Collections were made in May 2006 during low tide in the littoral zone along an extensive rocky shoreline peninsula between Playa Blanca and Boca Grande, Coiba Island, within Coiba National Park, Veraguas, Panama (07°23′50″ N, 81°39′00″ W) (Figure 1A). The collection site was situated on a large expanse of sedimentary rock weathered to sea-level, extending approximately 0.1 km² from the tip of the peninsula (Figure 1B). At low tide, many tidal pools and exposed boulders provided ample habitat for a variety of organisms on the underside of boulders and in tidal pools.

All cyanobacterial and algal samples were found growing in tidal pools located nearby the boulders where sea hares were collected. Three cyanobacterial samples, fixed to the sandy bottom, rocks or bedrock, were collected including a cf. Lyngbya sp. cyanobacterium in a mat assemblage, a cf. Lyngbya sp. cyanobacterium with a hair-like morphology, and a Symploca sp. (Figures 2A–C, respectively). Three algal species were collected including a red alga, Spyridia sp. and two green algae, Chaetomorpha sp. and Cladophora sp. (Figures 2D–F, respectively). A 4.5 L bag of each species was collected and then frozen at −20°C for chemical analyses. Voucher specimens were preserved in EtOH:seawater (70:30) and maintained at −20°C for identification. Additional samples of cyanobacteria and algae were maintained until required for feeding assays in aerated aquaria in fresh seawater with 12:12 h light:dark at ambient temperature at the Liquid Jungle Lab, Canales de Tierra, Pacific coast of Veraguas or at Naos Marine and Molecular Laboratories at the Smithsonian Tropical Research Institute, Panama City.

Nine S. rickettsi (Figure 3A) were collected in the field grazing on cf. Lyngbya sp. with the hair-like morphology (Figure 2B) and were allowed to continue to graze in the laboratory in aquaria (ambient temperature, 12:12 h light:dark) for 5 days, fasted for 24 h (to allow gut evacuation), euthanized in freezing seawater and stored at −20°C until tissue analysis was performed. Feeding studies were not conducted on these collections because of the extensive literature on the feeding preferences and metabolite sequestration of S. rickettsi (formerly S. striatus) (Kato and Scheuer, 1974; Rose et al., 1978; Paul and Pennings, 1991; Gallimore and Scheuer, 2000; Capper et al., 2006a; Cruz-Rivera and Paul, 2006).

Eighteen D. nicaraguana (Figure 3B) were collected in the same area as the food choices from the underside of uneven boulders in the intertidal zone. Three of these D. nicaraguana were used without laboratory feeding to assess their exposure to the food choices in the wild. They were fasted, euthanized, and stored as described above. Two additional animals died in the laboratory and were kept as vouchers. The remaining 13 animals were maintained in aquaria as above and were used in both multiple choice and no-choice feeding assays.

Dolabrifera nicaraguana (n = 12, average initial weight 2.6 ± 0.4 g) were placed individually in separate 2 L aquaria in fresh seawater. Each sea hare was provided a choice between five food types offered simultaneously: “mat-like” and “hair-like” morphologies of cf. Lyngbya sp. (average initial wet weight 809 ± 174 mg and 506 ± 117 mg, respectively); Symploca sp. (average initial wet weight 407 ± 109 mg); the red alga, Spyridia sp. (average initial wet weight 151 ± 32 mg); and the green alga, Chaetomorpha sp. (average initial wet weight 448 ± 121 mg). Food items were blotted, weighed, and placed at the bottom of 2 L aquaria with fresh seawater. There were twelve treatment (herbivore) aquaria and twelve control (no-herbivore) aquaria. The no-herbivore control aquaria were used to assess changes in algal mass throughout the experiment in the absence of herbivores (Cronin and Hay, 1996). Food items were removed after 60 h, blotted and reweighed. The amount of food consumed for each of the five food types was calculated for each replicate using the equation:

where T_i and T_f are the initial and final weights of the treatments, and C_i and C_f are the initial and final weights of the controls (Cronin and Hay, 1996). The Friedman test was used to detect significant differences in consumption of the different food types by D. nicaraguana (Erickson et al., 2006; Capper and Paul, 2008). A Nemenyi post hoc test was used to examine all possible pair-wise combinations for significant differences between groups (P < 0.05) (Conover, 1998).

To confirm food preferences, a no choice assay, using the same D. nicaraguana from the multiple-choice feeding assays above, was initiated after animals were fasted for 36 h. The no choice assays consisted of two D. nicaraguana feeding groups, including the cyanobacterium cf. Lyngbya sp. with “hair-like” morphology (n = 6, average initial wet weight 70 ± 22 mg) or the green alga Cladophora sp. (n = 6, average initial wet weight 41 ± 8 mg). The green alga Cladophora sp. was used instead of the green alga Chaetomorpha sp. because we had already established in the multiple-choice assay that the D. nicaraguana did not preferentially feed on Chaetomorpha sp. Food items were blotted dry, weighed, and divided into treatment and no-herbivore control group aquaria, as described above. The cf. Lyngbya sp. fed D. nicaraguana had an average initial animal weight of 2.1 ± 0.4 g while the Cladophora sp. fed D. nicaraguana had an average initial animal weight of 2.3 ± 0.3 g. A control group was included to assess changes in cyanobacterial (n = 6, average initial wet weight 65 ± 34 mg) and algal mass (n = 6, average initial wet weight 40 ± 12 mg) throughout the experiment in the absence of herbivores (Peterson and Renaud, 1989). After 60 h, food items were removed, blotted and weights taken as described above. Post-assay, sea hares were fasted for 24 h to allow gut evacuation, euthanized in freezing seawater, and stored at −20°C. The proportion of food consumed was calculated using equation 1. A Mann–Whitney U test was used to determine whether there were significant differences between algal and cyanobacterial fed groups.

Low-resolution mass spectra (MS) were obtained in MeOH and analyzed by direct injection on a JEOL LCmate mass spectrometer (Tokyo, Japan). Nuclear magnetic resonance (NMR) spectra were collected using a JEOL Eclipse 400 MHz spectrometer (United Kingdom). MS/MS analyses were conducted using a Finnigan LTQ MS (Thermo-Electron Corporation). High performance liquid chromatography (HPLC) was carried out in reverse phase using a Prontosil-120 C₁₈ analytical column (4.6 mm × 250 mm, Bischoff Chromatography, Leonberg, Germany) and a Merck Hitachi HPLC (Tokyo, Japan) containing dual pumps (L-7100) and a diode array detector (L-7455) monitoring at 210 nm.

Sea hare D. nicaraguana (n = 3) not used in the feeding assays, D. nicaraguana (n = 6) used in the multiple choice assay and in the cf. Lyngbya sp. no-choice feeding assay, and D. nicaraguana (n = 6) used in the multiple choice assay and in the green alga Cladophora sp. no-choice feeding assay were weighed and dissected, separating digestive gland, skin (including parapodia, foot, and head), mucus gland (also referred to as albumen gland), and grouping the remaining internal organs (Figure 3C and Supplementary Table 4). S. rickettsi (n = 9) were weighed and dissected, separating digestive gland, skin, and grouping the remaining internal organs (Supplementary Figure 1 and Supplementary Table 4). To obtain sufficient material for chemical analysis, tissue samples were grouped, with one or two individuals per replicate, as shown in Supplementary Table 4. Tissue samples were then lyophilized, weighed, and pulverized prior to dissolving in 1:1 EtOAc:MeOH, sonicating for 20 min, and extracting twice over 48 h. Samples were rinsed in 1:1 EtOAc:MeOH, filtered under vacuum, and dried via rotary evaporation. The excrement collected from D. nicaraguana and S. striatus during their 24 h fast, an egg mass from S. striatus, and mucus from D. nicaraguana were also dried and extracted as described above.

The cf. Lyngbya sp. cyanobacterium with “hair-like” morphology was lyophilized, ground, and 35 g of powdered sample was extracted by soaking in 1:1 EtOAc:MeOH and decanting three times over 72 h. The EtOAc:MeOH extract (given the code 9401) was collected and dried under rotary evaporation, producing 1.68 g of extract. The cf. Lyngbya sp. residue was then extracted by soaking in 1:1 EtOH:H₂O and decanting twice over 48 h. The EtOH:H₂O extract (given the code 9402) was collected, dried under rotary evaporation and then lyophilized to remove any remaining water, producing 0.4 g of extract. Both extracts were analyzed by MS, with the EtOAc:MeOH extract (9401) exhibiting mass peaks of m/z 887.1, 717.5, 703.5, 537.4, 313.3, 279.2, 245.2, 227.2 and 213.2. The EtOH:H₂O extract (9402) had two non-solvent peaks m/z 717.5 and 285.2.

The EtOAc:MeOH extract was fractionated using a Supelco Discovery DSC-18 reverse phase (RP) solid phase extraction (SPE) cartridge, with a bed weight of 10 g and 60 mL tube volume. The C₁₈ RP-SPE cartridge was first equilibrated in 1:1 MeOH:H₂O. After, loading 500 mg of the EtOAc:MeOH (9401) extract onto the cartridge, the sample was sequentially eluted under manually applied pressure with 150 ml each of 1:1, 3:2, 7:3, 4:1, and 5:1 MeOH:H₂O followed by 150 ml each of 100% MeOH, 100% EtOAc, and 100% acetone. The resulting eight fractions (coded 9401A through 9401H) were dried and assessed for bioactivity.

To rapidly assess a range of biological activity, in vitro bioassays already available through the Panama International Cooperative Biodiversity Group (ICBG) were employed as previously described (Moreno et al., 2011; Pavlik et al., 2013) to detect biological activity using P. falciparum (malaria), L. donovani (leishmaniasis), Trypanosoma cruzi (Chagas’ disease), MCF-7 breast cancer cells, and Vero mammalian cells (used to estimate overall cytotoxicity). Samples were initially screened at 10 μg/mL, with active samples then being further tested to determine the concentration to inhibit 50% growth (GI₅₀). Selectivity indices were calculated by dividing the GI₅₀ value for Vero cell cytotoxicity by the GI₅₀ value of the same sample toward the disease (e.g., P. falciparum, L. donavani, T. cruzi, MCF-7 cells).

Fraction 9401D, eluted with 4:1 MeOH:H₂O, was found to exhibit activity against P. falciparum in the malaria assay (GI₅₀ of 1.0 μg/mL) and was thus used for compound isolation via RP-HPLC. After dissolving in MeOH and filtering at 0.45 μm (Altech 17 mm PTFE syringe filters), compound isolation was accomplished using an isocratic system of 55% acetonitrile (CH₃CN) and 45% H₂O, with a flow rate of 1 ml/min to yield veraguamides M (1) and N (2), eluting at 23 and 31 min, respectively.

MS/MS was carried out on compounds 1 and 2 using a Finnigan LTQ MS (Thermo-Electron Corporation) using Tune Plus software version 1.0, as previously described (Mevers et al., 2011). Spectral files were converted to mzXML files (publicly accessible at http://gnps.ucsd.edu under MassIVE accession no. MSV000080055) and analyzed using previously described algorithms (Liu et al., 2009; Ng et al., 2009; Mohimani et al., 2011). Compounds 1 and 2 were analyzed by ¹H and ¹³C NMR, recorded in methylene chloride-d₂ (CD₂Cl₂, Cambridge Isotope Laboratories, Inc., Andover, MA, United States) (Supplementary Figures 2–5).

Veraguamide M (1): amorphous solid; [α]²³_D –49.5 (c 0.5, CHCl₃); ¹H NMR (400 MHz, CD₂Cl₂) δ 6.25 (d, J = 8.6 Hz, 1H), 4.88 (d, J = 7.7 Hz, 1H), 4.77 (p, J = 4.8 Hz, 2H), 4.69 (dd, J = 8.5, 5.3 Hz, 1H), 4.05 (d, J = 10.2 Hz, 1H), 3.97 (d, J = 10.4 Hz, 1H), 3.76 (dt, J = 9.4, 6.7 Hz, 1H), 3.55 (dt, J = 9.4, 6.9 Hz, 1H), 3.40 (s, 2H), 2.93 (d, J = 11.2 Hz, 6H), 2.35–2.15 (m, 3H), 2.15–1.88 (m, 3H), 1.81–1.62 (m, 1H), 1.67–1.35 (m, 4H), 1.32–1.08 (m, 4H), 1.07 (d, J = 6.5 Hz, 3H), 1.07–0.79 (m, 20H); ¹³C NMR (100 MHz, CD₂Cl₂) δ 173.3, 172.2, 171.0, 169.8, 165.9, 83.8, 76.9, 76.4, 68.6, 66.3, 64.0, 57.3, 52.6, 50.5, 47.3, 42.1, 38.8, 35.7, 35.2, 29.6, 29.4, 28.8, 28.3, 28.1, 26.0, 25.5, 25.1, 24.6, 23.7, 19.9, 19.7, 18.1, 16.4, 15.7, 14.1, 13.9, 11.4, 11.3, 10.7; ESIMS/MS m/z 689.44, 590.36, 524.35, 490.36, 470.31, 452.28, 411.33, 339.23, 297.23, 228.19. HRESIMS [M + H]⁺ m/z 717.4797 (calcd for C₃₉H₆₅N₄O₈, 717.4802).

Veraguamide N (2): amorphous solid; ¹H NMR (400 MHz, CD₂Cl₂) δ 6.22 (t, J = 10.5 Hz, 1H), 4.86 (dd, J = 13.9, 7.2 Hz, 1H), 4.80–4.73 (m, 2H), 4.71–4.64 (m, 2H), 4.07–3.85 (m, 2H), 3.80–3.67 (m, 1H), 3.59–3.48 (m, 1H), 3.41 (s, 2H), 2.98–2.87 (m, 7H), 2.29–2.10 (m, 2H), 2.04–1.89 (m, 2H), 1.66–1.59 (m, 2H), 1.44 (s, 2H), 1.22 (dd, J = 14.0, 8.4 Hz, 4H), 1.12–0.77 (m, 26H); ¹³C NMR (100 MHz, CD₂Cl₂) δ 173.3, 171.0, 170.8, 169.9, 165.7, 83.9, 77.3, 77.0, 76.5, 68.6, 66.3, 64.0, 57.2, 52.6, 47.2, 42.0, 38.9, 35.7, 35.2, 29.6, 29.4, 28.7, 28.2, 28.0, 25.9, 25.61, 25.57, 25.2, 23.7, 19.92, 19.87, 19.7, 18.2, 18.1, 17.8, 16.4, 15.7, 14.1, 11.4, 11.3, 10.8; HRESIMS [M + H]⁺ m/z 703.4641 (calcd for C₃₈H₆₃N₄O₈, 703.4646).

Extracts of sea hare tissues, egg mass, and excrement were analyzed via MS to determine the presence of compounds 1 and 2. In addition, ¹H-NMR of extracts of the sea hare digestive gland and skin tissues were obtained in CDCl₃ to compare with those of the pure compounds 1 and 2. HPLC was used to evaluate compound sequestration using an isocratic system of 68% CH₃CN and 32% H₂O at 1 mL/min. Compounds 1 and 2 eluted at 13 and 15 min, respectively, and were compared with chromatograms of the digestive glands and skin extracts from the feeding assay D. nicaraguana, control D. nicaraguana, and S. rickettsi.

In the multiple choice feeding assay, D. nicaraguana individuals were found to exhibit significant differences in preferences for the cyanobacterial and algal food types (Friedman test, P < 0.001). The Nemenyi post hoc test revealed no significant difference in preference between the two cf. Lyngbya species but found significant differences in consumption of the “hair-like” assemblage cf. Lyngbya sp. and the other food choices, including red alga Spyridia sp., cyanobacterium Symploca sp., and green alga Chaetomorpha sp. (P < 0.05, unadjusted; Figure 4A). Following the multiple choice assay, a no choice assay was performed using the “hair-like” cyanobacterium cf. Lyngbya sp. and the green algae Cladophora sp. In isolation, D. nicaraguana consumed more “hair-like” cf. Lyngbya sp. than it did the green alga Cladophora sp. (Figure 4B, Mann–Whitney U test, P = 0.008).

Given that there were significant preferences for the “hair-like” cf. Lyngbya sp. by D. nicaraguana, and that S. rickettsi, known to sequester cf. Lyngbya sp. secondary metabolites, was found grazing on the “hair-like” cf. Lyngbya sp., the “hair-like” collection of cf. Lyngbya sp. was used for further activity-guided isolation. Two extracts were obtained (EtOAc:MeOH and EtOH:H₂O) and analyzed by MS with the EtOAc:MeOH extract prioritized for further compound isolation. Fractions were tested for biological activity, with fraction D (eluted with 4:1 MeOH:H₂O) exhibiting activity against the malaria parasite P. falciparum (73.5% inhibition of parasite growth at 10 μg/mL) and MCF-7 cancer cells (70% cell death indicated by negative growth), and with little to no activity in the leishmaniasis or Chagas’ disease assays (18.3 and 9.7% inhibition, respectively), and with low cytotoxicity (GI₅₀ = 27 μg/mL). Compound isolation continued with RP HPLC, yielding two compounds (1 and 2) with molecular weights of 717 and 703 (Supplementary Figure 6).

Using HRESIMS, [M + H]⁺ peaks consistent with molecular formulas of C₃₉H₆₄N₄O₈ and C₃₈H₆₂N₄O₈, were obtained for compounds 1 and 2, respectively. Literature comparisons revealed several possible compounds with these molecular formula from cyanobacteria (Nakao et al., 1998; Mevers et al., 2011; Salvador et al., 2011) and ¹H and ¹³C NMR revealed substantially overlapping signals (Supplementary Figures 2–5). Therefore, MS/MS data was obtained for compound 1 and analyzed using software designed for sequencing cyclic peptides (Liu et al., 2009; Ng et al., 2009; Mohimani et al., 2011) and previously used with compounds from the same family [e.g., veraguamide E (3) (Mevers et al., 2011; Salvador et al., 2011)]. Using both manual and computational comparisons of the MS/MS fragmentation patterns, the locations of structural modifications were determined, as compared with veraguamide A (Supplementary Figures 7, 8). Compound 1 was found to have two rearrangements as compared with veraguamide E (3) including substitution of an N-Me-Ile in place of an N-Me-Val in the first residue clockwise from the HMoya, as well as an N-Me-Val in place of the N-Me-Ile for the fourth residue clockwise from the HMoya, thus resulting in designation as a new compound given the trivial name veraguamide M. Stereochemical assignment was not possible with the limited quantities of 1, although the sign and magnitude of the optical rotation is consistent with that reported for veraguamide E (3) (Salvador et al., 2011), making it likely that the absolute configuration of veraguamide M (1) is identical to that reported for compound 3. Similar analyses were performed for compound 2, resulting in the identification of another new compound named veraguamide N. Veraguamide N (2) was found to have one less methyl group, resulting in a 2-hydroxyisovaleric acid (Hiva) group in place of the 2-hydroxy-2-methylpentanoic acid (Hmpa) group found in the third residue clockwise from the HMoya group of compound 1 (Figure 5).

The animal tissues were analyzed under the same conditions as the compounds, veraguamides M (1) and N (2), and demonstrated that there was strong evidence of dietary compound sequestration by S. rickettsi, and possible evidence of compound sequestration by D. nicaraguana (Table 1 and Supplementary Figure 9). Sequestration of veraguamide M (1) and veraguamide N (2) was confirmed through MS, ¹H-NMR, and HPLC analysis in all three digestive gland replicates of S. rickettsi (Table 1 and Supplementary Figure 9). Additionally, veraguamide M (1) and veraguamide N (2) were evident in the MS and ¹H-NMR spectral data of the digestive gland of S. rickettsi (Table 1 and Supplementary Figure 9). These compounds were also evident in the MS of the other internal organs of S. rickettsi (Table 1). Excrement from the 24-h fasting period was evaluated using MS and showed evidence of veraguamide M (1), although an egg mass collected during the study did not show evidence of either compounds using MS (Table 1).

Dolabrifera nicaraguana, previously not known to sequester compounds from its diet, showed possible evidence of compound sequestration in our study (Table 1). Separate HPLC and MS analyses of the digestive gland of the D. nicaraguana fed the “hair-like” cf. Lyngbya sp. and of the tissues of D. nicaraguana not used in feeding assays exhibited low abundance signals indicative of possible evidence of the presence of veraguamide M (1) (Table 1). Veraguamide N (2) was not detected in D. nicaraguana tissues (Table 1). Other D. nicaraguana tissue samples, including skin, other internal organs, mucus gland, and mucus did not show evidence of compound sequestration, as assessed by MS and HPLC.

Veraguamide M (1) showed moderate activity against P. falciparum with a GI₅₀ value of 4.2 μM and moderate cytotoxicity to mammalian Vero cells with a GI₅₀ value of 29.3 μM (Table 2), resulting in a selectivity index of 7.0. Veraguamide N (2) showed moderate activity against P. falciparum with a GI₅₀ value of 4.3 μM and L. donovani with a GI₅₀ value of 6.9 μM, moderate cytotoxicity to mammalian Vero cells with a GI₅₀ value of 34.1 μM (Table 2), resulting in selectivity indices of 7.9 and 5.0 for antimalarial and anti-leishmanial activity, respectively.

Since both sequestered metabolites showed activity toward P. falciparum, sea hare tissues were also screened for bioactivity. D. nicaraguana tissue samples, including the digestive gland, skin and other internal organs, from animals used in the feeding assays and control animals were shown to be inactive (Table 1). In contrast, an extract from the S. rickettsi digestive gland showed similar antimalarial activity as the cyanobacterial compounds, veraguamides M (1) and N (2), with a GI₅₀ value of 3.0 μg/mL (Table 2). The S. rickettsi digestive gland extract was also cytotoxic to mammalian Vero cells with a GI₅₀ of 10 μg/mL, consistent with the detection of the two veraguamides by MS and NMR (Table 1) in the digestive gland extract. Extracts from the S. rickettsi skin, other internal organs, and excrement were shown to be inactive (Table 1).

Multiple choice feeding assay results revealed D. nicaraguana significantly preferred both cf. Lyngbya sp. morphologies “hair-like” and “mat” assemblage more than the cyanobacterium Symploca sp., the green alga Chaetomorpha sp., or the red alga Spyridia sp. In no-choice assays, the “hair-like” cf. Lyngbya sp. cyanobacterium was also significantly preferred to the green alga Cladophora sp. Whilst Prince and Johnson (2006) observed D. nicaraguana consuming cf. Lyngbya sp. in a no-choice feeding experiment, they also observed it consuming the green alga Enteromorpha clathrata. As no collections were made of E. clathrata, this alga could not be used in feeding preference assays reported herein. As a generalist grazer, it is likely that D. nicaraguana may capitalize on variety of food types, including diatoms, algal mats, and bacterial biofilms (Miller, 1969; Marshall and Willan, 1999; Rudman, 2003; Cimino and Ghiselin, 2009; Nimbs et al., 2017), as part of wider dietary repertoire. These grazing habits could be beneficial to D. nicaraguana, allowing it to survive varying environmental conditions and food availability. In a previous study in Panama City, at Punta Culebra on Naos island, scores of D. nicaraguana were reported to emerge to forage during the daytime once the tide fell below their tidal pools (Himstead and Wright, 2018). In our study, D. nicaraguana were also collected during this tidal period, although we did not find any individuals on the food treatment types used in our feeding assays, but rather nearby on the undersides of boulders. While D. nicaraguana preferred cf. Lyngbya sp. in our laboratory-based feeding assays, further research is required to assess the diet of D. nicaraguana in the wild.

Although there was possible evidence for sequestration of veraguamide M (1) in the digestive gland of D. nicaraguana, the low abundance signals found in the HPLC and MS analyses were not definitive. Additionally, D. nicaraguana tissue samples were biologically inactive, further suggesting minimal sequestration of veraguamides M (1) or N (2). It is possible that if the D. nicaraguana feeding assays ran for longer periods of time [e.g., 10 to 20 days (Pennings and Paul, 1993; Capper et al., 2005)] that these compounds may have bioaccumulated and demonstrated stronger evidence for sequestration. Additionally, detection of sequestered secondary metabolites may also depend on the ability of the organism to store or detoxify these compounds into less toxic metabolites (Capper et al., 2005). While some sea hares can sequester secondary metabolites with no apparent harm (Paul and Pennings, 1991), D. nicaraguana may not be able to tolerate these secondary metabolites and may, therefore, metabolize them into less harmful compounds.

In contrast, it is well known that Stylocheilus spp. sequester dietary-derived compounds from Lyngbya spp. and store them in their digestive glands (Pennings and Paul, 1993). Herein, S. rickettsi was found to sequester and store the cf. Lyngbya sp. compounds veraguamide M (1) and N (2), mainly in its digestive gland, but also in skin, other internal organs, and excrement. However, only the digestive gland of S. rickettsi exhibited biological activity, demonstrating the accumulation of sequestered compounds in the digestive gland. Moreover, the digestive gland extract showed similar antimalarial activity as the cyanobacterial compounds, while it was more cytotoxic toward mammalian Vero cells than were either of the sequestered compounds, suggesting there may be additional cytotoxic compounds stored in the S. rickettsi digestive gland. This is plausible given that Stylocheilus species process diverse secondary metabolites from their food (Paul and Pennings, 1991; Pennings and Paul, 1993; de Nys et al., 1996; Pennings et al., 1996; Capper et al., 2005) and concentrate the bulk of sequestered compounds in their digestive gland (Pennings and Paul, 1993). S. rickettsi provided an excellent opportunity to compare sequestration to an animal whose capability of sequestering was uncertain, in this case D. nicaraguana. In addition, this is the first study to demonstrate these phenomena in S. rickettsi, a species located in the Eastern Pacific. Additional studies incorporating the isolated compounds into artificial diets for sea hare feeding experiments are needed to fully assess consumption and subsequent compound sequestration by the sea hares.

Ongoing phylogenetic assessments of marine benthic cyanobacteria have revealed that these cyanobacteria are much more diverse than previously thought (Engene et al., 2011). Although cf. Lyngbya sp. and related species may be morphologically similar, their genetic diversity has allowed them to produce effective and biodiverse chemical defenses (Engene et al., 2013a). Stylocheilus are known to graze on several of these new genera (Capper et al., 2016), including cyanobacteria recently reclassified using molecular sequencing as Okeania sp., Moorena producens, Lyngbya sp., and Dapis sp. (Engene et al., 2013a,b, 2018; Tronholm and Engene, 2019). Stylocheilus are also known to exhibit an altered feeding preference based on secondary metabolite type (Capper and Paul, 2008; Capper et al., 2016). Thus, it may be that Stylocheilus species select their food based on morphology of cyanobacteria and the secondary metabolites they produce.

Stylocheilus species are located in tropical oceans and have been geographically separated for at least the last three million years, with the closure of the Isthmus of Panama (Bacon et al., 2015; O’Dea et al., 2016), diverging into three allopatric species found in the Indo-Pacific, Western Atlantic, and Eastern Pacific (Bazzicalupo et al., 2020). Although these geographical barriers likely resulted in Stylocheilus speciation, multiple species from this genus are known to evade predation by preferring metabolite rich Lyngbya spp. cyanobacterial assemblages, thus benefiting from the structural diversity of the defensive compounds found in these diverse cyanobacterial species. Results of this study are consistent with previous findings and confirm the food preference of S. rickettsi for these filamentous cyanobacteria, consistent with other species of Stylocheilus (Pennings and Paul, 1993; Pennings et al., 1996; Capper and Paul, 2008).

Veraguamides are encompassed within the kulolide superfamily of related cyclodepsipeptides produced by marine benthic cyanobacteria (Boudreau et al., 2012). In 2011, veraguamides A to L were published in parallel efforts (Mevers et al., 2011; Salvador et al., 2011) (two representative structures shown in Figure 5). Veraguamides A-G were isolated from Symploca cf. hydnoides from reef habitats in Guam (Salvador et al., 2011) and veraguamides A-C and H-L were isolated from Okeania sp. PAC-17-FEB-10-2 from an intertidal area offshore of a small island within Coiba National Park, Panama (Mevers et al., 2011). Interestingly, veraguamides A to C were isolated from two different cyanobacterial genera, as well as from geographically isolated locations in the Indo-Pacific and Eastern Pacific. It is possible that the cyanobacterial source of veraguamides M (1) and N (2), which we refer to as cf. Lyngbya sp., may be Okeania sp., since the morphologically similar cyanobacteria were both collected in Coiba National Park in Panama and contain similar compounds. However, it was not possible to carry out phylogenetic testing on the voucher specimens collected for the present study.

In 1998, two related depsipeptides within the kulolide superfamily, kulomo‘opunalide-1 (5) and -2 (6), were isolated from the marine gastropod Philinopsis speciosa, collected in the intertidal area offshore of O‘ahu, Hawai‘i (Nakao et al., 1998). P. speciosa is a generalist carnivore, predating on opisthobranch molluscs, including S. striatus (Nakao et al., 1998; Zamora-Silva and Malaquias, 2016). In feeding experiments, Nakao et al. (1998) demonstrated that P. speciosa fed on S. striatus, and they were able to isolate kulolide-1, a related depsipeptide, from both the predator and its prey. In addition, they noted that S. striatus feeds on cf. Lyngbya sp., although isolation and/or identification was not performed from the cyanobacterial food source (Nakao et al., 1998). However, inclusive of this current study, there are now three distinct reports of isolation of veraguamides from at least two genera of marine cyanobacteria (Mevers et al., 2011; Salvador et al., 2011). Thus, the dietary source of the veraguamide-type compounds, kulomo‘opunalide-1 (5) and -2 (6) and kulolide-1, were likely from marine cyanobacteria, sequestered by S. striatus and then accumulated by its predator P. speciosa. Additionally, P. speciosa had higher concentrations than S. striatus, suggesting that P. speciosa bioaccumulated this dietary-derived compound (Nakao et al., 1998).

Numerous compounds isolated from cyanobacteria have been shown to exhibit strong antimalarial properties (Supplementary Table 1; Fennell et al., 2003; Linington et al., 2007; McPhail et al., 2007; Barbaras et al., 2008; Gutierrez et al., 2010; Shao et al., 2015). Several cyanobacterial compounds, such as nostocarboline (Barbaras et al., 2008), dolastatin-10 (Fennell et al., 2003), and dragomabin (McPhail et al., 2007) have selectivity indexes two to three orders of magnitude greater than veraguamides M (1) and N (2). There are only a few known cyanobacterial compounds, however, with anti-leishmanial properties against L. donovani (Barbaras et al., 2008; Balunas et al., 2010, 2012; Sanchez et al., 2010). Herein, veraguamide N (2) was shown to exhibit modest anti-leishmanial properties (Supplementary Table 2). In addition, some of the previously isolated veraguamides showed moderate to weak cytotoxicity against cancer cell lines, although none of the previous veraguamide isolates were assessed for anti-leishmanial or antimalarial properties (Mevers et al., 2011).

Marine chemical ecology studies have important applications in the field of natural products based drug discovery (Simmons et al., 2005; Matthew et al., 2007; Wan et al., 2021). There are several examples of potential anticancer agents, sourced from marine cyanobacteria, which have been shown to be sequestered by the sea hare S. striatus (Luesch et al., 2002) including malyngamide A (Paul and Pennings, 1991), lyngbyatoxin A (Cardellina et al., 1979; Capper et al., 2005) and aplysiatoxin (Kato and Scheuer, 1974; Supplementary Table 3). The findings presented in this study provide further examples of dietary-derived cf. Lyngbya sp. compounds, veraguamide M (1) and N (2), sequestered by S. rickettsi and possibly by D. nicaraguana, which are also active against the clinically important diseases malaria and leishmaniasis.

Both sequestered compounds are active toward protozoan parasites, with this bioactivity evident in the digestive gland of the S. rickettsi. However, it remains untested whether dietary-acquired secondary metabolites provide defense against protozoan parasites in molluscs. Marine parasites, however, affect all trophic levels from individual to ecosystems, regulating host abundance, modifying traits, and can indirectly affect species interactions and community structure (Coen and Bishop, 2015). Molluscs have a complex response to parasites, including anatomic barriers, immune cell response, and physiological elements such as the complement system including proteins mainly synthesized in the liver (Al-Khalaifah and Al-Nasser, 2019).