A modular plasmid toolkit applied in marine bacteria reveals functional insights during bacteria-stimulated metamorphosis

ABSTRACT A conspicuous roadblock to studying marine bacteria for fundamental research and biotechnology is a lack of modular synthetic biology tools for their genetic manipulation. Here, we applied, and generated new parts for, a modular plasmid toolkit to study marine bacteria in the context of symbioses and host-microbe interactions. To demonstrate the utility of this plasmid system, we genetically manipulated the marine bacterium Pseudoalteromonas luteoviolacea, which stimulates the metamorphosis of the model tubeworm, Hydroides elegans. Using these tools, we quantified constitutive and native promoter expression, developed reporter strains that enable the imaging of host-bacteria interactions, and used CRISPR interference (CRISPRi) to knock down a secondary metabolite and a host-associated gene. We demonstrate the broader utility of this modular system for testing the genetic tractability of marine bacteria that are known to be associated with diverse host-microbe symbioses. These efforts resulted in the successful conjugation of 12 marine strains from the Alphaproteobacteria and Gammaproteobacteria classes. Altogether, the present study demonstrates how synthetic biology strategies enable the investigation of marine microbes and marine host-microbe symbioses with potential implications for environmental restoration and biotechnology. IMPORTANCE Marine Proteobacteria are attractive targets for genetic engineering due to their ability to produce a diversity of bioactive metabolites and their involvement in host-microbe symbioses. Modular cloning toolkits have become a standard for engineering model microbes, such as Escherichia coli, because they enable innumerable mix-and-match DNA assembly and engineering options. However, such modular tools have not yet been applied to most marine bacterial species. In this work, we adapt a modular plasmid toolkit for use in a set of 12 marine bacteria from the Gammaproteobacteria and Alphaproteobacteria classes. We demonstrate the utility of this genetic toolkit by engineering a marine Pseudoalteromonas bacterium to study their association with its host animal Hydroides elegans. This work provides a proof of concept that modular genetic tools can be applied to diverse marine bacteria to address basic science questions and for biotechnology innovations.

M arine bacteria are a valuable and currently under-utilized resource for environ mental restoration (1)(2)(3)(4)(5)(6) and bioprospecting (7,8), especially considering their influence on biogeochemical cycles (9) and their vital role in evolution through symbioses with eukaryotes (10). While advances in metagenomic sequencing have enabled a deeper exploration of microbial diversity and gene content (11,12), genetic tools to explore functions in marine bacteria remain scarce.
Effective genetic engineering approaches in model microbial species, such as Escherichia coli, utilize standardized and modular cloning toolkits (13)(14)(15)(16)(17)(18)(19), which leverage aligned plasmid parts based on the ordered pairings of restriction site overhangs to enable innumerable mix-and-match plasmid assembly options. However, such modular genetic tools have not yet been applied to most marine bacterial species. Thus, adapting and applying standardized molecular cloning tools for studying marine bacteria can provide a framework for addressing functional questions for basic science and biotech nology.
Marine bacteria are of specific interest as targets for genetic tool development due to their ability to produce diverse bioactive metabolites (20), their prominent associa tions in aquatic microbiomes, and their involvement in host-microbe symbioses (21)(22)(23). Alphaproteobacteria and Gammaproteobacteria, in particular, are the most abundant orders in the ocean (12) and are prominent members of the microbiomes of animals such as phytoplankton (12), tubeworms (21), and corals (24).
Of particular interest as targets for genetic manipulation are marine Pseudoalteromo nas species because they produce a number of bioactive secondary metabolites (8,(25)(26)(27)(28)(29) and are often found in association with marine invertebrates (30-36). Pseudoal teromonas species are known to engage in a transient symbiosis called bacteria-stimu lated metamorphosis, whereby surface-bound bacteria promote the larval-to-juvenile life cycle transition in invertebrates such as tubeworms and corals (37,38). Pseudoaltero monas luteoviolacea stimulates the metamorphosis of the tubeworm Hydroides elegans (39,40) by producing syringe-like protein complexes called Metamorphosis-Associated Contractile structures (MACs). MACs stimulate tubeworm metamorphosis by injecting an effector protein termed Mif1 into tubeworm larvae (40)(41)(42). Genes encoding the MACs structure are found in the P. luteoviolacea genome as a gene cluster encoding structural components, such as the macB baseplate and macS sheath, as well as the protein effector gene mif1 (41). Despite the significant insights gained by using genetics in P. luteoviolacea, new genetic tools are needed to further dissect the function of MACs and their stimulation of tubeworm metamorphosis.
In this work, we utilize a modular plasmid toolkit, and contribute new Marine Modification Kit (MMK) plasmid parts, to study bacteria-stimulated metamorphosis in the Gammaproteobacterium, P. luteoviolacea. We demonstrate the broader utility of this approach by conjugating MMK plasmids into marine Alphaproteobacteria and Gammap roteobacteria that have been shown previously to be involved in diverse host-microbe interactions.

Toolkit-enabled quantitative promoter expression in P. luteoviolacea
To test the application of modular genetic tools in marine bacteria, we identified a set of preexisting parts from the Yeast Toolkit and Bee Toolkit platforms (17,18) and used Golden Gate Assembly (14) for rapid, modular construction of plasmids ( Fig. 1A through C). Each type of part is defined by its functional role (e.g., promoter and coding sequence [CDS]) and directional 4 bp overhangs generated by flanking Type IIS (BsaI) restriction sites. The modular parts include Type-1 and Type-5 stage-2 connectors with BsmBI recognition sites (17,18), a Type-2 promoter with ribosome binding site (RBS), a Type-3 protein CDS (e.g., gfp and Nanoluciferase), a Type-4 terminator, an optional Type-6 repressor and Type-7 promoter with RBS, and a Type-8 backbone. Preexisting Type-8 backbones are available with different origins of replication (ColE1 and RSF1010) and antibiotic resistance markers (ampicillin, kanamycin, or spectinomycin resistance) (17,18). For this work, we selected a broad-host-range (BHR) plasmid backbone contain ing a kanamycin resistance gene, a reporter CDS (fluorescent gfp-optim1, mRuby, or Nanoluciferase [NLuc]), T7 terminator, and a stage-2 assembly connector. The back bone selected has an RSF1010 origin of replication which is known to replicate in a broad range of Gram-positive and Gram-negative bacterial hosts at a copy number of 10-12 per chromosome and also contains a promiscuous origin of transfer and the plasmid-encoded mobilization genes repA, repB, repC, and mobC (43,44). An auxotrophic MFDλpir strain was used as the E. coli donor, thus obviating the need to generate antibiotic-resistant recipient strains to counter select E. coli donor cells after conjugation (45).
To apply the modular genetic tools in a marine symbiosis model, we tested the expression of five promoters in P. luteoviolacea. We assembled plasmids with each promoter fused to NLuc and conjugated the plasmids into P. luteoviolacea. We utilized two existing BHR promoters, PA3 and CP25, previously shown to work in diverse bee gut microbes (17,46,47). We also created a Ptac lacO promoter part (pMMK201), which is a hybrid of the lac and trp promoters amplified from the pANT4 plasmid (48). When P. luteoviolacea with the plasmids were grown in exponential, stationary, or biofilm growth phases, we observed at least 10-fold more luminescence signal compared to the background with all BHR promoters tested (Fig. 1D).
Previous observations have shown that the production of MACs is greatest during the exponential phase of growth when P. luteoviolacea is cultured in rich media (40). However, the expression of mac genes in live cultures has not been previously quantified. To observe the expression of two native mac promoters, we constructed two plasmids with P. luteoviolacea promoters driving the expression of the MACs structural genes; promoters from the MACs sheath (macS promoter, pMMK203) and baseplate (macB promoter, pMMK202) genes. The macSp luciferase reporter strain was elevated 1,000-fold in exponential growth as compared to 100-fold in stationary and 10-fold in biofilm phase, when compared to the detection limit (Fig. 1E). In contrast, the macB, baseplate pro moter exhibited similar levels of luminescence among each phase, approximately 10-fold higher than the detection limit (Fig. 1E).

Functional CRISPRi knockdown of secondary metabolite biosynthesis in P. luteoviolacea
While previous studies in P. luteoviolacea have used gene knockouts to interrogate gene function, these approaches are time-consuming and low-throughput. We therefore tested whether P. luteoviolacea is amenable to gene knockdown via CRISPR interference (CRISPRi) ( Fig. 2A and B) (49,50). As a proof of concept, we targeted the vioA gene that encodes a key enzyme in the biosynthesis of violacein (51), which gives P. luteoviolacea its characteristic purple pigment (Fig. 2B). An assembled plasmid containing dCas9 and a single-guide RNA (sgRNA) targeting vioA (pMMK603) was conjugated into P. luteoviolacea resulting in the visible absence of the purple pigment associated with violacein produc tion on the plate (Fig. 2C). We also created a plasmid containing dCas9 and a sgRNA targeting gfp to test whether the presence of the CRISPRi machinery adversely affected wild-type (WT) P. luteoviolacea or violacein production. No difference was observed between the growth and cell morphology of P. luteoviolacea containing gfp or vioA sgRNA CRISPRi plasmids compared to WT (Fig. S1). WT P. luteoviolacea produced violacein as expected, while P. luteoviolacea with CRISPRi with the gfp sgRNA produced a statistically comparable amount of violacein (adjusted P = 0.26, n = 8, Dunn's multiple comparison test). A significant reduction of violacein production was observed between cultures of P. luteoviolacea strains expressing the vioA and gfp targeting CRISPRi plasmids (adjusted P = 0.02, n = 8, Dunn's multiple comparison test) (Fig. 2D). The lack of violacein in the vioA knockdown strain was comparable to that of a P. luteoviolacea strain with an in-frame deletion of vioA (adjusted P = 0.26, n = 8, Dunn's multiple comparison test) (Fig.  2D). These results demonstrate the successful implementation of CRISPRi for gene knockdown in P. luteoviolacea.

Functional CRISPRi knockdown and visualization of P. luteoviolacea during a tubeworm-microbe interaction
We next tested whether CRISPRi would be functional in the context of a marine hostmicrobe interaction by targeting the macB gene, which encodes the MACs baseplate, an essential component of the MACs complex that induces tubeworm metamorphosis (39, 40) (Fig. 3A). Biofilm metamorphosis assays were performed comparing P. luteoviolacea strains with sgRNAs targeting macB (pMMK604) or the sgRNA targeting gfp control (Fig.  3B). The strain with sgRNA targeting macB exhibited significantly reduced levels of tubeworm metamorphosis compared to the gfp-sgRNA control (adjusted P < 0.0001, Dunn's multiple comparisons test, n = 12) (Fig. 3B). The reduction of metamorphosis Research Article mBio stimulation in the macB-sgRNA knockdown strain was comparable to that of a P. luteoviolacea strain with an in-frame deletion of macB carrying the gfp-sgRNA control plasmid (adjusted P ≥ 0.99, Dunn's multiple comparison test, n = 12) (Fig. 3B). These results demonstrate that CRISPRi paired with a modular plasmid system is a viable tool for interrogating gene function during a marine host-microbe interaction.
To date, bacteria have not been visualized during or after the stimulation of metamor phosis in Hydroides. To test whether marine bacteria harboring a toolkit plasmid are amenable to live-cell imaging when in association with juvenile tubeworms, we created biofilms of P. luteoviolacea containing plasmids encoding CP25-gfp-T7 (gfp) or CP25-Nanoluc-T7 (NLuc) and added competent Hydroides larvae. After incubation for 24 h, biofilms of gfp-expressing P. luteoviolacea were clearly observed when visualized by fluorescence microscopy (Fig. 3C). P. luteoviolacea stimulated Hydroides metamorphosis while carrying a modular plasmid and fluorescent bacteria were observed being ingested by the Hydroides juveniles. Bacteria can be seen collecting in the pharynx (Fig.  3C, yellow arrows), then moving in a peristaltic fashion toward the gut (Movie S1). In contrast, bacteria containing a CP25-NLuc-T7 plasmid were difficult to visualize by light microscopy, in the absence of the gfp fluorescent marker (Fig. 3D). Taken together, the Research Article mBio modular plasmid system enables live imaging and experimentation during a marine host-microbe interaction.

Applying the modular toolkit in marine Alphaproteobacteria and Gammap roteobacteria
Given the success of genetic manipulation of P. luteoviolacea, we tested whether other marine Proteobacteria might be amenable to conjugation and retention of a modular genetic toolkit plasmid. To this end, we isolated or acquired representative bacteria that are known to engage in symbioses with marine plants or animals in the ocean ( Fig.  4A; Tables S1 and S2). To enable genetic selection using antibiotics, we determined the minimum inhibitory concentration for each bacterial strain tested against kanamy cin (Table S1). When conjugation was performed using the BHR (RSF1010) plasmid backbone, CP25 promoter, gfp reporter, and T7 terminator, we observed the expression of gfp in 12 marine strains across two proteobacterial classes, four orders, and 10 genera (  Research Article mBio

Modular genetic tools provide insights about bacteria-stimulated metamor phosis
We tested a modular plasmid toolkit on a genetically tractable marine bacterium, P. luteoviolacea, that promotes the metamorphosis of the tubeworm Hydroides elegans (40,41,55) and produces several bioactive secondary metabolites (26,29,56,57). We expand the tools available for functional interrogation of bacteria-stimulated metamorphosis in P. luteoviolacea by quantifying gene expression by a luminescence assay (Fig. 1D and E), and using CRISPRi to knock down the secondary metabolite, violacein ( Fig. 2C and D), as well as a metamorphosis-associated gene, macB (Fig. 3B) N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N a a a a a a a N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N  i i i  d d d d d d d  a a a a a a  Research Article mBio gene expression appeared static across the growth conditions tested. Although MACs are known to produce two effectors that stimulate tubeworm metamorphosis and kill eukaryotic cells (41,58), the environmental conditions that promote MACs production remain poorly characterized. The tools developed here could help to characterize the conditions under which P. luteoviolacea MACs are produced or assembled and could help in the development of MACs or other contractile injection systems for use in biotechnol ogy (59,60). The modular tools in this work open new capabilities for interrogating bacteriology, including the ability to quantify gene expression in live cultures, knock down gene expression for rapid functional testing, and visualize bacteria during an in vivo interaction. Whether, and how, bacteria and the animal are harmed or benefit from the interaction during bacteria-stimulated metamorphosis remains a prominent question in the field (38,61,62). Swimming Hydroides larvae initially encounter and are stimulated to undergo metamorphosis by the bacterial biofilm. And MACs were previously visualized within P. luteoviolacea biofilms by tagging the MACs baseplate with super-folder GFP (40). However, less attention has been put on the interaction between Hydroides and the bacteria after metamorphosis. Previous work by Gosselin et al. has shown that Hydroides is able to feed on bacteria as the sole food source (63). In the present work, we visualize live bacteria surrounding and being ingested by Hydroides juveniles (Fig. 3C) (21). The visualization of transgenic bacteria in Hydroides will enable future lines of research that can help dissect the role of microbiome seeding in bacteria-stimulated metamorphosis. More broadly, our results showcase the feasibility of using a modular plasmid toolkit to test hypotheses about bacteria-stimulated metamorphosis and provide a framework for the interrogation of other bacteria and their products that promote host-microbe symbioses (36,64,65).

Toolkit compatibility in marine bacteria
In this work, we explore genetic tractability in 12 ecologically relevant marine bacteria that belong to two Proteobacterial classes (Fig. 4). The Gammaproteobacteria strains conjugated successfully in this study are a selection of symbiosis-associated strains representing five genera (Fig. 4A) (66-72). To our knowledge, this is the first report of genetic tractability in strains from the genera Endozoicomonas, Nereida, and Cobetia (Fig.  4B). Endozoicomonas species are among the most abundant bacterial symbionts in some corals and other marine hosts (73)(74)(75). Related strains of Cobetia have been implicated in thermotolerance against bleaching in coral experiments with probiotic consortium treatments (76). The conjugation of the representative Endozoicomonas and Cobetia strains in this study is a considerable step toward exploring function in coral host-micro biome interactions at a critical time to encourage the restoration of coral reefs (6,77,78). The genetic conjugation of Pseudoalteromonas sp. PS5 in this study presents an opportunity to explore secondary metabolite production, including the coral metamor phosis-inducing compound, tetrabromopyrrole (Fig. 4) (36, 79). The Alphaproteobacteria strains tested for compatibility with MMK plasmids fall within the Roseobacter group (Fig. 4A), an ecologically important group of bacteria known to play a role in sulfur and carbon cycling on marine phytoplankton (80)(81)(82). Roseobacter strains have also been explored as probiotics for the aquaculture industry (83)(84)(85). We tested the toolkit with the tractable, phytoplankton-associated species of Phaeobacter gallaeciensis (86), and Ruegeria pomeroyi (87), and demonstrated conjugation with invertebrate microbiomeassociated strains Phaeobacter sp. HS012 (88) and Leisingera sp. 204H (89) (Fig. 4). Using modified Shimia sp. may be of interest for future coral microbiome studies (90)(91)(92)(93). Species in the Nereida genus have been isolated from kelp (94) and are associated with gall formations (95,96). Tractability in this strain could help guide further understanding of microbe-seaweed interactions (97,98), kelp aquaculture, and the development of kelp probiotics (99). In summary, the development of methods and established tractability of several new strains and genera have significant implications for the future of bacterial genetic development in established and emerging symbiosis systems.

Future modifications
The modularity of the plasmid toolkit enables the potential for creating new plasmids that are compatible with the existing system to boost functionality. For example, the addition of backbone plasmid parts (Type-8) with different origins of replication and selectable markers could allow utilization in bacteria that are naturally resistant to the antibiotics used in this and prior works (17,18). We have created a Type-8 Tn10 transposon backbone for stable integration of toolkit parts into the genomes of marine bacteria and used this part to integrate a fluorescent gfp marker into the genome of Pseudoalteromonas sp. PS5 (79). Type-8 parts like this could be used to tag and track marine bacteria for studying host-microbe interactions in the future.
The current promoter driving dCas9 is constitutive. However, adding an inducible promoter driving dCas9 (e.g., P BAD for arabinose induction) would allow the CRISPRi system to be controllable. The expression of gfp was not uniformly observed in the Phaeobacter, Leisingera, and Nereida strains (Fig. 4B). However, the plasmid toolkit could be used to identify plasmid components that would produce uniform expression (e.g., different origins of replication, selectable markers, promoters, etc.). In the future, more strains may be tested for manipulation with the present toolkit plasmids for applying genetics in a broader array of bacteria types.

Conclusion
The modular plasmid toolkit described here provides a basis for streamlining the genetic manipulation of marine bacteria for basic and applied purposes. These tools reveal new possibilities to study marine microbes in the context of plant and animal interactions, or with challenging and diverse non-model bacteria, ultimately helping us harness marine microbes for research, bioproduction, and biotechnology.

Bacterial culture
A list of strains used in this study, isolation sources, accession numbers, and minimum inhibitory concentration can be found in Table S1. Environmental strains of marine bacteria were isolated and cultured on Marine Broth (MB) 2216 (BD Difco) and or natural seawater tryptone (NSWT) media (1 L 0.2 µm filtered natural seawater from Scripps Pier, La Jolla, CA, 2.5 g tryptone, 1.5 g yeast extract, 1.5 mL glycerol). MB and NSWT media are used interchangeably throughout the study; however, the experiments were always conducted using only one media type. Marine bacteria were incubated between 25°C and 30°C, and cultures were shaken at 200 rpm. All liquid cultures were inoculated with a single colony and incubated between 16 and 18 h, unless otherwise indicated. E. coli SM10λpir and S17-1λpir were cultured in LB (Miller, BD Difco) at 37°C, shaking at 200 rpm. E. coli MFDλpir (45) was cultured in LB supplemented with 0.3 mM Diaminopimelic acid (DAP). For E. coli, antibiotic selections with ampicillin, kanamycin, and chloramphenicol were performed using a concentration of 12.5 µg mL −1 .

Plasmid construction and assembly
Golden Gate Assembly-compatible parts from the BTK, YTK (17,18), and MMK used in this work can be found in Table S3. New plasmid parts were made by PCR amplifying insert and backbone fragments and combining them with Gibson Assembly with a 2:1 ratio (20 fmol insert: 10 fmol backbone) (100). PCR amplification was performed with custom primers (Table S4), a high-fidelity DNA polymerase (PrimeSTAR GXL, Takara), and purified using a DNA Clean and Concentrator kit (Zymo Research). Part plasmids were assembled to make a stage 1 plasmid using Golden Gate Assembly, with T4 DNA ligase (Promega) and either BsaI or BsmBI (New England Biolabs), depending on the construct. Single-tube assembly was performed by running the following thermocycler program (BsaI/BsmBI): 37/42°C for 5 min, 16°C for 5 min, repeat 30×, 37/55°C for 10 min, and 80°C for 10 min. The assemblies were directly electroporated into S17-1λpir cells, confirmed by colony PCR (EconoTaq PLUS Green, LGC Biosearch) with internal primers, and then electroporated into MFDλpir cells for conjugation. To facilitate assembly for and expression of CRISPRi parts in P. luteoviolacea, we moved the BsmBI cut site in the dCas9 part plasmid (pBTK614) to a location where the existing bla gene will be retained in the assembled plasmid (pMMK601), and thus also conferring resistance to ampicillin. In the sgRNA plasmid (pBTK615), we replaced the existing PA1 promoter with the Ptac promoter (including −35 and −10 sequences but excluding lacO), which drives the sgRNA expression (pMMK602). The CRISPRi assemblies were electroporated directly into SM10λpir cells and shuttled to MFDλpir cells for conjugation.

Biparental conjugation in marine bacteria
E. coli donor strains (MFDλpir or SM10λpir) containing the mobilizable plasmids were grown under antibiotic selection in LB with the appropriate supplements (including 0.3 mM DAP for E. coli MFDλpir). Conjugations were performed as previously described (17) with modifications for culturing marine bacteria. Briefly, several colonies of the recipient strains were inoculated and grown overnight in liquid culture. Recipient and donor cultures were spun down (4,000 × g for 2 min) in a 1:1 OD 600 ratio. All donor supernatant was removed leaving only the cell pellet. All but 100 µL of the recipient supernatant is removed, and the cell pellet is resuspended. The recipient suspension was transferred to the donor pellet, which was resuspended with the recipient cells. Two 50 µL spots are plated onto NSWT (supplemented with 0.3 mM DAP for MFDλpirmediated conjugations) and incubated overnight at 25°C with the lids facing up. The next day, spots were scraped up with a pipette tip and resuspended in 500 µL of liquid marine media and 100 µL was plated onto marine media containing antibiotic selection, according to the minimum inhibitory concentration (Table S1). Streptomycin-resistant P. luteoviolacea (Table S1) were conjugated with E. coli SM10λpir, and counterselection was performed with 100-200 µg/mL streptomycin. All other marine bacteria (Table S1) were conjugated with E. coli MFDλpir, and transconjugant selection was performed in the absence of DAP. Several of the bacteria take longer to grow or do not reach a high optical density (i.e., Endozoicomonas, Ruegeria, and Nereida) in culture. Slower-growing marine bacteria were conjugated by growing larger 50 mL initial volumes of culture and spinning down the entire culture to reach 1:1 (donor:host) ratios.

Phylogeny
Strains or close representative strains used in this study were compiled into a genome group on PATRIC v3.6.12 (101). A whole genome phylogenetic codon tree composed of 100 single-copy genes (102) was performed using the Phylogenetic Tree Service (103)(104)(105). A maximum likelihood phylogeny was generated using the best protein model found by RaxMLv8.2.11 (106), which was LG. Bootstraps were generated using the rapid bootstrapping algorithm with the default of 100 resamples (54). The tree was visualized with FigTree v1.4.4. and was rooted at the mid-line.

Growth curve
Pseudoalteromonas luteoviolacea ∆vioA and WT were grown on MB agar plates and incubated overnight at 25°C. P. luteoviolacea strains expressing CRISPRi plasmids were grown on MB agar plates with 200 µg mL −1 of kanamycin and grown overnight at 25˚C. Single colonies were picked and inoculated into 5 mL of MB liquid media with the respective antibiotics listed above. Two biological replicate cultures were inoculated for each strain by picking different colonies from the agar plate and inoculating separate 5  Readings were done on the kinetic luminescence mode at 2 min intervals for 20 min in total, using a 400-ms integration time, a 1-mm height read, and no other optimization or shaking settings. The detection limit is defined as the average expression of P. luteoviolacea cells expressing a non-luminescent plasmid across growth conditions. Raw data were normalized to the OD 600 of the culture used and plotted with an N = 3 biological replicates.

Violacein extraction
The specified P. luteoviolacea strains were struck onto NSWT media containing 200 µg mL −1 of streptomycin and kanamycin and incubated overnight at 25°C. Single colonies were inoculated into 5 mL of liquid media containing the same antibiotic concentrations.
Cultures were incubated at 25°C, shaking at 200 rpm between 18 and 20 h. Cultures were removed from the incubator and standardized to an OD 600 of 1.5. The cells were pelleted, and the supernatant was removed. The cell pellet was resuspended in 200 µL of 100% ethanol. The resuspended cells were pelleted and the supernatant containing the crude extract was recorded on a BioTek Synergy HT plate reader (Vermont, USA) using the Gen5 program (v2.00.18) with an endpoint reading at 580 nm.

Microscopy
Microscopy was performed using a Zeiss Axio Observer.Z1 inverted microscope equipped with an Axiocam 506 mono camera and Neofluar10x/0.3 Ph1/DICI (Hydroides co-cultures) or Apochromat 100×/1.4 Oil DICIII (bacteria only) objectives. The Zeiss HE eGFP filter set 38 was used to capture GFPoptim-1 expression and Zeiss HE mRFP filter set 63 was used to capture mRuby2 expression. For Nanoluciferase controls, images were captured using the same fluorescence exposure times as the gfp optim-1 and mRuby2 labeled strains of the same species. Bacterial culture (2 µL) was added to freshly prepared 1% saltwater low-melt agarose (Apex catalog #20-103, Bioresearch products) pads on glass slides and coverslips were placed on top. Hydroides elegans were prepared in visualization chambers (Lab-Tek Chambered Coverglasses catalog #155411PK) with bacteria and imaged.

Hydroides elegans culture
Hydroides elegans adults were collected from Quivira Basin, San Diego, CA, USA. The larvae were cultured and reared as previously described (40,107). Larvae were main tained in beakers containing filtered artificial seawater (35 PSU) and were given new beakers with water changes daily. The larvae were fed live Isochrysis galbana and cultures were maintained as described previously. The larvae were used for metamorphosis assays once they reached competency (between 5 and 7 d old) (108).

Hydroides elegans metamorphosis assays
Biofilm metamorphosis assays were performed using previously described methods (39,40,109). Briefly, bacteria were struck onto MB plates with 300 µg mL −1 kanamycin as appropriate and were incubated overnight at 25°C. Up to three single colonies were inoculated into liquid broth and incubated overnight (between 15 and 18 h), shaking at 200 rpm. Cultures were pelleted at 4,000 × g for 2 min, the spent media were removed, and the cell pellets were washed twice with filtered artificial sea water (ASW). The concentration of the cells was diluted to OD 600 of 0.1, and four 100 µL aliquots of the cell concentrate were added to 96-well plates. The cells were given between 2 and 3 h to form biofilms, then the planktonic cells were removed and the adhered cells were washed twice with filtered ASW. Between 20 and 40 larvae were added to each well in 100 µL of filtered ASW. Metamorphosis was scored after 24 h. Three biological replicates were performed on different days using separate Hydroides larvae originating from different male and female animals.
Chambered metamorphosis assays were performed using the same preparation principles as described above with the following modifications. Visualization chambers (Lab-Tek, catalog # 155411) were used for setting up the metamorphosis assay, then subsequently imaged. Inductive strains containing constitutively expressed gfp/mRuby/ NLuc plasmids were struck out onto MB media containing 300 µg mL −1 kanamycin. Several colonies were inoculated into 5 mL MB media with antibiotics. Cultures were grown for 18 h and cells were washed and allowed to form biofilms as described above. Cell concentrations ranging between OD 600 0.1 and 0.5 were used to elicit optimal metamorphosis. Larvae were concentrated and the resident filtered ASW was treated with 300 µg mL −1 kanamycin. Larvae were imaged 24 h later.

ACKNOWLEDGMENTS
We thank current and former Shikuma Lab members who helped with cloning and providing feedback for this paper, including Taylor