Metagenomic characterisation of viral communities in corals: Mining biological signal from methodological noise1

Running title: Methodological biases in coral viromics

Elisha M. Wood-Charlson1,4,5, Karen D. Weynberg1, Curtis A. Suttle2, Simon Roux3,5,

Madeleine J. H. van Oppen1

1 Australian Institute of Marine Science, Townsville, QLD, Australia

2 Departments of Earth, Ocean & Atmospheric Sciences, Microbiology & Immunology,

Botany and the Canadian Institute for Advanced Research, University of British Columbia,

Vancouver, British Columbia, Canada

3 LaboratoireArticle Micro-organismes: Genome and Environment, Université Blaise Pascal,

Clermont Université, France

4 Correspondence:

Elisha M. Wood-Charlson

Australian Institute of Marine Science

PMB 3 Townsville MC

Townsville, QLD 4810, Australia

E-mail: [email protected]

5 Current affiliations: Wood-Charlson, Center for Microbial Oceanography: Research and

Education, University of Hawai’i at Manoa, Honolulu, HI, USA; Roux, Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ, USA

This article has been accepted for publication and undergone full peer review but has not been through the

copyediting,Accepted typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/1462-2920.12803

1 This article is protected by copyright. All rights reserved.

Summary

Reef-building corals form close associations with organisms from all three domains of and therefore have many potential viral hosts. Yet, knowledge of viral communities associated with corals is barely explored. This complexity presents a number of challenges in terms of the metagenomic assessments of coral viral communities, and requires specialised methods for purification and amplification of viral nucleic acids, as well as virome annotation. In this mini-review, we conduct a meta-analysis of the limited number of existing coral virome studies, as well as available coral transcriptome and metagenome data, to identify trends and potential complications inherent in different methods. The Article analysis shows that the method used for viral nucleic acid isolation drastically affects the observed viral assemblage and interpretation of the results. Further, the small number of viral reference genomes available, coupled with short sequence read lengths might cause errors in identification. Despite these limitations and potential biases, the data show that viral communities associated with corals are diverse, with double- and single-stranded

DNA and RNA . The identified viruses are dominated by dsDNA-tailed , but there are also viruses that infect hosts, likely the endosymbiotic , Symbiodinium spp., host coral, and other in close association.

Keywords: coral, Symbiodinium, metagenome, virome, virus, methodological biases Accepted

2 This article is protected by copyright. All rights reserved. Introduction

Scleractinian (stony) corals deposit calcium carbonate during colony growth and build the three-dimensional structures that constitute coral reefs. This structural framework provides important habitat for a wide taxonomic diversity of macroscopic and microscopic reef organisms. The coral itself hosts a collective of organisms, referred to as the coral holobiont, that spans the three domains of life, as well as the viruses that infect them. Of all the components in the coral holobiont, viruses are the least studied. Transmission electron microscopy (TEM) was used in earlier publications to describe virus-like particles (VLPs) in corals (Wilson et al., 2001; 2005; Davy and Patten, 2006; Patten et al., 2008). With the development of next-generation sequencing, a metagenomics approach to viral community Article characterisation is now also available (Edwards and Rohwer, 2005; Kristensen et al., 2010;

Rosario and Breitbart, 2011). However, there are inherent challenges to the purification of viruses and viral genomes. The small size of virus particles and their genomes, combined with losses during purification typically results in low amounts of nucleic acids, on which whole genome amplification (WGA) is often used to obtain enough material for sequencing; this can introduce quantitative biases in the data (Angly et al., 2006; Duhaime et al., 2012).

Moreover, the diverse genome chemistry of viruses (single stranded (ss)RNA, double stranded (ds)RNA, ssDNA, or dsDNA) complicates nucleic acid isolation from mixed communities (Andrews-Pfannkoch et al., 2010; Weynberg et al., 2014).

Viral metagenomes prepared from coral tissues require careful post-sequencing processing.

Most marine viruses lack representation in sequence databases, with coral-associated viruses being particularly scant. While there have been some phage therapy trials for

bacteriaAccepted associated with coral disease (Atad et al., 2012; Cohen et al., 2013), cultured

3 This article is protected by copyright. All rights reserved. viruses associated with a healthy coral holobiont have not been reported. The paucity of representative viral sequences in databases results in many unidentified sequences in marine viral metagenomes (Breitbart et al., 2002; Angly et al., 2006; Wegley et al., 2007;

Dinsdale et al., 2008; Vega Thurber et al., 2008; Williamson et al., 2008; Correa et al., 2013).

Further, sequences with similarity to cellular genes are often present in viral metagenomes

(Roux et al., 2013a), which may be the result of host contamination, horizontal gene transfer (HGT) between viruses and their hosts, or gene transfer agents (GTAs) (Canchaya et al., 2003; Monier et al., 2009; Liu et al., 2011; Lang et al., 2012; Roux et al., 2013a). All of these factors are likely important in the coral holobiont, given its wide diversity of organisms that live in close association. Article

In spite of these caveats, holobiont metagenome and transcriptome sequence data can provide qualitative information about viruses associated with various members of the coral holobiont. Regardless of whether the focus of these studies is the coral host tissue, the algal symbiont, or the prokaryotes, it is difficult to separate the host organisms from the associated viral community. Therefore, raw sequence data from non-viral fractions will likely contain viral sequences. In this contribution, publically available coral transcriptomes, as well as prokaryote and viral metagenomes isolated from corals, were interrogated for viral sequences and analysed comparatively.

The data sets

Coral, prokaryotic, and viral metagenome and meta-transcriptome data sets were downloaded from publically available sequence archives and uploaded to Metavir (Roux et

al., 2011),Accepted an online tool for analysing viral genomic data. Sequences from 11 coral species

4 This article is protected by copyright. All rights reserved. (Supplementary Table 1) were added to Metavir as “Coral-associated viruses” for this study.

In this analysis, we also included “P. compressa” (Vega Thurber et al., 2008), “Coral Atoll”

(Dinsdale et al., 2008), and several data sets from the “Coral virus – generating metagenomes” (Weynberg et al., 2014; Supplementary Table 1). Metagenomes were analysed using Metavir’s BLAST-based comparison (e-value ≤ 10-5) to the 2014-09-10 NCBI viral refseq database and normalized to genome length using the built-in Genome-relative

Abundance and Average Size (GAAS) normalization tool (Angly et al., 2009). Metavir assigns taxonomy to a viral sequence using this BLAST-based comparison to sequences that are annotated in the NCBI viral refseq database.

Article Methodological biases: Purification and amplification techniques select for certain viral groups

Coral-associated data sets were screened for the presence of viruses identified to a viral family, where possible (Table 1). All data sets contained dsDNA viruses, but were variable for ssDNA, dsRNA, ssRNA, and retrotranscribing (RT) viral families, depending on the holobiont fraction targeted for sequencing. Coral transcriptomes (Supplementary Table 1) were generated from extracted holobiont RNA, and they produced viral assemblages dominated by RNA viruses (36-78%). While this may be expected, since transcriptomes target RNA sequences (Supplementary Table 2), it is possible that they may also contain

DNA viruses, if the sample was undergoing an active DNA virus replication event. RNA viral assemblages from coral transcriptomes contained ssRNA viruses (1-13% of identified virus), a dsRNA virus (2-31%), and ssRNA and dsDNA RT viruses (29-71%). Prokaryote metagenomes (Supplementary Table 1) targeted the prokaryotic community through Percoll

fractionationAccepted (Supplementary Table 2) to remove coral host and Symbiodinium cells. The

5 This article is protected by copyright. All rights reserved. prokaryote-containing fractions were extracted for DNA and amplified through a Phi29 DNA polymerase-based rolling circle WGA method to obtain enough material for sequencing

(Littman et al., 2011; Wegley et al., 2007). The resulting viral assemblages were surprisingly depleted for dsDNA viruses (2-8% of identified viruses), and had a notably larger representation of ssDNA viruses (31-96%). In fact, the Porites astreoides prokaryotic data set

(Wegley et al., 2007) was dominated by hits to a single ssDNA virus, an unclassified

Dragonfly-associated microphage (48%). With almost half of the viral assemblage from this prokaryotic metagenome annotating to a single ssDNA virus, it is important to acknowledge that Phi29-based amplification may preferentially amplify ssDNA templates (Kim and Bae,

2011). Several of the viral DNA metagenomes were also amplified by Phi29 (Supplementary Article Table 2), but they were highly variable in their proportions of dsDNA and ssDNA viruses. For example, the Northern Line Islands seawater DNA viral metagenomes (Dinsdale et al. 2008) were dominated by dsDNA (95-99% of identified viruses), and the coral Porites compressa

DNA viral metagenomes (Vega Thurber et al., 2008) ranged from 17-86% dsDNA and 5-81% ssDNA. It should be noted that all of these viral metagenome data sets were treated with chloroform prior to viral DNA extraction, which may directly influence the observed viral assemblages, as we discuss below.

For studies targeting viral metagenomes, there are inherent challenges to purifying the viruses from coral tissue. Removal of tissue from the surface of coral skeletons results in a thick mucilage that is often resistant to the filtration and concentration methods commonly used in marine viral ecology. Consequently, most of the published coral-associated viral metagenomes use chloroform to break up mucus and disrupt cell membranes, prior to

loadingAccepted onto a CsCl gradient for purification by centrifugation (Dinsdale et al., 2008;

6 This article is protected by copyright. All rights reserved. Marhaver et al., 2008; Vega Thurber et al., 2008; Soffer et al., 2013). Because certain virus types are sensitive to chloroform and other organic solvents (Ackermann, 2006), we developed an alternative, non-chemical approach to preparing coral-associated viral metagenomes, which can be used to isolate, purify, and amplify paired DNA and RNA genomes from a single coral tissue sample (Weynberg et al., 2014; Supplementary Table 2).

In order to assess the impact of sample preparation and amplification on the recovery of coral-associated viruses in greater detail, we conducted a comparative analysis on dsDNA and ssDNA viruses in these data sets, as these viral groups were best catalogued in the databases and were present in all of the coral-associated metagenomes (prokaryote or Article virus; Table 1). Taxonomic annotation was considered for viruses identified at the order or family level (instead of species or strain). The proportional abundance of DNA viruses in the orders and or ds/ssDNA families (as not all viruses are assigned to an order; ICTV, 2012) was calculated for each sample and grouped by the following sample methodologies: coral transcriptomes (n=12), prokaryotic metagenomes (n=3), and chloroformed (n= 14) and non-chloroformed viral metagenomes (n=6) (Table 1). The resulting ds- and ssDNA viral assemblages were clustered by non-metric multidimensional scaling (nMDS) using metaMDS (Vegan package in R), and the ordination plot showed significant clustering within the data sets (Figure 1, stress =0.123, p ≤ 0.001, ANOVA by distance matrices).

Using this statistical method, the coral transcriptomes (typically dominated by RNA viruses, with an average of 39% dsDNA and <0.1% ssDNA) formed a distinct cluster away from the

rest ofAccepted the data sets (Figure 1). The outlier, Pocillipora acuta 2012, is an assembled

7 This article is protected by copyright. All rights reserved. transcriptome with a higher proportion of dsDNA viral contigs that were not present in the unassembled transcriptomes. The composition of the P. compressa DNA viral metagenomes

(Vega Thurber et al., 2008), however, was highly variable (Figure 1, highlighted in yellow).

The chloroformed P. compressa viral assemblages that were dominated by dsDNA viruses clustered with the chloroformed assemblages from seawater collected at the Northern Line

Islands (Dinsdale et al., 2008) and tissue from the coral D. strigosa (Marhaver et al., 2008).

The ssDNA virus-dominated P. compressa data sets formed a separate group closer to the ssDNA virus-rich, prokaryote-targeted viral metagenomes. The non-chloroformed, sequence-independent, single-primer amplification (SISPA) PCR amplified DNA viral metagenomes from Weynberg et al. (2014)’s methods publication appear as a tight cluster Article close to, but distinct from, the others.

The non-chloroformed RNA viral assemblages clustered away from the DNA viral assemblages. This was especially evident in the paired Acropora tenuis DNA and RNA viral metagenomes (Weynberg et al. 2014), supporting the utility of this proposed methodology, as DNA and RNA viral communities should be quite distinct, though this dogma may be in flux with the recent discovery of a chimeric ssDNA/ssRNA genome, CHIV (Roux et al. 2013b).

Regardless, the RNA viromes contained variable proportions of DNA viruses. For instance,

72-75% of the identified viral sequences in the Montastraea cavernosa RNA viral metagenomes were identified as ssDNA (Correa et al., 2013). In the A. tenuis RNA viral metagenome, less than 20% of the total viral assemblage matched ds-/ssDNA viruses.

Instead, it was dominated by a ssRNA virus matching an RNA virus infecting the Heterocapsa circularisquama (HcRNAV, 56% of identified viruses; Weynberg

et al. 2014;Accepted Tomaru et al. 2014). A small number of reads in the M. cavernosa stressed RNA

8 This article is protected by copyright. All rights reserved. viral assemblage (0.3%) were also annotated HcRNAV, suggesting this virus may infect

Symbiodinium spp. (Correa et al., 2013).

The presence of sequences matching to DNA viruses in RNA viral metagenome preparations is perhaps indicative of an active replication event, or alternatively, the existing methods may still be limited in achieving a pure isolation of RNA viruses. In addition, it is possible that different amplification methods result in biases for different viral groups. The A. tenuis virome was amplified using a SISPA PCR-based approach, while the M. cavernosa RNA viral metagenomes were amplified using a whole transcriptome amplification (WTA) kit.

Unfortunately, until we are able to sequence an unamplified viral assemblage purified from Article coral host tissue, we will be unable to accurately quantify the different viral taxa in a single sample.

Methodological biases: Size matters – sequence lengths, viral diameters, and e-values

As observed in previous marine virus sequencing studies (Breitbart et al., 2002; Angly et al.,

2006; Wegley et al., 2007; Dinsdale et al., 2008; Vega Thurber et al., 2008; Williamson et al.,

2008; Correa et al., 2013), the majority of sequences in the coral-associated virus data sets did not have a significant match to a viral genome reference sequence (data available for viewing at Metavir). However, viral metagenomes where the majority of reads were longer than 200 bp showed a larger proportion of significant hits (>17-43%), as previously observed by Wommack et al. (2008) for aquatic viral metagenomes in general. Since early inferences regarding viruses associated with the coral holobiont were limited by short read data sets, these conclusions should be re-examined. Fortunately, technological developments are

rapidlyAccepted overcoming these limitations as newer sequencing platforms are generating more

9 This article is protected by copyright. All rights reserved. and longer reads, which allow assembly of longer contigs and potentially whole viral genomes (Labonte and Suttle, 2013). As the ever-growing quantity of sequence data is integrated into reference databases and online tools, including Metavir (Roux et al., 2011) and VIROME (Wommack et al., 2012), the sequence space for environmental viruses will continue to be filled.

The current underrepresentation of virus genomes in public databases can result in sequence hits to genes of certain viruses that are not closely related to the virus from which the sequences originated. This was well-illustrated by the OtV-1 and OtV-2 genomes (viruses in the size range of 100-120 nm that infect the prasinophyte alga Ostreococcus tauri), which Article have at least three annotated genes with their closest identity to genes found in

(Raoult et al., 2004), as Mimivirus was the only sequenced viral genome reported to encode these genes at the time (Weynberg et al., 2009; Weynberg et al., 2011). The M. cavernosa viral metagenomes were also reported to contained significant hits to genes in giant viruses of the family (n = 287; Correa et al., 2013). However, this study included a 0.22

µm filtration step that should remove most giant viruses (diameter ≥750 nm) prior to sequencing (Xiao et al., 2005). Hence, such observations should be verified by complementing results from coral virus metagenomes with those obtained through other methods, such as viral whole genome sequencing, flow cytometry and TEM.

Another possible pitfall in metagenomic analyses is that BLAST-based searches using an e- value threshold become less reliable as the size of the database decreases. Therefore, BLAST comparisons to restricted databases should use thresholds that are independent of the

databaseAccepted size (i.e. BLAST bit-score), or if using database-dependent thresholds (i.e. e-

10 This article is protected by copyright. All rights reserved. values), they need to be adjusted appropriately. The importance of using database-size independent thresholds, like bit-score, can be easily visualized by mapping sequence read hits to an annotated viral genome. As an example, we used a herpesvirus genome (Suid herpesvirus 1, also used in Vega Thurber et al. 2008), as herpesvirus-like sequences have been reported in coral viral metagenomes previously (Marhaver et al., 2008; Vega Thurber et al., 2008; Correa et al., 2013), but were not commonly observed in these viral assemblages through our analyses (Table 1). A recruitment plot of the 10 coral virus metagenomes from these studies, matched to a single annotated genome (Suid herpesvirus

1) using a range of e-value thresholds (e-value > 10-3, = 10-3 to 10 -5, and < 10-5) highlights that most of the robust matches (above bit-score = 50; Roux et al. 2013a) occur in repeat Article regions of the genome (Figure 2). In fact, when comparing the same data sets to the NCBI

RefSeq Virus database (January 2014) and taking only the top BLAST hit to the Suid herpesvirus 1 genome, this robust coverage decreases to just a few sequences (Figure 2, insert highlighted by arrows). However, several coding genes (annotated on Figure 2) had sequences with significant matches (bit-scores > 50), so there is supporting evidence that herpesvirus-like genes are present in coral-associated viral metagenomes. The presence of herpes-like viruses in coral metagenomes is not unexpected as the Herpesvirales contain several marine representatives that cause disease in a variety of marine organisms, mostly vertebrates (Hanson et al., 2011; Maness et al., 2011). For corals, the presence of herpesviruses has not been linked to disease and their impact on the coral holobiont remains unknown.

Main patterns in coral-associated viruses from next generation sequencing studies to-date Accepted

11 This article is protected by copyright. All rights reserved. Our analysis demonstrates that the only group of annotated viruses that occurred in all data sets were dsDNA viruses, mostly from the order Caudovirales, which are the tailed bacteriophages (ICTV, 2012). Further, the majority of sequences from the coral-associated viral metagenomes that had a significant match (e-value ≤ 10-5) to NCBI’s RefSeq Virus database (ranging from 1-45% of the data set) were dsDNA and ssDNA bacteriophages

(Dinsdale et al., 2008; Marhaver et al., 2008; Vega Thurber et al., 2008; Correa et al., 2013).

The dominance of bacteriophages in these data sets is not surprising as these likely infect members of the rich bacterial communities associated with the coral mucus layer, gastric cavity and the coral tissue (Ducklow and Mitchell, 1979; Rohwer et al., 2002; Sweet et al.,

2011; Agostini et al., 2012). Article

Several hits to eukaryotic viruses were observed in the coral data sets and likely infect either the coral itself or its eukaryotic microbial symbionts, such as the dinoflagellate endosymbionts, Symbiodinium, apicomplexans, protozoans and fungi. Matches with members in the were observed in all of the data sets (highlighted in Table

1), possibly reflecting viruses associated with Symbiodinium spp., as both large dsDNA and small ssRNA viruses are known to infect dinoflagellates (Nagasaki, 2008; Tomaru et al.,

2008).

Coral bleaching, or the loss of the algal and/or their photosynthetic pigments from the coral tissues, can lead to considerable coral mortality (Bruno and Selig, 2007), as the host coral relies on their Symbiodinium as a primary source of fixed carbon (Falkowski et al., 1993).

This obligate symbiosis breaks down during unusually high summer temperatures due to

heat- andAccepted high light-induced damaged to holobiont (Weis, 2008). A possible, but poorly

12 This article is protected by copyright. All rights reserved. tested hypothesis for the cause of coral bleaching is that a latent virus in the Symbiodinium genome enters the lytic cycle in response to a stress event, causing lysis of Symbiodinium cells (Wilson et al., 2001; Wilson et al., 2005; Lohr et al., 2007). The generation and analysis of viral metagenomes from corals and freshly isolated or cultured Symbiodinium will assist with the design of experiments for testing this hypothesis.

Concluding remarks

Viruses play important roles as pathogens and mutualists in many, if not all, forms of cellular life (Munn, 2006; Suttle, 2007; Roossinck, 2011) and undoubtedly have important but currently unknown functions in the coral stress response, coral disease, and the adaptive Article potential of the coral holobiont with respect to climate change (van Oppen et al., 2009;

Thurber and Correa, 2011). Metagenomic characterisation of coral-associated viruses is a first step towards elucidating these roles. Particularly relevant in the context of current climate change is the observation of algal viruses, most likely associated with the dinoflagellate endosymbionts of corals, in all of the data sets analysed here, as these viruses may play a role in coral bleaching. Further, understanding the potential role of the diverse community in maintaining a healthy bacterial community or as agents of disease will be facilitated by robust coral viral metagenomes from healthy and diseased individuals.

Our analysis shows that marine viral assemblages within the coral holobiont can differ based on laboratory and bioinformatic methodologies. A recently developed method for isolating nucleic acids from the coral holobiont minimises the biases in the resulting metagenomes

and capturesAccepted a large portion of the viral diversity (Weynberg et al., 2014). Improvements in

13 This article is protected by copyright. All rights reserved. sequencing and bioinformatic tools for virus-related research in corals and other areas of aquatic virus research (Culley and Suttle, 2010; Schoenfeld et al., 2010; Duhaime and

Sullivan, 2012; Hurwitz et al., 2013; Solonenko et al., 2013; Thurber and Correa, 2011), suggest that coral virology is poised to become more accessible, thereby providing critical data that can be applied to understanding the effect of viruses on the biology and ecology of coral reefs.

Acknowledgements

We thank Murray Logan for statistical support and Alexander Culley for insightful comments on this manuscript. We acknowledge funding from the Australian Research Council (Future Article Fellowship #FT100100088 to MvO, Super Science Fellowship #FS110200034 to KW) and from the Australian Institute of Marine Science.

Supplementary information is available from Environmental Microbiology at:

References

Ackermann, H.W. (2006). “Classification of bacteriophages.," in The Bacteriophages, ed. R. Calendar. 2nd ed (New York, NY: Oxford University Press), 8-16. Agostini, S., Suzuki, Y., Higuchi, T., Casareto, B., Yoshinaga, K., Nakano, Y., and Fujimura, H. (2012) Biological and chemical characteristics of the coral gastric cavity. Coral Reefs 31: 147-156. Andrews-Pfannkoch, C., Fadrosh, D.W., Thorpe, J., and Williamson, S.J. (2010) Hydroxyapatite-mediated separation of double-stranded DNA, single-stranded DNA, and RNA genomes from natural viral assemblages. Appl Environ Microbiol 76: 5039-5045. Angly, F.E., Felts, B., Breitbart, M., Salamon, P., Edwards, R.A., Carlson, C. et al. (2006) The marine viromes of four oceanic regions. PLoS Biol 4: e368. Angly, F.E., Willner, D., Prieto-Davó, A., Edwards, R.A., Schmieder, R., Vega-Thurber, R. et al. (2009) The GAAS metagenomic tool and its estimations of viral and microbial average genome size in four major biomes. PLoS Comput Biol 5: e1000593. Atad, I., Zvuloni, A., Loya, Y., and Rosenberg, E. (2012) Phage therapy of the white plague- like disease of Favia favus in the Red Sea. Coral Reefs 31: 665-670. Accepted

14 This article is protected by copyright. All rights reserved. Breitbart, M., Salamon, P., Andresen, B., Mahaffy, J.M., Segall, A.M., Mead, D. et al. (2002) Genomic analysis of uncultured marine viral communities. Proc Natl Acad Sci USA 99: 14250-14255. Bruno, J.F., and Selig, E.R. (2007) Regional decline of coral cover in the Indo-Pacific: Timing, extent, and subregional comparisons. PLoS ONE 2: e711. Canchaya, C., Fournous, G., Chibani-Chennoufi, S., Dillmann, M.-L., and Brussow, H. (2003) Phage as agents of lateral gene transfer. Curr Opin Microbiol 6: 417-424. Cohen, Y., Pollock, F.J., Rosenberg, E., and Bourne, D.G. (2013) Phage therapy treatment of the coral pathogen Vibrio coralliilyticus. Microbiol Open 2: 64-74. Correa, A.M.S., Welsh, R.M., and Vega Thurber, R.L. (2013) Unique nucleocytoplasmic dsDNA and +ssRNA viruses are associated with the dinoflagellate endosymbionts of corals. ISME J 7: 13-27. Culley, A.I., and Suttle, C.A. (2010) Charcterization of the diversity of marine RNA viruses. In Manual of Aquatic Viral Ecology. Wilhelm, S.W., Weinbauer, M.G., and Suttle, C.A. (eds): ASLO, pp. 193-201. Davy, J.E., and Patten, N.L. (2007) Morphological diversity of virus-like particles within the surface microlayer of scleractinian corals. Aquatic Microbial Ecol 47: 37-44. Dinsdale, E.A., Pantos, O., Smriga, S., Edwards, R.A., Angly, F., Wegley, L. et al. (2008) Microbial ecology of four coral atolls in the Northern Line Islands. PLoS ONE 3: e1584.

Ducklow,Article H.W., and Mitchell, R. (1979) Bacterial populations and adaptations in the mucus layers on living corals. Limnol Oceanogr 24: 715. Duhaime, M.B., and Sullivan, M.B. (2012) Ocean viruses: Rigorously evaluating the metagenomic sample-to-sequence pipeline. Virology 434: 181-186. Duhaime, M.B., Deng, L., Poulos, B.T., and Sullivan, M.B. (2012) Towards quantitative metagenomics of wild viruses and other ultra-low concentration DNA samples: a rigorous assessment and optimization of the linker amplification method. Environ Microbiol 14: 2526-2537. Edwards, R.A., and Rohwer, F. (2005) Viral metagenomics. Nat Rev Micro 3: 504-510. Falkowski, P.G., Dubinsky, Z., Muscatine, L., and McCloskey, L. (1993) Population control in symbiotic corals. Bioscience 43: 606–611. Hanson, L., Dishon, A., and Kotler, M. (2011) Herpesviruses that infect fish. Viruses 3: 2160- 2191. Hurwitz, B.L., Deng, L., Poulos, B.T., and Sullivan, M.B. (2013) Evaluation of methods to concentrate and purify ocean virus communities through comparative, replicated metagenomics. Environ Microbiol 15: 1428-1440. ICTV (2012) Virus taxonomy: classification and nomenclature of viruses: Ninth Report of the International Committee on Taxonomy of Viruses. San Diego: Elsevier Academic Press. Kim, K.H., Bae, J.W. (2011) Amplification Methods Bias Metagenomic Libraries of Uncultured Single-Stranded and Double-Stranded DNA Viruses. Appl Environ Microbiol 77: 7663- 7668. Kristensen, D.M., Mushegian, A.R., Dolja, V.V., and Koonin, E.V. (2010) New dimensions of the virus world discovered through metagenomics. Trends Microbiol 18: 11-19. Labonte, J.M., and Suttle, C.A. (2013) Previously unknown and highly divergent ssDNA viruses populate the oceans. ISME J 7: 2169-2177. Lang, A.S., Zhaxybayeva, O., and Beatty, J.T. (2012) Gene transfer agents: phage-like elements of genetic exchange. Nat Rev Micro 10: 472-482. Accepted

15 This article is protected by copyright. All rights reserved. Littman, R., Willis, B.L., Bourne, D.G. (2011) Metagnomic analysis of the coral holobiont during a natural bleaching event on the Great Barrier Reef. Environ Microbiol Reports 3: 651-660. Liu, H.Q., Fu, Y.P., Li, B., Yu, X., Xie, J.T., Cheng, J.S. et al. (2011) Widespread horizontal gene transfer from circular single-stranded DNA viruses to eukaryotic genomes. BMC Evol Biol 11: 276. Lohr, J., Munn, C.B., and Wilson, W.H. (2007) Characterization of a latent virus-like infection of symbiotic zooxanthellae. Appl Environ Microbiol 73: 2976-2981. Maness, H.T.D., Nollens, H.H., Jensen, E.D., Goldstein, T., LaMere, S., Childress, A. et al. (2011) Phylogenetic analysis of marine mammal herpesviruses. Vet Microbiol 149: 23-29. Marhaver, K.L., Edwards, R.A., and Rohwer, F. (2008) Viral communities associated with healthy and bleaching corals. Environ Microbiol 10: 2277-2286. Monier, A., Pagarete, A.n., de Vargas, C., Allen, M.J., Read, B., Claverie, J.-M., and Ogata, H. (2009) Horizontal gene transfer of an entire metabolic pathway between a eukaryotic alga and its DNA virus. Genome Res 19: 1441-1449. Munn, C.B. (2006) Viruses as pathogens of marine organisms - from to whales. J Mar Biol Assoc UK 86: 453-467. Nagasaki, K. (2008) Dinoflagellates, diatoms, and their viruses. J Microbiol 46: 235-243. Patten, N., Harrison, P., and Mitchell, J. (2008) Prevalence of virus-like particles within a

staghornArticle scleractinian coral (Acropora muricata) from the Great Barrier Reef. Coral Reefs 27: 569-580. Raoult, D., Audic, S., Robert, C., Abergel, C., Renesto, P., Ogata, H. et al. (2004) The 1.2- megabase genome sequence of mimivirus. Science 306: 1344-1350. Rohwer, F., Seguritan, V., Azam, F., and Knowlton, N. (2002) Diversity and distribution of coral-associated bacteria. Mar Ecol Progr Ser 243: 10. Roossinck, M.J. (2011) The good viruses: viral mutualistic symbioses. Nat Rev Micro 9: 99- 108. Rosario, K., and Breitbart, M. (2011) Exploring the viral world through metagenomics. Curr Opin Virol 1: 1-9. Roux, S., Krupovic, M., Debroas, D., Forterre, P., and Enault, F. (2013a) Assessment of viral community functional potential from viral metagenomes may be hampered by contamination with cellular sequences. Open Biol 3: 130160. Roux, S., Enault, F., Bronner, G., Vaulot, D., Forterre, P., Krupovic, M. (2013b) Chimeric viruses blur the borders between the major groups of eukaryotic single-stranded DNA viruses. Nat Commun 4: 2700. Roux, S., Faubladier, M., Mahul, A., Paulhe, N., Bernard, A., Debroas, D., and Enault, F. (2011) Metavir: a web server dedicated to virome analysis. Bioinformatics 27: 3074-3075. Schoenfeld, T., Liles, M., Wommack, K.E., Polson, S.W., Godiska, R., and Mead, D. (2010) Functional viral metagenomics and the next generation of molecular tools. Trends Microbiol 18: 20-29. Soffer, N., Brandt, M.E., Correa, A.M.S., Smith, T.B., and Thurber, R.V. (2013) Potential role of viruses in white plague coral disease. ISME J 8: 271-283. Solonenko, S.A., Ignacio-Espinoza, J.C., Alberti, A., Cruaud, C., Hallam, S., Konstantinidis, K. et al. (2013) Sequencing platform and library preparation choices impact viral metagenomes. BMC Genomics 14: 320. Suttle, C.A. (2007) Marine viruses - major players in the global ecosystem. Nat Rev Microbiol

5: 801-Accepted 812.

16 This article is protected by copyright. All rights reserved. Sweet, M., Croquer, A., and Bythell, J. (2011) Bacterial assemblages differ between compartments within the coral holobiont. Coral Reefs 30: 39-52. Thurber, R.L.V., Correa, A.M.S. (2011) Viruses of reef-building scleractinian corals. J Exp Marine Biol Ecol 408: 102-113. Tomaru, Y., Katanozaka N., Nishida K., Shirai Y., Tarutani K., Yamaguchi M., et al. (2004) Isolation and characterization of two distinct types of HcRNAV, a single-stranded RNA virus infecting the bivalve-killing microalga Heterocapsa circularisquama. Aquat Microb Ecol 34: 207–218. Tomaru, Y., Shirai, Y., and Nagasaki, K. (2008) Ecology, physiology and genetics of a phycodnavirus infecting the noxious bloom‐forming . Fisher Sci 74: 701-711. van Oppen, M.J.H., Gates, R.D., and Leong, J.A. (2009) Coral-virus interactions: A double- edged sword? Symbiosis 47: 1-8. Vega Thurber, R.L., Barott, K.L., Hall, D., Liu, H., Rodriguez-Mueller, B., Desnues, C. et al. (2008) Metagenomic analysis indicates that stressors induce production of herpes-like viruses in the coral Porites compressa. Proc Natl Acad Sci USA 105: 18413-18418. Wegley, L., Edwards, R., Rodriguez-Brito, B., Liu, H., and Rohwer, F. (2007) Metagenomic analysis of the microbial community associated with the coral Porites astreoides. Environ Microbiol 9: 2707-2719.

Weis, V.M.Article (2008) Cellular mechanisms of Cnidarian bleaching: stress causes the collapse of symbiosis. J Exp Biol 211: 3059-3066. Weynberg, K.D., Wood-Charlson, E.M., Suttle, C.A., and van Oppen, M.J.H. (2014) Generating viral metagenomes from the coral holobiont. Front Microbiol: Virol 5: article 206. Weynberg, K.D., Allen, M.J., Ashelford, K., Scanlan, D.J., and Wilson, W.H. (2009) From small hosts come big viruses: the complete genome of a second Ostreococcus tauri virus, OtV‐1. Environ Microbiol 11: 2821-2839. Weynberg, K.D., Allen, M.J., Gilg, I.C., Scanlan, D.J., and Wilson, W.H. (2011) Genome sequence of Ostreococcus tauri virus OtV-2 throws light on the role of picoeukaryote niche separation in the ocean. J Virol 85: 4520-4529. Williamson, S.J., Rusch, D.B., Yooseph, S., Halpern, A.L., Heidelberg, K.B., Glass, J.I. et al. (2008) The Sorcerer II Global Ocean Sampling Expedition: Metagenomic characterization of viruses within aquatic microbial samples. PLoS ONE 3: e1456. Wilson, W.H., Francis, I., Ryan, K., and Davy, S.K. (2001) Temperature induction of viruses in symbiotic dinoflagellates. Aq Microb Ecol 25: 99-102. Wilson, W.H., Dale, A.L., Davy, J.E., and Davy, S.K. (2005) An enemy within? Observations of virus-like particles in reef corals. Coral Reefs 24: 145-148. Wommack, K.E., Bhavsar, J., and Ravel, J. (2008) Metagenomics: Read length matters. Applied and Environ Microbiol 74: 1453-1463. Wommack, K.E., Bhavsar, J., Polson, S.W., Chen, J., Dumas, M., Srinivasiah, S. et al. (2012) VIROME: a standard operating procedure for analysis of viral metagenome sequences. Stand Genom Sci 6: 427. Xiao, C., Chipman, P.R., Battisti, A.J., Bowman, V.D., Renesto, P., Raoult, D., and Rossmann, M.G. (2005) Cryo-electron Microscopy of the Giant Mimivirus. J Mol Biol 353: 493-496.

Accepted

17 This article is protected by copyright. All rights reserved. Figure Legends

Figure 1. Non-metric multidimensional scaling (nMDS) plot of coral-associated sequence data sets. The sequence data sets were grouped by holobiont target (coral transcriptome, n=12; prokaryote metagenome, n=3; and chloroformed (“CFM”, n=14) and non- chloroformed (“Non-CFM”, n=6) viral metagenomes. Abundance of sequences identified as ds- and ssDNA viruses (e-value ≤ 10-5) from Table 1 were normalized to the size of the data set and analysed using nMDS (stress = 0.123).

Figure 2. Recruitment plot of the viral metagenome data sets from Diploria strigosa, Porites compressa and Montastraea cavernosa against the Suid herpesvirus 1 genome (SuHV1, NCBI genome NC_006151, selected from Vega-Thurber et al. 2008). Sequences are coloured by e- Article value (red = e-value > 10-3, green = 10-3-10-5, and blue = < 10-5). Annotated regions are represented by the red blocks, and all repeat regions within non-coding regions are highlighted in white. BLAST bit-scores (size independent thresholds for database comparison) greater than 50 (area above black line in main figure and insert) are generally considered a significant match (Roux et al. 2013a), especially for small database comparisons. Stars mark annotated proteins with significant bit-scores. The insert (upper right) shows only sequences where the SuHV1 genome was the top BLAST-hit, with arrows indicating sequences with bit score > 50.

Supplementary Information

Supplementary Table 1. List of metagenomes and transcriptomes available through

Metavir.

Supplementary Table 2. Brief summary of methods used to produce the data sets analyzed

in this Accepted study.

18 This article is protected by copyright. All rights reserved. Article

EMI_12803_F1 Accepted

19 This article is protected by copyright. All rights reserved.

Article

EMI_12803_F2

Accepted

20 This article is protected by copyright. All rights reserved. Table 1. Viral families identified from coral-holobiont sequence data sets. Coloured boxes indicate that the viral family was identified by Metavir (e-value less than 10-5; NCBI Refseq 2014-09-10, GAAS normalization). Colours relate to statistical analysis in Figure 1, and information regarding the source of each data set is available in Supplementary Table 1. Hosts: Ar, ; B, bacteria; F, fungi; I, invertebrate; P, ; Pr, ; V, vertebrate. NA t0 pH DNA temp 2010 2012 DOC temp 2010 2012 2010 2012 control 2011 nutrient 2012 (database) (healthy) prokaryote (bleached) (pre-bleach) (post-bleach) RNA 2014 RNA 2014 DNA LiquidN2 SISPA LiquidN2 No LiquidN2 SISPA LiquidN2 No 2012 (contigs) 2012 M. annularis M. annularis M. faveolata A. tenuis tenuis A. P. compressa P. A. palmate palmate A. P. compressa P. A. millepora millepora A. millepora A. Palmyra seawater Palmyra Kiritimati seawater Kiritimati Kingman seawater Kingman Tubueran seawater Tubueran P. asteroides P. asteroides P. P. damicornis R damicornis P. damicornis P. M. cavernosa M. cavernosa A. hyacinthus hyacinthus A. hyacinthus A. P. compressa P. P. compressa P. A. tenuis tenuis A. tenuis A. D. strigosa D. strigosa M. cavernosa control M. cavernosa P. compressa P. P. compressa P. D. strigosa D. strigosa P. acuta acuta P. P. asteroides P. A. millepora millepora A. P. damicornis damicornis P. A. millepora millepora A.

Type Order Family Host damicornis P. P. damicornis P. ds DNA I Caudovirales B B B Herpesvirales V V I Unassigned V Ar I Asfarviridae V I Ar Corticoviridae B Ar I I, V Ar Pr Mimiviridae Pr Nimaviridae I Article I V Phycodnaviridae Pr B Polydnaviridae I V I, V Rudiviridae Ar Tectiviridae B B Unclassified ss DNA II Unassigned V Geminivirdae P Inoviridae B B P I, V Unassigned ds RNA III Unassigned F, P Hypoviridae F F, P, Pr F, I, P, V Unclassified ss RNA (+) IV V I Pr Picornaviridae V F, P Unassigned F, P Pr Astroviridae V V I, V Marnaviridae F P Togaviridae I, V P Unassigned ss RNA (-) V I, V

Unassigned Unassigned Arenaviridae V ss RNA (+) RT VI Unassigned Retroviridae V dsDNA RT P

Unclassified archaeal Ar Satellite Prokaryote Non-Chloroformed Holobiont Target Coral Transcriptome Chloroformed Virome Metagenom Virome Accepted Accepted