Specialized Metabolite-Mediated Predation Defense in the Marine Actinobacterium Salinispora

ABSTRACT

The obligate marine actinobacterial genus Salinispora has become a model organism for natural product discovery, yet little is known about the ecological functions of the compounds produced by this taxon. The aims of this study were to assess the effects of live cultures and culture extracts from two Salinispora species on invertebrate predators. In choice-based feeding experiments using the bacterivorous nematode Caenorhabditis elegans, live cultures of both Salinispora species were less preferred than Escherichia coli. When given a choice between the two species, C. elegans preferred S. areniolca over S. tropica. Culture extracts from S. tropica deterred C. elegans, while those from S. arenicola did not, suggesting that compounds produced by S. tropica account for the feeding deterrence. Bioactivity-guided isolation linked compounds in the lomaiviticin series to the deterrent activity. Additional assays using the marine polychaete Ophryotrocha siberti and marine nematodes further support the deterrent activity of S. tropica against potential predators. These results provide evidence that Salinispora natural products function as a defense against predation and that the strategies of predation defense differ between closely related species.

IMPORTANCE Bacteria inhabiting marine sediments are subject to predation by bacterivorous eukaryotes. Here, we test the hypothesis that sediment-derived bacteria in the genus Salinispora produce biologically active natural products that function as a defense against predation. The results reveal that cultures and culture extracts of S. tropica deter feeding by Caenorhabditis elegans and negatively affect the habitat preference of a marine annelid (Ophryotrocha siberti). These activities were linked to the lomaiviticins, a series of cytotoxic compounds produced by S. tropica. Microbial natural products that function as a defense against predation represent a poorly understood trait that can influence community structure in marine sediments.

INTRODUCTION

Marine bacteria play fundamental roles in nutrient cycling, primary production, bioremediation, and other ecosystem processes (1–3). These roles are particularly pronounced in ocean sediments, where bacterial abundances can exceed 10⁹ cells per cm³ (4). Numerous marine organisms have evolved to prey on this relatively abundant resource, including viruses, protozoans, nematodes, and annelids (5–8). In response, marine bacteria have evolved various strategies to avoid predation (9, 10), including morphological adaptations such as increasing cell size (11), reduced cell surface hydrophobicity (12), and the formation of colonies or biofilms that are relatively impervious to eukaryotic predators (13, 14). The production of natural products that deter predation represents another defense strategy (15). Microbial natural products have been linked to predation defense in terrestrial systems (16–19) and include the production of potent antinematodal compounds by soil Actinobacteria in the genus Streptomyces (20, 21). However, there are few examples in which the natural products produced by marine bacteria have been linked to predation defense. One notable exception is violacein, which is produced by Pseudomonas luteoviolacea, inhibits protist feeding at submillimolar concentrations, and can cause nanoflagellate death following the ingestion of 1 to 2 bacterial cells (22). Mat and bloom-forming cyanobacteria are also known to produce natural products that deter predation; however, in these cases the effects were observed on herbivores that feed on macroscopic growth forms (23–25).

The relatively high abundance of bacteria in marine sediments, coupled with their spatially structured communities and the diversity of potential predators, provides a strong ecological rationale for the production of natural products that function as feeding deterrents. The marine actinobacterial genus Salinispora (order Micromonosporales, family Micromonosporaceae) represents a useful candidate to test for predation defense in that it is readily cultured, broadly distributed in marine sediments, and a rich source of biologically active natural products (26). Salinispora currently comprises nine species (27) with over 170 different natural product biosynthetic gene clusters (BGCs) identified to date (28). Of these, several are fixed at the species level (29), suggesting they represent ecotype-defining traits (30). These include BGCs encoding the cytotoxic compounds lomaiviticin and salinisporamide A, which are fixed in S. tropica, and the kinase inhibitor staurosporine, which is found in all S. arenicola strains examined to date. Recently it was shown that S. arenicola and S. tropica maintain different competitive strategies, with S. arenicola employing interference competition and S. tropica exploiting relatively fast growth (31). Evidence that these two closely related species have diverged ecologically provides a mechanism to support their cooccurrence in marine sediments.

Bacterivorous nematodes are common in marine sediments (32) and, thus, represent a potential target of bacterial predation defense. Although Caenorhabditis elegans does not occur in the marine environment, it is highly chemotactic (33) and has been exploited to study nematode responses to toxins and pathogens (34) and to identify inhibitors produced by marine bacteria (52). C. elegans exhibits preference for the types of bacteria on which it feeds, suggesting that individuals will migrate among bacteria in search of an optimal food source (35). With their relatively short growth cycle, ease of culturing, and highly developed chemotaxis (33), C. elegans is an ideal organism to use in feeding assays. Similarly, various groups of sediment inhabiting marine annelids feed on bacteria (36), making them potential targets for microbial predation defense. Despite this ecological rationale, we could find little evidence that natural products produced by marine bacteria have been tested for this functional role.

This study addressed the hypothesis that Salinispora natural products function as feeding deterrents targeting bacterivorous eukaryotes. The results of live culture assays revealed that C. elegans prefers S. arenicola over S. tropica. Organic extracts from S. tropica deterred C. elegans feeding, suggesting the presence of a chemical defense, with the activity subsequently linked to known cytotoxic compounds in the lomaiviticin class. Salinispora tropica extracts similarly deterred a marine annelid and were lethal to marine nematodes, providing further support that S. tropica natural products function as feeding deterrents.

RESULTS

Caenorhabditis elegans feeding assay.

C. elegans feeding preference was first tested in a choice assay offering Escherichia coli, its normal food source, and either of two Salinispora species. In these assays, 10 S. tropica or nine S. arenicola strains (see Table S1 in the supplemental material) were spot inoculated onto agar plates along with E. coli. C. elegans eggs were placed equidistant between the E. coli and Salinispora colonies, and the average percentage of C. elegans hatchlings associated with either colony type was quantified after 24 h. Six replicate plates were prepared for each Salinispora species with spatial associations used as a proxy for feeding preference (initial monitoring over 24, 30, and 36 h revealed no major changes over time). The results revealed a significant preference for E. coli (Fig. 1) regardless of the Salinispora species tested (n = 9 for S. tropica, n = 10 for S. arenicola, P < 0.05 for both, Wilcoxon rank sum test), despite one S. arenicola strain (CNS-820) being preferred over E. coli (Fig. S1). We next asked if there was a difference in C. elegans feeding preference between S. tropica and S. arenicola. These assays were performed in a similar manner, except 10 strains from each of the two Salinispora species were randomly paired against each other. In each of two trials comparing the Salinispora species, significantly more C. elegans cells were associated with S. arenicola colonies (Fig. 1) (n = 6 per trial, P < 0.05 for both trials, Wilcoxon rank sum test). While these assays did not directly measure feeding, consumption was often visibly supported by the presence of orange material characteristic of the Salinispora carotenoid sioxanthin (37) in the digestive tracts of C. elegans associating with Salinispora colonies.

FIG 1.

C. elegans feeding assay: live cultures. (A) Percentage of C. elegans individuals associated with live S. tropica or E. coli colonies (average for nine strains over six independent tests). (B) Percentage of C. elegans individuals associated with live S. arenicola or E. coli colonies (average for 10 strains over six independent tests). (C) Percentage of C. elegans individuals associated with live S. tropica or S. arenicola colonies (trial 1, n = 6). (D) Percent of C. elegans individuals associated with live S. tropica or S. arenicola colonies (trial 2, n = 6). Error bars represent standard errors. Asterisks denote Wilcoxon rank-sum test P values of 0.00001.

To determine if the difference in C. elegans feeding preference between the two Salinispora species was chemically mediated, we repeated the feeding assays using organic extracts derived from Salinispora cultures. Since Salinispora spp. produce antibiotics and other biologically active compounds that could inhibit E. coli growth (26), an autoclaved E. coli cell paste to which extracts could be added was developed as an alternative food source. In preliminary studies, this cell paste was visibly consumed and shown to maintain C. elegans viability over time. Salinispora cultures grown on agar plates were extracted with ethyl acetate and the extracts dried, resolubilized in 50 μl dimethyl sulfoxide (DMSO), and incorporated into the E. coli cell paste at a final concentration of 1 mg/ml based on observed yields of ca. 0.5 mg of crude extract per ml of Salinispora agar plate culture. When given the choice between an E. coli cell paste containing S. tropica culture extracts and extracts of the culture medium (controls), C. elegans preferred the medium controls in three of the four strains tested (Fig. S2). On average, the medium controls were preferred over the S. tropica extracts (Fig. 2), supporting the hypothesis that metabolites produced by S. tropica contribute to the reduced preference of C. elegans for this species. Conversely, C. elegans preferred extracts from S. arenicola over the cell paste controls (Fig. 2), with significant differences detected for three of five strains tested (Fig. S2). The difference from the live culture assays, where E. coli was preferred over S. arenicola (Fig. 1B), could be due to compounds present in the S. arenicola colonies that are missed by the extraction process or changes in the nutritional value of E. coli following autoclaving. A similar trend was observed in the S. tropica assays, where differences between treatments and controls were greater for live cultures than culture extracts. Physical characteristics of the live colonies could also contribute to these patterns.

FIG 2.

C. elegans feeding assay: culture extracts. Shown is the percentage of C. elegans associated with autoclaved E. coli paste treated with S. tropica or S. arenicola culture extracts relative to controls (E. coli paste plus ethyl acetate extract of uninoculated culture medium). An ethyl acetate extract of the uninoculated culture medium (A1) was also compared to a solvent-only control. Results are presented out of 100%. Averages are shown for four S. tropica or five S. arenicola strains, each of which was tested in six independent assays. Error bars represent standard errors. Asterisks denote Wilcoxon rank-sum test P values of <0.05.

Efforts were made to identify the compounds responsible for the feeding deterrence exhibited by the S. tropica extracts. To generate a larger extract, the three deterrent S. tropica strains (Fig. S2) were grown in liquid culture (2 L) and extracted with XAD resin, and the activity of the extracts was confirmed in the C. elegans feeding assay. Strain CNY-012 produced the most material and, thus, was selected for bioassay-guided fractionation. Normal-phase flash chromatography resulted in one fraction with nematode deterrent activity (Fr6, 100% methanol [MeOH]) (Fig. 3A, Fig. S3). Further purification using reversed-phase high-performance liquid chromatography (HPLC) led to an active fraction that contained lomaiviticin C based on comparisons with reported UV and mass spectral data. Further comparison with an authentic standard supported this identification (Fig. 3) along with what appeared to be lomaiviticin A and breakdown products. Both lomaiviticins are known to be cytotoxic, with lomaiviticin A inhibiting numerous cell lines at subnanomolar concentrations (38). While lomaiviticin C is the least cytotoxic analogue reported to date (38), an authentic lomaiviticin C standard nonetheless exhibited C. elegans feeding deterrence at 0.05 mg/ml E. coli food paste (Fig. S3).

FIG 3.

Bioassay guided fractionation and compound identification. (A) Fractionation scheme (check indicates significant C. elegans feeding deterrence). (B) Evaporative light scattering detector and 254-nm absorbance chromatograms for the 100% MeOH fraction. (C) UV spectrum for the major peak at 4.41-min matches a lomaiviticin C standard (bottom). (D) High-resolution mass spectrum for the major peak in Fr 2 is consistent with lomaiviticin C. (E) Structure of lomaiviticin C.

We next quantified lomaiviticin C in the three deterrent S. tropica extracts. Based on area under the peak calculations, the extract from the 2-L culture of strain CNY-012 contained 8.2 (±0.37) μM lomaiviticin C, followed by strains CNS-197 and CNH-898, which contained 5.2 (±0.39) μM and 3.4 (±0.20) μM, respectively (Fig. S4). These concentrations are greater than the cancer cell line 50% inhibitory concentration cytotoxicity values reported for lomaiviticin C (0.2 to 0.9 μM) (38) yet below the value that deters C. elegans feeding (0.05 mg/ml or 36 μM). Nonetheless, it remains possible that the concentrations surrounding live colonies are considerably greater than those measured in liquid culture and that more potent analogs present in the extract (e.g., lomaiviticin A) contribute to the C. elegans feeding deterrence, although none could be confidently identified. This may also explain the poor correlation between lomaiviticin C concentration (Fig. S4B) and feeding deterrence (Fig. S2A). Notably, lomaiviticin C was not detected in the extract from strain CNB-440 (Fig. S5), the only S. tropica extract that did not deter C. elegans feeding (Fig. S2). Identifying additional S. tropica strains that fail to produce this compound and generate similar results would provide further support for the role of lomaiviticin C in C. elegans feeding deterrence.

Effects against marine predators.

An agar plate assay was designed to test the substrate preference of the annelid Ophryotrocha siberti (Dorvilleidae, Eunicida). Ophryotrocha is a bactivorous group that will migrate toward areas rich in bacteria and detritus while grazing (36, 39). This assay was used to determine if O. siberti avoided substrates containing Salinispora cultures or culture extracts relative to medium controls. The substrates were A1 medium seeded with Salinispora cultures or culture extracts, and preference was determined by monitoring the position of five O. siberti individuals following placement in the center of a petri dish, one-half of which was comprised of treatment or control substrates. Ophryotrocha siberti individuals preferred the medium controls to live cultures of either S. tropica CNB-440 (n = 5; P < 0.0005) or S. arenicola CNS-205 (n = 5; P < 0.05) (Fig. 4A and B). Substrates containing extracts from S. tropica CNB-440 were significantly less preferred than controls (n = 10, P < 0.001), while there was no significant difference between treatment and control for S. arenicola, (n = 10, P = 0.26) (Fig. 4C and D). As in the C. elegans assays, there was some reduction in the activity observed in extracts relative to live cultures.

FIG 4.

Ophryotrocha siberti substrate preference assay. (A and B) Percentage of O. siberti individuals on A1 media containing live S. tropica (CNB-440) (A) or S. arenicola (CNS-205) (B) cultures versus medium controls measured at 5-min intervals and averaged over 1 h. (C and D) Percentage of O. siberti individuals on A1 medium containing extracts of S. tropica (CNB-440) (C) or S. arenicola (CNS-205) (D) versus extract controls measured at 5-min intervals and averaged over 1 h. Error bars represent standard errors (n = 5); asterisks denote Wilcoxon rank-sum test P values of <0.05.

We performed similar substrate preference assays using nematodes collected from sediments at the Smithsonian field station at Carrie Bow Cay, Belize, where Salinispora spp. are known to occur (40). Based on 18S rRNA sequence analysis, these nematodes were identified as Robbea hypermnestra (GenBank accession number MZ787961). In preliminary trials, the nematodes died within several hours of contacting agar containing Salinispora culture extracts. We then chose to monitor nematode survivorship on treated and control agar plates over a time course of 6 h using extracts from S. tropica CNB-440 and S. arenicola CNS-205, which were available at the time. When subjected to agar substrates containing 1 mg/ml final concentration of crude extract from S. tropica strain CNB-440, marine nematode survivorship was reduced to 0% within 6 h (Fig. 5). In contrast, the extract from S. arenicola CNS-205 and the medium controls had little effect, with the nematodes surviving for several days.

FIG 5.

Marine nematode survivorship curves. Percent nematode survival on A1 agar medium containing crude extracts (1 mg/ml) from S. tropica strain CNB-440 or S. arenicola strain CNS-205 compared with a volumetric equivalent of A1 medium extract. Error bars represent standard deviations (n = 3).

DISCUSSION

It is well documented that bacteria inhabiting marine sediments produce biologically active natural products (41, 42). These compounds have largely been explored for their biomedical potential, leaving major gaps in our understanding of their ecological functions. This study assessed the effects of two sediment-inhabiting Salinispora species on the feeding and habitat preference of bacterivorous eukaryotes. The assays tested both live cultures and culture extracts to assess the role of natural products in defense against predation. The results differed for the two species, with S. tropica exhibiting higher levels of predator deterrence. This was somewhat surprising given previous evidence that S. tropica invests in growth while S. arenicola invests in the production of antibacterial compounds (31). While the higher growth rate reported for S. tropica was evident in these assays, the results suggest that the natural products produced by the two species serve distinct ecological functions, providing further support for their ecological differentiation. Collectively, it appears that the slower-growing species S. arenicola is better adapted to compete against bacteria while chemical defenses in the faster-growing species S. tropica preferentially target eukaryotic predation. While these concepts require further testing, the results indicate that the ecological functions of compounds produced by sediment-inhabiting Salinispora species can be complex and include defense against predatory eukaryotes.

While it is well understood that marine invertebrates and seaweeds maintain chemical defenses against predation (43, 44), much less is known about this function for bacterial natural products, likely due in part to the limited availability of appropriate bioassays. In response, we developed methods to test the effects of bacterial cultures and culture extracts on C. elegans feeding preference. Bioassay-guided fractionation led to the isolation of compounds in the lomaivitacin series as the likely candidates for the deterrent activity detected from S. tropica. While not marine, C. elegans was selected based on past use in feeding preference studies (45) after attempts to develop a similar assay with marine nematodes failed. Despite difficulties working with marine nematodes, it was possible to show that extracts from S. tropica were toxic to field-collected animals while extracts at similar concentrations from S. arenicola were not. Similarly, it was possible to develop a substrate preference assay using the marine annelid O. siberti and demonstrate that extracts from S. tropica were less preferred than controls while extracts from S. arenicola had no significant effect. Since the S. tropica strain available for testing (CNB-440) did not produce detectable lomaiviticns, it is likely that other compounds are responsible for this activity.

Several additional issues regarding the ecological relevance of the C. elegans results warrant discussion. First, the test concentrations used here were based on laboratory cultures, which may not reflect natural concentrations. Similarly, it is unknown if the lomaiviticins are produced in situ, as was previously shown for the Salinispora metabolite staurosporine (40). If they are produced, determining ecologically relevant test concentrations for laboratory-based bioassays remains a major challenging given uncertainties over diffusion gradients and the spatial scales at which compounds are encountered by potential predators. A targeted analysis for lomaiviticins in marine sediments known to harbor S. tropica would be a useful first step toward assessing in situ production.

It was not surprising that lomaiviticins were implicated in the feeding deterrence associated with S. tropica. These compounds are potent cytotoxins, causing double-stranded DNA breaks at nanomolar to subnanomolar concentrations (46). The lomaiviticin BGC is present in all S. tropica strains sequenced to date, whereas it is absent from all S. arenicola strains (28). The fixation of this BGC in S. tropica indicates that the encoded compounds serve an important ecological role. The results presented here suggest that role includes defense against predation. Inactivation of the lom BGC will be an important next step to link lomaiviticins to predation defense and rule out the possibility that other compounds are also involved. The relatively low activity of the S. arenicola cultures and culture extracts was surprising given that all of the strains tested possessed the BGC for the cytotoxic kinase inhibitor staurosporine, which the species is well known to produce (29). Thus, if staurosporine was produced, it did not reach levels that deter C. elegans feeding. More work will be needed to determine the potential role of this compound as a defense against predation.

While predation can have a major effect on planktonic bacterial community structure (47), considerably less is known about the effects of predation on bacterial communities in marine sediments. The results observed here suggest that chemically mediated effects on eukaryote feeding and substrate preference could play a role in structuring sediment microbial communities. These roles could include the creation of spatial barriers that defend producing strains from predatory eukaryotes while providing associated defenses for nondefended community members. Secreted allelochemicals may also help explain the high diversity of microbes detected in marine sediments by providing a mechanism for microscale niche segregation. While sediment-derived marine bacteria are well known as a source of natural products, much remains to be learned about their roles in predation defense.

MATERIALS AND METHODS

Live culture feeding assay.

Ten S. arenicola and nine S. tropica strains with sequenced and annotated genomes were selected to maximize their collective biosynthetic gene cluster diversity (see Table S1 in the supplemental material). Caenorhabditis elegans N2 was provided by the Troemel laboratory, UCSD, and maintained on nematode growth medium (NGM; Fisher Scientific) seeded with 300 μl of E. coli OP50 (provided by the Troemel lab). Cultures were kept at 20 to 25°C and transferred to new plates every 3 days. Axenic C. elegans cultures were generated using published protocols (48). Briefly, C. elegans cultures were washed with 1 ml deionized (DI) water, after which 700 μl was transferred to a 1.5-ml Eppendorf tube, and 300 μl of a 2:1 solution of 5% sodium hypochlorite and 5 M sodium hydroxide was added to lyse the adult C. elegans and E. coli cells. The slurry was vigorously vortexed every 2 min for 10 min and spun at 1,300 × g for 30 s, and the pellet was washed three times with 800 μl of autoclaved DI water to remove any remaining bleach/sodium hydroxide solution. The washed pellet was mixed with 800 μl of DI water, egg concentration determined using a dissecting scope, and diluted with sterile DI water to a final concentration of 8 eggs/μl.

Salinispora strains were cultured in medium A1 (10 g starch, 4 g yeast extract, 2 g peptone, 750 mL 0.2-μm filtered seawater, 250 mL deionized water) for 7 days. Twenty-five microliters from each strain was then spot inoculated 4 cm apart on a 150-mm by 15-mm petri plate containing 75 ml of medium A1 with 16 g agar/L, such that all strains from the same species (either 9 or 10) were on one plate. Six replicate plates were generated in this fashion for each species. After 7 days, 25 μl of E. coli strain OP50, grown in 50 ml LB Miller broth (Fisher Scientific) for 2 days, was spotted equidistant between the Salinispora colonies. After 1 day of room temperature growth, approximately 75 C. elegans eggs were seeded between the E. coli and Salinispora colonies. The number of live C. elegans hatchlings on the E. coli and Salinispora colonies was counted after 24 h using a dissection scope. The six replicate plates for each species were averaged to determine the preference between Salinispora colonies and E. coli food sources. Assays to test for preference between the two Salinispora species were prepared in a similar manner, except 10 S. tropica and 10 S. arenicola strains were paired and E. coli was not offered as a choice. The strains paired were randomized on six replicate plates for the first trial and the process repeated using a new set of strains for the second trial.

Extract feeding assay.

One milliliter of Salinispora culture (grown as previously described) was spread onto the surface of 25 individual 100-cm-diameter A1 agar plates (1.4% agar) and allowed to grow for 15 days. The agar plates were then chopped into 1- by 1-cm blocks, shaken with ethyl acetate (EtOAc) in a 2.5-L culture flask for 2 h, filtered (qualitative p8 fluted filter paper; Fisher) to remove cell and agar debris, dried on a rotary evaporator, and stored at –4°C. For the feeding assays, 1 mg of extract was dissolved in 50 μl DMSO, transferred to a sterile 10-ml glass culture tube, and combined with 1 ml of autoclaved E. coli OP50 that had been cultured in 50 ml LB medium for 2 days. This resulted in an extract test concentration of ca. 1 mg/ml. Medium controls were similarly prepared and tested at concentrations equivalent to the culture volume required to generate 1 mg of extract. Extract-treated and control E. coli cell paste was placed 1 cm apart in 60-mm by 15-mm petri dishes containing 4 ml of solid NGM and ca. 75 C. elegans eggs placed between the food sources. On average, 6 plates were tested per trial. The number of C. elegans organisms per food source was counted 24 h after C. elegans addition.

Bioactivity-guided isolation.

Cultures were grown as previously described, after which 25 ml was transferred into 4× 1-L medium A1 in 2.8-L culture flasks. After 12 days with shaking at 195 rpm, ca. 5 g of a presterilized, 1:1 mixture of XAD-7 and XAD-16 Amberlite resin (Fisher Scientific) was added to each flask. The cultures grew for another 3 days, after which the resin and cell mass was collected using cheese cloth and extracted with 500 ml acetone. The acetone was removed under vacuum and the resulting extracts tested at 1 mg/ml for C. elegans feeding deterrence. Active extracts were subjected to normal-phase, flash silica column chromatography using the following scheme: 100% hexanes, 50%:50% EtOAc-hexanes, 100% EtOAc, 5%:95% MeOH-DCM, 20%:80% MeOH-DCM, and 100% MeOH. The fractions were dried under nitrogen, resolubilized at concentrations equivalent to their yields per liter of culture, and tested for feeding deterrence. The active fraction was further separated using reversed-phase preparative HPLC with the following gradient: 0 to 95% acetonitrile (0.1% trifluoroacetic acid [TFA]), 100 to 5% H₂O (0.1% TFA) over 20 min with the first 2 min discarded. For liquid chromatography-mass spectrometry (LC-MS) analysis, fractions at 1 mg/ml in MeOH were run through a Phenomenex Kinetex (Torrance, CA, USA) C₁₈ reversed-phase HPLC column (2.6 mm; 100 by 4.6 mm) using an Agilent 1260 LC system (Santa Clara, CA. USA) under the following LC conditions with 0.1% TFA: 1 to 2 min, 10% acetonitrile (MeCN); 2 to 14 min, 10 to 100% MeCN; 14 to 15 min, 100% MeCN at a flow rate of 0.7 ml/min. Analyses were with an Agilent 6530 Accurate-Mass quadrupole time of flight mass spectrometry (Q-TOF) with the divert valve set to waste for the first 2 min. Q-TOF MS settings were the following: positive ion mode; mass range, m/z 200 to 1,600; MS scan rate, 3/s; MS/MS scan rate, 5/s; fixed collision energy, 20 eV; source gas temperature, 300°C; gas flow, 11 L/min; nebulizer, 45 lb per in.² and scan source parameters, VCap 3000, fragmentor 100, skimmer1 65, and OctopoleRFPeak 750. The MS was autotuned using Agilent tuning solution in positive mode before each measurement, and the data were analyzed with the MassHunter software (Agilent). A lomaiviticin C standard was provided by Seth Herzon.

Ophryotrocha substrate preference assay.

Dorvilleids identified as Ophryotrocha siberti based on morphology were collected from the effluent line of the moon jelly tanks at the Birch Aquarium, Scripps Institution of Oceanography. This identification was confirmed based on cytochrome c oxidase (COI) sequence analysis (GenBank accession number MZ820650) performed using standard techniques (49). Voucher specimens were deposited in the Scripps Institution of Oceanography Benthic Invertebrate Collection (A 1152). They were maintained in culture on a frozen spinach diet. Substrate preference assays were performed in 60-mm-diameter petri plates containing 4 ml medium A1 (1.2% agar) on one-half of the plate and 4 ml of A1 seeded with a Salinispora culture on the other half. Five O. siberti individuals were added to the center of a plate, which was then flooded with 1.5 ml of sterile seawater to facilitate movement. Due to a limited supply of O. siberti, only one representative Salinispora strain was used for each species (CNB-440 for S. tropica and CNS-205 for S. arenicola). Time-lapse photography was used to document the position of O. siberti individuals every 5 min for 1 h and substrate preference recorded as the average percentage of individuals per substrate over the 12 time point measurements. Five replicate plates were tested for each trial with randomized plate orientation to negate abiotic factors such as light. The assay was repeated using Salinispora extracts (added when agar had cooled to 55°C, final concentration of 1 mg/ml) and extract controls generated from 4 ml of medium A1.

Marine nematode assay.

Intertidal sediment collected by bucket at the Smithsonian field station, Carrie Bow Cay, Belize, in September 2015 was rigorously mixed by hand and the overlying seawater decanted through a 2-mm mesh to collect suspended biota. The samples were transferred to a petri dish and nematodes collected with the aid of a stereomicroscope and identified as Robbea hypermnestra based on morphology and 18S rRNA sequence analysis (49). Medium A1 plates containing extracts (1 mg/ml final concentration) of either S. tropica (CNB-440) or S. arenicola (CNS-205) were prepared along with controls containing extracts of medium A1 at a volumetrically equivalent concentration. Ten nematodes were added to triplicate treatment and control plates and survivorship monitored over 6 h based on movement and response to prodding.

Statistical analysis.

The nonparametric Wilcoxon rank sum test (50) was used to evaluate significant differences between treatments or between treatments and controls. These analyses were conducted in R (51) and graphs generated in Microsoft Excel.