Molecular networking prospection and characterization of terpenoids and C15-acetogenins in Brazilian seaweed extracts

Molecular networking (MN) can efficiently dereplicate extracts and pure compounds. Red algae of the genus Laurencia are rich in halogenated secondary metabolites, mainly sesquiterpenes and C15-acetogenins. Brown algae of the genus Dictyopteris produce mainly C11-hydrocarbons, sesquiterpenes and sulfur-containing compounds, while Dictyota and Canistrocarpus are reported to contain mainly diterpenes. This study performs an exploratory MN analysis of 14 extracts from algae collected in Brazil (including the oceanic islands) and characterizes the secondary metabolites from the analyzed species. The extracts and some isolated metabolites were analyzed by LC-MS using the FastDDA algorithm, and the MS/MS spectra were submitted to GNPS and displayed in Cytoscape 3.5.1. The GNPS platform generated 68 individual nodes and nine family networks. The MN exploratory analysis indicated chemical differences among species, and also in sampling sites for the same species. For some extracts, it was possible to identify mass values that could correspond to terpenoids and C15-acetogenins that have already been isolated from those or related species. An interesting chemodiversity was highlighted between Laurencia catarinensis from two nearby islands, and this was revealed and was also suggested by the family networks. Many nodes in the MN could not be characterized, and these metabolites can be used as targets for isolation in future works.


Introduction
Recently, new approaches have been developed to guide the isolation of new metabolites, with reduced costs and effort. MS/ MS-based Global Natural Products Social (GNPS) Molecular Networking (MN) is an example of a platform that can be easily integrated into natural product workows, to efficiently dereplicate extracts and pure compounds. 1,2 Using a special computational algorithm to compare the degree of spectral similarity between the MS/MS spectra of different compounds, this technique provides a visual overview of all the ions that were detected and fragmented in the MS experiment. 3 The key is based on structurally related natural products that share similar tandem mass fragmentation patterns, and molecular families that tend to cluster together and can be visualized as a network. 2,4 This dereplication strategy has been widely used for the identication of marine and microorganism-derived natural products. 4,5 The marine environment represents approximately 70% of the earth's surface, and has a huge biodiversity, characterized by a wide chemical diversity of natural products. 6,7 The Brazilian maritime area, known as the "Blue Amazon", has one of the longest coastlines in the world, and ve oceanic islands. 8 However, it is still under-explored as a source of natural products. This marine area is largely located in tropical zone, contributing to the development of a rich and diverse marine biota. Many marine algae species of the genera Laurencia, Dictyota, Canistrocarpus and Dictyopteris, can be found in this area.
While red algae of the genus Laurencia J. V. Lamouroux are rich in halogenated secondary metabolites, belonging mainly to the sesquiterpenes, diterpenes, triterpenes, and C 15 -acetogenins classes, 9 brown alga of the genus Dictyopteris produce mainly C 11 -hydrocarbons, sesquiterpenes and sulfur-containing compounds. 10 In addition, Dictyota and Canistrocarpus are chemically composed of diterpenes. [11][12][13][14][15][16] The metabolites produced by algae mainly serve as chemical defenses against herbivorous or pathogenic micro-organisms. 16 Many of these metabolites have been conrmed as presenting anticoagulant, antifouling, antimicrobial, antioxidant, herbivore-deterrent and cytotoxic activities. 9,10,[17][18][19][20] Results and discussion Samples were collected in 14 different locations. These samples were identied as belonging to seven species (Table 1). Analysis of the MN data revealed that 10 of the collected samples were distributed in nine family networks (F1-F9) of which: F1 presented 19 nodes, F2 showed 12 nodes, F3-F6 contained three nodes, and F7-F9 comprised two nodes (Fig. 1). Metabolites from extracts of red and brown algae clustered in different family networks, as was expected. Moreover, two extracts of brown (DJ1, DP2) and two of red algae (LD3 and LI4) were not included in any family network. The entire network was formed by 117 nodes, including 68 individual nodes. F1 ( Fig. 2A) 15 The node m/z 267.2134, detected at 7.32 min in the CC3 and DM1 LC-MS spectra, turned out to be a base peak ion, which was the fragment of the precursor m/z 385. 2359  ]. The information outlined above, in conjunction with the literature search led to two previously reported compounds. Since the completed structure is not always possible using tandem mass data alone, dictyotadiol (5) and dictyol B (6) were suggested as potential structures. 14,16 Similar MS/MS behavior was observed for the node at m/z 369.2395 [C 22 H 34 O 3 + Na] + (calcd 369.2406) that corresponded to the peak at 12.00 min in the CC1 and DM2 LC-MS data. This metabolite differs from 5 and 6 by 42 Da suggesting that it is an acetylated derivative. The above data, together with the data reported in the literature led to the structure of dictyol B acetate (7), another guaiane prenylated diterpene. 16 The node at m/z 271.2439 [C 20 H 31 O + H] + (13.09 min in CC1, CC2, DM2 and DM3) represents a base peak ion formed when However, the mass spectrum of this node gave the precursor ion with 2 Da lighter than 8 and 9 suggesting a prenylated guaiane diterpene analog.
In general, MN clustered in F1 metabolites are mainly from the brown algae Canistrocarpus cervicornis and Dictyota mertensii ( Fig. 1), with similar fragmentation patterns to the diterpenes already reported for the Dictyota species. Although the literature for C. cervicornis regards the presence of dolastane and secodolastane, curiously only the guaiane prenylated and xeniane types were observed in the nodes of F1. This suggest that C. cervicornis from Rocas Atoll is chemically diverse from other specimens collected on the Brazilian Coast. Compounds 8 and 9 were previously identied in D. mertensii, 12,13 while 1-4 and 6 were obtained from D. crenulata 14,15 and 5 and 7 were identied in D. dichotoma. 21 The nodes with m/z 270.2309, m/z 272.2452 and m/z 304.2349 were found in LC-MS, but these peaks correspond to isotopic patterns of the nodes with m/z 269.2265, m/z 271.2439 and m/z 303.2332, respectively. The foregoing data in conjunction to the data reported in the literature led to the structure of some chamigrane-type sesquiterpenes, such as 9hydroxy-4,10-dibromo-3-chloro-a-chamigrene (10) 22-24 and 2,10dibromo-3-chloro-8-hydroxy-b-chamigrene (11), 25 along to the irregular rearranged bisabolane-related sesquiterpene laucapyranoid A (12). 26 The metabolite detected at 7.74 and 8.11 min might be related, due to similar fragmentation behavior. Compound 10 was reported for many Laurencia species, such as  (19), which has already been reported in the chemical study of L. decumbens. 30 Node m/z 408.9829, in turn, was a fragment ion of the precursor m/z 442.9620 (RT 7.93 min), which appeared in the mass spectrum alongside m/z 444.9616, 446.9571 and 448.9570 at a ratio of 3 : 7 : 5 : 1, corresponding to the   (36 Da). As reported for F3 and F4, these ndings indicated a chamigrane sesquiterpene type, which was compatible with the structures of prepacifenol epoxide (20) 39 and johnstonol (21). 22 Since 20 and 21 were isolated from LC1 extract and presented the same retention times and fragmentation patterns, it is possible that this node corresponds to both compounds. Nevertheless, according to the literature, prepacifenol epoxide may be converted into johnstonol. 22 Compound 20 was also isolated from L. composita, 31 L. johnstonii, 20 L. nidica, 22 L. okamurai, 32 while compound 21 was reported for L. nidica, 22 L. okamurai 32 and L. pacica. 33 For F8 and F9, it was also not possible to establish any structure.
The individual node with m/z 392.9902 also corresponded to a metabolite produced by L. catarinensis from Xavier Island (LC1 extract). This mass spectrum showed m/z 390.9898, 392.9902 and 394.9876 at ratio of 1 : 2 : 1, corresponding to the isotopic pattern of a metabolite containing 2 bromine atoms. Two isomeric metabolites were observed at 4.96 and 6.39 min with this particular isotopic pattern and the mass value of m/z 390.9898. Its elemental composition [C 15 H 20 Br 2 O 2 + H] + (calcd m/z 392.9888) conrmed a dibrominated metabolite. The literature search led to the hypotheses of some C 15 -acetogenins containing vemembered cyclic ethers ring, such as laureepoxide (22) (one tetrahydrofuran ring), 34 (3E)-elatenyne (23) and elatenyne (24) (two isolated tetrahydrofuran rings), 30,35 and kumausallene (25) (two fused tetrahydrofuran rings), 25 along with laurobtusin (26), an acetogenin-containing sixmembered cyclic ether ring. 36 Compound 22 was reported to L. nipponica, 34 23 to L. majuscula, 35 24 to L. decumbens 30 and L. elata, 37 25 to L. nipponica 25 and 26 to L. obtusa. 36 The individual nodes at m/z 408.9581 and 410.9542 correspond to the metabolite from LC1 extract with a retention time at 9.80 min. These mass values were found in the same spectrum alongside m/z 412.9509 and m/z 414.9525 at a ratio of 3 : 7 : 5 : 1 corresponding to the isotopic pattern of a metabolite containing 2 bromine atoms and one chlorine atom. This ion was produced by the precursor at m/z as observed in the mass spectrum. Loss of H 2 O [C 15 H 21 Br 2 ClO 2 À H 2 O + H] + (calcd m/z 428.9654). These data were also indicative of a chamigrane-type sesquiterpene, and this compound was suggested to be pacifenol (27), since MS data and RT were in agreement with those from pacifenol isolated from LC1 (L. catarinensis from Xavier Island). This metabolite was reported for some Laurencia species, such as L. liformis, 38 L. majuscula 39 and L. pacica. 40 Overall, the diagnosis of the above dereplication data reveals how the marine ecosystem can vary according to geographical location. Marine chemistry is strongly based on the interaction and diversity of animal, plant and microorganism species. 40 LC1 and LC2, two extracts obtained from the same species of algae collected in relatively close islands from the Southern Brazilian Coast, interestingly supported this fact. For LC1, L. catarinensis collected in Xavier Island, three chamigrane-type sesquiterpenes (20, 21 and 27) were characterized; alongside with seven C 15 -acetogenins (18,19,22,23,24,25,26), while for LC2, L. catarinensis collected in Arvoredo Island, three compounds related to caespitol (12, 13 and 14) were suggested, together with four chamigrene-type sesquiterpenes (10,11,16,17). Previous chemical work with L. catarinensis collected in Arvoredo Island 41 found mostly caespitol-related metabolites, but no C 15-acetogenins. Meanwhile, from the ongoing investigation of L. catarinensis from Xavier Island, the three chamigranes characterized in the extract LC1 have been isolated and structurally identied and some fractions presented NMR data suggestive of C 15 -acetogenins. It was observed that Canistrocarpus cervicornis from Rocas Atoll (CC1-3) clustered mostly with Dictyota mertensii from three different oceanic islands (DM1-3), while extracts from Dictyopteris spp. presented mostly isolated nodes. Nevertheless, the isolation of further metabolites is needed, in order to allow a better evaluation of those hypotheses. It is also possible that some of the discussed nodes correspond to new metabolites with the same molecular formula structures and similar fragmentation patterns to compounds already reported in the literature.
Previous studies on the Laurencia species also showed a different chemical composition in relation to the collection sites. 42,43 Prevezols A and B, alongside acetogenins were identi-ed from L. obtusa collected in Greece, 44,45 while the same species collected on the Italian coast produced obtusalenes V, VI, VII and IX. 45 Furthermore, L. microcladia from some Greek islands showed different chemical proles according to the collection sites. [46][47][48][49] No compound could be identied in the extracts of Dictyopteris spp., Laurencia dendroidea and L. intricata.

Organism collection and extraction
Ten brown (Dictyopteris plagiogramma, Dictyopteris jolyana, Dictyota mertensii and Canistrocarpus cervicornis) and four red algae (Laurencia catarinensis, L. dendroidea, L. intricata) samples were obtained from different sites and in different seasons (Table 1), and dried under cold air. The dried material was exhaustively extracted with dichloromethane and methanol (2 : 1) at room temperature and the crude extracts were dried under vacuum.

LC-MS analysis
The fourteen crude extracts and the isolated metabolites were dissolved in acetonitrile with a nal concentration at 2.0 mg mL À1 , and then ltered over nylon lters. The samples and the blank were injected in an aliquot of 2.0 mL to an Acquity UPLC system equipped with a BEH C18 column (2.1 mm Â 50 mm; 1.7 mm), PDA detector and a quaternary bomb. Temperatures of the sample tray and column were set at 20 C e 40 C. The ow rate was 0.3 mL min À1 of a gradient mobile phase of CH 3 CN/ H 2 O (0.1% formic acid): from 80% to 70% of H 2 O in 1 min and then decreasing until 15% of CH 3 CN in 11 min. This proportion lasted 2 min; from 15% to 80% in 1 min; 5 min of 80% of H 2 O were used to return to the initial condition.
The mass spectrometer was set to observe m/z in a range of 100-1500 in positive ESI mode using the FastDDA algorithm to acquire MS/MS spectra. The capillary voltage was set at 4.0 kV and the cone voltage at 40 V. The cone gas ow was 200 L h À1 with an ion source temperature of 80 C. The desolvation temperature was 300 C, and the desolvation gas ow was 900 L h À1 . The analyses were accurate, and leucine enkephalin was used as reference (lock mass m/z 556.2771). The collision energy was between 25 and 35 eV and argon was used as collision gas for the tandem mass. The soware MassLynx V4.1 (Waters Corporation, Milford, USA) was used to process the LC-MS data.
Aer preliminary MN analysis of the extracts, the same LC-MS method was applied to isolated metabolites for comparison of retention times and MS/MS data.

Molecular networking
All chromatograms and MS/MS spectra obtained for the 14 crude extracts were digitally converted into .mzML les using MSConvert soware (http://www.proteowizard.sourceforge.net), and then submitted to the Global Natural Product Social MN (GNPS) platform (http://gnps.ucsd.edu). The MN was generated by interconnecting MS/MS spectra with precursor ion mass tolerance of 0.002 Da, fragment ion mass tolerance of 0.02 Da, minimum peak intensity equal to 50, a minimum matched of fragment ions and library search minimum matched peaks equal to four. The remaining parameters were kept the same as suggested by the GNPS platform. The MN generated was displayed on Cytoscape 3.6.0.

Conclusions
The MN exploratory analyses indicated chemical differences among the extracts from different brown and red algae species, and among different collection sites for the same species, including sites closer to the coast (LC1 and LC2) and on the oceanic islands. For some extracts it was possible to identify mass values compatible with metabolites already reported in the literature, such as terpenoids (F1, F3, F4 and F7 besides individual nodes) and C 15 -acetogenins (F7 and individual nodes) already isolated from those or related species. Furthermore, one diterpene and three sesquiterpenes were isolated from Dictyota mertensii (DM2) and Laurencia catarinensis (LC2) extracts, respectively. Due to the limitations of mass spectrometry for structure assignment, the metabolites reported here could also have an isomeric skeleton of the proposed structure. Many nodes in the MN could not be characterized, and these metabolites could be used as targets for isolation in a future work.

Conflicts of interest
There are no conicts to declare.