Diversity of Mycobacterium species from marine sponges and their sensitivity to antagonism by sponge-derived rifamycin-synthesizing actinobacterium in the genus Salinispora


  • Present address: Marie E.A. Gauthier, Institute of Evolutionary Biology and Environmental Studies, University of Zurich, Zurich, Switzerland.

  • Editor: Jan-Ulrich Kreft

Correspondence: John A. Fuerst, School of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane, Qld 4072, Australia. Tel.: +61 7 3365 4643; fax: +61 7 3365 4273; e-mail: j.fuerst@uq.edu.au


Eleven isolates of Mycobacterium species as well as an antimycobacterial Salinispora arenicola strain were cultured from the sponge Amphimedon queenslandica. The 16S rRNA, rpoB, and hsp65 genes from these Mycobacterium isolates were sequenced, and phylogenetic analysis of a concatenated alignment showed the formation of a large clade with Mycobacterium poriferae isolated previously from another sponge species. The separation of these Mycobacterium isolates into three species-level groups was evident from sequence similarity and phylogenetic analyses. In addition, an isolate that is phylogenetically related to Mycobacterium tuberculosis was recovered from the sponge Fascaplysinopsis sp. Several different mycobacteria thus appear to co-occur in the same sponge. An actinobacterium closely related to S. arenicola, a known producer of the antimycobacterial rifamycins, was coisolated from the same A. queenslandica specimen from which mycobacteria had been isolated. This Salinispora isolate was confirmed to synthesize rifamycin and displayed inhibitory effects against representatives from two of three Mycobacterium phylotype groups. Evidence for antagonism of sponge-derived Salinispora against sponge-derived Mycobacterium strains from the same sponge specimen and the production of antimycobacterial antibiotics by this Salinispora strain suggest that the synthesis of such antibiotics may have functions in competition between sponge microbial community members.


Marine sponges are known to harbor diverse associated bacteria within their tissues, including actinobacteria (Webster et al., 2001; Sun et al., 2010), some of which may synthesize bioactive compounds including antibiotics and cytotoxic compounds (Kim et al., 2006; Izumikawa et al., 2010). Members of genus Salinispora are known to synthesize rifamycins (Kim et al., 2006), compounds with known antibiotic activity against Mycobacterium species such as Mycobacterium tuberculosis, against which rifamycin class compound rifampicin is used as a clinical antibiotic (Aristoff et al., 2010). Salinispora species have been isolated from marine sediments and also from marine sponges (Mincer et al., 2002; Kim et al., 2005; Sun et al., 2010) and are known to synthesize a wide range of bioactive compounds (Fenical & Jensen, 2006). Considering the occurrence of the antimycobacterial organism Salinispora in marine sponges, the question arises as to whether any selective pressure for the evolution of its antimycobacterial compounds has acted – for example a competitive advantage in an environment in which mycobacteria co-occur and even compete for similar resources. Such a habitat might be found in marine sponges. For example, a novel Mycobacterium species, Mycobacterium poriferae, has been isolated from the sponge Halichondria bowerbanki (Padgitt & Moshier, 1987), and both Mycobacterium and Salinispora species have been isolated from the sponge Hymeniacidon perleve (Sun et al., 2010). It is hypothesized here that such organisms in the sponge microbial community might be in active competition where the production of antibiotics and the genes needed for their synthesis in producers are positively selected, as are resistance genes in bacteria targeted by such compounds.

In relation to these questions, we isolated several Mycobacterium species from a specimen of the Great Barrier Reef (GBR) sponge Amphimedon queenslandica, and these were characterized by sequencing of genes encoding for 16S rRNA, the β-subunit of RNA polymerase (rpoB), and 65-kDa heat shock protein (hsp65). We examined their co-occurrence with Salinispora arenicola capable of synthesizing antimycobacterial compounds and their sensitivity to antagonism by the sponge-derived S. arenicola. Furthermore, polyketide synthase (PKS) genes of the sponge-derived mycobacteria were examined because polyketides are known to include antibiotics (Walsh, 2004) and PKS genes can catalyze the synthesis of mycobacterial outer membrane lipids that are relevant to intracellular host cell infection in pathogenic mycobacteria (Onwueme et al., 2005; Chopra & Gokhale, 2009).

Materials and methods

Sponge collection

A specimen of the sponge A. queenslandica, living on shallow intertidal reef flat, was collected at Shark Bay, Heron Island, at coordinates 23°27′S, 151°5′E in October 2008. It was transported in seawater to The University of Queensland, Brisbane, and maintained in a recirculating aquaculture system at The Center for Marine Studies for 5 days before microbiological processing. A specimen of Fascaplysinopsis (Queensland Museum species no. 3) was collected at a depth of 26 m at coordinates 15°49′S, 145°38′E, 56 km southeast of Cooktown, in January 2003. This specimen was frozen and stored at −20 °C.

Bacterial isolation

Approximately 1 cm3 of sponge tissue was excised from the middle of the entire sponge and washed three times with sterile artificial seawater (ASW) (Aquasonic, NSW, Australia). A sponge homogenate was prepared by cutting sponge tissue into small pieces and homogenizing with 5 mL of ASW using a sterile mortar and pestle. Fast-growing mycobacteria were isolated from A. queenslandica, after 100 μL of each of a twofold dilution series of the sponge homogenate (up to 1/16 dilution) was inoculated onto a glycerol–asparagine medium consisting of 10 mL of glycerol, 1.0 g of l-asparagine, 1.0 g of K2HPO4, 16.6 g of artificial sea salts, 9.0 g of Davis agar, and 1000 mL of distilled water, followed by incubation at 28 °C. This medium was supplemented with 50 μg mL−1 of cycloheximide and 20 μg mL−1 of nalidixic acid to inhibit the growth of fungi and fast-growing bacteria. Colonies appearing after 3 weeks were picked for single-colony purification. Salinispora strains were isolated from homogenates of A. queenslandica using starch–yeast extract–peptone (SYP) medium and the method described by Kim et al. (2005), with the exception that the pH of the medium was not adjusted. Some of the Mycobacterium strains were also isolated using this method.

A slow-growing Mycobacterium species was isolated from Fascaplysinopsis sp. using a low-nutrient broth enrichment medium consisting of 0.001% peptone, 0.001% yeast extract, 0.001%d-glucose, 20 mL of Hutner's basal salts (Schlesner, 1994), 10 mL of vitamin solution no. 6 (Schlesner, 1994), 5 mL of 0.1 M Tris/HCl buffer (pH 7.5), and 1000 mL of ASW. Fifty milliliters of the low-nutrient broth enrichment medium supplemented with 50 μg mL−1 of cycloheximide and 500 μg mL−1 of ampicillin in a 250-mL Erlenmeyer flask was inoculated with 1 mL of homogenate of Fascaplysinopsis sp. and incubated on a rotary shaker (260 r.p.m.) at 28 °C in the dark for 1 month. Subsequently, a portion of this broth enrichment was plated on 1/10 strength Marine 2216 medium supplemented with 50 μg mL−1 of cycloheximide and 200 μg mL−1 of ampicillin. Colonies appearing after 2 months were purified using the single-colony subculture technique on the same medium.

Sequencing of 16S rRNA, hsp65, rpoB, and ketosynthase genes

Genomic DNA was extracted from the isolates using a Wizard® Genomic DNA Purification Kit (Promega, WI) with the recommended protocol for Gram-positive bacteria. The 16S rRNA gene sequence was amplified with the 27f (AGAGTTTGATCMTGGCTCAG) and 1492r (TACGGYTACCTTGTTACGACTT) primer set. In addition to the 16S rRNA gene, rpoB and hsp65 genes were analyzed to determine their evolutionary relationship within the genus Mycobacterium. Amplification of the hsp65 gene was performed with the primers Tb11 (ACCAACGATGGTGTGTCCAT) and Tb12 (CTTGTCGAACCGCATACCCT) (Telenti et al., 1993) and amplification of the rpoB gene was performed with the primers MF (CGACCACTTCGGCAACCG) and TBBrpoB2 (TACGGCGTCTCGATGAASCC) (Devulder et al., 2005). Amplified gene products were cleaned using the Wizard® SV Gel and PCR Clean-Up System (Promega). Nucleotide sequence analysis of the purified PCR products was performed at the Australian Genome Research Facility using an AB3730xl DNA analyzer (Applied Biosystems, CA). For four isolates representing each Mycobacterium phylotype, the β-ketosynthase (KS) domain of type I PKS was retrieved via PCR with the degenerate primer set degKS2F.i (GCIATGGAYCCICARCARMGIVT) and degKS5R.i (GTICCIGTICCRTGISCYTCIAC) under the conditions described by Schirmer et al. (2005). The amplified products were visually assessed by gel electrophoresis, and amplicons of the correct size (700 bp) were cleaned and cloned into pGEM-T Easy Vector (Promega) following the manufacturer's instructions. Nucleotide sequencing was performed with the primers T7 (TAATACGACTCACTATAGGG) and SP6 (ATTTAGGTGACACTATAG) after purification of the plasmid using the Wizard®Plus SV Minipreps DNA Purification System (Promega). Nucleotide sequences were deposited in the GenBank database under accession numbers HM210415–HM210460.

Phylogenetic analysis

The sequences of the 16S rRNA gene from isolates were aligned against the reference sequences retrieved from The Ribosomal Database Project (Cole et al., 2009), using the greengenes program (DeSantis et al., 2006), followed by Lane masking to remove any hypervariable region from the alignment (Lane, 1991). The dataset was exported into the phylip program (Felsenstein, 1989) for sequence similarity analysis.

For the phylogenetic analysis based on the concatenation of the three genes, reference sequences were obtained from the NCBI database associated with the study of Mignard & Flandrois (2008). Sequence alignment for each gene was performed using clustal x (Larkin et al., 2007). Aligned sequences for the three genes were concatenated and aligned again as single sequences. Phylogenetic trees were generated using the mega4.1 program (Tamura et al., 2007) for the neighbor-joining and maximum parsimony methods and the treefinder program (Jobb et al., 2004) for the maximum likelihood method with the HKY model of substitution. Bootstrapping was performed using 1000 replicates.

For the KS gene, translated protein sequences were derived from nucleotide sequences using the orf finder available at the NCBI website (http://www.ncbi.nlm.nih.gov/projects/gorf/). Phylogenetic trees were reconstructed from a clustal x alignment of translated KS protein sequences including the reference sequences obtained from the NCBI-available genome annotations using the mega4.1 and treefinder programs with the JTT model of substitution for the maximum likelihood calculation.

Antagonism determination assay

The Salinispora isolate AQ1M05 was heavily inoculated on one quarter segment of an SYP agar plate and grown for 3 weeks at 28 °C. After confirming the confluent growth of the Salinispora isolate on the plate, three strains of fast-growing Mycobacterium species representing each phylotype (AQ1GA1, AQ4GA8, and AQ4GA9) were inoculated on the same plate by streaking a line from the vicinity of Salinispora colonies to the opposite edge to the Salinispora growth. The plates were incubated at 28 °C for 10 days, and the inhibitory effects of the Salinispora isolate on the growth of Mycobacterium isolates were determined.

Determination of rifamycin production by LC–MS/MS

The production of rifamycins by Salinispora isolate AQ1M05 was determined using the LC–MS/MS method described by Hewavitharana et al. (2007) with the following modification of the extraction method: AQ1M05 was grown in 50 mL of SYP medium at 28 °C for 3 weeks. Five milliliters of the broth culture was withdrawn, and the pH was adjusted to 2.0 with concentrated HCl. After the removal of the cell material by centrifugation, an equal volume of ethyl acetate was added, and the mixture was incubated on a rotary shaker for 1 h. The resulting ethyl acetate layer was removed and evaporated to dryness under vacuum. Subsequently, the extract residue was reconstituted in 20% v/v methanol/water and frozen at −20 °C until LC–MS/MS analysis. The frozen extract was thawed and filtered through 0.2-μm filters before LC–MS/MS analysis.

Results and discussion

Isolation and phylogenetic analysis

On the basis of the 16S rRNA gene sequences, 11 isolates belonging to the genus Mycobacterium were recovered from a specimen of A. queenslandica. Phylogenetic analysis based on the 16S rRNA gene showed that four isolates – AQ4GA8, AQ1M16, AQ1M04, and AQ11356 – were identical to M. poriferae, a species isolated previously from a North Atlantic sponge, based on a 100% similarity value (Padgitt & Moshier, 1987). The remaining isolates – AQ1GA1, AQ1M06, AQ1GA3, AQ1GA4, AQ4GA9, AQ1GA10, and AQ4GA22 – from A. queenslandica were also most closely related to M. poriferae, having similarity values between 99.0% and 99.3% to M. poriferae. Because the Amphimedon specimen had developed a few spots of tissue necrosis after transfer into an aquaculture environment, we hypothesized that the presence of mycobacteria might be a result of primary or secondary infection. However, mycobacteria could be isolated only from healthy tissue, but not from the affected tissue or aquaculture water. It is estimated that mycobacteria comprised c. 2400 CFU g−1 of A. queenslandica healthy sponge tissue. In addition, an isolate FSD4b-SM that is closely related to M. tuberculosis based on a 16S rRNA gene similarity value of 99.6% was recovered from Fascaplysinopsis sp. after 2 months of incubation on isolation plates following 1 month of enrichment in an ampicillin-containing broth.

Because the interspecies similarity of the 16S rRNA gene is relatively high within the genus Mycobacterium (Devulder et al., 2005), two additional conserved genes, rpoB and hsp65, were analyzed. Based on rpoB and hsp65 gene sequences, the M. poriferae-related isolates can be divided into three groups. Group I includes isolates AQ4GA8, AQ1M16, AQ1M04, and AQ11356, which have similarity values to the most closely related species M. poriferae of above 98.5% for rpoB and 99.5% for hsp65 genes. Group II consists of isolates AQ1GA1 and AQ1M06, which have similarity values to rpoB of Mycobacterium brumae of 95.1% and to hsp65 of Mycobacterium rutilum and Mycobacterium novocastrense of 92.5%. Group III consists of isolates AQ1GA3, AQ1GA4, AQ4GA9, AQ1GA10, and AQ4GA22, which have similarity values of 95.1% to rpoB of M. poriferae and Mycobacterium goodii and 95.8% to hsp65 of the isolates from Group I, a group closely related to M. poriferae. For the hsp65 gene, the sequence similarity value of 97% has been proposed as a baseline for Mycobacterium species identification (McNabb et al., 2004). Based on the hsp65 gene alone, the sequence similarity between any isolate from Group II or Group III to any of the reference Mycobacterium species in the NCBI database is below 97%, suggesting that they could be considered to be unique mycobacteria, possibly comprising novel organisms at the species level.

Phylogenetic trees of a concatenated alignment of the three genes showed that isolates from A. queenslandica formed a large clade with M. poriferae with a significant bootstrap confidence, suggesting that these isolates may represent a sponge-specific phylotype (Fig. 1). Within this M. poriferae clade, they formed three individual clusters (Groups I, II, and III), suggesting the separation of these isolates into three species-level groups, a separation consistent with sequence similarity analysis. One of these clusters, Group I, contains M. poriferae itself and the M. poriferae-like strains of our isolates. Surprisingly, an isolate (FSD4b-SM) apparently closely related to the M. tuberculosis complex was recovered from another GBR sponge, Fascaplysinopsis sp. This isolate has similarity values of 91.3% to the rpoB gene of Mycobacterium bovis, Mycobacterium africanum, and Mycobacterium parmense and 93.1% to the hsp65 gene of M. parmense. Phylogenetic trees showed a close association of the strain FSD4b-SM with the M. tuberculosis complex, forming a cluster with significant bootstrap values.

Figure 1.

 Phylogenetic tree of representatives of the genus Mycobacterium and of sponge isolates computed from the concatenated alignment of 16S rRNA gene, hsp65, and rpoB nucleotide sequences using the neighbor-joining method. Bold refers to strains isolated from either the sponge Amphimedon queenslandica (AQ) or the sponge Fascaplysinopsis sp. (FSD) in this study. Nocardia abscessus DSM 44432 was chosen as the outgroup and used to root the tree. Significant bootstrap values calculated using the neighbor-joining, maximum parsimony, and maximum likelihood methods are indicated at each node in order from right to left.

Antagonistic effects of S. arenicola strain against Mycobacterium isolates

The strain of antimycobacterial Salinispora (AQ1M05) was isolated from the same specimen of A. queenslandica that yielded the mycobacteria strains. The 16S rRNA gene sequence of AQ1M05 shares 100% similarity to that of the S. arenicola type strain CNH643, and phylogenetic analysis of 16S rRNA gene demonstrated that this strain belongs to the species S. arenicola (data not shown). This S. arenicola strain was confirmed to produce rifamycin B and an additional probable rifamycin-like compound by LC–MS/MS analysis (Fig. 2). The antagonistic effect of the S. arenicola strain AQ1M05 was therefore evaluated against the representatives of each of the three Mycobacterium phylotypes (AQ1GA1, AQ4GA8, and AQ1GA9). The S. arenicola strain AQ1M05 produced antagonistic effects indicated by a growth inhibition zone against the Mycobacterium isolates AQ1GA1 and AQ4GA9, but not against the M. poriferae-like strain AQ4GA8 (Fig. 3).

Figure 2.

 LC–MS/MS chromatograms of (a) rifamycin B standard and (b) Salinispora arenicola strain AQ1M05 extract. A peak at 2.75 min indicates the presence of rifamycin B, and an additional peak at 2.11 min suggests the presence of another probable rifamycin class compound.

Figure 3.

 Photograph of Mycobacterium isolates AQ1GA1 (bottom left), AQ4GA8 (bottom right), and AQ4GA9 (bottom center) growing on the same plate where the Salinispora arenicola isolate AQ1M05 had established mature confluent colony growth (top) before the inoculation of the Mycobacterium isolates. Zones of inhibition are indicated by red arrows.

One hypothesis implied by these results could be that such antibiotics may function in competitive interactions between Salinispora and mycobacterial members of the sponge microbial community. The apparent resistance of one M. poriferae-like strain to antimicrobials produced by the S. arenicola strain might be consistent with a scenario in which an M. poriferae-like mycobacterium developed resistance to the rifamycin antibiotics of a co-occurring actinobacterium within the sponge microbial community. However, such a hypothesis would need to be tested by comparative phylogenetics of antibiotic synthesis genes and antibiotic resistance genes in the proposed interacting partners.

PKSs of sponge-associated mycobacteria

Phylogenetic analysis of KS genes of the isolates identified within the M. poriferae clade (AQ1GA1, AQ1GA3, and AQ4GA8) revealed the presence of KS domains similar to those of phenolpthiocerol synthesis type I PKSs (PpsC and PpsB) known to occur in pathogenic Mycobacterium species (Chopra & Gokhale, 2009). However, the KS genes of M. poriferae clade members isolated here are more closely related to those of environmental mycobacteria, such as Mycobacterium gilvum and Mycobacterium vanbaalenii, than to those of pathogenic mycobacteria (Fig. 4). Pps-family enzymes are involved in the biosynthesis of outer membrane lipids known as dimycocerosate esters, which are virulence factors for clinically relevant mycobacteria to facilitate replication in the host cell environment (Onwueme et al., 2005). The functions of these pps gene homologues found in genomes of environmental mycobacteria including sponge-associated mycobacteria remain unknown. The analysis of outer membrane lipids of sponge-associated mycobacteria might provide an insight into the mechanisms of their survival within the sponge environment.

Figure 4.

 Neighbor-joining tree generated from KS amino acid sequences of sponge-derived Mycobacterium isolates and reference sequences from annotated genomes of Mycobacterium tuberculosis H37Rv, Mycobacterium marinum M, Mycobacterium vanbaalenii PYR-1, Mycobacterium gilvum PYR-GCK, and Mycobacterium avium 104. Bold indicates either mycobacterial strain and KS sequence from this study or mycobacterial PKS gene which has been confidently annotated. Pps=phenolpthiocerol synthase; Mas=mycocerosic acid synthase; Pks=polyketide synthase; Fas=fatty acid synthase. Significant bootstrap values calculated using the neighbor-joining, maximum parsimony, and maximum likelihood methods are indicated at each node in order from right to left. Mycobacterium tuberculosis Fas was used as the outgroup.

In contrast, KS genes of the M. tuberculosis-related isolate (FSD4b-SM) showed characteristics distinct from that of M. poriferae clade members, displaying no clear homology to PKSs of any Mycobacterium species. blast analysis showed that one of the KS sequences of this isolate was more closely related to those of bioactive compound producers such as Sorangium cellulosum and Amycolatopsis orientalis than those of Mycobacterium species. PKS genes that are more closely related to those of Streptomyces than to other mycobacterial PKSs are also found in the genome of Mycobacterium marinum (Stinear et al., 2008). Genome comparison of Mycobacterium species showed that the genome of M. tuberculosis has undergone downsizing events during the process of becoming a specialized human pathogen in contrast to M. marinum, which has retained adaptations to its environmental niches (Stinear et al., 2008). The presence of unique PKS genes in the M. tuberculosis-related isolate might suggest that this species is adapted to survival in marine microbial communities rather than being a specialized pathogen. This may have occurred via retention of genes that may be involved in survival in the marine environment. It has been proposed that M. tuberculosis evolved from an environmental progenitor by horizontal gene transfer (Rosas-Magallanes et al., 2006). A genome project for this sponge-derived M. tuberculosis-related species might conceivably provide an insight into the evolution of M. tuberculosis.


We have isolated several different types of mycobacteria including a strain closely related to the M. tuberculosis complex from marine sponges, illustrating their diversity and sponge specificity. The coisolation of the antimycobacterial actinobacterium S. arenicola with mycobacteria from the same specimen of A. queenslandica and demonstration of antagonism by this Salinispora against the sponge mycobacteria suggest that the proposed relationship might be applied as a model to study the microbial interactions within the sponge environment.


Research on sponge-associated bacteria in the laboratory of J.A.F. is funded by an Australian Research Council (ARC) Linkage project. Research on A. queenslandica in the laboratory of B.M.D. is supported by grants from ARC. This paper is an output from the Great Barrier Reef Seabed Biodiversity Project, a collaboration between the Australian Institute of Marine Science (AIMS), the Commonwealth Scientific and Industrial Research Organization (CSIRO), Queensland Department of Primary Industries & Fisheries (QDPIF), and the Queensland Museum (QM); funded by the CRC Reef Research Centre, the Fisheries Research and Development Corporation, and the National Oceans Office; and led by R. Pitcher (Principal Investigator, CSIRO), P. Doherty (AIMS), J. Hooper (QM), and N. Gribble (QDPIF). We also wish to thank the crew of the FRV Gwendoline May (QDPIF) and RV Lady Basten (AIMS). H.I. was supported by the University of Queensland Research Scholarship (UQRS) and University of Queensland International Research Tuition Award (UQIRTA).