Diversity and antibacterial activity of bacteria cultured from Mediterranean Axinella spp. sponges


  • M. Haber,

    1. Department of Zoology, George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • M. Ilan

    Corresponding author
    1. Department of Zoology, George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel
    • Correspondence

      Micha Ilan, Department of Zoology, George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv 69978, Israel.

      E-mail: milan@post.tau.ac.il

    Search for more papers by this author



Evaluation of the diversity and antibacterial activity of bacteria cultivated from Mediterranean Axinella sponges and investigating the influence of culture conditions on antibacterial activity profiles of sponge bacteria.

Methods and Results

Based on 16S rRNA gene sequence analysis, the 259 bacteria isolated from the three Mediterranean Axinella sponges A. cannabina, A. verrucosa and A. polypoides belonged to 41 genera from the four phyla Actinobacteria, Bacteroidetes, Firmicutes and Proteobacteria and included five potential newly cultured genera. In antagonistic streak assays, 87 isolates (34%) from 13 genera showed antibacterial activity towards at least one of the 10 environmental and laboratory test bacteria. The extracts and filtrates of 22 isolates grown under three different culture conditions were less often active as the isolates in the corresponding antagonistic streak assays. Changes in antibacterial activity profiles were isolate- and culture condition-specific.


Axinella sponges are a good source to cultivate phylogenetic diverse and hitherto novel bacteria, many of which with antibacterial activity. Analysis of induced antibacterial activities might enhance the role of sponge bacteria in efforts to isolate new antibiotics in the future.

Significance and Impact of the Study

This study was the first to investigate the diversity and antibacterial activity of bacteria isolated from A. cannabina and A. verrucosa. It highlights the potential importance of induced activity and the need for employing multiple culture conditions in antibacterial screening assays of sponge-associated bacteria.


Sessile marine invertebrates, such as sponges (phylum Porifera), frequently use secondary metabolites to protect themselves from competitors for space (Engel and Pawlik 2000; Pawlik et al. 2007), predators (Pawlik et al. 1995; Burns et al. 2003; Sokolover and Ilan 2007), pathogens (Kelman et al. 2001; Mayer and Hamann 2005) and the settlement of fouling organisms (Kubanek et al. 2002; Tsoukatou et al. 2002; Haber et al. 2011). In addition to their ecological importance, these compounds can have promising pharmaceutical and biotechnological activities. Marine sponges are a major source for such structural diverse novel natural products (see reviews by Mayer and Hamann 2005; Blunt et al. 2011 and references therein). However, as is the case for many natural products from marine invertebrates, the promising sponge metabolites are often present in minute quantities in the organisms thus preventing harvest from nature without harming the survival of the sponge populations (Proksch et al. 2002; Imhoff et al. 2011). The structural similarity of many marine natural products to those of known microbial origin has lead to an increased interest in the microbial symbionts of marine invertebrates, as they might be the true producers of compounds isolated from their hosts, and therefore might be a solution to the supply problem in later stages of the pharmaceutical pipeline (Salomon et al. 2004; Waters et al. 2010).

Marine sponges are hosts to a diverse microbial community including Archaea (Margot et al. 2002), diatoms (Webster et al. 2004), dinoflagellates (Garson et al. 1998), fungi (Paz et al. 2010) and bacteria (Schmitt et al. 2012). The potential bacterial production of natural products previously identified as sponge metabolites has been investigated by several techniques such as cell disassociation followed by sorting and chemical extraction of different cell fractions (Unson et al. 1994), antibody staining (Gillor et al. 2000) and MALDI-Tof imaging (Simmons et al. 2008; Yarnold et al. 2012). However, metabolites might be produced by one cell type and accumulated in another, especially if the target molecules are small and can be easily transported between cells (Simmons et al. 2008). Unequivocal evidence of the origin of a compound is therefore only obtained by either localizing the biosynthetic gene cluster or by isolating the compound of interest from pure cell cultures (Salomon et al. 2004; Taylor et al. 2007; Simmons et al. 2008). A bacterial origin of some sponge metabolites has successfully been confirmed by both approaches, for example the complete gene cluster for psymberin was isolated from the metagenome of the sponge Psammocinia aff. bulbosa and supported a bacterial origin based on the genetic features (Fisch et al. 2009), and manzamine A originally found in the sponge Acanthostrongylopora sp. has later been isolated from pure cultures of its symbiotic actinobacterium Micromonospora sp. strain M42 (Hill et al. 2005; Peraud 2006).

Sponge bacterial communities differ from those of the surrounding seawater (Hentschel et al. 2002). Several phylogenetic clades specific to sponges have been reported, mainly by culture-independent methods (Taylor et al. 2007; Simister et al. 2012). Overall, sponge-associated bacteria are affiliated with 17 formally described bacterial phyla and several more, yet uncultured, candidate phyla (Simister et al. 2012). Despite their close phylogenetic relatedness, there is still a high diversity between sponge bacterial communities as there is only little overlap between communities from different sponge species on a bacterial species, genus and family level and most taxa are sponge species-specific (Schmitt et al. 2012; Giles et al. 2013). From a pharmacological viewpoint, unique bacteria are likely the source of hitherto unknown natural products, as they need to adapt to unique microhabitats (Jensen and Fenical 1996). Actinobacteria are among the dominant bacterial groups in sponges (Montalvo et al. 2005; Taylor et al. 2007; Vicente et al. 2013). They are known to produce a variety of interesting bioactive compounds, especially antibiotics. New antibiotics are continually needed because of the development and natural existence of resistant pathogens, the evolution of new diseases and the toxicity of some of the current compounds (Demain 1999).

In this study, we investigated the culturable bacterial community of three sympatric Mediterranean sponges: Axinella cannabina, A. polypoides and A. verrucosa, and evaluated their antibacterial activity. Axinella sponges are well known for their specific associations with Archaea (Margot et al. 2002), but a recent deep sequencing study also indicates the presence of a diverse bacterial community (White et al. 2012). Recent phylogenetic studies suggest that the family Axinellidae and its name-giving genus Axinella are polyphyletic (Alvarez et al. 2000; Gazave et al. 2010). The three species studied here are grouped in distinct clades with different chemical composition (Gazave et al. 2010). Isocyanids have been isolated from A. cannabina (Cimino et al. 1975), while pyrroles are present in A. verrucosa (Aiello et al. 2006; Haber et al. 2010). Both compound classes are absent in A. polypoides (Cimino et al. 1975). Thus, these three sponges represent different microhabitats, which we hypothesized should select for different adaptations in their bacterial symbionts including the production of a variety of secondary metabolites. The antibacterial activity of the isolated bacteria was screened in an antagonistic assay against environmental and laboratory test bacteria. To assess the potential for biotechnological exploitation, 22 of the more active isolates were then grown under three different conditions, and their antibacterial activity was re-examined.

Material and methods

Sponge sample collection and bacterial isolation

Samples were taken from three specimens of each of the two sponge species A. polypoides and A. verrucosa and from a single A. cannabina specimen. The samples were collected by Scuba diving at Sdot Yam, Israel (32°29.77N; 34°53.23E), at 30 m depth in November, 2006, stored underwater in ziplock bags and brought to the laboratory within 3 h after collection. Samples were cleaned from macro-epibionts (such as barnacles and hydrozoans) and rinsed with sterile artificial seawater (40 g l−1 Instant Ocean®; United Pet Group, Blacksburg, VA, USA) (ASW). The next steps were performed under sterile conditions in a laminar flow hood. A representative piece of about 1 cm3 from each sample was cut into small pieces and thoroughly ground with sterile mortar and pestle in 9-ml sterile ASW. The liquid part of each of the obtained suspensions gave the initial 100 dilutions and was further diluted to 10−1, 10−2 and 10−3. For each sample, 100 μl of the latter three dilutions was plated on marine agar plates (37·4 g l−1 marine broth 2216, 18 g l−1 bactoagar, pH 7·2) with three replicates per dilution for A. verrucosa and A. polypoides and two replicates for A. cannabina. 100 μl of the 100 and 10−1 dilutions was plated out on SCA [10 g l−1 soluble starch, 1 g l−1 casein (dissolved in 10 ml 1 mol l−1 NaOH), 0·5 g K2HPO4, 20 g l−1 NaCl, 20 g l−1 bactoagar, pH 7·2], modified ISP2 (4 g l−1 yeast extract, 10 g l−1 malt extract 4 g l−1 dextrose, 20 g l−1 NaCl, 20 g l−1 bactoagar, pH 7·2) and M1 media (10 g l−1 soluble starch, 4 g l−1 yeast extract, 2 g l−1 peptone, 20 g l−1 NaCl, 18 g l−1 bactoagar, pH 7·2). Again three replicates per dilution were prepared for A. verrucosa and A. polypoides and two for A. cannabina. M1 was not used for A. cannabina. Plating was always done until the agar surface was dry. SCA and modified ISP2 agar plates were supplemented with 10 μg ml−1 nalidixic acid (against Gram-negative bacteria), 25 μg ml−1 nystatin and 10 μg ml−1 cyclohexamide (both against fungi), while M1 was supplemented with 5 μg ml−1 rifampin (against Gram-negative bacteria) and 100 μg ml−1 cyclohexamide. Antibiotic stocks were 0·22 μm filter-sterilized before use, with the exception of nystatin, which does not completely dissolve in water, and were added to the autoclaved medium at a temperature of around 45°C under the laminar flow hood. Antibiotics were used to select for slow growing Gram-positive bacteria such as Actinobacteria, a group known for producing a wide array of bioactive compounds (Bérdy 2005; Fenical and Jensen 2006).

After inoculation, plates were incubated at room temperature (20–24°C). Plates for Gram-positive bacteria were kept in the dark to avoid light degradation of the antibiotics. Plates were sealed with Parafilm® (Brand, Wertheim, Germany) after 1 day to keep them humid and avoid contamination. Strains were selected from marine agar plates after 7 days, and from SCA, modified ISP2 and M1 plates after 19 and 39 days. The selection was based on diversity of colony morphology. Colonies were transferred to fresh plates with an inoculation loop by gently touching the surface and spread applying isolations streaks to allow the growth of single colonies. All bacteria from SCA, modified ISP2 and M1 plates were transferred to modified ISP2 plates without antibiotics to decrease the number of different media used. If these bacteria did not grow on modified ISP2 plates, they were picked again from the original plate and transferred to plates with their original growth medium without antibiotics. Colonies were restreaked on fresh plates until pure cultures were obtained. Stocks were prepared in 25% glycerol in liquid medium and stored at −80°C.

Molecular identification of isolated bacteria

For identification, the bacteria were grouped based on morphology and medium. After PCR amplification of the 16S rRNA gene, application of restriction fragment length polymorphism (RFLP) analysis enabled grouping of isolates, which were then compared based on 16S rRNA gene sequencing of a representative member of each RFLP pattern and its molecular identification.

DNA for PCR amplification was obtained by transferring bacterial cells of a single colony with a sterile toothpick into a 1·7-ml Eppendorf tube containing 30 μl of TE buffer (10 mmol l−1 Tris-HCl, 1 mmol l−1 EDTA) followed by three cycles of heating to 100°C for 5 min and freezing 1 h at −80°C) to lyse the cells. From the resulting suspension, 1 μl was added to the PCR mix consisted of 2·5 μl 10× PCR Buffer, 2·0 μl 2·5 mmol l−1 dNTPs, 0·25 μl of each 100 mmol l−1 primer (63f and 1387r) (Marchesi et al. 1998), 0·2 μl Dream Taq polymerase (Fisher Scientific GmbH, Schwerte, Germany) (5 units μl−1) and 19·8 μl molecular water. PCRs were run in a Bioer XP cycler (Bioer Technology, Tokyo, Japan) with the following temperature profile: 3 min 94°C, 30 cycles of 1 min 94°C, 1 : 20 min 54°C, 2 min 72°C and a final extension of 5 min 72°C. The RFLP analysis followed the approach by Bergman et al. (2011). Three micro litre of the PCR products was added to a reaction mix consisting of 1 μl 10× Buffer R+, 5·5 μl molecular water and 0·5 μl of the restriction enzyme BsuRI (=HaeIII) (10 units μl−1) (Fermentas) and incubated for 3 h at 37°C. The restriction patterns of the members within a group were compared with each other following electrophoresis at 60V for 2 h on a 2% (w/v) agarose TBE (89 mmol l−1 Tris-Base, 89 mmol l−1 boric acid, 2 mmol l−1 EDTA) gel spiked with 2 μl ethidium bromide. Gels were observed under UV, a digital photo was taken for documentation and used to determine the RFLP groups. The PCR product from a representative of each banding pattern present in a morphological group was sent to MCLAB (San Francisco, CA, USA) for sequencing with the forward primer 63f. Obtained raw sequence data were visualized and manually edited using the 4Peaks software, version 1.7.2 (A. Griekspoor and Tom Groothuis, available at mekentosj.com). The edited sequences were compared with the closest validly described bacterial species using the EzTaxon-e server (Kim et al. 2012). Isolates with less than 97% 16S rRNA gene similarity were considered as undescribed species and as new genera if the similarity was below 95%. All bacterial 16S rRNA gene sequences obtained in this study were submitted to GenBank, and the accession numbers are JN699088-158 and JN699062 for A. polypoides, JN699163-228 and JN699063-64 for A. verrucosa and JN699065-87 for A. cannabina, respectively.

To confirm the taxonomic assignment, a phylogenetic analysis was performed. For each sponge-associated isolate, the 16S rRNA gene sequence of the closest validly described species was downloaded from the EMBL database, as well as all available sequences of isolates previously cultured from the three studied sponge species. Thermotoga maritima (Phylum: Thermotogae) was used as an outgroup. Sequences were aligned using Infernal 1.0 (Nawrocki et al. 2009) as implemented in the ribosomal database project, version 10 (Cole et al. 2009). The obtained alignment was manually improved and ambiguously aligned positions as detected by Gblocks, version 0.91b, (with default settings, but allowing gaps in the final alignment if present in <50% of the sequences) (Castresana 2000) were removed. The resulting alignment was used to find the best neighbour-joining tree with mega 5.0 (Tamura et al. 2011) using Kimura-2-parameter distances, pairwise deletion of ambiguous positions and 1000 bootstrap replicates. Figtree, version 1.3.1, as available online (tree.bio.ed.ac.uk/software/figtree) was used to visualize the obtained tree.

Coverage analysis

A coverage analysis of the isolated bacteria was performed for each sponge species based on the obtained genus data. All bacteria with the same RFLP pattern in a morphological group were assigned to the same genus as the sequenced isolate. Good's coverage and the nonparametric diversity estimator SChao1 were calculated using the algorithm by Kemp and Aller (2004) as available online (http://www.aslo.org/lomethods/free/2004/0114a.html).

Antibacterial screening

Antibacterial activity of the isolated sponge-associated bacteria was tested in a streak assay. The sponge-associated bacteria were grown at room temperature from glycerol stocks on fresh agar plates using an isolation streak to ascertain the purity of the stocks. Single colonies were inoculated onto a fresh plate as a single middle line dividing the plate into two equal-sized halves. The media used were as follows: marine agar for bacteria isolated from this medium and modified ISP2 for bacteria isolated from SCA, modified ISP2 or M1. Bacteria that previously did not grow on modified ISP2 agar were inoculated on M1. Inoculated plates were incubated at 30°C for 1–4 days until a well-grown bacteria middle line was visible and then used in the assay.

Six environmental (three of each Gram type) and four laboratory (two of each Gram type) test strains that are employed routinely in antibacterial screening assays were used to test their susceptibility towards the sponge-associated bacteria. The test bacteria were chosen from a larger pool based on their ability to grow well on marine agar, modified ISP2 and M1 plates. Isolation of the environmental bacteria has been described elsewhere (Haber et al. 2011). The environmental test bacteria consisted of the three Gram-positive test strains Kocuria sp. ESY10 (16S rRNA gene sequence Genbank accession number GU479627), Sporosacina sp. NB90 (GU479626), Lysinibacillus sp. ESY9 (GU059941), and the three Gram-negative test strains Pseudomonas sp. NB86 (GU479630), Sulfitobacter sp. ESY17 (GU479629), Vibrio sp. ESY17 (GU479628). The laboratory test bacteria were the Gram-positive Bacillus subtilis, Staphylococcus aureus and the Gram-negative Escherichia coli GM1655, Pseudomonas aeruginosa Pao1. All test bacteria were inoculated from glycerol stocks kept at −80°C on fresh plates of either marine broth for the six environmental bacteria or LB medium for the four laboratory cultures using an isolation streak to ascertain culture purity.

For the assay, 5 ml of the appropriate liquid medium was inoculated from a single colony of the test bacteria and incubated at 30°C overnight. The next day, the bacterial cultures (late log phase) were used in the streak assay. On the back of the plates with the well-grown sponge-associated bacterial middle streak, up to three perpendicular lines per half plate were marked. From the overnight liquid culture of a test bacterium, 10 μl was placed with a pipette on the plate at the middle of a marked line and streaked with an inoculation loop perpendicular towards the middle line of the already grown sponge-associated bacteria, but without reaching and touching it. The perpendicular streak was always first directed towards the line of the sponge-associated bacteria, followed by three times back and forth spreading to ensure homogenous seeding. After inoculation of all test bacteria, plates were briefly allowed to dry before being wrapped with Parafilm®. Two controls were performed for each medium, one before and one after the inoculation of plates without sponge-associated bacteria as a middle line, to detect contaminations of the test bacteria cultures and control for normal growth as a reference for the test.

Incubation of plates lasted 48 h at 30°C, and inhibition of test bacteria was noted after 24 and 48 h. Inhibitions were seen as part of the test bacteria line, where the test bacterium did not grow. Inhibitions were scored to the closest mm and ranked as -: no inhibition, +: 1–5 mm, ++: 6–10 mm, and +++: 11 and more mm of inhibition. Tests that showed activity of the sponge-associated bacteria were repeated to confirm the activity (Fig. 1a).

Figure 1.

(a) Result of a streak assay after incubation showing various degrees of inhibition against four test bacteria (two marked with white arrows). (b, c) Liquid cultures of two sponge bacteria after incubation under shaking (Sh) and standing (St) conditions for the analysis of culture condition on production of antibacterial compounds.

Influence of culture condition on production of antibacterial compounds

To assess the influence of culture condition on the production of antibacterial compounds, 22 bacteria with activity against at least two test bacteria in the streak assay were chosen. Their selection was based to maximize phylogenetic diversity and included at least one bacterium from each of the three activity profiles (inhibition of only Gram-positive, only Gram-negative, and of both Gram-types). The chosen bacteria belonged to the Actinobacteria genera: Curtobacterium (one isolate), Microbacterium (1) and Streptomyces (7); the Firmicutes genera: Bacillus (4) and Terribacillus (1); the Proteobacteria genera: Pseudovibrio (5) from its Alpha-class as well as Pseudoalteromonas (2) and Microbulbifer (1) from its Gamma-class. These bacteria were grown under three conditions: (i) on agar plates, (ii) in liquid medium with shaking at 150 rpm and (iii) in liquid medium without shaking (Fig. 1b,c). The culture medium for each bacterium was the same as used previously in the antibacterial screening. Liquid media were prepared the same way as solid media without the addition of agar. All cultures were grown in the dark at 30°C for 5 days, with the exception of the four Streptomyces isolates MVI.12, MVI.20, MVII.16 and MVII.23. These were incubated for 10 days to allow spore formation, which has been linked to the production of antibiotics in Streptomyces bacteria.

Agar plate cultures were extracted after incubation by cutting them into 0·5 cm3 pieces and submerging them for 24 h in 30 ml methanol/dichloromethane 1 : 1 under gentle shaking, followed by a second extraction with 20 ml of the same solvent mixture for 4 h. Extracts were filtered over a Whatman filter no. 595 1/2 into the same bulb and evaporated under reduced pressure and gentle heating in the water bath (<40°C) using a Rotavapor (Büchi, Flawil, Switzerland). The obtained extracts were transferred to glass vials over a cotton wool filter using the same solvent mixture. The extracts were dried as described above, weighed and stored at −20°C until use in the assay.

Stationary and shaken liquid media cultures were grown in 100-ml medium in 250-ml Erlenmeyer flasks. Their optical density at 620 nm was measured at the end of incubation to confirm growth. A sample of 10 ml of each culture was taken with a sterile 10-ml syringe and filtered into a 15-ml Falcon tube using a 0·2-μm filter. The filtrates were stored at −20°C until tested in the assays.

Antibacterial activity of extracts and filtrates was tested on marine agar plates inoculated with overnight cultures of the same test bacteria as used in the streak assay. 200 μg of agar plate extracts was transferred to blank paper discs with a 1 : 1 mixture of methanol/dichloromethane. After the solvent mixture evaporated, the paper discs were applied onto the inoculated agar plates. Filtrates of liquid cultures were thawed, and a 10 μl drop was spotted onto the plates seeded with the test bacteria. Once the filtrates were absorbed, the assay plates were covered, wrapped with Parafilm® and incubated at 25°C. Inhibition zones were noted after 24 and 48 h. A maximum of nine different extracts and filtrates were tested per plate. Negative controls of media without any bacteria were run in parallel for all culture conditions. Assays using the filtrates and extracts of these controls were always negative.


Culturable diversity of sponge-associated bacteria

Pure cultures were obtained for 259 isolates: 110, 115 and 34 from A. verrucosa, A. polypoides and A. cannabina, respectively. After RFLP analysis, the 16S rRNA gene was partially sequenced for 164 isolates: 68, 73 and 23 bacterial isolates from A. verrucosa, A. polypoides and A. cannabina, respectively. The isolates belonged to 41 genera from four phyla based on the phylogenetic analysis and comparison to validly described species (Fig. S1). Most isolates belonged to the phylum Proteobacteria followed by the phyla Firmicutes, Actinobacteria and Bacteroidetes in order of decreasing abundance (Fig. 2). At least one bacterium of each phylum was among the isolates from each of the three sponge species.

Figure 2.

Total number of isolates of the bacterial genera cultured from the three sponge species based on 16S rRNA gene sequencing of representatives of each restriction fragment length polymorphism–morphotype. (■) Axinella cannabina; (□) Axinella verrucosa; (image_n/jam12401-gra-0001.png) Axinella polypoides.

Most sequences (122 of 164, 74·4%) had ≥99% similarity to the closest validly described bacterial species, and 40 of these sequences were completely identical. Strong indication for newly cultured bacterial species (indicated by <97% sequence similarity to validly described species) was found for 18 isolates (11% of all sequenced isolates), eight from A. verrucosa and 10 from A. polypoides (Table 1). The lowest similarity for a bacterium isolated from A. cannabina was 97·04% (see Table S1–S3 for detailed results of all isolates). With similarities of <95% to validly described species, seven of these 18 isolates likely represent newly cultured genera. Four isolates of the phylum Bacteroidetes were subsequently characterized in detail, which lead to the establishment of three new genera (Haber et al. 2013a,b,c).

Table 1. EzTaxon results for isolated sponge bacteria with 16S rRNA gene sequence similarity of <97% to previously validly described species
Isolate n PhylumClosest type strainAccession numberSim. (%)
  1. V, MV: Bacteria from Axinella verrucosa; P, MP: Bacteria from Axinella polypoides; Sim.: Similarity (accessed Jan. 9th 2013); n: number of bacteria within the same RFLP pattern-morphotype group. RFLP, restriction fragment length polymorphism.

  2. None of the isolates from Axinella cannabina qualified.

VI.143BacteroidetesActibacter sediminis JC2129T EF670651 91·38
VI.181BacteroidetesAureibacter tunicatorum A5Q-118T AB572584 91·39
VIII.042BacteroidetesNamhaeicola litoreus DPG-25T JN033800 91·72
VI.012BacteroidetesTenacibaculum aiptasiae a4T EF416572 95·85
VII.071BacteroidetesTenacibaculum lutimaris TF-26T AY661691 96·42
MVII.132FirmicutesBacillus safensis FO-036bT AF234854 95·56
VI.131ProteobacteriaEndozoicomonas elysicola MKT110T AB196667 94·80
VII.051ProteobacteriaEndozoicomonas montiporae CL-33T FJ347758 96·49
MPI.09A3ActinobacteriaRhodococcus canchipurensis MBRL 353T JN164649 96·25
PIII.031BacteroidetesLutibacter aestuarii MA-My1T HM234096 91·96
PIII.021BacteroidetesPersicobacter diffluens NCIMB 1402TD1266087·60
MPII.141FirmicutesPaenibacillus agaridevorans DSM 1355T AJ345023 96·44
PIII.061ProteobacteriaAmphitrea atlantica M41T AM156910 96·77
PII.021ProteobacteriaEndozoicomonas elysicola MKT110T AB196667 96·77
PII.031ProteobacteriaEndozoicomonas elysicola MKT110T AB196667 96·79
PIII.121ProteobacteriaEndozoicomonas elysicola MKT110T AB196667 94·95
PIII.142ProteobacteriaEndozoicomonas montiporae CL-33T FJ347758 96·73
PIII.071ProteobacteriaShewanella irciniae UST040317-058T DQ180743 96·94

The most commonly cultured isolates from all three sponge species belonged to the genera Bacillus and Pseudovibrio with a total of 64 and 58 of 259 isolates, respectively. In addition, the genera Micromonospora, Mycobacterium, Ruegeria and Endozoicomonas were also isolated from all three sponge species. Twelve genera in A. polypoides, nine in A. verrucosa and two in A. cannabina were only isolated from these particular host sponges, indicating some level of sponge specificity in the genus distribution between these three sponge species. However, abundance of these genera among the isolates was low (≤5 isolates), and all Bacteroidetes genera were among these sponge species-specific genera.

Based on the biodiversity, estimate SChao1 67, 57 and 59% of the culturable bacterial diversity on the genus level have been isolated from A. polypoides, A. verrucosa and A. cannabina, respectively. Good's coverage, the estimation of the percentage of phylotypes isolated from the sample, was 90, 87 and 74% for the isolation from A. polypoides, A. verrucosa and A. cannabina, respectively, thus indicating that major culturable genera of these sponges for the used cultivation approach have been isolated.

Antibacterial activity of the sponge-associated bacteria

Only the actinobacterium MVIII.14, isolated from A. verrucosa and closely related to Sphaerisporangium rubeum, did not grow enough to perform the antibacterial screening assay, all remaining 258 isolates were tested against the 10 test bacteria. Susceptibility of test bacteria varied according to Gram-characteristic and origin (laboratory vs environmental test bacteria). Gram-positive test bacteria were more often inhibited than Gram-negative ones (16·7 vs 4·9% of the assays) and environmental bacteria were more often inhibited than laboratory bacteria (19·5% of the assays vs 12·5% in Gram-positive and 7·4 vs 1·2% in Gram-negative test bacteria, respectively) (Fig. S2). Antibacterial activity against at least one of the 10 test bacteria was found in 87 bacterial isolates (33·7% of all isolates). The active isolates belonged to 13 of the 41 isolated genera (Table 2). Bacteria with the same RFLP-morphotype did not necessarily share the same activity profiles (see Table S4 for the detailed activity profile of each isolate). Most of the active isolates were affiliated with the phylum Firmicutes (44 isolates), followed by the phyla Actinobacteria (22), Proteobacteria (20) and Bacteroidetes (1 isolate). The maximum number of inhibited test bacteria was eight, observed for two bacterial isolates (MPII.15 and MPII.N1) from the same A. polypoides specimen with a shared RFLP-morphotype and affiliated with the genus Streptomyces. Both showed exactly the same activity pattern.

Table 2. Overview of the phylogenetic affiliation of the bacteria active in the streak assay isolated from the three sponge species. Mean number of inhibited test bacteria refers to the average number of test bacteria inhibited by the active isolates of the genus. The maximal number of test bacteria inhibited by a single isolate of a given genus is given in brackets
Phylogenetic affiliation of isolatesActive/total isolatesMean number of inhibited test bacteria (maximum)Axinella verrucosa isolatesAxinella polypoides isolatesAxinella cannabina isolates
  1. a

    Does not include MVIII.14, which could not be tested.

Firmicutes44/85 21/3921/382/8
Aneurinibacillus 1/15·0 (5)1/1
Bacillus 35/643·0 (6)19/3314/242/7
Terribacillus 6/122·0 (3)0/16/11
Halobacillus 1/33·0 (3)1/20/1
Paenibacillus 1/22·0 (2)1/2
Other genera0/3 0/20/1
Actinobacteria22/53a 11/21a11/250/7
Streptomyces 19/214·7 (8)11/138/8
Curtobacterium 1/17·0 (7)1/1
Microbacterium 2/53·5 (6)2/40/1
Other genera0/26a 0/8a0/120/6
Proteobacteria20/106 10/3910/490/18
Pseudovibrio 11/582·5 (5)8/253/220/11
Ruegeria 3/81·3 (2)0/13/40/3
Microbulbifer 3/132·7 (3)2/91/4
Pseudoalteromonas 3/53·3 (5)3/5
Other genera0/22 0/40/140/4
Bacteroidetes1/14 1/100/30/1
Tenacibaculum 1/31·0 (1)1/3
Other genera0/11 0/70/30/1
All87/258a3·2 (8)43/109a42/1152/34

Five of the 13 genera with antibacterial active isolates were present among the isolates from both media types. The genera Bacillus and Pseudovibrio had active isolates from both media types. In both genera, significantly more active bacteria were isolated from antibiotic supplemented media than from marine agar (Pseudovibrio: marine agar: four active of 47 isolates; and antibiotic supplemented media: seven of 11; Bacillus: marine agar: six of 21; and antibiotic supplemented media: 31 of 46) (Fisher Exact test, 2-tailed, < 0·001 for both genera). In the three other genera (Microbacterium, Terribacillus and Halobacillus), active members were only among the isolates from the antibiotic supplemented media, whereas isolates from marine agar were inactive.

The abundance of antibacterial active bacteria varied between the three sponge species. Axinella verrucosa had a slightly higher proportion than A. polypoides with 40% of 110 isolates (13% from marine agar, 64% from antibiotic supplemented media) compared with 36% of 116 (13% from marine agar and 61% from antibiotic supplemented media) for A. polypoides. Axinella cannabina had the lowest proportion of active bacteria with only 6% of 34 isolates (0% from marine agar, 20% from antibiotic supplemented media). The antibacterial active isolates of A. polypoides belonged to ten genera, while in A. verrucosa six different genera showed antibacterial activity and both active isolates from A. cannabina belonged to the same genus (Table 2).

Influence of culture condition on production of antibacterial compounds

The 22 chosen isolates for this analysis belonged to eight genera from three phyla. All isolates grew on agar plates, but only 13 isolates grew in all three conditions. Six isolates did not grow in liquid medium (Streptomyces strains MVI.20, MVI.12; Terribacillus strain MPII.16; Pseudovibrio strains MVI.16, MPI.20, MPII.18), two isolates grew in shaken liquid cultures, but failed to grow in standing liquid cultures (Microbulbifer strain VIII.17; Streptomyces strain MPII.19), and one grew only in standing, but not in shaken liquid culture (Pseudovibrio strain MVII.11). Based on OD and visual observation, growth was generally better in shaken than in standing liquid cultures for the 13 bacteria that grew in both liquid conditions, with two exceptions (Streptomyces strain MVII.16; Bacillus strain MVIII.16).

Antibacterial activity was less often observed in the extracts and filtrates of the 22 chosen sponge-associated bacteria than in the initial streak assays of the same bacteria (Table 3). In plate culture extracts, the activity initially seen in the streak assay was confirmed in 36 of 100 cases. None of the 33 observed activities against the five Gram-negative test bacteria and only 36 of the 67 activities against Gram-positive test bacteria were confirmed. Nine additional activities, absent in the streak assay, against Gram-positive test bacteria were found upon examination of plate extracts. Culture filtrates from standing cultures inhibited the test bacteria in 14 of the 73 cases initially seen in the streak assay (two cases against Gram-negative, 12 against Gram-positive). Six cases of new activities, three against a Gram-positive and three against a Gram-negative test bacterium, were observed. In the filtrates from the shaken liquid cultures, only 9 of 76 cases (two against a Gram-positive and seven against a Gram-negative bacterium) of activity initially detected in the streak assay were confirmed. Four new activities were observed, one against a Gram-negative test bacterium and three against a Gram-positive test bacterium.

Table 3. Comparison of the activities in the filtrates and extracts of the 22 tested sponge isolates to their initial streak assay results
Culture conditionTested isolatesGram-negative test bacteriaGram-positive test bacteria
Streak assay activitiesConfirmed activitiesMissing activitiesNew activitiesStreak assay activitiesConfirmed activitiesMissing activitiesNew activities
Shaking filtrate15232211537463
Standing filtrate142322135012383
Plate extract223303306736319

The filtrates of five isolates inhibited more than one test bacterium: two Streptomyces (MVII.16, MVII.23) and a Bacillus strain (MVII.19) from A. verrucosa as well as two Pseudoalteromonas strains (PI.14, PII.10) from A. polypoides. In three isolates (MVII.16, MVII.19 and PII.10), only the filtrates of the standing cultures were active, while one isolate (MVII.23) only the shaken culture filtrate was active, and in another isolate (PI.14), both filtrates showed exactly the same activity profile. However, in the latter (PI.14), the corresponding extract from the agar plate culture was completely inactive, while the filtrates were active against seven of the ten test bacteria, three activities representing new ones, and one activity found in the streak assay was missing. The exact activity profiles of all isolates tested are given in the Table S5.


In the present study, we investigated the diversity and antibacterial activity of bacteria cultured from three sympatric Mediterranean Axinella sponges. Bacteria of the phyla Proteobacteria (from the Alpha- and Gamma- classes), Firmicutes, Actinobacteria and Bacteroidetes were isolated from all three sponge species. These four bacterial phyla represent the most frequently isolated bacteria from marine sponges (see Table 1 in Taylor et al. 2007). Previously, a member of the bacterial phylum Verrucomicrobia was isolated from Western Mediterranean A. polypoides, using the medium M1 supplemented with the same antibiotics as used here (Scheuermayer et al. 2006). However, subsequent molecular examination failed to detect members of the phylum Verrucomicrobia within the sponge material, and Scheuermayer (2006) concluded that the bacterium was likely a seawater contaminant. The absence of Verrucomicrobia representatives in sponges studied here is therefore not surprising.

While identical on the phylum level, the culturable microbial communities of the three investigated sponge species differed from one another on the genus level as 23 of the 41 isolated bacterial genera were present only in one of the three sponge species. The genera, which were limited to a single sponge species, were present in low abundance in the cultures. Their relative abundance inside the sponges remains unclear, as bacteria dominating cultures might be present only in low abundance inside the sponge. The picture of a large proportion of the microbial community being sponge species-specific or shared only between a few species is consistent with the general notion of sponge bacterial communities based on molecular investigation at a wider geographical scale (Schmitt et al. 2012; Giles et al. 2013). One factor that may explain some of the interspecies difference in the associated microbial communities cultured from these Axinella species is the presence of antimicrobial compounds. Previous studies showed that the organic extracts of A. verrucosa and A. polypoides have antibacterial activity (Amade et al. 1987; Haber et al. 2011). The antimicrobial activity in A. verrucosa was caused by bromopyrroles, a class of compounds absent in A. polypoides. The different antibacterial compounds might therefore play a role in shaping the host microbial community, even if it is not known whether they are produced by the sponge or its symbionts.

Containing 23 and 28 different culturable genera, respectively, both A. verrucosa and A. polypoides proved to be a source of diverse bacteria. Eighteen sequenced isolates obtained from these two sponges potentially belonged to newly cultured species, of which seven could be newly cultured genera and one also a potentially newly cultured family (based on similarity criteria of 97, 95, and 90%, respectively, to described bacteria, Webster et al. 2010). Four of these seven isolates have been chosen for a detailed analysis and were confirmed to belong to three new genera of the phylum Bacteroidetes (Haber et al. 2013a,b,c). The comparatively low number (14) of different genera obtained from A. cannabina might be skewed due to the availability of only a single specimen from which isolates were cultured. The observed diversity is not surprising given that sponges are host to a huge diversity of associated bacteria with as many as 3000 different operational taxonomic units on a 95% similarity level detected in a single sponge species by molecular methods (Webster et al. 2010).

The bacterial cultures of the three investigated sponge species were dominated by the Alpha-Proteobacteria genus Pseudovibrio and the Firmicutes genus Bacillus, which together accounted for 42–53% of the cultured isolates in the different sponge species. Pseudovibrio isolates dominated the Marine Agar plates, whereas the Bacillus isolates were the most abundant isolates on the other media, likely due to the selection with nalidixic acid. Members of both genera have been frequently isolated from various sponges including A. polypoides (Thiel and Imhoff 2003; Muscholl-Silberhorn et al. 2008). All Pseudovibrio isolates in the present study were closely related to the sponge isolate Pseudovibrio denitrificans. Other closely related strains have been shown to dominate the bacterial cultures from several Mediterranean, Caribbean, Great Barrier Reef and other sponges worldwide including other Axinella sponges (e.g. Webster and Hill 2001; Lafi et al. 2005; Enticknap et al. 2006; Kennedy et al. 2009; O'Halloran et al. 2011), and based on comparison of morphotypes also the bacterial cultures of Western Mediterranean A. polypoides (Muscholl-Silberhorn et al. 2008). Isolates affiliated with other genera were also closely related to bacteria previously isolated from other sponges from various geographic locations (e.g. isolates of the genera Endozoicomonas, Shewanella and Leptobacterium), though none fell into a previously recognized sponge-specific clade. Interestingly, the Endozoicomonas and Pseudovibrio isolates of A. verrucosa were most similar to isolates from the closely related Caribbean sponge Axinella corrugata, which is likewise known for the presences of bromopyrrole compounds with feeding-deterrent (Wilson et al. 1999) and antibacterial activity (Newbold et al. 1999), suggesting an adaptation of these strains to this chemical microhabitat.

The cultivation of novel and chemically prolific micro-organisms is one of the most effective methods by which natural products are discovered (Fenical and Jensen 2006; Cragg and Newman 2013). New bacteria that differ only by few bases in their 16S rRNA gene sequence to known bacteria can represent new chemotypes (Fenical and Jensen 2006; Bull and Stach 2007). Even bacteria with completely identical 16S rRNA genes can be chemically very different, especially if they were isolated from different sources as exemplified by the comparison of sponge-associated Micromonospora strains to those isolated from the surrounding sediment by Vicente et al. (2013). As the discovery of new antibacterial leads from terrestrial source is nearing a saturation curve, previously untouched habitats are now being explored (Laatsch 2006) including symbionts from sponges. With 25% of the isolates showing <99% similarity to their closest type strains and a third of all isolates inhibiting the growth of at least one test bacterium in the antagonistic streak assay, the here isolated bacteria represent a good basis for further chemical investigations. The overall proportion of antibacterial isolates is similar to that found among several Mediterranean sponges including A. polypoides from the Western Mediterranean Sea (30·5% by Muscholl-Silberhorn et al. (2008)) and considerably higher than those reported by Hentschel et al. (2001) and Chelossi et al. (2007) from other Mediterranean sponges (11·3 and 14·4%, respectively).

Several factors make a direct comparison of these proportions difficult. As our results show, the number and species of the used test strains greatly influence the number of obtained active bacteria. In the present study, environmental bacteria were more often inhibited than the frequently used laboratory strains. These results are in line with the observation by Vaara (1993) that 90% of natural antibiotics fail to inhibit E. coli and Pseudomonas aeruginosa. Other key factors are the media and methods used for isolation of the sponge-associated bacteria that greatly influenced the diversity of the cultured isolates. In previous studies (Pimentel-Elardo et al. 2009, 2010) and in the present one, Streptomyces strains were isolated from A. polypoides using media supplemented with antibiotics, whereas other studies failed to isolate such bacteria from this sponge species, even though a great variety of other media and treatments for the enrichment of spore-forming bacteria (e.g. desiccation and phenol treatment) were used (Thiel and Imhoff 2003; Muscholl-Silberhorn et al. 2008). The absence of Streptomyces isolates from the cultures obtained from A. cannabina partly explains the lower proportion of antibacterial active isolates from this sponge. Also the media, on which the sponge-associated bacteria are tested, can influence their metabolism including the production of antibacterial compounds. Here, the proportion of antibacterial active Pseudovibrio and Bacillus isolates obtained from the ISP2, M1 and SCA media was significantly higher than from Marine Agar. Whereas in Bacillus, at least in part this might be caused due to the isolation of different species, for the Pseudovibrio isolates this is not the case, because all but one isolate were more than 99·5% similar along the sequenced 16S rRNA gene part to P. denitrificans DN34T. Muscholl-Silberhorn et al. (2008) observed loss of activity in Pseudovibrio isolates after subcultivation on tryptic soy agar, while O'Halloran et al. (2011) did not observe any antibacterial activity loss after retesting Pseudovibrio isolates cultured on starch–yeast extract-peptone seawater agar. It remains to be seen whether the use of different media could maintain or induce activity, but it seems obvious that a crucial point in the design of a screening study is the timing of the assay. If isolates are directly tested after their isolation or after several transfers, the outcome of the assays might differ. Here, isolates for tests were regrown from stocks that had been subcultivated several times.

In many assays, the sponge-associated bacteria are killed before an overlay assay is performed (e.g. Muscholl-Silberhorn et al. 2008). This procedure prevents an interaction between the isolate and the test bacterium. Any possibly induced production of antibacterial active compounds is thus missed. Such microbial antagonism is still a much-neglected approach in marine natural product discovery (Burgess et al. 1999) and industrial fermentation, even though it can open the hidden potential of biosynthetic pathways not expressed under standard laboratory conditions (Bader et al. 2010). Microbial antagonism has been described from bacteria residing in various marine invertebrates including sponges (Mearns-Spragg et al. 1998; Kanagasabhapathy and Nagata 2008). Apart from other factors (e.g. change of pH, nutrients availability), an induced response could also explain the much higher frequency of antibacterial activity in the antagonistic streak assay than found in the liquid culture filtrates and especially in agar plates extract. The isolation and characterization of compounds, which induce antibacterial activity, would greatly facilitate this kind of research, as recently shown by quorum-sensing molecules that induced antibacterial compound production in a sponge-associated Pseudoalteromonas strain (Guo et al. 2011).

Incidences of activity seen in extracts from agar plates, but absence in the streak assay with the same bacteria, can be due to the presence of antibacterial compounds within the cell that are not released or remain attached to the cell wall. Such compounds might have been extracted from the cells, but would have been missed in the streak assay, as the test bacterium did not touch the sponge-associated isolate and a minimal inhibition of 1 mm was used to assess activity. Several of the 22 isolates that were grown under different culture conditions, produced their activity in the absence of test bacteria only in one set of conditions, underlining the importance of employing multiple test conditions. The restriction of the production of antibacterial active metabolites to one set of conditions has been reported from various bacteria such as Bacillus licheniformis (Yan et al. 2002) and marine Roseobacter strains (Bruhn et al. 2007), which produced their compounds only under static conditions or in the presence of a substrate they could attach to. As it is not known whether sponge-associated bacteria float in the mesohyl or are fixed in a matrix, testing different physical conditions is important to trigger otherwise silent biosynthetic pathways.

An interesting outcome of our combined approach to assess the culturable diversity and antibacterial activity of sponge-associated bacteria from these three Axinella sponges was that none of the potentially newly cultured species showed antibacterial activity in our assays. These bacteria would have been missed, if isolates had been chosen solely based on antibacterial activity, whereas closely related members of the previous antibacterial active bacteria genera isolated from Western Mediterranean A. polypoides (Bacillus, Pseudovibrio, Microbulbifer, Streptomyces) (Thiel and Imhoff 2003; Muscholl-Silberhorn et al. 2008) were also obtained here and among the antibacterial active isolates (Table 2, Fig. S1). Given that some of the potentially newly cultured bacteria might represent newly cultured families, a chemical characterization could still lead to discovery of new natural products, especially considering that they represent a novel, yet unexplored source (Jensen and Fenical 1996; Hill 2004).

In conclusion, the study results highlight the promise of sponge-associated bacteria for marine natural product research especially, the isolation of novel, antibacterial active Streptomyces isolates, including those that produce their active compounds under different conditions, is encouraging in this respect, as it is the most important bacterial genus for antibiotic production (Bérdy 2005). Future steps should include the chemical characterization of the antibacterial active compounds, preferably after dereplication using databases (e.g. based on HPLC-MS data). This procedure will facilitate exploiting the biotechnological potential of the here isolated sponge-associated bacteria as well as a chemical analysis of the novel bacteria. Given the constant need for new bioactive compounds due to the acquired resistance of pathogens, new culture approaches that increase the obtained bacterial diversity should be considered, to tap into the pharmacological potential of such sponge-associated bacteria.


We thank Dr. Sigal Shefer for technical assistance, Adi Lavik and Ray Keren for discussions and critical reading of the manuscript. This study was partially supported by a Marie-Curie fellowship to M. H. as part of the Marie-Curie Research Training network awarded to M. I. (339 FP6, BIOCAPITAL; contract no. MRTN-CT-2004-512301).

Conflict of interest

No conflict of interest is declared.