Introduction

Sponges (phylum Porifera) are one of the most primitive multicellular animals and are sessile, filter-feeders. Sponges can form close associations with microbes that in some cases comprise up to 60% of the biomass of the sponges (Vacelet, 1975; Vacelet and Donadey, 1977; Wilkinson, 1978; Hentschel et al., 2006). Molecular techniques have revealed that these bacterial communities are remarkably diverse and include many novel bacteria (Taylor et al., 2007; Mohamed et al., 2008c; Webster and Blackall, 2009). Symbiotic relationships between sponges and microorganisms are considered to contribute to the health and nutrition of sponges (Mohamed et al., 2010). Many studies have shown the metabolic phenotypes of sponge-associated microbes (Taylor et al., 2007; Mohamed et al., 2008d, 2010), although little evidence exists confirming the contribution of symbiotic microbes to sponge well-being or survival (Webster and Blackall, 2009). Exceptions are the translocation of photosynthate from cyanobacteria to the host sponge (Wilkinson, 1979) and a decrease in health status of sponges associated with loss of cyanobacteria (Thacker, 2005).

Bacteria are known to effectively communicate with each other (Hooshangi and Bentley, 2008). One of the mechanisms for communication is termed quorum sensing (QS), in which bacteria monitor and respond to their own population density. In this process, diffusible signal molecules, sometimes called autoinducers, are produced at basal levels. The signals can accumulate proportionally to population density, and at critical threshold concentration, the signal binds to receptor proteins, which trigger expression of a specific spectrum of genes and phenotypes (Fuqua et al., 1994). Bacterial processes such as biofilm formation, bioluminescence, motility, virulence factor secretion, antibiotic production, sporulation and competence for DNA uptake can be controlled by QS systems (Ng and Bassler, 2009).

In Proteobacteria, acyl-homoserine lactones (AHLs) are a major type of molecular cue for QS, and were originally described as autoinducers. Bacteria that are able to produce AHLs have been isolated from marine sponges (Taylor et al., 2004; Mohamed et al., 2008a). Sponges harbor a higher proportion of AHL producers in their culturable bacterial communities compared with that of surrounding seawater (Mohamed et al., 2008a). AHL regulatory systems among these bacteria may have roles during bacterial colonization of the sponges or in controlling activities within the sponge symbiont community (Cicirelli, Hill and Fuqua, in preparation). They may also provide one means by which sponges could interact or interfere with bacteria (Taylor et al., 2007).

Autoinducer-2 (AI-2) is another well-known molecular cue and has been extensively analyzed in Vibrio harveyi and Vibrio cholerae, where it is involved in regulation of bioluminescence and virulence-associated traits (Miller et al., 2002; Lenz et al., 2004; Henke and Bassler, 2004a, 2004b). The activated methyl cycle is a crucial metabolic pathway to recycle homocysteine from the major methyl donor S-adenosyl methionine. LuxS, an S-ribosylhomocysteinase, catalyzes part of the cycle and functions to convert S-ribosylhomocysteine to homocysteine; meanwhile, it can also, as a side reaction, synthesize 4,5-dihydroxy-2,3-pentanedione, the precursor of AI-2. 4,5-Dihydroxy-2,3-pentanedione can spontaneously give rise to several furanone derivatives, collectively referred to as AI-2. The dual roles of luxS in metabolism and in AI-2 formation have led to controversy regarding its function in QS (Doherty et al., 2006; Rezzonico and Duffy, 2008). V. harveyi produces a third type of autoinducer CAI-1, an alpha-hydroxyketone (Tiaden et al., 2010), in addition to the AHL autoinducer (3-OH-C4-HSL) and AI-2. The three different types of autoinducers in V. harveyi bind to their respective membrane-bound receptors and converge on a central signal transduction pathway to regulate gene expressions (Ng and Bassler, 2009).

Thus far, no studies have been performed on the AI-2 systems among sponge-associated bacteria. During the process of isolating bacteria from the sponges Mycale laxissima and Ircinia strobilina, we obtained many Vibrio strains. We therefore focused on this investigation of the presence of luxS genes in sponge-associated bacteria on Vibrio spp. because of their high incidence in the sponges. Degenerate primers of luxS based on known Vibrio luxS sequences were designed and applied to all the Vibrio isolates. The AI-2 activity of all the Vibrio isolates was tested and a novel luxS sequence was cloned and analyzed further. The same set of primers was used to PCR-amplify the luxS genes directly from sponge bacterial communities.

Materials and methods

Sample collection

Three individuals of each of M. laxissima and I. strobilina were collected by SCUBA at Conch Reef, Key Largo, FL, USA in late October 2008 at a depth of ca 20 m. The water salinity was measured to be 36 p.p.t. using a portable refractometer (Fisher Scientific, Pittsburgh, PA, USA) at a water temperature ca 28 °C. Sponge samples were transported to the laboratory within 1 h of collection and sponges were rinsed three times with sterile artificial seawater for bacterial isolation and frozen at −20 °C for later DNA extraction. Water samples were collected within 1 m of the sponges at a depth of ca 20 m in a sterile 20-l container. Approximately 6 l of water was filtered through each of three 0.22 um pore-size Sterivex filters (Millipore, Billerica, MA, USA). The filters for extraction of DNA from bacteria in the water samples were flushed with 1X SET buffer (0.1 M NaCl, 10 mM Tris-Hcl, pH 8.0 1 mM EDTA) as described by Somerville et al. (1989) and stored at −20 °C.

Bacterial isolation and enumeration

Sponges were rinsed with sterile artificial seawater to remove transiently associated bacteria and then processed for isolation of culturable bacteria. Sponge tissue (1 cm3) was ground in artificial seawater using a sterile mortal and pestle and 10-fold serial dilutions were plated on Difco Marine Agar 2216 (BD Biosciences, Franklin Lakes, NJ, USA). Plates were incubated at 30 °C for 1 week at which time bacterial colonies were counted on plates with suitable dilutions (30–300 colonies). Approximately 50 colonies were selected for identification by 16S ribosomal RNA (rRNA) gene sequencing (below) to determine the proportion of vibrios in the culturable community. Serial dilutions of water samples were processed similarly for bacterial isolation. Total bacterial counts were determined using methods described by Mohamed et al. (2008d).

Identification of bacterial isolates by 16S rRNA gene sequence analysis

A single colony of each isolate was transferred to 15 ml of Marine Broth 2216 and incubated at 30 °C for 24–48 h. Bacterial genomic DNA was extracted from isolates using the UltraClean microbial kit (MoBio Laboratories, Carlsbad, CA, USA). Representatives of each morphotype based on careful observation of colony appearance were subcultured and cryopreserved at −80 °C in Marine Broth 2216 supplemented with 30% glycerol. Almost full-length 16S rRNA gene fragments were PCR-amplified using universal primers 27F and1492R as described by Enticknap et al. (2006). PCR products were sequenced using an ABI PRISM 3130xl genetic analyzer (Applied Biosystems, Foster City, CA, USA) and primers 27F and1492R. Sequences were assembled using online software CAP3 (http://pbil.univ-lyon1.fr/cap3.php) with a manual check. Chimeric sequences were identified by using the CHECK_CHIMERA program of the Ribosomal Database Project (Maidak et al., 2001) and the sequences were analyzed initially by using the BLASTn tool at the National Center for Biotechnology Information website (NCBI).

Extraction of genomic DNA from sponges and surrounding water samples

Freeze-dried sponge tissue (1 cm3) was ground using a sterile mortar and pestle. Total genomic DNA was extracted by a published method (Pitcher et al., 1989), modified for sponge tissues (Enticknap et al., 2006; Mohamed et al., 2008d). DNA was extracted from the filters obtained from surrounding seawater samples as described by Somerville et al. (1989).

Amplification of partial luxS gene sequences from Vibrio isolates

On the basis of sequences of known Vibrio luxS genes downloaded from Genbank, degenerate primers were designed for amplification of luxS gene fragments using online primer design software consensus-degenerate hybrid oligonucleotide primers (Rose et al., 2003). The primers were designated VluxsF (5′-TGCTGGACTCCTTCACCGTNGAYCAYAC-3′) and VluxsR (5′-TGCATGGCGGCGGTNCCRCAYTGTT-3′). These primers were designed based on an alignment of all available Vibrio luxS gene sequences. PCR mixtures consisted of 50 μl containing two units Platinum Taq polymerase (Invitrogen Life Technologies, Carlsbad, CA, USA), 1 × PCR Buffer, 2 mM MgCl2, 200 mM dNTPS (Fermentas, Glen Burnie, MD, USA), 0.2 μM each primer and 10–20 ng of genomic DNA or distilled water as a negative control. PCR cycling conditions for Vibrio luxS gene amplification consisted of 94 °C for 5 min followed by 30 cycles of 1 min at 94 °C, 1 min at 52 °C, and 1 min at 72 °C. A final 10 min extension step was done at 72 °C. PCR reactions were performed in a PTC-200 cycling system (Bio-Rad, Hercules, CA, USA). The PCR products with expected size ∼400 bp were purified using QIAquick gel extraction kit (Qiagen, Valencia, CA, USA) and sequenced as described above for 16S rRNA gene fragments using primer VluxsF.

Measurement of AI-2 activity

AI-2 activity was detected by using the reporter strain V. harveyi TL-26, in which all three signaling pathways are disrupted. Receptor genes for AHL and CAI-1 are mutated as well as the luxS gene. Thus, TL-26 can respond only to exogenously added AI-2 (not AHLs or CAI) and cannot produce AI-2 (Long and Bassler; personal communication). All the Vibrio isolates and V. harveyi TL-26 were grown overnight in Luria–Bertani medium with 2% NaCl. Culture supernatants of test strains were filter sterilized (0.22 μm) and 60 μl of sterile supernatant was added to 140 μl AB medium (Greenberg et al., 1979) containing V. harveyi TL-26 (inoculated 1:5000 from an overnight culture). Agrobacterium tumefaciens KYC55, which does not produce AI-2 (Zhu et al., 2003), was used as a control to test if this assay is specific. Vibrio mimicus ATCC 33653 was used as a positive control. The culture was incubated at 30 °C with shaking at 180 r.p.m. for 16 h and light production was measured by using a FLUOstar OPTIMA fluorescence microplate reader (BMGlabtech, Cary, NC, USA). The induction of luminescence by each tested supernatant was expressed relative to that in a negative control comprising sterile Luria–Bertani medium instead of culture supernatant.

AI-2 activity of JZ08IS72, one Vibrio isolate from the sponge I. strobilina, at different growth phases was measured. Briefly, a culture of JZ08IS72 grown in Luria–Bertani medium with 2% NaCl, incubated at 30 °C and shaken at 200 r.p.m. was sampled at several different points over a continuous period of 30 h. AI-2 activity was measured as described above.

Full-length cloning of luxS gene from isolate JZ08IS72

The full-length luxS gene (hereafter referred to here as luxS72) of isolate JZ08IS72 from the sponge I. strobilina was cloned using a GenomeWalker Universal Kit according to the manufacturer's instructions (Clontech Laboratories, Inc., Mountain View, CA, USA). Briefly, genomic DNA of JZ08IS72 was extracted using a CTAB method (Murray and Thompson, 1980). The DNA was digested with restriction enzymes provided by the kit and ligated to Genome Walker adaptors to construct a genomic library. The library was screened by luxS72 specific primers using a nested PCR method. The primer used for the first round PCR amplification of the 5′ end of the gene was luxsR1 (5′-GGTATGGATACCCTTTTCAGAAAGCAG-3′) and for the second round was luxsR2 (5′-CTGAGTAAAGCGAAGGTCAAACACAGT-3′). The primer used for the first round PCR amplification for the 3′ end was luxF1 (5′-CGTTGAGATCATCGACATCTCTCCTAT-3′) and for second round was luxF2 (5′- TACATGAGTCTGATCGGTACACCTACT-3′). The primers targeting the adaptor sequences were provided by the kit. All the PCR conditions were the same as recommended by the manufacturer. The PCR products were gel purified and sequenced. The sequences were aligned using Clustal X 2.0.12 (http://www.clustal.org/) and assembled using online software CAP3 (http://pbil.univ-lyon1.fr/cap3.php) and also checked manually. The assembled sequences were further analyzed by BLAST.

Heterologous expression of luxS72 in Escherichia coli DH5α

The coding sequence of luxS72 was PCR amplified using forward primer 5′-GCCCTGCAGCTGAAACAGGAA ACAGCTATGCCTTTACTAGATAGC-3′ (the PstI recognition site is underlined and the Shine–Dalgarno sequence from lacZ is both underlined and in italics) and reverse primer 5′-GGCCTCGAGTTACACCTTGAGGG-3′ (the XhoI recognition site is underlined). The PCR product was cloned into pCR4-TOPO vector (Invitrogen Life Technologies) and then subcloned into expression vector pSRKGm (Khan et al., 2008), named pSRKGm luxS72 and subsequently introduced into E. coli DH5α using standard techniques in Sambrook et al. (1989) and induced for expression using 1 mM isopropyl β-D-1-thiogalactopyranoside. At each cloning step, five clones were sequenced to identify clones with correct inserts. The activity of expressed luxS72 in E. coli DH5α was measured using the TL-26 AI-2 reporter as described above; and E. coli M4100, which can produce AI-2 (Sperandio et al., 1999), was used as positive control.

Clone library construction of luxS genes amplified from sponge-associated bacterial communities

Degenerate primers VluxsF and VluxsR were used to PCR-amplify putative Vibrio luxS genes from total genomic DNA extracted from sponges M. laxissima and I. strobilina, and also from surrounding seawater using the same PCR conditions used to amplify luxS genes from bacterial isolates. The PCR products were gel purified using a QIAquick gel extraction kit (Qiagen). Corresponding areas from the negative control sample in which the PCR was performed with no added template DNA were excised and taken through the same cloning procedure to provide strict negative controls. Purified PCR products were ligated into the pCR-XL-TOPO vector and transformed into One Shot TOP 10 chemically competent E. coli cells using a TOPO XL PCR cloning kit (Invitrogen Life Technologies). Plasmid DNA was isolated from individual clones and purified using a SprintPrep 384 kit (Agencourt Bioscience, Beverly, MA, USA). Sequencing was performed as described using M13 forward sequencing primer.

Phylogenetic analysis of 16S rRNA gene sequences and luxS gene sequences

16S rRNA gene sequences from isolates were edited using PreGap3 and Gap4 from the Staden package and analyzed using BLASTn at the NCBI website. Isolates were presumptively identified according to the identity of the closest cultured relative. Partial 16S rRNA gene sequences of all the Vibrio isolates were aligned using Clustal × 2.0.12 (http://www.clustal.org/) and a phylogenetic tree was constructed using software MEGA 4.0 (http://www.megasoftware.net/). luxS gene sequences from all the Vibrio isolates were converted to amino acid (aa) sequence using open reading frame finder and BLASTx at the NCBI website. luxS gene sequences from clone libraries were edited and the vector sequences were removed using the VecScreen tool at the NCBI website and were converted into aa sequences as described for isolates. The luxS gene sequences from isolates and clones were aligned and the phylogenetic tree was constructed as described for the 16S rRNA gene. To examine the novelty of the new cluster of isolates with luxS genes obtained in this study, the novel luxS genes were analyzed using BLASTn and the top 100 hits in Genbank were obtained. In addition, luxS genes in Vibrio spp. were searched for in Genbank using the key words ‘Vibrio AND luxS’. These two data sets were pooled and aligned together with all the luxS sequences obtained from this study. In order to present a comprehensive but legible tree, the following approach was used: if the reference sequences from GenBank shared higher than 99% similarity with each other, then only one representative was chosen for inclusion in the tree and the total number of sequences that shared over 99% similarity to that representative was given in brackets. The name and accession number of these sequences is presented in Supplementary Table.

Rarefaction analyses of the luxS genes clone libraries were performed using the mothur program (Schloss et al., 2009).

Nucleotide sequence accession numbers

16S rRNA gene sequences from isolates were submitted to GenBank under accession no. HM117107–HM117132. luxS gene sequences from isolates were submitted to GenBank under accession no. HM117133–HM117158 and luxS gene sequences from clone libraries were submitted to GenBank under accession no HM117021–HM117106.

Results

Partial luxS sequence and detection of AI-2 activity of Vibrio isolates from sponges

In total, 14 out of 47 isolates from M. laxissima, 10 out of 40 isolates from I. strobilina and 2 out 36 isolates from surrounding seawater were identified as Vibrio spp. by 16S rRNA gene sequences analysis (>1300 bp). For M. laxissima, the plate count was 1.2±0.2 × 107 colony-forming units (c.f.u.) cm−3 of sponge tissue (mean±s.e.), for I. strobilina, the plate count was 2.6±0.8 × 107 c.f.u. cm−3 of sponge tissue and for surrounding seawater, the plate count 2.0±0.5 × 105 c.f.u. ml−1. Total Vibrio isolates in 1 cm3 sponge tissue were calculated at ca 3.5 × 106 c.f.u. and 6.5 × 106 c.f.u. for M. laxissima and I. strobilina, respectively. For surrounding seawater, the culturable Vibrio count was ca 1.1 × 104 c.f.u. ml−1. Total bacterial counts obtained by microscopic enumeration of 4,6-diamidino-2-phenylindole-stained cells was 1.1±4.4 × 109 cells cm−3 and 2.8±0.4 × 109 cells cm−3 for M. laxissima and I. strobilina, respectively (Mohamed, 2007; Mohamed et al., 2008d).

As shown in the phylogenetic tree (Figure 1), a variety of different Vibrio species were retrieved. Six and nine isolates from the two sponges were closely related to strains of V. campbelli and V. harveyi, respectively. The other isolates from sponges were closely related to various Vibrio species, such as Vibrio agarivorans, Vibrio shilonii and Vibrio fortis. JZ08ML52 shared only 96% identity to its closest relative Vibrio rumoiensis (AY332407), indicating this strain might be a new species. 16S rRNA genes of two isolates from seawater (JZ08SW2 and JZ08SW6) shared 96–97% identity to that of Vibrio coralliilyticus (DJ440004), indicating that these isolates may also be new Vibrio spp.

Figure 1
figure 1

Rooted neighbor-joining phylogenetic tree of partial 16S rRNA gene sequences (>1300 bp) of Vibrio strains that were recovered from M. laxissima (JZ08ML), I. strobilina (JZ08IS) and seawater (JZ08SW). All the sequences retrieved in this study are in bold and accession numbers are listed in parentheses for each entry. Isolates in box represent the Vibrio isolates that carry novel luxS gene sequences shown in Figure 2. Reference sequences are shown with GenBank accession numbers listed after each sequence name. Bootstrap values >50% are shown at nodes. The scale bar indicates 0.10 substitutions per nucleotide position.

Full size image

Degenerate primers VluxsF and VluxR based on Vibrio luxS sequences were able to amplify partial luxS gene sequences from all the Vibrio isolates. A phylogenetic tree of the luxS gene sequences from these vibrios is shown in Figure 2. Using the approach described above to examine the novelty of the luxS cluster MI, about 130 sequences were analyzed. The tree based on Vibrio luxS sequences only (Figure 2) was very similar to that obtained by including all luxS sequences (Supplementary Figure 1). The majority of luxS genes from these sponge-derived Vibrio isolates are closely related to that of V. harveyi, which is consistent with the close relationship between these isolates and V. harveyi shown by 16S rRNA phylogeney. However, in several cases the phylogeny of luxS genes does not match that indicated by the corresponding 16S rRNA genes. For example, the 16S rRNA gene sequences of isolates JZ08ML69 and JZ08ML52 are closely related to those of V. harveyi (AB497065) and V. rumoiensis (AY332407), respectively; but the luxS gene sequences of these two isolated shared over 99% identity to each other. Similarly, the 16S rRNA genes of isolate JZ08ML48 grouped with V. fortis (AJ514917), but its luxS gene sequence clustered with the luxS genes from isolates JZ08ML43, JZ08ML44, JZ08ML49, JZ08ML63 and JZ08IS73.

Figure 2
figure 2

Phylogenetic tree using neighbor-joining method based on the predicted 96 aa residues encoded by luxS genes from Vibrio isolates. Sequences isolated from M. laxissima (JZ08ML), I. strobilina (JZ08IS) and seawater (JZ08SW) in this study are in bold and the accession number of each of these luxS sequences is listed after the isolate name. The numbers in [] indicates the numbers of sequences in Genbank that shared over 99% identity to the sequence on the tree and these sequences are listed in Supplementary Material Table 1. The novel luxS cluster is named Cluster MI. Bootstrap values >50% is shown at nodes. The scale indicates the number of aa substitutions per site.

Full size image

A novel cluster (named cluster MI) of luxS gene sequences, comprising those from isolates JZ08ML43, JZ08ML44, JZ08ML48, JZ08ML49, JZ08ML63, JZ08IS73, shared only 88–90% identity to their closest relative in Genbank (boxed in Figure 2). JZ08IS72 was chosen as a representative of this group for further analysis to confirm the novelty of these luxS genes.

All the isolates are evaluated for AI-2 activity using reporter strain V. harveyi TL-26. All isolates were able to induce light production (Figure 3) whereas A. tumefaciens KYC55 did not induce light production (data not shown), showing that all the Vibrio isolates, all of which have luxS genes, are able to synthesize AI-2-type molecules.

Figure 3
figure 3

AI-2 activities of all Vibrio isolates measured by reporter strain V. harveyi TL-26. Fold changes were calculated by comparing the light production induced by supernatant of individual log phase cultures to that using Luria–Bertani (LB) medium instead of culture supernatant. V. mimicus ATCC 33653 was used as positive control. Supernatant from A. tumefaciens KYC55 failed to induce light production. Bars are means from triplicate experiments; error bars represented the s.e.m.

Full size image

JZ08IS72 was grown in shaken culture in Luria–Bertani medium with 2% NaCl at 30 °C. AI-2 activity in the culture supernatant was detected from early exponential phase (2 h after inoculation) and onward. It reached the maximal level during transition to stationary phase (6 h after inoculation) and decreased afterward with some fluctuations as shown in Figure 4.

Figure 4
figure 4

AI-2 production profile during growth of strain JZ08IS72. Strain JZ08IS72 was grown in Luria–Bertani (LB) with 2% NaCl. The growth (open squares) and AI-2 activity (closed circles) were monitored at the indicated time intervals. The error bars represent s.e. of triplicate samples.

Full size image

Characterization of luxS sequence of JZ08IS72 and expression in E. coli DH5α

Using a genome walking method, the full length of luxS72 was cloned. Results showed that an open reading frame encoding a putative luxS gene designated luxS72 had 516 nucleotides and encoded 171 aa. It shared 88% identity with its closest luxS relative from Vibrio parahaemolyticus 16 (accession number ZP_05120871) at the aa level, although no significant similarity was found between luxS72 and that of V. parahaemolyticus 16 at the nucleotide level, indicating high divergence of DNA sequence of luxS72. Alignment of luxS72 to that of several other Vibrio species indicated that the variance of aa largely occurred in the C-terminus of the protein and luxS72 has all the conserved aa identified in Hilgers and Ludwig (2001) (see Supplementary Material Figure S2).

By walking upstream and downstream of luxS72 gene, partial sequence of the glutamate-cysteine ligase gene (closely related to that of V. harveyi HY01 (ZP_01988480)) was found upstream of luxS72 and partial sequence of a putative Mg2+ and Co2+ transporter corB gene was found downstream of luxS72 (data not shown).

In order to confirm the function of the novel luxS72, it was cloned into the pSRKGm expression vector, placing it under control of the isopropyl β-D-1-thiogalactopyranoside-inducible lacZ promoter (PLac) and expressed in E. coli DH5α, which does not produce AI-2 because of a mutation in the luxS gene (Surette and Bassler, 1998). Results shown in Figure 5 demonstrate that luxS72 was successfully expressed in the heterologous host. Significant AI-2 activity was detected compared with the empty vector and control (P<0.05) and the heterologous production of AI-2 because of expression of luxS72 was comparable to a luxS-proficient strain E. coli MC4100 (P>0.05).

Figure 5
figure 5

Heterologous expression of luxS72 in E. coli DH5α. luxS72 was cloned into expression vector pSRK-Gm and chemically transformed into E. coli DH5α. The transformant was grown in Luria–Bertani (LB) at 37 °C up to an OD600 of ca1.2 and induced by isopropyl β-D-1-thiogalactopyranoside (1 mM) for 3 h. E. coli MC4100 was used as a positive control. All the strains were grown up to an OD600 of 1.80±0.02. The AI-2 activity was measured as described in Materials and methods section. The bars (from the left) represent the result of AI-2 activity from E. coli DH5α, the empty vector control, E. coli DH5α expressing luxS72, and the positive control E. coli MC4100. Experiments were performed in triplicate and error bars indicate the s.e.m.

Full size image

Sequence analysis of luxS genes from clone libraries

luxS genes were successfully PCR-amplified from marine sponges M. laxissima and I. strobilina and also from surrounding seawater from Key Largo, using degenerate primers VluxSF and VluxSR. Phylogenetic relationships of the luxS-deduced aa sequences from these two marine sponges and surrounding seawater samples were analyzed. Phylogenetic relationships based on nucleotide sequences were also determined. Overall, these two trees based on aa sequences and nucleotide sequences were generally congruent with each other (data not shown). In total, 33, 30 and 23 clones of luxS genes were recovered and sequenced from the sponges M. laxissima, I. strobilina and seawater, respectively. To determine the relative richness and coverage of luxS genes sequenced from M. laxissima, I. strobilina and surrounding seawater, rarefaction analyses were performed. The rarefaction curves for luxS genes from M. laxissima and I. strobilina approached a plateau, and nearly plateaued for surrounding seawater at the distance of 0.01 (see Supplementary Material Figure S3). At the distance of 0.03, all the three rarefaction curves reached a plateau, indicating sufficient sampling of these clone libraries (data not shown). The phylogenetic analysis of these sequences is shown in Figure 6. In all, 31 out of 33 clones from M. laxissima, referred to as ML group 1 shared 97–99% identity to luxS gene of V. mimicus VM603 (ZP_05718770), and the other two clones, referred to as ML group 4 shared 100% identity to the luxS gene sequence of V. harveyi ATCC BAA-1116 (YP_001446656). In total, 28 out of 30 clones from I. strobilina and nine clones from seawater, referred to as IS and SW group 5, were closely related to the luxS gene of V. harveyi 1DA3 (ZP_06175640), sharing 96–99% identity and the other two clones from I. strobilina (IS group 3) shared 94% identity to the luxS gene of V. parahaemolyticus RIMD 2210633 (NP_798916). In all, 14 out of 23 clones from seawater, referred to as the SW group 2, were closely related to the luxS gene of V. mimicus VM603 (ZP_05718770), sharing 98–99% identity.

Figure 6
figure 6

Phylogenetic tree using the neighbor-joining method based on the predicted 117 aa residues encoded by the luxS gene amplified from the bacterial communities of M. laxissima (ML), I. strobilina (IS) and surrounding seawater (SW). Sequences obtained in this study are in bold. Bootstrap values > 50% are shown at each node. The scale bar indicates the number of aa substitutions per site. Numbers in () after each group name indicate the number of clones in each group. See Supplementary Material Table 1 for accession numbers of each clone. Numbers in [] after reference sequences are the same as in Figure 2 legend. See Supplementary Material Table 2 for accession numbers of each clone.

Full size image

Discussion

The marine sponges M. laxissima and I. strobilina contained several different Vibrio strains, identified by 16S rRNA gene sequence analysis. Associations of vibrios with other marine organisms range from pathogenic to mutualistic. Close relatives of the Vibrio isolates described in this study include V. harveyi, a major pathogen of marine animals (Zhang et al., 2001) and V. shiloli (AF007115), well established as a causative agent of coral bleaching (Kushmaro et al., 1996, 1997). Pathogenicity is often strain-specific rather than species-specific so the close relationship between the sponge-associated Vibrio species and known pathogens does not necessarily indicate that our Vibrio isolates are potential pathogens of their host sponges. Known mutualist relationships between Vibrio spp. and marine invertebrates include Vibrio fischeri-bobtail squid (McFall-Ngai, 2008). Sponge-associated vibrios might offer the host sponge or the associated bacterial communities some benefits. One possibility is that nitrogen-fixing vibrios may supply the holobionts with fixed nitrogen. Shieh and Lin (1994) described vibrios as nitrogen-fixing bacteria associated with sponge Halichondria. However, none of the vibrios isolated in this study are closely related to previously described nitrogen-fixing Vibrio spp. and none of the nifH gene sequences retrieved from the microbial communities of M. laxissima and I. strobilina is closely related to the nifH genes of vibrios (Mohamed et al., 2008b). The role of the Vibrio spp. associated with M. laxissima and I. strobilina therefore remains speculative.

Using degenerate luxS primers based on Vibrio luxS gene sequences, partial luxS gene sequences were detected in all Vibrio isolates from the two marine sponges and surrounding seawater. A novel cluster containing luxS genes from six isolates shared only low identity (88–90%) to known luxS gene sequences. Our results thus extend the known diversity of this QS gene. luxS genes are highly divergent among gram-negative and gram-positive bacteria but might have similar functions. For example, the luxS gene of Bacillus subtilis shares only ca 40% identity to that of V. cholerae and V. harveyi; however, AI-2 produced in B. subtilis can induce the bioluminescence of V. harveyi BB170 and LuxS-dependent QS is required to form a differentiated biofilm and also for swarming on solid surfaces (Lombardia et al., 2006). Further, even among the vibrios, the luxS gene of V. cholerae shares only 87% identity to that of V. harveyi yet the lux operon of V. harveyi can function well in V. cholerae (Lenz et al., 2004). This suggests that the novel luxS cluster obtained in this study might be involved in the same pathway as that of V. harveyi or V. cholerae. The LuxS72 aa sequence has all of the uniformly conserved residues for other S-ribosylhomocysteinases, including those implicated in metal binding (Hilgers and Ludwig, 2001). It is therefore not surprising that this protein can function to drive synthesis of AI-2 when expressed in E. coli DH5α.

A comparison of the plate counts obtained on marine agar and the total cell counts present in the sponges revealed that ca 1% of sponge-associated bacteria were culturable from both sponges. Vibrios comprised a major group of bacteria in the culturable assemblage from both sponges (ca 30% and 25% for M. laxissima and I. strobilina, respectively). We therefore focused on Vibrio spp. for this investigation of the presence of luxS genes in sponge-associated bacteria because of the high incidence of Vibrio spp. in the total bacterial assemblage and because there is a precedent for luxS signaling to be important in colonization of a marine invertebrate by a Vibrio sp., in the V. fischeri/Euprymna scolopes symbiosis (Lupp and Ruby, 2004). We acknowledge that the proportion of Vibrio spp. in the total sponge-associated bacterial community may be low (our data establish a lower limit of ca 3 Vibrio cells/10 000 prokaryote cells in these two sponges) and that the ecological significance of luxS signaling in these communities remains to be established. A recent study obtained 25 Vibrio isolates from healthy and diseased corals and showed that all the Vibrio isolates were able to activate an AI-2 reporter, although no luxS genes were amplified in this study (Tait et al., 2010). The investigators speculated that vibrios might use several different types of QS signal molecules to regulate processes that are used to colonize corals during adverse environmental conditions. Studies by Bansal et al. (2008) and Englert et al. (2009) showed that E. coli cells are attracted by AI-2 and AI-2 can be as effective an attractant as L-aspartate. On the basis of these studies, one hypothesis is that E. coli might use AI-2 to join groups of bacteria that produce AI-2 (Defoirdt, 2010). It is also possible that chemotaxis to AI-2 might serve to attract free-living bacteria to biofilms (Hegde et al., 2011). AI-2 is better understood in Vibrios than in other bacteria. AI-2 has been well characterized in V. harveyi and V. cholerae as a signaling molecule. It controls bioluminescence, metalloprotease production and type III secretion in V. harveyi and represses biofilm and virulence and activates protease activity in V. cholerae (Ng and Bassler, 2009). In light of these observations, it is possible that similar functions are controlled by AI-2 signaling in sponge-associated Vibrios.

Interestingly, in some cases there was a large difference in the levels of AI-2 activity between closely related isolates (for example, isolates JZ08ML70 and JZ08ML71 show around fivefold difference in AI-2 activity. The luxS gene has been showed to be regulated both at the transcriptional and translational levels (Zhang et al., 2008; Udekwu, 2010). The difference in the levels of AI-2 activity between closely related isolates, (for example, JZ08ML70 and JZ08ML71 show around fivefold difference in AI-2 activity) could be due to the difference in the gene regulation.

AI-2/luxS QS systems among vibrios can regulate genes involved in biofilm formation, virulence factors, bioluminescence and colonization (Ng and Bassler, 2009). In the well-characterized V. fischeri—E. scolopes colonization system, double mutants of luxS and ainS (an AHL synthase gene) reduces the already diminished level of colonization of ainS mutants by 50% (Lupp and Ruby, 2004). Almost all the Vibrio isolates obtained in this study (Zan, Fuqua and Hill, in preparation) and in previous report by Mohamed et al. (2008a) also produce AHL. The role of the AI-2/luxS system in the colonization of sponges by vibrios provides an interesting area for future work.

The AI-2 activity of JZ08IS72 peaked 6h after inoculation, decreased during the transition to stationary phase and there was almost no AI-2 activity detectable after 18h following inoculation. In contrast, the AI-2 activity of V. mimicus ATCC 33653 detected in parallel with that of JZ08IS72 did not show this pattern. In V. mimicus ATCC 33653, AI-2 activity peaked between 8 and 10h after inoculation and remained relatively constant with some fluctuations thereafter (data not shown). There are no other reports of AI-2 systems from Vibrio species showing a peak in production followed by a decline, not even in V. harveyi and V. cholerae, in which the AI-2 QS pathway is well studied. However, in Salmonella enterica serovar Typhimurium and E. coli K-12 strain MG1655, AI-2 production peaks in mid-to-late exponential phase and quickly declines either in the transition to stationary phase or in stationary phase (Xavier and Bassler, 2005; Doherty et al., 2006). This is because of the internalization of AI-2 by the Lsr transporter in S. enterica serovar Typhimurium and by the b1513 transporter (a homolog of the Lsr transporter) in E. coli K-12 strain MG1655 (Xavier and Bassler, 2005). No homolog of the Lsr transporter was found in the genomes of V. harveyi and V. cholerae (data not shown). This suggests that a potential new pathway involving in internalization of AI-2 or perhaps degradation may exist in JZ08IS72, which warrants further investigation. It is also possible that other environmental factors might affect the AI-2 activity of JZ08IS72 during stationary phase, such as the lack of a preferred carbon source, neutral pH, and low osmolarity (Surette et al., 1999). The ecological relevance of the peak-shaped pattern is not clear.

In the luxS gene clone library from M. laxissima, ML group 1 was clearly distinct from SW group 2 with a high bootstrap value of 82% and did not cluster with any other known luxS gene sequences, implying that ML group 1 might be a sponge-specific luxS gene group. IS group 3 is a relatively small group that is distantly related to any known Vibrio luxS sequence, suggesting that this group comprises a sponge-specific luxS gene group. Clones from I. strobilina and seawater in IS and SW group 5 were closely related to each other, suggesting that strains carrying these genes may be derived from the surrounding seawater.

Conclusions

All the Vibrio spp isolated from two marine sponges were shown to have luxS gene and were demonstrated to induce light production in a reporter strain. A novel cluster of luxS gene (cluster MI) sequences was obtained. The relationship between AI-2/luxS QS and the colonization on sponge hosts by vibrios deserves further investigation. This work provides the first information about luxS genes and AI-2 activity in sponge-associated bacterial communities.