FEMS Microbiology Ecology, 91, 2015 doi: 10.1093/femsec/fiu006 Advance Access Publication Date: 5 December 2014 Research Article
Bacterial communities associated with four ctenophore genera from the German Bight (North Sea) ¨ Peplies3 and Antje Wichels2 Wenjin Hao1,∗ , Gunnar Gerdts2 , Jorg 1
Binzhou Medical University, Yantai 264003, China, 2 Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Biologische Anstalt Helgoland, PO Box 180, 27498 Helgoland, Germany and 3 Ribocon GmbH, 28359 Bremen, Germany ∗ Corresponding author. Binzhou Medical University, 346 Guanhai Road, Yantai 264003, China. Tel: 0086 18553592064; E-mail: [email protected]
One Sentence Summary: The observed bacterial communities in ctenophores form specific associations. Editor: Dr Gary King
ABSTRACT Intense research has been conducted on jellyfish and ctenophores in recent years. They are increasingly recognized as key elements in the marine ecosystem that serve as critical indicators and drivers of ecosystem performance and change. However, the bacterial community associated with ctenophores is still poorly investigated. Based on automated ribosomal intergenic spacer analysis (ARISA) and 16S ribosomal RNA gene amplicon pyrosequencing, we investigated bacterial communities associated with the frequently occurring ctenophore species Mnemiopsis leidyi, Beroe sp., Bolinopsis infundibulum and Pleurobrachia pileus at Helgoland Roads in the German Bight (North Sea). We observed significant differences between the associated bacterial communities of the different ctenophore species based on ARISA patterns. With respect to bacterial taxa, all ctenophore species were dominated by Proteobacteria as revealed by pyrosequencing. Mnemiopsis leidyi and P. pileus mainly harboured Gammaproteobacteria, with Marinomonas as the dominant phylotype of M. leidyi. By contrast, Pseudoalteromonas and Psychrobacter were the most abundant Gammaproteobacteria in P. pileus. Beroe sp. was mainly dominated by Alphaproteobacteria, particularly by the genus Thalassospira. For B. infundibulum, the bacterial community was composed of Alphaproteobacteria and Gammaproteobacteria in equal parts, which consisted of the genera Thalassospira and Marinomonas. In addition, the bacterial communities associated with M. leidyi display a clear variation over time that needs further investigation. Our results indicate that the bacterial communities associated with ctenophores are highly species- specific. Key words: bacterial community composition; ctenophore; ARISA fingerprint; ribosomal amplicon pyrosequencing
INTRODUCTION Ctenophores (comb jellies) represent a distinct phylum of gelatinous invertebrates that are ubiquitous in all marine environments. Blooms and invasions of ctenophores have been documented in many estuarine, coastal and open-ocean ecosystems worldwide over the past several decades (Purcell and Arai 2001; Sullivan, Van Keuren and Clancy 2001; Purcell 2005; Condon and Steinberg 2008; Fuentes et al., 2010). It has been hypothesized that ctenophores benefit from multiple changes in the ocean
that are attributed to anthropogenic impacts, such as eutrophication, overfishing and global warming (Mills 2001; Graham and Bayha 2007; Purcell 2012; Condon et al., 2013). The numbers of blooms continue to increase globally, and some predictions suggest that these gelatinous animals may ultimately become dominant in the oceans in the future instead of fish-dominated systems (Purcell, Uye and Lo 2007; Pitt, Welsh and Condon 2009; Richardson et al., 2009). Ctenophores and jellyfish play an important role in the marine ecosystem by substantially affecting the structure of the
Received: 16 July 2014; Accepted: 25 October 2014 C FEMS 2014. All rights reserved. For permissions, please e-mail: [email protected]
FEMS Microbiology Ecology, 2015, Vol. 91, No. 1
planktonic food web (Schneider and Behrends 1998; Brodeur et al., 2002; Sommer et al., 2002; Purcell 2003). In addition, they release a large amount of nutrients and dissolved organic matter (DOM) (Nemazie, Purcell and Glibert 1993; Schneider and Behrends 1998; Pitt et al., 2009; Condon et al., 2011) through their metabolic activities, thus presumably directly stimulating bacterial growth and potentially influencing bacterial community composition (BCC) (Hansson and Norrman 1995; Condon et al., 2011). Condon et al. (2011) suggested that the microbial loop was promoted by the jelly-like DOM produced by ctenophores, which was readily available for heterotrophic bacteria. Furthermore, the bacterial community of the water column shifted from Alphaproteobacteria to Gammaproteobacteria in response to the addition of the jelly DOM. These findings directly indicate the ability of ctenophores to affect carbon cycling and the structure of free-living bacterial communities. Studies of bacterial communities associated with marine animals have largely focused on sponges, corals, bryozoans and crustaceans (Frias-Lopez et al., 2004; Wang et al., 2004; Kittelmann and Harder 2005; Ritchie 2006; Webster et al., 2010). It has been demonstrated that bacterial communities associated with marine invertebrates differ from those in the water column and display host-specificity (Rohwer et al., 2002; Thakur, Anil and Muelle 2004; Webster et al., 2010). They can also vary between habitats and seasons (Friedrich et al., 2001; Wichels et al., 2006; Sharp et al., 2007; Bickel and Tang 2014). The outer surface of marine organisms represents the major physiological interface with the environment. The bacterial colonization of a ‘living’ surface may be influenced by several factors, including the age of the colonized organism and the release of organic metabolites or extracellular polymers, which has been exemplified for various marine invertebrates such as ascidians (Wahl, Jensen and Fenical 1994), corals (Neulinger et al., 2008), sponges (Thakur et al., 2003) and bryozoans (Kittelmann and Harder 2005). These epibiotic bacteria have been reported to pose direct positive or negative effects on the colonized organism, such as the interference with gas and nutrient exchange (Wahl et al., 2011) and susceptibility to diseases (Ritchie 2006), but are also involved in the development and evolution of the organism (McFall-Ngai and Ruby 1991; Nyholm and McFall-Ngai 2004). To date, the bacterial communities associated with ctenophores have not been fully characterized and the functional roles have not been determined. Two studies have recently been published examining this area of research. Daniels and Breitbart (2012) was the first to publish on the community composition and found that specimens of two ctenophore species (M. leidyi and Beroe ovata) from Tampa Bay contained fewer bacterial operational taxonomic units (OTUs) by T-RFLP (terminal restriction fragment length polymorphism) and a lower diversity as revealed by 16S rRNA clone library analysis when compared with the water column. Interestingly, Dinasquet, Granhag and Riemann (2012) subsequently applied 454 pyrosequencing of 16S rRNA genes and found a similar BCC associated with M. leidyi in the Gullmar fjord off the west coast of Sweden. The aim of this study was to investigate the BCC of different ctenophore species in the German Bight (North Sea) over a period of one year. Mnemiopsis leidyi, which originally inhabited the Atlantic of North and South America (GESAMP 1997; Purcell et al., 2001), has invaded the Black, Azov, Marmara and Aegean Seas during the last two decades (Purcell et al., 2001) and was recently identified as an invasive species at Helgoland Roads (Boersma et al., 2007; Hamer, Malzahn and Boersma 2011). Conversely, Beroe sp., B. infundibulum and P. pileus are indigenous ctenophores at Helgoland Roads (Greve 1970).
MATERIALS AND METHODS Sample collection and preparation Individual specimens of ctenophores M. leidyi, Beroe sp., B. infundibulum and P. pileus were collected at Helgoland Roads in the German Bight (54◦ 11.3 N, 7◦ 54.0 E) from October 2009 to October 2010. The samples were obtained three times per week using a 500 μm mesh trawl towed by the research vessel Aade and transferred to the laboratory within 2 h. Specimens were observed under a dissecting microscope for morphological identification based on Greve (1975), Faasse and Bayha (2006) and Byern, Mills and Flammang (2010). Ten intact individuals of each species were collected each sampling day and were rinsed five times with sterile seawater (0.2 μm filtered and autoclaved) to remove transient and loosely associated microorganisms from the surfaces of the ctenophores. All specimens were checked to ensure that there were no visible gut contents, stored at −20◦ C and freeze-dried prior to DNA extraction.
DNA extraction Total genomic DNA was extracted from freeze-dried tissue using CTAB (cetyltrimethylammonium bromide) according to the modified protocol of Gawel and Jarret (1991). Ctenophore samples were homogenized using a sterile mortar and pestle. Aliquots from ground samples (1 mg) were transferred to 2 mL pre-heated (60◦ C) CTAB buffer (3% CTAB, 1.4 M NaCl, 20 mM EDTA, 100 mM Tris-HCl, 0.2% mercaptoethanol) and incubated at 65◦ C for 30 min. In total, 1 mL STE buffer (6.7% saccharose, 50 mM Tris, 1 mM EDTA, pH 8), lysozyme (10 mg/mL) and proteinase K (10 mg/mL) were added to the samples and incubated at 50◦ C for an additional 30 min. DNA extraction was performed twice using phenol–chloroform–isoamylalcohol (25 : 24 : 1), followed by DNA precipitation with isopropanol overnight at −20◦ C. The DNA extracts were washed with 75% ethanol and finally dried on a sterile bench. All DNA extracts were dissolved in 30–50 μl sterile water and served as template DNA for PCR. The quantity and quality of extracted DNA were determined by microphotometry using an Infinite 200 NanoQuant microplate reader (Tecan, Maennedorf, Switzerland).
Automated ribosomal intergenic spacer analysis (ARISA) To generate bacterial community profiles, the ARISA method, a PCR method based on the length polymorphism of the internal transcribed spacer (ITS) (Fisher and Triplett 1999; Ranjard, Brothier and Nazaret 2000) was applied. The ITS region was amplified with the forward primers L-D-Bact-132-a A-18 (5 -CCG GGT TTC CCC ATT CGG-3 ) and the fluorescently infrared labelled reverse primer S-D-Bact-1522-b-S-20 (5 -TGC GGC TGG ATC CCC TCC TT-3 ) (Ranjard et al., 2000). The PCR reaction and cycling conditions were performed as previously described (Krause et al., 2012), and 50 ng of genomic DNA template was applied in each PCR reaction. Based on the intensities of PCR products on agarose gels (1.4%, 1 h, 80 V), the original or diluted PCR products were mixed with an equal volume of stop mix and separated on 5.5% polyacrylamide gels prepared according to the manufacturer’s protocol (LI-COR Biosciences, Lincoln, NE, USA). A 50–1500 bp standard was applied as a size reference (all materials: LI-COR Biosciences, Lincoln, NE, USA). The samples were run on a LI-COR 4300 DNA Analyzer (LI-COR Biosciences, Lincoln, NE, USA) for 14 h at 1500 V. A digital image generated by the infrared fluorescence detector was used for further analyses.
Hao et al.
16S ribosomal amplicon pyrosequencing Based on the significant differences in community structure as revealed by the ARISA fingerprints, 96 DNA subsamples from M. leidyi, 44 from Beroe sp. 44 from B. infundibulum and 18 samples from P. pileus were chosen and pooled for each ctenophore. The tag PCR approach and the sequencing approach were performed with these four pooled samples by LGC Genomics (Berlin, Germany). The V1–V6 region of the 16S RNA gene was amplified (ca. 900 bp) using the following primer set: forward GM3 5 -AGAGTTTGATCMTGGC-3 and reverse 907R 5 CCGTCAATTCMTTTGAGTTT-3 (Krause et al., 2012; Teeling et al., 2012). Each PCR reaction (20 μl) contained 1U Kappa 2G robust polymerase (Biocat, Heidelberg, Germany), Kappa buffer A (Biocat, Heidelberg, Germany), Biostab PCR Optimizer (II) (SigmaAldrich, Steinheim, Germany), 25 mmol MgCl2 (Sigma-Aldrich, Steinheim, Germany), 10 mmol dNTPs and 10 μmol of each primer. As a template, 10–20 ng of DNA was added. The protocol consisted of 30 cycles: 1 min at 96◦ C, 15 sec at 96◦ C, 20 sec at 45◦ C and 60 sec at 72◦ C. Next, an exonuclease digestion containing 2 μl PCR product and 10 Units Exo I (NEB) for 30 min at 37◦ C and 15 min at 80◦ C was performed. Finally, tag PCR was performed, and the conditions for tag PCR were as described above; however, instead of genomic DNA the digested PCR products (220-fold dilution) and tagged primers were applied. The first three annealing cycles were run at 45◦ C, and the last three were run at 55◦ C. The sequencing was performed in a 454 Roche Genome Sequencer FLX + Titanium with a quarter plate.
Statistical analysis Analysis of ARISA data The ARISA gel images were analysed using BioNumerics 6.6 software (Applied Maths, Sint-Martens-Latem, Belgium). The normalization of band patterns was conducted by referencing the size standard. Bands with intensities lower than 5% of the maximum value of the respective lane were neglected. Binning to band classes was performed according to Kovacs (2010) and Brown (2005), and bands smaller than 300 bp were omitted from subsequent analyses. The generated ARISA band classes, each representing one OTU, were further transformed as a binary presence absence table and subsequently analysed without further transformation.
Multivariate analyses The ARISA band class tables were used to investigate the differences among four ctenophore species regarding the associated bacterial community structure. A permutational multivariate ANOVA (PERMANOVA) with fixed factors was applied based on the Jaccard coefficient (PRIMER Version 6, PRIMER-E Ltd, Lutton, UK) (Clarke and Warwick 2001; Clarke and Gorley 2006). Principal coordinate analysis (PCO) was performed to visualize the patterns of the bacterial community that were influenced by species. For all multivariate analyses, we used PRIMER 6 with the add-on package PERMANOVA+ (PRIMER-E Ltd, Plymouth, UK).
Processing of pyrosequencing data Sequence reads from 16S amplicon pyrosequencing were processed by the bioinformatics pipeline of the SILVA rRNA gene database project (Pruesse 2007) as first described by Teeling et al. (2012) and Ionescu et al. (2012). The input data were sorted and trimmed to remove any adapter or barcode by the sequencing
company. Dereplication (identification of identical reads ignoring overhangs) and clustering at a 98% sequence identity threshold (OTU definition based on the non-redundant subset of reads) were applied to filter and collapse the sequence data. Subsequently, one reference read of each OTU was classified against the SILVA taxonomy using the SILVA classifier and the result was mapped onto all reads that were assigned to the respective OTU before. This yields quantitative information (number of individual reads per taxonomic path) within the limitations of PCR and sequencing technique biases. Reads without significant classification results according to the SILVA rules were assigned to the meta group ‘No Relative’. Sequence data were deposited in the NCBI Sequence Read Archive (accession numbers SRR1391031, SRR1391032, SRR1391060 and SRR1391071). The relative abundances of identified taxonomic groups were calculated as percentages, and the BCC profiles were represented as bar charts to compare between ctenophore species. However, it should be noted that the numbers of the sequence reads only reflect the rDNA abundances in the amplicon pool and are therefore an approximation of the natural abundance.
Phylogenetic analysis R The sequences were phylogenetically analysed using the ARB software package (Ludwig et al., 2004). The sequence reads representing OTUs of the Marinomonas cluster were added to the ARB SSU database (release May 2005). The DNA sequences of Marinomonas collected from two other publications (Daniels and Breitbart 2012; Dinasquet et al., 2012) using ctenophores and the bacterial community were added. In addition, the alignment was refined by comparison of the closest relatives in NCBI retrieved by BLASTN in the event that they were not found in the database. The sequence alignment was performed with the integrated Fast Aligner. Finally, sequences with more than 1200 nucleotides were used to calculate a neighbour-joining tree. 1000 replicates were performed and the robustness is expressed in % as bootstrap values on the nodes of the tree. Partial sequences were added using the ARB ‘parsimony interactive’ tool.
RESULTS BCC of different ctenophore species Four ctenophore species, including M. leidyi, Beroe sp., B. infundibulum and P. pileus, occurred at Helgoland Roads from October 2009 to October 2010. However, M. leidyi was present throughout the entire sampling period except for May 2010 and all other species were absent from January to March in 2010 (Table 1). In total, 496 ctenophore specimens were collected and analysed. Mnemiopsis leidyi was abundant except for May 2010, and a total of 354 individuals were collected. By contrast, Beroe sp. and P. pileus occurred in lower numbers. In total, 44 individuals of Beroe sp. were collected except in August 2010, and 42 individuals of P. pileus were collected in May, June and July 2010. Bolinopsis infundibulum only appeared in June 2010 when 56 individuals were collected. Based on the ARISA fingerprints, the PCO plot depicted the BCC of all specimens of the four ctenophore species (M. leidyi, Beroe sp., B. infundibulum and P. pileus) (Fig. 1). The bacterial communities of the different ctenophore species were well separated in PCO ordination, which was confirmed by PERMANOVA main test (Table 2). The pairwise comparisons further illustrated that the four ctenophore genera contained distinctly different bacterial communities (P = 0.001) (Table 3). The highest amount of
FEMS Microbiology Ecology, 2015, Vol. 91, No. 1
Table 1. Amount of ctenophores collected at Helgoland Roads during the sampling period from November 2009 to October 2010. (The numbers shown here are summed up by weeks of each month). Month. Year Species
M. leidyi Beroe sp. B. infundibulum P. pileus
52 2 0 0
38 5 0 0
0 0 10 0
1 3 28 57
98 6 3 0
74 0 0 0
42 9 0 0
49 19 0 0
Table 3. PERMANOVA pairwise comparisons of BCC associated with four ctenophore species based on Jaccard dissimilarities of ARISA profiles. Comparison
Beroe sp. vs B.infundibulum Beroe sp. vs M. leidyi Beroe sp. vs P. pileus B. infundibulum vs M. leidyi B. infundibulum vs P. pileus M. leidyi vs P.pileus
3.4608 4.0271 3.1212 5.7974 3.6843 4.7464
0.001 0.001 0.001 0.001 0.001 0.001
998 999 997 999 998 996
Significant results (P (perm) < 0.05) are highlighted in bold. Displayed are pairwise a posteriori comparisons of the factor ‘species’.
Figure 1. PCO presenting the bacterial communities associated with four ctenophore species based on Jaccard coefficient from ARISA profiles.
variation occurred between M. leidyi and the other three species (t statistic). The bacterial richness (alpha diversity) (Fig. S1, Supporting Information), as estimated by the ARISA-OTU numbers, was variable among all four species. The highest OTU richness was recorded in Beroe sp. (S = 36), followed by B. infundibulum (S = 22) and M. leidyi (S = 21), whereas the lowest richness was observed in P. pileus (S = 5).
Temporal variation of the BCC of M. leidyi Mnemiopsis leidyi was the only species that was collected throughout most of the sampling period except for May (no occurrence) and June 2010 (1 specimen). Hence, the seasonal trend in the BCC was only analysed for this species. In the PCO plot (Fig. 2), the BCCs of summer samples (July and August) clearly clustered together and were separated from those of autumn and winter (September, October, November and December.). Consistent with the PCO ordination, PERMANOVA main test and pairwise comparisons of BCCs revealed significant dif-
ferences regarding the ‘month’ factor (P = 0.001) (Tables 4 and 5). This indicated a significant seasonal shift in the bacterial community of M. leidyi.
Pyrosequencing data analysis Based on the findings of the ARISA analyses, samples from each ctenophore species were pooled and subjected to 16S ribosomal amplicon pyrosequencing. PCO ordination and PERMANOVA analyses again displayed significant differences among the BCCs of the four ctenophore species (Fig. 3, Tables 6 and 7) regarding to this subset samples. The similar variation for the subset of samples and the total samples (Sq. root, Tables 6 and 2) indicates that the subset of samples represents the bacterial community associated with each ctenophore species. In total, 244 885 raw 16S rDNA pyrosequencing reads were obtained from the four ctenophore species: M. leidyi, Beroe sp., B. infundibulum and P. pileus. After removing low-quality sequences, 128 244 reads were used for subsequent analysis, including 21 721 of Beroe sp., 21 118 of B. infundibulum, 28 318 of M. leidyi and 57 087 of P. pileus. Singletons (n = 1) and rare reads (