Chemical Defense against Herbivory in the Brown Marine Macroalga Padina gymnospora Could Be Attributed to a New Hydrocarbon Compound

Brown marine macroalga Padina gymnospora (Phaeophyceae, Ochrophyta) produces both secondary metabolites (phlorotannins) and precipitate calcium carbonate (CaCO3—aragonite) on its surface as potential defensive strategies against herbivory. Here, we have evaluated the effect of natural concentrations of organic extracts (dichloromethane—DI; ethyl acetate—EA and methanol—ME, and three isolated fractions) and mineralized tissues of P. gymnospora as chemical and physical resistance, respectively, against the sea urchin Lytechinus variegatus through experimental laboratory feeding bioassays. Fatty acids (FA), glycolipids (GLY), phlorotannins (PH) and hydrocarbons (HC) were also characterized and/or quantified in extracts and fractions from P. gymnospora using nuclear magnetic resonance (NMR) and gas chromatography (GC) coupled to mass spectrometry (CG/MS) or GC coupled to flame ionization detector (FID) and chemical analysis. Our results showed that chemicals from the EA extract of P. gymnospora were significantly important in reducing consumption by L. variegatus, but the CaCO3 did not act as a physical protection against consumption by this sea urchin. An enriched fraction containing 76% of the new hydrocarbon 5Z,8Z,11Z,14Z-heneicosatetraene exhibited a significant defensive property, while other chemicals found in minor amounts, such as GLY, PH, saturated and monounsaturated FAs and CaCO3 did not interfere with the susceptibility of P. gymnospora to L. variegatus consumption. We suggest that the unsaturation of the 5Z,8Z,11Z,14Z-heneicosatetraene from P. gymnospora is probably an important structural characteristic responsible for the defensive property verified against the sea urchin.


Introduction
For a little over 20 years, studies in marine chemical ecology have reaffirmed the performance of various molecular types of secondary metabolites of marine macroalgae as chemical defense against herbivores. Through experimental approaches, several terpenoid types of green [1,2], brown [3,4] and red [5,6] macroalgae have been evidenced as a chemical defense against different types of herbivores, such as sea gastropods, mollusks, urchins, fishes and others; while phlorotannins played a role in protecting only brown macroalgae, also against varied types of herbivores [7]. However, on a smaller scale or with rare examples, other molecular types of brown macroalgae have also been evidenced as a defense against herbivores, such as hydrocarbons [8], terpenoid bromoquinone [9], and solvent had no effect on these results, since the solvent-(CTRL ME) and non-solvent (CTRL)-containing foods were equally consumed by L. variegatus (Figure 1). Among the fractions from the EA extract, only FA3 was significantly less consumed than its respective control ( Figure 2, p < 0.05, ANOVA), while FA1 and FA2 were not consumed differently from their controls (Figure 2, p > 0.05, ANOVA).   solvent had no effect on these results, since the solvent-(CTRL ME) and non-solvent (CTRL)-containing foods were equally consumed by L. variegatus (Figure 1). Among the fractions from the EA extract, only FA3 was significantly less consumed than its respective control ( Figure 2, p < 0.05, ANOVA), while FA1 and FA2 were not consumed differently from their controls ( Figure 2, p > 0.05, ANOVA).   Mineralized (MIN) and demineralized (DEM) tissues of P. gymnospora were less consumed than their respective controls (green macroalga U. fasciata) (Figure 3, p < 0.05, ANOVA). However, both tissue-types of P. gymnospora were not consumed in a significantly different way by L. variegatus (Figure 3, p > 0.05, ANOVA).

Chemical Analyses
The chemical structure of the P. gymnospora major compound isolated from FA3 w suggested by NMR ( 1 H, APT experiments, HSQC and HMBC), GC/MS, GC/FID, TLC a RI analysis and comparison of our data with that from the literature [36][37][38][39][40]. It was id tified as the new compound 5Z,8Z,11Z,14Z-heneicosatetraene (C21:4) (Figure 4), a hyd carbon (HC) similar to 1Z,6Z,9Z,12Z,15Z-heneicosapentaene (C21:5) previously isola from the brown macroalga Fucus vesiculosus [40]. This new compound is an isome 1E,3Z,6Z,9Z-heneicosatetraene (C21:4) that was already identified in the moth Uteth ornatrix [38,39].  Figure 5B. The similarity between both mass data ( Figure 5A,B) strongly supports t 5Z,8Z,11Z,14Z-heneicosatetraene and arachidonic acid possess an identical molecu fragment of at least ca. 164 mass units as common structural characteristic. In addit The RI C of the major compound in FA3 agrees with literature data for an unsaturated HC with C21:4 (RI L , 2021) [37]. In addition, when the FA3 and FA standards chromatograms were compared, the peak of FA3 major HC (C21:4) did not match with any known FA peak ( Figure 6A,B). As mentioned before, the NIST library suggested 5Z,8Z,11Z,14Zeicosatetraenoic acid or arachidonic acid as the main compound in FA3 ( Figure 5B). In this way, NIST information guided the double bond positions at C-5,8,11,14 mainly because of the ions at m/z = 164 and 175 for 5Z,8Z,11Z,14Z-heneicosatetraene ( Figure 5A) that are similar to m/z = 166 and 177 for 5Z,8Z,11Z,14Z-eicosatetraenoic acid or arachidonic acid ( Figure 5B). This search in the library showed a great structural similarity between both HC, but as previously indicated, the major HC is a C21:4 (considering mass fragments and RI C ), such as this new compound eluted in GC/MS of the esterification products before the oleic (C18:1, n-9) and the stearic acids (C18:0) ( Figure 6B). The base peaks in the MS of 5Z,8Z,11Z,14Z-heneicosatetraene were registered at m/z = 79/91 ( Figure 5A), and no molecular peak could be observed, probably because of its high instability at 70 eV ( Figure 5A). The molecular ion at m/z = 288 [M.+] of the synthetic compound 1Z,3Z,6Z,9Zheneicosatetraene (C21:4) has been previously determined [38].   [37]. In addition, when the FA3 and FA standards chromatograms were compared, the peak of FA3 major HC (C21:4) did not match with any known FA peak ( Figure 6A,B). As mentioned before, the NIST library suggested 5Z,8Z,11Z,14Z-eicosatetraenoic acid or arachidonic acid as the main compound in FA3 ( Figure 5B). In this way, NIST information guided the double bond positions at C-5,8,11,14 mainly because of the ions at m/z = 164 and 175 for 5Z,8Z,11Z,14Z-heneicosatetraene (Figure 5A) that are similar to m/z = 166 and 177 for 5Z,8Z,11Z,14Z-eicosatetraenoic acid or arachidonic acid ( Figure 5B). This search in the library showed a great structural similarity between both HC, but as previously indicated, the major HC is a C21:4 (considering mass fragments and RIC), such as this new compound eluted in GC/MS of the esterification products before the oleic (C18:1, n-9) and the stearic acids (C18:0) ( Figure 6B). The base peaks in the MS of 5Z,8Z,11Z,14Z-heneicosatetraene were registered at m/z = 79/91 ( Figure  5A), and no molecular peak could be observed, probably because of its high instability at 70 eV ( Figure 5A). The molecular ion at m/z = 288 [M.+] of the synthetic compound 1Z,3Z,6Z,9Z-heneicosatetraene (C21:4) has been previously determined [38].  Considering all datasets presented and the fact that FA3 was extracted from a TLC spot with an Rf of 0.86, characteristic of HC [40], we have suggested that FA3 is composed of a mixture of FA and HC, but the major compound of this fraction was identified as the new natural unsaturated hydrocarbon 5Z,8Z,11Z,14Z-heneicosatetraene ( Figure 4, Table  1). Table 1. Gas chromatography coupled to Mass Spectrometry (GC/MS) analysis of fatty acids (FAs) and hydrocarbons (HC) from P. gymnospora in the fraction FA3. RI = Retention index. The FAs identification was made comparing the mass spectrum of FAME FAs and the calculated Retention Index (RIc in HP, 25 m × 0.20 mm i.d. × 0.33 μm, MS) with those from literature (RIL) for FAs (a) [41] and HCs (b) [37]. Results are presented as mean (%) ± standard deviation (n = 3). RT = Retention time.

Peak
Compound (FA and HC) RIc RIL Molecular Formula Lipid FA3 Considering all datasets presented and the fact that FA3 was extracted from a TLC spot with an Rf of 0.86, characteristic of HC [40], we have suggested that FA3 is composed of a mixture of FA and HC, but the major compound of this fraction was identified as the new natural unsaturated hydrocarbon 5Z,8Z,11Z,14Z-heneicosatetraene ( Figure 4, Table 1). Table 1. Gas chromatography coupled to Mass Spectrometry (GC/MS) analysis of fatty acids (FAs) and hydrocarbons (HC) from P. gymnospora in the fraction FA3. RI = Retention index. The FAs identification was made comparing the mass spectrum of FAME FAs and the calculated Retention Index (RIc in HP, 25 m × 0.20 mm i.d. × 0.33 µm, MS) with those from literature (RI L ) for FAs (a) [41] and HCs (b) [37]. Results are presented as mean (%) ± standard deviation (n = 3). RT = Retention time.
The PERMANOVA analyses determined significant differences regarding the FAs derivative compositions among samples (DI, EA, ME, FA1, FA2 and FA3, p < 0.001, PSEUDO-F = 123.16 and df = 5 (Table S2, Supplementary Materials). The pair-wise test showed that these differences are basically found in the FA3 and in the other samples (DI, EA, ME, FA1, FA2, p > 0.05 (Tables S3 and S4, Supplementary Materials).
Glycolipids TLC densitometry analyses showed that P. gymnospora ME extracts presented a higher content of SFL (sulfoquinovosyldiacylglycerols-SQDG, digalactosyldiacy lglycerols-DGDG) and CMH (ceramide monohexoside) GLY than DI and EA extracts (p = 0.0001). Monogalactosyldiacylglycerol (MGDG) did not vary among ME, DI and EA extracts (p = 0.5695). Finally, both MGDG and DGDG were found in significantly higher content in DI and EA extracts than SQDG and GLY (p = 0.0001).

Discussion
In the present study, we have investigated whether the marine brown macroalga Padina gymnospora exhibits chemical defense against consumption by the sea-urchin Lytechinus variegatus. Using an experimental laboratory approach, we have evidenced that the dichloromethane (DI) and ethyl acetate (EA) extracts show a defensive property against L. variegatus. To our knowledge, this is the first demonstration that P. gymnospora has a specific compound as a chemical defense against herbivory. However, it reaffirms t that the Padina species can be chemically defended against consumers, since extracts of Padina crassa, P. australis and P. japonica inhibited consumption by the abalones Haliotis discus hannai, H. discus discus and H. gigantea, and the snails Chlorostoma lischkei, Omphalius rusticus and O. pfeifferi carpenteri [15]. Additional evidence includes the defensive action of a P. tenuis extract against the fish Zebrasoma flavescens [10] and non-polyphenolic or non-polar secondary metabolites of Padina sp. That can inhibit fish communities in the field [23].
Feeding assays have also revealed that the methanolic extract (ME) of P. gymonospora was the most consumed by L. variegatus and contained low amounts of phlorotannins (PH). Our data showed that a low PH content in the P. gymnospora ME extract (0.23% ± 0.03 DW) used in feeding assays and from samples of the natural populations (0.11% ± 0.01 to 0.38% ± 0.02 DW) agree with the literature for tropical brown macroalgae (<0.5%) [35,46,47]. That said, these amounts are very low to inhibit herbivory [47]. For example, the PH extracted from tropical S. furcatum were a deterrent against the amphipods Parhyale hawaiensis only at concentrations of 2 and 5% in artificial food, but the natural amount (0.5%) did not deter feeding by this amphipod [47]. In this way, our data agree with some previous studies using temperate brown macroalgae that showed that phlorotannins are not used as chemical defense against herbivory [48,49].
The P. gymnospora ME extracts also contained sulfoquinovosyl diacylglycerols (SQDG), glycosphingolipids (GLY) and digalactosyldiacylglycerols (DGDG), but in higher amounts than in DI and EA extracts. Some studies have shown that isolated GLYs are macroalgal phagostimulants [11,18,50], while another studies have shown that isolated GLYs have inhibited herbivory [48]. DGDG enriched with unsaturated FAs from Padina arborecens have stimulated consumption by the marine gastropod Haliotis discus [18], as well as DGDG and 1,2-diacylglyceryl-4 -O-(N,N,N-trimethyl)-homoserine (DGTH) isolated from ME extract from Ulva pertusa have stimulated consumption by H. discus [50]. The present data lead us to infer that P. gymnospora GLY are probably phagostimulant components of the ME extract, but further studies with these isolated compounds are necessary to confirm this assumption.
In the present study, saturated FAs were detected as major chemicals in non-defensive extracts or fractions evaluated against L. variegatus, while fractions enriched with the unsaturated 5Z,8Z,11Z,14Z-heneicosatetraene inhibited the consumption by this sea urchin. Previous studies have shown that unsaturated FAs exhibited more effective defensive property than saturated FAs against bacteria [59][60][61] and against herbivores [48,55,56]. These data about defensive activity of unsaturated molecules led us to suggest that the unsaturation portion of the major compound 5Z,8Z,11Z,14Z-heneicosatetraene found in P. gymnospora would be important for the defensive action verified here against the L. variegatus.
Regarding the ecological role of the calcification (CaCO 3 ) as physical protection of P. gymnospora, both MIN and DEM tissues of this brown macroalga were not differentially consumed by L. variegatus. The calcified species Padina tenuis was also readily eaten by the sea hare Dolabella auricularia, probably due to its soft thallus being easily bitten by this mollusc [28]. Their low susceptibility to ingestion by this sea urchin was evidenced by the fact that P. gymnospora individuals were less consumed by L. varigatus than the corresponding control U. fasciata. Similar results were also previously reported about Padina durvillei which was less consumed by the sea urchin Echinometra vanbrunti than Ulva rigida in field trials [24]. In fact, the use of calcium carbonate as protection against herbivory is far from a consensus, not only for Padina species. For example, for two highly calcified species Corallina vancouveriensis and Corallina officinalis var. chilensis, the reductions in calcium carbonate content did not cause a significant increase in urchin grazing [62]. Alternatively, the mechanism by which CaCO 3 inhibits herbivory may simply be due to the decreased nutritional value of the macroalgae or chemicals that stimulate the consumption [28]. This second possibility may be true for P. gymonospora studied here, since it exhibits FA and GLY that could stimulate the consumption by L. variegatus and overrides the effect of CaCO 3 . The obtained results lead us to infer that in P. gymnospora chemicals probably provided by the new compound HC 5Z,8Z,11Z,14Z-heneicosa-tetraene is more important than the physical one (CaCO 3 mineralization) to defend this macroalga against L. variegatus.

Samples
Padina gymnospora specimens were collected from the intertidal zone at Rasa beach (Rio de Janeiro State; Brazil; 22 • 43 58 41 • 57 25 W). After collection, living macroalgal samples were stored in filtered seawater inside a dark isothermal chamber and transported to the laboratory. Thereafter, P. gymnospora individuals were maintained inside an 8 L aquarium with seawater enriched with Provasoli medium [63], under controlled conditions as already described elsewhere [35].

Feeding Bioassays
Feeding assays were carried out using an echinoid species, the generalist consumer Lytechinus variegatus, which feeds on marine algae [64], and usually avoids food items that possess structural aspects and/or chemical defenses [65].
Two kinds of L. variegatus feeding bioassays were carried out: (1) evaluation of defensive effect of natural concentrations of P. gymnospora extracts (dichloromethane, DI; ethyl acetate, EA; methanol, ME) and isolated fractions (FA 1 , FA 2 and FA 3 ) against this sea-urchin in artificial foods; (2) susceptibility of mineralized (MIN) and demineralized (DEM) tissues of P. gymnospora to this sea-urchin.
For the assays 1, artificial foods were prepared according to the usual method [66]. The artificial Control foods were prepared by adding 0.72 g of agar to 20.0 mL of distilled water, heating in a microwave oven until boiling point. This mixture was then added to 16.0 mL of distilled water containing 2.0 g of freeze-dried Ulva sp. (Chlorophyta), a highly preferred food item [42]. Control food (without extracts or isolated fractions), but with solvent (CTRL ME) and without solvents (CTRL), were also prepared in order to make sure that any eventual solvent residue was not an artifact interfering in the bioassay result. Treatment foods were similarly prepared, but the crude extract or fraction was first dissolved in CH 2 Cl 2 and added to the 2.0 g of freeze-dried Ulva sp. and then the solvent was removed by rotary evaporation. This procedure was necessary to obtain a uniform coating of the metabolite on the algal particles prior to addition to agar [43] before adding and following the food preparation.
Before the bioassays, individuals of L. variegatus were maintained in a recirculating laboratory aquarium at constant temperature (20 • C), salinity (35 PSU) and aeration. After an acclimation of 24 h, the bioassays were carried out. Treatments and controls were hardened into a nylon screen and cut into small pieces (10 × 10 squares), which were then simultaneously offered to the sea urchin L. variegatus (n = 10). The defensive property was estimated by comparing the number of consumed squares between treatment artificial foods (DI, EA, ME, FA 1 , FA 2 , FA 3 and solvent-ME) and controls foods with (WS) and without solvents (WTS).
For the second bioassay-type (2), mineralized (MIN) pieces of P. gymnospora tissues were treated with HNO 3 5% (3 × 30 min) for obtaining demineralized tissue (DEM) according to the usual method [35]. Both tissues (MIN and DEM) were washed in filtered seawater, inserted in Petri dishes and inspected with a stereomicroscope (Olympus SZX7, Tokyo, Japan) [35]. Afterwards, the water excess in MIN and DEM tissues of P. gymnospora and respective control (Ulva sp.) thallus was removed using a filter paper to obtain the wet weight of each tissue. Treatment (MIN or DEM) and respective control were simultaneously offered to the sea urchin L. variegatus (n = 10). After the end of the assay, the water excess from treatments and controls was removed again to obtain their wet weight. The mean difference between treatment and control was expressed as percentage (%) of consumption. Specimens of L. variegatus were maintained in the laboratory aquarium under the conditions as described in the first assay.
Three fractions were obtained from the EA extract, that were solubilized in methanol (Merck (Readington Township, NW, USA), using preparative one-dimensional TLC silica gel 60 F 254 S (Merck, Readington Township, NW, USA) for neutral lipids with the following mobile phase: hexane, diethyl ether and acetic acid (Merck, Readington Township, NW, USA) at 90:7.5:1 (v/v/v) [44]. These fractions were analyzed for the lipid content (FAs) using GC/MS, but only FA derivatives (FAs) and hydrocarbons (HC) were found. For this purpose, the isolated fractions from P. gymnospora EA were named FA1, FA2 and FA3, because they showed mainly fatty acid derivatives in its constitution with TLC and GC/MS analysis. These fractions presented the following TLC reference factors (Rfs): FA1 (0.20), FA2 (0.52) and FA3 (0.86). The yields were 0.6%, 0.5% and 0.1%, respectively.
For FAs analyses, 1 mg/mL of the P. gymnospora extracts (n = 3) and fractions (n = 3) were submitted to esterification: samples were dissolved in toluene (C 7 H 8 , Merck, Readington Township, NW, USA), and treated overnight, at 50 • C, with a 1% sulphuric acid (H 2 SO 4 ) solution in methanol (Merck, Readington Township, NW, USA). After that, a 5% aqueous sodium chloride (NaCl) solution was added to the reaction medium and the esters were extracted with hexane (Merck, Readington Township, NW, USA), using a Pasteur pipette to collect separated phases. The hexane layer was washed with 2% potassium bicarbonate (KHCO 3 ) in distilled water, and the mixture was evaporated under a nitrogen-saturated atmosphere. Thereafter, the sample was re-suspended in 50 µL of hexane, and a 1 µL aliquot was analysed using GC/MS. The analyses were performed in a Shimadzu QP2010 Plus GC instrument coupled to a Mass Spectrometry Detector (Shimadzu Corporation, Kyoto, KR, Japan), equipped with a Hewlett-Packard Ultra 2 polysiloxane capillary column (Hewlett-Packard Company, Palo Alto, CA, USA) (25 m × 0.20 mm i.d. × film thickness 0.33 µm). The injector temperature was maintained at 250 • C, and 1 µL aliquots were injected in the split mode ratio of 1:1. The column oven temperature was programmed to increase from 40 • C to 160 • C at 30 • C/min; and from 160 • C to 233 • C at 1 • C/min; and from 233 • C to 300 • C at 30 • C/min. After that, temperature was maintained at 300 • C for 10 min. Helium was used as a carrier gas at a constant flow rate of 1 mL/ min. Electron impact spectra were recorded in positive mode at 70 eV with a scan time of 1 s. Mass fragments were detected in full scan mode from 40 to 600 (m/z). The FA compounds in P. gymnospora extracts and fractions were identified by comparing their mass spectra with the mass spectra of FAME 37-methylated FA mix standards (Supelco, Sigma-Aldrich Company, Saint Louis, MO, USA) and the standard series of n-alkanes (C 7 -C 30 , Sigma-Aldrich Company, Saint Louis, MO, USA) obtained on the same equipment in identical conditions. Two injections were performed with (n = 3) and without esters extraction. The Retention Indices (RIs) of the FA3 were determined relative to the retention times of a series of n-alkanes (C 7 -C 30 ) with linear interpolation. GC/MS software version 2.53 (Shimadzu Corporation, Kyoto, Japan) was used for data processing.
The FA3 fraction was also analyzed in a Shimadzu GC 2010 (Shimadzu Corporation, Kyoto, Japan) coupled to a Flame Ionization Detector (FID), equipped with a DB5 (Agilent J & W, Santa Clara, CA, USA) fused silica capillary column (30 m × 0.25 mm i.d. × film thickness 0.25 µm). The oven temperature was programmed to 50 • C to 240 • C at 3 • C min/min, then hold at 240 • C for 20 min. Injector and detector temperatures were set and maintained at 220 • C and 290 • C, respectively. An analyzed sample was dissolved in CHCl 3 (Merck, Readington Township, NW, USA), and 1 µL aliquots were injected in the split mode with a ratio of 1:40 using H 2 as the carrier gas (1.44 mL/min). The relative amounts of the components were calculated based on GC peak areas without correction factors. FA 3 was also analyzed with 1D and 2D NMR techniques (NMR; 1 H and 13 C/400 MHz and 500 MHz, CD 3 OD. Both P. gymnospora extracts obtained with solvents of different polarities (DI, EA and ME; n = 5) and extracts from natural populations (n = 5) (which were submitted to acetone:water, 7:3, extraction) were analyzed using the FC method [45]. The quantification of PSs was performed by adding 1N FC reagent (Sigma-Aldrich, Saint Louis, MO, USA) to a 400 µL aliquot of diluted extract (100 µg/mL). The quantification was performed in a spectrophotometer Libra S-80 (Biochrom, Cambridge, UK) at 750 nm, using a calibration curve obtained with a phloroglucinol standard (Sigma-Aldrich, Saint Louis, MO, USA) at 10 µg/mL, 20 µg/mL, 30 µg/mL and 40 µg/mL (ABS = 0.1021 × Conc − 0.1197; r 2 = 0.99). PSs based on phloroglucinol content are represented as % of dry weight.

Statistical Uni-and Multivariate Analyses
Analyses of the results from the L. variegatus feeding bioassays with artificial foods containing P. gymnospora extracts (DI, EA, ME; n = 5), fractions (FA 1 , FA 2 and FA 3 ; n = 5) and fronds (MIN/DEM) were performed with a one-way ANOVA (with post hoc Tukey). The percentages of FAs derivatives (% of content) were compared among treatments (DI, EA, ME, FA 1 , FA 2 and FA 3 , n = 3) by the analysis of variance (PERMANOVA) with a Euclidean distance matrix and 999 permutations (significant results, p < 0.05).
To compare FAs compositions among treatments (DI, EA, ME, FA1, FA2 and FA3, n = 3), chromatogram peak areas (% of content) of the compounds obtained using GC/MS of each treatment were compared using Bray-Curtis similarity (Cluster and principal component analyses, PCA). FA data matrix included all the FAs derivatives reported in the Supplementary Materials (Figure). The one-way ANOVA (post hoc Tukey) was also used to evaluate the differences of PH and GLY levels among P. gymnospora extracts (DI, EA and ME; n = 5). Significant results were confirmed when p < 0.05 (α = 5%). Univariate statistical analyses were performed using STATISTICA software (version 6.0; StatSoft, Inc., Tulsa, OK, USA), and multivariate statistical analyses were conducted using the PRIMER software program (version 6.0; PRIMER-E Ltd., Ivybridge, UK).