Spatial distribution of sponge-associated bacteria in the Mediterranean sponge Tethya aurantium

Authors


  • Editor: Patricia Sobecky

Correspondence: Johannes F. Imhoff, Leibniz-Institut für Meereswissenschaften, IFM-GEOMAR, Düsternbrooker Weg 20, D-24105 Kiel, Germany. Tel.: +49 (0) 431 6004450; fax: +49 (0) 431 6004452; e-mail: jimhoff@ifm-geomar.de

Abstract

The local distribution of the bacterial community associated with the marine sponge Tethya aurantium Pallas 1766 was studied. Distinct bacterial communities were found to inhabit the endosome and cortex. Clear differences in the associated bacterial populations were demonstrated by denaturing gradient gel electrophoresis (DGGE) and analysis of 16S rRNA gene clone libraries. Specifically associated phylotypes were identified for both regions: a new phylotype of Flexibacteria was recovered only from the sponge cortex, while Synechococcus species were present mainly in the sponge endosome. Light conduction via radiate spicule bundles conceivably facilitates the unusual association of Cyanobacteria with the sponge endosome. Furthermore, a new monophyletic cluster of sponge-derived 16S rRNA gene sequences related to the Betaproteobacteria was identified using analysis of 16S rRNA gene clone libraries. Members of this cluster were specifically associated with both cortex and endosome of T. aurantium.

Introduction

The phylum Porifera contains an estimated 15 000 species in three taxonomic classes: Calcarea (calcareous sponges), Hexactinellida (glass sponges) and Demospongiae (demosponges) (Hooper & van Soest, 2002). As sessile filter-feeding organisms sponges pump large amounts of water through their aquiferous channel system. They take up bacteria, single-celled algae and other food particles from the filtered water by phagocytosis within the choanocyte chambers, which are located within the inner part of the sponge, the endosome (=choanosome). The endosome is protected against strong currents and high light intensities by an outer region, the cortex (or ectosome) (Sará, 1987); this has also been used as a basis for taxonomic classification. As a protective device (Burton, 1928), the cortex is found to be particularly thick and well structured in species living in shallow waters subject to strong currents and high light intensities. By contrast, species living in more protected habitats have a thin, almost indistinct cortex (Sará, 1987).

Tethya aurantium (Fig. 1a) is characterized by a globular shape and a thick and well-developed cortex, clearly differentiated from the endosome by texture and colour of the tissue (Fig. 1b). In the Mediterranean Sea T. aurantium cooccurs with the very similar species of Tethya citrina, but inhabits different niches and can be distinguished by the development of its cortex. Tethya aurantium generally inhabits areas that are more exposed to light and current in depths of 1–40 m, while T. citrina prefers more sheltered places and possesses a thinner cortex than T. aurantium (Sará, 1987).

Figure 1.

 (a) Photograph of the Mediterranean sponge Tethya aurantium Pallas in situ. A cross section (b) shows the morphologically different regions endosome (en) and cortex (co). Electron microscopic photography shows no apparent accumulation of bacteria in the cortex (c) and a moderate number of different bacterial morphotypes within the endosome (d and e).

Associations between microorganisms and sponges have been systematically studied using microscopy and isolation methods since the 1970s (Vacelet, 1970, 1971, 1975; Vacelet & Donadey, 1977; Manz et al., 2000; Webster & Hill, 2001; Lafi et al., 2005). These studies have shown that bacteria are abundant within the mesohyl of sponges and can form up to 40% of the sponge volume (Wilkinson, 1978). More recent studies on sponge–microbe associations were based mainly on culture-independent molecular methods (Hinde et al., 1994; Althoff et al., 1998; Burja et al., 1999; Burja & Hill, 2001; Webster & Hill, 2001; Webster et al., 2001). Comparison of 16S rRNA gene clone libraries obtained from several sponges of different geographical origin have revealed unexpected conformity between the different sponge species, and a uniform sponge-associated bacterial community was proposed (Hentschel et al., 2002).

The aim of this study was to characterize the microbial community associated with T. aurantium. We demonstrate specific differences between bacteria associated with the cortex and endosome. Furthermore, we report on a new phylogenetic cluster of sponge-associated Betaproteobacteria and a possible association of T. aurantium with Cyanobacteria.

Materials and methods

Sampling sites

The Limski kanal is a semiclosed fjord-like bay in the Adriatic Sea near Rovinj (Istrian Peninsula, Croatia) (45°7, 972′N; 13°43, 734′E). It extends along an east–west axis, with an c. length of 11 km, a maximum width of about 650 m and a depth of up to 32 m. The bottom is generally muddy, with insular sites consisting of stones and shallow rocky slopes along the sides. It is characterized by a very high sedimentation rate and rapid water exchange (Kuzmanovic, 1985). In a recent study, 42 sponge species were found in the Limski kanal including three species of the genus Tethya: T. aurantium Pallas 1766, Tethya limski Müller & Zahn 1969 and T. citrina Sará & Melone 1965 (Brümmer et al., 2004).

Sponge sampling

Specimens of T. aurantium were collected by SCUBA diving from a depth of 5–15 m in April 2003, June 2004 and May 2005. The sponges were placed into sterile plastic bags, cooled in an isolation box, immediately transported to the laboratory and processed within 3 h. In the laboratory the sponges were washed carefully three times in filter-sterilized seawater (0.2 μm) prior to cutting. The sponges were separated into cortex and endosome sections and washed again separately in sterile seawater. Tissues were cut into small pieces of c. 20–30 mg each, frozen in liquid nitrogen and kept frozen (−80°C) until further investigation.

Ambient seawater was collected into sterile glass bottles (Duran; 1 L) prior to sponge sampling. The water was cooled on the way back to the laboratory and immediately filtered through a cellulose-acetate filter (0.2 μm pore size; Sartorius). The filters were placed in a cryovial, frozen in liquid nitrogen and stored at −80°C until further investigation.

Electron microscopy

Sponge samples were prepared for scanning electron microscopy by fixation with 1% glutaraldehyde in seawater, replacement of water with an ethanol series and subsequent critical-point drying. After mounting, samples were sputtered with Au/Pd and observed with a Zeiss DSM 940 scanning electron microscope.

DNA extraction

Genomic DNA was extracted and purified using the QIAGEN DNeasy® Tissue Kit following the manufacturer's protocol for Gram-positive bacteria and animal tissue.

PCR and cloning procedure

Amplification of ribosomal DNA was performed using puReTaq Ready-To-Go PCR Beads (Amersham Biosciences). For amplification of the nearly complete 16S rRNA gene the eubacterial primers 27f and 1492r (Lane, 1991) were used. The conditions for this PCR were: initial denaturation (2 min at 94°C) followed by 30 cycles of primer annealing (40 s at 50°C), primer extension (90 s at 72°C) and denaturation (40 s at 94°C), a final primer annealing (1 min at 42°C) and a final extension phase (5 min at 72°C). PCR products were checked for correct length on a 1% Tris-borate-EDTA (TBE) agarose gel (1% agarose, 8.9 mM Tris, 8.9 mM borate, 0.2 mM EDTA), stained with ethidium bromide and visualized under UV illumination.

DNA was purified using the High Pure PCR Product Purification Kit (Roche) prior to ligation into pCR®4-TOPO® vector and transformation into One Shot® Competent Escherichia coli cells using the TOPO TA® Cloning Kit (Invitrogen). Inserts were amplified as described above using the M13f/M13r primer set (M13f: 5′-GTAAAACGACGGCCAG-3′; M13r: 5′-CAGGAAACAGCTATGAC-3′) (0.1 μM each). Correct insert size was verified using agarose gel electrophoresis.

PCR for denaturing gradient gel electrophoresis (DGGE) was performed using the primers 342-GCf and 534r (Muyzer et al., 1993). The temperature profile was as follows: initial denaturation (2 min at 94°C) followed by 15 cycles of touchdown primer annealing (40 s at 65–50°C), primer extension (1 min at 72°C) and denaturation (40 s at 94°C), an additional 20 cycles of primer annealing (40 s at 50°C), primer extension (1 min at 72°C) and denaturation (40 s at 94°C), and a final primer annealing (1 min at 42°C) with a final extension phase (5 min at 72°C). PCR products were checked for correct length on a 2% TBE-agarose gel. Excised DGGE bands (see below) were reamplified using primers GC/M (5′-GGGGGCAGGGGGGC-3′) and 534r (Muyzer et al., 1993) as described above with an annealing temperature of 50°C and for 30 cycles.

Double gradient DGGE

DGGE was conducted in a double gradient gel (Petri & Imhoff, 2001), containing a linear 6–8% polyacrylamide (acrylamide : bisacrylamide ratio 37.5 : 1) and a 50–80% denaturing gradient (100% corresponds to 7 M urea/40% deionised formamide). The gel was run in a Tris-EDTA-acetic acid (TAE) buffer (10 mM Tris, 5 mM acetic acid, 5 mM EDTA, titrated to pH 7.5) at a voltage of 80 V for 15 h. The gel was stained in 1 × SYBR Gold (Invitrogen) (Tuma et al., 1999) in TAE and documented digitally.

Bands exclusively present in sponge samples were cut out and DNA was extracted for sequencing. Excised bands were transferred into 50 μL of molecular grade water, crushed with sterile pistils and incubated overnight at 4°C. The supernatant (1 μL) was used as template for reamplification with primers GC/M and 534r as described above and sequenced.

Cluster analysis

For statistical comparison of the DGGE banding patterns, similarity cluster analysis (Clarke & Warwick, 1994) and analysis of similarity (ANOSIM) (Clarke, 1993) were performed using the program primer 5 v5.2.2 (PRIMER–E Ltd). Similarity was calculated using the Bray–Curtis index and cluster analysis was conducted with complete linkage. Subsequently, one-way ANOSIM with all possible permutations was performed. In accordance with the PRIMER manual (Clarke & Gorley, 2001) ANOSIM R-values of >0.75 were interpreted as well separated, R>0.5 as overlapping, but clearly different, and R<0.25 as barely separable at all. Replicate samples were grouped according to source (e.g. endosome, cortex, seawater) as factors for the analyses.

Sequencing and phylogenetic analysis

After verification of correct insert size, clones (for each sample 29–41) were sequenced using the ABI PRISM® BigDye Terminator Ready Reaction Kit (Applied Biosystems) and an ABI PRISM® 310 Genetic Analyser (Perkin Elmer Applied Biosystems). Sequence primers used were: plasmid primers M13f and M13r as well as the 16S rRNA gene-specific primers 342f (Muyzer et al., 1993), 534r (Muyzer et al., 1993), 790f (5′-GATACCCTGGTAGTCC-3′), 907f (5′-GGCAAACTCAAAGGAATTGAC-3′), 1093f (5′-TCCCGCAACGAGCGCAACCC-3′) and 1093r (5′-GGGTTGCGCTCGTTGCGGGA-3′). Sequence data were edited with Lasergene Software SeqMan (DNAStar Inc.) and checked for possible chimeric origin using the program check_chimera (http://35.8.164.52/html/index.html) of the Ribosomal Database Project (http://rdp.cme.msu.edu/index.jsp) (Maidak et al., 1999). Putative chimeric sequences were removed from phylogenetic analyses. Next relatives were determined by comparison to 16S rRNA genes in the NCBI GenBank database using blast (Basic Local Alignment Search Tool) searches (Altschul et al., 1990) and the RDPII Sequence Match Program (http://rdp.cme.msu.edu/seqmatch/seqmatch_intro.jsp). Sequences were aligned using the FastAlign function of the alignment editor implemented in the ARB software package (http://www.arb-home.de) (Ludwig et al., 2004) and refined manually employing secondary structure information. For phylogenetic calculations the PhyML software (Guindon & Gascuel, 2003) as well as the online version of PhyML (Guindon et al., 2005) were used. Trees were calculated by the maximum-likelihood (ML) method (Felsenstein, 1981) using the GTR model and estimated proportion of invariable sites as well as the Gamma distribution parameter with near full-length sequences (≥1200 and ≥1000 bp for trees in Fig. 4b and c, respectively). Calculated trees were imported into ARB and short sequences were subsequently added by use of the ARB parsimony method without changing the tree topologies. Phylogenetic positions of short sequences (<1200 bp/<1000 bp) were additionally verified by phylogenetic analysis (ML, 100 bootstraps) of full and partial sequences.

Figure 4.

Figure 4.

 Phylogenetic trees constructed from 16S rRNA gene sequences related to the Betaproteobacteria (a), Cyanobacteria (b) and Bacteroidetes (c). Sequences obtained from Tethya aurantium are shown in bold type. Adriatic seawater clone sequences are underlined. Numbers of represented clones in each OTU are given in brackets after the clone names. Clusters of T. aurantium-associated bacterial phylotypes found in both years (Tethya-I, Tethya-II and Tethya-III) are framed with dotted line. The new sponge-specific cluster related to the Betaproteobacteria is framed with a dashed line (a). All trees were generated using the maximum likelihood method. Tree (a) is based on sequences of 500–1500 bp length. The DGGE sequence (<500) was added without changing the tree topology using the parsimony method in ARB. Trees (b) and (c) are based on almost complete sequences (b, ≥1000 bp; c, ≥1200 bp). Partial sequences (<1000/<1200 bp) were added without changing the tree topology using the parsimony method in ARB and are indicated by dashed branches. The phylogenetic positions of partial sequences were verified by calculation of all length sequences separately. Bootstrapping analysis (100 datasets) was conducted. Values equal to or greater than 50% are shown. Bootstrap values in parentheses refer to tree calculations including short sequences. The scale bars indicate the number of substitutions per nucleotide position.

Figure 4.

Figure 4.

 Phylogenetic trees constructed from 16S rRNA gene sequences related to the Betaproteobacteria (a), Cyanobacteria (b) and Bacteroidetes (c). Sequences obtained from Tethya aurantium are shown in bold type. Adriatic seawater clone sequences are underlined. Numbers of represented clones in each OTU are given in brackets after the clone names. Clusters of T. aurantium-associated bacterial phylotypes found in both years (Tethya-I, Tethya-II and Tethya-III) are framed with dotted line. The new sponge-specific cluster related to the Betaproteobacteria is framed with a dashed line (a). All trees were generated using the maximum likelihood method. Tree (a) is based on sequences of 500–1500 bp length. The DGGE sequence (<500) was added without changing the tree topology using the parsimony method in ARB. Trees (b) and (c) are based on almost complete sequences (b, ≥1000 bp; c, ≥1200 bp). Partial sequences (<1000/<1200 bp) were added without changing the tree topology using the parsimony method in ARB and are indicated by dashed branches. The phylogenetic positions of partial sequences were verified by calculation of all length sequences separately. Bootstrapping analysis (100 datasets) was conducted. Values equal to or greater than 50% are shown. Bootstrap values in parentheses refer to tree calculations including short sequences. The scale bars indicate the number of substitutions per nucleotide position.

Figure 4.

Figure 4.

 Phylogenetic trees constructed from 16S rRNA gene sequences related to the Betaproteobacteria (a), Cyanobacteria (b) and Bacteroidetes (c). Sequences obtained from Tethya aurantium are shown in bold type. Adriatic seawater clone sequences are underlined. Numbers of represented clones in each OTU are given in brackets after the clone names. Clusters of T. aurantium-associated bacterial phylotypes found in both years (Tethya-I, Tethya-II and Tethya-III) are framed with dotted line. The new sponge-specific cluster related to the Betaproteobacteria is framed with a dashed line (a). All trees were generated using the maximum likelihood method. Tree (a) is based on sequences of 500–1500 bp length. The DGGE sequence (<500) was added without changing the tree topology using the parsimony method in ARB. Trees (b) and (c) are based on almost complete sequences (b, ≥1000 bp; c, ≥1200 bp). Partial sequences (<1000/<1200 bp) were added without changing the tree topology using the parsimony method in ARB and are indicated by dashed branches. The phylogenetic positions of partial sequences were verified by calculation of all length sequences separately. Bootstrapping analysis (100 datasets) was conducted. Values equal to or greater than 50% are shown. Bootstrap values in parentheses refer to tree calculations including short sequences. The scale bars indicate the number of substitutions per nucleotide position.

Nucleotide sequence accession numbers

The 16S rRNA gene sequences obtained in this study have been deposited in the EMBL database. They have been assigned accession numbers AM259730–AM259769, AM259770–AM259831 and AM259832–AM259898 for the sequences obtained from seawater, endosome and cortex, respectively.

Diversity estimation

Sequences with similarities >99.0% were defined as one phylotype, i.e. one operational taxonomic unit (OTU). The proportion of prokaryotic diversity represented by the clone libraries was estimated by rarefaction analysis combined with nonlinear regression, and by calculation of the chao1 estimator as proposed by Kemp & Aller (2004). Rarefaction analysis calculations were performed applying the algorithm described by Hurlbert (1971) with the program aRarefactWin (http://www.uga.edu/strata/software.html). Rarefaction curves were plotted and regressions performed using two different regression equations:

image(1)
image(2)

where x is the sample size, y the observed number of OTUs and a the number of OTUs to be expected with infinite sample size (i.e. total diversity) (Koellner et al., 2004). Equation (1) is the most common regression approach for rarefaction analysis and has been used by many authors (e.g. Webster et al., 2004; Yakimov et al., 2006). However, curves derived from regression (1) exhibited an apparent misfit to the data points of the rarefaction analysis. The increase in the regression curves was too steep at small OTU numbers and curves flattened visually too early at large sample sizes. The expected underestimation of the maximum species number proposed a modification of regression (1), namely Equation (2). To prove this, we developed an algorithm to simulate sampling of specimens for rarefaction analysis that has the potential to produce datasets ranging from highly diverse (each OTU occurring only once or twice) to almost uniform (one or two abundant OTUs): a cohort of random integer numbers ki (iN; 1≤i≤50) were created with the equation ki=5·Γαi(zi), where Γ is the standard Gamma function of zi (randomized; 0≤z<1) with shape parameter αi (randomized; 0.1≤αi<0.6). Of this cohort, n numbers k were taken so that inline image (randomized; 30≤s<70). This procedure is tantamount to sampling up to s specimens (i.e. clones) representing n OTUs of abundance ki. The randomized Gamma function causes the distribution of ki to be skewed more or less to the right, resembling abundance proportions in most natural communities. The simulated datasets were subjected to rarefaction analysis, and the resulting rarefaction curves were fitted by regressions (1) and (2), respectively. From both equations, the asymptote a and the PRESS (Predicted Residual Error Sum of Squares) statistic were obtained. PRESS is a gauge of how well a regression predicts new data. The smaller the PRESS statistic, the better is the predictive ability of the regression. To test whether the two equations differed statistically from each other, we calculated the ratios A=a2/a1 and P=PRESS1/PRESS2. If there is one highly abundant OTU (ki≥20) beside many OTUs of low abundance (ki<3) in the sampling datasets, the rarefaction curve tends to converge to a line without apparent limit. This occurred three times with our simulated datasets with a2a1 (104<a2<107; 6<a1<40). For these cases, ratio A was not calculated. Transformed ratios A′=ln(ln(A+1/100)) and P′=ln(P+1/100) did not show significant deviations from the normal distribution (Shapiro–Wilk W-test). A′ and P′ were tested against the null hypotheses inline image and inline image, respectively.

SigmaPlot v6.0 (SPSS) was used for plotting and regression analysis. Statistical tests were performed with Statistica v6.1 (StatSoft).

Results

Electron microscopy

Tethya aurantium cortex and endosome (Fig. 1) samples were studied separately and sponge samples from consecutive years were compared. Electron micrographs revealed large differences between cortex and endosome. Only low numbers of bacteria were associated with the sponge cortex region while bacteria were fairly abundant in the endosome (Fig. 1c–e). Different morphotypes, especially a high abundance of rod-shaped bacteria, were found associated with the sponge endosome.

Cortex- and endosome-specific bacteria

DGGE banding patterns clearly showed the presence of different bacterial communities in endosome and cortex of T. aurantium (Fig. 2). Additionally, both parts of the sponge differed in their DGGE banding patterns from surrounding seawater samples. Bacterial phylotypes specifically associated with distinct sponge regions were represented by DGGE bands that were exclusively present in all endosome or cortex samples, respectively, but not found in seawater (Fig. 2a). Other bands were found in all sponge samples (Fig. 2a). Whereas the patterns in subsamples of endosome and cortex of one individual were apparently identical, slight variations were found between endosome as well as cortex samples from consecutive years (2003–2005) (Fig. 2b and c). Nevertheless, bacterial communities inhabiting the same sponge region (endosome or cortex) were more similar than populations from different regions (Fig. 2a and b). In contrast to the subsamples of a sponge individual (2005) that showed identical DGGE banding patterns, seawater samples from different nearby locations varied between each other to some extent (Fig. 2c). The clusters of DGGE banding patterns of endosome, cortex and seawater samples, respectively, were confirmed to be well separated by ANOSIM (R-value>0.75).

Figure 2.

 (a) DGGE banding patterns of amplified bacterial DNA extracted from Tethya aurantium Pallas cortex region (2003, 2004 and 2005-a/b), endosomal region (2003, 2004 and 2005-a/b) and surrounding seawater (SW). Specialised bands, which are not present in seawater, occur in all sponge samples (TA-I) and in the cortex only (TA-II). (b and c) Dendograms of similarity cluster analysis with DGGE banding patterns of amplified bacterial DNA extracted from T. aurantium cortex (co) and endosomal (en) samples. Comparison of sponge individuals from 2003, 2004 and 2005 and surrounding seawater (b) show that, although not identical, a clear clustering of the banding patterns is seen. Regardless of the year, samples of endosomal regions are more similar to each other than to the cortex samples of the same individual. Replicate subsamples from one individual (May 2005) (c) show identical banding patterns and precise differences between endosome and cortex, as well as to several nearby seawater samples, which differ to some extent.

Methods of diversity estimation

Comparison of the two regressions (Table 1) showed that rarefaction analysis using regression (2) resulted in higher estimated maximum OTU numbers a2 as compared with a1 of regression (1). a2 is, in most cases, comparable with the richness estimation using the nonparametric estimator chao1, as proposed by Kemp & Aller (2004) (Table 1). Equality of parameters a1 and a2 and the statistics PRESS1 and PRESS2 of the two regressions could be rejected on a highly significant level (P<10−6) in both cases. We can therefore state that (i) Equation (2) fits significantly better than Equation (1) as a regression for fitting rarefaction curves and predicting total diversity, and that (ii) the total diversity a1 calculated with the conventional Equation (1) is systematically lower than a2 calculated with regression (2).

Table 1.   Observed and estimated total bacterial diversity of phylotypes (OTUs) in the different sponge and seawater samples. Rarefaction analysis and the nonparametric richness estimator chao1 were used for diversity estimation. Rarefaction analysis was conducted with a commonly used regression (1) and a modified regression (2). In all cases regression (2) gave higher r2 values and a higher expected diversity than regression (1). Except for the seawater clone library, chao1 coverage resembles total coverage estimated by the use of regression (2). Total diversity was best covered by the analysed clones for the clone library from the sponge cortex in 2004 (76–103% depending on method used)
Clone librarynOTURarefaction analysis (RA)Chao1
Regression (1)Regression (2)Schao1Cchao1
a1r2RA1CRA1a2r2RA2CRA2
  1. n, number of clones in clone library; OTU, number of phylotypes/OTU in clone library; a, asymptote of regression equation, giving the estimated total diversity; r2, square of correlation-coefficient; C, OTU/estimated diversity, a measure of coverage of a clone library; Cchao1=Observed phylotypes/predicted SChao1=Coverage.

Seawater 2004412746.20.9997658%56.50.9999348%88.231%
Sponge cortex total663041.10.9981673%67.70.9999644%59.051%
Sponge cortex 2003372649.50.9999252%56.60.9999846%60.043%
Sponge cortex 20042965.80.97321103%7.90.9981476%6.790%
Sponge endosome total652123.00.9924691%36.10.9995458%35.559%
Sponge endosome 2003331314.80.9959988%20.50.9996863%20.663%
Sponge endosome 20043299.10.9901099%11.80.9992076%12.473%

Diversity of sponge-associated bacteria

A total of 171 clone sequences were obtained (29–41 sequences for each clone library). Observed numbers of OTUs and the total diversity estimated by rarefaction analyses revealed high variability in the sponge-associated bacterial community (i) between sponge endosome and cortex, (ii) between sponge and ambient seawater and (iii) between the different sampling times. In the sponge samples from June 2004, in cortex and endosome, six and nine OTUs were identified, respectively (Table 1). By contrast, seawater collected at the same time displayed 27 identified OTUs (Table 1). Tethya aurantium sampled in April 2003 displayed a higher diversity: 26 and 13 OTUs were identified in cortex and endosome, respectively (four-fold higher compared with cortex in 2004). The total number of OTUs obtained from the sponge cortex (30) was higher than each of the values calculated individually for 2003 (26) and 2004 (6).

Applying rarefaction analysis, regression (2) always showed higher r2 values and led to a higher estimated total diversity (Fig. 3, Table 1). According to regression (1), the total bacterial diversity seems to be well covered by the gene libraries obtained from the sponge sampled in June 2004 (88–103%). However, rarefaction analysis using regression (2) demonstrated that these proportions are overestimated under regression (1). When applying regression (2), c. 48–76% of the diversity seems to be covered by the clone libraries in this study (Table 1). Only three phylotypes were obtained from endosome or cortex in both years. The majority of the sequences were found only in one of the years. Thus, the total diversity of endosome- and cortex-associated sequences (as determined by rarefaction analysis), respectively, is much higher if both years are included than if each year is considered separately. These differences in the diversity possibly demonstrate annual and seasonal variation in the bacterial communities.

Figure 3.

 Example analytical rarefaction curve plotted for one sponge-derived 16S rRNA gene clone library (sponge cortex 2004). The expected number of OTUs as determined by the analytical algorithm described by Hurlbert (1971) were plotted against the number of analysed clones (circles). Extrapolated regression curves (solid line and dashed line) are shown for the different regression Equations (1) and (2). The expected total diversity determined by the asymptotes (a1/a2) is indicated by dotted lines. Regression (1), which has been used in former studies (Webster et al., 2004; Yakimov et al., 2006), results in a lower expected total diversity compared with regression (2) (a1<a2) with also lower values for the nonlinear coefficient of determination (r21<r22).

Phylogenetic analysis

A high phylogenetic diversity was observed for the 171 bacterial sequences obtained from both sponge samples and surrounding seawater, including members of the Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria and Deltaproteobacteria, Bacteroidetes, Verrucomicrobia, Planctomycetales, Gemmatimonadales, Acidobacteria, Actinobacteria, Spirochaeta and Cyanobacteria (Table 2). However, the majority of sequences was affiliated with the Proteobacteria (38%), the Cyanobacteria (27%) and the Bacteroidetes (25%).

Table 2.   List of phylogenetic affiliations of 16S rRNA gene clone sequences obtained from the cortex and endosome of Tethya aurantium sampled in 2003 and 2004 as well as surrounding seawater (2004)
CloneSourceLength
[bp]
Next relativeAcc.Overlap
[bp]
IdentityPhylogenetic
affiliation
  1. Numbers in brackets indicate no. of sequences in the phylotype represented by listed clones.

TAA-10-30v [1]Cortex 20031498Sponge clone L35 (Latrunculia apicalis)AY32141988193%Betaproteobacteria- related
TAA-5-01v [12]Cortex 20041489Sponge clone L35 (Latrunculia apicalis)AY32141988294%Betaproteobacteria- related
DGGE TA-Ia [2]Cortex 03/04142Sponge clone HRV40 (Halichondria panicea)UBZ8859211399%Betaproteobacteria- related
TAA-10-50v [1]Cortex 20031376Mangrove bacterioplankton clone DS143DQ234225137696%Gammaproteobacteria, Oceanospirallales
TAA-10-78v [1]Cortex 20031393Sponge clone E01-9c-26 (Axinella verrucosa)AJ581351131295%Gammaproteobacteria, Oceanospirallales
TAA-10-33v [2]Cortex 20031388Uncultured bacterium clone A314926AY907761121693%Gammaproteobacteria
TAA-10-62v [1]Cortex 20031372Gammaproteobacterium 17X/A02/237AY576771137591%Gammaproteobacteria
TAA-10-90v [2]Cortex 20031387unc. sediment proteobacterium SIMO-2184AY71155081093%Gammaproteobacteria
TAA-10-60v [1]Cortex 20031393Uncultured bacterium clone SDKAS1_6AY734243129091%Gammaproteobacteria, Coxiella group
TAA-10-18v [1]Cortex 20031303Marine Alphaproteobacterium clone MB11B07AY033299131399%Alphaproteobacteria
TAA-5-11v [2]Cortex 20041405Seamount Alphaproteobacterium clone JdFBGBact_40AF323257117488%Alphaproteobacteria
TAA-10-03v [1]Cortex 20031367Mucus bacterium 23AY654769117595%Alphaproteobacteria
TAA-10-13v [1]Cortex 20031335Hypersaline bacterium clone E6aH10DQ10360976592%Alphaproteobacteria
TAA-10-23v [2]Cortex 20031404Holophaga sp. oral clone CA002AF385537118493%Deltaproteobacteria
TAA-5-46v [1]Cortex 20041435Synechococcus sp. WH 8016AY172834143699%Cyanobacteria, Synechococcus group
TAA-10-02v [3]cortex 20031339Synechococcus sp. WH 8016AY172834132299%Cyanobacteria, Synechococcus group
TAA-10-19v [1]Cortex 20031335Antithamnion sp. plastid DANNX54299133595%Cyanobacteria, chloroplasts
TAA-10-96v [3]Cortex 20031387Tenacibaculum lutimaris strain TF-42AY661693130398%Bacteroidetes, Flavobacteriaceae
TAA-10-10v [4]Cortex 20031368Flavobacteriaceae bacterium CL-TF09AY962293136896%Bacteroidetes, Flavobacteriaceae
TAA-10-14v [2]Cortex 20031392Unc. Bacteroidetes bacterium C319a-R8C-C8AY678510138497%Bacteroidetes, Flavobacteriaceae
TAA-10-29v [1]Cortex 20031386Flavobacteriaceae str. SW334AF493686125296%Bacteroidetes, Flavobacteriaceae
TAA-10-74v [1]Cortex 20031390Bacterium K2-15AY345434139090%Bacteroidetes, Flavobacteriaceae
TAA-10-77v [1]Cortex 20031381Uncultured bacterium clone LC1408B-77DQ270634118091%Bacteroidetes, Flavobacteriaceae
TAA-5-15v [3]Cortex 20041430Bacteroidetes bacterium PM13AY548770120189%Bacteroidetes, Flavobacteriaceae
TAA-10-32v [1]Cortex 20031385Microscilla furvescensAB078079134091%Bacteroidetes, Flexibacteriaceae
TAA-5-103v [10]Cortex 20041375Microscilla furvescensAB078079130191%Bacteroidetes, Flexibacteriaceae
DGGE TA-II [2]Cortex 03/04137Bacteroidetes clone GCTRA14_SAY70146113597%Bacteroidetes, Flexibacteriaceae
TAA-5-25v [1]Cortex 20041485Flexibacter aggregansAB078038139988%Bacteroidetes, Flexibacteriaceae
TAA-10-06v [1]Cortex 20031385Uncultured marine eubacterium HstpL64AF15964087692%Planctomycetales
TAA-10-04v [1]Cortex 20031387Uncultured bacterium clone FS142-21B-02AY704401118390%Planctomycetales
TAA-10-09v [1]Cortex 20031369Uncultured Pirellula clone 6N14AF029078136997%Planctomycetales
TAA-10-101v [1]Cortex 20031419sponge clone TK19 (Aplysina aerophoba)AJ347028141693%Gemmatimonadales
TAA-10-43 [2]Cortex 20031375Unc. marine bacterium SPOTSFEB02_70m35DQ009431135792%Actinobacteria, Acidimicrobiaceae
TAA-10-01v [1]Cortex 20031382Uncultured Actinobacterium clone Bol7AY193208136596%Actinobacteria
TAI-8-03v [1]Endosome 20031436Sponge clone L35 (Latrunculia apicalis)AY32141988194%Betaproteobacteria- related
TAI-2-47f [12]Endosome 20041436Sponge clone L35 (Latrunculia apicalis)AY32141988294%Betaproteobacteria- related
DGGE TA-Ib [2]endosome 03/04142Sponge clone HRV40 (Halichondria panicea)UBZ8859211399%Betaproteobacteria- related
TAI-2-153v [2]Endosome 20041384Sponge clone 34P16 (Phyllospongia papyracea)AY84523195086%Gammaproteobacteria
TAI-8-61v [1]Endosome 20031370Uncultured marine bacterium clone SPOTSAUG01_5m75DQ009136137099%Gammaproteobacteria
TAI-8-75 [2]Endosome 2003864Uncultured Gammaproteobacterium KTc1119AF23512086099%Gammaproteobacteria
TAI-8-99k [1]Endosome 2003462Uncultured Gammaproteobacterium clone PI_4j5bAY580744438100%Gammaproteobacteria
TAI-8-76v [4]Endosome 20031370Uncultured marine bacterium clone SPOTSAPR01_5m185DQ009135137099%Gammaproteobacteria
TAI-8-20v [2]Endosome 20031388Unidentified Gammaproteobacterium OM60U70696138999%Gammaproteobacteria
TAI-8-64v [1]Endosome 20031339Photobacterium phosphoreum strain RHE-01AY435156130399%Gammaproteobacteria
TAI-2-166v [5]Endosome 20041443Cyanobacterium 5X15AJ289785144399%Cyanobacteria, Synechococcus group
TAI-8-58v [6]Endosome 20031332Cyanobacterium 5X15AJ289785133199%Cyanobacteria, Synechococcus group
TAI-8-74v [9]Endosome 20031340Synechococcus so. Almo3AY172800132699%Cyanobacteria, Synechococcus group
TAI-2-160v [6]endosome 20041419Synechococcus sp. RS9920AY172830140299%Cyanobacteria, Synechococcus group
TAI-8-17v [3]Endosome 20031372Flexibacter sp. IUB42AB058905137595%Bacteroidetes, Flavobacteriaceae
TAI-8-94v [1]Endosome 20031380Uncultured marine bacterium ZD0255AJ400343138196%Bacteroidetes, Flavobacteriaceae
TAI-8-51v [1]Endosome 20031388Uncultured CFB group bacterium clone AEGEAN_179AF406541136897%Bacteroidetes, Flavobacteriaceae
TAI-2-145v [1]Endosome 20041438Uncultured marine bacterium clone Chl1.12DQ071033141999%Bacteroidetes, Flavobacteriaceae
TAI-2-81v [1]Endosome 20041372Uncultured marine bacterium clone SPOTSAPR01_5m235DQ009115135397%Bacteroidetes, Flavobacteriaceae
TAI-2-123v [1]Endosome 20041385Flexibacter aggregans strain:IFO 15974AB078038145788%Bacteroidetes, Flexibacteriaceae
TAI-2-130f [3]Endosome 20041418Uncultured marine eubacterium HstpL83AF159642100899%Planctomycetes
TAI-2-28v [1]Endosome 20041407Uncultured Verrucomicrobia Arctic96BD-2AY028221119395%Verrucomicrobia
TAI-8-67v [1]Endosome 20031369Uncultured bacterium clone ELB16-004DQ015796136998%Actinobacteria
TAU-7-56k [1]Seawater 2004430Uncultured Gammaproteobacterium clone SIMO-2629DQ18960436799%Gammaproteobacteria
TAU-7-53p [1]Seawater 2004937Uncultured marine bacterium clone SPOTSOCT00_5m102DQ00913888298%Gammaproteobacteria
TAU-7-30p [4]Seawater 2004835uncultured Gammaproteobacterium CHAB-III-7AJ24092184098%Gammaproteobacteria
TAU-7-93 [1]Seawater 2004496Uncultured Gammaproteobacterium OCS44AF00165049599%Gammaproteobacteria
TAU-7-25 [1]Seawater 2004407Uncultured Gammaproteobacterium NAC11-19AF24564240797%Gammaproteobacteria
TAU-7-100v [1]Seawater 20041509Uncultured Gammaproteobacterium KTc1119AF235120149199%Gammaproteobacteria
TAU-7-63p [1]Seawater 2004866Uncultured bacterium clone MP104-1109-b35DQ08879984399%Gammaproteobacteria
TAU-7-71p [3]Seawater 2004866Uncultured bacterium MabScd-NBAB19392983399%Alphaproteobacteria
TAU-7-36p [1]Seawater 2004858Uncultured Proteobacterium clone SIMO-855AY71239274899%Alphaproteobacteria
TAU-7-38 [1]Seawater 2004414Uncultured bacterium clone CD3B11AY03839141097%Alphaproteobacteria
TAU-7-44p [1]Seawater 2004862Uncultured Alphaproteobacterium clone PI_4t1gAY58054780995%Alphaproteobacteria
TAU-7-79v [1]Seawater 20041438Unidentified eukaryote clone OM21 plastid 16S rRNA geneU32671125196%Cyanobacteria, chloroplasts group
TAU-7-26p [2]Seawater 2004747Environmental clone OCS50 chloroplast geneAF00165674698%Cyanobacteria, chloroplasts group
TAU-7-39p [1]Seawater 2004833Unidentified haptophyte OM153U7072076895%Cyanobacteria, chloroplasts group
TAU-7-57p [1]Seawater 2004801Uncultured diatom clone Hot Creek 8AY16875180295%Cyanobacteria, chloroplasts group
TAU-7-73p [1]Seawater 2004756Unidentified eukaryote clone OM20U3267075598%Cyanobacteria, chloroplasts group
TAU-7-97p [3]Seawater 2004803Environmental clone OCS20AF00165480098%Cyanobacteria, chloroplasts group
TAU-7-68v [4]Seawater 20041436Neoptilota densa plastidDQ028877133494%Cyanobacteria, chloroplasts group
TAU-7-74p [1]Seawater 2004864Environmental clone OCS162AF00165961792%Cyanobacteria, chloroplasts group
TAU-7-50p [1]Seawater 2004803Uncultured Bacteroidetes bacterium clone SIMO-780AY71231758096%Bacteroidetes, Flavobacteriaceae
TAU-7-28p [2]Seawater 2004690Uncultured Bacteroidetes bacterium clone CONW90AY82842058299%Bacteroidetes, Flavobacteriaceae
TAU-7-43 [1]Seawater 2004476Uncultured Bacteroidetes bacterium clone CONW90AY82842045795%Bacteroidetes, Flavobacteriaceae
TAU-7-02v [1]Seawater 20041490Uncultured marine bacterium clone Chl1.12DQ071033145299%Bacteroidetes, Flavobacteriaceae
TAU-7-69v [3]Seawater 20041487Unc. marine bacterium SPOTSAPR01_5m235DQ009115146897%Bacteroidetes, Flavobacteriaceae
TAU-7-61p [1]Seawater 2004813Unc. Bacteroidetes bacterium 3iSOMBO27AM16257681796%Bacteroidetes, Flexibacteriaceae
TAU-7-58 [1]Seawater 2004277Uncultured bacterium gene for 16S rRNA, clone:JS624-8AB12110628893%Spirochaetes
TAU-7-55p [1]Seawater 2004804Bacillus sp. C93DQ09100879799%Firmicutes, Bacillus group

Some phylotypes were found in T. aurantium from both years. They form monophyletic clusters (i) related to the Betaproteobacteria (Tethya-I, Fig. 4a), (ii) affiliated with the Cyanobacteria (Tethya-II, Fig. 4b) and (iii) within the Bacteroidetes (Tethya-III, Fig. 4c). Within each of these clusters, sequences share >99% similarity. Of special interest is cluster Tethya–I, which is represented by 26 clone sequences (19%) and was found in the cortex and endosome of all T. aurantium individuals. Tethya-I forms a monophyletic cluster with sponge-derived 16S rRNA gene sequences from different sponges from Antarctic (Webster et al., 2004) and Australian waters (Taylor et al., 2004), as well as the Mediterranean Sea (Althoff et al., 1998). This sponge-specific monophyletic cluster is related to the Betaproteobacteria and branches deeply (dotted frame in Fig. 4a). In addition, the DGGE band TA-I, assigned to the sponge-specific cluster Tethya-I, was unique to all T. aurantium samples but was not found in seawater. Moreover, no clone sequences belonging to the Betaproteobacteria were obtained from seawater in this study.

Within the Cyanobacteria, three phylotypes comprising 30 Synechococcus species sequences (22%) were obtained from T. aurantium-derived clone libraries. One of the phylotypes was found repeatedly in consecutive years (Tethya-II, Fig. 4b). Synechococcus species sequences were found mainly in the endosome samples (Table 2), but not in the surrounding seawater. All cyanobacteria-like sequences obtained from the surrounding seawater belonged to chloroplasts of different algae. Only one chloroplast sequence was also found in the sponge cortex in April 2003.

Within the Bacteroidetes one bacterial cluster (Tethya-III) of Flexibacteriaceae was repeatedly found in T. aurantium (Fig. 4c, Table 2). The sequences of the cluster shared highest similarity with Microscilla furvescens (AB078079, 91%) and were found in the cortex only. Additionally, DGGE band TA-II, present exclusively in cortex samples and absent from endosome and seawater, was assigned to this cluster (Fig. 4c). A further cluster of T. aurantium-derived Flexibacteriacea was found in both endosome and cortex in June 2004 samples (Fig. 4c). Similarity between the clusters was 90%. Flexibacteriaceae related (92%) to putative vertically transmitted sponge-symbionts (Enticknap et al., 2006) were found in the sponge cortex from 2004 only (TAA-5-15v, Fig. 4c). All other T. aurantium-derived sequences within the Bacteroidetes belonged to the Flavobacteriaceae and Saprospiraceae, regardless of their origin from sponge cortex or endosome (Fig. 4c). All sequences showed a high degree of similarity to sequences retrieved from various marine environments (Fig. 4c, Table 2).

Thirty T. aurantium-derived 16S rRNA gene sequences (22%) were closely related to other sponge-derived 16S rRNA gene sequences. Twenty-six of these belonged to cluster Tethya-I, with the remaining four sequences affiliated to the Gemmatimonadales, the Gammaproteobacteria, the Actinobacteria and the Acidobacteria: Clone TAA-10-101v, obtained from T. aurantium cortex in 2003, clustered monophyletically with the sponge-specific cluster ‘uncertain-I’ described by Hentschel et al. (2002) within the Gemmatimonadales (represented by sponge clone TK19, Table 2). Clone TAA-10-78v from the T. aurantium cortex (April 2003) was related to sequences obtained from sponges as well as other marine habitats within the Gammaproteobacteria (data not shown). blast search results showed closest association to sponge-associated bacteria (Table 2), but phylogenetic analysis could not identify monophyly of the sponge-derived gammaproteobacterial sequences. TAI-2-153v, representing two sequences obtained from the sponge endosome in 2004, was most similar to the Phyllospongia papyracea-associated Gammaproteobacteria clone 34P16 (Ridley et al., 2005) (sequence similarity of 86%, Table 2). Again, phylogenetic analysis did not support monophyletic clustering of the sponge-derived sequences. Actinobacteria as well as Acidobacteria derived from sponges have been reported previously (Hentschel et al., 2002; Imhoff & Stöhr, 2003; Kim et al., 2005; Schirmer et al., 2005) and sponge-specific clusters have been described (Hentschel et al., 2002). Tethya aurantium-derived sequences TAI-8-67v and TAA-10-43v showed similarities of 84–95% to different sponge-associated Actinobacteria sequences but did not group with any of the three described sponge-specific Actinobacteria clusters (Hentschel et al., 2002). TAA-10-23v, which represents two clones obtained from the T. aurantium cortex, showed <90% sequence similarity to the sponge-specific cluster Acido-I (Hentschel et al., 2002) and did not cluster monophyletically with it.

Several sequences obtained from T. aurantium were not specifically related to sequences found in sponges, but were either closely related to seawater-derived bacteria or to bacteria associated with other marine macroorganisms (Vergin et al., 1998; Cho & Giovannoni, 2004; Brown et al., 2005) [i.e. TAI-8-61v, TAI-8-64v, TAI-8-20v (Gammaproteobacteria) and TAI-2-130v, TAA-10-09v (Planctomycetales); Table 2]. Bacteria affiliated with the Gammaproteobacteria (TAU-7-100v/TAI-8-75p) and the Flavobacteriaceae (Bacteroidetes) (TAI-2-145v/TAU-7-02v and TAI-2-81v/TAU-7-69) were found in both T. aurantium endosome and seawater in this study. No sequence obtained from the sponge cortex showed similarity to seawater-derived sequences obtained in this study.

Alphaproteobacteria were exclusively found in the T. aurantium cortex and in ambient seawater in June 2004 (Table 2). The sponge-derived sequences were related to bacterioplankton-derived sequences (Suzuki et al., 2001), coral mucus-associated bacteria (O. Koren & E. Rosenberg, unpublished data, GenBank accession no. AY654769) and a filamentous bacterium from a wastewater treatment plant (Levantesi et al., 2004). Alphaproteobacteria closely related to those found in seawater in this study (TAU-7-44p and TAU-7-38) have been obtained from sponges (Halichondria panicea and Halichondria okadai) previously (Althoff et al., 1998; I. Okano et al., unpublished data, GenBank accession no. AB054143).

Discussion

Although sponges do not possess organs or real tissues, cortex and endosome are clearly differentiated with respect to structure and function. For the first time, our studies on Tethya aurantium have revealed that they also differ in their bacterial communities. Distinct phylotypes, represented by DGGE bands and 16S rRNA gene clone sequences, were affiliated with the different regions of the sponge.

Tethya aurantium supports a relatively low diversity of specifically associated bacteria. Only three bacterial phylotypes were found in sponge specimens from both years investigated. For diversity estimation we applied rarefaction analysis in combination with two nonlinear regressions. The commonly used regression (1) was shown to underestimate systematically the total diversity, while regression (2) results in estimates comparable with those from chao1. Regression (2) was demonstrated to be significantly better suited as a regression for fitting rarefaction curves and predicting total diversity, and therefore is recommended to be used for diversity estimation of clone libraries in future studies. Under application of the more conservative regression (2), rarefaction analysis displays only two (cortex) and four (endosome) expected additional bacterial phylotypes (OTUs) not identified in this study for the T. aurantium specimen from 2004.

One characteristic cluster closely related to the Betaproteobcteria (cluster Tethya-I) is associated with both endosome and cortex of T. aurantium. Betaproteobacteria, with the exception of ammonium oxidizers (Voytek & Ward, 1995), are not abundant in open oceans (Giovannoni & Rappé, 2000), but are characteristic of freshwater habitats (Methe et al., 1998; Schweitzer et al., 2001) and have also been observed in coastal waters (Rappe et al., 1997; Fuhrman & Ouverney, 1998). Nonetheless, Betaproteobacteria-affiliated bacteria have previously been found in sponges by culture-independent methods (Althoff et al., 1998; Webster et al., 2001, 2004; Thoms et al., 2003; Taylor et al., 2004). In Rhopaloides odorabile, they were located intracellularly in some cases by fluorescence in situ hybridization (Webster et al., 2001). Tethya-I clusters monophyletically with those other sponge-derived sequences (Althoff et al., 1998; Webster et al., 2004; Taylor et al., 2004, 2005), forming a sponge-specific monophyletic cluster (Fig. 4a). Our studies indicate that this cluster branches deeply within the Betaproteobacteria, but its exact phylogenetic affiliation remains uncertain. Phylogenetic analysis using the backbone tree and the parsimony method implemented in the ARB program (Ludwig et al., 2004) demonstrated a consistent affiliation with the Betaproteobacteria, although within this group the phylogenetic position of the Tethya-I cluster largely depends on the (DNA)-filter used. By contrast, complete phylogenetic calculations including a reasonable number of representatives of the Betaproteobacteria, Gammaproteobacteria and Alphaproteobacteria, using the ML method (Felsenstein, 1981), placed the sponge-specific cluster outside the known Betaproteobacteria (Fig. 4a).

Finding of Tethya-I-related sequences in Halichondria panicea from the Adriatic Sea but not in individuals from the North Sea or the Baltic Sea (Althoff et al., 1998) possibly indicates a synergistic sponge–microbe association. Yet, the occurrence of members of the Betaproteobacteria-related sponge-specific cluster is not limited to the Mediterranean Sea, as formerly unaffiliated sequences obtained from the Antarctic sponges Latrunculia apicalis and Mycale acerata (Webster et al., 2004) can now be assigned to the cluster. As Betaproteobacteria are abundant in freshwater habitats (Methe et al., 1998; Schweitzer et al., 2001) a freshwater origin cannot be excluded for the T. aurantium-derived sequences obtained from the river-fed Limksi kanal. However, no Betaproteobacteria were found in the seawater surrounding the T. aurantium habitat and furthermore no freshwater-derived bacteria closely related to the Tethya-I cluster have been described.

Cyanobacterial associations in sponges have been known for many years. Within the Synechococcus group one specifically sponge-associated cluster and several additional sponge-associated Cyanobacteria have been identified (Steindler et al., 2005). Besides symbiotic associations, the genus Synechococcus is also known as a member of marine picoplankton communities and serves as food for different filter-feeding animals, including sponges (Pile et al., 1996). For T. aurantium no cyanobacterial associations have been reported in the literature. Interestingly, we found Synechococcus species sequences making up a major part of the endosome-derived clone library (n=26%, 39%) (Fig. 4b, Table 2). They represent three closely related phylotypes, only distantly related to the sponge-specific group of Synechococcus species strains. They show high sequence similarity to several cultured and uncultured Synechococcus strains, also including sponge- and sponge-larvae-associated uncultured strains obtained from Chondrilla sp. and Mycale laxissima (Usher et al., 2004; Enticknap et al., 2006). The close relationship to putative vertically transmitted sponge-associated Synechococcus strains indicates constant and obligate association to sponges. However, unlike the T. aurantium-derived sequences in this study, the closely related cyanobacterial symbionts from Chondrilla sp. were also found in seawater. It was hypothesized that these Cyanobacteria have become symbiotic with sponges relatively recently (Usher et al., 2004). Owing to the close phylogenetic relationship to planktonic Synechococcus strains, a seawater origin cannot be excluded for the T. aurantium-associated strains either. Cyanobacteria in sponges have generally been found in the thin (few millimetres) outer tissue regions, where light energy is available for photosynthesis. The endosome of T. aurantium is covered by a thick and dense cortex region, but spicule bundles might function as a natural light conductor, similar to fibre-optic systems, as was postulated for the growth of the sponge-associated green alga Ostreobium sp. in Tethya seychellensis (Gaino & Sará, 1994).

A specifically cortex-associated bacterial cluster was identified within the division Bacteroidetes, affiliated with the family Flexibacteriaceae (clone cluster Tethya-III and DGGE band TA-II, Fig. 2). Although several previous studies have demonstrated Bacteroidetes belonging to the family Flavobacteriaceae to be associated with sponges (Webster et al., 2001, 2004; Lafi et al., 2005), only very recently have sponge-associated Flexibacteriaceae been obtained from sponge larvae by culture-independent methods (Enticknap et al., 2006). These putative vertically transmitted sponge symbionts are closely related to sequences obtained from T. aurantium cortex in 2004 and distantly related to cluster Tethya-III found in both years. Cluster Tethya-III has been observed in sponges for the first time in this study. The bacteria were not detected in the ambient seawater and in the sponge endosome. Given the occurrence in the sponge cortex exclusively and presence in specimens from consecutive years as well as the phylogenetically relatively large distance (<91%) to known sequences, we assume specific association between the Microscilla-like bacteria and T. aurantium.

As sponges are filter-feeding animals, a seawater origin of sponge-derived bacterial sequences cannot be excluded despite repeated washing steps in sterile seawater prior to DNA extraction. Several sequences obtained from seawater in this study share high similarity with clone sequences found in both T. aurantium and other sponges. Thus, some of the endosome-associated bacterial sequences may solely resemble DNA of ambient seawater bacteria ingested in the choanocyte chambers. Additionally, the high variability of different phylotypes observed either implies seasonal differences in the sponge-associated bacterial communities or possibly reflects seasonal microbial population dynamics in the ambient seawater. The similarity between the sponge endosome-associated bacterial community and the ambient seawater bacterioplankton as demonstrated in DGGE and phylogenetic analysis again emphasizes the need to differentiate between sponge-specific and merely ingested bacteria.

Hentschel et al. (2002) found 70% of all sponge-derived sequences clustering together in different phylogenetic clades. By contrast, a minor fraction (22%) of the T. aurantium-derived sequences were closely related to other sponge-associated bacteria. Apart from cluster Tethya–I related to the Betaproteobacteria, only two additional sequences obtained from T. aurantium cluster with other known sponge-derived sequences. The monophyly of the sponge-specific cluster indicates common ancestry for members of this group. Furthermore, limitation of these bacteria to associations with the phylum Porifera can be hypothesized. However, because abundance and diversity of bacteria associated with different sponges depends to a large extent on the sponge species and possibly on seasonal influences, a uniform specifically sponge-associated bacterial community as proposed by Hentschel et al. (2002) probably does not exist.

We suggest a specific association of both the Betaproteobacteria-related cluster Tethya–I and the Flexibacteriaceae cluster Tethya-III with the sponge T. aurantium. The new cluster of specifically associated Flexibacteriaceae has so far been exclusively found in T. aurantium and its presence in sponges of other taxonomic affiliation and geographical regions remains to be investigated. The unusual association of Synechococcus species strains with the T. aurantium endosome and the putative light conduction by sponge spicule bundles will be studied in future research.

Acknowledgements

We gratefully acknowledge the help of R. Batel and the personnel at the Ruder Boskovic Institut (Rovinj, Croatia) and F. Brümmer (University of Stuttgart, Germany) during sampling of the sponges. This work was supported by the German Federal Ministry of Education and Research (bmb+f) as part of the Kompetenz-Zentrum BIOTECmarin (grant no. 03F0345B)

Ancillary