Chemical Dereplication of Sarragassum cinnerum
Metabolomic profiling of S. cinnerum crude extract, dereplicated eleven compounds, using LC–HRESIMS. The identified metabolites 1-11 belonged to different chemical classes, including tetrahydrofuran, hydroquinone, plastoquinone, sterols, meroditerpenoids, and sulfoglycolipid (Supplementary Table S1, Figures S1-2, Figure1)
Analyzing of S. cinnerum crude extract, several hits were proposed (Supplementary Table S1, Figure S1-2, Figure1). The molecular ion mass peaks at m/z 215.1283, and 277.2162 [M-H]+, for the predicted molecular formulas C11H20O4 and C18H30O2 gave hits of the tetrahydrofuran antibacterial, and bisnorditerpenes type cytotoxicity against P388 cells (5R,7S,8S)-Communiol A 1, and Hedaol A 2, respectively, that were previously isolated from Sargassum spp.10,11. The mass ion peaks at m/z 307.2624, and 343.2276 corresponding to the suggested molecular formulas C20H34O2, and C22H30O3 [M+H]+ fit a fatty acid, and hydroquinone anti-inflammatory derivative compound Arachidonic acid 3, and Sargachromanol A 4, that was previously isolated from Sargassum pallidum, and Sargassum siliquastrum, respectively12,13. The ion mass peaks at m/z 395.2950, 425.3420, 427.3576, and 487.3060 [M+H]+ for the predicted molecular formulas C27H38O2, C29H44O2, C29H46O2, and C29H42O6 gave hits of the antiviral plastoquinones 2-Geranylgeranyl-6-methylbenzoquinone 5, which was isolated from Sargassum micracanthum14, the anticancer steroidal nucleus of 24-Ethylcholesta-4,24(28)-dien-3,6-dione 6, Saringosterone 7, which were isolated from Sargassum carpophyllum, and Sargassum asperfolium, respectively15,16, and the antioxidant meroditerpenoids of Nahocol A 8, which were isolated from Sargassum siliquastrum17. Two major ion peaks with the m/z values of 445.3682 and 459.2749 [M+H]+ with molecular formulas C29H48O3 and C27H38O6 were detected and dereplicated as 24xi-Hydroperoxy-24-vinylcholesterol 9, and Sargathunbergol A 10, respectively, which were isolated earlier from Sargassum carpophyllum, and Sargassum thunbergii, respectively15,18.
In addition, the mass ion peaks at m/z 553.2681 [M-H]+, for the predicted molecular formula C25H46O11S was dereplicated sulfoglycolipid derivative 1-O-(11-Hexadecenoyl)-3-O-(6'-sulfo-α-D-quinovopyranosyl) glycerol 11, which was previously detected in Sargassum hemiphyllum (Figure1)19.
Phytochemical investigation of Sarragassum cinnerum
Based on the physicochemical and chromatographic properties, the spectral analyses from UV, 1H, and DEPT-Q NMR, as well as comparisons with the literature and some authentic samples, the crude methanolic extract of S. cinnerum afforded the new aryl cresol 12–13, along with the known O-Cresol 1420, m-Cresol 1521. Additionally, arachidonic acid 1622, eicosenoic acid 1722, 1-O-arachidonyl-glycerol 1823, 1-O- arachidonyl-3-O-(α-D-glucopyranosyl) glycerol 1923, 7-β-methyl androsterol 2024, and 1-deoxy-β-D-psicosofuranose 2125, were identified (Figure 2). All characterized compounds 14, and 15 were isolated herein for the first time from the genus Sarragassum (Figures S3-28, Figure 2).
Analysis of the HRESIMS, 1D and 2D NMR data of compounds 12–13 suggested a possible plastoquinones core scaffold11. The HRESIMS data for compound 12 showed an adduct pseudo-molecular ion peak at m/z 314.2607 [M + H]+ (calc. for C22H34O, 314.2604), suggesting 7 degrees of unsaturation. The 1H and DEPT-Q 13C NMR data (Table 1, Figures S3–S4), along with the Heteronuclear Single Quantum Correlation Experiment (HSQC) data (Figure S5), suggested six characteristic resonances appeared: three aromatic methine groups at δH 6.68 (1H, s) δC 91.1, δH 5.18 (1H, m) δC 116.0, δH 6.98 (1H, d, J = 8.0 δC 123.6, δH 7.13 (1H,d, J = 8.0) δC 123.1, three quaternary carbons at δC 153.8, 140.6, and 134.5, and one methyl group at δH 1.34 (1H, s) δC 29.8, suggesting the characteristic core structure for a tri-substituted benzene unit11.
Table 1. DEPT-Q (400 MHz) and 1H (100 MHz) NMR data of compounds 12, 13 in DMSO-d6; Carbon multiplicities were determined by the DEPT-Q experiments
Position
|
12
|
13
|
|
δC
|
δH (J in Hz)
|
δC
|
δH (J in Hz)
|
1
|
153.8, qC
|
|
153.8, qC
|
|
2
|
123.1 , CH
|
7.13, d (8.0)
|
123.1 , CH
|
7.13, d (8.0)
|
3
|
123.6 , CH
|
6.98, d (8.0)
|
123.6 , CH
|
6.98, d (8.0)
|
4
|
134.5 , qC
|
|
140.6 , qC
|
|
5
|
140.6, qC
|
|
116.0 , CH
|
6.68, s
|
6
|
116.0 , CH
|
6.68, s
|
134.5 , qC
|
|
7
|
29.8, CH3
|
1.34, s
|
31.9, CH3
|
1.23, s
|
1'
|
33.4 , CH2
|
2.26, m
|
33.7 , CH2
|
2.26, m
|
2'
|
20.3, CH2
|
2.03, overlapped
|
20.5, CH2
|
2.03, overlapped
|
3'
|
27.1, CH2
|
2.01, overlapped
|
27.1, CH2
|
2.01, overlapped
|
4'
|
127.9 , CH
|
5.31-5.35, m
|
127.9 , CH
|
5.31-5.35, m
|
5'
|
128.8 , CH
|
5.31-5.35, m
|
128.8 , CH
|
5.31-5.35, m
|
6'
|
25.5, CH2
|
2.78, overlapped
|
25.5, CH2
|
2.78, overlapped
|
7'
|
128.0 , CH
|
5.31-5.35, m
|
128.1 , CH
|
5.31-5.35, m
|
8‵
|
128.3 , CH
|
5.31-5.35, m
|
128.4 , CH
|
5.31-5.35, m
|
9‵
|
25.6, CH2
|
2.78, overlapped
|
25.6, CH2
|
2.78, overlapped
|
10‵
|
128.2 , CH
|
5.31-5.35, m
|
128.3 , CH
|
5.31-5.35, m
|
11‵
|
129.4, CH
|
5.31-5.35, m
|
129.4, CH
|
5.31-5.35, m
|
12‵
|
24.9 , CH2
|
1.52, overlapped
|
24.9 , CH2
|
1.52, overlapped
|
13‵
|
28.9 , CH2
|
1.24, overlapped
|
28.9 , CH2
|
1.24, overlapped
|
14‵
|
22.4 , CH2
|
1.25, overlapped
|
22.4 , CH2
|
1.25, overlapped
|
15‵
|
14.5 ,CH3
|
0.89, t (6.6)
|
14.3 ,CH3
|
0.85, t (6.6)
|
qC, quaternary, CH, methine, CH2, methylene, CH3, methyl carbons
NMR data also showed eight aliphatic methylene groups at δH 1.20-2.8 δC 20.5-33.7 (Table 1), three olefinic methine groups at δH 5.31-5.35 (6H, m) δC 127.9-129.4. These signals are suggestive characteristics for 4,7,11-pentadecenyl moiety, where the Heteronuclear Multiple-Bond Correlation (HMBC) experiment of 12 (Figure 3) confirmed the position of the three olefinic methine groups at 4,7,11 of the alkene side chain. Moreover, HMBC experiment showed the 3J-HMBC correlation of the proton H-1‵ δH 2.26 (δC 33.4) with the quaternary carbonyl carbon C-4′ (δC 134.5). Accordingly, compound 12 was identified as 4-(1-(4,7,11-pentadecenyl)-o-cresol.
The molecular formula of compound 13 was identical to that of 12 based on HRESIMS (C22H34O). The 1H and 13C NMR data was also very close to those of compound 12 for the 4,7,11-pentadecenyl moiety, but differed in the resonated chemical shifts of the aromatic attached methyl group of the core tri-substituted benzene unit (Table 1). Comparing the DEPT-Q 13C NMR data of compound 13 with those of 12 showed a downfield shifting of carbons C-7 (ΔδC +2.1), compared with those of compound 12 (Table 1). This suggested a positional difference of the location of the aromatic attached methyl group in the tri-substituted benzene unit versus 12 (Table 1, Figure 2, Figures S8–S12). The assignment of the location of the aromatic attached methyl group in 13 was aided by the HMBC experiment. A 3J-HMBC correlation (Figure 4) of compound 13 proton H-7 δH 1.23 (δC 31.9) with the quaternary carbonyl carbon C-4 (δC 140.6) and a 4J-HMBC correlation of the proton H-7 δH 1.23 (δC 31.9) with the methylene carbon C-1‵ (δC 33.7) confirmed the meta-location of aromatic attached methyl group at the cresol moiety. Accordingly, compound 13 was identified as 4-(1-(4,7,11-pentadecenyl)-m-cresol.
Antiproliferative activity of the isolated metabolites
The isolated compounds 12-21 were in vitro screened for their ability as antiproliferative against hepatic, breast, and colorectal carcinoma cell lines (HEPG2, MCF7, and CACO2, respectively). Results showed that compounds 12, 13, and 19 were able to inhibit the growth of all tested cell lines moderately with IC50 values ranged from 11.2 ± 0.6 to 21.6 ± 1.3µM (Table 2).
Table 2. In vitro antiproliferative activity of the isolated compounds 12-21 expressed as IC50 ± (SSEM) µM
IC50 (µM)
|
Code
|
HEPG2
|
MCF7
|
CACO2
|
12
|
14.5 ± 0.8
|
17.6 ± 0.9
|
18.2 ± 0.7
|
13
|
13.1 ± 1.1
|
12.7 ± 1.3
|
11.2 ± 0.6
|
14
|
> 50
|
> 50
|
> 50
|
15
|
> 50
|
> 50
|
> 50
|
16
|
> 50
|
> 50
|
> 50
|
17
|
> 50
|
> 50
|
> 50
|
18
|
> 50
|
> 50
|
> 50
|
19
|
18.5 ± 1.4
|
21.6 ± 1.3
|
15.7 ± 0.9
|
20
|
> 50
|
> 50
|
> 50
|
21
|
> 50
|
> 50
|
> 50
|
Doxorubicin
|
4.2±0.3
|
3.8±0.2
|
3.4±0.1
|
IC50 value of compounds against each cancer cell line, which was defined as the concentration (µM) that caused a 50% inhibition of cell growth in vitro, data were expressed as mean ± SEM (n = 3). One-way analysis of variance (ANOVA) followed by Dunnett’s test using PASW Statistics® version 18 (Quarry Bay, Hong Kong) was applied. GraphPad Prism software version 6 (La Jolla, CA, USA) was used for statistical calculations. * statistically significant at p < 0.05. Doxorubicin a positive control
Virtual screening-based target identification
Characterization of the biological target for a certain molecule is a true challenge, however, the continuous development of in silico tools including molecular modeling and virtual screening has significantly improve the success rate of finding suitable molecular targets. Many online target identification platforms are currently available, and their search protocols are either structural based or ligand-based. PharmMapper is one of these online platforms that can screen and suggest the most probable protein targets of a query molecule based on its pharmacophore model26. The basic principle of pharmacophore-based screening is that the binding of certain molecules with their protein targets is mainly determined by key pharmacophore maps (i.e. spatial arrangement of structural features). Thus, molecules that their shapes are able to fit with these pharmacophore maps have the highest probability to bind the same protein target. Consequently, PharmMapper was used to propose a proper protein target for compounds 12, 13 and 19. 5-LOX and 15-LOX were found to be the top-scoring hits for these metabolites. In addition, they have a direct link to tumorigenesis8,27-30. Consequently, they were selected for further in vitro and in silico validations.
LOX inhibition assay
To validate the preliminary virtual screening prediction, compounds 12, 13 and 19 were assayed for their 5-LOX and 15-LOX inhibitory activities. Interestingly the three compounds achieved potent enzyme inhibition toward 5-LOX (IC50 1.3 ± 0.1 to 2.1 ± 0.4 µM, Table 3). However, their activity against 15-LOX was weaker, particularly compounds 12 and 13 (IC50 25.3 ± 0.4 and 23.6 ± 0.3 µM, respectively) that were more selective for 5-LOX (Table 3).
Table 3. Docking scores, binding free energies, Ki and IC50 values of compounds 12, 13, and 19 together with the co-crystalized inhibitors NDGA and AA
Compound
|
ΔGVina*
|
ΔGFEP**
|
ΔGKDEEP***
|
Ki #
|
IC50#
|
5-LOX
|
15-LOX
|
5-LOX
|
15-LOX
|
5-LOX
|
15-LOX
|
5-LOX
|
15-LOX
|
5-LOX
|
15-LOX
|
12
|
-9.3
|
-5.1
|
-8.1
|
-4.4
|
-7.7
|
-4.6
|
0.9 ± 0.1
|
17.4 ± 0.2
|
1.6 ± 0.3
|
25.3 ± 0.4
|
13
|
-8.9
|
-5.5
|
-8.0
|
-4.7
|
-7.5
|
-4.5
|
0.7 ± 0.2
|
14.3 ± 0.4
|
1.3 ± 0.1
|
23.6 ± 0.3
|
19
|
-9.1
|
-7.7
|
-7.9
|
-7.1
|
-7.6
|
-7.2
|
1.4 ± 0.2
|
4.2 ± 0.1
|
2.1 ± 0.4
|
6.7 ± 0.3
|
NDGA##
|
-7.2
|
-6.9
|
-7.0
|
-6.5
|
-6.8
|
-6.5
|
6.9 ± 0.1
|
6.1 ± 0.2
|
8.8 ± 0.3
|
-9.5 ± 0.5
|
AA##
|
-7.6
|
-7.0
|
-6.2
|
-6.4
|
-7.1
|
-6.3
|
-
|
-
|
-
|
-
|
Lipoxygenase (LOX), nordihydroguaiaretic acid (NDGA), arachidonic acid (AA)
*Vina docking scores calculated in kcal/mol
**MDS-derived binding free energies calculated in kcal/mol by FEP method
***Neural networking-derived binding free energies calculated in kcal/mol by KDEEP software
#In vitro inhibition constant (Ki) and inhibition concentration 50 (IC50) expressed as µM
# #The reported co-crystalized ligands
Moreover, they showed inhibitory constants (Ki) ranged from 0.7 ± 0.2 to 17.4 ± 0.2 µM (Table 3), and these values were most agree with the competitive inhibition of both enzymes31.
The results of enzyme inhibition assay were also correlated with those of the antiproliferative ones for HepG2 and MCF-7 and CACO2. Surprisingly, overexpression of 5-LOX has been reported in breast, liver and colorectal cancers27-29. Also 15-LOX has been reported to be over expressed in a number of tumors like prostate and breast cancers. Hence, these enzymes can be considered promising targets for cancer therapy.
Molecular Docking and dynamic simulation
5-LOX has a hydrophobic active site9, that harbor a catalytic iron (Fe+2), and such hydrophobicity is essential to allow efficient binding with the hydrophobic substrate, arachidonic acid (AA) (Figure 2)9. Compounds 12, 13, and 19 have extended unsaturated hydrophobic side chains that resemble AA (Figure 5).
Molecular docking experiments revealed that these compounds could bind with the 5-LOX’s active site efficiently with binding scores ranged from -8.9 to -9.3 kcal/mol (Figure 5), and their bindings were even better than the co-crystalized ligands (Table 3). Additionally, the phenolic moiety of both compounds was involved in H-bonding with HIS-600 similarly to the co-crystallized redox-type inhibitor, nordihydroguaiaretic acid (NDGA) (Figure 5).
Compounds 12 and 13’s hydrophobic side chains were able to adapt themselves inside the hydrophobic U-shaped active site, where they took convergent orientations, but slightly different from that of AA (Figure 5). LEU-368, ILE-406, LEU-414, and LEU-607 were the main amino acid residues involved in the hydrophobic interactions with their side chains, while PHE-359, TRP-599, and PRO-569 interacted with their aromatic moieties. The binding mode of compound 19 was quite different, where its polar carbohydrate head interacted with LYS-409, GLN-413, and ILE-673 through four strong hydrogen bonds (< 2.5̊A), while it’s hydrophobic tail interacted with LEU-368, LEU-414, TRP-599, and LEU-607 (Figure 5).
Subsequent Molecular Dynamic Simulation (MDS) experiments (50ns) revealed that the three compounds 12, 13, and 19 were able to stabilize the enzyme’s active site.
Compounds 12 and 13’s positions remained to change over the first 32.4 ns (RMSD ~ 3.4̊A). Afterward, they became stable till the end of the simulation (average RMSD values of 2.67 and 2.59̊A, respectively), where their extended hydrocarbon chains became more relaxed and straight (Figure 6). The H-bonds between their phenolic group and HIS-600 remained unchanged throughout the MDS. Starting from 22.6ns, GLN-363’s side chain became also involved in H-bonding with the phenolic group of both compounds (Figure 6). Additionally, compound 12’s tail remained imbedded inside a hydrophobic pocket consists of the side chains of TRP-147, PHE-151, LEU-368, LEU-373, and LEU-414, while compound 13’s tail, settled inside another hydrophobic pocket consists of TRP-147, LEU-414, ILE-415, and VAL-433 (Figure 6).
Similarly, the hydrophobic part of compound 19 was compacted at the beginning of MDS, and gradually become more extended till stabilization at 25.4ns (RMSD = 2.75̊A), where PHE-359, PRO-569, and TRP-599 became involved in hydrophobic interactions with the molecule’s tail. Furthermore, the side chain of LYS-409 became involved in an additional H-bonding with the molecule’s hydrophilic carbohydrate part (Figure 6). Further binding free energy calculations (ΔGFEP and ΔGKDEEP) revealed that compounds 12, 13, and 19 got higher binding free energy values than that of the co-crystalized ligands (Figure 7), and were in good accordance with the in vitro enzyme inhibition (Figure 7).
On the other hand, docking scores of the three compounds against 15-LOX were significantly lower, particularly for compounds 12 and 13 (Figure 7) that showed only hydrophobic interactions (with PHE-184, TYR-185, PHE-365, LEU-374, LEU-415, LEU-419, LEU-420, VAL-603, LEU-609, and LEU-610) inside the enzyme’s active site. Upon MDS (Figure 7), both compounds showed unstable positioning inside 15-LOX active site, where the surrounding hydrophobic amino acid residues were able to keep them in position till ~ 23.4ns (RMSD ~ 3.5 Å). Afterward, their position inside the active site began to change dramatically and their RMSDs reached reach about 7.1 Å and remained to fluctuate till the end of MDS with gross averages of 8.1 and 7.5̊A, respectively over the 50ns of simulation. This obvious instability of compounds 11 and 12 reflected on their binding free energies and in vitro inhibitory activities that were significantly lower in comparison with 5-LOX (Figure 7).
Such instability was not the case with compound 19, where the hydrophilic carbohydrate moiety was able to form a network of H-bonds (7 H-bonds) with TYR-185, GLN-425, ARG-429, and ASP-602. Furthermore, these hydrophilic interactions remained unchanged over the course of MDS, and thus compound 20’s RMSD was at equilibrium (~2.6̊A) till the end of MDS. Such structural and dynamic information can explain the convergent inhibitory activity of compound 19 against both 5-LOX and 15-LOX (Figure7), and the moderate selective inhibition of 5-LOX by compounds 12 and 13 over 15-LOX.