Bacterial lipopolysaccharide induces settlement and metamorphosis in a marine larva

Significance

New surfaces in the sea are quickly populated by dense communities of invertebrate animals, whose establishment and maintenance require site-specific settlement of larvae from the plankton. Larvae selectively settle in sites where they can metamorphose and thrive largely due to inductive cues from bacteria residing on these surfaces. However, the nature of the cues used to identify “right places” has remained enigmatic. Here, we demonstrate that lipopolysaccharide, the main component of the outer membrane of Gram-negative bacteria from the bacterium Cellulophaga lytica, induces metamorphosis for a marine worm. We then discuss the likelihood that lipopolysaccharide provides the variation necessary to explain settlement site selectivity for many of the bottom-living invertebrate animals that metamorphose in response to bacterial biofilms.

Abstract

How larvae of the many phyla of marine invertebrates find places appropriate for settlement, metamorphosis, growth, and reproduction is an enduring question in marine science. Biofilm-induced metamorphosis has been observed in marine invertebrate larvae from nearly every major marine phylum. Despite the widespread nature of this phenomenon, the mechanism of induction remains poorly understood. The serpulid polychaete Hydroides elegans is a well established model for investigating bacteria-induced larval development. A broad range of biofilm bacterial species elicit larval metamorphosis in H. elegans via at least two mechanisms, including outer membrane vesicles (OMVs) and complexes of phage-tail bacteriocins. We investigated the interaction between larvae of H. elegans and the inductive bacterium Cellulophaga lytica, which produces an abundance of OMVs but not phage-tail bacteriocins. We asked whether the OMVs of C. lytica induce larval settlement due to cell membrane components or through delivery of specific cargo. Employing a biochemical structure–function approach with a strong ecological focus, the cells and OMVs produced by C. lytica were interrogated to determine the class of the inductive compounds. Here, we report that larvae of H. elegans are induced to metamorphose by lipopolysaccharide produced by C. lytica. The widespread prevalence of lipopolysaccharide and its associated taxonomic and structural variability suggest it may be a broadly employed cue for bacterially induced larval settlement of marine invertebrates.

More than two-thirds of the earth’s surface is covered by the oceans. The solid surface beneath these salt waters is greater than that of all the jungles, rain forests, boreal forests, prairies, deserts, tundras, and mountains combined. Yet, we know more about the dynamics of the biological world living above the ocean’s shores than we do about the enormously diverse benthic communities that extend from the higher reaches of the intertidal to the greatest depths of the seas. With global climate change causing massive melting of the earth’s glaciers, sea levels are rising, and the amount of land covered by the seas is increasing dramatically. The establishment and maintenance of benthic marine animal communities depends heavily on recruitment of minute larvae from the plankton. Larval recruitment is an essential process not only for all marine benthic communities, but also for creating harvestable populations of mariculture species, such as clams, oysters, and shrimp. It is also the initial process in the establishment of costly biofouling communities on the hulls of ships and in the pipes that bring cooling seawater to coastal electrical generators (1). A critical element of this process that remains poorly understood is how larvae select habitats suitable not just for attachment and metamorphosis, but long term survival, growth, and reproduction (2, 3).

In recent decades we have learned that the cues most invertebrate larvae employ to select right places to metamorphose are almost universally biological, the majority originating from microbes residing in surface biofilms (4, 5). Biofilm- or bacteria-induced settlement has been shown for larvae of sponges (6–8), cnidarians (9, 10), bryozoans (11, 12), molluscs (13–16), annelids (17), echinoderms (18, 19), crustaceans (20, 21), and urochordates (22) (i.e., all of the major invertebrate phyla). Although the number of larval types recorded to settle and metamorphose in response to bacterial films is now so great as to suggest a nearly universal mechanism for both settlement induction and larval response, we remain largely ignorant about the diversity of bacteria that stimulate larvae to settle, the chemical nature of the bacterial cues, and the mechanisms involved in larval recognition of bacteria.

Larvae of the serpulid polychaete Hydroides elegans provide an excellent model for investigating bacteria-stimulated settlement (17, 23). H. elegans is a cosmopolitan member of the benthic and biofouling communities in warm water bays and ports throughout the world (23). As is the case for a vast majority of marine invertebrates, an external stimulant induces an extremely rapid series of morphogenetic events that quickly transform the swimming larva of H. elegans into a bottom-living, tube-dwelling juvenile that will grow rapidly into a reproductive adult (23). Larvae of H. elegans typically do not settle in the absence of a biofilm, and this response follows direct contact with surface biofilms (17, 23–28). Investigations of the complex biofilm communities that induce settlement and metamorphosis in H. elegans and the cues that they produce must take place in the context of this ecology. Inductive cues must be naturally entrained within the biofilm and fast acting, because even in a low turnover environment such as Pearl Harbor, Hawaiʻi, current speeds and turbulence at the biofilm interface would otherwise rapidly disperse and remove dissolved compounds (29–31).

Evidence has accumulated that it is particular bacterial species residing in the biofilms that induce the metamorphic response in the larvae of H. elegans (5, 28, 32). Inductive bacteria from a broad range of phyla, including both Gram-negative and Gram-positive strains, are inductive when cultured in monospecies biofilms (24, 28, 32–35). However, most biofilm-bacterial species do not induce settlement in H. elegans (28, 36). To date, we have identified at least two different induction mechanisms from two different Gram-negative bacteria, Pseudoalteromonas luteoviolacea and Cellulophaga lytica. For P. luteoviolacea strain HI1, a highly inductive biofilm species isolated from Pearl Harbor, Hawaiʻi and genetically characterized in our laboratory, we identified specific structural elements derived from phage-tail proteins that are in some manner involved in settlement induction (37–39). However, examination of other inductive bacterial species revealed that induction of larval settlement cannot alone be explained by phage-tail bacteriocins. C. lytica, another biofilm bacterium, induces larvae of H. elegans to settle by an entirely different mechanism. Although cell-free preparations from broth cultures of C. lytica induce the tubeworm larvae to metamorphose, analysis of the genome of C. lytica (40) yielded none of the genes that transcribe the phage-tail elements of bacteriocins (34). Additionally, transmission electron microscopy (TEM) of inductive supernatants from C. lytica revealed an abundance of outer membrane vesicles (OMVs) and none of the bacteriocins found in similar preparations from P. luteoviolacea (34).

We have previously suggested that such membrane vesicles should be considered for their potential to be a common mechanism of interaction between biofilm bacteria and invertebrate larvae (34). Gram-negative bacteria ubiquitously produce OMVs (41, 42), but their structures and contents are variable per species and strain. Interactions between microbes and eukaryotes are mediated largely by a suite of conserved molecules knowns as microbe-associated molecular patterns (MAMPs) (43, 44). Two major MAMP classes, lipopolysaccharide (LPS) and peptidoglycan, comprise important components of the envelope of both whole cells and OMVs. Additionally, OMVs may enclose genetic material, virulence factors, proteins, small bioactive metabolites, and lipids (41). Consequently, OMVs provide the mechanism for multiple ecological roles fulfilled by Gram-negative bacteria, including cell-to-cell signaling and transfer of DNA, proteins, and small signaling molecules between cells (41, 42, 45–53). The ubiquity of OMVs means that the response to their detection by a eukaryote is entirely context dependent.

In the present study, we asked whether the OMVs of C. lytica induce larval settlement because they are accessible pieces of bacterial cell membranes or because they deliver a specific cargo. Therefore, to gain key insight into the source of the molecular structure of the larval settlement inducer produced by C. lytica, we utilized an approach that incorporated both chemical ecology and natural products chemistry (54–56). To determine if it is a protein, nucleic acid, or lipid, we subjected the OMV fractions to a battery of broad spectrum enzymatic treatments. To determine if secondary metabolites were involved, we employed bioassay-guided fractionations (informed by in silico analyses of the bacterial genome) on both OMVs and whole cells from C. lytica cultures. All treatments and chemical isolations were assessed for ecological relevance by testing their ability to induce metamorphosis in larvae of H. elegans. These approaches allowed us to reject all but bacterial LPS as cues for larval settlement. As the outermost element of Gram-negative bacterial cells, membrane LPS provides the surface-contact stimulus indicated by behavioral studies of larval settlement in H. elegans (26). We conclude LPS is a key element in selective larval settlement by H. elegans, and we predict this will hold true for many other benthic marine invertebrate animals.

Results

OMV Production, Size, and Metamorphic Bioactivity as a Function of Growth Phase in C. lytica.

OMVs isolated from broth cultures of C. lytica induced metamorphosis in larvae of H. elegans at all growth phases (midlog, late log, and stationary phase) with the highest average response to OMVs isolated during stationary phase and lowest from midlog phase OMVs [H(3) = 13.02, P < 0.001] (Fig. 1A). Notably, although all size classes (<3 kDa, 3 to 30 kDa, 30 to 100 kDa) induced metamorphosis in larvae of H. elegans [H(3) = 11.78, P < 0.001] (Fig. 1B), the highest abundance of vesicles was also observed during stationary phase (Fig. 1C).

Fig. 1.

Metamorphosis of larvae of H. elegans induced by OMVs from C. lytica. (A) Larvae of H. elegans metamorphose when exposed to OMVs from each growth phase (midlog, late log, and stationary) of a broth culture of C. lytica [H(4) = 13.02, P < 0.001]. (B) Larvae of H. elegans metamorphose when exposed to OMVs from different size classes [H(4) = 11.78, P < 0.001]. Metamorphosis was counted after 24-h exposure, with filtered sterile seawater (SW) serving as a negative control, and a multispecies (wild-type) biofilm serving as a positive control. Error bars indicate SE, n = 4. Letters show significant differences according to Dunn’s test. a = no significant difference from the wild-type biofilm (WT biofilm), b = significant difference from all. (C) Negatively stained TEM image of OMVs from the midlog and stationary stage of a culture of C. lytica. (Scale bars, 200 nm.)

Enzymatic Interrogations of OMVs for Larval Metamorphic Cues.

Nucleic acids and proteins.

The enzymatic interrogations of OMVs from C. lytica with nucleases and proteases had no impact on the metamorphosis-inducing capacity of the OMVs (Fig. 2A). The absence of nucleic acids in the OMV samples was confirmed by electrophoresis and UV absorbance, and the enzymatic destruction of proteins was confirmed by SDS/PAGE. Interestingly, both proteases had to be removed from the treatment by ultracentrifugation prior to larval assay, because the residual enzymes themselves induced metamorphosis (SI Appendix, Fig. S2).

Fig. 2.

Induction of metamorphosis in larvae of H. elegans by enzyme-treated OMVs from C. lytica. OMVs and seawater (SW) controls were exposed to either: (A) nucleases (DNase, 200 U, or RNase, 200 U) and proteases (trypsin, 20 µg·mL⁻¹) followed by either Proteinase K or Pronase E (100 µg·mL⁻¹) [H(12) = 33.42, P < 0.001]; (B) lipase (100 U), phospholipase A₁ (PLA1; 100 U), phospholipase A₂ (PLA2; 100 U), and phospholipase C (PLC; 100 U) [H(12) = 33.38, P < 0.001]; (C) lysozyme (500 U) [H(6) = 23.75, P < 0.001]. Metamorphosis was counted after 24-h exposure. Filtered sterile SW served as a negative control, and a multispecies (wild-type, WT) biofilm, a C. lytica biofilm, and an untreated C. lytica OMV sample were used as positive controls. Error bars indicate SE, n = 4. Letters indicate significant differences according to Dunn’s test.

Lipids.

The enzymatic interrogation of OMVs from C. lytica with lipases and phospholipases significantly decreased OMV induction of metamorphosis indicating the involvement of a bioactive lipid [H(3) = 33.38, P < 0.001] (Fig. 2B).

Peptidoglycan.

Enzymatic interrogation of OMVs from C. lytica with lysozyme (to break down peptidoglycan) had no impact on metamorphosis of larvae of H. elegans (Fig. 2C).

Bioassay-Guided Fractionation of C. lytica.

Initial extracts of the OMVs, cell-free supernatant, and cell pellets from C. lytica were significantly lower than the induction observed from wild-type biofilms but not C. lytica biofilms (Fig. 3). However, none of the HPLC fractions or purified compounds were as active as the parent samples (SI Appendix, Part C). The compounds isolated through the bioassay-guided fractionation of C. lytica were all extracted from moderately active HPLC fractions of the cells of C. lytica (SI Appendix, Figs. S2–S6). The amino acids (phenylalanine and leucine)—as well as niacin, thymidine, adenosine, and a 2,5-diketopiperazine of phenylalanine and proline—were isolated and had minor levels of bioactivity (SI Appendix, Fig. S5). Bioactivity of isolated compounds was consistent with their commercial counterparts (SI Appendix, Figs. S5 and S6). It should be noted that induction of larval metamorphosis did not begin until after more than 6 h of exposure to any of the fractions, which is exceptionally slow.

Fig. 3.

Induction of metamorphosis in larvae of H. elegans in response to methanol extracts (10 µg·mL⁻¹) of C. lytica culture components. (A) OMV pellet [H(4) = 13.35, P < 0.001]; (B) CLS containing OMVs [H(4) = 12.46, P < 0.001]; (C) CLP [H(4) = 17.4, P = 0.001]. Extracts were solubilized in ethanol, transferred to wells, evaporated to remove solvent, and sterile seawater (SW) and larvae were added. Solvent control: ethanol (EtOH); negative control: filtered sterile SW; positive controls: Wild-type biofilm and monospecific C. lytica biofilm. Metamorphosis was counted after 24-h exposure. Error bars indicate SE, n = 4. Letters indicate significant differences according to Dunn’s test.

Analysis of the genome of C. lytica with AntiSMASH (57–59) identified one potential region of secondary metabolite significance from 445,027 to 465,863 bp. A protein-BLAST of this contig identified it as a phytoene synthase (SI Appendix, Tables S1 and S2). Analysis of the genome of C. lytica with NaPDos (60) identified three potential ketosynthase domains, (SI Appendix, Table S3); however, closer examination showed that these sequences are involved in fatty acid biosynthesis and are unrelated to the production of bioactive secondary metabolites.

Analysis of Cell Envelope Components for Larval Metamorphic Cues.

A peptidoglycan focused extract of the cell envelope at first appeared to induce settlement in larvae of H. elegans, but the activity disappeared upon treatment with lipase, suggesting the isolated peptidoglycan was contaminated with a lipid. This activity was unaffected by treatment with lysozyme (Fig. 4). Commercial muramyl dipeptides and muramyl tripeptides, tested across a concentration range of 1 to 100 µM, resulted in no metamorphosis of larvae of H. elegans when tested separately or in combination (SI Appendix, Fig. S8).

Fig. 4.

Metamorphosis in the larvae of H. elegans in response to an SDS extract of cells of C. lytica. Metamorphosis was counted 24 h after exposure to the SDS extract (10 µg·mL⁻¹). SDS extract treated with lysozyme (500 U); SDS extract treated with lipase (100 U). Negative control: filtered sterile SW. Positive controls: wild-type biofilm; C. lytica biofilm; and C. lytica OMV sample. Error bars indicate SE, n = 4. Letters indicate significant differences according to Dunn’s test.

Because lipases significantly reduced settlement induction by OMV samples and no significantly inductive lipophilic small molecules were found in the natural products isolations, our focus returned to the remaining major lipophilic component of the bacterial outer membrane, LPS.

Extracted and purified LPS successfully induced metamorphosis in the larvae of H. elegans [62%; H(5) =19.5, P = 0.002] (Fig. 5A). Importantly, metamorphosis in larvae of H. elegans began less than 1 h after exposure to LPS samples. Enzymatic treatment of the LPS sample was consistent with the results gained from the OMV experiments; only lipase reduced the induction of metamorphosis in larvae of H. elegans (Fig. 5A). A dose–response curve generated for this LPS fraction yielded a peak of metamorphosis at 5 µg·mL⁻¹. Metamorphosis was not induced when the concentration of the fraction dropped below 0.5 µg·mL⁻¹, and toxicity and death were observed in larvae exposed to concentrations of LPS at or above 50 µg·mL⁻¹ (Fig. 5B). The absence of peptidoglycan and proteins from this preparation was confirmed using SDS/PAGE, comparing Coomassie blue staining with silver staining (SI Appendix, Fig. S7). Additionally, the absence of proteins was indicated by a lack of absorption of the sample at 260 nm (SI Appendix, Part E). The absence of other contaminating glycolipids was confirmed with SDS/PAGE gels stained with the lipophilic stain Pro-Q Emerald 300 (SI Appendix, Fig. S8) as well as thin-layer chromatography (TLC) developed with 5-hydroxy-1-tetralone. The Pro-Q Emerald 300 stain suggests that there could be at least 2 LPS molecules within the purified LPS sample.

Fig. 5.

Metamorphosis of larvae of Hydroides elegans exposed to purified LPS. (A) Induction of metamorphosis by LPS treated with lysozyme (500 U) or lipase (100 U) [H(6) = 27.06, P < 0.001]; negative control: filtered sterile seawater (SW). Positive controls: wild-type biofilm and biofilm of C. lytica. Error bars indicate SE, n = 4. Letters indicate significant differences according to Dunn’s test. (B) Metamorphosis dose–response to LPS isolated from C. lytica. Purity of extracted LPS can be found in SI Appendix, Part E.

Discussion

The broadly distributed marine polychaete H. elegans serves as a useful proxy for undoubtedly thousands of benthic marine invertebrate species whose larvae settle and metamorphose in response to specific biofilm-dwelling bacteria (6, 12, 23, 24, 34, 61–64). However, it can be difficult to know whether bioactive chemicals identified in the laboratory are the same as those that act in situ or simply stimulate processes downstream of the receptor (65–69). The practice of isolating and identifying compounds involves the exposure of the larva to substances with which it may not naturally interact (65). While experimentally valuable, artificial induction of larval metamorphosis does not enable us to learn more of the true induction of metamorphosis that is so essential to benthic marine organisms. Fortunately, the wealth of knowledge accumulated for larval settlement conditions provides sound criteria for the acceptance or rejection of the ecological relevance of any isolated cue. Our conditions for accepting an ecologically relevant cue to induce metamorphosis in H. elegans require it to be bacterial in origin, fast acting, and entrained to the biofilm (26). It was with this focus that we asked the question: What chemical component of marine biofilm bacterium C. lytica cues the larva of H. elegans to settle and metamorphose?

OMVs, a key component for bacterial interactions with the environment, can be considered a secretion or a delivery system (42, 70). Previous studies have shown that OMVs produced during different culture growth stages or of different sizes can vary in composition and biological activity (42, 70, 71). Our data, however, demonstrated bioactivity in all growth stages, with a steady but nonsignificant increase over time, suggesting that the bioactive compound produced by C. lytica may be either constitutively produced or has a long half-life and may proportionately increase in cultures over time. To determine whether OMVs induce metamorphosis through the delivery of “cargo,” we subjected OMV fractions from C. lytica to a battery of enzymatic treatments and bioassay-guided fractionation to identify components that impact larval settlement. Enzymatic interrogations of the OMVs strongly indicated that induction of metamorphosis was not due to a nucleic acid or protein cargo. Lipases, however, did inhibit induction of metamorphosis by the OMVs, a result that was consistent with earlier research by Alegado et al. (72) that had identified a sulfonolipid from another Bacteroidota that impacted development in a choanoflagellate.

The reduction in metamorphosis as a result of lipase treatment triggered an investigation for the presence of a bioactive lipid through bioassay-guided fractionation. The decision was made to use reverse-phase chromatography for the bioassay-guided search for compounds, because it allowed the broadest net that could capture both a bioactive lipid such as that from Alegado et al. (72) and also small bioactive molecules, such as those implicated in the metamorphosis of coral (64). The reverse-phase HPLC separation of extracts of cultures, cell-free supernatants, and OMVs from C. lytica failed to isolate and identify LPS as the bioactive agent. LPS compounds are large macromolecules with significant regions of nonpolarity (lipid A) that would not elute with the reverse-phase system chosen. Detection of LPS by the bioassay-guided fractionation would have required the use of a normal-phase system and is an important reminder of the selection effect of the separation system applied.

The bioassay-guided fractionation identified a few low-activity common small molecules, such as amino acids, but these were judged to not have ecological relevance. First, it is known that induction of metamorphosis can be stimulated in larvae of H. elegans by artificial inducers (25, 67). Second, previous studies into the induction of metamorphosis in H. elegans by dissolved free amino acids (DFAAs) rejected DFAA due to delayed onset of metamorphosis, low levels of metamorphosis, nonspecificity of DFAAs, undetectability of DFAAs in field studies, and the presence of contaminating bacterial biofilms in the initial experimental set-up (73–75). Third, the turbulent flow present at the field site, Pearl Harbor where H. elegans is abundant, would quickly dissipate FAAs and keep them well below inductive levels (29–31). Our results agree with these conclusions; when we tested amino acids, metamorphosis was rare and delayed. Additionally, all the extracts, HPLC fractions, and purified compounds were less active than their corresponding parent samples. For these reasons, we concluded that none of the low-activity metabolites from the bioassay-guided fractionations were involved in natural metamorphic induction. The in silico searches further confirmed the results from the bioassay-guided fractionation approach; no gene clusters or domains from known secondary metabolite biosynthetic pathways were identified that could transcribe compounds likely to induce complex larval metamorphosis.

The results of the enzymatic interrogations of the OMVs, in combination with the discovery that, in culture, OMVs were active settlement inducers regardless of their size or during which bacterial growth phase they were collected, suggested that OMVs induced metamorphosis because they were small pieces of cell membranes, rather than because they were acting as a delivery system. For this reason, we next focused our investigations on studies of the bacterial cell membranes. Cell membranes can contain a few larger lipid classes that may have been missed by the bioassay-guided fractionation, including phospholipids, glycolipids, and glycosphingolipids.

Phospholipids, although considered a major source for activation-dependent signaling molecules (76), can be almost immediately dismissed from the list of potential settlement inducers. TLC chromatography of the active LPS sample showed a complete absence of phospholipid. Furthermore, not only is phospholipid signaling overwhelmingly intracellular (76, 77), but its activity also arises from secondary messenger molecules derived from the membrane phospholipids. These secondary messenger molecules are produced when phospholipases selectively hydrolyze phospholipids (76, 77). Consequently, the phospholipases should have increased production of the secondary messenger molecules and resulted in higher metamorphosis and not the significant reduction in metamorphosis that was observed (Fig. 2). Alternatively, the reduced activity of OMVs treated with phospholipase could reflect the requirement for an intact or negatively charged membrane (77). Selective externalization of phospholipids in the outer membrane can alter the biophysical properties of the membrane affecting their binding properties (76). Additionally, intact membranes may hold active cue molecules in particular conformations that would not occur if their components were in a free solution. If that conformation is required to activate the larval receptor, removal of the compound from the membrane would result in a similar loss of activity. This explanation, however, was quickly rejected by the discovery of inductive activity among the remaining classes of compounds obtained by extraction of the cell envelope. Rather, the inhibition of induction by lipase treatments is most likely the result of nonspecific enzyme action.

The remaining classes of compounds included the important MAMP groups, peptidoglycan and LPS. Of these two groups, peptidoglycan has been shown to be an active component of nonpathogenic bacterial signaling in other systems (78, 79) and, for this reason, was chosen as the next component examined for its capacity to induce larval settlement. Results of larval assays utilizing isolated peptidoglycan, commercial peptidoglycan monomers, or OMVs subjected to lysozyme treatments were all negative. These results are not entirely unexpected since the peptidoglycan layer resides beneath the outer membrane and would not be the first point of interaction between a larva and a bacterium or an OMV.

LPS, which could be described as a “prototypical MAMP,” was found to induce metamorphosis in larvae of H. elegans when provided at a concentration that reflects the concentration of cells in an active biofilm of C. lytica (24). Although nonpathogenic roles for LPS are difficult to find, some have been identified in both the plant (80) and marine (nonmetamorphic) environments (43, 81). In addition to being a major MAMP, LPS molecules comprise the vast outer portion of the outer membrane of Gram-negative bacteria and, as such, would be the first bacterial molecules to interact with a larva upon contact, which is in line with how the process naturally occurs (26). Similarly, LPS forms the outer portion of the outer membrane of OMVs and is thus consistent with the observed activity of OMVs from C. lytica. It is important to note here that, although isolations of LPS can include residual traces of both peptidoglycan and other membrane lipids (phospholipids, glycosphingolipids, and so forth) (82), purity checks using both SDS/PAGE and TLC revealed none of these classes of compounds in the active LPS extract. Additionally, lipase and phospholipase treatments of the isolated LPS inhibited the inductive activity, consistent with the results from the OMV samples and suggesting that nonspecific enzyme activity occurred. In other words, the discovery that the inductive cue for settlement and metamorphosis of larvae of H. elegans is comprised of bacterial LPS is consistent with all of the observations made previously and in this study.

While LPS may largely explain the widespread occurrence of inductive activity across Gram-negative biofilm bacteria, it cannot account for the same activity found in Gram-positive bacteria (34) that lack both an outer membrane and its LPS element. Instead, the cell envelope of Gram-positive bacteria consists of a thickened cell wall of peptidoglycan that can be populated by teichoic and lipoteichoic acids (LTA) (83, 84). LTA is a major MAMP of Gram-positive bacteria; it is a component of nearly all Gram-positive membranes, as well as a functional equivalent to LPS (85). Thus, we predict that LTA from the inductive Gram-positive bacteria will be the active moiety in induction of metamorphosis. However, given the great predominance of Gram-negative bacteria in marine biofilms (86) may render this possibility irrelevant (i.e., settlement induction in nature may always be by Gram-negative bacteria).

The strong immunogenicity of LPS, combined with its ubiquitous expression by Gram-negative bacteria and inherent structural variability, serve to make it both a likely target of evolutionarily conserved innate immune receptors in larvae as well as suggest a possibly widespread mechanism for bacterial induction of marine invertebrate metamorphosis. LPS is a highly variable macromolecule with this variability tied closely to bacterial taxonomic lines and growth conditions (87–91). Indeed, our purified LPS sample appears to contain at least three different LPS molecules, which is not unusual (92). LPS molecules are composed of three main components: lipid A (endotoxin), an inner core oligosaccharide (kDO), and an outer core oligosaccharide (O-antigen), with variation occurring primarily in the lipid A and O-antigen components (92). While beyond the scope of this paper, determining the location of the inductive portion of the macromolecule or whether it is a specific moiety of a specific LPS molecule produced by C. lytica will comprise our next focus of study. If variations in the structure of LPS are responsible for the observed inductive activity of different species and strains of bacteria, it can explain why many—but definitely not all—marine Gram-negative bacteria induce settlement and metamorphosis in the larvae of H. elegans (17, 23, 28) and other invertebrate species. It may also explain why not all strains of the same bacterium induce larvae of a single invertebrate species to settle (37). Most importantly, the inherent variation of LPS molecules could explain the differential settlement of different invertebrates in different habitats.

Materials and Methods

Larval Culture and Bioassays.

Dioecious adult H. elegans were collected from docks in Pearl Harbor and maintained in the laboratory in continuously flowing, unfiltered seawater for several weeks without a noticeable decrease in fecundity. Gametes were obtained and larval cultured according to Nedved and Hadfield (23). Briefly, worms were induced to spawn by removal from their tubes. Gametes were mixed. Fertilization and development proceeded rapidly to a feeding trochophore stage in 1 d. Larvae were fed the single-celled alga Isochrysis galbana Tahitian Strain at a concentration of ∼60,000 cells·mL⁻¹. The larvae developed to the metamorphically competent nectochaete stage by day 5 and were utilized in metamorphosis experiments for 1 to 2 d. Because larvae of H. elegans will not metamorphose without exposure to bacterial biofilms or their products, wild, complex natural films developed on glass slides readily served as positive larval controls, and maintaining the larvae in double-filtered, autoclaved seawater served as negative larval controls. Assays also included a biofilm of C. lytica for reference. Assays were carried out in 24-well plates, and the percent of larvae that metamorphosed was determined at 20 to 24 h. Significant differences (P < 0.05) were calculated using Kruskal–Wallis with pairwise comparisons performed with Dunn’s test and false-detection rate correction (93) in Graphpad Prism 9. All experiments included four replicates/treatment with a minimum of 20 larvae per replicate and were repeated independently four times.

Bacterial Culture and OMV Isolation.

C. lytica was grown as previously described by Freckelton et al. (34) on 1/2 FSWt media (94). Briefly, cryogen bacterial stocks were streaked onto 1/2 FSWt agar plates and incubated for 24 to 48 h at 25 °C. Single colonies were then used to inoculate 10-mL cultures of 1/2 FSWt broth and incubated for 12 h at 25 °C. An aliquot of this starter culture was then used to inoculate 200-mL cultures and incubated for a further 14 h at 25 °C. Cells were harvested by low-speed centrifugation (4,000 × g, 25 min 4 °C). The cell-free broth was then filtered (0.2 µm; Millipore) to remove any remaining cells and then ultracentrifuged (200,000 × g, 2 h, 4 °C) (34, 51) to isolate OMVs. All samples were examined by TEM using negatively stained preparations to confirm the presence of OMVs, as previously described (34).

Determination of Relative Concentration of OMVs.

OMV production is highly dependent on growth conditions, and pelagic and biofilm bacterial populations can produce widely disparate amounts of OMVs (41, 42). The amount of OMVs produced by a bacterium can likewise be impacted by the growth phase of the population or the presence of other species. This variability results in difficulties in predicting an ecologically relevant concentration for testing. Consequently, isolated OMV suspensions were diluted and tested across three orders of magnitude. The lower (0.001×) dilution reflects that of the number of cells in 1-h biofilm (10⁵ cells/mm²) (24) and the highest concentration (1×) reflect the concentration used to inoculate those biofilms (10⁸ cells·mL⁻¹). For clarity of presentations all figures present the data for the lowest dilution only, unless otherwise stated. Additional OMV dilutions are presented in the SI Appendix, Part A.

To maintain consistent relative concentrations of OMVs between culture batches, an aliquot of each sample was stained with the fluorescent lipophilic styryl dye FM-4-64 (Molecular Probes; T-3166) following the protocol described by Schooling and Beverage (46). Briefly, samples were suspended in Hank’s buffered saline solution and stained with 20 µM FM-4-64 in the dark for 20 min at 25 °C. Fluorescence was read using a Molecular Devices SpectraMax M5 spectrophotometer (EX: 544 EM:680). Cholesterol was used to calibrate the dye’s response.

Analysis of OMVs from Different Growth Phases of Cultures of C. lytica.

Cultures of C. lytica were monitored for growth rate by measuring the OD₆₀₀ of an aliquot hourly to determine the growth phase. This process was repeated with triplicate cultures, and samples were collected at midlog, late log, and stationary phases. OMVs were isolated from each sample time point as described above and diluted to reflect the concentration of a culture at 10⁵ cells·mL⁻¹ (as above) in keeping with their FM-46-4 readings.

Analysis of OMV Size Classes.

Cell-free, filtered (0.2 µm), conditioned media was used to isolate OMVs. OMVs from C. lytica were separated into size classes by ultrafiltration (Millipore, Centricon Plus-70; 3 kDa, 30 kDa, 100 kDa). Aliquots (2 × 70 mL) of conditioned media were applied to each ultrafilter and then centrifuged (4,000 × g, 30 min, 4 °C). Aliquots of the retentate and filtrate were taken for metamorphosis assays and TEM analysis. The remaining filtrate was then applied to the next ultrafilter (100 kDa > 30 kDa > 3 kDa).

Enzyme Treatment of OMVs.

Isolated OMVs were combined with enzymes (nucleases, proteases, lipases, or lysozyme) and incubated in a water bath at the optimal temperatures and time periods described in Table 1. Lysozyme is an enzyme that catalyzes the breakdown of 1,4-β-linkages between N-acetylmuramic acid and N-acetyl-d-glucosamine residues in peptidoglycan (95). Enzyme action was halted by heat treatment alone, except for the proteases, which had to be removed from the treatment by ultracentrifugation prior to larval assay, because the residual enzymes themselves induced metamorphosis (SI Appendix, Part B and Fig. S1). For each enzyme or set of enzymes, a seawater control was also performed to ensure that the responses observed were not a result of the enzymes or heat treatments alone. Successes of nuclease treatments were assessed by spectrometry and electrophoresis. Successes of protease treatments were assessed by spectrometry and SDS/PAGE.

Table 1.

Enzyme treatments of OMVs from C. lytica

Cell component targeted	Enzyme (source)	Incubation			Thermal inactivation
		Concentration	Time, h	°C	°C	Time, min
Nucleic acid	DNase (105)	200 U	2.5	37 °C	75 °C	10
	RNase (106)	200 U	2.5	37 °C	75 °C	10
Proteins	Proteinase K (107)	100 µg·mL−1	2.5	55 °C	100 °C	10
	Pronase E (108)	100 µg·mL−1	2.5	55 °C	100 °C	10
	Trypsin (109)	20 µg·mL−1	2.5	37 °C	100 °C	10
Peptidoglycan	Lysozyme (110)	500 U	2.5	37 °C	100 °C	10
Lipid	Lipase (111)	100 U	2.5	55 °C	100 °C	10
	Phospholipase A1 (112)	100 U	2.5	55 °C	100 °C	10
	Phospholipase A2 (113)	100 U	2.5	55 °C	100 °C	10
	Phospholipase C (114)	100 U	2.5	55 °C	100 °C	10

Bioassay-Guided Fractionation of C. lytica.

Cell pellets (CLP), cell-free supernates (CLS) and OMVs were extracted with organic solvents directly to identify metamorphosis-inducing molecules. CLP and OMV samples were extracted with methanol (MeOH) followed by a dichloromethane-methanol mixture (DCM: MeOH, 1:1). Methanol was used as the first extracting solvent to both quickly quench the cells and to remove any residual water. As this aqueous methanol extract would have primarily targeted polar molecules, this extraction was followed by a DCM: MeOH to recover any nonpolar molecules. Some nonpolar molecules will be extracted by both solvent systems but should exist in higher concentration in the DCM: MeOH extract. CLS (3.5 L) was extracted by liquid:liquid partition with ethyl acetate (EtOAc), with nonpolar compounds extracted into the EtOAc layer.

The initial OMV extracts were subjected to metamorphosis bioassays first at 10 µg·mL⁻¹. Any samples that produced a toxic response in the larvae were reassayed at 1 µg·mL⁻¹ and 0.1 µg·mL⁻¹. Extracts were added to wells by dissolving in MeOH at the concentration required for the addition of no more than 50 µL of MeOH to sample wells. MeOH was allowed to evaporate before drying overnight in a lyophilizer to remove any residual MeOH. An MeOH solvent control was added to the assay to ensure any observed induction of metamorphosis was not due to the residual presence of MeOH.

Additionally, each extract (CLP, CLS, OMV) was subjected to reverse-phase HPLC (Thermo Scientific 3000), and all fractions were subjected to metamorphosis bioassay with larvae of H. elegans. CLP HPLC fractions were generated for bioactivity testing through an acetonitrile gradient (10 to 25%, 20 min; 25 to 40%, 10 min; 40 to 100%, 20 min; 100%, 10 min). Active fractions were further purified using HPLC (Phenomenex, Luna, 5 µm, C18, 100 Å, 250 × 100 mm, MeOH 10%, 30 min). For CLS, extracts were fractionated using a Phenyl-Hexyl column (Phenomenex, Luna, 5 µm, Phenyl-Hexyl, 250 × 100 mm) and a water: acetonitrile gradient. ¹H NMR and liquid chromatography/electrospray ionization/mass spectrometry (LC-ESI-MS) data were collected on all active fractions using a Bruker (400 MHz) and an Agilent 1260 HPLC system coupled to a 6120 quadrupole LC-MS spectrometer, respectively. Active fractions underwent ¹H NMR and high-resolution MS analysis to assess purity and, if fractions were still mixtures of compounds, they underwent additional fractionation. This process was repeated until pure fractions were obtained. Specific purification details of isolated compounds are provided in the SI Appendix, Part C.

In Silico Exploration for the Presence of Secondary Metabolite Gene Clusters in the Genome of C. lytica.

AntiSMASH and NaPDos are bioinformatic platforms that allow genome mining for secondary metabolite biosynthesis gene clusters in bacterial genomes (57–59) and allow rapid genome-wide identification of secondary metabolite genes, C- or KS-domains, or amino acids. In silico analyses of the genome of C. lytica (CP009239.1) using both AntiSMASH 5.0 (57–59) and NaPDos (60) were run with default, relaxed parameters to determine the potential for C. lytica to produce secondary metabolites (SI Appendix, Part D and Tables S1–S3).

Extraction of Cell-Envelope Components of C. lytica.

Because OMVs are composed of pieces of the bacterial outer membrane and because of the need for larger quantities of compounds for treatments and assays, the components of the cell envelope (whole membrane, peptidoglycan, and LPS) were assayed directly by extraction from whole cells.

Extraction of Cell-Envelope Peptidoglycan of C. lytica.

An extract containing peptidoglycans was obtained from cultures of C. lytica with a protocol adapted from Leulier et al. (82). Culture growth was halted by chilling cultures on ice to 4 °C. Cells were collected by centrifugation, washed once with a cold 3.5% NaCl solution, and centrifuged again. Cells were then suspended in hot (95 °C) aqueous SDS solution (4%) for 30 min and then cooled to room temperature (1 h). The SDS was removed by ultracentrifugation (20,000 × g, 40 min) and the pellet was resuspended in water and washed again for a total of five washes. The pellet was then resuspended in water and incubated with both Dnase I, Rnase, Proteinase K, Pronase E, and Trypsin (Table 1) to remove any contaminating proteins or nucleic acids. Enzymes were removed following the same ultracentrifuge protocol described above. Metamorphic activity in the peptidoglycan extract was further interrogated with either a lysozyme or lipase treatment. Induction of metamorphosis in larvae of H. elegans was conducted before and after lipase and lysozyme treatments. Peptidoglycan monomers in the form of muramyl dipeptides and tripeptides (Sigma Aldrich) were also assayed for induction of metamorphosis in larvae of H. elegans. Details can be found in SI Appendix, Part F.

Extraction of Cell-Envelope LPS of C. lytica.

LPS was extracted using the Westphal hot phenol method (96) as updated by Apicella et al. (97, 98). The LPS fraction was assessed for metamorphic induction in larvae of H. elegans across a range of concentrations determined from the inoculation density of metamorphically inductive biofilms of C. lytica (10⁵ cells·mm⁻²) and estimates of the amount of LPS present per cell range from 25 to 62 fg·cell⁻¹ (99–101). This calculation estimates that active biofilms of C. lytica contain between 2 µg and 6 µg LPS per well in a metamorphosis assay. Consequently, a range from 0.5 µg·mL⁻¹ to 10 µg·mL⁻¹ LPS was utilized to develop a dose–response series for assessment of induction of metamorphosis in larvae of H. elegans. Commercial samples of LPS (Sigma Aldrich) were also assayed for induction of metamorphosis in larvae of H. elegans. Details can be found in the SI Appendix, Part F.

To reduce issues with solubility of the LPS extract and to determine a weight-for-volume concentration, an aliquot of the concentrated LPS sample was dried completely using a speed-vac. This mass was then used to calculate the concentration of the remaining solution. The LPS sample was also subjected to enzymatic interrogation with lipase, lysozyme, and pronase E, as described above.

Analysis of LPS Extract Purity.

Aqueous LPS solutions were checked for contaminating proteins by measuring the absorption at 260 nm. Electrophoretic behavior and additional protein contamination checks were performed by SDS/PAGE (Coomassie blue and silver staining) (82, 102). The presence of glycolipids was identified with additional SDS/PAGE gels stained with Invitrogen Pro-Q Emerald 300 Lipopolysaccharide Gel Stain Kit according to the manufacturer’s instructions. The presence of contaminating lipid classes in the extracted LPS was checked by utilizing the rapid fluorometric microdetection system of Watanabe and Mizuta (103). Briefly, the LPS extract was run on an alumina backed silica TLC plate (Aadvance Instruments, 60 Å F254) alongside LPS and glucose standards, as well as the phospholipid standard, phosphatidyl ethanolamine (Sigma Aldrich). Commercial LPS standards were checked for contaminants with thin layer chromatography and UV spectroscopy. TLCs were run using the solvent system isobutyric acid:concentrated ammonium hydroxide:water in the ratio (vol/vol/vol) of 57:4:39 (104). The TLC was developed by heating with acidified 5-hydroxy-1-tetralone (103). Under UV light (365 nm) hexoses, including oligosaccharides and polysaccharides containing hexoses appear as green spots; acidic and neutral glycosphingolipids appear as a yellow spot and phospholipids appear as blue spots.

Examination of Bacterial Preparations with TEM.

Bacterial preparations and cell-free preparations were applied to glow-discharged formvar grids, negatively stained with 2% uranyl acetate aqueous solution, and air dried. Samples were imaged using a 120 kV Hitachi HT7700 electron microscope with an AMT XR-41 camera.

Data Availability

Raw data have been deposited in Figshare, https://doi.org/10.6084/m9.figshare.10324772 (115).