Hawaiian bobtail squid, E. scolopes, were obtained from sand flats in Oahu (Maunalua Bay, 21°16′51.42″ N, 157°43′33.07″ W), Hawaii and maintained in aquaria as previously described (Schleicher and Nyholm, 2011). Eggs laid in captivity from one adult female were collected, flash frozen on the 11th day of development, and stored at -80°C. Ten eggs were thawed for bacterial isolation and their outer capsules and embryos were removed and discarded with sterile forceps. The JCs were isolated, surface sterilized with 70% ethanol, and rinsed with filter-sterilized squid Ringers (FSSR, 530 mM NaCl, 25 mM MgCl₂, 10 mM CaCl₂, 20 mM HEPES, pH = 7.5). The 10 JCs were pooled and homogenized in FSSR, then serially diluted and plated on seawater tryptone (SWT) medium (5 g/L tryptone, 3 g/L yeast extract, 3 mL/L glycerol, 700 mL/L Instant Ocean sea salts, 15 g/L agar, 300 mL/L DI water). Leisingera sp. JC1 colonies appeared dark blue on this medium and were streaked to isolation.

Genomic DNA was extracted using the MasterPure DNA Purification kit (Epicentre, Madison, WI, USA) from an overnight liquid culture of Leisingera sp. JC1 grown shaking at 30°C in SWT. DNA was quantified using a Qubit 2.0 fluorometer (Life Technologies, Agawam, MA, USA) and checked for quality on a 1% agarose gel and using a NanoDrop 1000 spectrophotometer (Thermo Scientific, Agawam, MA, USA). A paired end library was prepared from 1 ng of genomic DNA using the Nextera XT DNA library kit (Illumina, Inc., San Diego, CA, USA) and quantified using the Qubit fluorometer and bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The library was sequenced on an Illumina MiSeq sequencer using 2 bp × 250 bp reads at the Microbial Analysis Resources and Services (MARS) facility at the University of Connecticut (Storrs, CT, USA).

Reads were trimmed using the CLC Genomic Workbench (Qiagen, Hilden, Germany) and a draft genome was assembled using the A5 assembler (Tritt et al., 2012). Coverage was determined by mapping trimmed reads to the draft genome assembly using CLC Genomic Workbench. The genome was annotated using the Rapid Annotation using Subsystem Technology (RAST, Aziz et al., 2008)¹ server and analyzed with the Antibiotic and Secondary Metabolite Analysis Shell 3.0 (antiSMASH, Weber et al., 2015)² for potential secondary metabolite biosynthesis gene clusters. The draft genome assembly has been deposited in DDBJ/EMBL/GenBank under accession LYUZ00000000. The version described in this paper is version LYUZ01000000.

Initial 16S identity suggested JC1 belonged to the genus Leisingera (data not shown). To validate this conclusion and to evaluate its relationship to the previously sequenced ANG isolates, a further taxonomic analysis was undertaken that used 17 previously described Leisingera genomes (Collins et al., 2015). A 33 gene multilocus sequence analysis was carried out following the methodology described in Collins et al. (2015). After generating alignments for each of the 33 genes using MUSCLE (Edgar, 2004), a concatenated alignment was generated using in-house python scripts. An optimal model of evolution was determined using the Akaike information criterion with correction for small sample size as implemented in jModelTest v2.1.4 (Darriba et al., 2012). The best-fitting model reported was GTR + Gamma estimation + Invariable site estimation. A maximum-likelihood (ML) phylogeny was generated from the concatenated multi-sequence alignment using PhyML v3.0_360-500M (Guindon et al., 2010). PhyML parameters consisted of GTR model, estimated p-invar, four substitution rate categories, estimated gamma distribution, sub-tree pruning and regrafting enabled with 100 bootstrap replicates. In addition to the maximum-likelihood analysis, a Bayesian inference analysis was also conducted using MrBayes v3.2.4 x64 (Ronquist et al., 2012). A mixed model with gamma estimation and invariable sites was used. The mixed model settled on a GTR submodel with only one parameter difference from the default GTR model with a posterior probability > 0.8. The standard GTR model accounted for the remainder of the model probability. The analysis used two cold chains with three heated chains each and ran for one million generations. After the run finished, convergence was assessed using average standard deviation of split frequencies of the cold chains, potential scale reduction factors of parameters, and minimum effective sample sizes of parameters. All criteria indicated the runs had converged.

Average nucleotide identity (ANI) was calculated using JSpecies 1.2.1 (Richter and Rosselló-Móra, 2009). The calculations were made using the MUMmer aligner with its default options. Contig files were generated for this analysis using the seqret function of the EMBOSS package (Rice et al., 2000). The reciprocal comparisons were averaged for reporting. Estimates of in silico DDH were made using the Genome-to-genome distance calculator 2.1 (Meier-Kolthoff et al., 2013) using the BLAST+ alignment method and the formula 2 algorithm outputs.

Select genomes were compared using the BLAST Ring Generator (BRIG) v1.0 (Alikhan et al., 2011). Default BLAST options were used. A whole genome alignment was generated using the Mauve program v2.3.1 (Darling et al., 2010). The progressiveMauve algorithm was used with default options.

Homoserine lactone (HSL) production was detected using a well-diffusion assay with the HSL-sensing bacterium Agrobacterium tumefaciens NTL4 pZLR4 (Cha et al., 1998) as previously described (Ravn et al., 2001; Collins et al., 2015). In brief, A. tumefaciens NTL4 was grown in 3 mL of LB with 30 μg/mL gentamicin for 24 h at 30°C. This culture was used to inoculate 50 mL of AB minimal media with 0.5% casamino acids and 0.5% glucose (Chilton et al., 1974), and allowed to grow for another 24 h at 30°C. This culture was used to inoculate 100 mL of AB minimal media to which 1.2% agar had been added and a final concentration of 0.5% casamino acids, 0.5% glucose, and 75 μg/mL 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) was added after autoclaving. The inoculated molten agar was allowed to solidify in Petri dishes and wells were cut into the media using a sterile borer.

Leisingera sp. JC1 and Leisingera sp. ANG1 were grown overnight at 30°C in 3 mL of SWT broth with 30 μM FeCl₃, 0.5% glucose, and 0.5% casamino acids to induce HSL production. Cells were pelleted and the supernatant was collected and filtered through a 0.22 μm filter (Thermo Scientific, Agawam, MA, USA). Cell-free supernatant (60 μl) was added to the wells in the A. tumefaciens plates. The N-3-oxohexanoyl homoserine lactone standard was serially diluted and added to wells of an A. tumefaciens plate as a control and for semi-quantitative comparison. All plates were incubated at 28°C for 24 h before imaging.

To qualitatively detect siderophore production, Leisingera sp. JC1 was plated in triplicate on chrome azurol S (CAS) indicator agar, modified for marine bacteria as previously described (Whistler and Ruby, 2003), and incubated at 28°C for 24 h before imaging. Sequestration of iron from CAS causes a color change from blue to orange, indicating siderophore production.

To confirm the presence of indigoidine biosynthesis genes in JC1, genomic DNA was extracted and quantified as described for genomic sequencing above. Primers were designed (Supplemental Table S1) to amplify the igiCDR genes based on the draft genome assembly and using Primer3 software (Untergasser et al., 2012). PCR amplification was performed using the standard GoTaq Green Master Mix (Promega, Madison, WI, USA) protocol with 30 cycles and 55°C annealing temperature.

Leisingera sp. JC1 was cultured for extraction using SWT media (as described above except without addition of glycerol, delineated hereafter as SWT_ng). A three step culturing process was employed to produce sufficient scale for secondary metabolite extraction, while ensuring that the bacterium was in late stationary phase for optimal production of secondary metabolites (Ruiz et al., 2010). First, small scale cultures were prepared by inoculating a JC1 colony into 5 mL of media in a 24 deep well plate, which was incubated for 3 days at room temperature while shaking at 200 rpm. Then, medium scale cultures were prepared by transferring 1.5 mL of the small scale cultures into 125 mL baffled flasks with 50 mL media, which were incubated for 3 days at room temperature while shaking at 125 rpm. Lastly, large scale cultures were prepared by transferring 15 mL of medium scale cultures into 1 L baffled flasks with 500 mL of media, which were incubated for 3 days at room temperature while shaking at 125 rpm.

All extraction solvents were ACS grade and purchased from Sigma Aldrich (St. Louis, MO, USA).

All HPLC grade solvents and reagents were purchased from Sigma-Aldrich. LC–MS data were collected on an Agilent ESI single quadrupole mass spectrometer coupled to an Agilent 1260 HPLC system with a G1311 quaternary pump, G1322 degasser, and a G1315 diode array detector (Agilent Technologies, Santa Clara, CA). A gradient elution was used from 10% methanol in H₂O to 90% methanol in H₂O over 25 min using an Agilent Eclipse XDB-C₁₈ RP-HPLC column (4.6 mm × 150 mm, 5 μm) and a flow rate of 1 mL/min. Indigoidine enriched extracts were prepared at 5 mg/mL in DMSO. Indigoidine eluted at retention time (t_R) 10.7 min in agreement with literature (Yu et al., 2013).

To observe inhibition of vibrio strains and ANG isolate strains by Leisingera sp. JC1 (Supplementary Table S3), a zone of inhibition (ZOI) assay was used. The vibrio strains V. anguillarum 775, V. parahaemolyticus KNH1, V. fischeri ES114, V. harveyi B392, and Photobacterium leiognathi KNH6 were grown for 2.5 h (to stationary phase) at 30°C in YTSS (4 g/L tryptone, 2.5 g/L yeast extract, 15 g/L Instant Ocean sea salts) broth and then serially diluted from 10⁷ to 10⁴ CFU/mL in YTSS broth to observe density dependent inhibition. Each dilution was plated in triplicate on YTSS agar using a sterile swab to form a lawn. All ANG isolates tested were grown overnight (~4 × 10⁸ CFU/mL) in SWT broth at 30°C and plated on SWT agar using a sterile swab to form a lawn. Leisingera sp. JC1 was grown overnight to a density of ~1 × 10⁸ CFU/mL in SWT when testing with ANG isolates and in YTSS when testing with vibrio strains. This overnight broth of Leisingera sp. JC1 was spotted (10 μL) on the surface of each lawn in quadruplicate. All plates were incubated at 28°C for 24 h before imaging and ZOI measurements around the Leisingera sp. JC1 colonies. SWT or YTSS broth (10 μL) were spotted on each lawn as media controls, and 10 μL of the overnight culture of Leisingera sp. JC1 was spotted in quadruplicate on SWT or YTSS agar without any bacterial lawns as a growth control.

To quantify inhibition, an average of three ZOI diameters were measured and an average of three diameters of the JC1 colonies were measured using ImageJ (Schneider et al., 2012). Due to slight variations in JC1 colony size across trials, the measurements were normalized by subtracting the average JC1 colony diameter from the average ZOI diameter. To determine if differences in ZOIs across lawn densities per organism were statistically significant, one-way ANOVAs were performed. If the results of the one-way ANOVA indicated statistically significant differences, multiple comparisons post hoc Tukey tests were performed to determine which lawn densities were significantly different.

Leisingera sp. JC1 extracts were tested for antibacterial activity against V. fischeri ES114, V. anguillarum 775, and V. parahaemolyticus KNH1. High throughput assays with these bacterial strains were developed based on similar assays with natural product extracts and human pathogens (Zgoda and Porter, 2001), including obtaining CFU counts and growth curves for each of the vibrio strains as well as determining proper incubation times and temperatures and finding appropriate controls. These assays were performed in 96-well plates (Corning Costar, Corning, NY, USA) with SWT media and incubated at 28°C while shaking at 200 rpm. The bacterial inocula were prepared by adding select colonies into 5 mL of media and adjusted to OD₆₀₀ 0.1 (approximately 1–2 × 10⁸ CFU/mL as per Clinical and Laboratory Standards Institute, 2012). Colony forming unit (CFU) counts were manually confirmed to ensure accurate approximation for each vibrio strain.

Extracts were screened as previously described (Zgoda and Porter, 2001) with the following modifications. Briefly, master mix was prepared by addition of 1.6 mL adjusted vibrio inoculum, 7.84 mL sterile water, and 6.4 mL of SWT media. To each well, 198 μL of master mix was added with 2 μL of either positive control (chloramphenicol, final testing concentration 2.5 μg/mL), negative control (DMSO), or extract prepared in DMSO (screened at final concentration of 500 μg/mL; MIC performed using serial dilutions). Sterility control wells consisted of 98 μL sterile water, 100 μL of SWT media, and 2 μL of DMSO. All controls and samples were tested in technical triplicates with experiments repeated a minimum of three times to confirm results. Plates were read at 600 nm every 2 h from 0 to 10 h with a final reading at 24 h using a Synergy H1 Hybrid Reader (Biotek, Winooski, VT, USA). Results are given as percent control activity (PCA) calculated in comparison with DMSO, the negative control.

Direct analysis in real time-mass spectrometry (DART-MS) analysis was performed using a JEOL AccuTOF with DART ion source (IonSense, Inc., Saugus, MA, USA). High purity helium 5.0–6.0 grade (greater than 99.999% purity) was heated to 300°C and used for ionization. Five locations were selected on JC1 colonies in the presence or absence of V. fischeri, including (A) center of colony, (B) midpoint between center and edge of colony, (C) edge of colony, (D) ZOI (in the absence of V. fischeri sample was obtained from a point equidistant from colony edge), and (E) outside ZOI. At each location a sterile single use syringe needle (BD Medical, Franklin Lakes, NJ, USA) was placed in the sample and then placed between the DART ion source and the MS inlet. Positive ion MS data were obtained over a m/z range of 60–700 and relative percent abundance was obtained for the indigoidine ion. Standards were run after sampling each colony and mass spectral data were monitored in real time to ensure no residual indigoidine remained after each sample. DART-MS is only semi-quantitative due to the potential for differential ionization, suppression of ions, and/or changes in sample concentration in the DART ion source (Sanchez et al., 2011). Therefore, relative indigoidine ion abundance was used to generate heatmaps representing a gradient from less abundance (black) to more abundance (red).

JC1 bacterial inoculum was prepared by adding JC1 colonies into 5 mL of SWT_ng media in a 24 deep well plate, incubated for 24 h at room temperature while shaking at 200 rpm. Bacterial inocula for the vibrios were prepared by adding bacterial colonies of each species separately into 5 mL of SWT_ng media in 24 deep well plates, incubated for 2 h at 28°C while shaking at 200 rpm. All bacterial inocula (JC1, V. fischeri, V. anguillarum, V. parahaemolyticus) were adjusted to OD₆₀₀ 0.1 prior to use.

Co-cultures of JC1 with individual vibrios were prepared by adding 1 mL of adjusted JC1 inoculum to 10 mL SWT_ng media in 125 mL baffled flasks, incubated for 24 h at room temperature while shaking at 125 rpm, followed by addition of 200 μL of V. fischeri, V. anguillarum, or V. parahaemolyticus. After addition of the vibrio strain, co-cultures were incubated for an additional 24 h at room temperature while shaking at 125 rpm. Monocultures of JC1, V. fischeri, V. anguillarum, and V. parahaemolyticus were prepared by adding 1 mL of adjusted inoculum to 10 mL SWT_ng media in 125 mL baffled flasks, incubated for 48 h while shaking at 125 rpm.

All co-cultures and monocultures were extracted using the indigoidine enriched protocol described above. LC–MS data was obtained on the Agilent LC–MS system described above, using an isocratic method to ensure minimal baseline variation (10% acetonitrile in H₂O with 0.1% formic acid over 15 min at a flow rate of 1 mL/min with 20 μL injection volume). Extracts were prepared at 5 mg/mL in DMSO. Indigoidine was detected and quantitated via measurement of area under the curve at UV absorbance 299 nm and confirmed by MS.

Analysis with the antibiotic and Secondary Metabolite Analysis Shell (antiSMASH, Weber et al., 2015) predicted several potential secondary metabolite biosynthesis gene clusters (Supplementary Table S4). These results included three separate siderophore clusters, one bacteriocin, one HSL, one type 1 polyketide synthase (T1 PKS), one other PKS (not type 1,2,3, or trans-AT), and two clusters classified as “other.” Of these two “other” clusters, one contains the biosynthesis cluster for the known antimicrobial metabolite, indigoidine (Cude et al., 2012), while the other contains a previously described putative hybrid polyketide synthase/non-ribosomal peptide synthetase (PKS/NRPS) gene cluster known to be conserved amongst roseobacters (Martens et al., 2007). This PKS/NRPS gene cluster encodes a polyketide synthase, glycosyl transferase, non-ribosomal peptide synthetase, and phosphopantetheinyl transferase, but the product of this cluster has not yet been identified. The top homologous gene cluster of the T1 PKS is 45% similar to a cluster in the ANG isolate, Leisingera sp. ANG-M7. While some roseobacters are capable of producing the novel secondary metabolite TDA (Bruhn et al., 2006, 2007; Geng et al., 2008), genes for synthesis of this molecule were not found nor was the molecule detected via LC–MS (data not shown).

AntiSMASH predicted one luxIR homolog in Leisingera sp. JC1, flanked by an acyltransferase, crotonyl-CoA reductase, helicase, and oxidoreductase, similar to the previously published gene arrangement in bacterial isolates from the ANG (Collins et al., 2015). Production of HSLs by JC1 was confirmed in the A. tumefaciens NTL4 reporter assay, in which cell-free supernatant of a JC1 culture did induce β-galactosidase activity, indicating the presence of HSLs (Figures 2A,B). When compared to a dilution series of the N-3-oxohexanoyl HSL, JC1 produced a halo similar to that seen by 25 nM of HSL standard. The HSL production of JC1 was also slightly less than that of a closely related ANG isolate, Leisingera sp. ANG1.

Understanding the gene regulation by quorum sensing will be an important avenue of research for Leisingera sp. JC1 and the other E. scolopes ANG isolates due to the different habitats these bacteria experience. It is hypothesized that cephalopod ANGs are colonized via horizontal transmission from the environment (Kaufman et al., 1998), and potential symbionts must switch from living at very low cell densities in the seawater to very high cell densities in the ANG tubules (Collins et al., 2012). When ANG bacteria are deposited into the JC layers of eggs, these bacteria again experience a switch from the very high densities of the ANG to a lower density in the eggs. Due to this change in environments and cell densities, quorum sensing may play a role in gene regulation for ANG/egg JC bacteria.

Quorum sensing is also important in host-microbe interactions involving other roseobacters. For example, quorum sensing regulates motility and biofilm formation during host colonization in the sponge symbiont Ruegeria sp. KLH11 (Zan et al., 2012) and is necessary for colonization of the alga, Ulva australis by Phaeobacter gallaeciensis 2.10 (Rao et al., 2007). In other roseobacters, quorum sensing regulates secondary metabolite production, such as TDA in Phaeobacter gallaeciensis (Berger et al., 2011). In the indigoidine producing roseobacter, Leisingera sp. Y4I, there are two quorum sensing systems that regulate indigoidine production, pgaIR and phaIR (Cude et al., 2015). The JC1 luxI homolog has a 72% amino acid similarity to pgaI (RBY4I_1689) in Y4I, and the JC1 luxR homolog has an 81% amino acid similarity to pgaR (RBY4I_3631) in Y4I. The second set of luxIR homologs in Y4I, phaIR (RBY4I_3464 and RBY4I_1027), is not present in JC1. PgaI synthesizes the C8-HSL, produced by several proteobacteria, while PhaI synthesizes the 3OHC_12:1-HSL, which may be species specific. JC1 lacking the phaIR system may reflect its divergence from Leisingera sp. Y4I. Further analyses will be needed to understand if indigoidine production in Leisingera sp. JC1 is regulated by quorum sensing.

Three separate siderophore biosynthesis gene clusters were detected in the genome, as described above, and production of iron chelators was confirmed by plating on CAS agar (Figures 2C,D). Appearance of an orange halo around colonies indicates that iron was sequestered from the chrome-azurol S dye in the media. Siderophores are high-affinity iron chelators, and can provide a growth advantage to cells in iron-limited environments, such as in seawater and in colonization of hosts. Although, the presence of siderophore biosynthesis genes in the genomes of currently sequenced roseobacter clade members is rare, 10 of the 12 previously sequenced E. scolopes ANG roseobacter symbionts did have the genes and/or demonstrate production of siderophores (Collins et al., 2015). Similar to the majority of the squid-associated roseobacter clade, the Leisingera sp. JC1 genome contains siderophore biosynthesis genes, indicating that siderophore production may play a role in the ANG symbiosis.

The indigoidine biosynthesis gene cluster in Leisingera sp. JC1 contains all six biosynthesis genes previously described for Leisingera sp. Y4I (Cude et al., 2012) and shares a similar genome arrangement (Figure 3). To confirm the presence of individual members of the indigoidine biosynthesis gene cluster, primers were designed to three different components of the pathway, the non-ribosomal peptide synthetase (igiD), the transcriptional regulator (igiR), and one of the three indigoidine modification genes (igiC). The presence of these genes in JC1 genomic DNA was confirmed by PCR (Supplementary Figure S1).

Indigoidine biosynthesis genes have been detected in a diverse group of bacteria, including the Actinobacteria (Streptomyces), and Alpha-, Beta-, and Gamma-proteobacteria. The JC1 indigoidine biosynthesis operon shares the closest homology to the operon in Leisingera sp. Y4I, with 90–95% amino acid similarity for all gene products (Table 2). Other indigoidine biosynthesis operons share the non-ribosomal peptide synthetase, igiD, but many lack the same accessory genes required to modify indigoidine. When compared to other indigoidine producing strains, the igiD of JC1 is functionally homologous to other NRPS genes, sharing 49–53% amino acid similarity with Vogesella indigofera, Streptomyces lavendulae, and Dickeya dadantii 3937 (Table 2). A comparison with the genome of Leisingera sp. Y4I also confirmed that the indigoidine gene cluster is shared between these strains although absent from related ANG isolate Leisingera sp. M7 (Supplementary Figures S4 and S5).

Because of the distinctive morphology and the genetic evidence for indigoidine biosynthesis, Leisingera sp. JC1 was cultured and extracted to obtain chemical evidence of indigoidine production. Using a three-step culture process, a deep blue liquid culture was obtained. However, upon extraction using a typical resin-based organic extraction protocol, most of the blue color was insoluble in organic solvents and little evidence of indigoidine production was observed via liquid chromatography-mass spectrometry (LC–MS, see Figures 4D,E), integrating to only 0.8% of the JC1 normal extract and indicative of negligible indigoidine extraction using this method. Therefore, an indigoidine enriched extraction protocol was utilized to pellet the insoluble indigoidine away from other media and cellular components followed by dissolving the sample in DMSO (Yu et al., 2013), resulting in an indigoidine enriched extract with 91.1% indigoidine. Analysis via LC–MS confirmed the presence of indigoidine (Figure 4A) in the indigoidine enriched extract (Figures 4B,C) with a peak eluting at 10.7 min with an [M-H]^- of 247.0, consistent with the molecular weight and fragmentation pattern of indigoidine (248.2 g/mol) and in agreement with literature precedent (Yu et al., 2013). In addition, indigoidine was detected in two other JC and ANG isolates that exhibited a similar dark blue coloration in culture (data not shown).

While performing ZOI assays, there was a dramatic change in colony pigmentation of Leisingera sp. JC1 when grown alone (Figure 5B) as compared to growth under challenge with various vibrio strains (Figures 5C–G). Deep blue pigment production was observed uniformly when JC1 was grown in monoculture and appeared to localize to the outer edges of the colonies when presented with vibrio strains. Direct analysis in real time-mass spectrometry (DART-MS) is an ambient ionization technique in which samples can be analyzed without sample preparation or extraction (Sanchez et al., 2011). DART-MS was utilized to chemically confirm the visual observations of localization of indigoidine production of JC1 in monoculture and co-culture with V. fischeri over the course of 7 days (Figure 7). Five locations were selected on each colony including center (A), midpoint (B), edge (C), ZOI (D), and outside ZOI (E). In the absence of V. fischeri, indigoidine was uniformly produced throughout JC1 colonies (locations A–C). However, in the presence of V. fischeri there was little to no indigoidine production in the center or midpoints of JC1 colonies, but intense indigoidine detected around colony edges (Figure 7, location C only). Indigoidine was only minimally detected in the ZOI or outside the ZOI for either monoculture or co-culture. Trends in the localization of indigoidine production were even more apparent upon measurement after 7 days.

There are several examples of pigment production being induced when in the presence of other bacteria, such as in Staphylococcus aureus when co-cultured with Pseudomonas aeruginosa (Antonic et al., 2013) or production of a red pigment by Streptomyces lividans TK23 when co-cultured with Tsukamurella pulmonis TP-B0596 (Onaka et al., 2011). Pigment production can also be induced under other stress response conditions, such as protection from UV radiation (Tong and Lighthart, 1997). Pigment production has also been tied to photosynthesis (Orf and Blankenship, 2013), however, Leisingera sp. JC1 lacks genes associated with photosynthesis or carbon fixation (data not shown).

When grown alone, Leisingera sp. JC1 exhibited a uniform blue–black pigmentation across the colony which was confirmed by mass spectrometry to be essentially uniform production of indigoidine. Secondary metabolite biosynthesis is an energy intensive endeavor and production of antimicrobial compounds would typically be thought to be reserved for defensive situations. Since indigoidine is produced throughout the colony when in monoculture, and given its relatively moderate antibacterial activity as suggested by assays with the indigoidine enriched extract, it is also possible that indigoidine serves multiple functions for Leisingera sp. JC1. However, Leisingera sp. JC1 localized indigoidine production to the outer edges of the colony when co-cultured with V. fischeri and other vibrios. If utilized as a defensive compound, indigoidine may be localized to points of direct interaction with other microorganisms. Secondary metabolite production can be localized to susceptible locations such as in plants, sponges, and other sessile terrestrial and marine organisms (Amsler et al., 2001; Furrow et al., 2003; Van Dyck et al., 2010). The role of Leisingera sp. JC1 has yet to be examined directly in the ANG symbiosis but localized production of indigoidine or other secondary metabolites may play a role in egg defense or inhibition of other bacteria from colonizing the ANG (see conclusions below).

After observing localized production of indigoidine when grown on solid media with V. fischeri, additional co-culture experiments were undertaken in liquid media using several of the vibrios from the antimicrobial assays above. Leisingera sp. JC1 was grown in monoculture and in the presence of V. fischeri, V. anguillarum, and V. parahaemolyticus, followed by extraction and measurement of indigoidine production (Figure 8). Monocultures of all three vibrios were also grown and extracted as controls. Addition of V. fischeri to established cultures of Leisingera sp. JC1 resulted in a 1.38 fold increase in indigoidine. Co-cultures with V. anguillarum and V. parahaemolyticus resulted in a decrease in indigoidine production of approximately 0.5 fold for both organisms. Vibrio monocultures confirmed that these species do not produce indigoidine. Changes in indigoidine production were also visually evident with darker, more intense blue observed for extracts cultured with V. fischeri in comparison with JC1 monoculture, as well as lighter blue extracts observed for V. anguillarum and V. parahaemolyticus co-cultures (Figure 8A).

The increase in indigoidine production of JC1 with V. fischeri is consistent with the antibacterial activity observed for Leisingera sp. JC1 on both solid and liquid media (Figures 5 and 6), strengthening the hypothesis that indigoidine may play a protective role in association with E. scolopes. In addition, the downregulation of production with V. anguillarum and V. parahaemolyticus also supports the liquid culture bioassay data (Figure 6). The differential antimicrobial activity and indigoidine production between the three vibrios may be due to the purported role of the ANG and JC bacteria in the host. The ability of Leisingera sp. JC1 to inhibit V. fischeri may be related to the fact that the ANG is located directly posterior to the light organ, which harbors high densities of the sole symbiont, V. fischeri (McFall-Ngai, 2014). Each day 95% of viable V. fischeri cells in the light organ are expelled directly into the mantle cavity of the host as part of the regulatory mechanisms of that association (Boettcher et al., 1996; Nyholm and McFall-Ngai, 1998). A study from another squid, Doryteuthis pealeii (Kaufman et al., 1998) suggests that ANG bacteria are environmentally transmitted during development. Given that V. fischeri is not detected in the ANG (Collins et al., 2012), the inhibitory effect of Leisingera sp. JC1 and other ANG isolates may prevent V. fischeri and other vibrios from colonizing the ANG and thus help shape the consortium during development. Alternatively, inhibition against vibrios may play a role in egg defense since eggs are exposed to seawater for approximately three weeks and vibrios are known to be common members of the bacterioplankton.