-Cl as a reagent. Key 1 H NMR signals used for configurational assignments were determined by respective 1 H− 1 H COSY correlations and the already completed full assignments of 1 H NMR data for 3, 4, and 5 (see Supplementary  Figures S61-S72). The

Plant Growth Response Assays
Arabidopsis thaliana Col-0, a model organism for plant growth and development, was used to test each isolated compound according to a previously described protocol Huang et al., 2019). Plants grown in 2% (v/v) DMSO were used as the negative control. Seeds of Col-0 incubated in 1 µM of 6benzylaminopurine (BAP) were selected as the positive control. Test samples were dissolved in 2% (v/v) DMSO at various test concentrations for the experiment.

In vitro AChE Activity Assays
For compounds 1−11, the in vitro acetylcholinesterase (AChE) activity was assessed by the colorimetric method in 96-well plates according to a previously reported method (Santos et al., 2012). Donepezil was selected as the positive control with IC 50 value of 11.9 nM .

In vitro Cancer Cell Cytotoxicity Assays
The in vitro cytotoxic activities of all isolated compounds against four human cancer cell lines (CCRF-CEM, MDA-MB-231, HCT-116, and AGS) were evaluated by the MTT method as previously described Li W. et al., 2019). 7-Ethyl-10hydroxycamptothecin (1.3, 10.8, 9.9, and 4.2 nM, respectively) was used as the positive control against four above-mentioned human cancer cell lines.

Antibacterial Activity Assays
Compounds 1−11 were evaluated for their antibacterial activities against methicillin-resistant Staphylococcus aureus ATCC43300 and Escherichia coli ATCC25922 in 96-well plates according to the method described by Gu et al. (2018). Ciprofloxacin was selected as a positive control against the above-mentioned bacteria with MIC values of 0.5 µg/mL.

In vitro Antimalarial Activity Assays
The in vitro antimalarial activity of the compounds was evaluated against the parasite (Plasmodium falciparum W2), which was cultured continuously according to a previously described method (Lazaro et al., 2006). Chloroquine, atovaquone, and artemisinin were used as positive controls against the above parasite with EC 50 values of 112, 2.5, and 160 nM, respectively.

Structure Elucidation
Compound 1 was isolated as white crystals.  (Table 1), including a carbonyl at δ C 169.5 (C-1), six aromatics at δ C 153.9 (C-8), 146.6 (C-5), 124.5 (C-4a), 123.9 (C-6), 115.1 (C-7), and 108.2 (C-8a), and seven alkyl carbons that were two oxygenated methines at δ C 79.4 (C-3) and 65.6 (C-4 ), four hydrocarbon methylenes at δ C 38.6 (C-3 ), 34.3 (C-1 ), 26.3 (C-4), and 20.8 (C-2 ), and a methyl at δ C 23.7 (C-5 ). In total, the NMR data suggested the presence of a dihydroisocoumarin skeleton contained in 1. The 1 H− 1 H COSY correlations of H-5 /H-4 /H-3 /H-2 /H-1 further indicated a continuous spin system in the molecule, identified as -CH 2 -CH 2 -CH 2 -CH(OH)-CH 3 (Figure 2). Furthermore, according to the key HMBC correlations presented in Figure 2, including from H-2 to C-3, from H-4 to C-1 , C-3, C-5, C-4a, and C-8a, as well as from 8-OH to C-7, C-8, and C-8a, the planar structure of 1 was established as shown. Finally, on the basis of X-ray single-crystal diffraction analysis (Figure 3), the absolute configuration of 1 was established as 3R,4 S, and this new molecule was given the trivial name aspergimarin A. Compound 2 was obtained as a brown oil, and the molecular formula of C 18 H 22 O 8 was assigned to this molecule by the anion HRESIMS peak at m/z 365.1232 [M − H] − (calcd for C 18 H 21 O 8 , 365.1242). The 1 H and 13 C NMR spectroscopic data of 2 ( Table 1) resembled those of 1, and it was determined that the core structure of an oxygenated hydrocarbon-extended dihydroisocoumarin was shared between these molecules. The 1 H− 1 H COSY correlation between H-2 and H-3 , along with key HMBC correlations from H-3 (δ H 2.46) to C-1 (δ C 171.8) and C-4 (δ C 173.5) (Figure 2), indicated the existence of the linear chain -OCO-CH 2 -CH 2 -CO 2 H in the structure of 2. Furthermore, according to the same biosynthetic pathway with 1 based on and the key HMBC correlation observed from H-4 (δ H 4.83) to C-1 , the planar structure of 2 was established. The absolute configuration for 2 was suggested as being 3R,4 S based on the biosynthetic logic that it would match that of 1. The CD spectra of 1 and 2 (Figure 4) are able to be overlapped, with the same Cotton effects observed, which further support the configurational assignment. Accordingly, the resolved structure of compound 2 was afforded the trivial name aspergimarin B.
Compound 3 was also obtained as a brown oil, and its molecular formula of C 20 H 28 O 8 was determined by the potassium cation adduct peak in the HRESIMS spectrum at m/z 435.1416 [M + K] + (calcd for C 20 H 28 KO 8 , 435.1416). The CD spectrum of 3 (Figure 4) and the 1 H and 13 C NMR spectroscopic data ( Table 1) were similar to those of both 1 and 2, indicating   that this molecule is another oxygenated hydrocarbon-extended dihydroisocoumarin analog with the same absolute configuration at C-3 and C-4 . It was determined from a 1 H− 1 H COSY correlation of H-4 /H-5 and the key HMBC correlations from H-2 (δ H 2.38) to C-1 (δ C 170.4) and C-4 (δ C 43.7), as well as from 3 -OH (δ H 4.53) to C-2 (δ C 46.9), C-3 (δ C 69.8) and C-6 (δ C 27.4) (Figure 2), that there is a different secondary carbon side chain [-OCO-CH 2 -C(CH 3 )(OH)-CH 2 -CH 2 OH] in the molecule of 3 as compared to 2. Furthermore, the key HMBC correlation observed from H-4 (δ H 4.83) to C-1 allowed for the planar structure of 3 to be completed. Since the biosynthetic logic of 3 with relation to 1 and 2, together with the matching CD data of these molecules allowed the partial absolute configuration to be assigned as 3R,4 S, only one stereocenter remained uncertain. Using the modified Mosher's method (Figure 5) (Gu et al., 2018) the absolute configuration of C-3 of 3 was established as being R. Therefore, the absolute configuration of 3 was determined to be 3R,4 S,3 R.
Compound 4 was obtained as a yellow oil, and the molecular formula of this molecule was determined to be C 16 H 22 O 5 was based on a peak observed in the HRESIMS spectrum at m/z 295.1537 [M + H] + (calcd for C 16 H 23 O 5 , 295.1540). The data of 4 from spectroscopic analysis, including UV, IR, and NMR, were extremely similar to those reported for the known compound, penicimarin C (9) (Qi et al., 2013). The exceptions noted were determined to be due to the substitution of one methoxy group for a proton at C-7. The placement of the additional methoxy group at C-7 was determined from key HMBC correlations observed from H-5 (δ H 7.06) to C-4 (δ C 32.7), C-7 (δ C 152.2), and C-8a (δ C 119.0), and from 7-O-CH 3 (δ H 3.81) to C-7, together with the 1 H− 1 H COSY correlation of H-5/H-6 (δ H 7.28) (Figure 2). Accordingly, the planar structure of 4 was unambiguously established as shown. Compared to the CD data of 1 and some previously described values for dihydroisocoumarins (Choukchou-Braham et al., 1994), the observed CD spectrum of 4 (Figure 4) indicated the R configuration at C-3. The absolute configuration at the side chain was suggested as 4 S to match that of 1 according to biosynthetic logic, and this was confirmed by use of the modified Mosher's method ( Figure 5). Thus, the absolute configuration of 4 was determined to be 3R,4 S.
Compound 5 was isolated as a colorless oil, and its molecular formula was established as being C 14 H 18 O 5 according to the associated sodiated molecular ion peak in the HRESIMS spectrum at m/z 289.1034 [M + Na] + (calcd for C 14 H 18 NaO 5 , 289.1046). This formula for 5 corresponds to an additional OH with respect to 1, and the 1 H and 13 C NMR data ( Table 2) suggested shared carbon skeletons with the substitution of an additional hydroxy group. From the chemical shift differences calculated between 5 and 1, the additional hydroxy group of 5 was suggested to be at C-4. This assignment was further supported by key HMBC correlations from H-4 (δ H 4.71) to C-5 (δ C 116.3) and from H-5 (δ H 7.03) to C-4 (δ C 67.5), together with the 1 H− 1 H COSY correlation of H-4/H-3 (δ H 4.47) (Figure 2). The absolute configuration at C-3 was determined to be R by the comparison of CD data with 1 and 4 (Figure 4). The absolute configuration of chiral centers at both C-4 and C-4 in 5 was established as being S by use of the modified Mosher's method ( Figure 5). Therefore, the absolute configuration of 5 was determined to be 3R,4S,4 S. Frontiers in Microbiology | www.frontiersin.org Compound 6 was isolated as a yellow amorphous powder, and its molecular formula was determined to be C 14 H 16 O 5 by an associated sodiated molecular ion peak in the HRESIMS spectrum at m/z 287.0879 [M + Na] + (calcd for C 14 H 16 NaO 5 + , 287.0890). The spectroscopic data of 6, including UV, IR, and NMR, resembled those previously reported for the known molecule penicilloxalone B (10) (Ren et al., 2019). From the 1 H and 13 C NMR data, it was obvious that the aromatic substitution patterns of 6 and 10 differed, with 6 bearing protons at C-5 and C-6 while 10 has protons at C-6 and C-7. Furthermore, the key HMBC correlations observed from H-5 (δ H 6.63) to C-4 (δ C 31.4), C-7 (δ C 144.4), and C-8a (δ C 108.4) together with the 1 H− 1 H COSY correlation of H-5/H-6 (δ H 7.00) (Figure 2) corroborated that the two phenolic hydroxy groups of 6 were situated at C-7 and C-8 (δ C 150.0). The absolute configuration of 6 was determined to be 3R, the same as for 10, by comparison of the observed CD spectrum for this molecule with reported data for 10 (Figure 4).

Effects of Compounds 1−11 on Plant
Growth of Arabidopsis thaliana Columbia-0 The isolated compounds 1−11 were subjected to bioassays for testing plant growth response using A. thaliana Col-0, a model plant growth organism. At 100 µM, for both 1 and 5, root growth inhibitory activity was observed against Col-0. Interestingly, while 1 showed only significant inhibitory effect on the lateral root growth of Col-0, 5 caused notable inhibition of both lateral root and primary root growth, as shown in Figure 6. Compounds 2-4 and 6-11 did not show any obvious activity in the same plant growth response assay at 100 µM.

Results of Compounds 1−11 Against Four Additional Bioassays
All compounds isolated in this study (1−11) were also tested for their in vitro inhibitory activity of AChE in a biochemical assay, and phenotypic tests for cytotoxicity against four human-derived cancer cell lines, namely, CCRF-CEM (acute lymphoblastic leukemia T lymphocyte), MDA-MB-231 (breast cancer), HCT-116 (colon cancer), and AGS (gastric adenocarcinoma), and antibacterial activity toward methicillin-resistant S. aureus ATCC43300 and E. coli ATCC25922, and antimalarial activity against P. falciparum W2. None of these compounds exhibited inhibition of AChE (IC 50 > 100 µM), cytotoxic activities against any of the cell lines tested (IC 50 > 50 µM), antibacterial activities toward the two bacteria (MIC > 50 µg/mL), or antimalarial activity against the parasite (EC 50 > 100 µg/mL).

CONCLUSION
In summary, six new dihydroisocoumarin derivatives, aspergimarins A−F (1−6) were obtained along with five known analogs (7−11) from the fermentation of a fungus Aspergillus sp. NBUF87, isolated from the sponge Hymeniacidon sp. collected from the Paracel Islands in the South China Sea. Structurally, compounds 2 and 3 are esters of the C-4 hydroxy group in the C-3 side chain of 1, representing relatively rare isocoumarin derivatives according to previous literature reports (Saeed, 2016). In a plant growth response assay using A. thaliana Col-0 as a model organism, only 1 and 5 showed inhibitory activity against root growth, while the others were inactive at up to 100 µM. This finding indicates that the substitution pattern of hydroxy groups in these dihydroisocoumarins may play an important role in root growth inhibitory activity, and further studies remain necessary to interrogate this phenomenon. Since none of the isolated compounds, including 1 and 5, were found to be broadly AChE inhibitors, anticancer agents, antibacterial agents, or antimalarial agents, it is proposed that these root growth inhibitors act through more elaborate signaling pathways. Moreover, 1 and 5 were here as the root growth inhibitors of Col-0, suggesting that they have an important impact in agricultural production.