Environmental Microbiology Reports (2017) 9(3), 310–315 doi:10.1111/1758-2229.12541 Correspondence of coral holobiont metabolome with symbiotic bacteria, archaea and Symbiodinium communities Emilia M. Sogin,1*† Hollie M. Putnam,1‡ Introduction Craig E. Nelson ,2 Paul Anderson3 and Microbial symbiotic partners mediate biochemical trans- Ruth D. Gates1 formations that contribute to host performance, altering 1Hawai‘i Institute of Marine Biology, University of Hawai‘i individual survival and ecosystem function (McFall-Ngai at Manoa, Kane‘ohe, HI, USA. et al., 2013). Reef-building corals partner with a diverse 2Department of Oceanography and Sea Grant College assemblage of bacterial, archaeal and eukaryotic organ- Program, Center for Microbial Oceanography: Research isms (i.e. microbiome) that comprise the coral holobiont and Education, University of Hawai‘i at Manoa, and cycle organic molecules necessary for life in Honolulu, HI, USA. oligotrophic waters (Rohwer et al., 2002; Gates and 3Department of Computer Science, College of Ainsworth, 2011). Coral-associated micro-algae from the Charleston, Charleston, NC, USA. dinoflagellate genus Symbiodinium help to sustain large organic biomasses on reefs through carbon production Summary (Muscatine and Porter, 1977), while coral-associated bacterial and archaeal assemblages participate in sulfur Microbial symbiotic partners, such as those associ- (Raina et al., 2009) and nitrogen cycling (Radecker€ ated with Scleractinian corals, mediate biochemical et al., 2015). The breakdown of these symbiotic relation- transformations that influence host performance and ships (e.g. bleaching and disease), leads to mortality survival. While evidence suggests microbial commu- and drastic changes in reef structure (Hughes et al., nity composition partly accounts for differences in 2003). Because not all symbiotic partners contribute coral physiology, how these symbionts affect meta- equally to holobiont performance (Stat et al., 2008), bolic pathways remains underexplored. We aimed to investigating how different microbes influence coral assess functional implications of variation among physiology is essential to understanding the resilience in hospite coral-associated microbial partners .To and resistance of reef ecosystems in a changing ocean this end, we characterized and compared metabolo- environment. To this end, we sought to link microbial mic profiles and microbial community composition community composition to holobiont biochemical profiles from nine reef-building coral species. These data using deep sequencing of marker genes and a holistic demonstrate metabolite profiles and microbial com- metabolomic approach. munities are species-specific and are correlated to Porites one another. Using spp. as a case study, we Results and discussion present evidence that the relative abundance of dif- ferent sub-clades of Symbiodinium and bacterial/ We sampled the microbial community and metabolite com- archaeal families are linked to positive and negative position from nine coral species within four genera from a metabolomic signatures. Our data suggest that while single fringing reef in Mo’orea, French Polynesia (Fig. S1; some microbial partners benefit the union, others are Supporting information provides full experimental method- more opportunistic with potential detriment to the ology). We characterized both Symbiodinium and bacterial/ host. Consequently, coral partner choice likely influ- archaeal communities from each sample by sequencing ences cellular metabolic activities and, therefore, hol- the internal transcribed spacer 2 (ITS2) and the V3-V4 obiont nutrition. region of the 16s ribosomal RNA gene, respectively. Using custom bioinformatics pipelines (FileS1 and FileS4), our Received 9 December, 2016; revised 25 January, 2017; accepted analysis identified 80 Symbiodinium operational taxonomic 12 March, 2017. *For correspondence. E-mail esogin@mpi-bremen. units grouped to 97% similarity (OTUs; assigned to clades de; Tel. (149) 421 2028 823; Fax (149) 421 2028 580. Present A, C, D and G) and 618 bacterial/archaeal families through addresses: †Symbiosis Department, Max Planck Institute for Marine Microbiology, Bremen, DE. ‡Biological Sciences, University of database classification (Supporting information Fig. S2; Rhode Island, Kingston, RI, USA. Table S1); the majority of these families belong within the VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd The coral metabolome links to symbiont communities 311 gamma-, alpha- and delta-proteobacteria but with repre- using only the P. lobata and P. r u s samples still detected sentatives spread across 130 classes (Supporting informa- congruency between microbial communities and metab- tion Fig. S2B). In parallel, we assessed coral metabolite olite composition (Fig. 1 and I). This suggests different profiles from crude extracts using proton-nuclear magnetic microbial partners may play similar roles in holobiont resonance spectroscopy (1H-NMR) based metabolomic nutrition and provide functional redundancy in the sys- techniques (Sogin et al., 2014). Our spectral binning tem, such that metabolic processes are retained (Yin approach identified 197 peaks from aligned NMR profiles, et al., 2000). While our approach advances the field by which were used in subsequent downstream analyses. We providing a framework to investigate the roles of the unambiguously identified 8 compounds through database symbiotic partners in a complex milieu, further metage- and literature matching, and assigned the remaining sig- nomic approaches will provide a detailed comparison nals to either carbohydrate, lipid or aromatic metabolite between the functions of different microbes (Dinsdale classes based on spectral identity (Supporting information et al., 2008). Our findings lend weight to growing infor- Fig. S3; Bross-Walch et al., 2005). mation on the metabolic role of mixed microbial commu- Our sequencing and metabolomic profiling data nities, with various Symbiodinium types characterized by allowed us to test for linkages between microbial com- different chemical profiles (Klueter et al., 2015) and munity assemblages and metabolite composition. This shifts in associated bacterial/archaeal communities alter- novel in hospite approach provides direction for future ing the metabolic potential of the holobiont (Wegley efforts seeking to identify specific roles of uncultured et al., 2007). complex mixtures of microbial communities in reef cor- Using this reduced dataset containing only Porites sp. als. Multivariate analysis of Symbiodinium and coral with similar metabolite profiles, we also calculated metabolite profiles grouped samples into coral species- spearman rank correlation coefficients between the and genera-specific clusters (Fig. 1A and C) while bac- abundance of individual NMR peaks and independently teria/archaea communities only sometimes separated by the relative abundances of both Symbiodinium and bac- genera (Fig. 1B). These results confirm previous studies terial/archaeal symbionts. We observed significant corre- showing that closely related host taxa share similar sym- lations between coral-associated symbionts and biotic communities and biochemical profiles (Rohwer metabolites (multiple comparisons corrected P < 0.05), et al., 2002; Putnam et al., 2012; Sogin et al., 2014). further supporting the hypothesis that microbial Using a Procrustes rotation to statistically compare the communities alter holobiont metabolomic profiles multidimensional structure of each dataset, our study (Fig. 2). Hierarchical clustering on correlation patterns demonstrates congruency among coral Symbiodinium group microbial partners based on their association pat- and bacterial communities with metabolomic profiles terns with metabolite peaks, highlighting taxa that are (Fig. 1 D and E). Furthermore, these results are robust positively and negatively associated with small com- to our choice of microbial phylogenetic binning. ANOSIM pounds. While additional efforts are required to deter- tests of bacterial/archaeal community structure differ- mine if these associations are benefiting or harming the ences among coral species (R 5 0.75, P 5 0.001) and holobiont, we can interpret these relationships in context Procrustes correspondence tests of multivariate congru- of the current physiological understanding of coral micro- ency with metabolite composition (P 5 0.001) were con- bial communities. Therefore, using these data, we sistent when bacterial/archaeal OTUs were constructed explore the hypothesis that some symbionts positively using a 97% sequence similarity criterion; we use contribute to, while others may act in an opportunistic or family-level classifications herein as a conservative parasitic manner, with respect to holobiont nutrition approach. (Lesser et al., 2013). Corals host a diverse assemblage of symbiotic part- Clade C Symbiodinium fixes and translocates more ners; therefore, finding correlative relationships between organic carbon than other clades (Stat et al., 2008); our metabolites and microbial members across multiple host observation that OTU75 assigned to sub-clade C15 cor- species is not surprising. Considering different corals relates positively with carbohydrates but negatively with exhibit broadly different physiologies (Loya et al., 2001), lipids suggests this sub-type produces metabolically it is essential to isolate the influence of the host to iden- active carbohydrates not used in lipid metabolism (Fig. tify the effect of microbial communities