Neuronal Modulators from the Coral-Associated Fungi Aspergillus candidus

Three new p-terphenyl derivatives, named 4″-O-methyl-prenylterphenyllin B (1) and phenylcandilide A and B (17 and 18), and three new indole-diterpene alkaloids, asperindoles E–G (22-24), were isolated together with eighteen known analogues from the fungi Aspergillus candidus associated with the South China Sea gorgonian Junceela fragillis. The structures and absolute configurations of the new compounds were elucidated on the basis of spectroscopic analysis, and DFT/NMR and TDDFT/ECD calculations. In a primary cultured cortical neuronal network, the compounds 6, 9, 14, 17, 18 and 24 modulated spontaneous Ca2+ oscillations and 4-aminopyridine hyperexcited neuronal activity. A preliminary structure–activity relationship was discussed.

As part of our continuing search for bioactive molecules from marine invertebrates and the associated fungi [15][16][17], a strain of A. candidus was isolated from the internal tissues of the gorgonian coral Junceela fragillis, collected from the Xisha area of the South China Sea. Chemical investigation of the fermentation extract of this fungus resulted in the isolation of three new p-terphenyl derivatives and three new indole-diterpene alkaloids, together with eighteen known analogues. Further, p-Terphenyls are regarded as the main metabolites of A. candidus. More than 230 analogues have been reported up to date, with the chemical diversity being attributed to the substituents on rings A and C [8,18,19]. The indole-diterpene alkaloids are a cluster of characteristic metabolites from this genus and were firstly reported from the titled fungi [20]. These metabolites are structurally constructed with an indole molecule and a saculatane diterpenoid. The biosynthetic process
Phenylcandilide A (17), obtained as a yellow amorphous solid, had a molecular formula of C 18 H 18 O 5 as deduced from the HRESIMS, indicating 10 degrees of unsaturation. The NMR data of 17 showed similarity to those of 4-methyl-candidusin A (16) [32], regarding the signals for the phenol ring C and benzofuran ring B. The C-2 to C-5 diene fragment of the phenol ring A in 16, however, was degraded to a methyl and a hydroxymethyl groups in 17. The distinct HMBC correlations from H 3 -2 to C-1 and C-6, and H 2 -5 to C-6 and C-1 confirmed the location of 1-Me and 6-CH 2 OH ( Figure 2). The structure of 17 was determined and nominated as phenylcandilide A. phous solid. Its molecular formula was determined as C26H28O5 by HRESIMS, requir 13 degrees of unsaturation. The IR spectrum displayed absorptions for hydroxy (33 cm −1 ) and substituted benzol (1609, 1462, 834, and 815 cm −1 ) functionalities. The presen of benzol rings was supported by the strong UV absorptions at 276, 248, and 209 nm. T NMR spectra of 1 displayed resonances for twenty sp 2 carbons and six sp 3 carbons, tak into account the ten degrees of unsaturation. The remaining three degrees of unsaturat assigned to the ring system of this molecule were in agreement with that of the terphe framework. Its NMR data were almost identical to those of the co-isolated prenylt phenyllin B (12) [11], except for the appearance of an additional methoxy group (δH 3 3H, s; δC 55.6, CH3). The methoxy group was assigned as 4-OMe via its HMBC correlatio to C-4, and further confirmed by its NOE correlation with H-5 ( Figure S1). The structu of compound 1 was therefore determined as 4-O-methyl-prenylterphenyllin B.
Phenylcandilide A (17), obtained as a yellow amorphous solid, had a molecular f mula of C18H18O5 as deduced from the HRESIMS, indicating 10 degrees of unsaturati The NMR data of 17 showed similarity to those of 4-methyl-candidusin A (16) [32], garding the signals for the phenol ring C and benzofuran ring B. The C-2 to C-5 die fragment of the phenol ring A in 16, however, was degraded to a methyl and a droxymethyl groups in 17. The distinct HMBC correlations from H3-2 to C-1 and C-6, a H2-5 to C-6 and C-1 confirmed the location of 1-Me and 6-CH2OH ( Figure 2). The struct of 17 was determined and nominated as phenylcandilide A. Phenylcandilide B (18) was obtained as a yellow amorphous solid. Its molecular f mula was established as C20H20O6 on the basis of the HRESIMS. Comparison of its NM data ( Table 1) with those of 17 revealed a similarity in the structures. A difference w recognized for the signals of the hydroxymethyl group in 17 being replaced by a met acetate subunit in 18, confirmed by the IR at 1740 cm -1 . The location of the methyl acet subunit was indicated by the diagnostic HMBC correlations from H2-5 to C-4, C-1 and 6, and 4-OMe to C-4 ( Figure 2). The structure of compound 18 was thus determined a nominated as phenylcandilide B.  Phenylcandilide B (18) was obtained as a yellow amorphous solid. Its molecular formula was established as C 20 H 20 O 6 on the basis of the HRESIMS. Comparison of its NMR data ( Table 1) with those of 17 revealed a similarity in the structures. A difference was recognized for the signals of the hydroxymethyl group in 17 being replaced by a methyl acetate subunit in 18, confirmed by the IR at 1740 cm -1 . The location of the methyl acetate subunit was indicated by the diagnostic HMBC correlations from H 2 -5 to C-4, C-1 and C-6, and 4-OMe to C-4 ( Figure 2). The structure of compound 18 was thus determined and nominated as phenylcandilide B. Asperindole E (22) was obtained as an optically active, white powder. Its molecular formula was established as C 27 H 31 NO 5 by HRESIMS, implying 13 degrees of unsaturation. The IR spectrum of 22 displayed absorptions for hydroxy (3360 cm -1 ) and substituted benzol (1632, 1468, 800, and 742 cm −1 ) functionalities. The presence of an α,β-unsaturated ketone moiety was suggested by the characteristic IR absorption at 1658 cm −1 . The strong UV absorptions at 279, 268, 230 and 210 nm were in agreement with the presence of a benzol ring or an α,β-unsaturated ketone moiety in the structure. The NMR spectra (Table 2) demonstrated a great similarity to the known metabolites of asperindoles A-D, previously obtained from an ascidian-derived fungus Aspergillus sp. [20], suggesting the same indole-diterpene framework for these molecules. Signals for the acetyl group in asperindole B were not observed for 22. The structure of 22 was suggested to be the deacetyl analogues of asperindole B, which was fully confirmed by 2D NMR experiments, particularly HMBC and NOESY ( Figure 3). The absolute configurations of 22 were the same as that of asperindoles A-D on the basis of the similar ECD spectra ( Figure 4). The structure of compound 22 was thus determined as asperindole E. Asperindole F (23), obtained as an optically active, white powder, has a molecular formula of C 31 H 36 ClNO 7 as determined by HRESIMS. The presence of a chlorine atom in the molecule was indicated by the isotopic peaks at m/z 568/570 [M − H] − with a ratio of 3:1. The NMR spectra of 23 ( Table 2) resembled those of asperindole C (20) [20] (Table S1), except for the acetyl group which is absent, showing the same difference pattern as that between asperindole E (22) and B. Compound 23 is the deacetylated derivative of asperindole C, and was named as asperindole F. This assignment was further confirmed by NMR and ECD experiments (Figures 3 and 4).   Asperindole G (24) was obtained as an optically active, white powder. Its HRESIMS gave the same molecular formula as that of asperindole F (23). As expected, the NMR data of 24 (Table 2) showed similarity to those of 23. However, two sp 3 carbon atoms (δ 94.0, C; δ 30.7, CH2) in 23 were replaced by two sp 2 carbon atoms (δ 145.0, C; δ 111.5, CH) in 24, suggesting a cleavage of the ether bridge of ring G in 23 to form a C-6 to C-7 double bond,   Asperindole G (24) was obtained as an optically active, white powder. Its HRESIMS gave the same molecular formula as that of asperindole F (23). As expected, the NMR data of 24 (Table 2) showed similarity to those of 23. However, two sp 3 carbon atoms (δ 94.0, C; δ 30.7, CH2) in 23 were replaced by two sp 2 carbon atoms (δ 145.0, C; δ 111.5, CH) in 24, suggesting a cleavage of the ether bridge of ring G in 23 to form a C-6 to C-7 double bond, Asperindole G (24) was obtained as an optically active, white powder. Its HRESIMS gave the same molecular formula as that of asperindole F (23). As expected, the NMR data of 24 (Table 2) showed similarity to those of 23. However, two sp 3 carbon atoms (δ 94.0, C; δ 30.7, CH 2 ) in 23 were replaced by two sp 2 carbon atoms (δ 145.0, C; δ 111.5, CH) in 24, suggesting a cleavage of the ether bridge of ring G in 23 to form a C-6 to C-7 double bond, and a 28-hydroxymethyl group in 24. The assignment for the C-6 to C-7 double bond was confirmed by the HMBC correlations of H-6 with C-4 and C-12, and H-9 with C-7. In addition, the α-hydroxyisobutyrate moiety was attached to the C-28 methylene group of the side-chain instead of the C-27 of the bridged 1,3-dioxane ring (as those in 19-23), as confirmed by the diagnostic HMBC correlations between H 2 -28 and the ester carbonyl carbon (δ 175.5, C). Compound 24 was expected to have the same stereochemistry as that of 19-23 due to their obvious correlations in biogenetic origin, even though the ECD spectrum of 24 was significantly different from those of 20-23. However, a very weak NOE correlation was observed between H-9 and H 3 -26 in the NOESY spectrum of 24. This suggested that an epimerization might occur at C-9, which is adjacent to the C-10 ketone group. In order to determine the absolute configuration of C-9, TDDFT-ECD [33] and DFT-NMR calculations [34] were performed on the (3S,4R,9R,13S,16S,27S) and (3S,4R,9S,13S,16S,27S) epimers. Ring F, containing an α,β-unsaturated carbonyl chromophore, was expected to have a major impact on the high-wavelength ECD transitions, and thus a large difference was expected between the ECD spectra of the two epimers.  Figure 6). This small difference suggested that 24 had (9R) absolute configuration, and the difference in the experimental ECD spectra of 20-23 and 24 derives from the different chromophore systems and different planar structures. It is well-documented that even small structural changes can result in markedly different or mirror-image ECD spectra for homochiral derivatives by changing the preferred conformation or electronic properties of the molecule [36]. The (9R) and (9S) epimers were further distinguished by 13 C NMR DFT calculations, which has been proven an efficient method to distinguish diastereomeric natural products [17,34]. For the NMR calculations, the above MMFF conformers were re-optimized at the B3LYP/6-31 + G(d,p) level yielding 16 and 14 low-energy structures above the 1% Boltzmann distribution, respectively. Despite the DMSO solvent and the presence of the halogen, both causing larger deviations in the computed data, the calculated 13 C NMR shifts of the (9R) epimer had a slightly smaller MAE average value than those of the (9S) epimer, and the DP4+ statistical analysis [35,37] resulted in an 83.75% confidence for the (9R) epimer (Table S2). Since both the computed ECD and NMR data suggested (9R) configuration, the H-9 and H 3 -26 NOE cross-peak must be an artefact, and the absolute configuration of 24 was determined to be (3S,4R,9R,13S,16S,27S). The interatomic distance of H-9 and H 3 -26 is 3.7 Å in the (9S) epimer and it is above 5.0 Å in the (9R) epimer.
All the isolated compounds were evaluated for their neuronal modulatory activities by testing their effect on spontaneous Ca 2+ oscillations (SCOs), and the seizurogenic agent 4-aminopyridine (4-AP) induced hyperexcitation in primary cultured neocortical neurons (Table 3). SCOs play a crucial role in mediating neuron development, and are closely associated with neuronal excitable and inhibitory neuronal transmission [38][39][40]. The compounds with modulatory activity on SCOs may have potential in drug candidates for treating neurological diseases such as epilepsy, pain and depression [41].  All the isolated compounds were evaluated for their neuronal modulatory activities by testing their effect on spontaneous Ca 2+ oscillations (SCOs), and the seizurogenic agent 4-aminopyridine (4-AP) induced hyperexcitation in primary cultured neocortical neurons (Table 3). SCOs play a crucial role in mediating neuron development, and are closely associated with neuronal excitable and inhibitory neuronal transmission [38][39][40]. The compounds with modulatory activity on SCOs may have potential in drug candidates for treating neurological diseases such as epilepsy, pain and depression [41].  All the isolated compounds were evaluated for their neuronal modulatory activities by testing their effect on spontaneous Ca 2+ oscillations (SCOs), and the seizurogenic agent 4-aminopyridine (4-AP) induced hyperexcitation in primary cultured neocortical neurons (Table 3). SCOs play a crucial role in mediating neuron development, and are closely associated with neuronal excitable and inhibitory neuronal transmission [38][39][40]. The compounds with modulatory activity on SCOs may have potential in drug candidates for treating neurological diseases such as epilepsy, pain and depression [41]. In the present study, we found that four compounds 6, 17, 18 and 24 inhibited SCO activity, and 4-AP induced hyperexcitability by decreasing the SCO amplitude and frequency in the primary cultured cortical neuronal network. However, compounds 9 and 14 produced a more complicated Ca 2+ response. Their concentration dependently increased the SCO frequency with the concurrent suppression of the SCO amplitude at concentrations below 10 µM and 3 µM, respectively, and transiently increased the intracellular Ca 2+ concentration which recovered to basal level within 5 min at concentrations of 30 µM and 10 µM, respectively (Table 3, and Figures S85-S90). For the cluster of p-terphenyl derivatives, all the active compounds have hydroxyls for both R 2 and R 4 and those with hydrogens for both R 1 and R 3 displayed the strongest activity. Substitution for one of the R 2 /R 4 pair of hydroxyls or one of the R 1 /R 3 pair of hydrogens will decrease the activity. Interestingly, the degradation of ring A to a hydroxymethyl group may lead to an increase in activity.
For the cluster of indole-diterpene alkaloids, ring cleavage on the ether bridge of ring G seems critical for the activity since all those compounds that have ring G are not active.

Fungal Material
The fungal strain SG-8-5 was isolated from the internal tissues of the gorgonian coral J. fragillis, which was collected from the Xisha area of the South China Sea, and identified as A. candidus by 18sRNA sequence (GenBank accession number AB008396.1). The fungus was deposited in Tongji University, Shanghai, China.

Neuronal Modulatory Activity Assay In Vitro
The neuronal modulatory activities of 1-24 were evaluated by testing the effect on spontaneous Ca 2+ oscillations (SCOs) and seizurogenic agent 4-aminopyridine (4-AP)induced hyperactive SCOs frequency and amplitude in primary cultured neocortical neurons as described previously [16,38]. Neocortical neurons at 9 days in vitro (DIV) were used to investigate the influence of tested compounds on intracellular Ca 2+ concentration ([Ca 2+ ] i ). Briefly, the neurons were loaded with Fluo-4 for 1 h at 37 • C in Locke's buffer. After recording the baseline spontaneous Ca 2+ oscillations for 5 min, different concentrations of compounds were added to the corresponding well, and the [Ca 2+ ] i was monitored for 15 min using FLIPR tetra® . To test anti-epileptic potential of the meroterpenoids, 4-AP (10 µM) was added and the monitoring for [Ca 2+ ] i was continued for an additional 10 min. The presented data were values of F/F 0 , where F is the fluorescence intensity at any time point whereas F 0 is the basal fluorescence. An event with ∆ F/F 0 over 0.1 unit was considered to be an SCO. The frequency and amplitude of SCOs were quantified using Origin software (V7.0) from a time period of 5 min after first or second addition of compound or vehicle (0.1% DMSO).

Computational Section
Mixed torsional/low-frequency mode conformational searches were carried out by means of the Macromodel 10.8.011 software by using the Merck molecular force field (MMFF) with an implicit solvent model for CHCl 3 [42]. Geometry re-optimizations were carried out at the B3LYP/6-31 + G(d,p) level in vacuo and the ωB97X/TZVP level with the PCM solvent model for MeCN. TDDFT-ECD calculations were run with various functionals (B3LYP, BH&HLYP, CAMB3LYP, and PBE0) and the TZVP basis was set as implemented in the Gaussian 09 package, with the same or no solvent model as in the preceding DFT optimization step [43]. ECD spectra were generated as sums of Gaussians with 3000 and 2700 cm −1 widths at half-height, using dipole-velocity-computed rotational strength values [44]. NMR calculations were performed at the mPW1PW91/6-311 + G(2d,p) level [45]. Computed NMR shift data were corrected with I = 185.2853 and S = −1.0306 [46]. Boltzmann distributions were estimated from the B3LYP and ωB97X energies. The MOLEKEL software package was used for visualization of the results [47].

Conclusions
From the fungi A. candidus, associated with the South China Sea gorgonian Junceela fragillis, twenty-four metabolites having p-terphenyl and indole-diterpene frameworks were obtained with their structures and absolute configurations being elucidated on the basis of spectroscopic analysis and computational calculations. It was found that small structural changes can result in a markedly different ECD spectra, and 13 C NMR DFT calculations are an efficient method to distinguish diastereomeric natural products. Six compounds could modulate SCOs and 4-aminopyridine hyperexcited neuronal activity in the in vitro biotest. A preliminary structureactivity relationship was discussed, which may give a reference for further investigation or chemical optimization of SCO modulators.