Sponge-Specific Unknown Bacterial Groups Detected in Marine Sponges Collected from Korea Through Barcoded Pyrosequencing

Abstract

The bacterial diversity of 10 marine sponges belonging to the species Cliona celata, an unidentified Cliona species, Haliclona cinerea, Halichondria okadai, Hymeniacidon sinapium, Lissodendoryx isodictyalis, Penares incrustans, Spirastrella abata, and Spirastrella panis collected from Jeju Island and Chuja Island was investigated using amplicon pyrosequencing of the 16S rRNA genes. The microbial diversity of these sponges has as of yet rarely or never been investigated. All sponges, except Cliona celata, Lissodendoryx isodictyalis, and Penares incrustans, showed simple bacterial diversity, in which one or two bacterial OTUs occupied more than 50% of the pyrosequencing reads and their OTU rank abundance curves saturated quickly. Most of the predominant OTUs belonged to Alpha-, Beta-, or Gammaproteobacteria. Some of the OTUs from the sponges with low diversity were distantly (88%~89%) or moderately (93%~97%) related to known sequences in the GenBank nucleotide database. Phylogenetic analysis showed that many of the representative sequences of the OTUs were related to the sequences originating from sponges and corals, and formed sponge-specific or -related clades. The marine sponges investigated herein harbored unexplored bacterial diversity, and further studies should be done to understand the microbes present in sponges.

Introduction

Sponges are well known as reservoirs of microbes; they are inhabited by high numbers of bacteria, which were estimated to be up to 109 cells per milliliter of sponge tissue [14], accounting for as much as 50% of the biomass of the sponges [31]. The microbes in sponges have various roles, including as food sources [29], pathogenic agents [2], symbiotic residents [49], sponge skeleton stabilizers, metabolic waste consumers, secondary metabolite producers [11], and nutrient cycle mediators [40].

Numerous studies have been performed to reveal the diversity in sponges. Culture-dependent studies based on the isolation of pure cultures revealed various kinds of bacteria, [44,48] and novel strains were isolated and proposed as new taxa [28,51]. Many studies based on culture-independent methods have also been performed to reveal so-called unidentified majorities [1] and unknown bacterial diversity in sponges [13,36]. The massive parallel 454 pyrosequencing, a next-generation sequencing method, has been applied to investigate the bacterial diversity in various environments [7,30,34] as well as in marine sponges [17,18,24,33,34,45,46].

Marine sponges harbor various kinds of bacteria, comprising a total of 25 bacterial phyla [34]. The most diverse phyla contained in sponges are Proteobacteria, Chloroflexi, and Poribacteria, but many less diverse phyla such as Acidobacteria, Actinobacteria, Cyanobacteria, Gemmatimonadetes, Bacteroidetes, Spirochaetes, Firmicutes, Nitrospirae, TM7, SBR1093, and OS-K have also been observed [34].

However, some sponges, known as low microbial abundance (LMA) sponges, have low microbial content of only about 105-106 cells per gram or milliliter [12]. LMA sponges also have lower taxonomic diversity (less number of taxa) than high microbial abundance (HMA) sponges, and were reported to contain only up to five phyla in a study of five specimens of three different LMA sponges, Callyspongia vaginalis, Niphates digitalis, and Raspailia topsenti [10].

In the present study, the bacterial diversity in 10 sponges collected from Jeju Island and Chuja Island, Korea, was investigated using pyrosequencing to reveal the unidentified bacterial diversity of sponges, including Cliona celata, Haliclona cinerea, Lissodendoryx isodictyalis, Spirastrella abata, Spirastrella panis, and Penares incrustans, which have rarely been explored to date. Most of the sponges, except Cliona celata, Lissodendoryx isodictyalis, and Penares incrustans, were found to contain very simple diversity comparable to that of LMA sponges, and some of the predominant OTUs were found to be unrelated to known sequences.

 

Materials and Methods

Sample Collection and DNA Extraction

Sponge specimens were collected from Mueung-ri, Daejeongeup, Seogwipo City, Jeju Province (Jeju Island), Korea (n = 6) in Feb. 2011 by scuba diving, and from around Chuja Island in the South Sea of Korea (n = 4) in Nov. 2009 by scuba diving or from the intertidal region. Samples were collected aseptically and delivered to the laboratory. Small pieces (about 1 cm3 in volume) of sponge tissues were washed with sterilized seawater, frozen at -70℃ for 24 h, and then lysophilized at -50℃, 0.033 Mbar for 24 h. The freeze-dried tissues were aseptically homogenized in a mortar. The G-spin genomic DNA extraction kit (Intron, South Korea) was used for DNA extraction.

Barcoded Pyrosequencing

The V1 to V3 region of the 16S rRNA genes were amplified using primer sets (V1-9F: 5’-X-AC-GAGTTTGATCMTGGCTCAG-3’ and V3-541R: 5’-X-AC-WTTACCGCGGCTGCTGG-3’; X indicates the barcode sequences that were composed of various combinations of six nucleotides to tag different samples). The Genome Sequencer FLX titanium (Roche, Germany) system was used for pyrosequencing, according to the method in the manufacturer’s manuals (Macrogen, Korea).

Bioinformatic Analysis of Reads

After pyrosequencing, reads were sorted according to tags, and then the primer and tag sequences were cut out. Sequences less than 300 bp in size and with any ambiguous bases, “N”, were discarded from further analyses. Sequences with reverse complement orientation were adjusted into the same orientation. Sequences were aligned and the front and rear parts of sequences that did not match with the other sequences were cut. Putative chimera sequences were detected with the chimer.uchime [6] command of the Mothur package [32]. The QIIME package ver. 1.7.0 was used for to determine the OTUs and for their taxonomic assignment [22]. Sequence names were marked in accordance with the samples and merged into one file. The uclust approach in the QIIME package was used for clustering the OTUs and the determination of representative sequences. A sequence similarity of 97% was used as the criterion for taxonomic assignments. The representative sequences of OTUs were compared with the 97_otus and the 97_otu_taxonomy files of the gg_13_05 version of the Greengenes database [27] and taxonomy was assigned using the RDP classifier [23]. Sequences related to chloroplasts and mitochondria were discarded from further analyses. Diversity estimators, including Chao1, Shannon, and Simpson indices, were determined for subsamples made with sequences of the same numbers. Unifrac analysis [26] was performed to compare the samples. For the Unifrac analysis, similarities of 97%, 94%, 91%, 88%, and 85% were used as criteria for OTU clustering. The PyNAST program [3] was used for the alignment of representative sequences obtained from the clustering, after which the FastTree program [25] was used for construction of a phylogenetic tree using the aligned sequences. UPGMA trees were made from the Unifrac analysis. Some representative sequences were compared against the nucleotide database of GenBank and related sequences were used for the construction of a phylogenetic tree. A neighborjoining tree was constructed with the MEGA5 program [39], in which the Kimura 2-parameter method [21] was used to calculate a distance matrix.

 

Results and Discussion

Sponges and Their Bacterial Communities

All 10 specimens were so-called siliceous sponges (the class Demospongiae), which have spicules made out of silicon dioxide [41]. Cliona celata is an ecologically/biotechnologically important species that is well known as an excavating and cosmopolitan sponge [50]. The sterol composition and novel fatty acids were discovered from Haliclona cinerea sponges from the Black Sea [8,19]. Lissodendoryx isodictyalis is an encrusting sponge with blue-gray color and a strong odor [35]. Lipids with cytotoxicity and other bioactivities were isolated from Spirastrella abata sponges from Korea [15]. The diversity of cultured and uncultured bacteria of Spirastrella abata and Spirastrella panis have been investigated with restriction fragment length polymorphism and denaturing gradient gel electrophoresis (DGGE) [4,16]. Bioactive triterpenoids and penaramides were discovered from Penares incrustans sponges [37,42]. However, the latter five species have not yet been studied extensively.

As far as we know, bacterial diversity (especially uncultured) of sponges such as Cliona celata, Haliclona cinerea, Lissodendoryx isodictyalis, Spirastrella abata, Spirastrella panis, and Penares incrustans have not yet been or have rarely been published as of yet. Nucleotide records of the GenBank database were surveyed using three or four keywords: (16S) AND (rRNA) AND (“genus name”), or (16S) AND (rRNA) AND (“genus name”) AND (“species name”) of each sponge. The numbers of records that came up for each sponge species or genus name were as follows: Cliona celata, 0; Cliona, 6; Haliclona cinerea, 0; Haliclona, 1839; Halichondria okadai, 122; Halichondria, 608; Hymeniacidon sinapium, 131; Hymeniacidon, 1,245; Lissodendoryx isodictyalis, 0; Lissodendoryx, 1,598; Penares incrustans, 3; Penares, 6; Spirastrella abata, 0; Spirastrella panis, 1; Spirastrella, 5. When species names were used, six sponge species among eight showed no or no more than three records. Even when genus names were used, three sponge genera including Cliona, Penares, and Spirastrella showed no more than six records. This shows that the bacterial diversity of most of the sponges in this study was rarely investigated. Therefore, this study may provide the first observation of the comprehensive bacterial diversity of the sponges examined.

Unifrac Analysis of Sponges According to Bacterial Diversity

A total of 28,913 sequences were assigned taxonomically after non-bacterial reads including chloroplasts and mitochondria-related sequences were discarded. UPGMA trees from Unifrac analyses showed somewhat different tree topologies in accordance with 16S rRNA gene similarities (91%~85%). One group composed of Sppa2S, Spab1S, and Spab1M showed the same topology in all trees, and one specimen, Pein1M, was always distinguished from the others (Fig. 1). The relationships among other specimens was not robust in the tree topologies that were made based on different criteria of 16S rRNA gene similarities, because the bacterial communities from each of the sponges were highly different from each other. A previous study with 32 sponges showed the bacterial patterns to have no correlation with sponge phylogeny. Sponges from different genera were grouped closer than those from the same genus based on their bacterial compositions [34]. However, Sppa2S, Spab1S, and Spab1M belonged to the same genus and showed a closely related relationship in a clade. Although Spab1S and Spab1M were collected from different sample sites, their bacterial profiles were very similar, which implies that taxon-specific bacterial communities exist in some sponges.

Table 1.aC, Chuja Island; M, Mureung-ri.

Fig. 1.Weighted UPGMA tree from Unifrac analysis showing relationships between samples according to bacterial profile. OTUs were determined based on 97% similarity. Numbers at the branched points indicate percentages of frequency of each cluster detected in the five UPGMA trees, which were constructed based on different 16S rRNA gene similarities (97%, 94%, 91%, 88%, and 85%). The scale bar represents Unifrac distance.

Diversity Estimation of Sponges

The sponges used in this study generally showed low and moderate diversity, except for the specimen Pein1M. Several sponges showed extremely low diversity, which could be expressed by Shannon indexes below 3.00 and even down to 0.81 (Table 2). The simplest bacterial community was that of the specimen Haok1M, identified as Halichondria okadai, which contained only seven OTUs. In a previous study, specimens of the sponge Halichondria panacea in the same genus also showed simpler diversity than seawater, with some specimens showing less than five bands in the DGGE patterns [47]. Another sponge showing extremely low diversity was Haci1M, a specimen of the sponge Haliclona cinerea. Two species in the same genus, H. heliophila and H. tubifera, were also reported to have low diversity and species-specific bacterial communities [9]. Spab1M and Spab1S also had low levels of diversity with a single highly abundant OTU. Low diversity in the sponge genus Spirastrella was reported in a previous study [16], where only five bands were observed in the DDGE patterns. The specimen Clsp1S showed much lower diversity than Clce1S, another specimen of the same genus, Cliona. Rank-accumulated relative abundance curves showed distinguished patterns among the sponge specimens (Fig. 2). The sponges with low diversity had single predominant OTUs with high proportions from 50.8% to 92.6%. Some of them had only a few or dozens of OTUs, but few OTUs showed up in high proportions. The sponges with relatively high diversity such as Liis2M, Clce1S, and Pein1M showed different curve shapes. Pein1M showed evenly distributed abundances among the OTUs. Clce1S also showed a similar moderate slope, except for the OTU ranking first. Liis2M showed evenly distributed abundances among OTUs with several highly abundant OTUs, but the curve saturated quickly. The bacterial diversities of marine sponges examined in previous studies were much higher than those of the present study [17,18], and many more bacterial phyla were detected. In the present study, however, several sponges had very low diversity and were composed of very few OTUs, with one or two OTUs sometimes occupying almost all reads. This might be derived from several factors, such as different sampling sites, experimental procedure, number of reads, and sponge condition. There are reports of low diversity in sponges in which very few OTUs were found in many sponges [34], so it may not seem to be a rare phenomenon.

Table 2.aReads were subsampled from each sample and used to calculate diversity estimators. Data from Haok1M were not used for diversity index calculation because the number of reads was too small. OTUs were determined based on 16S rRNA gene similarity of 97%.

Fig. 2.Rank-cumulative relative abundance curve in which relative abundances of OTUs were accumulated sequentially in order of abundance from the largest. The first 19 OTUs are shown. OTUs were determined based on 97% similarity.

Bacterial Diversity at Phylum and Class Levels

Pein1M was clearly distinguished from the other sponges at the phylum level. It had a bacterial community spanning to 14 phyla, in which the bacteria were evenly distributed (9 phyla were more than 1% in proportion). Clce1S also contained many phyla (16 phyla), but only five of the phyla were more than 1%, and Proteobacteria occupied up to 83.0%. A much higher proportion of Proteobacteria was observed in all specimens except Pein1M (12.5%), ranging from 70.7% to 100.0%. These high proportions were unusual, but not impossible according to previous studies. A high proportion of Proteobacteria (up to 72.6%) was also observed in a previous study [18]. In the low microbial abundance sponges Crella cyathophora, Stylissa carteri, and Niphates digitalis, 78%, 87%, and 93% abundances of Proteobacteria, respectively, were roughly calculated from data on the investigation of bacterial communities [10]. Diversity at the class level of bacteria detected herein is shown in Fig. 3. Reads related to Alphaproteobacteria were predominant in Spab1S, Spab1M, Sppa2S, Clsp1S, and Haok1M. In contrast, Gammaproteobacteria was the major class in Hysi1M and Liis2M, and they also contained a significant amount of Alphaproteobacteria. Clce1S and Haci1M contained Betaproteobacteria as the major class, and Clce1S also contained Gammaproteobacteria as a significant member.

Fig. 3.Taxonomic assignment results at the class level. Only phyla with an abundance of more than 1% in at least one sample are shown. “Unclassified” means that the sequence was related to reads of which the position has not been verified taxonomically in that level. Square brackets indicate that the name was not verified taxonomically.

BLASTN and Phylogenetic Analysis of Predominant OTUs

Some representative sequences of highly abundant OTUs were compared against the nucleotide database in GenBank (Table 3). The criterion of 94% similarity was used for OTU determination to reduce the sequence number for comparison. Above all, representative sequences were related to sequences mainly from sponges. Some sequences were related to other marine invertebrates such as coral and octacoral, and to marine environments such as seawater, marine sediments, and microbialites. Most of the sequences were taxonomically unidentified from genus even to class level based on QIIME analysis.

Table 3.Ratios in each specimen, taxonomic assignment, and results of the BLASTN comparison with the nucleotide database of GenBank are summarized. aClassification was shown at the genus level. “Unclassified” means that the sequence was related to reads of which the position has not been verified taxonomically in that level. Square brackets indicate that the name was not verified taxonomically. OTU was determined based on a 16S rRNA gene similarity of 94%.

For example, the sequence of denovo103, which was highly predominant in the clade Spirastrella, was related to a clone (Accession No. JQ515708) reported from the coral Montastraea faveolata [20]. Denovo103 was distantly related (93% similarity) to the known sequence JQ515708 and was taxonomically identified only at the class level as Alphaproteobacteria. The representative sequence of denovo694, the most abundance OTU from Haok1M, showed 100% similarity with a sequence (AB054177) obtained from the same sponge species Halichondria okadai from an unpublished study. Interestingly, the second most abundant OTU from Haok1M (denovo197) also showed similarity with a sequence (AB054180, 98%) from the same unpublished study. The two sequences, denovo694 and denovo197, showed high similarity with clones in public databases, but they were very different from strains that have been cultivated and identified taxonomically (in Alphaproteobacteria and Gammaproteobacteria with 90% and 86% similarity, respectively).

Denovo624, the representative sequence of the most abundant OTU of the sponge HaciM, showed the best hit with a sequence (JF824778, 98% similarity) obtained from the same sponge genus, Halichondria tubifera, and belonged to the Betaproteobacteria. Another sequence (KC492704) belonging to Betaproteobacteria was reported to be important in a different sponge, Crambe crambe, although the similarity was low (88%). The betaproteobacterial cells (KC492704) were proven to predominantly inhabit the sponge through cloning-sequencing, pyrosequencing, catalyzed reporter deposition fluorescent in situ hybridization, and transmission electron microscopy, implying that they have potentially important roles in sponges [5]. Denovo205 was the most abundant OTU in Clce1S and had the closest similarity with a sequence (DQ889898, 88%) that had been detected from the octocoral Erythropodium caribaeorum in an unpublished study. The low similarity of denovo205 with other sequences (<88%) implies that new sponge specimens can be reservoirs of novel bacteria that have been unexplored so far.

Many monophyletic and sponge-specific phylogenetic clusters in Bacteria and Archaea have been described in previous papers [38,40]. Among them, the cluster within Betaproteobacteria is one of the largest sponge-specific clusters that branched deeply from cultivated strains and is composed only of sequences obtained directly from environments [38]. Denovo205 and denovo624 were also included in this sponge-specific cluster, as shown in Fig. 4.

Fig. 4.Neighbor-joining tree of predominant OTUs. A 16S rRNA gene similarity of 94% was used for OTU determination to reduce the number of OTUs. Bootstrap values above 70% are shown at the branches.

Through this study, it was revealed that the bacterial communities of some sponges were very simple but differed from one another, and that some predominant OTUs of those sponges showed high dissimilarity with known sequences, phylogenetically forming a sponge-specific clade. Considering that about 8,500 species of sponges have been confirmed taxonomically and the total number of species was estimated to be double that many [43], while only dozens of studies have been performed to reveal their bacterial diversities, much unexplored microbial diversity of sponges remains to be revealed. Many more studies should be carried out to further understand the microbes in sponges.