Investigating Host-Symbiont Crosstalk in the Coral Reef Demosponge Amphimedon Queenslandica

Investigating host-symbiont crosstalk in the coral reef demosponge Amphimedon queenslandica

Xueyan Xiang BSc (Biotechnology)

0000-0003-1843-0474

A thesis submitted for the degree of Doctor of Philosophy at The University of Queensland in 2021 School of Biological Sciences Abstract

In the last two decades, the widespread application of -omic approaches has revealed the astonishing diversity and ubiquity of microbial communities that inhabit animals. This growing understanding of host-symbiont interactions has revolutionised our view about how symbiotic microbes influence their host animal. Studies in humans and many other animals have revealed that symbiotic microbes can regulate and contribute to the metabolism of different nutrients. These metabolic activities can produce metabolites that affect the host's physiology, development, growth and immunity. In this thesis, I use genomic and transcriptomic approaches to explore host-symbiont crosstalk in the coral reef demosponge Amphimedon queenslandica. This sponge provides a tractable system because it houses a low complexity and low abundance microbiota dominated by just three proteobacterial symbiont types.

Complete and reliable holobiont genomic and transcriptomic data can help to investigate interactions between the sponge and its symbionts. To better understand interactions between A. queenslandica and its symbionts, I first resequenced, reassembled and reannotated the existing draft genomes of A. queenslandica and its three primary proteobacterial symbionts – AqS1, AqS2, and AqS3 – using additional in vitro reconstituted chromatin (Chicago) and short paired-end Illumina data. This markedly improved the genome assemblies and gene functional annotations of all four of these symbiotic partners (Chapter 2). The N50 length of A. queenslandica genomic scaffolds increased almost eight-fold, from 120 to 950 kb. The scaffolds N50 of AqS1, AqS2 and AqS3 also substantially increased to 103, 148 and 90 kb, respectively. This reanalysis revealed the relative contribution of these genomes to holobiont functioning, with the three symbionts disproportionately contributing to a majority of the holobiont's metabolic pathways.

I then turned to assess the gene activity in this hologenome by developing an approach to accurately analyse holobiont transcriptomes. The efficacy of the various approaches used previously to capture host (eukaryote) and bacterial symbiont transcriptomes indicated biases in capturing both pools of mRNA. In this context, I sought to identify a dual RNA-Seq approach that sufficiently captures both host and symbiont transcriptomes. I compared the transcriptomic data generated from poly(A) captured mRNA-Seq (Poly(A)-mRNA-Seq) and ribosomal RNA depleted RNA-Seq (rRNA- depleted-RNA-Seq) using bacterial-enriched adult tissues of A. queenslandica holobiont, and from previously generated bacterial-unenriched A. queenslandica PolyA-mRNA-Seq data, focusing on the transcriptome depth and coverage (Chapter 3). For the host sponge, no significant difference was found in transcriptomes generated from Poly(A)-mRNA-Seq and rRNA-depleted-RNA-Seq. However, the rRNA-depleted-RNA-Seq performed better than the Poly(A)-mRNA-Seq in capturing

i representative symbiont transcriptomes. I also found that bacterial cell enrichment enabled adequate capture of the symbiont transcriptomes, although it reduced the A. queenslandica host transcriptome. This comparison demonstrated that RNA-Seq by ribosomal RNA depletion is an effective and reliable method to obtain an accurate and representative holobiont transcriptome. This dual RNA-Seq approach – rRNA-depleted-RNA-Seq – allows for the accurate generation of A.queenslandica holobiont transcriptome data.

I used these new A. queenslandica holobiont genomes and transcriptomes to identify potential host- symbiont metabolic interactions (Chapter 4). For example, genomic and transcriptomic analyses of the sponge A. queenslandica and its three primary symbiotic bacteria, AqS1, AqS2, and AqS3, confirmed host and symbionts are all involved in active carbohydrate metabolism. The host and symbionts also collaborate in assimilating dissolved inorganic nitrogen, sulfur, and phosphate. The symbiotic bacteria can uniquely de novo synthesise – and thus potentially provide the host sponge with – essential amino acids, some vitamins (B1, B2, and B5) and cofactors (folate and siroheme). This analysis of metabolic pathway enzymes suggests the sponge holobiont works cooperatively to assimilate the major marine nutrients, and cycle nutrients in the aquatic ecosystems.

A. queenslandica hologenomes and transcriptomes also revealed potential animal-bacterial interkingdom signalling (Chapter 5). For instance, the capacity of the symbiont AqS1 to uptake and degrade sponge-derived γ-aminobutanoate (GABA) suggests that GABA may act as a quorum- sensing (QS) signal between the host and symbionts, which has the potential to influence the holobiont behaviour. The sponge symbionts appear to generate some metabolites that are not produced by the sponge, yet can act as ligands for G protein-coupled receptors (GPCRs) of the sponge. Notably, some of these are neurotransmitters in neural animals, including dopamine, tyramine, tryptamine, acetate, and propionate. Given this may provide a mechanism for the symbionts to regulate sponge signalling pathways and influence sponge physiology, I tested one of these – dopamine – experimentally. I show that the experimental introduction of dopamine can influence larval phototactic swimming behaviour, potentially via a dopamine-related GPCR. Together these results reveal the potential for sponge-bacterial interkingdom signalling communication.

In summary, the A. queenslandica holobiont -omics analyses have revealed the molecular-level mechanisms of sponge-microbe cooperation in nutrient assimilation, potential metabolic interdependence amongst the partners of the holobiont, and signalling communication between the host and symbionts. These results reveal potential crosstalk within the A. queenslandica holobiont that establishes and maintains the sponge-bacteria symbiosis.

Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co-authors for any jointly authored works included in the thesis.

iii

Publications included in this thesis

No publications included

Submitted manuscripts included in this thesis

No manuscripts submitted for publication

Other publications during candidature

Peer-reviewed papers 1. Hall, M. R., K. M. Kocot, K. W. Baughman, S. L. Fernandez-Valverde, M. E. A. Gauthier, W. L. Hatleberg, A. Krishnan, C. McDougall, C. A. Motti, E. Shoguchi, T. Wang, X. Y. Xiang, M. Zhao, U. Bose, C. Shinzato, K. Hisata, M. Fujie, M. Kanda, S. F. Cummins, N. Satoh, S. M. Degnan and B. M. Degnan (2017). The crown-of-thorns starfish genome as a guide for biocontrol of this coral reef pest. Nature 544(7649): 231-234. 2. Kukekova, A. V., J. L. Johnson, X. Y. Xiang, S. H. Feng, S. P. Liu, H. M. Rando, A. V. Kharlamova, Y. Herbeck, N. A. Serdyukova, Z. J. Xiong, V. Beklemischeva, K. P. Koepfli, R. G. Gulevich, A. V. Vladimirova, J. P. Hekman, P. L. Perelman, A. S. Graphodatsky, S. J. O'Brien, X. Wang, A. G. Clark, G. M. Acland, L. N. Trut and G. J. Zhang (2018). Red fox genome assembly identifies genomic regions associated with tame and aggressive behaviours. Nature Ecology & Evolution 2(9): 1514-1514. 3. Rando, H. M., M. Farre, M. P. Robson, N. B. Won, J. L. Johnson, R. Buch, E. R. Bastounes, X. Y. Xiang, S. H. Feng, S. P. Liu, Z. J. Xiong, J. Kim, G. J. Zhang, L. N. Trut, D. M. Larkin and A. V. Kukekova (2018). Construction of Red Fox Chromosomal Fragments from the Short-Read Genome Assembly. Genes 9(6). 4. Gao, W., Y. B. Sun, W. W. Zhou, Z. J. Xiong, L. N. Chen, H. Li, T. T. Fu, K. Xu, W. Xu, L. Ma, Y. J. Chen, X. Y. Xiang, L. Zhou, T. Zeng, S. Zhang, J. Q. Jin, H. M. Chen, G. J. Zhang, D. M. Hillis, X. Ji, Y. P. Zhang and J. Che (2019). Genomic and transcriptomic investigations of the evolutionary transition from oviparity to viviparity. Proceedings of the National Academy of Sciences of the United States of America 116(9): 3646-3655.

Conference abstracts

1. X. Y. Xiang, D. Poli, S. M. Degnan and B. M. Degnan, Comparison of RNA-Seq by Poly (A) capture and ribosomal RNA depletion for sponge holobiont transcriptome, 3rd International Symposium on Sponge Microbiology, Shanghai, China, 25-29 Oct 2018.

Contributions by others to the thesis

Sandie M. Degnan contributed to the conception and design of this research, advised on methods and results interpretation, and critically revised the entire thesis.

Bernard M. Degnan advised on design of this research, methods and results interpretation, and critically revised the entire thesis. He also collected the sponge samples from the field and prepared the DNA material for Chicago and Illumina sequencing.

Dovetail Genomics (Santa Cruz, USA) prepared and sequenced the Chicago libraries.

Institute for Molecular Bioscience Sequencing Facility (IMB, Brisbane, Australia) prepared and sequenced the Illumina DNA libraries.

Davide Poli collected the sponge samples and prepared RNA material for dual RNA-Seq.

Ramaciotti Centre for Genomics (Sydney, Australia) prepared the dual RNA-Seq libraries and performed transcriptome sequencing, which dataset was used in chapter 3.

IMB prepared the dual RNA-Seq libraries and Australian Genome Research Facility (AGRF, Brisbane, Australia) performed transcriptome sequencing, which dataset was used in chapter 4 and 5.

Nick Rhodes provided high-performance computing support constantly.

Eunice Wong advised on sponge larval phototactic swimming experiment setting.

Elizabeth Ryan advised on statistical analyses for the larval phototactic data.

Statement of parts of the thesis submitted to qualify for the award of another degree

No works submitted towards another degree have been included in this thesis

Research involving human or animal subjects

No animal or human subjects were involved in this research

vii

Acknowledgments

First and foremost, I would like to express my deepest gratitude to my remarkable supervisor Sandie Degnan, for all the guidance and inspiration you have given me along the long way. This thesis would not have been possible without you. Your insight and wisdom can always lead the way when I get lost in the endless analyses. Your constant source of patience, support and encouragements are always there when I am down. I have learnt so much from you in aspects of both work and life, and I can’t thank you enough.

I would like to express my heartfelt gratitude to my co-supervisor and joint lab Head, Bernie Degnan. My PhD journey began with your guidance. Thank you for giving me the freedom to trying new things, forgiving my mistakes and encouraging me to interpret results from different perspectives. I greatly appreciated your tremendous support since I first joined the lab.

I also wish to thank my other co-supervisor, Philip Bond, for welcoming me into your lab and guiding me the proteomic experiments. I would also like to extend my thanks to my panel of readers, Paul Ebert and Jan Engelstaedter, for your frank and honest advice regarding my project at each milestone.

To all the members of the Degnan Lab, past and present, thank you for your valuable knowledge, advice and help over the years. Special thanks to Kerry Roper and Gemma Richards for making my first steps into the sponge lab, Davide Poli, Laura Rix, Eunice Wong, and Huifang Yuan for your help and advice on my sponge experiments, Haojing Shao for assisting with bioinformatics, Joanne Leerberg and Chris Challen for unwavering support as Lab Manager.

I am indebted to Sandie Degnan, Bernie Degnan, Laura Rix and Sheen Wong, for taking the time to read and critique the various chapters of this thesis. Especially, Sandie and Bernie have patiently revised all my chapters several times and gave me vast suggestions and comments to improve my writing. Thank you both for being so generous with the time you have spent training my writing skills and providing helpful feedback. I greatly appreciate your help in my chapters.

Numerous UQ colleagues have generously lent their time and expertise in various fields. In particular, Nick Rhodes from Queensland Facility for Advanced Bioinformatics is always there with constant support for computational and bioinformatics resources. I am sincerely grateful for your invaluable help over the years. Andrew Laloo and Amanda Nouwens assisted with proteomics experiment. I appreciated your help on this even the result was not positive at last. Steven Robbins gave me the advice to interpret the sponge-bacterial metabolic pathways, and Elizabeth Ryan

viii advised on statistical analyses for the larval phototactic data. And thanks to the staff of Heron Island Research Station for their assistance on my field trip and the members of UQ Hacky Hour for help on my bioinformatics problems.

This research was funded by Australian Research Council grants awarded to Sandie M. Degnan. I acknowledge the University of Queensland for awarding me the Research Training Program Scholarship that covered my tuition fees and provided me living allowance stipends. I am grateful to my supervisors Sandie and Bernie Degnan for financially supporting my field trip on Herson Island, EMBL Australia PhD Course, and participation in various educational programs. Thank the School of Biological Sciences for supporting my conference trip to ISSM conference and the UQ Graduate School for supporting my placement at CSIRO. These opportunities were invaluable for my personal and professional development.

I would like to express my deep gratitude to Denis Bauer for giving me the opportunity to do the placement in your Transformational Bioinformatics group at CSIRO. Special thanks to Natalie Twine and Arash Bayat for your advice and help on my placement project.

Special thanks to Tahsha, Bec, Will, Laura, Ben, Shun, Carmel, Markus, Davide, Marie, Mathias, Olivia, Sheen, Eunice, Enya, Haojing, Huifang, and Bin. Thank you for creating such a friendly and enjoyable lab environment. We have shared so much great time in the Degnan lab.

Finally, I am eternally grateful to my family, Mum, Dad, and Yinhua, for always being there with endless support and unconditional love. Words cannot express my gratitude for everything you have done for me.

Financial support

This research was funded by the Australian Research Council (ARC) grants awarded to Sandie M. Degnan.

This research was supported by an Australian Government Research Training Program Scholarship.

Conference travel and career development opportunity were supported by the University of Queensland (UQ) School of Biological Sciences, UQ Research Training Career Development Scholarship.

Keywords

Marine invertebrate, symbiosis, hologenomics, dual RNA-Seq, inter-kingdom, metabolism, signalling, phototransduction

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 060205, Marine and Estuarine Ecology, 40%

ANZSRC code: 060808, Invertebrate Biology, 35%

ANZSRC code: 060203, Ecological Physiology, 25%

Fields of Research (FoR) Classification

FoR code: 0608, Zoology, 60%

FoR code: 0601, Biochemistry and Cell Biology, 25%

FoR code: 0605, Microbiology, 15%

Table of Contents

Chapter 1 General Introduction ...... 1

1.1 Host-symbiont interactions are ubiquitous and vital in the animal kingdom ...... 1 1.1.1 Host-symbiont interactions in humans ...... 1 1.1.2 Host-symbiont interactions in other vertebrates ...... 2 1.1.3 Host-symbiont interactions in invertebrates ...... 3 1.2 Interconnecting the host and symbionts ...... 4 1.2.1 Research methods to interconnect the host and symbionts ...... 4 1.2.2 Hosts and symbionts need to communicate with each other ...... 5 1.3 Marine sponge holobionts ...... 7 1.3.1 Sponges host diverse and often abundant microbial symbiont communities ...... 7 1.3.2 The role of sponges in marine ecosystems ...... 7 1.3.3 Symbiont bacterial assisted sponge nutrient assimilation ...... 8 1.4 The Amphimedon queenslandica holobiont to study host-symbiont crosstalk ...... 10 1.4.1 The existence of draft genomes for both the host and the primary symbionts ...... 10 1.4.2 The nutrient status on Heron Island Reef ...... 12

Chapter 2 An improved Amphimedon queenslandica hologenome assembly reveals how three proteobacterial symbionts can extend the metabolic phenotypic of their marine sponge host ...... 113

2.1 Abstract ...... 13 2.2 Introduction ...... 14 2.3 Materials and Methods ...... 15 2.3.1 Holobiont sampling and sequencing ...... 15 2.3.2 Hologenome assemblies ...... 16 2.3.3 Hologenome annotation ...... 16 2.3.4 Functional annotation of holobiont gene models ...... 16 2.4 Results ...... 16 2.4.1 Hologenome sequencing and assembly ...... 16 2.4.2 Hologenome annotation ...... 19 2.4.3 Functional annotation of the A. queenslandica hologenome...... 20 2.5 Discussion ...... 24 2.5.1 Improvement of the A. queenslandica hologenome ...... 24

xii

2.5.2 Functional partitioning of the A. queenslandica hologenome ...... 26 2.6 Conclusion ...... 28

Chapter 3 Ribosomal RNA-depletion provides an efficient method for successful dual RNA-Seq expression profiling of a marine sponge holobiont ...... 29

3.1 Abstract ...... 29 3.2 Introduction ...... 29 3.3 Materials and Methods ...... 32 3.3.1 Sample collection and sequencing ...... 32 3.3.2 Read processing and alignment ...... 33 3.3.3 Comparative assessment of gene depth and gene coverage ...... 33 3.3.4 Comparison with an RNA-Seq dataset generated previously from non-bacterial- enriched sponge sample ...... 33 3.3.5 Comparison of gene function among different data sets ...... 34 3.4 Results ...... 34 3.4.1 Sequencing and alignment profile ...... 34 3.4.2 Correlation between samples ...... 35 3.4.3 Gene coverage ...... 37 3.4.4 Distribution of read depth in gene coding regions ...... 41 3.4.5 Comparison of biological functions of expressed genes in the different RNA-Seq data sets ...... 43 3.5 Discussion ...... 48 3.5.1 Transient polyadenylation in bacteria reduces mRNA capture in Poly(A)-RNA-Seq ...... 48 3.5.2 Advantages and challenges of rRNA-depleted-RNA-Seq for dual RNA-Seq experiments ...... 49 3.6 Conclusion ...... 51

Chapter 4 Metabolic crosstalk between Amphimedon queenslandica and its primary bacterial symbionts ...... 52

4.1 Abstract ...... 52 4.2 Introduction ...... 53 4.3 Materials and Methods ...... 54 4.3.1 Functional annotation of holobiont gene models ...... 54 4.3.2 Assigning KEGG modules to host and symbionts ...... 55

xiii

4.3.3 Biological material collection and symbiotic bacteria enrichment ...... 55 4.3.4 RNA-Seq library preparation and sequencing ...... 56 4.3.5 Transcriptome data from previous experiments ...... 56 4.3.6 Read processing and alignment ...... 56 4.3.7 Classification of holobiont gene expression levels into quartiles ...... 58 4.3.8 Metabolic reconstruction of KEGG pathways ...... 58 4.4 Results ...... 59 4.4.1 Holobiont transcriptome data ...... 59 4.4.2 Gene expression quartile analyses ...... 60 4.4.3 Symbiotic bacterial contribution to the holobiont metabolism ...... 60 4.4.4 Carbohydrate metabolism in the A. queenslandica holobiont ...... 64 4.4.5 Amino acid biosynthesis in the A. queenslandica holobiont ...... 72 4.4.6 Fatty acid biosynthesis in the A. queenslandica holobiont ...... 78 4.4.7 Metabolism of cofactors and vitamins in the A. queenslandica holobiont ...... 79 4.4.8 Dissolved inorganic nitrogen assimilation capacity of the A. queenslandica holobiont ...... 83 4.4.9 Sulfur assimilation capacity of A. queenslandica holobiont ...... 84 4.4.10 Phosphorus assimilation capacity of A. queenslandica holobiont ...... 86 4.5 Discussion ...... 87 4.5.1 Nutrient assimilated by the sponge holobiont ...... 87 4.5.2 Metabolic complementation between the host sponge and symbiotic microorganisms ...... 91 4.6 Conclusion ...... 92

Chapter 5 Interkingdom signalling between a sponge and its primary symbiotic bacteria ...... 93

5.1 Abstract ...... 93 5.2 Introduction ...... 94 5.3 Materials and Methods ...... 96 5.3.1 Reconstruction of KEGG pathways ...... 96 5.3.2 The capability of the sponge A. queenslandica to respond to dopamine ...... 96 5.4 Results ...... 98 5.4.1 A. queenslandica has the potential to signal its symbionts ...... 98 5.4.2 Signalling molecules produced by symbionts ...... 103 5.4.3 A dopamine receptor assay in A. queenslandica ...... 109

xiv

5.5 Discussion ...... 113 5.5.1 No evidence for an AHL-QS system in the A. queenslandica holobiont ...... 113 5.5.2 Small signalling molecules may mediate interactions between partners in the A. queenslandica holobiont ...... 114 5.6 Conclusion ...... 117

Chapter 6 General Discussion ...... 119

6.1 Overview of project objectives ...... 119 6.1.1 Genetic resources of A. queenslandica holobiont ...... 119 6.1.2 Collaboration between host and symbionts in nutrient assimilation ...... 120 6.1.3 Host-symbiont metabolic complementation ...... 122 6.1.4 Host-symbiont interkingdom signalling communication ...... 122 6.2 Concluding remarks and recommendations for future study ...... 124

Reference ...... 127

Appendices ...... 161

Supplementary files (Chapter2) ...... 161 Supplementary files (Chapter3) ...... 166 Supplementary files (Chapter4) ...... 167

List of Figures

Figure 2-2. The percentages of per biological categories for A. queenslandica (Aq), AqS1, AqS2 and AqS3 ...... 23

Figure 3-1. Taxonomic distribution of reads, which is the percent of reads aligned to each species in the sponge holobiont for rRNA-depleted- and Poly(A)-RNA-Seq libraries...... 35

Figure 3-2. Correlation of expressed genes between biological replicates estimated for the A. queenslandica holobiont...... 36

Figure 3-3. Percentage of genes to which reads aligned in A. queenslandica (Aq) and the 3 primary symbionts (AqS1, AqS2 and AqS3) genomes for rRNA-depleted-RNA-Seq, and Poly(A)-RNA-Seq, and bacterial unenriched-Poly(A)-RNA-Seq data...... 39

Figure 3-4. Boxplot of the number of reads mapped to each expressed gene in rRNA-depleted- RNA-Seq, Poly(A)-RNA-Seq, and Poly(A)-RNA-Seq data without bacterial enrichment data (Unenriched-Poly(A)-RNA) for A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3)...... 42

Figure 3-5. Expressed GO percentage in each sample for A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3)...... 45

Figure 3-6. Distribution of the percentage of expressed genes in each GO term...... 46

Figure 3-7. The average expressed gene numbers of some of the most differentially represented GO terms in the rRNA-depleted-, Poly(A)- and bacterial unenriched-Poly(A)-RNA data sets...... 47

Figure 3-8. Distribution of the expressed gene percentage for each GO of each sample...... 49

Figure 4-1. Overview of the glycolysis pathway in the A. queenslandica holobiont...... 66

Figure 4-2. Expression levels of A. queenslandica genes involved in glycolysis and pentose phosphate pathways throughout development and in cell types...... 67

xvi

Figure 4-3. The expression levels of A. queenslandica holobiont genes involved in pyruvate oxidation and Krebs cycle in the adult holobiont...... 68

Figure 4-4. Pentose phosphate pathway (PPP) in the A. queenslandica holobiont...... 70

Figure 4-5. Expression levels of A. queenslandica genes involved in some potential metabolic complementation between the host and symbionts throughout development and in cell types...... 72

Figure 4-6. The completeness of amino acid biosynthesis in the A. queenslandica holobiont...... 73

Figure 4-7. The threonine, lysine, and methionine biosynthesis pathway of the A. queenslandica holobiont...... 76

Figure 4-8. The expression levels of A. queenslandica holobiont genes involved in the riboflavin metabolism pathway in the adult holobiont ...... 80

Figure 4-9. Pantothenate biosynthesis in the A. queenslandica holobiont...... 82

Figure 4-10. Expression levels of A. queenslandica genes involved in sulfur metabolism throughout development and in cell types...... 86

Figure 4-11. Schematic representations of a working model for polyphosphate accumulation in the A. queenslandica holobiont...... 87

Figure 4-12. Schematic overview of metabolic interactions between all of the members of the A. queenslandica holobiont...... 88

Figure 5-1. The GABA shunt pathway of the A. queenslandica holobiont ...... 100

Figure 5-2. Expression levels of A. queenslandica genes involved in GABA shunt pathway and GABA transporter throughout the life cycle and in adult cell types...... 101

Figure 5-4. Tyramine, dopamine, tryptamine and histamine synthesis in the A. queenslandica holobiont...... 104

Figure 5-5. Partial alignment of aromatic amino acid decarboxylase and histidine decarboxylase ...... 105 xvii

Figure 5-6. Alignment of putative polyphenol oxidase sequences of the three primary symbiotic bacteria with the Escherichia coli polyphenol oxidase...... 106

Figure 5-7. Alignment of partial A. queenslandica putative hemocyanins with that of two crustaceans...... 107

Figure 5-8. Effect of dopamine receptor agonist (DRA) and antagonist (DRAA) on larval phototaxis...... 111

Figure 5-9. P-values of the effects of DRA and DRAA on the sponge larval negatively phototactic swimming behaviour...... 113

xviii

List of Tables

Table 2-1. Genome assembly statistics of A. queenslandica (Aq), and the three primary symbionts, AqS1, AqS2 and AqS3...... 18

Table 2-2. Comparison of BUSCO analyses of old and new A. queenslandica (Aq), AqS1, AqS2 and AqS3 assemblies...... 18

Table 2-3. Comparison of coding and non-coding gene number and length between old and new assemblies for A. queenslandica (Aq), and the three primary symbionts, AqS1, AqS2 and AqS3. .... 19

Table 2-4. The number of A. queenslandica (Aq), AqS1, AqS2 and AqS3 genes classified into six broad biological categories by KEGG Mapper...... 20

Table 3-2. The average median number of reads aligned per expressed gene ...... 42

Table 3-3. Number of annotated GO terms for genes of A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3)...... 43

Table 3-5. Average percentage of GO terms represented in the expressed genes of A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3) in the three RNA-Seq data sets...... 46

Table 4-1. TPM normalised read count quartile values of transcriptome data for A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3) ...... 59

Table 4-2. The number of genes expressed in each quartile of the holobiont transcriptome for A. queenslandica (Aq) and the three dominant symbionts (AqS1, AqS2 and AqS3)...... 60

Table 4-3. Complete KEGG modules of A. queenslandica holobiont contributed by the symbiotic bacteria ...... 62

xix

Table 5-1. The average percentage of larvae in quartile 1 (Q1) varying since introduced into the chamber...... 112

Table 5-2. The average percentage of larvae in quartile 4 (Q4) varying since introduced into the chamber...... 112

List of Abbreviations used in the thesis

Abbreviation Definition AA amino acid ABAT GABA aminotransferase ACP acyl carrier protein ADPGK ADP-dependent glucokinase AGRF Australian Genome Research Facility AHL N-acyl homoserine lactone AMP antimicrobial peptide CDS protein-coding sequences CMF-ARS calcium- and magnesium-free artificial saltwater

CO2 carbon dioxide DAPI 4’,6-Diamidino-2-Phenylindole, Dihydrochloride DDC aromatic amino acid decarboxylase DE differential expression DIN dissolved inorganic nitrogen DOC dissolved organic carbon DOM dissolved organic matter DON dissolved organic nitrogen DRA dopamine receptor agonist DRAA dopamine receptor agonist antagonist DRD dopamine-like receptor EV extracellular vesicle FFAR free fatty acid receptor FSW filtered seawater GABA γ-aminobutyric acid GABA-AT GABA aminotransferase GabD succinate semialdehyde dehydrogenase GAD glutamate decarboxylase GAPDH glyceraldehyde-3-phosphate dehydrogenase gDNA genomic DNA GHD glutamate dehydrogenase GI gastrointestinal glmer generalised linear mixed model Glu glutamate GO Gene Ontology GPCR G protein-coupled receptor GS glutamine synthetase GTP guanosine 5'-triphosphate HDC histidine decarboxylase HGT horizontal gene transfer

xxi

Abbreviation Definition HMA high microbial abundance HPLC liquid chromatography IBD inflammatory bowel diseases KO KEGG Orthology LC-MS liquid-chromatography-mass spectrometry LMA low microbial abundance LPS lipopolysaccharide MAMP microbe-associated molecular pattern NA noradrenaline NAD nicotinamide adenine dinucleotide nanoSIMS nanoscale secondary ion mass spectrometry NO nitric oxide NOS nitric oxide synthase OTU operational taxonomic unit PGRP peptidoglycan recognition protein PLFA phospholipid fatty acid PLP pyridoxal phosphate Poly(A)-RNA-Seq poly(A) captured mRNA-Seq polyP polyphosphate POM particulate organic matter PON particulate organic nitrogen PPi-PFK pyrophosphate-dependent phosphofructokinase PPO polyphenol oxidase PPO polyphenol oxidase PPP pentose phosphate pathway PRPP phosphoribosyl diphosphate PRR pattern-recognition receptor PTS phosphotransferase proteins PYG glycogen phosphorylase QS quorum-sensing rRNA ribosomal RNA rRNA-depleted-RNA-Seq ribosomal RNA depleted RNA-Seq rTCA reductive citric acid cycle SBP sedoheptulose 1,7-bisphosphate SCFA short-chain fatty acid SGS salivary gland surface SIP stable isotope probing SOB sulfur-oxidizing bacteria SQOR sulfide:quinone oxidoreductase

xxii

Abbreviation Definition sRNA small noncoding RNAs TAAR trace amine-associated receptor TCA tricarboxylic acid cycle TLR Toll-like receptors tmRNA transfer-messenger RNA TOC total organic carbon TPM transcripts per million tRNA transfer RNA vitamin B1 thiamine-phosphate vitamin B2 riboflavin vitamin B5 pantothenate

xxiii

Chapter 1 General Introduction

1.1 Host-symbiont interactions are ubiquitous and vital in the animal kingdom

The advent and widespread application of genomic approaches have revealed the surprising diversity and ubiquity of the microbial world. Thus, animal multicellularity is now viewed as an entirety of the animal itself plus its diverse inhabited microorganisms (Gilbert et al. 2012). Together these organisms comprise an entity known as the "holobiont" (Theis et al. 2016). A growing understanding of host-symbiont interactions has revolutionised our view about how symbionts influence and interact with host animals (McFall-Ngai et al. 2013, Simon et al. 2019). The symbiotic microbes have critical impacts in host biology, such as reproduction, development, immunity, growth and physiology (Yen and Barr 1971, Engelstadter and Hurst 2009, Pradeu 2011, McFall-Ngai et al. 2013, Leulier et al. 2017, van de Guchte et al. 2018). In the last two decades, the composition and diversity of the symbiotic microbial communities and their critical roles on the host have been widely explored across the animal kingdom (Reviewed in McFall-Ngai et al. 2013, Gilbert et al. 2018, O’Brien et al. 2019, Simon et al. 2019), from early invertebrates (van de Water et al. 2018, O’Brien et al. 2020) to human (Lloyd-Price et al. 2017, Almeida et al. 2019, Nayfach et al. 2019, Pasolli et al. 2019).

1.1.1 Host-symbiont interactions in humans

The human microbiome is the best-studied animal holobiont system so far. The Human Microbiome Project (Group et al. 2009, Proctor et al. 2019) has revealed that different microbial communities inhabit different surfaces of the human body and that the number of bacteria is in the same order as the number of human cells in an average human (Sender et al. 2016). The human gut microbiota alone contains about 3.9 x 10 13 bacterial cells (estimate number for an adult man) (Sender et al. 2016). The human microbiome is highly personalised and body site-specific (Huttenhower et al. 2012, Gilbert et al. 2018). Recent studies have made significant progress in characterising the microbial communities from numerous body sites (oral, nasal, vaginal, gut, and skin) at species level across different populations (Lloyd-Price et al. 2017, Almeida et al. 2019, Nayfach et al. 2019, Pasolli et al. 2019). Tens of thousands of human microbial genomes are assembled, and millions of genes are annotated (Lloyd-Price et al. 2017, Almeida et al. 2019, Nayfach et al. 2019, Pasolli et al. 2019); 46 million non-redundant genes are identified for oral and gut microbes (Tierney et al. 2019). These symbiotic microbes play essential roles in human health (Gilbert et al. 2018, Proctor et

1 al. 2019), and even gene-level variability in the microbiome is associated with human health and disease (Zeevi et al. 2019).

The microbiome plays critical roles in the human immune system through complex interactions (Belkaid and Hand 2014, Zheng et al. 2020). Microbial colonisation of mucosal tissues during early life plays a pivotal role in the maturation of the host immune system (Gensollen et al. 2016). Multi- omics approaches and mechanistic experiments have expanded our understanding of the microbial structure and function role in both healthy and diseased states (Gilbert et al. 2018, Proctor et al. 2019). Changes in the human microbiome, especially the gut microbiome, are correlated with a wide range of diseases, including obesity, inflammatory bowel diseases (IBD), major depressive disorder and cancer; these are mainly metabolic, immunological and mental health disorders (Reviewed in Gilbert et al. 2018, Gopalakrishnan et al. 2018). For example, a significant decrease in Bacteroidetes and an increase in Firmicutes in the gut can result in obesity (Ley 2010). Mammalian gut bacteria can assist in the breakdown of refractory or toxic dietary compounds (Flint et al. 2008, Koropatkin et al. 2012, Brune 2014), producing many vital metabolites, such as short-chain fatty acids (SCFAs), bile acids, choline, and specific vitamins (Nicholson et al. 2012, Rowland et al. 2017). Importantly, some of the microbe-derived metabolites (e.g., SCFAs, bile acids) can directly modulate the host immunity and influence the host healthy states, which is one of the modes of microbiome-immune interactions (Lloyd-Price et al. 2019, Zheng et al. 2020). The microbiome- immune interactions facilitate the developing translational therapies (Gilbert et al. 2018). For example, bacterial probiotics can promote immune checkpoint blockade therapy for patients with cancer (Sivan et al. 2015, Matson et al. 2018, Routy et al. 2018).

1.1.2 Host-symbiont interactions in other vertebrates

The gut (Youngblut et al. 2019, Song et al. 2020) and skin (Reviewed in Ross et al. 2019) microbial communities have been particularly well explored in a wide range of vertebrate species. Recent large scale 16S rRNA sequence studies have characterised and compared the gut microbiome patterns across all vertebrate lineages, including members of Mammalia, Aves, Reptilia, Amphibia, and Actinopterygii (Youngblut et al. 2019, Song et al. 2020) and the skin microbiome in Mammalia (Ross et al. 2018). Various factors can affect their microbial diversity and abundance, such as host diet, evolutionary history (Youngblut et al. 2019, Song et al. 2020), geographic location, body region, biological sex, and living environments (Ross et al. 2018, Ross et al. 2019). Host phylogeny is the most significant factor influencing nonhuman mammalian skin microbiota (Ross et al. 2018). The nonflying mammalian gut microbial diversity is strongly correlated with host diet and phylogeny, and many microbial species are specific to a particular kind of mammal (Youngblut et

2 al. 2019, Song et al. 2020). In contrast, many gut microbes appear to be shared across different species in flying mammals and birds, whose gut microbiomes are only very weakly correlated to diet or host phylogeny (Song et al. 2020).

Compared with humans, many fewer microbial genomes have been recovered for the other vertebrates. A few species' gut microbial genomes are available so far, including the experimental model, mice (Liu et al. 2020) and ruminant livestock (Seshadri et al. 2018, Stewart et al. 2018, Stewart et al. 2019). The rumen microbiota can ferment indigestible plant polysaccharides into nutrients (e.g. SCFAs), which can then be used by the host ruminant livestock for growth (Seshadri et al. 2018, Stewart et al. 2018, Stewart et al. 2019). The ruminant microbial genomes reveal the taxon-specific metabolic pathways of polysaccharide degradation, SCFA production, methanogenesis, and de novo synthesis of vitamin B12 (Seshadri et al. 2018). These genomic resources expand our understanding of the functions of specific symbiotic microbes.

1.1.3 Host-symbiont interactions in invertebrates

There is a growing recognition that host-symbiont interactions are widespread across the animal kingdom (McFall-Ngai et al. 2013, Douglas 2018). To date, the studies on invertebrate-microbe interactions have focused mainly on a select group of holobionts, that include evolutionarily- or ecosystem-important animals (e.g. Hydra, squid, coral and sponge), biomedical models (e.g. Drosophila melanogaster, Caenorhabditis elegans, zebrafish and mosquito), and economically important animals (e.g. honeybee and shrimp) (Douglas 2019, Gao et al. 2020, Holt et al. 2020, O’Brien et al. 2020, Vanwonterghem and Webster 2020). In these holobiont systems, the symbiotic microbes have been shown to exert critical influences in host metabolism, growth, development, reproduction, and immunity (Douglas 2019, Gao et al. 2020, Holt et al. 2020, O’Brien et al. 2020, Vanwonterghem and Webster 2020). Among the biomedical models, Drosophila, C. elegans, zebrafish are traditional animals to study development, neurobiology and behaviour, and genetic diseases (Douglas 2019). The laboratory strains harbour low diversity microbiomes and can be experimentally manipulated to obtain axenic (microbe-free) or gnotobiotic (with defined microbial community) hosts, which can facilitate microbiome research (Douglas 2019). In the mosquito, fertilisation between sperm from Wolbachia-infected males and eggs from uninfected females results in the death of the zygotes (Yen and Barr 1971), which can be used to control mosquito populations and block pathogen transmission (Gao et al. 2020). The microbiome status of economically important animals, honeybee and shrimp, can affect the host health and growth (Zheng et al. 2017, Holt et al. 2020).

The host-symbiont interactions are also widely studied in a handful of evolutionarily- or ecosystem- important marine invertebrates, such as squid, Hydra, corals and sponges. The single symbiont, Vibrio fischeri can regulate the function of the light organ of its host squid Euprymna scolopes (Henares et al. 2012). Hydra belongs to the basal phylum Cnidaria; it has a simple nerve net and can produce neuropeptides that can control its microbial community (Klimovich and Bosch 2018). In turn, the symbiotic bacteria can modulate body contractions in Hydra, which may interact with the neuronal receptors (Murillo-Rincon et al. 2017, Klimovich and Bosch 2018). Research on the interaction between Hydra nerve cells and symbiotic bacteria might provide new insight into the origin and first role of the nervous system (Klimovich and Bosch 2018). Corals and sponges play essential roles in marine nutrient cycling, and their symbiotic microbiomes make critical contributions to host health and nutrition (Apprill 2017). For example, the disruption of the symbiosis between the coral host and its endosymbiotic algae can result in coral bleaching (Rosenberg et al. 2009). The symbiotic microbes are involved in nutrient assimilation in sponge holobiont (Rix et al. 2017, Rix et al. 2020).

1.2 Interconnecting the host and symbionts

1.2.1 Research methods to interconnect the host and symbionts

Multi-omics analyses facilitate the understanding of the interconnection between the host and symbionts. Integrated analyses of metadata with multi-omics datasets are increasingly being used to decipher the diverse symbiotic microbial compositions and metabolic activities (Abram 2015, Heinken and Thiele 2015, Lagier et al. 2016, Proctor et al. 2019). 16S rRNA gene amplicon pyrosequencing is widely performed to characterise the composition and diversity of the symbiotic microbiome (Poretsky et al. 2014). Beyond this simple community characterisation, metagenomics has revolutionised the understanding of the functional role of specific organisms in the holobiont system (Lagier et al. 2016). Some symbionts can be isolated from the host by serial dilution and cultured with specific dietary compounds (Matthies et al. 2009) or microbes (Strandwitz et al. 2019), whose whole genomes can be directly sequenced. But the vast majority of symbiotic microbes are uncultured microorganisms (Lewis et al. 2010). With advances in sequencing technology and computational methods, metagenomic sequencing has been used to obtain microbial genomes directly without isolation or culturing from high-complexity microbial communities (Parks et al. 2017, Tully et al. 2018, Nayfach et al. 2019, Pasolli et al. 2019, Stewart et al. 2019).

Whole-genome and metagenomic sequencing of these symbionts can offer insight into the potential metabolic capacities of various microbiota and their interactions with the host (Walker et al. 2014,

Browne et al. 2016, Lagier et al. 2016, Moitinho-Silva et al. 2017, Proctor et al. 2019). Metatranscriptomics and metaproteomics can further reveal information about gene expression and regulatory networks, and metabolomics can reveal the intermediates and end-products of metabolism (Abram 2015, Rowland et al. 2017, Fettweis et al. 2019, Lloyd-Price et al. 2019, Zhou et al. 2019). Stable isotope probing (SIP) combined with these '-omics' techniques can be used to explore the fate of specific compounds within complex microbial systems (Tannock et al. 2014). Finally, mathematical models can be used to integrate these omics datasets, encompassing metagenomics, metatranscriptomics, metaproteomics, metabolomics and SIP-omics, to provide unprecedented insight into the function and of the host-bacteria system (Abram 2015, Heinken and Thiele 2015). With these integrated methodologies, metabolic pathways and regulatory networks involving symbiotic microbes, have begun to be unveiled especially in humans and other bilaterians (Moitinho-Silva et al. 2017, Proctor et al. 2019, Strandwitz et al. 2019).

1.2.2 Hosts and symbionts need to communicate with each other

To maintain the animal-microbiome symbiotic homeostasis and execute their respective regulatory functions in the holobiont system, hosts and symbionts need to communicate with each other. Hosts and symbionts can communicate with each other through a variety of signal molecules, including many hormones (Hughes and Sperandio 2008), neurotransmitters (Strandwitz 2018), and quorum- sensing (QS) signal molecules (Li et al. 2019). To demonstrate these signal communications, some examples of the signal interplay between the host and symbiont are provided below, with a focus on QS signal molecules and neurotransmitters.

Host signal to symbionts

In some systems, animal hosts have been found to communicate with their symbiotic bacteria through QS (Li et al. 2019). The host may be able to modify the bacterial QS signal molecules, called autoinducers (AIs), and thus can regulate the symbiotic bacterial phenotype (Pietschke et al. 2017). For example, the cnidarian Hydra vulgaris can control colonisation of its primary bacterial coloniser Curvibacter sp. via modifying the autoinducer, N-(3-oxododecanoyl)-L-homoserine lactone, to N-(3-hydroxydodecanoyl)-L-homoserine lactone (Pietschke et al. 2017). The host may also be able to activate the symbiotic bacterial QS by directly releasing its own signal molecules (Davidson et al. 2004) that may even be QS mimics (Ismail et al. 2016, Li et al. 2019). As another example, in the squid-Vibrio symbiosis system, the squid host Euprymna scolopes produces the signal molecule nitric oxide (NO). NO can participate in QS of the bacterial symbiont Vibrio fisheri, through LuxU, and eventually initiate its colonisation (Davidson et al. 2004) and regulate its

5 bioluminescence (Henares et al. 2012).

Host hormonal signals can cross-kingdom communicate with symbiotic bacterial two-component systems to modulate the bacterial gene expression (Sperandio et al. 2003, Hughes and Sperandio 2008). The major animal hormone signal molecules include amino acid derivatives (amines), peptides, and steroids (Neave 2007). Many of these molecules are capable of inter-kingdom signalling with symbiotic microbes (Hughes and Sperandio 2008). For example, human released amine hormones, adrenaline and noradrenaline, can activate bacterial histidine sensor kinase QseC, which is a part of the two-component system, and thus regulate bacterial virulence gene expressions (Hughes and Sperandio 2008). Through this mechanism, the host can modulate the gut microbial composition to adapt to stress (Lyte et al. 2011, Collins et al. 2012).

Symbiotic signal to the host

As well as receiving signals from the host, symbiotic bacteria can send signals to their host through metabolites that can act as neuroactive molecules. The most widely studied symbiotic metabolites are those produced by the human gut microbiota (Martin et al. 2018). The gut microbiota releases various signal metabolites, including NO (Sobko et al. 2006), acetylcholine, serotonin, dopamine, norepinephrine, γ-aminobutyric acid (GABA), trace amines (Galland 2014, Mazzoli and Pessione 2016) and SCFAs (Silva et al. 2020). These are neuroactive molecules that can, both directly and indirectly, communicate with the host central and enteric nervous, endocrine, and immune systems (Carabotti et al. 2015, Mazzoli and Pessione 2016, Martin et al. 2018, Silva et al. 2020). For example, the SCFAs produced by microbiota in the human large intestine can bind to G protein- coupled receptor (GPCR) free fatty acid receptor 2 and 3 (FFAR 2/3), which are found on a wide range of cell types, including colonocytes, immune cells and enteroendocrine cells (Brown et al. 2003, Le Poul et al. 2003, Nilsson et al. 2003, Morrison and Preston 2016, Silva et al. 2020). Through the interplay of these metabolic signals, the gut microbiota can affect the host physiology, behaviour and health (Collins et al. 2012, Martin et al. 2018, Silva et al. 2020).

Microbe-associated molecular patterns (MAMPs) produced by symbionts might also act as signals by interacting with the animal immune pattern-recognition receptors (PRRs) in both vertebrates and invertebrates (Chu and Mazmanian 2013), such as Toll-like receptors (TLRs) and peptidoglycan recognition proteins (PGRPs). MAMPs are conserved microbe-specific molecules, such as bacterial flagellin, peptidoglycan (PGN), lipopolysaccharides (LPS) (Newman et al. 2013), that can trigger host PRR signalling pathways and promote a range of host immune responses (Parker et al. 2018); in some cases, they have also been found to induce host tissue development (McFall-Ngai et al.

2012) (McFall-Ngai et al. 2013). For example, PGN released from symbiotic gut bacteria activates Drosophila melanogaster PGRPs, which can regulate the production of antimicrobial peptides (AMPs), and eventually, maintain intestinal homeostasis and prevent overactive immune responses (Chu and Mazmanian 2013). In the squid-Vibrio symbiosis system, PGN and LPS from the symbiont V. fisheri promote the light organ developmental programs of the host E. scolopes (McFall-Ngai et al. 2012). This MAMP-PRR communication supports and maintains animal- microbe symbiotic relationships.

1.3 Marine sponge holobionts

1.3.1 Sponges host diverse and often abundant microbial symbiont communities

Marine sponges (phylum Porifera) often contain strikingly dense and complex microbial communities (Thomas et al. 2016, Webster and Thomas 2016, O’Brien et al. 2020, Steinert et al. 2020). In some high microbial abundance (HMA) sponges, microbial communities can constitute up to 40% of the sponge tissue volume, mainly located in the sponge's extracellular matrix (mesohyl) (Vacelet and Donadey 1977). 16s rRNA gene amplicon pyrosequencing has been widely used to characterise the sponge microbial composition and diversity at the phylum-level (Reveillaud et al. 2014, Moitinho-Silva et al. 2017, O’Brien et al. 2020, Steinert et al. 2020). Diverse bacterial phyla are known to be associated with sponges (Thomas et al. 2016, Webster and Thomas 2016, O’Brien et al. 2020, Steinert et al. 2020). Phylogenetic surveys have reported dominant groups, including Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Nitrospirae, Poribacteria, Proteobacteria, and Planctomycetes (Pita et al. 2018, Engelberts et al. 2020, Steinert et al. 2020). Depending on the species, each new generation of host sponge acquires certain symbiotic microbial species from the maternal adult (vertical transmission) or from the ambient environment (horizontal transmission), or both (Schmitt et al. 2008, Sipkema et al. 2015, Fieth et al. 2016, Webster and Thomas 2016, Björk et al. 2019). The compositions of symbiont communities can vary during ontogenetic (Fieth et al. 2016, Wu et al. 2018, Sacristán-Soriano et al. 2019) and geographical (Griffiths et al. 2019) changes.

1.3.2 The role of sponges in marine ecosystems

Marine sponges are functionally essential components of the benthic environment from the tropics to the poles (Bell 2008). In addition to their biomass dominance and their roles in bioerosion and stabilisation of benthic substrates, these sessile, filter-feeding animals are crucial ecosystem engineers in many marine habitats because of their impressive seawater filtration capacity

(thousands of litres per kilogram of sponge per day) (Bell 2008, Hentschel et al. 2012). The constant flow of seawater through the sponge that is created by ciliated choanocytes, provides food to, and removes metabolic end products from, the adult sponges. Sponges rapidly uptake and fix dissolved organic matter (DOM) (de Goeij et al. 2013), the most abundant resource produced on coral reefs. DOM provides the sponge with some simple biochemicals (amino acids, simple sugars, vitamins, fatty acids), and complex biopolymers (proteins, polysaccharides, lignins) (Repeta 2015), for the sponge biomass growth and cell proliferation. In turn, sponges release detritus or particulate organic matter (POM) in the form of shed cells and other material, which is subsequently ingested by other reef fauna (de Goeij et al. 2013, Pawlik et al. 2016, Rix et al. 2016, Rix et al. 2020). Thereby, sponges make nutrients and energy retained in DOM available to organisms at higher trophic levels.

Recent studies have reported that, in addition to the host sponges (Achlatis et al. 2019), the symbiotic bacteria are also participating in nutrient cycling in the marine environment (de Goeij et al. 2008, van Duyl et al. 2008, Rix et al. 2017, Gantt et al. 2019). Isotope tracing experiments have revealed that both the host sponge cells and symbiotic bacteria are actively assimilating DOM (de Goeij et al. 2008, Rix et al. 2016, Rix et al. 2017, Rix et al. 2020). Notably, in the HMA sponge Aplysina aerophoba, the symbiotic microbes account for the majority (65-87%) of DOM assimilation (Rix et al. 2020). Metagenomic and metatranscriptomic approaches are beginning to reveal the metabolic pathways involved by the sponge symbionts (Fiore et al. 2015, Germer et al. 2017, Moitinho-Silva et al. 2017, Weigel and Erwin 2017, Engelberts et al. 2020). An increasing number of studies have reported various metabolic capacities of the sponge-associated microbes, including carbon metabolism (Kamke et al. 2013, Gauthier et al. 2016, Moitinho-Silva et al. 2017, Bayer et al. 2018), and nitrogen transformation (Fiore et al. 2015, Li et al. 2015, Moitinho-Silva et al. 2017, Engelberts et al. 2020). These experimental and bioinformatic results shed light on the roles of sponge-associated microbes in the nutrient assimilation in the marine ecosystem. To demonstrate the nutrient assimilation capacities of sponge-associated microbes, some examples of metabolic pathways within sponge-associated microbes are provided below, with a focus on the metabolism of carbon and nitrogen, which are the common elements in living organisms (Alberts et al. 2002) and the essential nutrients in the open ocean (Repeta and Boiteau 2017).

1.3.3 Symbiont bacterial assisted sponge nutrient assimilation

Carbon metabolism in sponge holobionts

Oceanic dissolved organic carbon (DOC) is one of the largest carbon reservoirs on earth (Lønborg

8 et al. 2020) and marine sponges play an important role in assimilating these DOC (de Goeij et al. 2013). Isotope tracing experiments have reported that DOC assimilation involves both the host sponge and symbiotic microbes (Rix et al. 2017, Rix et al. 2020). Recent genomic and transcriptomic sequence analyses have provided insight into sponge holobiont carbon metabolism pathways. The gene repertoires of sponge symbiotic bacteria suggest their capacities for heterotrophic metabolism (Hallam 2006, Siegl et al. 2011, Moitinho-Silva et al. 2017). Genes involved in glycolysis, the tricarboxylic acid (TCA) cycle, as well as the pentose phosphate pathway (PPP), are present in the genomes of many heterotrophic sponge-associated bacteria (Hallam 2006, Siegl et al. 2011, Kamke et al. 2013, Li et al. 2015, Moitinho-Silva et al. 2017). These metabolic pathways indicate the potential capabilities of symbionts to assimilate a wide range of carbon and sugar compounds for energy and nutrition needs. The gene repertoires of some sponge-associated bacteria also show their potential autotrophic carbon assimilative capacity via different strategies (Hallam 2006, Siegl et al. 2011, Li et al. 2015). The crenarchaeote, which is symbiotically-associated with the axinellid sponge, Cenarchaeum symbiosum, appears to use a modified 3-hydroxypropionate cycle for carbon fixation (Hallam 2006). The candidate phylum Poribacteria appear to use the reductive TCA cycle and the reductive acetyl CoA pathway (Wood– Ljungdahl pathway) to fix carbon (Siegl et al. 2011). The symbionts of the deep-sea sponge Neamphius huxleyi have two alternative paths, the reductive TCA cycle and the Calvin-Benson cycle, to synthesise organic matter from carbon dioxide (CO2) (Li et al. 2015). These predicted carbon metabolic pathways of sponge symbionts reveal the carbon cycling capacity of sponge- associated bacteria in aquatic ecosystems.

Nitrogen metabolism in sponge holobionts

Nitrogen is an essential nutrient for all living organisms as it is the major component of amino acids and nucleic acids (Kuypers et al. 2018). It is also one of the main nutrients that limit biological productivity in the ocean (Gruber 2008, Kuypers et al. 2018). Nitrogen fixation and transformation (nitrification, denitrification, and ammonia oxidation) have been quite extensively studied in sponge holobionts over the past 40 years (Zhang et al. 2019). Sponges (holobiont) can both assimilate and release dissolved inorganic nitrogen (DIN, e.g. ammonia, nitrite, and nitrate) (Jiménez and Ribes 2007, Fiore et al. 2013), and dissolved and particulate organic nitrogen (DON and PON respectively) (Rix et al. 2017, Zhang et al. 2019, Rix et al. 2020). Both stable isotope and omics analyses have reported the sponge-associated microbes are involved in nitrogen cycling on coral reefs (Fiore et al. 2015, Li et al. 2016, Moitinho-Silva et al. 2017, Weigel and Erwin 2017, Rix et al.

2020). The most abundant chemical form of nitrogen, dissolved nitrogen gas (N2), is only bioavailable to some nitrogen-fixing bacteria and archaea and generally unavailable to other marine

9 organisms (Gruber 2008). The nitrogen-fixing bacteria, e.g., cyanobacteria, are identified in a wide range of sponges and present nitrogenase activity (Wilkinson and Fay 1979, Weisz et al. 2007, Zhang et al. 2014, Fiore et al. 2015, Ribes et al. 2015). Stable isotope analyses have reported that nitrogen fixed by nitrogen-fixing symbionts can be transferred to the host sponge in either inorganic or organic forms (Wilkinson et al. 1999, Mohamed et al. 2008, Fiore et al. 2013). However, nitrogen fixation by symbiotic bacteria is not ubiquitous in sponges (Ribes et al. 2015).

Metagenomic and metatranscriptomic studies have also revealed sponge-associated bacteria can utilise ammonia, nitrite and nitrate (Thomas et al. 2010, Siegl et al. 2011, Fiore et al. 2015, Moitinho-Silva et al. 2017) through ammonia oxidation, nitrite oxidation, and denitrification (Hoffmann 2009). Many genes involved in inorganic nitrogen cycling (e.g. nar genes encoding a nitrate oxidoreductase, nirK genes encoding for nitrite reductases, and formamidase-encoding gene) are active in sponge-associated symbionts (Fiore et al. 2015, Moitinho-Silva et al. 2017). For example, a metatranscriptomic study of Cymbastela concentrica holobiont reported a notable example of nitrogen cycling between sponge and symbionts, and among the symbionts (Moitinho- Silva et al. 2017). Heterotrophic Phyllobacteriaceae (CcPhy), nitrite-dependent chemolithoautotrophy Nitrospira (CCNi), and ammonia-dependent chemolithoautotrophy Thaumarchaeota (CcThau) all coexist in C.concentrica tissue (Esteves et al. 2016). Sponge metabolic products (creatine and creatinine) can be degraded into glycine and urea by CcPhy and CcThau (Moitinho-Silva et al. 2017). Urea cannot be broken down by CcPhy, but can be transported into CcNi and then decomposed into CO2 and ammonia (Moitinho-Silva et al. 2017). CcPhy derives ammonium for nitrogen assimilation, while chemolithoautotrophy CcThau acquires energy through ammonia oxidation, releasing nitrite, which can be deoxidated to nitrate in CcNi to generate energy (Moitinho-Silva et al. 2017). Nitrogenous wastes excreted by sponge are efficiently recycled in the sponge system and benefit for the existence of symbionts in sponge tissues (Moitinho-Silva et al. 2017).

1.4 The Amphimedon queenslandica holobiont to study host- symbiont crosstalk

1.4.1 The existence of draft genomes for both the host and the primary symbionts

Currently, our understanding of the molecular-level mechanisms of sponge-microbe cooperation in nutrient assimilation is still limited by the lack of genomic information for both sponge hosts and their symbiotic microbes. Availability of a hologenome provides a fundamental advantage to

10 deciphere integrated host-symbiont mechanisms. Plenty of sponge symbiont genomes have been assembled through metagenomic binning strategy (Tian et al. 2014, Gauthier et al. 2016, Slaby et al. 2017, Karimi et al. 2018, Engelberts et al. 2020), and the gene repertoires have revealed the versatile metabolism capacities of symbiotic bacteria, such as carbon, nitrogen, or sulfur metabolism (Gauthier et al. 2016, Moitinho-Silva et al. 2017, Slaby et al. 2017, Karimi et al. 2018, Engelberts et al. 2020). But only a few sponge genomes are available so far (Reviewed in Slaby et al. 2019), including Amphimedon queenslandica (Srivastava et al. 2010), Ephydatia muelleri (Kenny et al. 2020), and Tethya wilhelma (Francis et al. 2017). Among these genome-available sponges, the genome assemblies of the predominant symbiotic microbes are only published for A. queenslandica (Gauthier et al. 2016).

The existence of draft genomes for both the host (Srivastava et al. 2010) and the primary symbionts (Gauthier et al. 2016) of the tropical demosponge A. queenslandica makes this holobiont a valuable system for studying host-symbiont metabolism. This genomic data is complemented by 16S rRNA gene amplicon sequencing and electron microscopy that together have been used to characterise the bacterial symbiont community at different developmental stages throughout the life cycle of the sponge host (Fieth et al. 2016). Importantly, that work revealed a relatively low diversity microbiome inhabiting A. queenslandica, dominated (comprising 90% of the bacterial community) by just three proteobacterial OTUs (AqS1, AqS2 and AqS3) in the adult sponge (Fieth et al. 2016, Gauthier et al. 2016), draft genomes of which have been assembled and annotated (Gauthier et al. 2016). More specific, AqS1 and AqS3 belong to the order Chromatiales and Oceanospirillales, respectively; and AqS2 is a betaproteobacterium (Fieth et al. 2016, Gauthier et al. 2016). These three symbionts are vertically inherited in A. queenslandica and maintain proportional dominance through life cycle except for larval settlement and metamorphosis stage, when an influx of environmentally-derived bacteria occurs (Fieth et al. 2016). Especially, gammaprotebacterium AqS1 is the most prevalent symbiont in A. queenslandica across the life cycle, and appears to be commonly phagocytosed by sponge cells especially during the swimming larval stages (Fieth et al. 2016). AqS1 and AqS2 are equipped with complete or near-complete glycolysis pathway, TCA cycle, and pentose phosphate pathway, which suggests their capability to assimilate carbohydrate (Gauthier et al. 2016). They can also assimilate nitrogen by biosynthesis of most amino acids (Gauthier et al. 2016), including the essential amino acids of the sponge (Srivastava et al. 2010, Munroe et al. 2019, Song et al. 2020). Their genomic information indicates their capacities to synthesise some B vitamins. These foundational genomic resources suggest the carbon and nitrogen assimilating ability of the symbiont bacteria and the symbiotic bacteria might provide the host sponge essential amino acids and B vitamins, which might be a potential source of host-symbiont

11 metabolic interactions in the A. queenslandica holobiont.

1.4.2 The nutrient status on Heron Island Reef

Meanwhile, the characterised environmental nutrient status on Heron Island Reef, where A. queenslandica adults are collected (Hooper J 2006), is an additional resource to study host- symbiont cooperation in nutrient assimilation. The variability of the major nutrients, such as total organic carbon, dissolved organic carbon, total nitrogen, sulfate, and phosphorous, have been recorded throughout the year (Watson et al. 2017). Carbon and sulfate are the most abundant nutrients, and nitrogen and phosphorous are the limited recourses in the seawater in all seasons (Watson et al. 2017). The average total organic carbon (TOC) concentrations vary between 1 mg/L and 4 mg/L across the whole year, with the average sulfate concentrations range from 1.7 mg/L to 5 mg/L (Watson et al. 2017). While the highest average total nitrogen and phosphorous concentration in the year is only 0.17 mg/L and 0.04 mg/L, respectively (Watson et al. 2017). These nutrient availabilities might facilitate the understanding of sponge-bacterial metabolic interactions if we learnt about the nutrition assimilation activities of the sponge and its symbiotic bacteria.

Chapter 2 An improved Amphimedon queenslandica hologenome assembly reveals how three proteobacterial symbionts can extend the metabolic phenotypic of their marine sponge host

2.1 Abstract

Amphimedon queenslandica is an Indo-Pacific demosponge that hosts a relatively low complexity extracellular microbiome; three maternally-inherited bacterial symbionts (AqS1, AqS2 and AqS3) can comprise as much as 90% of the total bacterial symbiont abundance in adult sponges. Here I present improved assemblies and updated annotation of this hologenome. The use of in vitro reconstituted chromatin (Chicago) and short paired-end Illumina data, combined with existing reference genome assemblies, has improved the scaffold length N50 of the A. queenslandica genome by approximately 8-fold and has reduced sponge and symbiont scaffold number by 71% and 44%, respectively. With the liftover of sponge gene models from previous assemblies, there was no appreciable improvement in sponge protein-coding sequences (CDS). With prokaryotic genome annotation by Prokka, the symbiotic reassemblies have a decrease in the number of CDS in AqS1 (27%), and an increase in the number of CDS in AqS2 (20%) and AqS3 (21%). The improved assemblies of this low microbial abundance sponge holobiont allow for a more complete view of gene function in host and symbionts, and interactions between the animal and bacterial partners. Remarkably, I find that the three bacterial symbionts contribute 45.2% of the total number of KEGG-annotated metabolic genes, compared to 54.8% from the sponge host, highlighting the extent to which the symbioses serve to extend the metabolic phenotype of the holobiont. The symbionts contribute enormously (72% of total gene number) to the total membrane transport capabilities of the holobiont, further indicating the potential for exchange of diverse molecules between host and symbionts, and between symbionts, via the extracellular matrix that houses the bacteria. All three symbiont genomes encode genes for antimicrobial drug resistance, which may assist the vertically-inherited symbionts in maintaining their dominance over bacteria that are horizontally acquired by young sponges once they settle onto the benthic substrate and begin metamorphosis.

2.2 Introduction

Marine sponges (poriferans) play crucial roles in nutrient recycling in nutrient-poor ecosystems such as coral reefs (de Goeij et al. 2013, Folkers and Rombouts 2020). Sponges often contain strikingly dense and complex prokaryotic communities (Webster and Thomas 2016, O’Brien et al. 2020, Steinert et al. 2020) that may play important roles in the assimilation of nutrients by extending the metabolic phenotype of the sponge host (de Goeij et al. 2008, van Duyl et al. 2008, Rix et al. 2017, Gantt et al. 2019, Rix et al. 2020). In this context, holobiont genome assemblies and functional annotation provide a foundation for exploring mechanisms of evolution, physiology, metabolism and mechanisms of interaction in sponge symbioses (Fiore et al. 2015, Germer et al. 2017, Moitinho-Silva et al. 2017, Weigel and Erwin 2017). Already, significant insight has been achieved by determining the gene content of numerous bacterial symbiont genomes. For example, genes involved in glycolysis, the tricarboxylic acid (TCA) cycle, and the pentose phosphate pathways are encoded in the genomes of several sponge-associated bacteria (Hallam 2006, Siegl et al. 2011, Li et al. 2015, Moitinho-Silva et al. 2017, Bayer et al. 2018), indicating the potential capability of symbionts to assimilate a wide range of carbon and sugar compounds for energy and nutritional needs of the sponge holobiont. Further, ammonia, nitrite and nitrate excreted by the sponge can likely be assimilated by sponge-associated bacteria through ammonia oxidation, nitrite oxidation, and denitrification (Fiore et al. 2015, Moitinho-Silva et al. 2017, Zhang et al. 2019, Folkers and Rombouts 2020). The field of sponge-bacterial symbiosis research will benefit from the increased availability of both host and symbiont genomes, so that potential sources of metabolic interactions and complementation between specific partners can be revealed.

The haploscerid demosponge A. queenslandica hosts a low complexity and low abundance microbiome dominated by just three proteobacterial symbionts, AqS1, AqS2 and AqS3, which are inherited vertically via the maternal transfer of core symbionts into oocytes and early embryos (Fieth et al. 2016). These three symbionts are dominant in A. queenslandica through life cycle except for larval settlement and metamorphosis stage, when an influx of environmentally-derived bacteria occurs (Fieth et al. 2016). In adult sponges, these three symbionts alone can comprise up to 90% of the microbiome; their presence is stable throughout the year in the Heron Island reef field site, although the relative proportions can shift a little across seasons (Gauthier et al. 2016). The draft genomes of the three symbionts have been published (Gauthier et al. 2016). Gene content of the symbiont genomes indicates, for example, the potential for sulfur oxidation, carbon monoxide oxidation and inorganic phosphate assimilation, all of which may support the metabolism of the host and contribute to nutrient recycling in the coral reef habitat of the sponge (Gauthier et al.

2016).

In addition, the content, structure and organisation of the A. queenslandica host genome are well characterised (Srivastava, Simakov et al. 2010). Revised A. queenslandica gene models, Aqu2.1, increased the total gene number in this sponge by 46%, providing a more comprehensive set of genes for functional and comparative studies (Fernandez-Valverde et al. 2015). This genome, in combination with developmental and cell-specific transcriptomes, has shed light on the origin of allorecognition (Grice et al. 2017) and immunity genes (Yuen et al. 2014), regulatory evolution (Gaiti et al. 2017, de Mendoza et al. 2019) and cell type and animal body plan evolution (Levin et al. 2016, Sebé-Pedrós et al. 2018, Sogabe et al. 2019).

To better decipher the nature of interactions between A. queenslandica and its dominant bacterial symbionts, here I revised the assemblies of the genomes of A. queenslandica, AqS1, AqS2 and AqS3 by HiRise pipeline, which can correct contig order and orientation errors, and improve the contiguity in genome assemblies (https://dovetailgenomics.com/ga_tech_overview/) by incorporating additional in vitro reconstituted chromatin (Chicago) data (Putnam et al. 2016). The resultant improved hologenome assembly allows for a more comprehensive view of the potential metabolic and signalling processes and functional interactions between host sponge and symbiotic bacteria.

2.3 Materials and Methods

2.3.1 Holobiont sampling and sequencing

Two A. queenslandica adults were collected in January 2016 from Shark Bay, Heron Island Reef, Australia (Latitude −23.44, Longitude 151.92) and their total genomic DNA (gDNA) were extracted as described in (Srivastava et al. 2010). The Chicago method was used to reconstitute in vitro chromatin for gDNA from one sponge; this reconstituted chromatin was further digested and sheared for long-range mate-pair library preparation (Putnam et al. 2016). The sample was prepared in triplicate, and three Chicago libraries (2 x 101 bp read pairs with insert sizes ranging between 1 to 300 kb) were sequenced on an Illumina HiSeq2000 platform at Dovetail Genomics (Santa Cruz, USA). An Illumina library was prepared using the gDNA from the second sponge, using a TruSeq® Nano DNA LT Library Preparation Kit. This library was sequenced on an Illumina NextSeq500 platform, to produce four lanes of 2 x 75 bp pair-end Illumina reads with an average insert size of 350 bp.

2.3.2 Hologenome assemblies

The HiRise workflow (Putnam et al. 2016) was used to reassemble the A. queenslandica sponge and the three dominant symbiotic bacterial genomes. First, both the Chicago mate-pair reads, and the Illumina pair-end reads were aligned to the existing assembly of the sponge A. queenslandica genome (Aq1) (Srivastava et al. 2010) and of the three primary symbiotic bacterial genomes (Gauthier et al. 2016) using a modified version of SNAP (1.0dev.67_as) (https://github.com/robertDT/dt-snap). This generated the bam files specifically for the HiRise assembler (Putnam et al. 2016). Then, reads that paired aligned to each species were extracted with bedtools (v2.26.0) (Quinlan and Hall 2010) and used to improve the existing sponge holobiont genomes using HiRise (version HiRise_July2015_GR) (Putnam et al. 2016). The completeness of the new genome assemblies was assessed by BUSCO (version 3.0.2) (Simao et al. 2015). Specifically, the new A. queenslandica assembly, hereafter referred to as Aq3, was compared with the metazoa_odb9 set and the three new bacterial assemblies were compared with the proteobacteria_odb9 set.

2.3.3 Hologenome annotation

For A. queenslandica, the gene models of the improved genome assembly were generated from the published Aqu2.1 A. queenslandica gene models (Fernandez-Valverde et al. 2015) by a liftover annotation pipeline (https://github.com/wurmlab/flo), hereafter referred to as Aqu3.1. The three bacterial genomes were annotated by Prokka (version 1.12) (Seemann 2014).

2.3.4 Functional annotation of holobiont gene models

Holobiont gene models were assigned GO terms and InterPro IDs using Blast2GO (version 5.2.4) with the default Blast2GO 3-steps (Blast, Mapping, Annotation) Gene Ontology annotation workflow (Conesa and Gotz 2008). Blastp-fast was used to search the non-redundant protein sequences (nr) database with e-value 1.0e-3. Functional pathways involving holobiont gene models were annotated by GhostKOALA searching against the genus_prokaryotes and family_eukaryotes Kyoto Encyclopedia of Genes and Genomes (KEGG) GENES database (Kanehisa et al. 2016).

2.4 Results

2.4.1 Hologenome sequencing and assembly

The A. queenslandica hologenome assemblies were improved significantly by the integration of the 16 new sequencing data. Specifically, three Chicago libraries produced a total of 277 M read pairs for the hologenome (2 x 101 bp) (Supplementary File 2-1; Supplementary File 2-2); of these, 178 M mapped to the sponge host A. queenslandica, and for the symbionts, 0.26 M mapped to AqS1, 0.023 M to AqS2, and 0.027 to AqS3 (Supplementary File 2-3). A single paired-end library (insert size 350 bp) yielded a total of 121 M Illumina read pairs for the hologenome (2 x 75 bp) (Supplementary File 2-2); of these, 59 M mapped to the sponge host Aq, and for the symbionts, 3.6 M mapped to AqS1, 0.9 M to AqS2, and 0.6 M to AqS3 (Supplementary File 2-3). With the addition of the long- range Chicago reads, the A. queenslandica contig assembly improved from 166.7 to 167.7 Mb total scaffold length, with the scaffold N50 increasing almost 8-fold, from 120 to 950 kb (Table 2-1). The maximum scaffold length increased from 1.89 to 4.60 Mb, the average scaffold length increased from 12 to 43 kb, and the scaffold number decreased from 13,397 to 3,871. The scaffold N50s of the three primary symbiotic bacterial assemblies, AqS1, AqS2, and AqS3, also increased to 102,706, 147,679 and 89,982 bp, respectively (Table 2-1)

Assembly quality and completeness of all four genomes were evaluated by BUSCO, which revealed a slight improvement in the symbiont BUSCO estimates compared to previous versions (Table 2-2). The proportion of missing genes in AqS2 improved in the new assembly, decreasing from 28.5 to 23.1% (Table 2-2).

Table 2-1. Genome assembly statistics of A. queenslandica (Aq), and the three primary symbionts, AqS1, AqS2 and AqS3. The Aq1 and v1 assemblies are the original genomes of A. queenslandica (Srivastava et al. 2010) and the three primary symbionts (Gauthier et al. 2016), respectively. Aq3 and v2 assemblies are the improved genomes of A. queenslandica and the three primary symbionts in this study.

Aq AqS1 AqS2 AqS3 Genome version Aq1 Aq3 v1 v2 v1 v2 v1 v2 Total length (bp) 166,679,601 167,703,835 4,203,406 4,207,334 1,608,701 1,625,801 3,161,579 3,170,385 Number of scaffolds 13,397 3,871 127 116 239 68 233 151 Average scaffold length (bp) 12,441 43,323 33,097 36,270 6,730 23,908 13,569 20,995 Maximum scaffold length (bp) 1,888,931 4,599,197 265,783 287,762 82,499 441,393 172,753 311,513 Scaffold N50 length (bp) 120,365 950,481 79,254 102,706 12,238 147,679 35,167 89,982 Scaffold N50 309 49 15 14 32 3 26 11

Table 2-2. Comparison of BUSCO analyses of old and new A. queenslandica (Aq), AqS1, AqS2 and AqS3 assemblies. Complete BUSCO genes are those that match a complete gene in the BUSCO reference group; Fragmented are genes that only partially match a gene in the BUSCO reference group, and Missing means that there are no genes that match. The Aq1 and v1 assemblies are the original genomes of A. queenslandica (Srivastava et al. 2010) and the three primary symbionts (Gauthier et al. 2016), respectively. Aq3 and v2 assemblies are the improved genomes of A. queenslandica and the three primary symbionts in this study.

Aq AqS1 AqS2 AqS3 Genome version Aq1 Aq3 v1 v2 v1 v2 v1 v2 Complete (%) 86.8 86.6 95.0 95.1 64.7 69.2 89.1 89.6 Fragmented (%) 2.1 2.4 0.5 0.5 6.8 7.7 2.7 2.7 Missing (%) 11.1 11.0 4.5 4.4 28.5 23.1 8.2 7.7 Total BUSCO genes 978 978 221 221 221 221 221 221

2.4.2 Hologenome annotation

The genes of the A. queenslandica and symbiotic bacterial genomes were annotated using different methods as appropriate. For the sponge host A. queenslandica, 42,644 gene models were generated in Aqu3.1 from the published 44,001 Aqu2.1 gene models (Fernandez- Valverde et al. 2015) using a LiftOver annotation pipeline FLO (https://github.com/wurmlab/flo) (Table 2-3). About 3% of Aqu2.1 gene models were lost through the LiftOver process. Of the 42,644 predicted gene model, 36,841 (86.4%) gene models were annotated with biological functions by Blast2GO (Supplementary File 2-5) (Conesa and Gotz 2008).

Table 2-3. Comparison of coding and non-coding gene number and length between old and new assemblies for A. queenslandica (Aq), and the three primary symbionts, AqS1, AqS2 and AqS3.

Aq AqS1 AqS2 AqS3 Aqu2.1 Aqu3.1 v1 v2 v1 v2 v1 v2 CDS 44,001 42,644 4,767 3,478 1,349 1,621 2,418 2,933 rRNA na na 4 3 3 3 5 1 tRNA na na 47 49 42 47 59 59 tmRNA na na 0 1 0 1 0 2 Average CDS length (bp) 2,420 2,182 765 987 934 846 1,045 936

For the bacterial symbionts, CDS, ribosomal RNA (rRNA), transfer RNA (tRNA), and transfer-messenger RNA (tmRNA) of the three genomes were annotated by Prokka (version 1.12) (Seemann 2014). Totals of 3478, 1621, and 2933 CDS were annotated in AqS1, AqS2, and AqS3, respectively (Table 2-3); together, the CDS of the three symbionts comprise 15.8% of the total hologenome CDS. In addition, the AqS1 genome contains 3 rRNA genes, 1 tmRNA and 49 tRNA, AqS2 has 3 rRNA, 1 tmRNA and 47 tRNAs, and AqS3 has 1 rRNA, 2 tmRNA and 59 tRNAs.

The annotated CDS in the three symbionts differ between current and previous genomic versions. For AqS1, 4767 CDS were identified in the original genome assembly (Gauthier et al. 2016). The new AqS1 assembly has fewer CDS (1289), but the average CDS length has

19 increased by 222 bp (Table 2-3). In contrast, 272 and 515 more CDS are annotated in the reassembled AqS2 and AqS3 genomes, respectively, with average CDS lengths that are 88 and 109 bp shorter in the respective new assemblies (Table 2-3).

2.4.3 Functional annotation of the A. queenslandica hologenome.

Gene functions for the host sponge, and the three symbiotic bacteria were characterised by identifying orthologues in the KEGG database using GhostKOALA (Kanehisa et al. 2016). Using this method, a total of 11,761 A. queenslandica genes (representing 27.6% of the Aqu3.1 gene set) were assigned to 4,846 KEGG Orthology (KO) groups (Table 2-4). For the three bacterial symbionts, 1,614 (46.4%) AqS1 CDS were assigned to 1,234 KO groups, 895 (55.2%) AqS2 CDS were assigned to 803 KO groups, and 1,032 (35.2%) AqS3 CDS were assigned to 905 KO groups (Table 2-4). Of the total hologenome genes that could be assigned to a KO group, 23.1% are from the three bacterial symbiont genomes.

Table 2-4. The number of A. queenslandica (Aq), AqS1, AqS2 and AqS3 genes classified into six broad biological categories by KEGG Mapper.

KEGG Biological category Aq AqS1 AqS2 AqS3 Cellular Processes 1,822 122 44 41 Environmental Information Processing 1,592 211 92 56 Genetic Information Processing 1,754 197 144 166 Human Diseases 2,403 89 42 62 Metabolism 1,772 655 388 419 Organismal Systems 2,344 50 21 30 Total KEGG annotated genes 11,761 1,614 895 1032

Most of these KEGG-annotated genes were hierarchically grouped based on function into six broad biological categories following (Kanehisa et al. 2017): Cellular processes, Environmental information processing, Genetic information processing, Human diseases, Metabolism, and Organismal systems (Table 2-4; Figure 2-1A). The sponge host A. queenslandica has the highest number of genes in all six major categories, as expected because of its much larger genome size and gene content overall (Table 2-1; Table 2-3). Despite this, the three symbionts together contribute a number of Metabolism genes (1,462) that is only a little less than the number (1,772) attributed to the sponge host (Table 2-4). In all, the three bacterial symbionts together are contributing 45.2 % of the total number of 20 metabolic genes to the holobiont, despite contributing only 23.1 % of all KEGG-annotated genes in the hologenome. This is a result of the fact that, of the KEGG-annotated genes grouped into the six broad biological categories in each of the four genomes, 49.47% (AqS1), 53.08% (AqS2) and 54.13% (AqS3) are attributed to Metabolism in the three symbionts, and just 15.16% are attributed to Metabolism in the host (Table 2-4; Figure 2-2).

Cellular processes Human diseases A. queenslandica AqS1 AqS2 AqS3 Environmental Information processing Metabolism Genetic information processing Organismal systems Percentage 0 25% 50% 75% 100% A Cellular Processes Environmental Information Processing Genetic Information Processing Human Diseases Metabolism Organismal Systems

B Cell growth and death Cell motility Cellular community - eukaryotes Cellular community - prokaryotes Transport and catabolism Membrane transport Signal transduction Signaling molecules and interaction Folding, sorting and degradation Replication and repair Transcription Translation Cancer: overview Cancer: speciﬁc types Cardiovascular disease Drug resistance: antimicrobial Drug resistance: antineoplastic Endocrine and metabolic disease Immune disease Infectious disease: bacterial Infectious disease: parasitic Infectious disease: viral Neurodegenerative disease Substance dependence Amino acid metabolism Biosynthesis of other secondary metabolites Carbohydrate metabolism Energy metabolism Glycan biosynthesis and metabolism Lipid metabolism Metabolism of cofactors and vitamins Metabolism of other amino acids Metabolism of terpenoids and polyketides Nucleotide metabolism Xenobiotics biodegradation and metabolism Aging Circulatory system Development and regeneration Digestive system Endocrine system Environmental adaptation Excretory system Immune system Nervous system Sensory system 23.1%

Figure 2-1. A. The percentages of A. queenslandica (Aq), AqS1, AqS2 and AqS3 genes contributed to each of the six broad biological categories by KEGG Mapper. B. The percentages of Aq, AqS1, AqS2 and AqS3 genes contributed to per sub-categories. The y-label colours of the sub-categories in B are consistent with the y-label colours of the categories in A. Of the total hologenome genes that could be assigned to a KO group, 23.1% are from the three bacterial symbiont genomes.

AqS3

AqS2

AqS1

0 25% 50% 75% 100% Percentage

Cellular Processes Genetic Information Processing Metabolism Environmental Information Processing Human Diseases Organismal Systems

Figure 2-2. The percentages of per biological categories for A. queenslandica (Aq), AqS1, AqS2 and AqS3. The percentages are calculated based on the genes that are hierarchically grouped into the six broad biological categories in Table 2-4.

This disproportionately high contribution by the bacterial symbionts to the metabolic capacity of the hologenome is further reflected in the high diversity of specific metabolic pathways attributable to symbionts, including both central and secondary metabolism (Figure 2-1B; Supplementary File 2-4). Indeed, most specific sub-categories of metabolism are over- represented in the symbiont genomes; the exception is just 1 out of 11 metabolic sub- categories, namely lipid metabolism, of which contributes a number of genes commensurate with the proportional representation of the symbiont genomes in the total hologenome (Figure 2-1B; Supplementary File 2-4). Based on the proportional contribution of genes to the KEEG functions, all three of symbionts contribute fairly equally to the metabolic sub- categories; the exception is Xenobiotics biodegradation and metabolism, to which AqS2 contributes much less than AqS1 and AqS3, reflective of the smaller genome size of the former.

A breakdown of KEGG sub-categories in the remaining (non-metabolic) biological parent categories reveals other functions that also are disproportionately represented in either A. queenslandica or in one or more of the three symbiotic bacteria (Figure 2-1B; Supplementary File 2-4). KEGG sub-categories associated specifically with the metazoan condition – that is, sub-categories related to multicellular systems, disease, development, intercellular signalling

23 and gene regulation – mostly are overrepresented by sponge genes as expected. One exception is the Disease sub-category Antimicrobial drug resistance, which is populated only by bacterial genes representing all of the three symbionts (Figure 2-1B; Supplementary File 2-4). Interestingly, genes annotated as Membrane transport within the parent category Environmental information processing are disproportionately represented in the symbiont genomes. The symbionts contribute enormously (72% of total gene number) to this sub- category, most particularly AqS1 and AqS2, reflecting the relatively high number of carrier and transport membrane proteins encoded in these genomes (Figure 2-1B; Supplementary File 2-4) (Gauthier, Watson et al. 2016). In contrast, nearly all the genes related to signalling in the same parent category – Signaling molecules and interactions, and Signal transduction – are restricted to A. queenslandica (Figure 2-1B; Supplementary File 2-4).

2.5 Discussion

Using additional in vitro reconstituted chromatin (Chicago) (Putnam et al. 2016) and short paired-end Illumina sequence data, I have improved the genome assemblies of the coral reef demosponge A. queenslandica and its three primary symbionts, AqS1, AqS2, and AqS3, and assigned biological functions to the CDS in the four revised genome assemblies. This hologenome provides an improved view of potential functional capabilities and interactions of the sponge and its symbionts, the latter of which comprise as much as 90% of the microbiome in adult sponges (Fieth et al. 2016, Gauthier et al. 2016).

2.5.1 Improvement of the A. queenslandica hologenome

The original Aq1 assembly and annotation was the first draft poriferan genome to become available, and has contributed substantially to understanding the origin of the animal kingdom and the predicted nature of the last common animal ancestor (Srivastava et al. 2010). The later inclusion of deep transcriptomic data from Amphimedon adults, larvae and juveniles resulted in a 46% increase in the estimated number of gene models from 30,060 (Aqu1) to 44,001 (Aqu2.1), and a marked improvement in estimating transcription start and stop sites (Fernandez-Valverde et al. 2015). This allowed for the analysis of promoters and other regulatory regions, and long non-coding RNAs (Gaiti et al. 2015, Fernandez-Valverde and Degnan 2016, Gaiti et al. 2017, de Mendoza et al. 2019).

The new A. queenslandica genome assembly (Aq3) reported herein has reduced the number

24 of scaffolds by 71.1%, increased the scaffold N50 length by 7.9 times, and increased the maximum scaffold length from 1.89 to 4.60 Mb. This has resulted in an increase in the total scaffold length from 166.7 to 167.7 Mb. The incorporation of multiple Aq1 scaffolds into a single Aq3 scaffold resulted in a 3.1% reduction in the estimated number of gene models from 44,001 to 42,644 when Aqu2.1 gene models were lifted-over onto the new assembly. These improvements enhance the value of the A. queenslandica genome as a comparative system for understanding the origin of animal complexity. In recent years, the availability of a number of other high-quality draft sponge genomes, including a recent chromosome-level assembly of the freshwater sponge Ephydatia muelleri, has further improved our understanding of genome structure and predicted genome content of the last common ancestor to extant poriferans (Srivastava et al. 2010, Nichols et al. 2012, Fortunato et al. 2014, Ryu et al. 2016, Ereskovsky et al. 2017, Francis et al. 2017, Kenny et al. 2020). This growing suite of poriferan genomes also provides a more robust comparative framework for broader metazoan analyses.

The A. queenslandica holobiont is considered to be relatively simple compared to many other sponges (Hentschel et al. 2012, Thomas et al. 2016, Webster and Thomas 2016, O’Brien et al. 2020, Steinert et al. 2020), and is characterised as a low complexity and low abundance microbial symbiotic community (Fieth et al. 2016, Gauthier et al. 2016). This relative simplicity – combined with the ability to trace symbiont dynamics through the sponge life cycle, including a vertical inheritance from the maternal sponge during oogenesis and early embryogenesis – allows for a detailed characterisation of the bacterial symbionts inhabiting the sponge and their contribution to normal and perturbed holobiont function during embryogenesis, larval development, metamorphosis, growth and maturation (Fieth et al. 2016).

The low complexity microbiome in A. queenslandica is dominated by just three proteobacterial symbionts that are vertically inherited, and thus comprise a large proportion of the bacterial biomass throughout the entire sponge life cycle (Fieth et al. 2016). In adult sponges, these three symbionts alone can comprise up to 90% of the microbiome (Fieth et al. 2016, Gauthier et al. 2016). The resequencing of these three symbionts has resulted in improved draft genome assemblies that have marginally increased the estimated genome sizes of AqS1, AqS2, and AqS3 by 0.1, 1.1 and 0.3%, and the scaffold N50 lengths 1.3, 12.1 and 2.6-fold, respectively. The reassemblies have resulted in a marked increase in the number

25 of CDS in AqS2 (20.2%) and AqS3 (21.3%), and a decrease in the number of CDS in AqS1 (27%).

BUSCO assessment of the quality and completeness of the new hologenome assemblies reveals that original and revised draft genomes are almost identical in terms of the percentage of conserved genes present in these genomes, except for AqS2 which improved by about 5%. AqS2 had the most substantial increase in genome assembly quality, with a scaffold N50 increase from 12.2 to 147.7 kb (12.1-fold increase). At this stage, I am unable to determine if the genes missing in these BUSCO analyses are related to the quality of the current genome assemblies, or are an actual reduction in the size and gene content of the symbiotic bacterial genome (Baumgartner et al. 2017, Boscaro et al. 2017). However, the lack of substantial improvement in the BUSCO scores, despite a marked improvement in the assembly, is consistent with these values being a true reflection of the number of BUSCO genes present in these genomes, rather than a reflection of genome quality.

These improved bacterial genomes provide a more accurate foundation for predicting the metabolic capabilities of the core symbionts, and how they interact with, and contribute to, the A. queenslandica holobiont. This, in turn, provides a valuable contribution to the general understanding of sponge-microbe interactions (Hentschel et al. 2012, Fiore et al. 2015, Webster and Thomas 2016, Germer et al. 2017, Moitinho-Silva et al. 2017, Weigel and Erwin 2017, Zhang et al. 2019, Engelberts et al. 2020).

2.5.2 Functional partitioning of the A. queenslandica hologenome

The overall improvement of the assemblies and gene models of A. queenslandica, AqS1, AqS2, and AqS3 allows me to provide the first-ever comprehensive view of specific biological functions and contributions of each core partner to a sponge hologenome. This was done by assigning hologenome CDS to KEGG Orthology (KO) groups, which then could be assigned a function according to the known function of orthologues in the KEGG database (Kanehisa et al. 2016). In total, 27.6% of the Aqu3.1 gene models can be assigned to a KO group, while 46.4, 55.2 and 35.2% v.2 models of AqS1, AqS2 and AqS3, respectively, can be assigned to a KO group. Of the total number of assignable CDS in the hologenome, 23.1% are bacterial.

The KEGG analyses of the four genomes reveal the distribution of gene functions and the

26 metabolic capacities across the bacterial symbionts and the host sponge A. queenslandica. Because genes can be assigned specifically to one of the four core partners in the symbiosis, the relative contributions of each partner can be compared and contrasted. Although, as expected based on genome size and gene number, A. queenslandica has the most genes in all six broad KEGG categories – Cellular processes, Environmental information processing, Genetic information processing, Human diseases, Organismal systems and Metabolism – the three bacterial symbionts together contribute a remarkable 45.2% of the total number of Metabolism genes to the hologenome. This is despite only 23.1% of the KO group genes in the hologenome being derived from the bacterial symbionts, and highlights the extent to which the symbioses extend enormously the metabolic phenotype of the sponge holobiont.

Indeed, more than 40% of KEGG-annotated CDS in each of the symbionts encode proteins involved in metabolism. The bacterial symbiont disproportionately high contribution to the hologenome's metabolic capacity is further reflected in the high diversity of metabolic pathways encoded by the three symbionts. All the 11 metabolic sub-categories in KEGG have greater than 23.1% bacterial representation, and almost all of them are over-represented in the symbiont genomes, the exception is lipid metabolism (23.6%), of which is represented to a proportion that would be expected based on hologenome composition. Increasing omics studies have reported the functional metabolic potentials of sponge-associated microbes, including their capacities of carbon, nitrogen, sulfur and phosphorus metabolism (Kamke et al. 2013, Fiore et al. 2015, Gauthier et al. 2016, Moitinho-Silva et al. 2017, Bayer et al. 2018, Zhang et al. 2019). These metabolic capacities of sponge-associated microbes indicate their vital roles to support the nutrient cycle in marine sponges (Zhang et al. 2019).

The KEGG analysis also highlights the potential modes by which host and symbionts may interact through the exchange of diverse gene products, metabolites and nutrients; here, the Environmental information processing category provides particular insights. The symbiont genomes, particularly AqS1 and AqS2, encode a disproportionately large number – 72% of the total hologenome gene number in this category – of carrier and transport membrane proteins, suggesting these are the primary mechanisms by which the symbionts interact with each other and with the host sponge. In contrast, the largest proportion of KO group genes in A. queenslandica encode signalling molecules, receptors and signal transducers, suggesting that the sponge itself relies more heavily on receptor-mediated interactions with other members of the holobiont.

Also of note is the presence in all three symbiont genomes of genes encoding antimicrobial drug resistance. I hypothesise that these gene products may assist the vertically-inherited symbionts in maintaining their dominance over the environmental bacteria that are horizontally acquired by young A. queenslandica when the dispersive larvae settle onto the benthic substrate and begin metamorphosis (Fieth et al. 2016). Although the horizontal uptake of these diverse bacteria significantly briefly disrupts the microbiome composition at this stage of the life cycle, the three vertically-inherited symbionts regain their dominance by the time metamorphosis is complete, and the juvenile body plan is established (Fieth et al. 2016). I further hypothesise that the antimicrobial drug resistance capabilities of all three core symbionts may help to explain why treating A. queenslandica larvae with an antibiotic cocktail only partially and inconsistently reduces the number of symbionts, making it impossible to achieve aposymbiotic larvae that would be such a valuable experimental tool (S. Degnan, personal communication).

2.6 Conclusion

This reassembly of the A. queenslandica hologenome has improved both qualities of the sponge host and its three primary bacterial symbionts, allowing us to attribute relative contributions of each to specific holobiont functions based on gene content. This, in turn, provides a strong foundation for understanding the specific contributions of each of these four partners to the development and functioning of the holobiont through its life cycle and under changing environmental conditions. With this in mind, further details of the metabolic contributions of each the core partners separately, and of putative instances where two or more partners intersect their contributions to a single metabolic pathway, are explored more deeply in Chapter 4.

Chapter 3 Ribosomal RNA-depletion provides an efficient method for successful dual RNA-Seq expression profiling of a marine sponge holobiont

3.1 Abstract

Investigations of host-symbiont interactions can benefit enormously from a complete and reliable holobiont gene expression profiling. The most efficient way to acquire holobiont transcriptomes is to perform RNA-Seq on both host and symbionts simultaneously. However, optimal methods for capturing both the host and symbiont mRNAs are not yet well-defined, particularly when the host is a eukaryote and the symbionts are bacteria or archaea. Traditionally, poly(A)-enriched libraries have been used to capture eukaryotic mRNA, but the ability of this method to adequately capture bacterial mRNAs is unclear because of the short half-life of the bacterial transcripts. To determine the optimal RNA-Seq approach for animal host–bacterial symbiont analysis, I compared transcriptome depth and coverage achieved by two different mRNA capture and sequencing strategies applied to the marine demosponge Amphimedon queenslandica holobiont, for which genomes of the animal host and three numerically most abundant bacterial symbionts are available. The two strategies were: (i) poly(A) captured mRNA-Seq (Poly(A)-RNA-Seq); and (ii) ribosomal RNA depleted RNA-Seq (rRNA-depleted-RNA-Seq). For the host sponge, I found no significant difference in transcriptomes generated by the two different mRNA capture methods. However, for the symbiont transcriptomes, I confirmed the expectation that the rRNA-depleted-RNA-Seq performs much better than the Poly(A)-RNA-Seq. I also found that bacterial cell enrichment enables adequate capture of the symbiont transcriptomes, although it reduces the A. queenslandica host transcriptome. This comparison demonstrates that RNA-Seq by ribosomal RNA depletion is an effective and reliable method to analyse holobiont transcriptomes.

3.2 Introduction

Increasing interest in the microbes that live in association with multicellular organisms has revealed that symbioses are ubiquitous, and that plants and animals are best considered as holobionts comprising both a multicellular host and diverse microbiota (Bordenstein and Theis 2015, Rosenberg and Zilber-Rosenberg 2016). Host-associated microbes can have a critical influence on host reproduction, development, immunity, physiology and fitness (Yen and Barr 1971, Engelstadter and Hurst 2009, Pradeu 2011, Ezenwa et al. 2012, McFall-Ngai et al. 2013, Rosenberg

29 and Zilber-Rosenberg 2016, Leulier et al. 2017). To investigate these influences, the mechanisms of host-symbiont interactions are increasingly being explored by multi-omics approaches (Heinken and Thiele 2015, Rowland et al. 2017).

With recent significant reductions in sequencing costs, RNA-sequencing (RNA-Seq) has become a widely used method to study host-symbiont interactions, and is especially valuable when used to profile host and symbiont transcriptomes simultaneously. Several recent studies have applied dual transcriptome analysis to reveal potential modes of crosstalk between symbiosis partners. For instance, transcriptome analysis of the whitefly (Bemisia tabaci) complex revealed complementation in amino acid biosynthesis pathways between the whitefly and its bacterial endosymbiont Candidatus Portiera aleyrodidarum (Upadhyay et al. 2015). Dual RNA-Seq across the life cycle of the nematode Brugia malayi, and its bacterial endosymbiont Wolbachia identified developmental pathways involving both nematode and bacterial genes (Grote et al. 2017). Metatranscriptome analysis contributed to the reconstruction of metabolic pathways of the sponges and its symbiotic microbes (Germer et al. 2017, Moitinho-Silva et al. 2017).

However, the accurate and simultaneous analysis of the transcriptional state of both animal and bacterial partners in a symbiosis currently is not optimised for many holobiont systems. Transcriptomes generated via RNA-Seq tend to be dominated by highly abundant ribosomal (r)RNAs, making it necessary first to remove rRNA and enrich for coding sequence mRNAs. This need for rRNA removal creates challenges for studies wishing to capture both eukaryote and bacterial mRNAs, mainly because of differences in eukaryote and bacterial transcript processing, lifespan and decay. In the eukaryote animal hosts, mRNA transcripts generally have a half-life of many hours (Yang et al. 2003, Sharova et al. 2009) and are stabilised by the addition of a long tail of adenines (poly(A) tails; ~250 long) at the 3' end of the transcript (Westermann et al. 2012, Perez- Ortin et al. 2013, Dendooven et al. 2020). These relatively stable mRNAs with long poly(A) tails can be easily captured and enriched away from rRNAs by the use of oligo (dT) primers, to generate libraries for sequencing; this is standard practice for eukaryote mRNA-Seq. In contrast, bacterial symbiont mRNAs generally have a much shorter half-life of only a few minutes on average (Selinger et al. 2003), and are transiently polyadenylated with short poly(A) tails (<50 As) (Westermann et al. 2012, Dendooven et al. 2020) that tag the transcript for degradation, rather than for stabilisation (Dreyfus and Regnier 2002). Thus, the most common choice for rRNA depletion in bacterial studies is to use a subtractive hybridisation method via commercial kits (see Petrova et al. (2017) for comparison of different kits) before RNA-Seq analyses.

These features suggest that holobiont transcriptome analyses will require different methods to

30 capture mRNAs from eukaryote and bacterial partners before RNA-Seq. In studies of marine sponges, well recognised as exemplar animal-bacterial holobionts, numerous strategies have been used to try and capture transcriptomes from both host and symbionts (Fiore et al. 2015, Germer et al. 2017, Moitinho-Silva et al. 2017). To reconstruct metabolic networks linking the host sponge Cymbastela concentrica, a diatom and three proteobacterial symbionts (Moitinho-Silva et al. 2017), combined a eukaryote poly(A) capture with a separate bacterial rRNA depletion method. The need to treat eukaryote and bacterial RNA separately is labour- and cost-intensive. Thus, a single method that obtains accurate mRNA representation from host and microbe is preferred. To this end, both sponge host and bacterial symbiont transcriptomes were obtained from the giant barrel sponge Xestospongia muta holobiont by applying only a eukaryote rRNA depletion step (Fiore et al. 2015); it appears in that study bacterial rRNAs were retained, but there is no analysis as to whether these interfered with the depth of mRNA reads acquired. In contrast, transcriptomes of both the sponge Vaceletia sp. and its bacterial symbionts were assessed by using only poly(A) capture method (Germer et al. 2017). Based on the differences in eukaryotic and bacterial mRNA stability and processing described above, an oligo (dT) enrichment method is likely to capture only a portion of the bacterial mRNAs. Germer et al. (2017) did not explore the consequences of this for an accurate representation of the symbiotic partners in their study.

Accurate determination of the fraction of expressed eukaryote and bacterial genes that are captured by poly(A) enrichment compared to rRNA depletion methods requires that transcripts can be mapped back to assembled genomes of all partners. When this is not possible because genomes are not available for both host and symbionts, it is unclear to what extent the transcriptomes represent an accurate picture of the contributions of various partners in the symbiosis. To quantitatively compare the efficacy of Poly(A)-RNA-Seq versus rRNA depletion-RNA-Seq for the capture of both animal host and bacterial symbiont transcriptomes, here I apply both strategies to an analysis of gene expression in the demosponge Amphimedon queenslandica holobiont. A. queenslandica hosts a low diversity microbiome that is dominated by three proteobacterial symbionts (AqS1, AqS2 and AqS3) in the adult stage (Fieth et al. 2016). The genomes of both the host sponge (Srivastava et al. 2010, Fernandez-Valverde et al. 2015) and of the three primary symbionts have been assembled and annotated (Gauthier et al. 2016), with recent improvements in quality by incorporation of in vitro reconstituted chromatin (Chicago) and short paired-end Illumina data (Chapter 2). These foundational genomic resources allow me to align transcript reads precisely to their associated genomes, making the A. queenslandica holobiont a useful system to compare the power of the two RNA-seq library construction methods.

3.3 Materials and Methods

3.3.1 Sample collection and sequencing

Adult A. queenslandica were collected in June 2018 from Heron Island Reef, Great Barrier Reef, Queensland, Australia (Latitude −23.44, Longitude 151.92) as described previously (Leys et al. 2008). Six separate tissue biopsies, each approximately 3 cm3, were sampled immediately after collection as a source of RNA for this study. To increase the proportional representation of the low abundance bacterial symbionts in the samples prior to total RNA extraction, bacteria were enriched by a series of centrifugation and filtration steps following the protocol of Thomas et al. (Thomas et al. 2010). Bacterial cell enrichments were visualised by microscopy to determine approximate proportions of sponge cells to bacteria cells using DAPI (4',6-Diamidino-2-Phenylindole, Dihydrochloride) nucleic acid stain following the manufacturer’s protocol.

Total RNA was extracted separately from the six replicated biopsies using TRIzol (Sigma-Aldrich) and contaminating DNA was removed with Deoxyribonuclease I (Invitrogen), both following the manufacturer' s protocol. RNA was quantified by Qbit® 2.0 Fluorimeter (Invitrogen), and quality was assessed by 1% TAE agarose gel electrophoresis. Illumina libraries were constructed separately from each of the six total RNAs using the TruSeq Stranded mRNA Library Prep Kit, with details as follows.

Three of the samples were subjected to the standard TruSeq Stranded mRNA-seq workflow, using the poly-A pulldown and standard protocol with 1 µg total RNA as input. The other three samples (2.5 µg total RNA each) were first rRNA-depleted using the Ribo-Zero Gold ribosomal RNA Removal Kit (Epidemiology, supplied by Illumina), which removes animal, Gram-positive and Gram-negative bacterial cytoplasmic and mitochondrial rRNA. 100 ng of the depleted RNA was then used as input to the mRNA-seq protocol, but omitting the poly-A pulldown and instead of starting at the RNA fragmentation step.

From the fragmentation step onwards, the TruSeq protocol was followed for all six samples, using random hexamer priming for the generation of cDNA. I used 12 PCR cycles for the amplification. Each library was sequenced on the Illumina NextSeq500 platform by four runs. Both library preparation and sequencing were performed at Ramaciotti Centre for Genomics, Sydney, Australia. Thus, all six samples were treated identically throughout except for their random assignment to either poly(A) capture or rRNA depletion.

3.3.2 Read processing and alignment

Raw paired-end sequences of 75 base pairs (bp) in length were processed using Trimmomatic (version 0.36) (Bolger et al. 2014) to crop the first 10 bp of each read, and to trim the reads using a 4 bp sliding window and an average quality threshold of 20. Both unpaired reads and resulting reads smaller than 60 bp were discarded. The remaining quality-filtered, paired-end reads were aligned to the updated genomes (Chapter 2) of A. queenslandica and those of the three most numerically dominant bacterial symbionts – AqS1, AqS2 and AqS3 – using HISAT2 (version 2.0.5) with default parameters (Pertea et al. 2016). Both unaligned reads, and reads that aligned to more than one of these four genomes and thus could not be unambiguously assigned, were discarded. The read mapping bam file of each sample was split into four bam files based on the scaffold id of each species by SAMtools (version 1.3) (Li et al. 2009).

3.3.3 Comparative assessment of gene depth and gene coverage

The number of reads that aligned to the sponge or bacterial protein-coding sequences were counted by htseq-count (version 0.11.2.) from HTSeq framework with default parameters except for -- stranded=reverse (Anders et al. 2015). These aligned reads were then converted to transcript counts per gene as a measure of gene depth. In addition, for each genome, gene coverage was estimated as the percentage of genes that were represented in the RNA-Seq data (that is, expressed gene number divided by total gene number). These two measures were compared within and between libraries prepared by each of the two different methods, and with previously generated adult transcriptomes for this sponge.

3.3.4 Comparison with an RNA-Seq dataset generated previously from non- bacterial-enriched sponge sample

To specifically assess the effect of bacterial enrichment on the ability to recover holobiont transcriptomes, I compared the data generated as described above with previously published deep RNA-Seq data (Fernandez-Valverde et al. 2015). In that study, a tissue biopsy from one adult sponge was used to prepare Poly(A)-RNA libraries, but with no bacterial enrichment step. The resultant stranded sequenced RNA-Seq reads were used in this comparison because the sequence strategy was consistent with the data generated in this study. I directly compared this previously published data with that of bacterial-enriched Poly(A)-RNA-Seq data from the current study. To do so, the raw paired-end reads were filtered by Trimmomatic (parameters: SLIDINGWINDOW:4:15 MINLEN:60 HEADCROP:7) (version 0.36) (Bolger, Lohse, & Usadel, 2014). The remaining high-

33 quality pair-end reads were aligned to the holobiont genomes as described above. Gene counts and gene coverage were also calculated from these data as described above.

3.3.5 Comparison of gene function among different data sets

To investigate the possibility of functional biases resulting from differential capture of transcripts, Gene Ontology (GO) analyses were performed for the genes captured by the three different library preparation methods, rRNA-depleted-RNA-Seq, Poly(A)-RNA-Seq and bacterial-unenriched Poly(A)-RNA-Seq. GO term annotations of the holobiont gene models were performed using Blast2GO (version 5.2.4) (Conesa and Gotz 2008), and the number of genes assigned to each annotated GO term was calculated for each species. Expressed genes in each sample were identified as those to which at least one read pair mapped, and expressed GO terms were considered as those for which there was at least one gene expressed that could be attributed to the GO term. To identify any differences in the GO functions among the different RNA-Seq libraries, the expressed GO percentages were estimated in each sample for the sponge host and for each of the three bacterial symbionts, separately. That is, for each of the four species, the number of expressed GO terms in a given sample was divided by the total number of GO terms annotated in that species; the three replicated samples per RNA-Seq dataset were averaged. As several genes might exert the same GO functions, for each expressed GO term, the number and proportion of expressed genes were also calculated for each RNA-Seq data set; the three replicated samples per RNA-Seq dataset were averaged.

3.4 Results

3.4.1 Sequencing and alignment profile

Using four lanes of Illumina NextSeq500, I generated 388 and 403 million reads for the three replicate rRNA-depleted-RNA-Seq samples (A, B, C) and the three Poly(A)-RNA-Seq samples (D, E, F), respectively. After filtering low-quality reads with Trimmamatic (Bolger et al. 2014), more than 86% of reads (~336 and ~354 million reads for rRNA-depleted and Poly(A) samples, respectively) remained for subsequent analysis. For the additional bacterial-unenriched Poly(A)- RNA-Seq transcriptomes, 341 million reads were generated for the four samples, and 320 million reads remained after the quality filter step. All these high-quality reads were then aligned to improved genomes of the sponge A. queenslandica, and its proteobacterial symbionts, AqS1, AqS2 and AqS3 by HISAT2 (version 2.0.5) (Pertea et al. 2016). On average, 69.32% rRNA-depleted- RNA-Seq reads, 81.24% Poly(A)-RNA-Seq reads, and 63.14% bacterial-unenriched Poly(A)-RNA-

Seq reads could be mapped to the four genomes. Of these mapped reads, 4.79% rRNA-depleted- RNA-Seq reads, 0.11% Poly(A)-RNA-Seq and few bacterial-unenriched Poly(A)-RNA-Seq reads mapped to more than one genome, and thus were discarded because they could not be unambiguously assigned. Finally, 222 million mapped rRNA-depleted-RNA-Seq reads, 287 million mapped Poly(A)-RNA-Seq reads, and 51 million mapped bacterial-unenriched Poly(A)-RNA-Seq reads remained for further analysis.

For the bacterial-enriched rRNA-depleted-RNA-Seq libraries, 76.99% of the mapped reads were attributed to the sponge host Aq, while 19.19%, 2.54%, and 1.28% were attributed to AqS1, AqS2, and AqS3, respectively (Figure 3-1). By comparison, for the bacterial-enriched Poly(A)-RNA-Seq libraries, almost all the aligned reads (99.17%) were attributed to the sponge host; only 0.60%, 0.18%, and 0.05% reads were attributed to bacterial symbiont genomes AqS1, AqS2, and AqS3, respectively (Figure 3-1). For the bacterial-unenriched Poly(A)-RNA-Seq library, 99.94% of mapped reads were attributed to the sponge host, and only 0.034% to the three symbiotic bacteria in total.

A B AqS3: 1.28% AqS2: 2.54%

AqS1:19.19%

AqS3: 0.05% AqS2: 0.18% AqS1: 0.60% Aq:77.00% Aq:99.17%

rRNA-depleted-RNA-Seq Poly(A)-RNA-Seq

Figure 3-1. Taxonomic distribution of reads, which is the percent of reads aligned to each species in the sponge holobiont for rRNA-depleted- and Poly(A)-RNA-Seq libraries. Aq is the genome of A. queenslandica; AqS1, AqS2 and AqS3 are the genomes of the three primary bacterial symbionts.

3.4.2 Correlation between samples

The reads aligned to each gene were considered as a measure of gene expression. Correlation of the gene expression between the technical triplicates was estimated by gene read counts (gene depth) between individuals within each experimental group. All the genes read counts were included in calculating the Spearman correlation efficiencies. For the host sponge A. queenslandica, the gene

35 expression correlation was strong among the triplicates from both rRNA-depleted- and Poly(A)- RNA-Seq groups, with Spearman correlation efficiencies ranging from 0.907 to 0.969 (Figure 3-2). For the three primary symbiotic bacteria, the correlation of gene expression varied between the biological replicates from each experimental group. For the rRNA-depleted-RNA-Seq data, the Spearman correlation efficiencies of AqS1 biological replicates ranged from 0.967 to 0.987, with AqS2 from 0.916 to 0.941, and AqS3 from 0.826 to 0.889. These were higher than those correlation efficiencies of the Poly(A)-RNA-Seq replicates, with AqS1 ranged from 0.915 to 0.920, AqS2 from 0.843 to 0.855, and AqS3 from 0.695 to 0.733. Overall, the strong correlations between the biological replicates from each experimental group suggested the gene expression of the four species in the A. queenslandica holobiont were comparable within the rRNA-depleted- and the Poly(A)-RNA-Seq samples.

A rRNA-depleted-RNA-Seq C C C C

0.930 0.987 0.941 0.889 B B B B

A 0.907 0.907 A 0.967 0.980 A 0.916 0.939 A 0.826 0.853 Correlation coeffcient 1.0

0.5 B Poly(A)-RNA-Seq 0.0 F F F F −0.5 0.969 0.920 0.855 0.733 −1.0 E E E E

D 0.966 0.965 D 0.915 0.916 D 0.853 0.843 D 0.695 0.695

Aq AqS1 AqS2 AqS3

Figure 3-2. Correlation of expressed genes between biological replicates estimated for the A. queenslandica holobiont. Aq is A. queenslandica; AqS1, AqS2 and AqS3 are the three primary symbiotic bacteria.

3.4.3 Gene coverage

To better understand the influence of RNA-Seq library preparation method for capture of both animal host and bacterial symbiont transcriptomes, I compared the percentage of genes to which RNA-Seq reads were mapped across the different methods (Figure 3-3). On average, a very similar 53.93% (+0.0443) and 57.38% (+0.0148) of A. queenslandica genes were covered by the rRNA- depleted-RNA-Seq data and Poly(A)-RNA-Seq data, respectively (two-tailed t-test, p-value 0.308), but lower than the bacterial unenriched-Poly(A)-RNA-Seq data, which gene coverage was 66.62% for A. queenslandica (Table 3-1). For the bacterial symbiont genomes, on average, 89.70% (+0.0082) AqS1 genes, 95.11% (+0.0029) AqS2 genes, and 80.36% (+0.0260) AqS3 genes were captured by rRNA-depleted-RNA-Seq libraries. These gene coverages were much higher than those captured by Poly(A)-RNA-Seq libraries, which, on average, was 70.07% (+0.0119) for AqS1, 65.99% (+0.0181) for AqS2, and 44.46% (+0.0259) for AqS3 (one-tailed t-test, p-values were 2.54e- 05 in AqS1, 4.98e-04 in AqS2, and 3.56e-05 in AqS3); and also higher than that of bacterial unenriched-Poly(A)-RNA-Seq sample, which was 18.14% in AqS1, 24.06% in AqS2, and 1.13% in AqS3 (Table 3-1). This gene coverage comparison indicated that rRNA-depleted-RNA-Seq and Poly(A)-RNA-Seq could capture the animal host transcriptomes at the same level, but rRNA- depleted-RNA-Seq was better at obtaining the bacterial symbiont transcriptomes. The efficacy of the unenriched-Poly(A)-RNA-Seq was much higher for capturing the sponge host transcriptomes than the bacterial symbiont transcriptomes, compared to the rRNA-depleted-RNA-Seq and Poly(A)- RNA-Seq.

Transcript quantification is one of the most common applications of RNA-Seq, which estimates the gene expression levels based on the number of reads mapped to each gene. This approach is widely used to identify differential expression (DE) genes between different treatments and contexts (Conesa et al. 2016). Many DE software packages filter genes with <5 reads before identifying DE genes because of both biological and statistical concerns, removing genes with low read counts prior to downstream analysis (Chen et al. 2016). In this study, I filtered the genes with <5 read pairs and then calculated the gene coverage, which was the percentage of genes often used for downstream analyses, such as DE analysis. On average, 50.94% (+0.0362) and 49.92% (+0.0104) A. queenslandica genes were covered with at least 5 read pairs in the rRNA-depleted-RNA-Seq data and Poly(A)-RNA-Seq data, respectively, with 58.66% of the genes represented in bacterial unenriched-Poly(A)-RNA-Seq data (Table 3-1). These values are consistent with the estimated proportion of genes in the genome expressed in adult A. queenslandica (Fernandez-Valverde et al 2015)

For the symbiont bacterial genes, on average, 86.23% (+0.0079) AqS1 genes, 93.65% (+0.0087) AqS2 genes and 75.63% (+0.0344) AqS3 genes were captured with at least 5 read pairs in rRNA- depleted-RNA-Seq data. By comparison, for the Poly(A)-RNA-Seq data, on average, only 54.22% (+0.0016) AqS1 genes, 41.07% (+0.0123) AqS2 genes, and 19.57% (+0.0130) AqS3 genes were captured. 3.22% AqS1 genes, 5.43% AqS2 genes, and 0.20% AqS3 genes were captured from the bacterial unenriched-Poly(A)-RNA-Seq sample (Table 3-1). These gene coverages indicated rRNA- depleted-RNA-Seq allowed a similar proportion of sponge genes and a higher proportion of bacterial genes that could be used for downstream analysis, compared to Poly(A)-RNA-Seq. For the unenriched-Poly(A)-RNA-Seq sample, the downstream analysis could be performed on the sponge transcriptome but not the symbiotic bacterial transcriptomes because if insufficient coverage of the latter.

1.00

0.75 ge a

ver 0.50 o Gene C

0.25

0.00

Aq AqS1 AqS2 AqS3

rRNA-depleted-RNA-Seq Threshold Poly(A)-RNA-Seq 1readpair unenriched-Poly(A)-RNA-Seq 5readpairs

Table 3-1. Average gene coverage of rRNA-depleted-RNA-Seq, and Poly(A)-RNA-Seq, and bacterial unenriched-Poly(A)-RNA-Seq data for A. queenslandica (Aq) and the three primary symbiont (AqS1, AqS2 and AqS3) genomes. The gene coverage was calculated by dividing the expressed gene number by total gene number. The three replicated samples per RNA-Seq dataset were averaged. Two cutoff thresholds were set to calculate the expressed gene number. A cutoff of 1 read pair means that genes with at least 1 read pair aligned were considered as expressed, and a cutoff of 5 read pairs means that only genes to which at least 5 read pairs mapped were considered as expressed.

Cutoff 1 read pair 5 read pairs Species Aq AqS1 AqS2 AqS3 Aq AqS1 AqS2 AqS3 rRNA-depleted-RNA-Seq 53.93% 89.70% 95.11% 80.36% 50.94% 86.23% 93.65% 75.63% Poly(A)-RNA-Seq 57.38% 70.07% 65.99% 44.46% 49.92% 54.22% 41.07% 19.57% Unenriched-Poly(A)-RNA-Seq 66.62% 18.14% 24.06% 1.13% 58.66% 3.22% 5.43% 0.20%

3.4.4 Distribution of read depth in gene coding regions

Sequence depth is a crucial measure of the robustness of RNA-Seq data. I calculated the gene depth by counts of aligned reads per gene, and compared the gene depth distribution of the rRNA- depleted-, Poly(A)- and bacterial unenriched-Poly(A)-RNA-Seq samples (Figure 3-4; Supplementary File 3-1). For the host sponge A. queenslandica, on average, the median gene depths were 110 (+15.95) and 113 (+1.73) for the rRNA-depleted- and Poly(A)-RNA-Seq samples, and was 83 for the bacterial unenriched-Poly(A)-RNA-Seq adult (Table 3-2). For each bacterial symbiont, the rRNA-depleted-RNA-Seq data showed the highest average gene depths, which were much higher than those of the Poly(A)-RNA-Seq data and bacterial unenriched-Poly(A)-RNA-Seq data. On average, the rRNA-depleted-RNA-Seq median gene depths were 446 (+89.37) in AqS1, 162 (+84.50) in AqS2, and 47 (+13.75) in AqS3 (Table 3-2); the Poly(A)-RNA-Seq median gene depths were 14 (+0.58) in AqS1, 7 (+0.58) in AqS2, and 4 (+0) in AqS3. The median gene depth of all the three primary symbiotic bacterial was just two for the bacterial unenriched-Poly(A)-RNA- Seq sample. The similar read depth distributions of A. queenslandica for rRNA-depleted- and Poly(A)-RNA-Seq samples implied that there was no noteworthy gene depth difference among the host sponge transcriptome captured by both methods, but sponge transcript depths were lower obtained by bacterial unenriched-Poly(A)-RNA-Seq (Figure 3-4). The symbionts' gene depths were much higher in the rRNA-depleted-RNA-Seq data than the Poly(A)-RNA-Seq data, indicating that the rRNA-depleted-RNA-Seq captured significantly deeper symbiont bacterial transcripts than Poly(A)-RNA-Seq and bacterial unenriched-Poly(A)-RNA-Seq.

1000 600 200

1000 750 150

400

500 100 Gene Depth 500

200

250 50

0 0 0 0

Aq AqS1 AqS2 AqS3

rRNA-depleted-RNA-Seq Poly(A)-RNA-Seq Unenriched-Poly(A)-RNA-Seq

Figure 3-4. Boxplot of the number of reads mapped to each expressed gene in rRNA-depleted-RNA-Seq, Poly(A)-RNA-Seq, and Poly(A)-RNA-Seq data without bacterial enrichment data (Unenriched-Poly(A)- RNA) for A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3).

Table 3-2. The average median number of reads aligned per expressed gene

Type Aq AqS1 AqS2 AqS3 rRNA-depleted-RNA-Seq 110 446 162 47 Poly(A)-RNA-Seq 113 14 7 4 Unenriched-Poly(A)-RNA-Seq 83 2 2 2

3.4.5 Comparison of biological functions of expressed genes in the different RNA-Seq data sets

A first step towards understanding the gene and biological functions in host and symbionts is to annotate expressed and enriched genes in RNA-Seq data sets using Gene Ontology (GO) (Ashburner et al. 2000). GO analyses were performed and compared between rRNA-depleted- RNA-Seq, Poly(A)-RNA-Seq and bacterial unenriched Poly(A)-RNA-Seq using Blast2GO (version 5.2.4) (Conesa and Gotz 2008) to determine if any of the RNA-Seq approaches resulted in a bias in the genes captured. For all the holobiont genes, 28,570 A. queenslandica genes, 2,417 AqS1 genes, 1,193 AqS2 genes, and 1,977 AqS3 genes were assigned to 6,357, 1,317, 904, and 1,030 GO terms, respectively (Table 3-3).

Table 3-3. Number of annotated GO terms for genes of A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3).

Species Annotated gene number Assigned GO number Aq 28,570 6,357 AqS1 2,417 1,317 AqS2 1,193 904 AqS3 1,977 1,030

To identify any differences in the GO functional annotations among the different RNA-Seq libraries, the percentage of expressed GO terms in each sample was calculated for each species. That is, for a given species, if one of the genes, assigned to a particular GO term, was expressed with at least 1 read pair in a given sample, this GO term was considered as expressed GO term. The expressed GO percentage was estimated as the percentage of GO terms that were represented in the RNA-Seq data (Table 3-4; Figure 3-5). For A. queenslandica, almost all the GO terms were present in the rRNA-depleted-RNA-Seq data (95.88% + 0.0085), Poly(A)-RNA-Seq data (97.43% + 0.0036) and bacterial unenriched-Poly(A)-RNA-Seq data (98.47%). For the bacterial symbionts, the expressed GO term percentages were 98.96% + 0.0022 in AqS1, 99.34% + 0.0011 in AqS2 and 98.32% + 0.0059 in AqS3 for the rRNA-depleted-RNA-Seq samples (Table 3-4). These percentages of expressed GO terms in rRNA-depleted-RNA-Seq samples were higher than in Poly(A)-RNA-Seq samples, which was 92.63% + 0.0055 in AqS1, 86.32% + 0.0103 in AqS2, and 79.58% + 0.0136 in AqS3 (one-tailed t-test, p-values were 3.68e-04 in AqS1, 9.37e-04 in AqS2, and 1.91e-04 in AqS3); and also higher than that of bacterial unenriched-Poly(A)-RNA-Seq sample, which was 38.72% in AqS1, 46.02% in AqS2, and 2.91% in AqS3 (Table 3-4). These GO term expression percentages

43 indicate that the rRNA-depleted-RNA-Seq, Poly(A)-RNA-Seq and bacterial unenriched-Poly(A)- RNA-Seq reveal a similar functional transcriptome profile of the host A. queenslandica. In contrast, rRNA-depleted-RNA-Seq provides a near-complete functional representation of the symbiotic bacteria, which is much higher than Poly(A)-RNA-Seq and bacterial unenriched-Poly(A)-RNA-Seq (Table 3-4; Figure 3-5).

Table 3-4. Average expressed GO percentage in rRNA-depleted-RNA-Seq, Poly(A)-RNA-Seq, and Poly(A)- RNA-Seq data without bacterial enrichment data (Unenriched-Poly(A)-RNA) for A. queenslandica (Aq) and the 3 primary symbionts (AqS1, AqS2 and AqS3). Expressed GO percentage was estimated as expressed GO number divided by the total GO number annotated in each species.

Type Aq AqS1 AqS2 AqS3 rRNA-depleted-RNA-Seq 95.88% 98.96% 99.34% 98.32% Poly(A)-RNA-Seq 97.43% 92.63% 86.32% 79.58% Unenriched-Poly(A)-RNA-Seq 98.47% 38.72% 46.02% 2.91%

1.00 ● ● ● ● ● ● ● ● ● ● ●

● ● 0.75

0.50 ●

● Percentage of expressed GOs of expressed Percentage

0.25

● 0.00

Aq AqS1 AqS2 AqS3

● rRNA−depleted−RNA−Seq ● Poly(A)−RNA−Seq ● Unenriched−Poly(A)−RNA−Seq

Figure 3-5. Expressed GO percentage in each sample for A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3).

GO functional profiles of expressed or over-represented (enriched) genes in these three RNA-Seq data sets were compared to determine if GO categories were equally represented in the different types of RNA-Seq libraries; the three replicated samples per RNA-Seq data set were averaged (Figure 3-6; Supplementary File 3-2). This approach revealed differences in the representation of specific GO functions in each of the libraries (Table 3-5; Figure 3-6). For the sponge A. queenslandica, on average, the expressed gene percentage of each GO term was higher in the bacterial unenriched-Poly(A)-RNA-Seq data (93.64%) compared to the rRNA-depleted-RNA-Seq data (86.57%) and the Poly(A)-RNA-Seq data (90.62%). For each bacterial symbiont, the rRNA- depleted-RNA-Seq data showed the highest GO term expressed gene percentages compared to the

Poly(A)-RNA-Seq data and bacterial unenriched-Poly(A)-RNA-Seq data (Figure 3-6; Table 3-5). The rRNA-depleted-RNA-Seq average GO term expressed gene percentages of the three symbionts were similar, 95.57%, 98.62% and 95.62% for AqS1, AqS2, and AqS3, respectively (Table 3-5). In contrast, the average GO term expressed gene percentages of Poly(A)-RNA-Seq and bacterial unenriched-Poly(A)-RNA-Seq for the three symbionts varied considerably. For the Poly(A)-RNA- Seq, the average GO term expressed gene percentages decreased from 84.49% in AqS1, to 77.45% in AqS2 and 67.85% in AqS3. For the bacterial unenriched-Poly(A)-RNA-Seq, this percentage decreased from 21.88% in AqS1 and 31.27% in AqS2 to 0.43% in AqS3 (Table 3-5). These GO term percentages suggest that for the symbionts, the Poly(A)-RNA-Seq method produces bias in the functional types of genes that are captured, as indicated by GO enrichment results (Supplementary File 3-3).

Aq AqS1 AqS2 AqS3

Percentage 100

0 GO Term

Type rRNA−depleted−RNA−Seq Poly(A)−RNA−Seq Unenriched−Poly(A)−RNA−Seq 0 7 0 5 0 5 0 5 14 10 10 10 Log2GeneNum Log2GeneNum Log2GeneNum Log2GeneNum

Figure 3-6. Distribution of the percentage of expressed genes in each GO term. The heatmaps present the average percentage of genes expressed in each GO term detected in A. queenslandica (Aq) and the three dominant symbionts (AqS1, AqS2 and AqS3) in the rRNA-depleted-, Poly(A)- and bacterial unenriched- Poly(A)-RNA data sets. The dot plot next to each heatmap shows the total gene numbers attributed to each GO term with the x-axis on a log2 scale.

Table 3-5. Average percentage of GO terms represented in the expressed genes of A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3) in the three RNA-Seq data sets.

Type Aq AqS1 AqS2 AqS3 rRNA-depleted-RNA-Seq 86.57% 95.57% 98.62% 95.62% Poly(A)-RNA-Seq 90.62% 84.49% 77.45% 67.85% Unenriched-Poly(A)-RNA-Seq 93.63% 21.88% 31.27% 0.43%

I used pairwise comparisons of expressed gene numbers to identify which functional groups of genes, as indicated by GO terms, are most differently captured by the different library types. For each pairwise comparison, the most differentially represented 30 GO terms were identified as those with the biggest differences of expressed gene numbers. Examples include 4 biological process GOs (oxidation-reduction process, proteolysis, transmembrane transport, and translation), 4 cellular component GOs (integral component of membrane, membrane, cytoplasm, and plasma membrane) and 13 molecular function GOs (i.e. catalytic activity, hydrolase activity, transferase activity, ATP binding, and DNA binding) (Figure 3-7). Overall, compared to unenriched-Poly(A)-RNA data, the rRNA-depleted-, and Poly(A)-RNA data sets (Figure 3-7) under-represented sponge genes involved in integral component of membrane, membrane, ATP binding, and protein binding (Figure 3-7). For the three symbiotic bacteria, compared to the rRNA-depleted RNA data, the Poly(A)- and bacterial unenriched-Poly(A)-RNA data underrepresented bacterial genes involved in oxidation-reduction, transmembrane transport, integral component of membrane, membrane, and ATP binding (Figure 3-7).

oxidation-reduction process translation proteolysis transmembrane transport integral component of membrane membrane cytoplasm plasma membrane ATP binding DNA binding hydrolase activity transferase activity metal ion binding catalytic activity structural constituent of ribosome oxidoreductase activity protein binding nucleotide binding transmembrane transporter activity methyltransferase activity nucleic acid binding 0 1000 2000 3000 4000 5000 0 100 200 300 4000 50 100 150 0 100 200 300 Aq AqS1 AqS2 AqS3 Biological Process Average expressed gene number Cellular Component Molecular Function rRNA-depleted-RNA-Seq Poly(A)-RNA-Seq Unenriched-Poly(A)-RNA-Seq

Figure 3-7. The average expressed gene numbers of some of the most differentially represented GO terms in the rRNA-depleted-, Poly(A)- and bacterial unenriched-Poly(A)-RNA data sets. For a given species, if the average expressed gene numbers for a GO term are not presented, that GO term was not in the top 30 most different represented GO terms among the three RNA library types.

3.5 Discussion

3.5.1 Transient polyadenylation in bacteria reduces mRNA capture in Poly(A)- RNA-Seq

Sequencing depth and coverage are central to assessing the efficacy of RNA-Seq (Sims et al. 2014). In this metatranscriptome analysis of the A. queenslandica holobiont, I sought to identify an RNA- Seq approach that could yield high depth and coverage for genes expressed in both the sponge host and its three proteobacterial symbionts. Under conditions used in this study – i.e. libraries from bacterial-enriched cell preparations from adult sponges – I found that rRNA-depleted- and Poly(A)- RNA-Seq capture sponge host transcripts at similar levels with at least 5 read pairs aligned (on average, 50.94% and 49.92% of genes in the genome). In contrast, although the rRNA-depleted- RNA-Seq captured a large proportion of the expressed genes in the symbionts (86.23% AqS1, 93.65% AqS2, and 75.63% AqS3 of genes per genome), only partial bacterial transcriptomes were obtained by Poly(A)-RNA-Seq (54.22% AqS1; 41.07% AqS2; 19.57% AqS3 of genes per genome). This difference reflects the different capacities of these two RNA-Seq approaches to efficiently capture bacterial mRNAs, which are relatively unstable (Selinger et al. 2003) and are polyadenylated just prior to degradation (Dreyfus and Regnier 2002). The Poly(A)-RNA approach captures a smaller bacterial mRNAs that may be those that are transiently polyadenylated prior to degradation.

Interestingly, the bacterial mRNAs captured by Poly(A)-RNA-Seq method do not appear to be a random snapshot of the bacterial transcriptome profile. Combining all the three bacterial enriched Poly(A)-RNA-Seq samples and the bacterial unenriched Poly(A)-RNA-Seq sample, 502, 396, and 29 GO terms were represented in the AqS1, AqS2, and AqS3 expressed genes, respectively; 23 of them were expressed in all the three symbionts (Figure 3-8). 59, 50, and 93 of the annotated GO terms were not detected in AqS1, AqS2, and AqS3, respectively; and one of them, GO:0002100 (tRNA wobble adenosine to inosine editing), was not expressed in all the three symbionts (Figure 3-8). This bias appears to be in contrast to bacterial polyadenylation reported as a ubiquitous post- transcriptional modification (Maes et al. 2017) and bacterial poly(A) polymerases not having mRNA sequence specificity (Haugel-Nielsen et al. 1996). Given that both exert a general RNA destabilising function in bacteria (Dreyfus and Regnier 2002, Maes et al. 2017), the differences of the detected and undetected GO terms in the poly(A)-RNA samples in this study might be explained by differential mRNA turnover rate. Bacterial mRNA turnover is modulated by ribonucleases (RNases), RNA binding proteins, and small noncoding RNAs (sRNA) (reviewed in Anderson and

Dunman 2009), and could be facilitated by poly(A) polymerase (reviewed in Hajnsdorf and Kaberdin 2018). These regulatory processes can play a vital role in bacterial stress adaptation, cell growth and virulence factor production (reviewed in Anderson and Dunman 2009, Li et al. 2017). When investigating the total Escherichia coli RNA content, only 110 genomic regions were detected as being polyadenylated; other polyadenylated RNAs might degrade too rapidly to be detected (Maes et al. 2017). Thus variable mRNAs decay rates (reviewed in Hui et al. 2014), could yield specific pools of bacterial mRNAs captured by poly(A)-RNA-Seq in this study.

Aq AqS1 AqS2 AqS3

Percentage 100

0 GO Term

Type rRNA−depleted−RNA−Seq Poly(A)−RNA−Seq Unenriched−Poly(A)−RNA−Seq 0 7 0 5 0 5 0 5 14 10 10 10 Log2GeneNum Log2GeneNum Log2GeneNum Log2GeneNum

Figure 3-8. Distribution of the expressed gene percentage for each GO of each sample. The heatmaps present the expressed gene percentage of each gene ontology (GO) of A. queenslandica (Aq) and the three dominant symbionts (AqS1, AqS2 and AqS3) in all the rRNA-depleted-RNA-Seq, Poly(A)-RNA-Seq and bacterial Unenriched-Poly(A)-RNA samples. The dot plot next to each heatmap shows the total gene numbers attributed to each GO term. The x-axis numbers are log2 (total gene count for each GO), and the y-axis is all the GO terms.

3.5.2 Advantages and challenges of rRNA-depleted-RNA-Seq for dual RNA-Seq experiments

The most important advantage of rRNA-depleted-RNA-Seq over Poly(A)-RNA-Seq for dual RNA- Seq is its higher efficiency to capture the global tanscriptome of the holobiont. The rRNA-depleted- RNA-Seq simultaneously captures a high proportion of the expressed genes in sponge host and bacterial symbionts, which is in contrast to Poly(A)- based RNA-Seq approaches. Importantly, there was not a marked difference between sponge host transcriptome representation in rRNA-depleted- and Poly(A)-RNA-Seq data sets, which were derived from bacterial cell enrichment. In contrast, all three bacterial symbiont transcriptomes are significantly better covered in the rRNA-depleted-RNA- Seq data. The high bacterial gene coverage in these data is similar to the human-pathogen system

49 where an rRNA-depleted-RNA-Seq method captured 88% of the Haemophilus influenzae genes from in vitro human bronchial epithelium; the human transcriptome was simultaneously captured (Baddal et al. 2015). Besides, rRNA-depleted-RNA-Seq also improves detection of low abundance transcripts (Petrova et al. 2017, Kim et al. 2019) and noncoding RNA (Westermann et al. 2016), which could reveal a more complex host-symbiont transcriptome profile.

However, dual RNA-Seq requires sufficient amounts of RNA material with good quality from both host and symbionts (Westermann et al. 2012, Wolf et al. 2018). This is often a challenging step for dual RNA-seq because of the different nature and content of RNA between symbiotic microbes and host aminals (Westermann et al. 2012, Westermann et al. 2017, Wolf et al. 2018). To overcome this obstacle, enrichment of the symbiotic bacteria prior to sequencing is applied in this study. Transcriptome coverages of both libraries, rRNA-depleted- and Poly(A)-RNA-Seq, do not differ markedly from previous deep transcriptomes generated from sponge cells, i.e. unenriched bacterial poly(A) RNA. Indeed, the output from rRNA-depleted- and Poly(A)-RNA-Seq datasets indicate that (i) a sufficient number of sponge cells are captured using this bacterial cell enrichment method to generate highly representative RNA-Seq libraries and (ii) there is no substantial loss of host transcript abundance or representation in the rRNA-depleted-RNA-Seq dataset. In contrast, all three bacterial symbiont transcriptomes are significantly better covered in the rRNA-depleted-RNA-Seq data. To maintain the transcriptomic integrity and minimize unwanted transcriptomic changes (Westermann et al. 2017), the biopsy should be kept at low temperature during these bacterial enrichment steps (Thomas et al. 2010, Westermann et al. 2017), and RNA quantitive and qualitative assessments are necessary prior to library preparation.

Among the challenges for generating quality rRNA-depleted-RNA-Seq data from host and symbionts simultaneously, an important one is ensuring complete (or near-complete) removal of rRNA in the depletion step. Ribosomal RNA is the predominant (>80%) RNA in both the bacterial and eukaryotic cells (Westermann et al. 2012). Many commercial ribosomal RNA depletion kits exist, and they have different efficiencies (Petrova et al. 2017). The Ribo-Zero Gold rRNA Removal Kit (Epidemiology) (Illumina) used in this experiment yielded a high degree of rRNA depletion (Petrova et al. 2017), and is known to remove more than 99% rRNA in some samples, but was discontinued in November 2018. The new riboPOOL kit (siTOOLs) offers optimally-designed biotinylated DNA probes to robustly remove the rRNA of diverse species (eukaryotes and prokaryotes), which might be a ideal replacement for the Ribo-Zero Gold rRNA Removal Kit. Its rRNA removal efficiency varied in different samples (siTOOLs , Kim et al. 2019), and trials using the A. queenslandica holobiont have showed much promise (Poli et al. In preparation). An alternative option to overcome the incomplete removal of rRNA is to increase the overall

50 sequencing depth. The remaining rRNA reads can be filtered in silico (Kopylova et al. 2012, Schmieder et al. 2012). Higher sequencing depth would increase the sequencing cost. Meanwhile, the ribosomal depletion step is relatively expensive. The poly(A)-RNA capture step can be done with the mRNA library prep kit itself, while another rRNA removal kit is required for rRNA- depleted-RNA-Seq.

3.6 Conclusion

In this chapter, I compared the sponge holobiont transcriptome completeness between poly(A)- RNA-Seq and rRNA-depleted-RNA-Seq. I find that an rRNA-depleted-RNA-Seq strategy can capture a more complete and reliable animal-bacterial transcriptome than the poly(A)-RNA-Seq. Elimination of the higher cost, bacterial enriched rRNA-depleted-RNA-Seq is an optimal choice to study sponge host-symbiont interactions.

Chapter 4 Metabolic crosstalk between Amphimedon queenslandica and its primary bacterial symbionts

4.1 Abstract

Research on sponges has unveiled the complex diversity and function of their symbiotic microbes, but for most species the metabolic crosstalk within sponge-microbe symbiosis is not yet well characterised. In this chapter, I use genomic and transcriptomic analysis to predict the metabolic capacities of the sponge host A. queenslandica and its three primary symbiotic bacteria, AqS1, AqS2, and AqS3, and identify potential metabolic crosstalk between host and symbionts. Glycolysis, Krebs cycle, pentose phosphate pathway and fatty acid biosynthesis pathways are complete and expressed in both sponge and symbionts, which indicates their carbon heterotrophic metabolisms. The symbionts AqS1 and AqS2 may import and use hexoses liberated by host polysaccharide breakdown, indicating the potential of host-symbiont cooperation in assimilating carbon. Host- symbiont collaboration also may take place in assimilating dissolved inorganic nitrogen (DIN), where nitrate can be reduced to nitrite by AqS1, which can be subsequently reduced to ammonia by A. queenslandica, and assimilated by the host sponge and the three symbionts. Both the sponge and the three primary symbionts can assimilate sulfide, which is limited in water column but can be generated by sulfate reduction of AqS1. AqS1 and AqS3 can accumulate phosphate with the regulation through Pho regulon. These pathways suggest the host sponge and symbiotic bacteria collaborate together to assimilate the major marine nutrients, carbon, nitrogen, sulfur and phosphorus. Different suites of sponge and symbiotic bacterial genes involved in amino acid biosynthesis and transport are present and expressed, suggesting amino acid complementation may happen between the host sponge and the three primary symbionts. Several vitamin (B1, B2 and B5) and cofactor (folate, siroheme) biosynthesis pathways are present only in the symbionts, but not in the host, suggesting that the symbiotic bacteria may mediate the sponge nutrient assimilation and development via the supply of B vitamins and cofactors. This chapter reveals the widespread potential of host and symbionts to work together to assimilate marine nutrients and identify other potential metabolic complementations. The genomic and transcriptomic data of the host sponge fine tune these potential host-symbiont metabolic crosstalk in sponge life cycle and cells, advancing our understanding about the critical roles of symbionts in sponge development.

4.2 Introduction

Marine sponges are crucial engineers within aquatic ecosystems due to their capacity to capture, retain and recycle nutrients in oligotrophic conditions (de Goeij et al. 2013, Folkers and Rombouts 2020). With their impressive seawater filtration capacities (Bell 2008, Leys et al. 2011) and high cell proliferation rates (Ayling 1983, Alexander et al. 2015), sponges can rapidly obtain both particulate (bacteria and plankton) (Kahn et al. 2015, Leys et al. 2017, Kahn et al. 2018, Kazanidis et al. 2018) and dissolved food sources (Rix et al. 2016, Rix et al. 2017, Rix et al. 2020) from seawater, and covert these nutrients to particulate organic matters (POM) by cell shedding (de Goeij et al. 2013, Alexander et al. 2014, Maldonado 2015), which makes the nutrients available to higher trophic level reef inhabitants (Vanwonterghem and Webster 2020). Thus, sponges carry out a critical benthic-pelagic coupling function for marine ecosystems by retaining and recycling pelagic nutrients toward other benthos (Bell 2008, Folkers and Rombouts 2020).

Sponges often contain strikingly dense and complex microbial communities (Thomas et al. 2016, Webster and Thomas 2016, O’Brien et al. 2020, Steinert et al. 2020). Both the host sponge and sponge-associated microorganisms can participate in the nutrient cycling (e.g. carbon, nitrogen and sulfur) in the marine environment (de Goeij et al. 2008, van Duyl et al. 2008, Rix et al. 2017, Achlatis et al. 2019, Gantt et al. 2019). For example, recent isotope tracing experiments have reported the microbial symbionts of a high-microbial abundance (HMA) sponge accounts for the majority (63-87%) of their assimilation of dissolved organic matter (DOM) (Rix et al. 2020), which is the most abundant source of heterotrophic food in the oceans (Zhang et al. 2018).

Functional metabolic potentials of the sponge-associated microorganisms have been reported using omics approaches. Metagenomic and metatranscriptomic analyses have reported the capacity of some sponge symbionts to degrade carbohydrate via glycolysis, tricarboxylic acid (TCA) pathway and pentose phosphate pathways (Kamke et al. 2013, Gauthier et al. 2016, Moitinho-Silva et al. 2017, Bayer et al. 2018). Autotrophic carbon fixation pathways, such as reductive citric acid (rTCA) cycle and Calvin-Benson cycle, are also identified in the genomes of some sponge symbionts (Li et al. 2015, Engelberts et al. 2020), suggesting their capacity to synthesise organic matter using CO2. These pathways have revealed the functional potential of some symbionts in mediating the carbon cycling in marine sponges (Zhang et al. 2019). Apart from carbon cycling, the sponge-associated microbes are involved in nitrogen cycling on coral reefs (Fiore et al. 2015, Li et al. 2016, Moitinho-Silva et al. 2017, Weigel and Erwin 2017, Rix et al. 2020). Genes involved in nitrogen fixation, assimilation, ammonia oxidisation and denitrification are present in the metagenome of the deep-sea sponge Neamphius huxleyi (Li et al. 2015). The symbionts of sponge

Hymeniacidon heliophile are capable of ammonia assimilation, oxidation, assimilatory nitrate reduction, dissimilatory nitrate reduction and denitrification (Weigel and Erwin 2017). Nitrogen transformation pathways are also present in many other sponges microbes, such as the symbionts of sponge Xestospongia muta (Fiore et al. 2015), Cymbastela concentrica (Moitinho-Silva et al. 2017), and Ircinia ramosa (Engelberts et al. 2020). All these carbon and nitrogen transformation capacities of the sponge symbionts indicate their potential nutrient cycling roles within the holobiont.

Meanwhile, mounting studies report that sponge symbionts could biosynthesise some amino acids, B vitamins and cofactors, which are essential for the host sponge metabolism (Fiore et al. 2015, Gauthier et al. 2016, Engelberts et al. 2020). For example, the symbionts of Xestospongia muta can release amino acids (lysine, histidine, and tryptophan) and B vitamins (riboflavin, biotin, thiamine, cobalamin), which may be imported and utilised by the sponge (Fiore et al. 2015). Biosynthetic pathways for most amino acids and two cofactors (thiamine and riboflavin) are present in the symbiont genomes of the sponge Aplysina aerophoba (Bayer et al. 2018). The symbionts of the sponge Theonella swinhoei Y can produce most amino acids and rare cofactors like coenzyme F420 (Lackner et al. 2017). Many of these potentially symbiont-derived amino acids appear to be essential for the sponge host (Munroe et al. 2019). Thus, the symbionts may augment the sponge central metabolism through synthesis of these vital vitamins and cofactors (Bayer et al. 2018).

Previous sponge holobiont studies have begun to shed light on symbiont metabolic potential, however, they are predominantly from the aspect of the sponge-associated microorganisms (Fiore et al. 2015, Gauthier et al. 2016, Moitinho-Silva et al. 2017, Engelberts et al. 2020) and many detailed metabolic pathways of sponge-microbe interactions remain to be determined. Understanding the metabolic capabilities of both a sponge host and its symbionts is helpful in investigating the cooperation between the sponge and sponge-associated microorganisms in mediating nutrient assimilation. In this chapter, I investigate the metabolic capabilities of the sponge A. queenslandica and its three dominant symbionts (AqS1, AqS2, and AqS3) simultaneously. Using combined hologenomic and transcriptomic data, I identify active metabolic pathways expressed in A. queenslandica and its symbionts to test for potential metabolic complementation in this holobiont, including the crosstalk between host and symbionts, and among symbionts.

4.3 Materials and Methods

4.3.1 Functional annotation of holobiont gene models

Functional annotation is a crucial step to associate biological functions to genomic sequences.

Holobiont gene models from Chapter 2 were assigned to GO terms and InterPro IDs using Blast2GO (version 5.2.4) with the default Blast2GO 3-step – Blast, mapping and annotation – Gene Ontology annotation workflow (Conesa and Gotz 2008). Blastp-fast was used to search the non- redundant protein sequences (nr) database with e-value 1.0e-3. Gene orthology was determined using Kyoto Encyclopedia of Genes and Genomes (KEGG) (KEGG Orthology; KO) annotated by GhostKOALA searching against the genus_prokaryotes and family_eukaryotes KEGG GENES database (Kanehisa et al. 2016).

4.3.2 Assigning KEGG modules to host and symbionts

To identify the potential interactions among the host sponge and three primary bacterial symbionts, KEGG pathways of the holobiont and KEGG modules of A. queenslandica were reconstructed by KEGG Mapper (Kanehisa et al. 2017), based on the holobiont KEGG search results. The KEGG modules that were complete in the holobiont but incomplete in A. queenslandica were considered to be symbiotic bacterial modules. These modules were further manually confirmed whether they were complete in the symbionts, and the complete prokaryotic modules were discarded, as the aim of this analysis was to determine potential host-symbiont interactions.

4.3.3 Biological material collection and symbiotic bacteria enrichment

To obtain gene expression profiling of the A. queenslandica holobiont, six adult A. queenslandica were collected in December 2018 from Heron Island Reef, Great Barrier Reef, Queensland, Australia (Latitude 23.44, Longitude 151.92) as described previously (Leys et al. 2008) and maintained at Heron Island Research Station. The sponges were randomly placed in 12 L aquaria subjected to the same ambient flow-through seawater from the reef flat where they were collected. After a 24 h adaptation period, a 3 cm3 tissue biopsy was taken from each sponge and immediately placed in calcium- and magnesium-free artificial saltwater (CMF-ARS) (Schippers et al. 2011) on ice and kept at 4˚C during the following procedure. To increase the proportion of bacterial cells and ensure adequate representation of symbiont transcripts in the RNASeq data, a series of centrifugation and filtration steps were performed following the protocol of Thomas et al. (Thomas et al. 2010). The enriched microbial pellets were frozen in liquid nitrogen and stored at -80˚C until analysis. Total RNA was extracted using a standard TRIzol - phenol-chloroform method (Sigma) according to the manufacturer's protocol. RNA quantity and quality were assessed using Qubit® RNA assay kit on the Qubit® fluorometer (Invitrogen-Life Technologies) and mRNA Pico Series II assay on the Agilent 2100 Bioanalyzer (Agilent Technologies), respectively.

4.3.4 RNA-Seq library preparation and sequencing

RNA-Seq libraries were generated using an Illumina TruSeq Stranded mRNA Library Prep kit and IDT for Illumina TruSeq RNA UD Indexes, according to the manufacturer's protocol with some modifications described as follows. First, rRNA depletion was performed using specifically designed 3'-biotinylated DNA oligos riboPOOL kit (siTOOLs Biotech) following the manufacturer's protocol. To remove the majority of small RNA fragments, rRNA-depleted RNA was purified with AxyPrep MAG PCR Clean-up beads (Axygen). The rRNA-depleted and purified RNA samples were then used to prepare RNA-Seq libraries with the Illumina TruSeq Stranded mRNA Library Prep kit, omitting the poly-A RNA pulldown step and starting with the RNA fragmentation step. These RNA fragments were synthesised into cDNA with random primers and SuperScript II Reverse Transcriptase (Invitrogen). For stranded RNA sequence, cDNAs were converted into dsDNA in the presence of dUTP. Following 3' adenylation and adaptor ligation, 15 cycles PCR were performed to create the final cDNA libraries. These libraries were quantified on Agilent BioAnalyzer 2100 with the High Sensitivity DNA Kit (Agilent). Library preparation was performed at the University of Queensland Institute for Molecular Bioscience Sequencing Facility. Individual libraries were sequenced on the Illumina NovaSeq 6000 platform in one lane at the Australian Genome Research Facility (AGRF).

4.3.5 Transcriptome data from previous experiments

To extend the A. queenslandica holobiont transcriptome profile, the RNA-Seq reads generated by the rRNA-depleted RNA-Seq method in Chapter 3 (three samples) were used to estimate the holobiont gene expression and potential host-symbionts interactions. These rRNA-depleted RNA- Seq data are herein named 'field holobiont' transcriptome, to distinguish them from the new generated 'aquaria holobiont' transcriptome data. To better decipher how these host-symbionts interactions varied at different developmental stages and cell types, the sponge Cel-Seq2 transcriptomes of 82 samples from 17 different developmental tissue stages from cleavage embryos to adults (referred to as 'developmental' transcriptome) (Anavy et al. 2014, Levin et al. 2016) and 31 samples from 3 cell types (archaeocytes, choanocytes and pinacocytes, referred to as 'cell type' transcriptome) (Sogabe et al. 2019) were also used in this study.

4.3.6 Read processing and alignment

The raw paired-end reads of the field holobiont transcriptome were processed by Trimmomatic (version 0.36) (Bolger et al. 2014) to crop the first 10 bp of each read, and to trim the reads using a

4 bp sliding window and an average quality threshold of 20. Both unpaired reads and resulting reads smaller than 60 bp were discarded. The raw single end transcript reads of six aquaria A. queenslandica adults from flow-through aquaria were also filtered by Trimmomatic (parameter: SLIDINGWINDOW:4:20 MINLEN:20 LEADING:20 TRAILING:20) (version 0.36) (Bolger et al. 2014). The remaining high-quality transcript reads were aligned to the improved genome assembles of the A. queenslandica, AqS1, AqS2, and AqS3 (Chapter 2) using HISAT2 (version 2.0.5) with default parameters (Pertea et al. 2016). Both unaligned reads and reads that aligned to more than one of these four genomes were discarded because their source was ambiguous. The read mapping bam file of each sample was split into four bam files based on the scaffold ID of each species by SAMtools (version 1.3) (Li et al. 2009). The number of reads that aligned to sponge gene exon or bacterial protein-coding sequences was counted by htseq-count (version 0.11.2) from HTSeq framework with default parameters except for stranded=reverse (Anders et al. 2015); the latter is important to account for the strand-specific RNA libraries prepared with dUTP (Shin et al. 2014).

The A. queenslandica developmental transcriptome data (Anavy et al. 2014) and cell type transcriptome data (Sogabe et al. 2019) were also used to further study how potential sponge- bacterial crosstalk pathways varied in different A. queenslandica developmental stages and cells. Raw transcriptome reads were processed by Trimmomatic (parameter: SLIDINGWINDOW:4:20 MINLEN:20 LEADING:20 TRAILING:20) (version 0.36) (Bolger et al. 2014). The remaining high-quality reads were then aligned to improved genome assemblies of the A. queenslandica hologenome using HISAT2 (version 2.0.5) with default parameters (Pertea et al. 2016). Reads counts aligned to the gene exon regions were calculated by htseq-count (version 0.11.2) from the HTSeq framework with default parameters (Anders et al. 2015).

4.3.7 Classification of holobiont gene expression levels into quartiles

Relative expression levels of the holobiont genes were defined by the expression quartiles based on the number of reads aligned to each species. Both the aquaria and field holobiont transcriptomic reads were used to estimate holobiont gene expression. For the host A. queenslandica and the three primary symbionts AqS1, AqS2, and AqS3, transcripts per million (TPM) normalisation (Li et al. 2010) was used to measure the gene expression levels for the nine enriched bacterial holobiont samples (six aquaria samples and three field samples). The mean TPM normalised gene count values of the nine samples were used to calculate quartile values. If the gene was not aligned by any reads in all the nine samples, the gene was discarded. If the gene was aligned by at least one read in any of the nine samples, the gene was considered to be expressed and thus classified into four groups based on the normalised read count quartile values (Table 4-1), ranging from lowest (Q1) to highest (Q4). The sponge gene expression levels of the developmental samples, and cell type samples were also classified into quartiles with the same method, except that these gene counts were normalised with R package DESeq2 (Love et al. 2014).

4.3.8 Metabolic reconstruction of KEGG pathways

To predict the active functional pathways in each species, expressed genes that had a KEGG annotation were used to reconstruct KEGG pathways by KEGG mapper (Kanehisa and Sato 2020). As the symbiotic bacterial genomes were assembled from metagenomic reads, their genome assemblies are not quite complete (95.1%, 69.2%, and 89.6% complete BUSCO in AqS1, AqS2, and AqS3, respectively). To define an active metabolic function in the symbiont, the corresponding KEGG module was required to be > 60% expressed, and all key enzymes (as defined by the literature) needed to be detected in the holobiont transcriptome data. These criteria matched those used in another recent sponge microbiome study (Engelberts et al. 2020). Many enzymes in A. queenslandica and the three primary symbionts can be encoded by several genes because they were annotated with the same biological function. In each KEGG pathway, the gene copy that was expressed at the highest levels in all the samples in each dataset was used to indicate the expression of this functional gene.

Table 4-1. TPM normalised read count quartile values of transcriptome data for A. queenslandica (Aq) and the three primary symbionts (AqS1, AqS2 and AqS3). The gene is classified into Q1 if its expression level is smaller than Quartile 1 value in each species. If its expression level is between the Quartile 1 value and Quartile 2 value, the gene is expressed at Q2. If its expression level is between the Quartile 2 value and Quartile 3 value, the gene is expressed at Q3. If its expression level is larger than Quartile 3 value, the gene is expressed at Q4.

Quartile Aq AqS1 AqS2 AqS3 Quartile 1 0.29 31.67 104.71 49.17 Quartile 2 2.22 93.07 182.30 132.80 Quartile 3 10.21 227.87 398.49 352.19

4.4 Results

4.4.1 Holobiont transcriptome data

To evaluate the expression of the A. queenslandica holobiont genes, field and aquaria A. queenslandica transcriptome data were analysed. For the field RNA samples, 388 million Illumina short paired-end reads (average read length 75 bp) were generated in total, and 336 million reads (average read length 65 bp) remained after filtering and trimming low-quality reads via Trimmomatic (Bolger et al. 2014). These high-quality reads were aligned to the improved genome assembly sequences of the sponge A. queenslandica and the three symbionts, AqS1, AqS2, and AqS3 (Chapter 2). On average, 66% of the reads aligned to one of the four species; specifically, 50.8%, 12.7%, 1.7%, and 0.8% of the total reads aligned to the genome of A. queenslandica, AqS1, AqS2, and AqS3, respectively. A total of 3.3% aligned to more than one species and were discarded. For the six aquaria adult RNA samples, 270 million Illumina short single-end reads were generated, and 269 million high-quality reads were obtained after the Trimmomatic (Bolger et al. 2014) filter and trim steps. On average, 49.6%, 19.1%, 2.9%, and 0.7% of the reads aligned to the genome sequences of A. queenslandica, AqS1, AqS2, and AqS3, respectively. A total of 4.4% of the reads aligned to more than one species and were discarded.

To better decipher how potential interactions between the host-symbionts varied in different developmental stages and cell types, the sponge developmental transcriptome data (from cleavage embryos to adults) (Anavy et al. 2014) and cell type transcriptome data (from archaeocytes, choanocytes, pinacocytes) (Sogabe et al. 2019) were also analysed. For the developmental transcriptome data, there were 262 million reads in total, and 242 million reads survived from the quality control step, and 160 million reads were aligned to the A. queenslandica genome sequences. For the cell type transcriptome data, there were 356 million reads in total, and 319 million reads

59 were survived from the quality control step, and 198 million reads were aligned to the A. queenslandica genome sequences.

4.4.2 Gene expression quartile analyses

Gene depth was determined by the number of reads that aligned per gene, which was used to describe the relative gene expression level. The read numbers aligned to the holobiont genes were counted by htseq-count (version 0.11.2.) from the HTSeq framework with default parameters (Anders et al. 2015). For the holobiont transcriptome data, 37,606 (88.19% of the total gene number) A. queenslandica genes, 3,295 (94.74%) AqS1 genes, 1,604 (98.95%) AqS2 genes, and 2,787 (95.02%) AqS3 genes were expressed (aligned by at least one read) in at least one of the three field samples or six aquaria samples (Table 4-2). The expression levels of these expressed genes were classified into four groups based on the quartile values (Table 4-1) of average TPM normalised read counts of the nine samples (referred as adult holobiont transcriptome data from now on). To determine active pathways in each species, the expression levels of genes assigned to specific KEGG pathways were analysed.

Table 4-2. The number of genes expressed in each quartile of the holobiont transcriptome for A. queenslandica (Aq) and the three dominant symbionts (AqS1, AqS2 and AqS3).

Quartile Aq AqS1 AqS2 AqS3 Not expressed 5038 183 17 146 Q1 9402 824 401 697 Q2 9401 823 401 696 Q3 9401 824 401 697 Q4 9402 824 401 697

4.4.3 Symbiotic bacterial contribution to the holobiont metabolism

Analysis of KEGG modules and pathways using A. queenslandica hologenome and transcriptome data reveals 28 KEGG modules, which belong to 19 KEGG pathways (Table 4-3), is incomplete in A. queenslandica, but are complete in the holobiont with the contribution from the symbiotic bacteria. These bacterial supported modules function in the metabolism of amino acids (14 modules), cofactors and vitamins (6 modules), carbohydrates (2 modules), energy (2 modules) and lipids (3 modules), suggesting that the symbionts support the holobiont metabolism in multiple ways. Given the widespread nature of holobiont metabolism, the detailed metabolic capacities of A.

60 queenslandica, AqS1, AqS2 and AqS3 and their potential to metabolically interact were further explored, with a focus on carbohydrate metabolism, and the biosynthesis of amino acids, fatty acids, and some cofactors and vitamins.

Table 4-3. Complete KEGG modules of A. queenslandica holobiont contributed by the symbiotic bacteria

Biological_process Biological_pathway KEGG_module Amino acid metabolism 00220 Arginine biosynthesis M00028 Ornithine biosynthesis glutamate => ornithine Amino acid metabolism 00220 Arginine biosynthesis M00844 Arginine biosynthesis ornithine => arginine 00260 Glycine serine and Amino acid metabolism M00018 Threonine biosynthesis aspartate => homoserine => threonine threonine metabolism 00260 Glycine serine and Amino acid metabolism M00555 Betaine biosynthesis choline => betaine threonine metabolism 00270 Cysteine and methionine Amino acid metabolism M00017 Methionine biosynthesis apartate => homoserine => methionine metabolism 00270 Cysteine and methionine Amino acid metabolism M00021 Cysteine biosynthesis serine => cysteine metabolism 00290 Valine leucine and M00019 Valine/isoleucine biosynthesis pyruvate => valine / 2-oxobutanoate => Amino acid metabolism isoleucine biosynthesis isoleucine 00290 Valine leucine and Amino acid metabolism M00432 Leucine biosynthesis 2-oxoisovalerate => 2-oxoisocaproate isoleucine biosynthesis 00290 Valine leucine and Amino acid metabolism M00570 Isoleucine biosynthesis threonine => 2-oxobutanoate => isoleucine isoleucine biosynthesis Amino acid metabolism 00300 Lysine biosynthesis M00016 Lysine biosynthesis succinyl-DAP pathway aspartate => lysine Amino acid metabolism 00340 Histidine metabolism M00026 Histidine biosynthesis PRPP => histidine 00400 Phenylalanine tyrosine and Amino acid metabolism M00022 Shikimate pathway phosphoenolpyruvate + erythrose-4P => chorismate tryptophan biosynthesis 00400 Phenylalanine tyrosine and Amino acid metabolism M00023 Tryptophan biosynthesis chorismate => tryptophan tryptophan biosynthesis 00400 Phenylalanine tyrosine and Amino acid metabolism M00025 Tyrosine biosynthesis chorismate => tyrosine tryptophan biosynthesis

The Table is continued on the next page

Biological_process Biological_pathway KEGG_module 00030 Pentose phosphate M00008 Entner-Doudoroff pathway glucose-6P => glyceraldehyde-3P + Carbohydrate metabolism pathway pyruvate 00630 Glyoxylate and Carbohydrate metabolism M00012 Glyoxylate cycle dicarboxylate metabolism Energy metabolism 00920 Sulfur metabolism M00595 Thiosulfate oxidation by SOX complex thiosulfate => sulfate Energy metabolism 00920 Sulfur metabolism M00596 Dissimilatory sulfate reduction sulfate => H2S Lipid metabolism 00061 Fatty acid biosynthesis M00082 Fatty acid biosynthesis initiation Lipid metabolism 00061 Fatty acid biosynthesis M00083 Fatty acid biosynthesis elongation 00564 Glycerophospholipid Lipid metabolism M00093 Phosphatidylethanolamine (PE) biosynthesis PA => PS => PE metabolism Metabolism of cofactors and 00730 Thiamine metabolism M00127 Thiamine biosynthesis AIR => thiamine-P/thiamine-2P vitamins Metabolism of cofactors and 00740 Riboflavin metabolism M00125 Riboflavin biosynthesis GTP => riboflavin/FMN/FAD vitamins Metabolism of cofactors and 00760 Nicotinate and M00115 NAD biosynthesis aspartate => NAD vitamins nicotinamide metabolism Metabolism of cofactors and 00770 Pantothenate and CoA M00019 Valine/isoleucine biosynthesis pyruvate => valine / 2-oxobutanoate vitamins biosynthesis => isoleucine Metabolism of cofactors and 00790 Folate biosynthesis M00126 Tetrahydrofolate biosynthesis GTP => THF vitamins Metabolism of cofactors and 00860 Porphyrin and chlorophyll M00846 Siroheme biosynthesis glutamate => siroheme vitamins metabolism Nucleotide metabolism 00230 Purine metabolism M00048 Inosine monophosphate biosynthesis PRPP + glutamine => IMP

4.4.4 Carbohydrate metabolism in the A. queenslandica holobiont

Aerobic cellular respiration

The gene repertoire of A. queenslandica and its three primary symbionts, AqS1, AqS2, and AqS3, reveals their potential aerobic cellular respiration. Aerobic cellular respiration is an important metabolic process that converts a food source into biochemical energy; it includes the conversion of glucose and other carbohydrates into ATP through glycolysis, pyruvate oxidation, Krebs cycle and oxidative phosphorylation (Bear 2019). Complete or near- complete glycolysis, pyruvate oxidation, Krebs cycle and oxidative phosphorylation pathways are present in the A. queenslandica holobiont, indicating heterotrophic metabolism of the host and of the three primary symbionts.

A. queenslandica and AqS1 have complete glycolysis pathways to oxidise hexose sugars and produce energy, while near-complete pathways are present in AqS2 and AqS3 (Figure 4-1). Hexokinase, which can initiate the metabolism of glucose by phosphorylation (Wilson 2003), is present in A. queenslandica but not in the three primary symbionts. However, the ppgK gene, encoding polyphosphate glucokinase (EC 2.7.1.63), is present in the genome of AqS1. This enzyme also catalyses the phosphorylation of glucose (Hsieh et al. 1996). The other nine enzymes involved in glycolysis are present in all four species, except for phosphoglucose isomerase (EC 5.3.1.9) and pyruvate kinase (EC 2.7.1.40), both of which are absent in AqS3. All of these host and symbiont genes are expressed in the adult holotranscriptome (Figure 4-1).

In A. queenslandica, most of the genes encoding the ten enzymes involved in glycolysis are highly expressed (Q3 or Q4) across the life cycle and in the three cell types (Figure 4-2). Interestingly, the A. queenslandica gene encoding the first enzyme of the glycolysis pathway, namely hexokinase (EC 2.7.1.1), is expressed highest at the oscula (filter-feeding juvenile) stage and tend to be expressed higher in late metamorphosis and adult stages than in embryonic and larval stages (Figure 4-2). It is also highly expressed (Q4) in the three adult cell types with the highest expression level in archaeocytes (Figure 4-2). ADP-dependent glucokinase (ADPGK, EC 2.7.1.147) can also phosphorylate glucose to glucose-6-phosphate. Two copies ADPGK are present in A. queenslandica and their expression levels tend to higher during the 6-24 hr post settlement than the other stages (Figure 4-2). For adult A.

64 queenslandica, one ADPGK is expressed highest in archaeocytes and another one expressed highest in pinacocyte (Figure 4-2). These genes gene levels indicate a higher carbon assimilation capacity in metamorphosis and adult stages.

Pyruvate produced by the glycolysis step can be further oxidised into carbon dioxide (CO2) and acetyl-CoA by the A. queenslandica holobiont. The three enzymes – pyruvate dehydrogenase (EC 1.2.4.1) (E1), lipoate acetyltransferase (EC 2.3.1.12) (E2) and dihydrolipoyl dehydrogenase (EC 1.8.1.4) (E3) – that form the pyruvate dehydrogenase complex (Wieland 1982), are present in all four genomes (Figure 4-3). This complex can convert pyruvate to acetyl-CoA with the help of thiamine pyrophosphate (TPP), lipoate coenzyme A, FAD and NAD+ (Wieland 1982). The pyruvate to acetyl-CoA transformation can also be catalysed by 2-oxoacid:ferredoxin oxidoreductases (EC 1.2.7.11, 1.2.7.3) (Gibson et al. 2015). The alpha and beta subunits of 2-oxoacid:ferredoxin oxidoreductase are present only in AqS1 and AqS3. The E3 component of the pyruvate dehydrogenase complex is also part of the oxoglutarate dehydrogenase complex (Sanderson et al. 1996), which is involved in the Krebs cycle. Similar to pyruvate oxidation, both oxoglutarate dehydrogenase complex and 2-oxoacid:ferredoxin oxidoreductase (EC 1.2.7.11, 1.2.7.3) can convert 2-oxoglutarate to succinyl-CoA. The former complex is present in all four species, but the latter enzyme is found only in AqS1 and AqS3 (Figure 4-3). The other seven enzymes involved in the Krebs cycle are present in all the four holobiont members.

All genes encoding the enzymes involved in pyruvate oxidation and the Krebs cycle are expressed in the transcriptomes of the adult host and three symbionts (Figure 4-3). Higher expression levels of the gene encoding 2-oxoacid:ferredoxin oxidoreductase compared to pyruvate/oxoglutarate dehydrogenase complex in AqS3 suggest the potential of higher reliance on the 2-oxoacid:ferredoxin oxidoreductase for pyruvate oxidation and the Krebs cycle (Figure 4-3). In the developmental and cell-type transcriptome, all host genes involved in pyruvate oxidation and Krebs cycle are highly expressed, except for the gene encoding 2- oxoglutarate dehydrogenase E2 component (EC 2.3.1.61), which was expressed at Q3 in most developmental stages and cell types. Overall, the host and symbionts all can independently undertake pyruvate oxidation and Krebs cycle, which are pivotal metabolic processes that link carbohydrate, lipid and amino acid metabolism (Bender and Mayes 2016).

Starch/Glycogen Glucose

2.7.1.1 11 2.4.1.1 1 2.7.1.63 5.4.2.2 Glucose-1-phosphate Glucose-6-phosphate Aq AqS1 AqS2 AqS3 12 2 5.3.1.9 15 2.7.1.4 Glucose Fructose Fructose-6-phosphate ENZYMES 1 2.7.1.1 3 14 3.2.1.26 2.7.1.11 1 Hexokinase 13 3.2.1.20 Fructose-1,6-phosphate 2 Phosphoglucose isomerase Sucrose 4 4.1.2.13 3 Phosphofructokinase 5.3.1.1 Dihydroxyacetone Glyceraldehyde 4 Aldolase phosphate 3-phosphate 5 5 Triosephosphate isomerase 6 1.2.1.12 6 Glyceraldehyde 3-phosphate dehydrogenase 1,3-Bisphosphoglycerate 7 Phosphoglycerate kinase 7 2.7.2.3 8 Phosphoglyceromutase 3-Phosphoglycerate 9 Enolase 5.4.2.11 8 5.4.2.12 10 Pyruvate kinase 2-Phosphoglycerate 11 Glycogen phosphorylase

9 4.2.1.11 12 Phosphoglucomutase

Phosphoenolpyruvate 13 Alpha-glucosidase

10 2.7.1.40 14 Beta-fructofuranosidase

15 Fructokinase Pyruvate

Figure 4-1. Overview of the glycolysis pathway in the A. queenslandica holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub- block is filled with colour, the enzyme is present and expressed in the adult holobiont, otherwise, it’s absent in that species. There are no cases in which a gene is present but not expressed in this pathway.

Quartile Normalized

2.7.1.1_Aqu3.1.19198_001 Glycolysis 2.7.1.147_Aqu3.1.18690_001 2.7.1.147_Aqu3.1.38787_001 Pentose phosphate 5.3.1.9_Aqu3.1.39626_001 pathway 2.7.1.11_Aqu3.1.33393_001 4.1.2.13_Aqu3.1.07554_001 Stage 4.1.2.13_Aqu3.1.36501_001 Cleavage 1.2.1.12_Aqu3.1.29717_001 Early 2.7.2.3_Aqu3.1.41809_001 Mid Late 5.4.2.12_Aqu3.1.40636_001 Larva 4.2.1.11_Aqu3.1.21650_001 Metamorphosis Adult 4.2.1.11_Aqu3.1.39258_001 Quartile 4.2.1.11_Aqu3.1.43363_001 0 2.7.1.40_Aqu3.1.08244_001 1 2 2.7.1.40_Aqu3.1.08245_001 3 2.7.1.40_Aqu3.1.12422_001 4 2.7.1.40_Aqu3.1.34807_001 Developmental 2.7.1.40_Aqu3.1.43531_001 4 1.1.1.49_Aqu3.1.42602_001 2 3.1.1.31_Aqu3.1.38418_001 0 -2 1.1.1.47_Aqu3.1.20222_001 Cell_type 1.1.1.44_Aqu3.1.36291_001 1.5 5.1.3.1_Aqu3.1.37539_001 1 5.3.1.6_Aqu3.1.40626_001 0.5 0 2.2.1.1_Aqu3.1.40413_001 -0.5 2.2.1.2_Aqu3.1.32110_001 -1 2.7.6.1_Aqu3.1.41066_001 -1.5 wn wn ing ing age age r r o o pole pole Ring Ring Spot Spot Adult Adult Cloud Cloud av av Br Br Oscula Oscula ent- ent- 7_hr_PS 7_hr_PS 1_hr_PS 1_hr_PS Chamber Chamber Cle Cle Late_ Late_ T T Late_spot Late_spot 12_hr_PS 24_hr_PS 12_hr_PS 24_hr_PS 6- 6- Competent Competent Pinacocyte Pinacocyte Choanocyte Choanocyte Archaeocyte Archaeocyte 11- 23- 11- 23- Precompetent Precompetent

Figure 4-2. Expression levels of A. queenslandica genes involved in glycolysis and pentose phosphate pathways throughout development and in cell types. The left heatmap presents these gene expression quartiles, and the right one presents the TPM normalised gene expression levels.

Quartile 0 1 2 3 4

Figure 4-3. The expression levels of A. queenslandica holobiont genes involved in pyruvate oxidation and Krebs cycle in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

A. queenslandica, AqS1, AqS2 and AqS3 all have electron transport chains for aerobic respiration; however, the electron transport complex I of the four species is not the same. Specifically, A. queenslandica can produce mitochondrial complex I, while AqS1 and AqS2 possess a bacterial complex I, a type I NADH dehydrogenase to catalyse electron transfer from NADH to ubiquinone in the respiratory chain. The genes encoding the 14 Nuo subunits of NADH-quinone oxidoreductase are also present in AqS1 and AqS2, as in E. coli (Falk- Krzesinski and Wolfe 1998). AqS3 employs Na+-transporting NADH: ubiquinone oxidoreductase as a redox-driven ion pump, which is analogous to mitochondrial complex I (Verkhovsky and Bogachev 2010). The genes encoding the six subunits of this complex (Nakayama et al. 1998) all are present in the AqS3 genome. The four subunits of the

68 succinate dehydrogenase (complex II), cytochrome c reductase (complex III), cytochrome c oxidase (complex IV), and ATP synthase (complex V) are present in all four species, except that genes encoding cytochrome c reductase are absent from the AqS2 genome. Nearly all genes encoding the subunits of these electron transport complexes are expressed in all partners of the adult holobiont, with most of the host genes being expressed at Q3 or Q4 across the life cycle and cell types.

Pentose phosphate pathway

The pentose phosphate pathway (PPP) is another method to oxidise glucose (Dashty 2013) and a crucial step to produce NADPH and ribose-5-phosphate for the ultimate synthesis of nucleic acids. It comprises two branches: the irreversible oxidative and the reversible non- oxidative (Figure 4-4) (Kowalik et al. 2017). The genomic and transcriptomic information of the A. queenslandica holobiont reveals its capacity to complete the oxidative or the non- oxidative PPP, as follows. A complete PPP is present in the A. queenslandica host, with all genes being highly expressed (Q4) in the adult holobiont (Figure 4-4), throughout development and in the three sponge cell types (Q4 or Q3) except the gene encoding hexose- 6-phosphate dehydrogenase (EC 1.1.1.47) (Figure 4-2). Some of the sponge genes involved in the PPP, but not in the glycolysis, are expressed higher during the late metamorphosis and adult stages than the other developmental stages, especially the two key enzymes, transaldolase (EC 2.2.1.2) and transketolase (EC 2.2.1.1) in the non-oxidative PPP (Figure 4-2).

For the three bacterial symbionts, the oxidative PPP is complete and active only in AqS1. In contrast, the non-oxidative PPP same with the host is incomplete in all the three symbionts because the key enzyme transaldolase (EC 2.2.1.2) is absent. Studies in other bacteria have revealed an alternative route for the non-oxidative PPP, which relies on the sedoheptulose 1,7-bisphosphate (SBP) pathway (Figure 4-4) that uses pyrophosphate-dependent phosphofructokinase (PPi-PFK, EC 2.7.1.90) instead of transaldolase. PPi-PFK can convert sedoheptulose 7-phosphate (S7P) to SBP (Reshetnikov et al. 2008, Koendjbiharie et al. 2020) and the genes encoding PPi-PFK are expressed in the three symbionts in the adult holobiont. Collectively these results reveal an active oxidative PPP in A. queenslandica and AqS1, and a functional non-oxidative PPP in A. queenslandica (via a transaldolase) and the three symbionts (via a 6-phosphofructokinase) (Figure 4-4).

6-Phosphogluconic 1.1.1.44 acid Ribulose 5-phosphate 1.1.1.343 Aq AqS1 AqS2 AqS3 3.1.1.31 5.1.3.1 5.1.3.6

6-Phospho- Xylulose 5- Ribose 5- Oxidative phase gluconolactone phosphate phosphate Non-oxidative phase

1.1.1.49 2.2.1.1 1.1.1.363

beta-D-Glucose Glyceraldehyde Sedoheptulose 2.7.1.90 Sedoheptulose 1,7- 6-phosphate 3-phosphate 7-phosphate bisphosphate

4.1.2.13 5.3.1.9 2.2.1.2

alpha-D-Glucose 5.3.1.9 beta-D-Fructose Erythrose 4- Xylulose 5- Glycerone 6-phosphate 6-phosphate phosphate phosphate phosphate 2.7.1.90 2.2.1.1 5.3.1.1 3.1.3.11 beta-D-Fructose Glyceraldehyde 1,6-bisphosphate 4.1.2.13 3-phosphate

Figure 4-4. Pentose phosphate pathway (PPP) in the A. queenslandica holobiont. The oxidative PPP is complete and expressed in A. queenslandica and AqS1, and a functional non-oxidative PPP is present in A. queenslandica (via a transaldolase, EC 2.2.1.2) and the three symbionts (via a 6- phosphofructokinase, EC 2.7.1.90). The EC code is designated in each enzyme block. The four sub- blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is filled with colour, the enzyme is present and expressed in the adult holobiont, otherwise, it’s absent in that species. There are no cases in which a gene is present but not expressed in this pathway.

Carbohydrate complementation in the A. queenslandica holobiont

The A. queenslandica holobiont can utilise other carbohydrates, such as glycogen, starch, and sucrose, and the three symbionts appear to utilise the glucose liberated by the sponge from glycogen and starch. Two PYG genes encoding glycogen phosphorylase (PYG, EC 2.4.1.1), which can degrade glycogen and starch, and liberate glucose in the form of glucose-1- phosphate (Komoda and Matsunaga 2015), are present in the host sponge and highly expressed (Q4) in the adult holobiont (Figure 4-1). These two PYG genes are highly expressed across most of the life cycle with the higher expression levels at late embryonic, precompetent larval and adult stages (Figure 4-5). In the adult cells, one PYG is highly expressed (Q4) in the three cell types with the highest expression in pinacocyte; the other one is expressed at Q3 in the three cell types with the highest expression in archaeocyte (Figure

4-5). But no gene encoding PYG is present in the three primary symbionts. The sponge liberated glucose-1-phosphate can be converted to glucose-6-phosphate by phosphoglucomutase (EC 5.4.2.2) and proceed directly to glycolysis (Komoda and Matsunaga 2015). The genes encoding phosphoglucomutase (EC 5.4.2.2) are expressed in both host and symbionts (Figure 4-1).

Besides glycogen breakdown, the host sponge can hydrolyse sucrose into D-fructose and D- glucose by (lysosomal) alpha-glucosidase (EC 3.2.1.20). Two malZ, encding alpha- glucosidase, and one GAA, encoding lysosomal alpha-glucosidase, are present in the sponge and highly expressed in adult (Figure 4-1) and all three cell types (Q3 or Q4) (Figure 4-5). These genes are expressed across the life cycle, with higher levels (Q4) observed during metamorphosis and in the adult (Figure 4-5). The symbiont AqS2 can also hydrolyse sucrose by beta-fructofuranosidase (EC 3.2.1.26). This gene is moderately expressed (Q2) in the adult holobiont.

The products of sucrose hydrolysis by A. queenslandica and AqS2 can be imported and catabolized by AqS1. The genes frcBCA encoding the three subunits of the fructose transport system are present and expressed in AqS1, while two phosphotransferase proteins (PTS) – Enzyme I (EI) and PTS fructose transporter subunit IIA (IIA) (Zuniga et al. 2005, Deutscher et al. 2006) – are expressed in AqS2 (Q2 and Q3, respectively). Both FrcBCA (Lambert et al. 2001) and fructose PTS transporters (Postma et al. 1993) can perform fructose uptake. Thus, AqS1 and AqS2 can import fructose via the fructose transport system and phosphoenolpyruvate: sugar phosphotransferase system (PTS), respectively. The other intermediate of the sucrose hydrolysis – glucose – can be taken up through multiple sugar transport system ATP-binding proteins (EC 3.6.3.19, encoded by malK in AqS1 and AqS2) or glucose/mannose transport system permease proteins (in AqS2). All genes encoding these transporters are expressed in the adult holobiont.

Stage PYG.1 Cleavage Early PYG.2 Mid MalZ.1 Late MalZ.2 Larva Metamorphosis GAA tile Adult r PhoA Quartile SLC19A2_3 0 Qua 1 SLC35A3_4 2 SLC46A1 3 NIT-6 4

PYG.1 Developmental PYG.2 4 MalZ.1 2 0 MalZ.2 ed

z -2 GAA Cell_type PhoA 1.5 mali r SLC19A2_3 1 0.5 No SLC35A3_4 0 SLC46A1 -0.5 NIT-6 -1 -1.5 wn ing age r o pole Ring Spot Adult Cloud av Br Oscula ent- 7_hr_PS 1_hr_PS Chamber Cle Late_ T Late_spot 12_hr_PS 24_hr_PS 6- Competent Pinacocyte Choanocyte Archaeocyte 11- 23- Precompetent

Figure 4-5. Expression levels of A. queenslandica genes involved in some potential metabolic complementation between the host and symbionts throughout development and in cell types. The upper heatmap presents these gene expression quartiles, and the bottom one presents the TPM normalised gene expression levels. In the row names, the number after the point represents the number of gene copy, e.g. PYG.1 and PYG.2 are two copies of genes encoding glycogen phosphorylase (PYG, EC 2.4.1.1). MalZ, alpha-glucosidase (EC 3.2.1.20); GAA, lysosomal alpha- glucosidase (EC 3.2.1.20); PhoA, alkaline phosphatase (EC 3.1.3.1); SLC19A2_3 and SLC35F3_4, thiamine transporters; SLC46A1, proton-coupled folate transporter; NIT-6, nitrite reductase [NAD(P)H] (EC 1.7.1.4).

4.4.5 Amino acid biosynthesis in the A. queenslandica holobiont

The symbionts can produce essential amino acids for A. queenslandica. The sponge host has incomplete biosynthesis pathways for the eight essential amino acids (histidine, leucine, isoleucine, lysine, methionine, phenylalanine, valine, and tryptophan) that are required by humans (Reeds 2000). However, complete or near-complete biosynthesis pathways for all of these eight essential amino acids are present in AqS1, AqS2 and AqS3 (Figure 4-6).

Amino Acid Biosynthesis Aq AqS1 AqS2 AqS3 Histidine Isoleucine Leucine Lysine Methionine Phenylalanine Threonine Tryptophan Valine Alanine Arginine Asparagine Aspartic_acid Cysteine Glutamic_acid Glutamine Glycine Proline Serine Tyrosine

Figure 4-6. The completeness of amino acid biosynthesis in the A. queenslandica holobiont. The pie charts show the percentage of enzymes present for each amino acid synthesise pathway encoded by each species. The yellow and green backgrounds indicate the human essential and nonessential amino acids, respectively. Aq is A. queenslandica, AqS1, AqS2 and AqS3 are the three primary symbionts.

A complete histidine biosynthesis pathway is present in the three symbiotic bacteria, AqS1, AqS2 and AqS3, but not in A. queenslandica (Supplementary File 4-1). The symbionts can form histidine by a series of reactions from phosphoribosyl diphosphate (PRPP) (Ingle 2011), the product of the pentose phosphate pathway. These symbiont enzymes are all expressed in the adult holobiont, except that phosphoribosyl-AMP cyclohydrolase (EC 3.5.4.19), which

73 converts N′-5′-phosphoribosyl-AMP into (N′-[(5′-phosphoribosyl)formimino]-5- aminoimidazole-4-carboxamide) ribonucleotide, is not present in the AqS2 genome.

The branch-chain amino acid (valine, leucine and isoleucine) biosynthesis pathways are present and expressed in the three symbiotic bacteria (Supplementary File 4-2). The seven enzymes involved in these pathways – acetolactate synthase (EC 2.2.1.6), ketol-acid reductoisomerase (EC 1.1.1.86), dihydroxy-acid dehydratase (EC 4.2.1.9), branched-chain amino acid aminotransferase (EC 2.6.1.42), 2-isopropylmalate synthase (EC 2.3.3.13), 3- isopropylmalate dehydratase (EC 4.2.1.33), and 3-isopropylmalate dehydrogenase (EC 1.1.1.85) – are all present in AqS1, AqS2, and AqS3, and expressed in the adult holobiont. In contrast, only dihydroxy-acid dehydratase, branched-chain amino acid aminotransferase and 2-isopropylmalate synthase, are present in A. queenslandica. These enzymes also are involved in other metabolic pathways, such as branched-chain-amino-acid transaminase, and in cysteine and methionine metabolism pathway (Venos et al. 2004).

The A. queenslandica holobiont has similar capacity for the synthesis of aromatic amino acids as for branch-chain amino acids. The complete shikimate pathway, which starts with phosphoenol-pyruvate and erythrose-4-phosphate and ends with chorismite (Herrmann and Weaver 1999), is present in AqS1, AqS2 and AqS3, but not in A. queenslandica (Supplementary File 4-3). Chorismite is the substrate for the synthesis of phenylalanine, tyrosine and tryptophan, and all symbiont genes involved in the biosynthesis of these three aromatic amino acids are present and expressed in the adult holobiont, except that phosphoribosylanthranilate isomerase (EC 5.3.1.24) involved in tryptophan de novo synthesis pathway, is not present in the AqS2 genome. These aromatic amino acids de novo synthesis pathways are incomplete in A. queenslandica because only the aroA (3-phosphoshikimate 1- carboxyvinyltransferase, EC 2.5.1.19) and trpE (anthranilate synthase component I, EC 4.1.3.27) genes are present. The phhA gene, encoding phenylalanine-4-hydroxylase (EC 1.14.16.1) is present in A. queenslandica. This enzyme can produce tyrosine if phenylalanine is available from the environment or from the symbiotic bacteria. The sponge phhA is highly expressed in the adult holobiont, across the life cycle and in the three adult cell types

Two complete de novo threonine synthesis pathways, which initiate from L-aspartate and L- glycine, are present in the A. queenslandica holobiont (Figure 4-7; Supplementary File 4-4). The three bacterial symbionts can synthesise threonine from aspartate as genes encoding five enzymes involved in this process are all present, except the gene encoding homoserine kinase

74 type II (EC 2.7.1.39) is absent in AqS1 (Figure 4-7). All symbiotic bacterial genes involved in this threonine synthesis are expressed in the adult holobiont (Figure 4-7). Threonine can also be produced by threonine aldolase (EC 4.1.2.48) from glycine. The ltaE gene, encoding threonine aldolase (EC 4.1.2.48), is expressed in the adult holobiont by A. queenslandica and AqS1, and in A. queenslandica developmental stages and cell types. All genes involved in glycine synthesis from glycerate-3P (Supplementary File 4-4), an intermediate product of glycolysis pathway, are expressed by the sponge host and three symbionts, and in all the developmental stages and the three adult cell types. In summary, threonine and glycine are nonessential amino acids for both the host A. queenslandica and the three symbiotic bacteria.

The three bacterial symbionts can biosynthesise methionine from aspartate. The genes involved in this process are present in their genomes, except that metY, encoding O- acetylhomoserine (thiol)-lyase (EC 2.5.1.49), is absent in AqS2 (Figure 4-7; Supplementary File 4-5). These bacterial genes are expressed in the adult holobiont. Only the metH gene, encoding methionine synthase (EC 2.1.1.13) that is the last enzyme of methionine biosynthesis pathway and converts homocysteine to methionine, is present in A. queenslandica and expressed in the adult (Figure 4-7). For lysine biosynthesis (Figure 4-7), the three symbionts can use L-aspartate as a precursor, except 2,3,4,5-tetrahydropyridine-2,6- dicarboxylate N-succinyltransferase (EC 2.3.1.117) and diaminopimelate epimerase (EC 5.1.1.7) are missing in AqS2. These bacterial genes are all expressed in the adult holobiont. For A. queenslandica, only the genes encoding two enzymes in this lysine biosynthesis pathway, namely 4-hydroxy-tetrahydrodipicolinate synthase (EC 4.3.3.7) and diaminopimelate decarboxylase (EC 4.1.1.20) are present and expressed in the adult holobiont. Overall, the three symbionts are equipped with complete or near-complete lysine, and methionine biosynthesis pathways, while A. queenslandica can produce lysine, and methionine if some key intermediates are available.

L-4-Aspartyl 2.7.2.4 Aq AqS1 AqS2 AqS3 phosphate L-Aspartate

1.2.1.11

L-Aspartate 4- 1.1.1.3 2.3.1.31 O-Acetyl-L- semialdehyde L-Homoserine homoserine

2.7.1.39 4.3.3.7 2.5.1.48 2.5.1.49

(2S,4S)-4-Hydroxy-2,3,4,5- O-Phospho-L- tetrahydrodipicolinate homoserine L- Homocysteine

1.17.1.8 4.2.3.1 2.1.1.13 2.1.1.14 2,3,4,5- Tetrahydrodipicolinate L-Threonine L-Methionine

2.3.1.117 4.1.2.48

N-Succinyl-2-L-amino- 6-oxoheptanedioate L-Glycine L-Lysine

2.6.1.17 4.1.1.20

N-Succinyl-LL-2,6- 3.5.1.18 LL-2,6- 5.1.1.7 Meso-2,6- diaminoheptanedioate Diaminoheptanedioate Diaminopimelate

Figure 4-7. The threonine, lysine, and methionine biosynthesis pathway of the A. queenslandica holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is filled with colour, the enzyme is present and expressed in the adult holobiont, otherwise, it’s absent in that species. There are no cases in which a gene is present but not expressed in this pathway.

The arginine synthesis pathway is complete or near-complete in the three symbiotic bacteria, but incomplete in A. queenslandica. There are eight enzymatic steps in the arginine synthesis from glutamate (Caldara et al. 2008). The genes involved in this process are present in the genomes of the three symbiotic bacteria, except that argD, encoding acetylornithine aminotransferase (EC2.6.1.11), is absent from AqS1 (Supplementary File 4-6). These bacterial genes are all expressed in the adult holobiont. Only one of the eight enzymes – acetylornithine deacetylase (EC 3.5.1.16) – is present in A. queenslandica, and expressed in the adult holobiont.

Asparagine can be synthesised by A. queenslandica, but not by the three primary symbionts. Asparagine synthetase (EC 6.3.5.4) can reversibly convert aspartate to asparagine (Lomelino et al. 2017), which is encoded by asnB. Two copies of asnB are present in A. queenslandica (Supplementary File 4-7), with one copy being moderately expressed (Q2) in the adult and most developmental stages, and variably expressed in the adult cell types: Q3 in archaeocyte, and Q1 in choanocyte and pinacocyte. The other copy is lowly expressed (Q1) in the adult and choanocytes, and moderately (Q2) pinacocytes, but not expressed most of developmental. L-asparaginase (EC 3.5.1.1) can convert asparagine into aspartic acid and is present in A. queenslandica and AqS2. Alternatively, aspartic acid and 2-oxoglutarate can be released by aspartate aminotransferase (EC 2.6.1.1) from oxaloacetate and L-glutamate. This enzyme is in A. queenslandica, AqS1 and AqS2. These genes are expressed in the adult holobiont. Based on the genomic and transcriptomic information, aspartic acid is likely an essential amino acid for AqS3, and asparagine is essential for AqS1, AqS2 and AqS3; both amino acids could be produced by A. queenslandica.

All the enzymes that synthesise the other six amino acids (alanine, cysteine, serine, glutamate, glutamine, and proline) are encoded by the genes in A. queenslandica, AqS1, AqS2 and AqS3. L-alanine can be produced by alanine dehydrogenase (EC 1.4.1.1), or alanine transaminase (EC 2.6.1.2). The gene encoding alanine dehydrogenase is only present in AqS1 while the gene encoding alanine transaminase is present in the genomes of A. queenslandica and AqS3 (Supplementary File 4-7). Both genes are highly expressed at Q3 or Q4 in the adult holobiont. These two enzymes catalyse reversible conversion between L-alanine and pyruvate (Saier and Jenkins 1967, Dave and Kadeppagari 2019). This conversion can also be performed by D-alanine transaminase (EC 2.6.1.21), which converts pyruvate to D-alanine, and alanine racemase (EC 5.1.1.1), which turns D-alanine to L-alanine. The genes encoding these two enzymes are present in the genomes of AqS1, AqS2 and AqS3 and are expressed in the adult. 3-phosphoglycerate dehydrogenase (EC 1.1.1.95) and phosphoserine aminotransferase (EC 2.6.1.52) are encoded by the genes of A. queenslandica, AqS1, AqS2 and AqS3. These two enzymes convert 3P-D-glycerate to phosphoserine. Phosphoserine is the precursor for cysteine catalysed by cysteine synthase (EC 2.5.1.47) and for serine catalysed by phosphoserine phosphatase (EC 3.1.3.3) (Supplementary File 4-5). All the holobiont genes encoding these four enzymes are expressed in the adult holobiont. The three symbionts can synthesise glutamate from 2-oxoglutarate using glutamate synthase (EC 1.4.1.13) (Supplementary File 4-7). A. queenslandica can produce glutamate by glutaminase (EC

3.5.1.2) from glutamine (Supplementary File 4-7). The holobiont genes involved in glutamate producing are highly expressed in the adult. The four species can convert glutamate to glutamine by glutamine synthetase (EC 6.3.1.2) (Supplementary File 4-7) and proline by a series reactions of glutamate 5-kinase (EC 2.7.2.11), glutamate-5-semialdehyde dehydrogenase (EC 1.2.1.41) and pyrroline-5-carboxylate reductase (EC 1.5.1.2) (Supplementary File 4-8). All the holobiont genes involved in these two processes are expressed in the adult.

In summary, the sponge host cannot independently synthesise histidine, leucine, isoleucine, lysine, methionine, phenylalanine, valine, tryptophan, and arginine. These can be made available from the symbiotic bacteria. In contrast, the three symbionts cannot generate asparagine but may obtain it from the host. Thus, the symbionts can supply the sponge essential amino acids via amino acid transporters or being engulfed (Leys et al. 2017, Shih et al. 2020, Song et al. 2020). Four copies of SLC7A9 (Shigeta et al. 2006) and two copies of TC.APA (Casagrande et al. 2008), which encode L-type amino acid transporter and basic amino acid/polyamine antiporter respectively, are identified in A. queenslandica and expressed in the adult holobiont. Genes (aapJ, aapQ, aapM, and aapP) encoding complete general L-amino acid transport system (Walshaw et al. 1997) are present in AqS1 and AqS2, and expressed in the adult holobiont. Several copies of genes (livK, livH, livM, livG, and livF) encoding complete branched-chain amino acid transport system (Adams et al. 1990) are in AqS1, and all of these genes expressed in the adult holobiont. All of these active amino acid biosynthesis pathways and transporters are consistent with amino acids being transported between the host sponge and the three symbiotic bacteria.

4.4.6 Fatty acid biosynthesis in the A. queenslandica holobiont

The three primary symbiotic bacteria can synthesise fatty acids via the type II fatty acid biosynthetic pathway (Lu et al. 2004). This pathway is complete in AqS3, and nearly complete in AqS1 and AqS2, but incomplete in A. queenslandica (Supplementary File 4-9). The three symbionts can form malonyl-CoA from acetyl-CoA with acetyl-CoA carboxylase (EC 6.4.1.2). In A. queenslandica, malonyl-CoA can be derived from malonate by acyl-CoA synthetase family member 3, encoded by gene ACSF3. Malonyl-CoA is then transferred to an acyl carrier protein (ACP) and converted to malonyl-ACP by ACP-S-malonyltransferase (FabD), which is present in AqS1 and AqS3; ACP is present in the sponge and the three

78 symbionts. Malonyl-ACP and another acetyl-CoA can be condensed to acetoacetyl-ACP by 3-oxoacyl-[acyl-carrier-protein] synthase III (FabH) in AqS1, and AqS2, and by acetoacetyl- ACP synthase (FabY) in AqS3. Acetoacetyl-ACP is the substrate for 3-oxoacyl-ACP reductase (FabG) to form 3-hydroxybutanoyl-ACP. 3-hydroxybutanoyl-ACP can then be converted to butyryl-ACP via 3-hydroxyacyl-ACP dehydratase (FabZ / FabA) and enoyl- ACP reductase (FabI / MECR). FabZ is present in AqS2 and AqS3, and FabA in AqS3. FabI is present in the three symbionts, and MECR is present in A. queenslandica. All symbiotic bacterial and host sponge genes involved in this pathway are expressed in the adult holobiont. The symbionts can extend butyryl-ACP to the longest octadecanoic-ACP via this type II fatty acid biosynthetic pathway. Enzymes involved in fatty acid elongation are present in A. queenslandica. The sponge genes encoding these fatty acid elongating enzymes are all highly expressed (Q4) in the adult holobiont (Supplementary File 4-10).

4.4.7 Metabolism of cofactors and vitamins in the A. queenslandica holobiont

The vitamin B1 (thiamine-phosphate) biosynthesis pathway is partial complete in AqS1 and AqS2, but incomplete in A. queenslandica and AqS3. AqS1 and AqS2 can biosynthesise vitamin B1 (thiamine-phosphate) as 60% of their genes involved in this process are present and expressed in the adult holobiont (Supplementary File 4-11). A. queenslandica and AqS3 may be able to use thiamine-phosphate produced by AqS1 and AqS2, as the genes encoding alkaline phosphatase (EC 3.1.3.1) and thiamine phosphate phosphatase (EC 3.1.3.100) are present in their genomes, respectively. These two enzymes convert thiamine-phosphate to thiamine. The AqS3 gene encoding thiamine phosphate phosphatase, is lowly expressed (Q1) in the adult holobiont. The sponge PhoA encoding alkaline phosphatase is expressed (Q2) in the adult holobiont. This sponge PhoA is expressed higher in the early and middle embryo stages compared to the other developmental stages (Figure 4-5). In the adult cells, PhoA is expressed at Q4 in choanocyte, and at Q3 in archaeocyte and pinacocyte (Figure 4-5). In addition, A. queenslandica and AqS3 could uptake thiamine from the environment, as genes encoding thiamine transporter (gene SLC19A2_3, and SLC35F3_4 in A. queenslandica; gene thiQ in AqS3) are identified and expressed in the adult holobiont. The sponge genes SLC19A2_3 and SLC35F3_4 are expressed at Q3 or Q4 across the life cycle, and in the three cell types (Figure 4-5). Meanwhile, A. queenslandica and AqS3 genes that encode enzymes to phosphorylate thiamine are also expressed in the adult holobiont.

The riboflavin (vitamin B2) biosynthesis pathway is complete in the symbiotic bacteria AqS1 and AqS3, and nearly complete in AqS2: only the gene encoding 5-amino-6-(5-phospho-D- ribitylamino) uracil phosphatase (EC 3.1.3.104) is absent in the genome of AqS2. The symbiotic bacterial genes that convert guanosine 5'-triphosphate (GTP) and ribulose 5- phosphate to vitamin B2 are all expressed in the adult holobiont (Figure 4-8). In contrast, the genes involved in this vitamin B2 biosynthesis pathway all are missing in the A. queenslandica genome. The A. queenslandica, AqS1, and AqS2 genes encoding riboflavin transporter are present and expressed in the adult holobiont. The sponge gene SLC52A3, encoding riboflavin transporter, is highly expressed (Q3 or Q4) across the life cycle and in the three adult cell types.

Quartile 0 1 2 3 4

Figure 4-8. The expression levels of A. queenslandica holobiont genes involved in the riboflavin metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

Pantothenate (vitamin B5) can be synthesised by the symbionts with the intermediates from the host in A. queenslandica holobiont system (Figure 4-9). Pantothenate can be yielded by pantoate--beta-alanine ligase (encoded by gene panC, EC 6.3.2.1) with the intermediates β- alanine and pantoate (Figure 4-9). Gene panC is identified in AqS2 and AqS3 and expressed in the adult holobiont. Within the four species, β-alanine can only be produced by A.

80 queenslandica from uracil, and the sponge genes involved in this process are all highly expressed in the adult (Q4) and developmental stages (Q3 or Q4). Most of these sponge genes also are expressed at Q3 or Q4 three adult cell types, except that the gene DPYD, encoding dihydropyrimidine dehydrogenase (NADP+) (EC 1.3.1.2), is expressed less (Q2) in the pinacocytes. AqS2 and AqS3 express the genes to produce pantoate from pyruvate. In this process, AqS2 and AqS3 may convert 2-dehydropantoate to pantoate via ketol-acid reductoisomerase (encoded by gene ilvC, EC 1.1.1.86) (Leonardi and Jackowski 2007), as 2- dehydropantoate 2-reductase (encoded by gene panE, EC 1.1.1.169) is not found in AqS2 and AqS3. The AqS1 capability to biosynthesise pantothenate is not clear because the genes panB and panC are not detected in the genome, but two copies of panE are present and expressed at Q3 or Q4 in the adult holobiont; this seems to indicate a high possibility of AqS1 pantothenate biosynthesis. All these expressions of genes involved in the pantothenate biosynthesis pathway reveal the sponge can provide β-alanine for the symbiotic bacteria to produce pantothenate.

The folate biosynthesis pathway (Supplementary File 4-12), which can produce folate from GTP, is incomplete in A. queenslandica and AqS2, but is active in AqS1 and AqS3. All genes involved in this process are expressed, except that the gene encoding alkaline phosphatase (EC 3.1.3.1) is not recovered in AqS1. Meanwhile, the gene SLC46A1, which encodes proton- coupled folate transporter, is identified in A. queenslandica and expressed at Q2 in the adult holobiont. The sponge gene SLC46A1 is expressed at Q2 or Q3 across the life cycle and Q3, Q1, and Q2 in archaeocytes, choanocytes, and pinacocytes, respectively (Figure 4-5). Thus, float can be biosynthesized by AqS1 and AqS3 and be imported by A. queenslandica.

Pyruvate Aq AqS1 AqS2 AqS3 2.2.1.6

2-Acetolactate

1.1.1.86

2,3-Dihydroxy-3- Uracil methylbutanoate

4.2.1.9 1.3.1.2

3-Methyl-2- 5,6- oxobutanoate Dihydrouracil

2.1.2.11 3.5.2.2

2- N-Carbamoyl- Dehydropantoate beta-alanine 1.1.1.169 or 3.5.1.6 1.1.1.86 Pantoate beta-Alanine

6.3.2.1

Pantothenate

Figure 4-9. Pantothenate biosynthesis in the A. queenslandica holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is filled with colour, the enzyme is present and expressed in the adult holobiont, otherwise, it’s absent in that species. There are no cases in which a gene is present but not expressed in this pathway.

Different strategies are used by the species within the A. queenslandica holobiont to produce the cofactor NAD. A. queenslandica holobiont genomic and transcriptomic data suggest that A. queenslandica can convert tryptophan to NAD; NAD can also be produced by AqS2 and AqS3 from aspartate as well (Supplementary File 4-14). The NAD biosynthesis pathway is complete in A. queenslandica and AqS3, and nearly complete in AqS2 as only the gene nadC, encoding nicotinate-nucleotide diphosphorylase (EC 2.4.2.19), is absent. All the A. queenslandica, AqS2 and AqS3 genes involved in the NAD biosynthesis pathway are

82 expressed in the adult holobiont. The NAD biosynthesis pathway is incomplete in AqS1.

AqS1 can produce NAD when vitamin B3 is available.

The holobiont genomic and transcriptomic data also reveal the capacity of AqS1 to biosynthesise siroheme. Siroheme is an essential cofactor that helps the enzymes sulphite and nitrite reductases accomplish the reduction of sulphate and nitrate (Tripathy et al. 2010). A siroheme synthesis pathway originating from glutamate is active in the symbiotic bacteria AqS1; all the AqS1 genes involved in this pathway are expressed in the adult holobiont, except that the gene hemB, which encodes porphobilinogen synthase (EC 4.2.1.24), is not found in the genome of AqS1 (Supplementary File 4-13). The gene cysG, which encodes the essential multifunctional enzyme siroheme synthase (EC 2.1.1.107, 1.3.1.76, 4.99.1.4) and is solely responsible for siroheme synthesis (Spencer et al. 1993), is only identified in AqS1 and is expressed at Q2 in the adult holobiont. Some other genes involved in this pathway are also identified and expressed in the adult holobiont by the sponge and the other two symbiotic bacteria, but these genes are also involved in the biosynthesis of heme (Bali et al. 2011) for which there is a complete or near-complete pathway active in the host sponge and the three symbionts. Thus, siroheme can be produced by symbiont AqS1, but not by A. queenslandica, AqS2 and AqS3.

4.4.8 Dissolved inorganic nitrogen assimilation capacity of the A. queenslandica holobiont

The holobiont genomic and transcriptomic data has revealed the capacity of A. queenslandica holobiont to assimilate dissolved inorganic nitrogen (DIN) from the surrounding seawater. DIN comprises nitrate, nitrite, and ammonium (Priya et al. 2017). Nitrate can be reduced to nitrite by nitrate reductase, which is subsequently reduced to ammonia by nitrite reductase (Allen et al. 2001). The two-component NarX-NarL sensor-response regulator pairs can response to nitrate and nitrite, and regulate the activity of fumarate reductase (Schröder et al. 1994, Goh et al. 2005) and nitrate reductase (Noriega et al. 2010). The genes, encoding NarX-NarL two-component regulatory systems, nitrate reductase and three (frdABC) of the four subunits (frdABCD) of fumarate reductase are present and expressed in the adult holobiont by AqS1. Gene nit-6, encoding nitrite reductase [NAD(P)H] (EC 1.7.1.4) is identified in A. queenslandica and expressed at Q3 in the adult holobiont. In the adult sponge cells, the sponge gene nit-6 expressed at Q2 and Q3 in archaeocytes and choanocytes,

83 respectively, but not at all in pinacocytes (Figure 4-5). Siroheme is required for the function of nitrite reductase [NAD(P)H] (EC 1.7.1.4) (Vega and Garrett 1975, Colandene and Garrett 1996), which can be generated by the symbiont AqS1 in the A. queenslandica holobiont. Thus, the sponge A. queenslandica can produce ammonia via nitrite reduction, using cofactor and substrate (nitrite) available from AqS1. The produced ammonia can be assimilated by synthesis glutamine through glutamine synthetase (GS, EC 6.3.1.2), or synthesis glutamate through glutamate dehydrogenase (GDH, EC 1.4.1.2/1.4.1.3/1.4.1.4). The gene glnA is highly expressed (Q4) in the adult holobiont by the sponge and the three primary symbionts. Genes encoding glutamate dehydrogenase (GDH) are also present in A. queenslandica, AqS1, and AqS2, and highly expressed at Q4 in the adult holobiont. The sponge genes encoding GS and GDH are also highly expressed (Q4) across the life cycles and in the three adult cell types.

4.4.9 Sulfur assimilation capacity of A. queenslandica holobiont

The symbiont AqS1 facilitates sulfur assimilation in the sponge holobiont. AqS1 is a sulphur- oxidising bacterium (Gauthier et al. 2016) equipped with near-complete SOX system and a reverse dissimilatory sulfate reduction and oxidation pathway. The SOX complex can oxidize thiosulfate, sulfide, sulfite, and elemental sulfur to sulfate (Reviewed in Ghosh and Dam 2009). The genes encoding SOX complex in AqS1 include soxAXYZBD but no soxC, and these genes are expressed in the adult holobiont. AqS1 can also reversely catalyse the sulfate to sulfide through a series reactions of sulfate adenylyl transferase (Sat), heterodisulphide reductase complex (HdrABC), adenylylsulfate reductase complex (AprAB), dissimilatory sulfite reductase complex (DsrABC), and Hdr-like proteins, e.g. HdrD (Santos et al. 2015, Wang et al. 2019). The AqS1 genes involved in this sulfate and sulfide redox are all expressed in the adult holobiont.

The sulfide generated by AqS1 can be used to synthesis cysteine by A. queenslandica and the three symbionts via cysteine synthase (EC 2.5.1.47), or to produce thiosulfate by A. queenslandica via sulfide:quinone oxidoreductase (SQOR, EC 1.8.5.8) (Jackson et al. 2012) and putative thiosulfate sulfurtransferase (rhodanese-like protein) (Libiad et al. 2014). The genes of A. queenslandica and the three symbionts encoding cysteine synthase (EC 2.5.1.47), and the A. queenslandica gene encoding SQOR and Rhodanese are all highly expressed (Q3 or Q4) in the adult holobiont. The A. queenslandica cysK, encoding cysteine synthase (EC 2.5.1.47), is expressed (Q2 or Q3) across the life cycle, with higher levels observed during

84 late metamorphosis (Figure 4-10). In the adult, it is expressed in Q3, Q1, and Q2 in archaeocytes, choanocytes, and pinacocytes, respectively (Figure 4-10). The A. queenslandica gene encoding SQOR is highly expressed (Q4) during most development stages with highest expression in adult, and the three adult cell types with highest expression in archaeocytes (Figure 4-10). The A. queenslandica gene encoding rhodanese is expressed the higher in cleavage (Q4) than the other developmental stages (Q3), and it is highly expressed (Q4) in the three adult cells with highest expression in archaeocytes (Figure 4-10).

A. queenslandica can also assimilate sulfate through transportation via prestin (SLC26A5) (Schaechinger and Oliver 2007) and formation of 3'-phosphoadenosine 5'-phosphosulfate (PAPS) via bifunctional 3'-phosphoadenosine 5'-phosphosulfate synthase (PAPSS, EC 2.7.7.4, 2.7.1.25). Two copies of SLC26A5 and one PAPSS are present in A. queenslandica and are highly expressed (Q4) in the adult holobiont, with highest expression in archaeocytes (Figure 4-10). They are also expressed across the life cycle, with higher levels observed during late metamorphosis and in the adult (Figure 4-10). Another type of sulfate transporter (CysZ) is present in AqS1 and expressed (Q2) in the adult holobiont, but CysZ cannot transport thiosulfate (Zhang et al. 2014). Sulfate-thiosulfate permease (CysPTWA) can transport sulfate and thiosulfate (Aguilar-Barajas et al. 2011).Although only gene cysA, encoding sulfate/thiosulfate import ATP-binding subunit (CysA) of CysPTWA, is present in AqS1, it is highly expressed (Q4) in the adult holobiont, which suggests AqS1 may import thiosulfate through CysA.

CysK Stage Developmental SQOR Cleavage 4

tile Rhodanese Early 2 r SLC26A5.1 Mid 0 Late -2 Qua SLC26A5.2 Larva PAPSS Metamorphosis Cell_type Adult 1.5 CysK 1 Quartile ed SQOR 0.5 z 0 Rhodanese 1 0 SLC26A5.1 2 -0.5 mali

r SLC26A5.2 3 -1 4 -1.5

No PAPSS wn ing age r o Ring Spot Adult Cloud av Br Oscula ent-pole 1_hr_PS Chamber Cle Late_ T Late_spot 6-7_hr_PS Competent Pinacocyte Choanocyte Archaeocyte 11-12_hr_PS 23-24_hr_PS Precompetent

Figure 4-10. Expression levels of A. queenslandica genes involved in sulfur metabolism throughout development and in cell types. The upper heatmap presents these gene expression quartiles, and the bottom one presents the TPM normalised gene expression levels. In the row names, the number after the point represents the number of gene copy, e.g., SLC26A5.1 and SLC26A5.2 are two copies of genes encoding sulfate transporter prestin (SLC26A5). CysK, cysteine synthase (EC 2.5.1.47); SQOR, sulfide:quinone oxidoreductase (EC 1.8.5.8); Rhodanese, putative thiosulfate sulfurtransferase (rhodanese-like protein); PAPSS, bifunctional 3'-phosphoadenosine 5'- phosphosulfate synthase (PAPSS, EC 2.7.7.4, 2.7.1.25).

4.4.10 Phosphorus assimilation capacity of A. queenslandica holobiont

The holobiont genomic and transcriptomic data has revealed the symbionts can assimilate inorganic phosphorus (phosphate, Pi) in the A. queenslandica holobiont. The phosphate (Pho) regulon is present in AqS1 and AqS3 and expressed in the adult holobiont (Figure 4-11). Under conditions of Pi limitation, phosphate regulon sensor protein (PhoR) can activate phosphate regulon transcriptional regulatory protein (PhoB) via phosphorylation, thus regulate the expression of downstream response genes, e.g. pstS, encoding the phosphate binding protein; pstABC, encoding the phosphate ABC transporters and phoU encoding the phosphate transport system regulatory protein (Lamarche et al. 2008, Santos-Beneit 2015, Peng et al. 2017). The symbiont AqS1 and AqS3 can import Pi through this Pho regulon system, which is incomplete in AqS2 (Figure 4-11). The imported Pi is then accumulated in the symbionts as polyphosphate (polyP) by polyphosphate kinase (EC 2.7.4.1) (Zhang et al.

2015). Gene ppk, encoding polyphosphate kinase, is only present in AqS1 and AqS3, and is expressed in the adult holobiont (Figure 4-11).

Figure 4-11. Schematic representations of a working model for polyphosphate accumulation in the A. queenslandica holobiont. The four sub-blocks in each protein block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is filled with colour, the enzyme is present and expressed in the adult holobiont, otherwise, it’s absent in that species. There are no cases in which a gene is present but not expressed in this pathway. PhoBR, the phosphate two-component regulator and sensor proteins; PstS, phosphate binding protein; PstABC, the phosphate ABC transporter; PhoU, the phosphate transport system regulatory protein; PPK, polyphosphate kinase; PolyP, polyphosphate. Figure adapted from (Peng et al. 2017).

4.5 Discussion

4.5.1 Nutrient assimilated by the sponge holobiont

In this chapter, I have explored the metabolic pathways of the host A. queenslandica and its three primary symbionts, AqS1, AqS2 and AqS3, through genomic and transcriptomic analyses and identified their carbohydrate, nitrogen, phosphate, and sulfur metabolic

87 capacities. These pathways suggest the host sponge and symbiotic bacteria collaborate together to assimilate the major nutrients, carbon, nitrogen, sulfur and phosphorus (Figure 4-12). These findings advance our understanding of the critical benthic-pelagic coupling functional roles (Bell 2008, Folkers and Rombouts 2020) of the sponge and symbiotic microorganisms.

Glutamine Nitrogen Nitrate Nitrite Ammonia Glutamate Fatty acids Purine and pyrimidine Amino acids ** ** ** * ** Carbon F, H, I, K, L, M, R, V, W A, C,*** D, E, G, P, Q, S, T, Y Phosphorus Glycogen Starch N PolyP Phosphate Vitamins and cofacters

B1 Glucose B Amphimedon queenslandica 2

B5 Folate Sulfur Thiosulfate Sulfate Sulfde Cysteine Siroheme

Sulfur globules Aq AqS1 AqS2 AqS3

Figure 4-12. Schematic overview of metabolic interactions between all of the members of the A. queenslandica holobiont. The dash lines present the nutrients obtained from environment; the solid lines present the reactions in the holobiont. The four sub-blocks in block from left to right represent the reactions in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is filled with colour, the reaction is present in that species, otherwise, the reaction is present but partial complete in that species. If the sub-block stroke is transparent, the reaction is not present in that species. Amino acids: A, Alanine; C, Cysteine; D, Aspartate (aspartic acid); E, Glutamic acid; F, Phenylalanine; G, Glycine; H, Histidine; I, Isoleucine; K, Lysine; L, Leucine; M, Methionine; N, Asparagine; P, Proline; Q, Glutamine; R, Arginine; S, Serine; T, Threonine; V, Valine; W, Tryptophan; Y, Tyrosine. Asterisk sign above the amino acids: *, the pathway is near-complete in AqS1; **, the pathway is near-complete in AqS2; ***, the pathway is not present in AqS3.

Both the host sponges and the symbiotic bacteria can perform heterotrophic carbon assimilation in the marine ecosystem. Complete glycolysis, TCA cycle, and PPP are present in A. queenslandica and the three primary symbionts, which together indicate their capacity for aerobic cellular respiration and heterotrophic carbon assimilation. Similar active symbiotic bacterial glycolysis, TCA cycle, PPP pathways have also been reported in Cymbastela concentrica holobiont, in which the symbiont CcPhy may utilise the DOC coming from the filtering activity of the host sponge (Moitinho-Silva et al. 2017), but it has

88 mainly focused on the symbionts. Isotope experiments have reported both the sponge and symbiotic bacterial are assimilating DOM at the same similar rate, which accounts for the majority of their heterotrophic diet (Rix et al. 2017, Rix et al. 2020).

The assimilated carbohydrates can be used for the sponge energy expenditure (Reiswig 1974, Weisz et al. 2008, Pfannkuchen et al. 2009) and also the cell production, which is a crucial step to makes the nutrients available to higher trophic level reef inhabitants (de Goeij et al. 2013, Alexander et al. 2014, Maldonado 2015, Vanwonterghem and Webster 2020). The metabolic pathways to utilize the assimilated carbon, such as biosynthesis of fatty acids, purines, pyrimidines, and amino acids, have been identified in both A. queenslandica and its three primary symbionts. The intermediate metabolites of carbohydrate catabolism, such as pyruvate, acetyl-CoA, 2-oxoglutarate, and PRPP, are vital material for fatty acid, purine, pyrimidine and many amino acid biosyntheses. Phospholipid fatty acid (PLFA) analysis in sponge Chondrilla sacciformis (HMA), Hemimycale arabica and Mycale fistulifera (LMA) reveal that these sponges can incorporate DOM into sponge- and bacteria-specific PLFAs, respectively (Rix et al. 2017). Isotopic analysis in the sponge Mycale grandis reveals that the host sponge directly assimilates amino acids synthesised by the symbiotic bacteria, originally from DOM, which is a vital pathway for sponge amino acids acquisition (Shih et al. 2020). Thus, both the sponge and symbiotic bacteria may assimilate carbohydrate and incorporate them into fatty acids and amino acids, which are necessary material to build blocks of the cell (Alberts et al. 2002).

The sponge genomic and transcriptomic information facilitate a more comprehensive understanding of the host contribution and variation in heterotrophic carbon assimilation during life cycle. The expression levels of genes encoding hexokinase, ADPGK, PYG and PPP during life cycle indicate a higher carbohydrate assimilation capacity in metamorphosis and adult stages. This higher carbohydrate assimilation capacity during oscula/juvenile and adult stages may correlate with the capacity to pump and filter sea water, which require an energetic expenditure (Pfannkuchen et al. 2009). Meanwhile, recent studies in many cancers have demonstrated that the activities of PPP and glycolysis might regulate the energy production and incorporation of nutrients into biomass needed to produce new cells (Heiden et al. 2009, Liberti and Locasale 2016). Increased PPP activity was found in cancer cells (Jiang et al. 2014). The increased PPP activity of A. queenslandica in late postlarvae juvenile and adult stages may also help explain the rapid sponge cell proliferation rate in

89 metamorphosing A. queenslandica (Sogabe et al. 2016) and in other sponges (Ayling 1983, de Goeij et al. 2013, Alexander et al. 2014, Maldonado 2015). During metamorphosis, an influx of environmental bacteria, many of which can be phagocytosed by sponge cells (Fieth et al. 2016), may provide the sponge with lots bacterial carbon (Kahn et al. 2015, Leys et al. 2017, Kahn et al. 2018, Kazanidis et al. 2018). At this stage, to reorganise the body plan (Degnan et al. 2015), most of the larval sponge cells undergo programmed cell death or transdifferentiate into a range of juvenile cell types (Nakanishi et al. 2014), and many new cells are proliferated (Sogabe et al. 2016).

Besides carbon, both the host sponge and symbiont bacteria can assimilate DIN from the surrounding seawater. Nitrate can be reduced to nitrite by symbiont AqS1, which is subsequently reduced to ammonia by A. queenslandica, with the help of the cofactor siroheme produced by AqS1. The resultant ammonia can be assimilated via glutamine or glutamate synthesis by the host and the three symbionts. The ammonia and nitrite concentrations in the A. queenslandica reef flat habitat are much lower than that of nitrate throughout the year (Watson et al. 2017), therefore, the symbiotic bacterial capacity to reduce nitrate is particular important for the sponge holobiont DIN assimilation. Nitrogen transformation pathways, including ammonia assimilation, ammonia oxidation, assimilatory/dissimilatory nitrate reduction and (de)nitrification, have been reported in many sponge-associated microbes (Fiore et al. 2015, Moitinho-Silva et al. 2017, Weigel and Erwin 2017, Zhang et al. 2019, Engelberts et al. 2020). However, ammonia is considered to be a sponge metabolic waste product (Davy et al. 2002) and assimilated by the symbiotic microbes (Fiore et al. 2013, Fiore et al. 2015). The microbe-produced organic nitrogen may then be translocated to the host sponge cells, which cannot assimilate ammonia directly (Fiore et al. 2015, Achlatis et al. 2018). With the genomic and transcriptomic data from both A. queenslandica and its primary symbionts, I identified the cooperation between the host and symbionts to assimilate DIN. Specifically, the expression levels of A. queenslandica genes encoding GS and GDH are high (Q4) across the life cycles and in the three adult cell types, which indicates the ammonia assimilation capacity of the sponge. The transcriptional activity of GS is also found in sponge Cliona varians (Riesgo et al. 2014), suggesting the ammonia assimilation capacity of the host sponge is not unique in A. queenslandica.

The associated symbionts play an important role in A. queenslandica holobiont sulfur assimilation from the water column, which is rich in sulfate, but limited in sulfide (Watson et

90 al. 2017). AqS1 (Gauthier et al. 2016) can oxidize environment thiosulfate and elemental sulfur to sulfate (Reviewed in Ghosh and Dam 2009) and also reduce sulfate to sulfide. The sponge hologenomic and transcriptomic data advance our understanding of the sulfur assimilation of A. queenslandica holobiont to a more fine-tuned level. Both A. queenslandica and the three primary symbionts have the genomic capacity to assimilate sulfide, which may be mainly generated by AqS1. Interestingly, during the life cycle, the sponge activity to generate thiosulfate is most active in adult, especially in archaeocytes (Figure 4-10; Figure 4-12). Sponge pluripotent archaeocytes can differentiate into a range of other cell types and significantly upregulate genes that control cell proliferation (Sogabe et al. 2019).

Phosphorous (P) is one of the limiting elements in oligotrophic marine habitats (Watson et al. 2017). The sponge associated symbionts can accumulate Pi by the formation of long-chain polyP through polyphosphate kinase (Zhang et al. 2015). In A. queenslandica holoboint, besides AqS1 (Gauthier et al. 2016), AqS3 can also accumulate Pi with the regulation through Pho regulon (Figure 4-11) (Santos-Beneit 2015).

4.5.2 Metabolic complementation between the host sponge and symbiotic microorganisms

The A. queenslandica holobiont genomes and transcriptomes have also revealed the potential for metabolic complementation between the sponge and symbionts. Various amino acid biosynthesis pathways are present in the three primary symbionts. The amino acid biosynthesis capacity has been identified in other sponge symbionts (Fiore et al. 2015, Lackner et al. 2017, Bayer et al. 2018). The genomic information of A. queenslandica reveal the host sponge is cannot synthesis histidine, leucine, isoleucine, lysine, methionine, phenylalanine, valine, tryptophan and arginine, but can synthesis threonine, which is essential amino acid for humans (Reeds 2000), and asparagine, which cannot be synthesised by the three primary symbionts. These metabolic pathways suggest amino acid complementation may happen between the host sponge and the three primary symbionts. The arginine biosynthesis complementation between A. queenslandica and symbiotic bacteria is confirmed through stable isotope tracing (Song et al. 2020). Thus, the symbionts can supply the sponge essential amino acids via amino acid transporters or being engulfed (Leys et al. 2017, Shih et al. 2020, Song et al. 2020).

The three primary symbiotic bacteria can provide the host sponge with some B vitamins and

91 cofactors. The A. queenslandica holobiont genomic and transcriptomic data reveal that the vitamin B1, B2, folate, and siroheme biosynthesis pathways are active in AqS1, vitamin B2 biosynthesis pathway is near-complete in AqS2, and vitamin B2 and folate biosynthesis pathways are complete in AqS3. The genomic capacities of symbiotic bacteria to produce B vitamins and cofactors has been found in other sponge holobiont systems (Fiore et al. 2015, Engelberts et al. 2020). The microbiome of sponge Ircinia ramose can biosynthesise vitamin

B1, B2, B5, B6, B7, and B12 (Engelberts et al. 2020). The microbes of sponge Xestospongia muta can produce vitamin B1, B2, B7, and B12 (Fiore et al. 2015). The genomic information of A. queenslandica indicates the host sponge is heterotrophic for these symbiont-derived B vitamins and cofactors. In particular, transporters for B1 and folate uptake are identified in A. queenslandica and are expressed throughout the life cycle (Figure 4-5). As AqS1, AqS2, and AqS3 are vertically inherited in A. queenslandica and maintain proportional dominance at most developmental stages (Fieth et al. 2016). These symbionts may influence the host development and nutrient assimilation through these B vitamins and cofactors, such as the siroheme role in nitrite reduction (mentioned above), which can produce ammonia. Notably, the expression level of A. queenslandica nit-6, encoding nitrite reductase, is highest at the beginning of metamorphosis (Figure 4-5). Ammonia is an important chemical cue to initiate larval settlement in some invertebrates (Coon et al. 1990, Hadfield and Paul 2001), so it may be that the symbiotic bacterial influence the sponge larval settlement through regulating ammonia levels in the holoboint.

4.6 Conclusion

In this study, I used new and existing A. queenslandica holobiont genomic and transcriptomic information to reveal the complex metabolic pathways of the host sponge A. queenslandica and its three primary symbiotic bacteria, AqS1, AqS2, and AqS3. These integrated pathways indicate the potential for metabolite complementation between the host sponge and the symbiotic bacteria and their potential collaboration in carbon, nitrogen, sulfur and phosphorous assimilation. These biological interactions reveal likely cooperation between the sponge and its symbionts to assimilate nutrients from the ambient seawater, and thus their dependence on each other in metabolism. The life cycle transcriptomic analysis of A. queenslandica reveals the critical roles of the symbiotic bacteria for the sponge development. The cell type transcriptomic data provides more comprehensive interactions between symbiotic bacteria and sponge cell.

Chapter 5 Interkingdom signalling between a sponge and its primary symbiotic bacteria

5.1 Abstract

Interkingdom signalling within a holobiont allows host and symbionts to communicate and regulate each other's physiological and developmental states. Analyses of the genomes and transcriptomes of the demosponge Amphimedon queenslandica and its dominant bacterial symbionts reveal that most known cell signalling functions – ligands, receptors and transducers – are encoded and expressed by the sponge host (Chapters 2 and 3). Here, however, I show that here a suite of signalling molecules known as neurotransmitters in neural animals can be produced by the aneural holobiont or by the symbionts alone. Specifically, A. queenslandica alone can synthesise and release γ-aminobutyric acid (GABA), which is involved in quorum sensing in other holobionts. The host-produced GABA can then be imported and degraded by the bacterial symbiont AqS1. Multiple GABA receptors are encoded by the A. queenslandica genome, indicating that this signalling molecule has the potential to influence sponge behaviours such as larvae settlement, so it is significant that the presence of GABA can be moderated by AqS1. The bacterial symbionts on their own can generate other metabolites – dopamine, tyramine, tryptamine, acetate, and propionate – that are used as neurotransmitters in neural animals. These could hypothetically be received by A. queenslandica G protein-coupled receptors (GPCRs) and thus could further regulate the host signalling pathways and influence host physiology. In this chapter, I show that the experimental introduction on one of these – dopamine – can influence larval phototactic swimming behaviour, potentially via a dopamine- related GPCR. This study investigated the potential signalling communication between the host sponge and its primary symbiotic bacteria to expand our knowledge of crosstalk interactions within sponge-bacteria symbiosis. The animal-bacterial conversations via host-GPCR – symbiont-derived ligand interactions might date back before the emergence of the nervous system.

5.2 Introduction

Sponges have lived with diverse and abundant symbiotic microbial communities (Thomas et al. 2016, Webster and Thomas 2016, O’Brien et al. 2020, Steinert et al. 2020) for over 600 million years (Müller et al. 2009). To maintain the sponge-bacteria symbiotic relationship, sponges and symbiotic bacteria have likely evolved inter-kingdom communication systems. In sponges, most research in this area has focused to date on quorum sensing (QS) signals (Gardères et al. 2014). Two classes of QS signals, termed autoinducers, N-acyl-homoserine lactones (AHLs) (Britstein et al. 2017, Mangano et al. 2018, Reen et al. 2019), and autoinducer-2 (AI-2), e.g. furanone derivatives (Zan et al. 2011), can be produced by at least some sponge-associated bacteria. AHLs are mediated by luxI (encoding AHL synthase), and luxR (encoding the signal receptor); AI-2 is mediated by luxS (encoding AI-2 synthases) (Hmelo 2017) and luxP (encoding the signal receptor) (Hmelo 2017). Screening of 211 Antarctic sponge-associated bacteria detected AHL with different lengths in more than half of these bacterial genera (Mangano et al. 2018). Screening of 15 sponge species from the Mediterranean and the Red Sea detected AHL in seven species but not in an Amphimedon species (Britstein et al. 2017). These studies suggest that an AHL-QS system is not ubiquitous in sponge holobionts. The communication mechanisms between sponges and symbionts without AHL QS signals are poorly understood so far.

Some signalling molecules that are used as neurotransmitters in neural animals are reported as being involved in inter-kingdom signalling between host animals and symbiotic bacteria. The most widely studied of these are human gut microbiota, which releases various signal metabolites including nitric oxide (NO) (Sobko et al. 2006), acetylcholine, serotonin, dopamine, noradrenaline (NA), GABA, trace amines (Galland 2014, Mazzoli and Pessione 2016) and short-chain fatty acids (SCFAs) (Silva et al. 2020). These signalling molecules enable bidirectional host-symbiont communication through the microbiota-gut-brain axis (Collins et al. 2012). Symbiotic microbes produce neuroactive molecules that can, both directly and indirectly, access the host central and enteric nervous, endocrine and immune systems, and influence the host physiology (Carabotti et al. 2015, Mazzoli and Pessione 2016, Martin et al. 2018, Silva et al. 2020). For example, Bacteroides, the major bacterial producer of GABA in the human gut, can alter the host GABA levels, which is associated with depression (Strandwitz et al. 2019). Conversely, the host can also shape the composition of the gut microbiota via these same signalling molecules (Collins et al. 2012). For instance, NA levels in the gut increase during stress, which results in enhanced growth of Escherichia coli and other Proteobacteria (Collins et al. 2012, Galland 2014).

One central mechanism of host-symbiont communication is through microbiota-derived GPCR-

94 active metabolites (Cohen et al. 2017, Husted et al. 2017, Chen et al. 2019, Colosimo et al. 2019, Pandey et al. 2019). Recent large-scale human gut microbiota metabolome screening against hundreds of GPCRs has identified a large number of metabolites produced by gut microbes, including phenylpropanoic acid, cadaverine, 9-10-methylenehexadecanoic acid, 12- methyltetradecanoic acid and trace amines. Each of these can act as a ligand for specific GPCRs and can trigger distinct physiological responses in the host (Chen et al. 2019, Colosimo et al. 2019). For example, a specific gut strain of Bacteroides thetaiotaomicron can produce the essential amino acid phenylalanine, which not only is an agonist for adhesion GPCRs (GPR56 and GPR97) but also can be converted by another gut strain, Morganella morganii, into the trace amine phenethylamine (Chen et al. 2019). Phenethylamine can readily cross the blood-brain barrier and activate dopamine receptors (Chen et al. 2019). Through this signalling interaction, the symbiotic bacteria can significantly impact local and systemic host physiology (Chen et al. 2019).

Sponges lack conventional nervous systems but have complex signalling pathways that enable them to respond to a range of stimuli (Leys 2015, Leys et al. 2019). GPCRs play an essential role in eukaryotic signal transduction (Jastrzebska 2013). A large number and diversity of GPCRs are present in sponges, including metabotropic glutamate (mGluRs), GABA (Francis et al. 2017, Kenny et al. 2020), adrenergic, serotonin, and dopamine receptors (Krishnan et al. 2014). Genomic and transcriptomic data reveal that the biosynthesis pathways of a wide range of small signalling molecules that function as neurotransmitters in neural animals are also present in sponges (Francis et al. 2017, Leys et al. 2019), such as glutamate and GABA (Elliott and Leys 2010, Srivastava et al. 2010, Francis et al. 2017, Leys et al. 2019). However, some of these signalling molecules, such as dopamine, serotonin, and adrenaline, cannot be generated by the demosponges A. queenslandica (Srivastava et al. 2010) and Tethya wilhelma (Francis et al. 2017, Leys et al. 2019). Serotonin or serotonin-like molecules found in sponge extracts (Hu et al. 2002, Hedner et al. 2006, Kochanowska et al. 2008) are thus considered to be produced by the bacterial symbionts (Leys 2015). The sponge's capacity to receive the signal (whether sponge- or bacterial- derived) via GPCRs indicates the potential for sponge-bacterial signal transduction through small signalling molecules. Yet, the capacity of sponge-associated bacteria to produce these small signalling molecules is poorly known.

In this chapter, potential sponge-bacteria inter-kingdom signals are studied by exploring the genomic and transcriptomic data of the host sponge A. queenslandica, and its three dominant symbionts, AqS1, AqS2, and AqS3. Specifically, I analyse the capacities of the A. queenslandica holobiont to produce and respond to QS signals and other small signalling molecules, including amino acids (glutamate and GABA), biogenic amines (catecholamine and trace amines), and

SCFAs (acetate and propionate). The gene repertoire of A. queenslandica, AqS1, AqS2, and AqS3 reveals the potential for host-symbiont signal interactions and regulatory mechanisms.

5.3 Materials and Methods

5.3.1 Reconstruction of KEGG pathways

The functional gene results of KEGG (Kanehisa et al. 2017) and Blast2GO (Conesa and Gotz 2008) annotations (Chapter 2) were explored to identify sponge-bacterial potential signal communications. The signalling pathways of A. queenslandica and the QS system of the three primary symbionts, AqS1, AqS2, and AqS3, were reconstructed based on the KEGG annotations resulting from KEGG mapper (Kanehisa and Sato 2020). As a large proportion of genes have no orthologues in the KEGG database, missing genes in some KEGG pathways were manually checked against the Blast2GO gene function annotation results (Chapter 2).

The adult A. queenslandica holobiont transcriptome data (Chapters 3 and 4) were analysed to confirm that holobiont genes involved in predicted inter-kingdom signal interactions are expressed. To better decipher how these host-symbiont interactions varied at different developmental stages and cell types, the sponge Cel-Seq2 transcriptomes of 82 samples from 17 different developmental tissue stages (from cleavage embryos to adults, referred as developmental transcriptome from now on) (Anavy et al. 2014, Levin et al. 2016) and 31 samples from 3 cell types (archaeocytes, choanocytes, pinacocytes, referred as cell type transcriptome from now on) (Sogabe et al. 2019) were also used in this study. A quartile analysis (as described in Chapter 4) was used to represent the gene expression levels.

5.3.2 The capability of the sponge A. queenslandica to respond to dopamine

To seek experimental evidence for the existence of a dopamine receptor in A. queenslandica, I analysed the phototactic swimming behaviour of sponge larvae when they were exposed to a dopamine receptor agonist (DRA) and antagonist (DRAA). Both the DRA (rotigotine hydrochloride) and DRAA (flupenthixol dihydrochloride) agents were obtained from Abcam (Melbourne, Australia).

Collection of A. queenslandica adults and larvae

Adult A. queenslandica sponges were collected from the Heron Island Reef, Great Barrier Reef, Queensland, Australia (Latitude −23.44, Longitude 151.92) as previously described (Leys et al.

2008). These adult sponge samples were then transported to a closed aquaria system at the University of Queensland and maintained in artificial seawater at approximately 25°C. Aquaria containing four adult sponges were heated by about 1-2°C above ambient water temperature to facilitate larval release (Leys et al. 2008). The released larvae were collected using a Pasteur pipette over a 1 hr period and were pooled for subsequent experimentation.

Optimisation of DRA and DRAA concentrations

To determine the maximum concentration of DRA and DRAA that the sponge larvae can survive, ten larvae were exposed to different concentrations of DRA and DRAA in 10 ml 0.22 μm filtered seawater (FSW) in 6-well plates. The DRA and DRAA were tested at 10-4, 10-5, 5x10-6, 10-6 and 10- 7 M; FSW was used as control. 10-3 M DRA and DRAA stock solutions were prepared first and then diluted to the test concentrations, respectively. Mortality and swimming ability of the larvae were observed after 30, 60, 90, 120, 150, 210, and 300 min post-exposure to the chemicals.

Sponge larvae swimming behaviour when exposed to DRA and DRAA

Given that A. queenslandica larvae are negatively phototactic (Leys and Degnan 2001, Say and Degnan 2020), the larval swimming behaviour assays were conducted in a 20 ml chamber with diffuse light presented at one side. Larvae aged between 4 to 6 h post-emergence (hpe) were used for these phototactic assays. Ten larvae were incubated in FSW containing either 10-5 M DRA for 2 min or 10-5 M DRAA for 3 min (the incubation time and the maximum survival concentration were tested prior with a small batch of larvae) and then transferred to the light side of the chamber containing 10-5 M DRA or DRAA in FSW. For the DRAA_DRA group, ten larvae were incubated in the 10-5 M DRAA in FSW for 3 minutes and then transferred to the light side of the chamber containing 10-5 M DRA in FSW. FSW was used as control. The larval swimming behaviours were recorded for 60 seconds. To quantify larval phototaxis, the chamber was equally divided into quartiles from the light side (Q1) to the dark side (Q4), and the larval number in each quartile was counted every 5 seconds (Wong 2020). For each group, there were seven replicates and ten larvae per replicate. The average larval number in each quartile at all the time points was calculated.

The effects of DRA and DRAA on larval phototactic swimming behaviour were examined using the generalised linear mixed model (glmer) with the R package lme4 (Bates et al. 2014). Given the normal larval negative phototaxis characteristics (Degnan et al. 2008), FSW-control larval number in Q1 decreased and in Q4 increased over time, hence the Q1 and Q4 series larval numbers were fitted to Poisson generalised linear mixed models separately. The chemical treatment and swimming time were treated as fixed effects; replicates and the swimming time of each chemical treatment

97 were treated as random effects. The differences between models were assessed with the likelihood ratio test (anova), and the gof function in R package aods3 (Lesnoff et al. 2018) was used to check the goodness of fit for models.

5.4 Results

5.4.1 A. queenslandica has the potential to signal its symbionts

The A. queenslandica genome encodes enzymes that allow for the synthesis of neurochemicals present in neural animals, including GABA, nitric oxide (NO) and noradrenaline (NA). The release of these molecules from A. queenslandica may be received by the symbionts as quorum sensing molecules and thus may be involved in communication between host and symbionts.

GABA signalling

The host sponge A. queenslandica can generate and release GABA. A. queenslandica can catalyse glutamate (Glu) to GABA by glutamate decarboxylase (GAD, EC 4.1.1.15) (Figure 5-1). The gene encoding GAD is only present in the sponge genome and expressed in the adult holobiont, but not in the symbiont genomes. Through the sponge life cycle, GAD is highly expressed; its expression is in Q3 from cleavage embryo to 1 hr post settlement, increases to Q4 from 6-7 hr post-settlement to oscula, and then decreases to Q3 in the adult (Figure 5-2). In adult cells, it is expressed in Q3 in choanocytes and pinacocytes, and Q4 in archaeocytes (Figure 5-2). Gene SLC6A12, encoding a GABA transporter (solute carrier family 6 member 12) (Zhou et al. 2012, César-Razquin et al. 2015), is also present in A. queenslandica and expressed in the adult holobiont. SLC6A12 is highly expressed (Q4) across the life cycle and in the three adult cell types (Figure 5-2).

Both the sponge and the symbiotic bacteria can use GABA via the GABA shunt pathway, which is complete in A. queenslandica and AqS1, but not in AqS2 and AqS3 (Figure 5-1). Both the sponge and AqS1 can degrade GABA to succinate semialdehyde via GABA aminotransferase (ABAT, EC 2.6.1.19 in A. queenslandica, and POP2, EC 2.6.1.96 in AqS1). AqS1 POP2 is expressed (Q3) in the adult holobiont. Ten copies of ABAT are present in A. queenslandica, but eight of them are not expressed or very low expressed (Q1) in few developmental stages across the life cycle and adult cells. The other two copies are highly expressed (Q3 or Q4) across the life cycle and in the three adult cell types (Figure 5-2). Succinate semialdehyde can be converted to succinate via succinate semialdehyde dehydrogenase (GabD, EC 1.2.1.79), which is present and expressed in both the sponge and the three primary symbionts. The enzyme GabD is NAD(P)+-dependent in A. queenslandica and AqS1, and NADP+-dependent in AqS2 and AqS3. A. queenslandica GabD is

98 highly expressed (Q4) across the life cycle and in the three adult cell types (Figure 5-2). ABAT and GabD are the two main enzymes involved in the GABA shunt pathway (Shelp et al. 1999, Feehily et al. 2013). The symbiont AqS1 appears to have the potential to import GABA via a spermidine/putrescine ATP-binding cassette (ABC) transport system, which is a homolog to the GABA uptake system, GtsABCD (transporter classification 3.A.1.11.6) (White et al. 2009, Eitinger et al. 2011). The AqS1 genes encoding this transport system are all expressed in the adult holobiont.

Nitric oxide signalling

The host A. queenslandica can generate and release signalling molecule nitric oxide (NO), which has the potential to react with the three primary symbotic bacteria. The gene NOS, encodes nitric oxide synthase (EC 1.14.13.39), is only present in A. queenslandica and highly expressed (Q4) in the adult holobiont. Nitric oxide synthase can release NO from L-arginine (Song et al. 2020), which is available from the three primary symbionts (Chapter 4). The A. queenslandica NOS is highly expressed (Q4) across the life cycle, with the highest expression at 6-7 hr post-settlement stage, and in the three adult cell types, with the highest expression in pinacocytes (Figure 5-3). In many bacterial systems, heme-nitric oxide/oxygen binding (H-NOX) proteins or nitric oxide sensing protein (NosP) (Hossain and Boon 2017) act as NO sensors (Bacon et al. 2017). However, H-NOX domains and NosP protein are not present in AqS1, AqS2, and AqS3. NO can also induce S- nitrosylation of glyceraldehyde-3-phosphate dehydrogenase (GAPDH, EC 1.2.1.12) via a non- enzymatic, covalent NAD+ modification (reviewed in Butterfield et al. 2010). The gap gene, encoding GAPDH, is present in AqS1, AqS2, and AqS3. All three symbionts express gap in the adult holobiont.

Glutamine Aq AqS1 AqS2 AqS3 H2O Glutaminase NH 3.5.1.2 3 Glutamate Glutamate α-Keto acid synthase NADP+ 1.4.1.13 NAD+ H+ + 1.4.1.- NADH + H Glutamate NH Glutamate CO 3 decarboxylase 2 dehydrogenase Glutamine 4.1.1.15 NADPH + H+ Isocitrate GABA

Pyruvate α-Ketoglutarate α-Ketoglutarate GABA pyruvate GABA transaminase aminotransferase GABA Krebs 2.6.1.96 2.6.1.19 shunt cycle Glutamate Alanine Succinyl-CoA Succinate NAD+ semialdehyde dehydrogenase Succinate Succinate semialdehyde 1.2.1.79 NADH + H+

Figure 5-1. The GABA shunt pathway of the A. queenslandica holobiont. Enzymes are indicated in bold; those specifically associated with the GABA shunt are in bold and highlighted in grey. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is filled with colour, the enzyme is present and expressed in the adult holobiont, otherwise, it’s absent in that species. There are no cases in which a gene is present but not expressed in this pathway. This figure is modified from (Shelp et al. 1999).

100

Stage GAD Developmental Cleavage 4 SLC6A12 Early 2 tile

r ABAT.1 Mid 0 ABAT.2 Late

Qua -2 GabD Larva Metamorphosis Cell_type GAD Adult 1.5 ed z SLC6A12 Quartile 1 ABAT.1 0.5 mali 0 r ABAT.2 1 0 No GabD 2 -0.5 3 -1 wn ing age r o

pole 4 Ring Spot -1.5 Adult Cloud av Br Oscula ent- 1_hr_PS 7_hr_PS T Chamber Cle Late_ Late_spot 12_hr_PS 24_hr_PS 6- Competent Pinacocyte Choanocyte Archaeocyte 11- 23- Precompetent

Figure 5-2. Expression levels of A. queenslandica genes involved in GABA shunt pathway and GABA transporter throughout the life cycle and in adult cell types. The upper heatmap presents these gene expression quartiles, and the bottom one presents the TPM normalised gene expression levels. In the row names, the number after the point represents the number of gene copy, e.g., ABAT.1 and ABAT.2 are two copies of genes encoding GABA aminotransferase (EC 2.6.1.19). GAD, Glutamate decarboxylase (EC 4.1.1.15); SLC6A12, solute carrier family 6 member 12 (GABA transporter); GabD, succinate semialdehyde dehydrogenase (EC 1.2.1.16/1.2.1.79).

101

Stage Cleavage NOS Early DBH.1 Mid Late DBH.2 Larva DBH.3 Metamorphosis

tile Adult

r DBH.4 DBH.5 Quartile 0 Qua DRD 1 TAAR 2 FFAR3.1 3 4 FFAR3.2 Developmental NOS 4 DBH.1 2 DBH.2 0 ed

z DBH.3 -2 DBH.4 Cell_type 1.5 mali DBH.5 r 1 DRD 0.5 No TAAR 0 FFAR3.1 -0.5 FFAR3.2 -1 -1.5 wn ing age r o pole Ring Spot Adult Cloud av Br Oscula ent- 1_hr_PS 7_hr_PS T Chamber Cle Late_ Late_spot 12_hr_PS 24_hr_PS 6- Competent Pinacocyte Choanocyte Archaeocyte 11- 23- Precompetent

Figure 5-3. Expression levels of A. queenslandica genes involved in some potential host-symbiont signalling molecule production and receiving throughout the life cycle and in adult cell types. The upper heatmap presents these gene expression quartiles, and the bottom one presents the TPM normalised gene expression levels. In the row names, the number after the point represents the number of gene copy. NOS, nitric oxide synthase (EC 1.14.13.39); DBH, dopamine beta-monooxygenase (EC 1.14.17.1); DRD, dopamine- like receptor; TAAR, trace amine-associated receptor; FFAR3, free fatty acid receptor 3.

Noradrenaline signalling

Noradrenaline (NA) is one of the primary neurotransmitters in the adrenergic signalling pathway (Goldstein 2010). It can be produced by A. queenslandica by dopamine beta-monooxygenase (EC 1.14.17.1), which can convert dopamine to NA. Dopamine beta-monooxygenase (EC 1.14.17.1) is encoded by the DBH gene. Five copies of DBH are present in A. queenslandica based on KEGG annotations, and expressed (Q2 to Q4) in the adult holobiont. These five DBH genes have different developmental and cell-type expression profiles (Figure 5-3). One DBH copy is not expressed or

102 low expressed at embryonic and larval stages, begins to increase expression at the beginning of settlement and reaches to the highest Q4 at adult; this copy is also highly expressed (Q4) in the three adult cells, with the highest expression in pinacocytes (Figure 5-3). One DBH copy is not expressed or low expressed at embryonic stages, and expressed higher (Q2 or Q3) at larval, metamorphosis and adult stages (Figure 5-3). The other three DBH copies are not expressed or low expressed across the life cycle (Figure 5-3). Taken together, the observations that DBH is expressed throughout the A. queenslandica life cycle and in multiple cell types is consistent with the sponge being able to produce NA.

NA can potentially phosphorylate the bacterial sensor histidine kinase QseC and regulate the bacterial motility through quorum sensing and two-component system (Hughes and Sperandio 2008). In AqS1 qseC, encoding sensor histidine kinase QseC, is expressed (Q2) in the adult holobioint. However, the genes that encode its response regulator QseB, downstream motility genes (flhDC) and Shiga toxin genes (stxAB) are not present in AqS1.

5.4.2 Signalling molecules produced by symbionts

The bacterial genomes reveal that the symbionts have the capacity to synthesise several putative signalling molecules, including dopamine, tyramine, tryptamine, phenethylamine and histamine (Figure 5-4). Dopamine can be converted from tyrosine through oxidation and decarboxylation by polyphenol oxidase (PPO, EC 1.10.3.1) and aromatic amino acid decarboxylase (DDC, EC 4.1.1.28), respectively. Tyramine, tryptamine, phenethylamine, and histamine are derived from tyrosine, tryptophan, phenylalanine, and histidine, which is catalysed by the pyridoxal 5'-phosphate (PLP) dependent decarboxylase (Komori et al. 2012). To classify these holobiont gene functions, conserved protein domains of these genes were predicted using the NCBI Conserved Domain Database (CDD) (Marchler-Bauer et al. 2017).

Two AqS1 and one AqS2 gene are classified into the DOPA decarboxylase family, which belongs to the pyridoxal phosphate (PLP)-dependent aspartate aminotransferase superfamily and contains DDC, histidine decarboxylase (HDC, EC 4.1.1.22), and GAD (EC 4.1.1.15) (Eliot and Kirsch 2004, Liang et al. 2019). To further predict the function of these three genes, conserved catalytic residues were identified by aligning their protein sequences with that of aromatic amino acid decarboxylase and histidine decarboxylase from rat, human, bovine and fruit fly using web-based Clustal Omega (Sievers and Higgins 2014) (Figure 5-5). The conserved catalytic residues Lys303 (the amino acid site is based on human residue sites) and His192 are present in all the three genes, which can bind PLP to form a Schiff-base structure, and decarboxylate the substrates (Liang et al. 2017),

103 respectively. The human HDC S354G mutation decreases affinity for histidine but acquires affinity for L-DOPA (Komori et al. 2012), demonstrating this residue is essential for substrate specificity. Ser354 is present in one AqS1 and one AqS2 gene, and Gly354 was found in one AqS1 gene. These results reveal that the ddc gene, which encodes DDC, exists in AqS1 and AqS2, and the hdc gene exists in AqS1.

Aq AqS1 AqS2 AqS3

NH2 Aromatic amino OH acid decarboxylase

NH2 HO HO 4.1.1.28 Tyramine Tyrosine

Hemocyanin Polyphenol oxidase 1.10.3.1 1.10.3.- O HO NH2 HO OH Aromatic amino acid decarboxylase NH2 HO HO 4.1.1.28 Dopamine L-DOPA

NH2 OH Aromatic amino

NH acid decarboxylase HN 2 HN

Tryptophan 4.1.1.28 Tryptamine

NH2 Aromatic amino OH acid decarboxylase

NH2 4.1.1.28 Phenethylamine Phenylalanine

N N NH2 OH Histidine decarboxylase

NH2 HN 4.1.1.22 HN Histidine Histamine

Figure 5-4. Tyramine, dopamine, tryptamine and histamine synthesis in the A. queenslandica holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is filled with colour, the enzyme is present and expressed in the adult holobiont, otherwise, it’s absent in that species. There are no cases in which a gene is present but not expressed in this pathway.

104

AqS1v2_02414 GAFVTGATMASFTALAAARRRVLLAHG-HDVDRDGLAKAPPITVAVGESAHATVFKALGL 204 sp|P16453|DCHS_RAT GVLQRTVSESTLIALLAARKNKILEMKAHEPNADESSLNARLVAYASDQAHSSVEKAGLI 206 sp|P19113|DCHS_HUMAN GVLQSTVSESTLIALLAARKNKILEMKTSEPDADESCLNARLVAYASDQAHSSVEKAGLI 203 sp|Q5EA83|DCHS_BOVIN GVLQSTVSESTLIALLAARKNKILEMKASEPEADESFLNARLVAYASDQAHSSVEKAGLI 203 sp|Q05733|DCHS_DROME GVLQTTASEATLVCLLAGRTRAIQRFHERHPGYQDAEINARLVAYCSDQAHSSVEKAALI 202 sp|P05031|DDC_DROME GVIQGTASESTLVALLGAKAKKLKEVKELHPEWDEHTILGKLVGYCSDQAHSSVERAGLL 236 sp|P14173|DDC_RAT GVIQGSASEATLVALLAARTKMIRQLQAASPELTQAALMEKLVAYTSDQAHSSVERAGLI 201 sp|P27718|DDC_BOVIN GVIQGTASEATLVALLAARTKVTRHLQAASPELMQAAIMEKLVAYASDQAHSSVEKAGLI 201 sp|P20711|DDC_HUMAN GVIQGSASEATLVALLAARTKVIHRLQAASPELTQAAIMEKLVAYSSDQAHSSVERAGLI 201 AqS1v2_02410 GVIQDTASSATLAAVLTMRERALDGKGNSR----GLAGQPALRIYASNEVHSSIDRALWF 189 AqS2v2_00368 GVIQDSASSGTLAAVLTARERATGWQGNEA----GLAGGPPLRMYYSRHAHASVPKAIML 189 *.: .: .:: .: :. : ..*::::* :

AqS1v2_02414 LGLGRAQVVKVEADS-QGRMRPDALSRL------QGPAIVCAQAGNVNTGAIDPIRRI 255 sp|P16453|DCHS_RAT SLVK---IKFLPVDD-NFSLRGEALQKAIEEDKQQGLVPVFVCATLGTTGVCAFDKLSEL 262 sp|P19113|DCHS_HUMAN SLVK---MKFLPVDD-NFSLRGEALQKAIEEDKQRGLVPVFVCATLGTTGVCAFDCLSEL 259 sp|Q5EA83|DCHS_BOVIN SLVK---MKFLPVDE-NFSLRGEALQKAIKEDRERGLVPIFVCATLGTTGVCAFDCLSEL 259 sp|Q05733|DCHS_DROME GLVR---MRYIEADD-DLAMRGKLLREAIEDDIKQGLVPFWVCATLGTTGSCSFDNLEEI 258 sp|P05031|DDC_DROME GGVK---LRSVQSE--NHRMRGAALEKAIEQDVAEGLIPFYAVVTLGTTNSCAFDYLDEC 291 sp|P14173|DDC_RAT GGVK---IKAIPSDG-NYSMRAAALREALERDKAAGLIPFFVVVTLGTTSCCSFDNLLEV 257 sp|P27718|DDC_BOVIN GGVR---LKAIPSDG-KFAMRASALQEALERDKAAGLIPFFVVATLGTTSCCSFDNLLEV 257 sp|P20711|DDC_HUMAN GGVK---LKAIPSDG-NFAMRASALQEALERDKAAGLIPFFMVATLGTTTCCSFDNLLEV 257 AqS1v2_02410 SGIGADNLVRIPTAGPMRGMILERLRDAIAADRAAGFLPAGIVAAVGGTSTGACDDIAAV 249 AqS2v2_00368 AGLGRANAVAIDLDA-DGAMDAGALERAIEADRAAGMKPAGVVATVGATSTGDADGLAAT 248 : : : * * . *. *: AqS1v2_02414 CAHVPAADIWVHVDGAFGLWVRALRSADALDIDPARKAKASMIADQGAGIEDADSWATDA 315 sp|P16453|DCHS_RAT GPICAREGLWLHVDAAYAGTAF------LRP------ELRGFLKGIEYADSFTFNP 306 sp|P19113|DCHS_HUMAN GPICAREGLWLHIDAAYAGTAF------LCP------EFRGFLKGIEYADSFTFNP 303 sp|Q5EA83|DCHS_BOVIN GPICAREGLWLHIDAAYAGTAF------LCP------EFRGFLKGIEYADSFTFNP 303 sp|Q05733|DCHS_DROME GIVCAEHHLWLHVDAAYAGSAF------ICP------EFRTWLRGIERADSIAFNP 302 sp|P05031|DDC_DROME GPVGNKHNLWIHVDAAYAGSAF------ICP------EYRHLMKGIESADSFNFNP 335 sp|P14173|DDC_RAT GPICNQEGVWLHIDAAYAGSAF------ICP------EFRYLLNGVEFADSFNFNP 301 sp|P27718|DDC_BOVIN GPICHEEGLWLHVDAAYAGSAF------ICP------EFRHLLNGVEFADSFNFNP 301 sp|P20711|DDC_HUMAN GPICNKEDIWLHVDAAYAGSAF------ICP------EFRHLLNGVEFADSFNFNP 301 AqS1v2_02410 SQVAQEESLYLHVDAAWAGSAM------ICP------EFRSLWHGAEHADSIVLNA 293 AqS2v2_00368 GAVVRRHGLYGHVDAAWAGSAA------LCP------EHRGLLDGLEQWDSYLFNP 292 :: *:*.*:. . :* ** ** :

AqS1v2_02414 HKWLNVPYDCGIALVRDAEALRGAMSVQADYLPQASRG----EP--F-EHTPEASRRMRS 368 sp|P16453|DCHS_RAT SKWMMVHFDCTGFWVKDKYKLQQTFSVNPIYLRHA--N--SGVATDFMHWQIPLSRRFRS 362 sp|P19113|DCHS_HUMAN SKWMMVHFDCTGFWVKDKYKLQQTFSVNPIYLRHA--N--SGVATDFMHWQIPLSRRFRS 359 sp|Q5EA83|DCHS_BOVIN SKWMMVHFDCTGFWVKDKYKLQQTFSVDPVYLRHA--D--SGVATDFMHWQIPLSRRFRS 359 sp|Q05733|DCHS_DROME SKWLMVHFDATALWVRDSTAVHRTFNVEPLYLQHE--N--SGVAVDFMHWQIPLSRRFRA 358 sp|P05031|DDC_DROME HKWMLVNFDCSAMWLKDPSWVVNAFNVDPLYLKHDMQ----GSAPDYRHWQIPLGRRFRA 391 sp|P14173|DDC_RAT HKWLLVNFDCSAMWVKKRTDLTEAFNMDPVYLRHSHQD--SGLITDYRHWQIPLGRRFRS 359 sp|P27718|DDC_BOVIN HKWLLVNFDCSAMWVKKRTDLTGAFRLDPVYLRHSHQD--SGLITDYRHWQLPLGRRFRS 359 sp|P20711|DDC_HUMAN HKWLLVNFDCSAMWVKKRTDLTGAFRLDPTYLKHSHQD--SGLITDYRHWQIPLGRRFRS 359 AqS1v2_02410 HKWLGAQLECSIHLLKDENALRNTLTITREYLGTHGQQSARGELRNLSDLSLQLGRRFRA 353 AqS2v2_00368 HKWLGTNFDLCAHYLKDPAAQIRTFGAQPDYLQTAGVEQ----PADFSSWTAPLGRRFRA 348 **: . : ::. :: ** .**:*:

Figure 5-5. Partial alignment of aromatic amino acid decarboxylase and histidine decarboxylase. AqS1v2_02410 and AqS2v2_00368 are putative aromatic amino acid decarboxylases in AqS1 and AqS2. AqS1v2_02414 is a putative histidine decarboxylase in AqS1. Rattus norvegicus (sp|P14173|DDC_RAT, sp|P16453|DCHS_RAT), Homo sapiens (sp|P20711|DDC_HUMAN, sp|P19113|DCHS_HUMAN), Bos Taurus (sp|P27718|DDC_BOVIN, sp|Q5EA83|DCHS_BOVIN) and Drosophila melanogaster (sp|P05031|DDC_DROME, sp|Q05733|DCHS_DROME) aromatic amino acid decarboxylase and histidine decarboxylase are used in this analysis. The colour triangles indicate catalytic residues, i.e., red: conserved PLP binding sites, blue: decarboxylation residue, green: substrate specificity, black: PLP binding sites.

105

The host sponge and its three primary symbionts can catalyse the o-hydroxylation of tyrosine to L- DOPA, which is the intermediate of dopamine biosynthesis, but different enzymes appear to be used by the host and symbionts to complete this step (Figure 5-4). The three symbiotic bacteria possess the yfiH gene, which encodes the copper-containing protein PPO (Figure 5-6) that converts tyrosine to L-DOPA (Araji et al. 2014). This conversion can also be catalysed by tyrosinase (EC 1.14.18.1) or tyrosine hydroxylase (EC 1.14.16.2), which is not present in A. queenslandica. Nevertheless, metalloprotein hemocyanin is homologous to PPO, and can potentially catalyse the o- hydroxylation of monophenols (Adachi et al. 2005, Morioka et al. 2006). Hemocyanin contains two copper atoms, which are coupled with six conserved histidine residues (Hazes et al. 1993). Six A. queenslandica genes contain the pfam00372 (hemocyanin, copper-containing domain), or pfam03723 (Hemocyanin, ig-like domain), and two of them have the six conserved histidine residues (Figure 5-7). Together, these results reveal that the host sponge and three primary symbionts can hydroxylate monophenols.

AqS3v2_00910 MPTETMTEATEAWVERAGLRWLEPKWNV-PGVCAVSLSRAGGSSAGAWAGLNLSDAVGDD 59 AqS2v2_00042 ------MWGFELRKPAWEPPPGVEARLSTRAGGVSDGRYGSCNLAEHVGDD 45 AqS1v2_00062 ------MSLDRFAPLEIIEARWPAPPAVRTLSTTRLGGTSQGAWSSLNLGLGSGDD 50 sp|P33644|POLOX_ECOLI ------MSKLIVPQWPQPKGVAACSSTRIGGVSLPPYDSLNLGAHCGDN 43 .* .* ::*** *:. **. **:

AqS3v2_00910 PEAVEVNRSRVRDVLGV-QSPDWIRQRHGIAVARAGSGPVIADAL--LTQRSRASRTPCA 116 AqS2v2_00042 AGAVGENRMRFAAALPGEPAVAWIRQEHGAEVLEAAAKLTGQEEPACDGIWTARPGQACV 105 AqS1v2_00062 PERVMLNRERLHRFLDLPVEALWLDQVHGNEVVFAGDILPGSKAPSADACIARPGDPPCA 110 sp|P33644|POLOX_ECOLI PDHVEENRKRLFAAGNLPSKPVWLEQVHGKDVLKLTGEPYAS--KRADASYSNTPGTVCA 101 ****. *: * ** * : *.

AqS3v2_00910 ILSADCLPVVLAARDGRAVGIAHAGWRGLLAGVVPNLLRALDLPSGELSAWLAPCIGPEA 176 AqS2v2_00042 VLTADCVPLLLAAADGSQVAAVHAGWRGLAAGIVAKACAAFG--GKEIAAYIGPCISAAN 163 AqS1v2_00062 VQSADCLPVVLCDEDASVVAIAHAGWRGLAKGVISACIEAMGCAPSRLSAWMGPAIGPAA 170 sp|P33644|POLOX_ECOLI VMTADCLPVLFCNRAGTEVAAAHAGWRGLCAGVLEETVSCFADNPENILAWLGPAIGPRA 161 : :***:*:::. .*..******* *:: .: .: *::.*.*.

AqS3v2_00910 FWVGPEVRRAYLMLWGEMAEGEFVACHVDRWQMNLAGLAERQLRLLGVEHILRGSHCTHR 236 AqS2v2_00042 YVVKEDVHERLRAAGG--EDG--LRRDDAGWHADLTAIARGQLRAAGVERIAVEGVCTFG 219 AqS1v2_00062 YEVGEDVRKAFAEPGL--DNA-FRPAGSGKWRANLFAIARHRMQSAGVHRIYGGDLCTFA 227 sp|P33644|POLOX_ECOLI FEVGGEVREAFMAVDA--KASAAFIQHGDKYLADIYQLARQRLANVGVEQIFGGDRCTYT 219 :* :*:. .::::*.:: **.:* . **.

AqS3v2_00910 DADTFYSYRREQ-TSGRQATLAWLE------260 AqS2v2_00042 TPRDFYSARRDGQASGRFATAVWRTA------245 AqS1v2_00062 DSRRFFSHRRDKGITGRMATLVWIDPRADRPSRPCGG 264 sp|P33644|POLOX_ECOLI ENETFFSYRRDK-TTGRMASFIWLI------243 *:* **: :** *: *

Figure 5-6. Alignment of putative polyphenol oxidase sequences of the three primary symbiotic bacteria with the Escherichia coli polyphenol oxidase. The copper-binding residues are indicated with black triangles.

106

Aqu3.1.10572_001 VI--VGPTLPD-----SNPTTADTLRAIETLSYWREDYDLNDHHYHWHLVYPWSGISTIK 66 Aqu3.1.15800_001 VTPTIPSTFPE-----LNPTTADTLKAIETLSYWREDYDLNDHHYHWHLVYPWSGISTIK 349 sp|P80888|HCY_PALVU M---THKE-----GTFNMSFTGTQKNREQRVAYFGQDIGMNIHHVTWHMDFPFWW----D 206 sp|P04253|HCY2_LIMPO H---VRPEFDESPILVDVQDTGNILDPEYRLAYYREDVGINAHHWHWHLVYPSTW----N 185 *. ::*: :* .:* ** **: :* .

Aqu3.1.10572_001 DKKIHRTIDRQGELFLYMHSQMVAQYNAERLSWSDIGLVDSWSYDQVVEPSYTPPPGLRD 126 Aqu3.1.15800_001 EEKIHRTIDRQGELFLYMHSQMLARYNAERLSWSDIGLVDPWSYDEVVEPSYTPPPGLRD 409 sp|P80888|HCY_PALVU DS-YGYHLDRKGELFFWVHHQLTARFDAERFSNW-MDPVDELHWDDIIHEGFAPHASYKY 264 sp|P04253|HCY2_LIMPO PKYFGKKKDRKGELFYYMHQQMCARYDCERLSNG-MHRMLPFNNFDEPLAGYAPHLTHVA 244 . **:**** ::* *: *:::.**:* :: : .::*

Aqu3.1.10572_001 E---YGARPPLQGWLEQHSPYMTEKLAT-VSKDTMIFWRNNINKGIIQGYFYTKKENGDQ 182 Aqu3.1.15800_001 E---YGARPPLQGWLEEHSPFMPEKLAT-VSKDTMIFWRNNINKGIIQGYFYTKKENGDQ 465 sp|P80888|HCY_PALVU GGE-FPTRPDNTHF------KNVDGVARVRDMEITENRIRDAIAHGYITATDG---- 310 sp|P04253|HCY2_LIMPO SGKYYSPRPDGLKL------RDLGDI-EISEMVRMRERILDSIHLGYVISEDG---- 290 :** ..: * .:.* ..* **. : .

Aqu3.1.10572_001 GKFHLTEDDAMNWVGIIVEAEAHQLQEVSPGSKEFIDSDLYGTLHNLGHDKFGEIGYQTY 242 Aqu3.1.15800_001 GKFNLTEDNAMNWVGIIVEAEAHQLQEVSPGSNEFIDSDLYGTLHNLGHDKFGEIGYQTY 525 sp|P80888|HCY_PALVU HTIDIRQPNGIELLGDIIES------SMYSSNPHYPGSLHNTAHGMLGRQGDPHG 359 sp|P04253|HCY2_LIMPO SHKTLDELHGTDILGALVES------SYESVNHEYYGNLHNWGHVTMARIHDPDG 339 ::..::*::*: * :. *.*** .* :..

Aqu3.1.10572_001 MSNKNRWGVMGFTSVAVRDPVFWNWHRHIDDFCQSIINKYKQHALKESAPPHVKLTGVQI 302 Aqu3.1.15800_001 MSNKNRWGVMGFTSVAVRDPVFWIWHRHIDDFRQSIVKKYKQHALKESAPPHVKLTGVQI 585 sp|P80888|HCY_PALVU KF-NMPPGVMEHFETATRDPSFFRLHKYMDNIFKEHTDSFPPYTHEDLEFPGVSVDNIAI 418 sp|P04253|HCY2_LIMPO RF-HEEPGVMSDTSTSLRDPIFYNWHRFIDNIFHEYKNTLKPYDHDVLNFPDIQVQDVTL 398 : *** ..: *** *: *:.:*:: :. .. :. *:.:.::

Figure 5-7. Alignment of partial A. queenslandica putative hemocyanins with that of two crustaceans. Palinurus vulgaris (sp|P80888|HCY_PALVU) and Limulus polyphemus (sp|P04253|HCY2_LIMPO) sequences of hemocyanins are used for the alignment. The copper-binding residue sites are indicated with black triangles.

The holobiont transcriptome data reveal that symbiotic bacteria have the potential to release these signalling molecules. AqS1 and AqS2 can convert tyrosine to dopamine through oxidation and decarboxylation by PPO and DDC, which are encoded by yfiH and ddc genes. AqS1 and AqS2 yfiH genes are lowly expressed (Q2 and Q1, respectively), while AqS1 and AqS2 ddc genes are more highly expressed (Q3 and Q2, respectively). These gene expressions suggest that the dopamine synthesis pathway is active in AqS1 and AqS2. DDC can also catalyse tyrosine and tryptophan to signalling molecules tyramine and tryptamine, respectively. Symbiont AqS1 can produce histamine with histidine decarboxylase (EC 4.1.1.22), which is encoded by hdc and expressed (Q3) by AqS1 in the adult holobiont.

The host sponge has the potential to receive signals produced by the symbionts via its rhodopsin GPCRs. Specifically, the A. queenslandica genome encodes a dopamine-like receptor (DRD) and trace amine-associated receptor (TAAR) based on KEGG annotations (Kanehisa et al. 2016). DRD has been reported in A. queenslandica previously (Srivastava et al. 2010, Krishnan et al. 2014), but not confirmed when conserved domain and structural, functional and phylogenetic information were

107 considered (Wong et al. 2019). As the sponge rhodopsin GPCRs have diverged considerably from those present in eumetazoans (Krishnan et al. 2014), the putative A. queenslandica DRD sequence may have diverged uniquely from other metazoans. TAAR is further supported by a BLAST search of the NCBI Swissprot database (Camacho et al. 2009), which is best aligned to the TAAR sequences from rat, mouse and human. Besides the affinity to trace amines (i.e., phenethylamine, tyramine, and tryptamine), TAAR can be activated by dopamine (Miller 2011, Grandy et al. 2016). The presence of putative DRD and TAAR genes in A. queenslandica is consistent with the host sponge being able to take up dopamine and trace amines released by the symbionts.

Analysis of developmental and cell-type transcriptomes reveals DRD and TAAR are widely expressed in A. queenslandica (Figure 5-3). DRD is highly expressed in the oscula juvenile (Q3) and less so in the larva and metamorphosis juvenile (Q2), and not detected in early development. In adult cells, this gene is expressed in archaeocytes and choanocytes, Q2 and Q3 respectively, but not in pinacocytes. TAAR is expressed in the adult holobiont and variably expressed during development, with highest expression (Q3) at cleavage embryonic stage. In the adult cells, it is highest expressed in archaeocytes (Q3). These sponge GPCR gene expression profiles reveal the ability of A. queenslandica to receive the symbiotic signalling varies across development and cell types.

The A. queenslandica holobiont can potentially produce some SCFAs. Acetate, propionate, and butyrate are the three most common SCFAs. Acetate and propionate can be synthesised by A. queenslandica and the three symbionts using acetyl-CoA synthetase (EC 6.2.1.1) from acetyl-CoA and propanoyl-CoA. Acetyl-CoA synthetase is encoded by the bacterial acsA gene and the sponge ACSS gene. Seven copies ACSS gene are present in the sponge genome, and six of which are expressed in the adult holobiont. Two copies acsA are in AqS1, and one copy acsA is in AqS2 and AqS3. All of the symbiotic acsA are highly expressed (Q3 or Q4) in the adult holobiont. Enzymes that could convert butanoyl-CoA to butanoate are not present in the genomes of A. queenslandica, AqS1, AqS2, and AqS3.

These SCFAs are signal molecules that bind the GPCR, free fatty acid receptor 3 (FFAR3) (Soto et al. 2014). Two copies FFAR3 are present in the A. queenslandica genome and expressed in the adult holobiont. One copy of FFAR3 is not expressed in the three cell types and lowly expressed (Q1 or Q2) across the life cycle (Figure 5-3). The other copy FFAR3 is transiently expressed (Q1) in the late embryo to precompetent larval stages, and in adult choanocytes (Figure 5-3). The expressions of these GPCR genes suggest that SCFA acetate and propionate may act as inter- kingdom signals between the sponge and the three symbiotic bacteria.

108

5.4.3 A dopamine receptor assay in A. queenslandica

Based on sequence analysis, it is not clear if the bioinformatically- predicted dopamine receptor is functional in A. queenslandica. Based on the A. queenslandica holobiont genome and transcriptome, genes encoding polyphenol oxidase (PPO, EC 1.10.3.1) and aromatic amino acid decarboxylase (DDC, EC 4.1.1.28) are active in AqS1 and AqS2 (details presented in the above results). These two enzymes can catalyse tyrosine into dopamine (Hadjiconstantinou and Neff 2008, Araji et al. 2014). Meanwhile, the gene encoding a dopamine receptor is expressed in the host A. queenslandica. These in silico results suggest the potential for the symbiotic bacteria to communicate with the host sponge through dopamine signalling. However, the A. queenslandica dopamine receptor sequence is divergent from that of other basal metazoan dopamine receptors (Krishnan et al. 2014).

To determine if the putative dopamine receptor in A. queenslandica is functional, I assessed if larval phototactic swimming behaviour changes when exposed to a dopamine receptor agonist (DRA; 10-5 M rotigotine hydrochloride) and antagonist (DRAA; 10-5 M flupenthixol dihydrochloride) in a unidirectional light chamber (Leys and Degnan 2001, Say and Degnan 2020, Wong 2020). In normal larval negative phototaxis in FSW, there is a rapid migration of larvae to the dark side of the chamber, resulting in a decreasing number of larvae in the brightest quartile Q1 and an increasing number of larvae in the darkest quartile Q4 over 1 min. The average larval percentage in Q1 dropped from 100% to 13% within 10 sec, which varied between 9% and 14% afterwards (Figure 5-8 and Table 5-1). This percentage in Q4 reached 76% at 25 sec, which varied between 71% and 79% afterwards (Figure 5-8 and Table 5-1).

The DRA has a significant inhibitory effect on normal larval negative phototaxis, with a lower proportion of larvae swimming to the dark side of the chamber (glmer Q1 p-value = 0.00556 and Q4 p-value = 0.00502) (Figure 5-9). In contrast, the DRAA increased larval negative phototaxis, although not significantly (glmer Q1 p-value = 0.0584 and Q4 p-value = 0.377) (Figure 5-9). The average larval percentage in Q1 dropped dramatically from 100% to 4% within 10 seconds and stabilised at 4% afterwards, and the percentage in Q4 soared to 86% within 25 seconds, and increased to the highest rate, 96% at 40 seconds and steadied afterwards (Figure 5-8; Table 5-1).

Larvae pre-incubated in the DRAA for 3 min and then transferred to the light side of a chamber containing the DRA (DRAA_DRA) display a more pronounced inhibitory effect compared to DRA, with a lower proportion of larvae swimming to the dark side of the chamber compared to FSW controls (glmer Q1 p-value = 2.44E-05 and Q4 p-value = 8.53E-07) (Figure 5-9). In the DRA and

109

DRAA_DRA treated group, the average larval percentages in Q1 dropped to 37% and 44% within 10 sec, respectively (Table 5-1). These percentages then varied between 29% and 19% in the DRA group and varied between 36% and 30% in the DRAA_DRA group, respectively (Table 5-1). These percentages were much higher than that of the control group (Figure 5-8; Table 5-1). The larval percentage variation of DRA treatment in Q1 is not significantly different from that of DRAA_DRA treatment (glmer p-value = 0.139) (Figure 5-9). The average larval percentages in Q4 increased to 56% and 31% since being introduced into the DRA and DRAA_DRA chamber for 35 seconds, respectively (Table 5-2). In the DRA group, these percentages then slightly varied between 56% and 59% (Table 5-2). In the DRAA_DRA group, these percentages ranged between 31% and 36% afterwards (Table 5-2). These percentages were much lower than those of the control group (Figure 5-8; Table 5-2). The larval percentage variation of DRA treatment in Q4 is also different from that of DRAA_DRA treatment (glmer p-value = 0.0315) (Figure 5-9).

In summary, these larval behaviour results with DRA and DRAA are consistent with the existence of functional dopamine-related receptor in A. queenslandica.

110

FSW DRA DRAA DRAA_DRA

100%

75%

Quartile Q1 50% Q2 Q3 Q4 % larvae in quartile % larvae

25%

0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 Time since introduced (s)

Figure 5-8. Effect of dopamine receptor agonist (DRA) and antagonist (DRAA) on larval phototaxis. The percentage of larvae in each quartile is the average of the seven replicates of 10 larvae each per treatment. Normal negative phototaxis is observed in FSW controls. In DRA and DRAA treatments, larvae are exposed to 10-5 M DRA or DRAA in FSW for 2 or 3 min, respectively before being placed in the assay chamber containing the same solution. In DRAA_DRA, larvae are incubated in 10-5 M DRAA in FSW for 3 min and then transferred to the light side of a chamber containing 10-5 M DRA in FSW.

111

Table 5-1. The average percentage of larvae in quartile 1 (Q1) varying since introduced into the chamber. It is the average larvae number percentage of the seven replicates. FSW was control; DRA and DRAA were exposed to dopamine receptor agonist (DRA) and antagonist (DRAA), respectively; DRAA_DRA was incubated in the DRAA solution for 3 minutes and then transferred to the light side of the DRA chamber.

Time since introduced 0 s 5 s 10 s 15 s 20 s 25 s 30 s 35 s 40 s 45 s 50 s 55 s 60 s FSW 100% 53% 13% 11% 9% 9% 9% 9% 10% 10% 13% 14% 14% DRA 100% 80% 37% 29% 23% 23% 21% 20% 19% 19% 19% 24% 24% DRAA 100% 49% 4% 4% 4% 4% 4% 4% 4% 4% 4% 4% 4% DRAA_DRA 100% 87% 44% 30% 31% 36% 34% 31% 34% 36% 33% 35% 35%

Table 5-2. The average percentage of larvae in quartile 4 (Q4) varying since introduced into the chamber. It is the average larvae number percentage of the seven replicates. FSW was control; DRA and DRAA were exposed to dopamine receptor agonist (DRA) and antagonist (DRAA), respectively; DRAA_DRA was incubated in the DRAA solution for 3 minutes and then transferred to the light side of the DRA chamber.

Time since introduced 0 s 5 s 10 s 15 s 20 s 25 s 30 s 35 s 40 s 45 s 50 s 55 s 60 s FSW 0% 0% 0% 26% 61% 76% 79% 79% 79% 80% 79% 76% 71% DRA 0% 0% 0% 3% 11% 33% 44% 56% 56% 59% 57% 56% 57% DRAA 0% 0% 3% 25% 69% 86% 93% 94% 96% 96% 96% 96% 96% DRAA_DRA 0% 0% 0% 0% 9% 17% 27% 31% 31% 33% 33% 33% 36%

112

DRAA_DRA *** ***

p−value DRAA . *** *** (0, 0.001) (0.001, 0.01) (0.01, 0.05) DRA Q4 (0.05, 0.1) ** *** * (0.1, 1)

FSW ** ***

FSW DRA DRAA DRAA_DRA

Figure 5-9. P-values of the effects of DRA and DRAA on the sponge larval negatively phototactic swimming behaviour. These p-values were generated by fitting the larval number of the Q1 series and the Q4 series with and without the treatment effect to the Poisson generalised linear mixed model, separately.

5.5 Discussion

5.5.1 No evidence for an AHL-QS system in the A. queenslandica holobiont

In this chapter, I focus on the capacity for inter-kingdom signalling in the A. queenslandica holobiont. First, I sought evidence for communication through AHL-QS inter-kingdom quorum signalling as observed in other sponge holobionts (Gardères et al. 2014). The AHL-QS system requires the expression of AHL synthases and cognate signal receptors (Li and Nair 2012). Although multiple AHL-QS systems, such as LuxI/R, LasI/R, LuxM/N, and TswI/R, have been reported in sponge symbionts (Venturi and Subramoni 2009, Britstein et al. 2016), no genes encoding AHL synthases and corresponding receptors are detected in the genomes of A. queenslandica, AqS1, AqS2, and AqS3. The absence of these genes suggests that AHLs are not QS signals in the A. queenslandica holobiont. Sponges from the Mediterranean, the Red Sea (Britstein et al. 2017), and the Antarctic (Mangano et al. 2018), including an Amphimedon species(Britstein et al. 2017), also appear to lack AHL-QS signalling systems, lending support to the conclusion that

113 this system is not ubiquitous in sponge holobionts.

5.5.2 Small signalling molecules may mediate interactions between partners in the A. queenslandica holobiont

Analysis of A. queenslandica holobiont genomes and transcriptomes reveals that this sponge and its three primary symbiotic bacteria can produce different types of small signalling molecules, including amino acids (glutamate (Chapter 4) and GABA), biogenic amines (catecholamine [adrenaline, noradrenaline and dopamine] and trace amines [phenethylamine, tyramine and tryptamine]), and SCFAs (acetate and propionate). These signalling molecules, which are used as neurotransmitters in neural animals, may be received by A. queenslandica GPCRs, including mGluRs, DRD, TAAR, and FFAR3. Thus, A. queenslandica and its symbiotic bacteria may dialogue through host-GPCR – symbiont-derived ligand interactions.

Amino acid signalling

The A. queenslandica symbionts potentially can use GABA synthesised and released by the sponge host. GABA is involved in several eukaryotic signal pathways and in the regulation of bacterial quorum sensing (Chevrot et al. 2006, Chalifoux and Carter 2011, Strandwitz et al. 2019). A. queenslandica can synthesise GABA by GAD, which is highly expressed during most of the development, particularly during metamorphosis. A. queenslandica GAD expression increases to Q4 at 6-7 hr post-settlement and reaches the highest at 23-24 hr post-settlement (Figure 5-2). GABA can affect larval settlement and metamorphosis in a range of marine invertebrates (Searcy-Bernal and Anguiano-Beltran 1998, Takami et al. 2002, Garcia-Lavandeira et al. 2005, Sun et al. 2014, Biscocho et al. 2018).

The GABA-stimulated bacterial attKLM regulation system (Chevrot et al. 2006) appears not to be present in the three primary symbionts of A. queenslandica. However, AqS1 appears to be able to import GABA and convert it to succinate semialdehyde and succinate through the GABA shunt pathway. Succinate is a crucial metabolite involved in several metabolic pathways, including TCA, oxidative phosphorylation, and amino acid metabolism (Tretter et al. 2016). GABA degradation is common in bacteria, acting as a nutrient source (Feehily and Karatzas 2013, Strandwitz et al. 2019). These results suggest that GABA may act as a QS signal in the A. queenslandica holobiont, with AqS1 potentially able to regulate the GABA level and thus influence sponge behaviours, such as larvae settlement and metamorphosis (Searcy-Bernal and Anguiano-Beltran 1998, Takami et al. 2002, Garcia-Lavandeira et al. 2005, Sun et al. 2014, Biscocho et al. 2018).

114

Metabolites of amino acids by symbiotic bacteria may regulate the host sponge signalling. The genomic and transcriptomic data of A. queenslandica holobiont reveal that A. queenslandica, AqS1, AqS2 and AqS3 can biosynthesise glutamate from 2-oxoglutarate and ammonia, and convert glutamate to glutamine (Chapter 4); AqS1 can degrade sponge produced GABA (mentioned above). Amino acid transporters are also present in A. queenslandica (solute carrier family 6 and 7), and in both AqS1 and AqS2 (general L-amino acid transport system or putative spermidine/putrescine ATP-binding cassette (ABC) transport system). These amino acid transporters give sponge and symbionts the potential to translocate glutamate and GABA within the holobiont. Glutamate and GABA could function as signalling molecules and bind to A. queenslandica metabotropic glutamate and GABA receptors, respectively. These activated GPCRs can, in turn, regulate the cellular sponge physiology through multiple signalling pathways, such as cAMP signalling pathway (Chalifoux and Carter 2011). Thus, the symbiotic bacteria have the potential to regulate the sponge physiology by altering the levels of glutamate and GABA within the holobiont.

Glutamate or GABA are ubiquitous communication molecules in pre-bilaterian metazoans (Krishnan and Schiöth 2015), e.g. sponges (Leys et al. 2019) and ctenophores (Moroz et al. 2014), and various bilaterians, e.g. fruit flies (Liebeskind et al. 2017) and humans (Mazzoli and Pessione 2016). Several gut bacteria produce glutamate or GABA (Mazzoli and Pessione 2016, Pokusaeva et al. 2017) that regulate the host emotional behaviour via the central nervous system (Bravo et al. 2011, Strandwitz et al. 2019). Glutamate and GABA might, therefore, be ancient and conserved inter-kingdom signalling molecules between animal and symbiotic bacteria.

Biogenic amines

The symbiotic bacteria can synthesise catecholamines (dopamine) and trace amines (phenethylamine, tyramine, and tryptamine) in the A. queenslandica holobiont. Dopamine can be produced by AqS1, and AqS2 from tyrosine via the enzymes PPO and DDC but not by A. queenslandica. It is the same for phenethylamine, tyramine and tryptamine. A complete dopamine biosynthesis pathway is also missing in other demosponges (Riesgo et al. 2014, Francis et al. 2017). Notably, the absence of DDC in demosponges (Riesgo et al. 2014, Francis et al. 2017) suggests they cannot produce dopamine, phenethylamine, tyramine and tryptamine. Dopamine isolated from demosponge Neopetrosia exigua holobiont is considered to be provided by an endosymbiotic Synechococcus-like cyanobacterium (Liu et al. 2004), suggesting that dopamine, phenethylamine, tyramine and tryptamine are probably produced by symbiotic bacteria in demosponge holobionts.

The production of catecholamine (dopamine) and trace amines (phenethylamine, tyramine, and

115 tryptamine) by symbiotic bacteria in the A. queenslandica holobiont may allow for inter-kingdom signalling. As the first test of this proposition, I assessed A. queenslandica larval phototaxis in the presence of a dopamine receptor agonist (DRA) and antagonist (DRAA). In this experiment, the DRA significantly weakened normal negative phototaxis, while larvae in the presence of the DRAA tended to be more negatively reactive to light. DRA might affect the sponge larval phototaxis through modulating intracellular calcium levels, as previous studies have shown light illumination results in an increase in intracellular calcium levels and alter the movement of posterior cilia which act as a rudder (Say and Degnan 2020, Wong 2020). DRA, which has a high affinity for both D1 and D2 dopamine receptors, may decrease intracellular calcium levels, as shown in other animals (Beaulieu and Gainetdinov 2011).

In addition to dopamine receptors, TAAR is also predicted to be present in A. queenslandica. The existence of DRD and TAAR GPCRs in A. queenslandica suggests that this sponge can receive dopamine and trace amines released by the symbionts. Our understanding of the functional roles of symbiotic-released dopamine and trace amines in sponges is limited. One recent study has identified a natural monoamine alkaloid, 3-bromo-4-methoxyphenethylamine, as QS inhibitory molecule in the sponge Sarcotragus spinosulus (Saurav et al. 2020). Various studies have reported the ability of gut microbiota to synthesise neurotransmitters, including dopamine, serotonin and histamine, that can bind directly to host receptors and affect host physiology, behaviour and health (Collins et al. 2012, Johnson and Foster 2018, Martin et al. 2018). For example, substantial levels of free dopamine are identified in the gut lumen of specific pathogen-free mice and can regulate colonic water absorption (Asano et al. 2012). The gut microbiota can alter the turnover of noradrenaline, dopamine and serotonin, and modulate the host brain development and behaviour (Diaz Heijtz et al. 2011).

The potential role of catecholamines (noradrenaline and adrenaline) in signalling between the host sponge and the symbiotic bacteria is unclear. A. queenslandica can produce NA, which may be sensed by AqS1 (QseC) (Clarke et al. 2006). Thus, NA may be cross-kingdom signalling between the sponge and bacterial symbionts. However, QseC-mediated two-component QS system (Rooks et al. 2017) is not complete in AqS1. Hologenomic and transcriptomic information suggest A. queenslandica cannot generate adrenaline. The same incomplete adrenergic signalling pathway has been reported in eight sponges; adrenergic receptors are present in some of these sponges while DBH is present in all (Riesgo et al. 2014). Meanwhile, I do not find sufficient support for the presence of adrenergic receptors in A. queenslandica, as previously reported (Krishnan et al. 2014). Several sponge Rhodopsin GPCRs align best to beta-adrenergic receptors when searching against Swissprot database (Camacho et al. 2009), albeit with low BLAST score (between 50-70). This is

116 similar to the BLAST score to other types Rhodopsin GPCRs, e.g. DRD, serotonin receptor. The main reason for these ambiguous BLAST search results is the considerable divergence between the sponge Rhodopsin GPCRs and that found in other basal metazoans (Krishnan et al. 2014).

Thus, more experiments are required to verify the existence and function of some Rhodopsin GPCRs, e.g., adrenergic and serotonin receptors (Wong et al. 2019), in A. queenslandica, and confirm the role of noradrenaline and adrenaline in the sponge-bacterial signalling. Adrenaline treatment in sponge Tethya wilhelma can decrease its frequency of the endogenous contraction rhythm (Ellwanger and Nickel 2006), which suggest adrenaline signalling in the sponge. In humans, NA is an important molecule for cardiac (Lymperopoulos et al. 2013) and brain (Brodie and Shore 1957) function and is also reported as stress signalling between host and gut microbiota (Hughes and Sperandio 2008). NA produced by humans (Collins et al. 2012) can activate the expression of pathogenic enterohemorrhagic Escherichia coli virulence genes (Sperandio et al. 2003) and increase its growth rate (Freestone et al. 2002). Gut microbiota can modulate the level of host NA to interact with the host immune and nervous systems (Rea et al. 2016, Strandwitz 2018). These findings support the hypothesis that NA might be a host-symbiont inter-kingdom signal in diverse animal holobionts.

SCFAs

The SCFAs (acetate and propionate) may also be inter-kingdom signalling molecules. A. queenslandica and its three primary symbionts can produce acetate and propionate. These SCFAs can act as signalling molecules that bind to a GPCR, FFAR3 (Soto et al. 2014), which is present in A. queenslandica. This SCFAs metabolism and signalling pathway raise the possibility that symbiotic bacterial-derived SCFAs can act as signals between sponge and bacteria. The most widely studied SCFAs are the end products of anaerobic fermentation by microbiota in the colon (den Besten et al. 2013). These bacterial released SCFAs can directly bind to FFAR2 and FFAR3 expressed on enteroendocrine cells and induce the release of gut hormones such as GABA and serotonin (5-HT), and influence brain physiology and behaviour (Collins et al. 2012, Silva et al. 2020).

5.6 Conclusion

In this chapter, I analysed the potential sponge-bacteria inter-kingdom signalling within the A. queenslandica holobiont. Genomic and transcriptomic data suggest that host GPCR- bacterial derived neurotransmitter interactions can exist in the A. queenslandica holobiont, instead of the AHL-QS system, which is present in some sponge holobionts (Britstein et al. 2017, Mangano et al.

117

2018). I have shown that symbiotic bacteria can potentially produce or utilise small signalling molecules, including amino acids (glutamate and GABA), biogenic amines (dopamine, phenethylamine, tyramine and tryptamine), and SCFAs (acetate, and propionate), which are known to act as neurotransmitters in neural animals (Collins et al. 2012). These small signalling molecules may bind to the A. queenslandica GPCRs and regulate the hosts physiological and developmental states. These potential signalling systems between the sponge and bacteria expanded our understanding of sponge-bacteria symbiosis and may provide insights into the evolution of the nervous system (Klimovich and Bosch 2018, Chen et al. 2019, Colosimo et al. 2019).

118

Chapter 6 General Discussion

6.1 Overview of project objectives

Genomic and transcriptomic studies in humans and many other animals over the last decade have established the consensus that symbiotic microbes can exert critical influences on the host metabolism, physiology and health (McFall-Ngai et al. 2013, Simon et al. 2019). Diverse microbial communities reside as symbionts in sponges, one of the earliest-diverging of the extant metazoan lineages (Thomas et al. 2016, Webster and Thomas 2016, O’Brien et al. 2020, Steinert et al. 2020), and participate in the sponge (holobiont) nutrient cycling in the marine environment (de Goeij et al. 2008, van Duyl et al. 2008, Rix et al. 2017, Gantt et al. 2019, Rix et al. 2020). However, our knowledge about the sponge-microbe crosstalk is still limited. For this thesis, I sought to explore the molecular mechanisms underlying the sponge-microbe metabolic interactions and signalling communications within the coral reef demosponge Amphimedon queenslandica holobiont through holo-genome and -transcriptome analyses.

6.1.1 Genetic resources of A. queenslandica holobiont

Complete and reliable holobiont genomic and transcriptomic data are valuable genetic resources for deciphering the molecular interactions between the sponge and the symbionts. The sponge A. queenslandica houses a low complexity and abundance microbiota dominated by just three proteobacterial symbionts, AqS1, AqS2, and AqS3 throughout most of the life cycle (Fieth et al. 2016). To improve the holobiont genome assemblies (Chapter 2), long-range linked reads were generated from Chicago libraries and assembled by HiRise workflow (Putnam et al. 2016) based on the previously published genomes of A. queenslandica (Srivastava et al. 2010) and the three primary symbionts, AqS1, AqS2, and AqS3 (Gauthier et al. 2016). With these long-range linked reads, the scaffolds N50 length of A. queenslandica increased almost eight-fold, from 120 kbps to 950 kbps. The scaffold N50 of the three primary symbiotic bacterial assemblies – AqS1, AqS2, and AqS3 – were also increased to 103 kbps, 148 kbps, and 90 kbps, respectively. This reassembly of the A. queenslandica hologenome allows for a more complete view of gene function in host and symbionts, and interactions between the animal and bacterial partners. Remarkably, I find that the three bacterial symbionts contribute 45.2% of the total number of KEGG-annotated metabolic genes, despite the three bacterial symbionts contributing only 23.1 % of all KEGG-annotated genes in the hologenome. This disproportionately high contribution from bacterial symbionts suggests the symbioses serve to extend the metabolic phenotype of the holobiont. The symbionts contribute

119 enormously to the total membrane transport capabilities of the holobiont, further indicating the potential for exchange of diverse molecules between host and symbionts, and between symbionts.

The most efficient way to acquire holobiont transcriptomes is to perform RNA-Seq on both host and symbionts simultaneously. Various strategies have been employed to capture both the host and symbiont mRNAs (Fiore et al. 2015, Upadhyay et al. 2015, Germer et al. 2017, Grote et al. 2017, Moitinho-Silva et al. 2017). However, the efficacy of these approaches is not yet well-defined when the host is a eukaryote, and the symbionts are prokaryotes; the biggest challenge is that, in the latter, mRNAs generally have a much shorter half-life of only a few minutes on average (Selinger et al. 2003). Thus, I compared the transcriptomic data generated from poly(A) captured mRNA-Seq (PolyA-mRNA-Seq) and ribosomal RNA depleted RNA-Seq (rRNA-depleted-RNA-Seq), from bacterial enriched adult tissues of A. queenslandica holobiont, focusing on the transcriptome depth and coverage (Chapter 3). For the host sponge, no significant difference was found in transcriptomes generated from the two different mRNA capture methods. However, the rRNA- depleted-RNA-Seq performed much better than the PolyA-mRNA-Seq for the symbiont transcriptomes. This comparison demonstrated that RNA-Seq by ribosomal RNA depletion clearly is the most effective method to achieve complete and reliable holobiont transcriptome. Once I confirmed the dual RNA-Seq method, more A.queenslandica holobiont transcriptome data were generated by ribosomal RNA depleted RNA-Seq to investigate sponge-bacteria crosstalk (Chapter 4).

6.1.2 Collaboration between host and symbionts in nutrient assimilation

Sponges are functionally essential components of the marine ecosystem because of their ability to retain and recycle pelagic nutrients toward other benthos (Bell 2008, Folkers and Rombouts 2020). Isotope tracing experiments have reported that both the host sponge cells and symbiotic microbes are actively integrating DOM (de Goeij et al. 2008, Rix et al. 2016, Rix et al. 2017, Rix et al. 2020). Many genomic and transcriptomic analyses have investigated molecular-level metabolic pathways, but only from the perspective of either the host sponges (Riesgo et al. 2014, Francis et al. 2017, Kenny et al. 2020) or the symbiotic microbes (Fiore et al. 2015, Germer et al. 2017, Moitinho-Silva et al. 2017, Weigel and Erwin 2017, Engelberts et al. 2020), but not both together. The genomes and transcriptomes from both host sponge A. queenslandica and its primary symbionts make this holobiont a valuable resource to study potential host-symbiont metabolic crosstalk to assimilate the major marine nutrients, carbon, nitrogen, sulfur and phosphorus (Chapter 4). Meanwhile, the host life cycle (Anavy et al. 2014, Levin et al. 2016) and cell-type transcriptome data (Sogabe et al. 2019) allow a better understanding of the sponge contribution and variation to these potential host-

120 symbionts interactions during at different developmental stages and cell types (Chapter 4).

Glycolysis, Krebs cycle, and PPP are complete and active in both the sponge and symbionts; fatty acid biosynthesis pathways and various sponge essential amino acid biosynthesis pathways are present in the symbionts. These metabolic pathways reveal the sponge and symbiotic bacterial molecular pathways that can convert external carbohydrates into energy or biomass growth materials (nucleotides, fatty acids and amino acids). The symbionts AqS1 and AqS2 may import and use hexoses liberated by host polysaccharide breakdown, indicating the potential of host-symbiont cooperation in assimilating carbon. Specially, during the life cycle, the host carbohydrate assimilation capacity is higher in metamorphosis and adult stages. This higher carbohydrate assimilation capacity during oscula/juvenile and adult stages may correlate with the capacity to pump and filter seawater, which requires an energetic expenditure (Pfannkuchen et al. 2009), and the rapid sponge cell proliferation rate in metamorphosing A. queenslandica (Sogabe et al. 2016).

I found bioinformatic evidence that the host and symbiont collaborate in assimilating dissolved inorganic nitrogen (DIN). Symbiont AqS1 can reduce nitrate to nitrite, which can be subsequently reduced to ammonia by A. queenslandica, with the help of cofactor siroheme produce by symbiont AqS1. The generated ammonia can be assimilated via glutamine or glutamate synthesis by the host sponge and the three symbionts. Specifically, the host glutamine and glutamate synthesis are active across the life cycles and in the three adult cell types, which further indicate the ammonia assimilation capacity of the sponge. The symbiotic bacterial capacity to reduce nitrate is particularly important for the sponge holobiont DIN assimilation, because the ammonia and nitrite concentrations in the A. queenslandica reef flat habitat are much lower than that of nitrate throughout the year (Watson et al. 2017).

The associated symbionts play an important role in A. queenslandica holobiont phosphorous and sulfur assimilation from the water column, which is limited in phosphate (Pi) and sulfide, but rich in sulfate (Watson et al. 2017). Both AqS1 and AqS3 can also accumulate Pi by the formation of polyphosphate. AqS1 (Gauthier et al. 2016) can oxidise environment thiosulfate and elemental sulfur to sulfate (Reviewed in Ghosh and Dam 2009) and also reduce sulfate to sulfide. The sulfide generated by AqS1 can be used to synthesis cysteine by A. queenslandica and the three symbionts, or to produce thiosulfate by A. queenslandica. The sponge hologenomic and transcriptomic data advance our understanding of the sulfur assimilation of A. queenslandica holobiont to a more fine- tuned level. During the life cycle, the sponge activity to generate thiosulfate is most active in adult, especially in archaeocytes, which significantly upregulate genes that control cell proliferation (Sogabe et al. 2019).

121

6.1.3 Host-symbiont metabolic complementation

The A. queenslandica holobiont genomes and transcriptomes have also revealed the potential for metabolic complementation within the sponge system during nutrient assimilation (Chapter 4). The symbiotic bacteria can de novo synthesise – and thus potentially provide the host sponge with – essential amino acids (Song et al. 2020), some vitamins (B1, B2, and B5), and cofactors (folate and siroheme). The capacities of symbiotic microbes to provide the animal host with essential amino acids, vitamins and cofactors have been reported in many other sponges (Fiore et al. 2015, Gauthier et al. 2016, Engelberts et al. 2020) and animal lineages (Zientz et al. 2004, Douglas 2009, Ponnudurai et al. 2017, Feng et al. 2019, Yang et al. 2020). The genomic information of A. queenslandica further reveals the host sponge can provide the three symbionts with the essential amino acid asparagine and is heterotrophic for these symbiont-derived B vitamins and cofactors. Moreover, there is evidence that within-pathway complementation may also exist among the different holobiont partners. For example, the host sponge can provide the symbionts β-alanine to generate vitamin B5.

The symbionts may influence the host development and nutrient assimilation through these metabolic complementations. AqS1, AqS2, and AqS3 are vertically inherited in A. queenslandica and maintain proportional dominance at most developmental stages (Fieth et al. 2016). AqS1 can regulate the holobiont nitrogen assimilation by provide A. queenslandica with siroheme and nitrite (mentioned above). A. queenslandica can reduce nitrite to ammonia. Ammonia is an important chemical cue to initiate larval settlement in some invertebrates (Coon et al. 1990, Hadfield and Paul 2001). The expression level of A. queenslandica nit-6, encoding nitrite reductase, is highest at the beginning of metamorphosis. Thus, it may be that the symbiotic bacterial influence the sponge larval settlement through regulating ammonia levels in the holoboint. On the other hand, the host sponge can import the symbiont-derived amino acid arginine and release nitric oxide (NO). NO can also induce larval metamorphosis of A. queenslandica (Song et al. 2020). All these metabolic complementations indicate the critical roles of the symbionts in the host developmental.

6.1.4 Host-symbiont interkingdom signalling communication

To maintain the animal-microbiome symbiosis, hosts and symbionts need to communicate with each other. One communication method identified in sponge-bacteria symbioses is through quorum sensing (QS) signals, such as N-acyl-homoserine lactones (AHLs) (Britstein et al. 2017, Mangano et al. 2018, Reen et al. 2019), which are the main metabolites for the bacteria to communicate with each other (Rémy et al. 2018). Screening of 15 sponge species from the Mediterranean and the Red

122

Sea detected AHL QS signals in seven species but not in an Amphimedon species (Britstein et al. 2017). I explored the capacity of the A. queenslandica holobiont to produce and communicate through AHL-QS system, but found no evidence of any of the necessary genes in the sponge or in the three primary symbionts, AqS1, AqS2, and AqS3 (Chapter 5).

Interestingly, however, the A. queenslandica holobiont genomic and transcriptomic analyses revealed another potential animal-bacterial interkingdom communication method. Namely, I found genome-encoded evidence of host-GPCR – symbiont-derived ligand interactions within the A. queenslandica system (Chapter 5). Sponges lack a nervous system, but they do have complex signalling pathways that enable them to respond to a range of stimuli (Leys 2015, Leys et al. 2019). Four of the five major families of GPCRs – Glutamate, Rhodopsin, Adhesion and Frizzled – are present in A. queenslandica (Krishnan et al. 2014). The genomic and transcriptomic analyses revealed that the three symbionts can biosynthesise glutamate, and the symbiont AqS1 may uptake and degrade sponge-derived GABA. Glutamate and GABA can function as signalling molecules and bind to A. queenslandica metabotropic glutamate and GABA receptors, respectively. Thus, the symbionts might communicate with the host sponge via controlling the glutamate and GABA level within the holobiont. Moreover, the symbiotic bacteria can produce catecholamine (dopamine), trace amines (phenethylamine, tyramine, tryptamine, and histamine), and SCFAs (acetate, and propionate), which may be received by A. queenslandica GPCRs. I, therefore, hypothesise that the symbiotic bacteria may communicate with the host through these GPCR ligands in the A. queenslandica holobiont.

Inter-kingdom signalling between the sponge host and its symbiotic bacteria can shed light on the evolution of metazoan neural signalling pathways. GPCRs control signal transduction in most animals with or without nervous systems (Krishnan and Schiöth 2015). Host-GPCR – symbiont- derived ligand interactions have been reported in Hydra holobiont, which belongs to the basal phylum Cnidaria characterised by a simple nerve net (Murillo-Rincon et al. 2017, Klimovich and Bosch 2018). In humans, a vast literature reports that symbiotic microbes can communicate with the host through producing GPCR-active metabolites (Cohen et al. 2017, Husted et al. 2017, Chen et al. 2019, Colosimo et al. 2019, Pandey et al. 2019) and ultimately can regulate host biology, including development, immunity and growth (Belkaid and Hand 2014, Gilbert et al. 2018, Proctor et al. 2019, Zheng et al. 2020). My PhD work back-dates animal-bacterial communication via host-GPCR – symbiont-derived ligand before the emergence of the nervous system.

123

6.2 Concluding remarks and recommendations for future study

This thesis represents one of the first studies to analyse the sponge-bacterial crosstalk with genomes of both host and primary symbionts, complemented by different experimental transcriptomes. The A. queenslandica holobiont genomic and transcriptomic analyses have proposed the molecular-level mechanisms of sponge-microbe cooperation in nutrient assimilation, the metabolic interdependence amongst the partners of the holobiont, and the signalling communication between the host and symbionts. The sponge genomic and transcriptomic data facilitate a more comprehensive and fine- tuned level understanding of the host contribution and variation in these potential host-symbiont interactions during the life cycle and different sponge cells. The findings of this thesis provide a framework for future studies to test hypotheses about sponge-bacteria metabolic interactions and signalling regulations. Here, I discuss some potential avenues of future pursuit to help shed light on the crosstalk within the A. queenslandica holobiont that establishes and maintains the sponge- bacteria symbiosis.

Qualitative and quantitative characterisation of the metabolic capacities of both the sponge and symbiotic bacteria, and the metabolic complementation between them, are necessary to experimentally validate the hypotheses generated by the genomic and transcriptomic analyses. I have noted herein that some metabolic pathways are near-complete or partial-complete in the genomes of A. queenslandica, AqS1, AqS2 or AqS3; these include fatty acid, arginine, lysine, methionine, and vitamin B5 biosynthesis pathways, which are difficult to confirm through in silico analysis. (Meta)proteomics and metabolomics are well-developed methods in other holobionts to identify and quantify proteins and metabolites, respectively (Proctor et al. 2019, Yang et al. 2020); they would be invaluable tools to explore protein and metabolite profiles in A. queenslandica. Another method to directly quantify the metabolic capacities and complementation within holobiont system would be stable isotope tracing (SIP), coupling with nanoscale secondary ion mass spectrometry (nanoSIMS) (Rix et al. 2020), liquid chromatography (HPLC) or liquid- chromatography-mass spectrometry (LC-MS) (Song et al. 2020); these techniques have been developed in A. queenslandica (Song et al. 2020, Rix et al. In preparation).

Another important line of inquiry for the future would be to experimentally validate the signalling communication between A. queenslandica and symbiotic bacteria. The first step would be to trace the production and translocation of signalling molecules (Chapter 5) in the sponge and the symbionts, ideally via SIP-MS as mentioned above, and then determine the responses of A. queenslandica to these signalling molecules. The sponge larval phototactic swimming experiment is a quick method to test whether the host can receive the symbiont-derived signalling molecules

124

(Chapter 5), but it cannot identify the specific responding genes. Gene-knockdown techniques, such as RNA interference (RNAi) (Rivera et al. 2011) and the CRISPR/Cas9 gene-editing (Ran et al. 2013), would be more robust methods to detect the gene functions of A. queenslandica GPCRs, especially some sponge Rhodopsin GPCRs, which are highly divergent from those of other basal metazoans (Krishnan et al. 2014). These validation experiments would test the hypothesis that animal-bacterial communicate through host-GPCR – symbiont-derived ligand interactions in a holobiont without a nervous system. Signalling interactions exist between nerve cells of Hydra, the first metazoans that contain neurons, and symbiotic bacteria (Augustin et al. 2017, Murillo-Rincon et al. 2017, Klimovich and Bosch 2018). Research on the signalling communication between the sponge and symbiotic bacteria might provide new insight into the origin and first role of the nervous system.

Exploring metabolite transfer between A. queenslandica and symbiotic bacteria would be another interesting direction for future pursuit. The A. queenslandica holobiont genomic and transcriptomic information have proposed that complex metabolic complementation and signalling communication may exist between the host and its symbionts. Some transporters in the sponge and the three symbiotic bacteria may allow for the exchange of metabolites with small molecular sizes, such as amino acid transporters, sugar transport proteins (Chapter 4). However, it is still largely unknown how these metabolites and signalling molecules exchange between the sponge and three symbionts. Electron micrographs have shown the host sponge can directly obtain the metabolites of symbiotic microbes via phagocytosis (Fieth et al. 2016, Leys et al. 2017). Another method by which host and symbionts may exchange metabolites is via the release of extracellular vesicles (EVs), which has been reported in human (Lee 2019, Macia et al. 2019). EVs are evolutionarily-conserved vesicles secreted by both prokaryotic and eukaryotic cells (van Niel et al. 2018). They can function in the exchange of diverse components – varying from small signalling molecules to larger lipids and proteins – between host and microbes (Lee 2019, Macia et al. 2019). Learning about the component exchange method amongst the different holobiont partners in the A. queenslandica holobiont is helpful to confirm the metabolite exchanges between the host and symbionts.

Characterisation of the crosstalk between A. queenslandica and symbiotic bacteria in various contexts, such as non-feeding (postlarva) to feeding (juvenile) transition, and different nutrient environment, would help to confirm the dynamic interactions between the sponge and symbiotic microbes. 16S RNA sequencing has revealed that the symbiotic bacterial communities of the late postlarvae and juvenile are similar (Fieth R and Degnan SM, unpublished data), but their nutrient sources are quite different. The postlarvae may obtain nutrient from phagocytosis of the bacterial symbionts that are carried inside the larvae and the new bacteria that are recruited from the

125 environment (Fieth et al. 2016), while the juveniles obtain diverse food from the seawater because they have a functional aquiferous system (Degnan et al. 2015). The A. queenslandica holobiont transcriptome of postlarvae and juvenile would explore the varied host-symbiont interactions in the context of the assimilation of different nutrient sources during development. The A. queenslandica holobiont genomic and transcriptomic data reveal the symbionts play important roles in carbon, nitrogen, sulfur and phosphorous assimilation (Chapter 4). In particular, symbiont AqS1 can reduce environmental nitrate to nitrite, sulfate to sulfide. The A. queenslandica reef flat habitat is rich in nitrate and sulfate, but limited in nitrite and sulfide (Watson, Krömer et al. 2017). The A. queenslandica holobiont transcriptome of different nutrient environments (manually control the nutrient concentration), might help to study dynamic host-symbiont metabolic cooperation in nutrient assimilation.

126

Reference

Abram, F. (2015). "Systems-based approaches to unravel multi-species microbial community functioning." Computational and Structural Biotechnology Journal 13: 24-32.

Achlatis, M., M. Pernice, K. Green, J. M. d. Goeij, P. Guagliardo, M. R. Kilburn, O. Hoegh- Guldberg and S. Dove (2019). "Single-cell visualization indicates direct role of sponge host in uptake of dissolved organic matter." Proceedings of the Royal Society B: Biological Sciences 286(1916): 20192153.

Achlatis, M., M. Pernice, K. Green, P. Guagliardo, M. R. Kilburn, O. Hoegh-Guldberg and S. Dove (2018). "Single-cell measurement of ammonium and bicarbonate uptake within a photosymbiotic bioeroding sponge." The ISME journal 12(5): 1308-1318.

Adachi, K., H. Endo, T. Watanabe, T. Nishioka and T. Hirata (2005). "Hemocyanin in the exoskeleton of crustaceans: enzymatic properties and immunolocalization." Pigment Cell Research 18(2): 136-143.

Adams, M. D., L. M. Wagner, T. J. Graddis, R. Landick, T. K. Antonucci, A. L. Gibson and D. L. Oxender (1990). "Nucleotide sequence and genetic characterization reveal six essential genes for the LIV-I and LS transport systems of Escherichia coli." J Biol Chem 265(20): 11436-11443.

Aguilar-Barajas, E., C. Díaz-Pérez, M. I. Ramírez-Díaz, H. Riveros-Rosas and C. Cervantes (2011). "Bacterial transport of sulfate, molybdate, and related oxyanions." BioMetals 24(4): 687-707.

Alberts, B., A. Johnson, J. Lewis, M. Raff, K. Roberts and P. Walter (2002). The chemical components of a cell. Molecular Biology of the Cell. 4th edition, Garland Science.

Alexander, B. E., M. Achlatis, R. Osinga, H. G. van der Geest, J. P. M. Cleutjens, B. Schutte and J. M. de Goeij (2015). "Cell kinetics during regeneration in the sponge Halisarca caerulea: how local is the response to tissue damage?" PeerJ 3: e820.

Alexander, B. E., K. Liebrand, R. Osinga, H. G. van der Geest, W. Admiraal, J. P. M. Cleutjens, B. Schutte, F. Verheyen, M. Ribes, E. van Loon and J. M. de Goeij (2014). "Cell Turnover and Detritus Production in Marine Sponges from Tropical and Temperate Benthic Ecosystems." PLOS ONE 9(10): e109486.

Allen, A. E., M. G. Booth, M. E. Frischer, P. G. Verity, J. P. Zehr and S. Zani (2001). "Diversity and detection of nitrate assimilation genes in marine bacteria." Appl Environ Microbiol 67(11): 5343-5348.

Almeida, A., A. L. Mitchell, M. Boland, S. C. Forster, G. B. Gloor, A. Tarkowska, T. D. Lawley and R. D. Finn (2019). "A new genomic blueprint of the human gut microbiota." Nature 568(7753): 499-504.

127

Anavy, L., M. Levin, S. Khair, N. Nakanishi, S. L. Fernandez-Valverde, B. M. Degnan and I. Yanai (2014). "BLIND ordering of large-scale transcriptomic developmental timecourses." Development 141(5): 1161-1166.

Anders, S., P. T. Pyl and W. Huber (2015). "HTSeq--a Python framework to work with high- throughput sequencing data." Bioinformatics 31(2): 166-169.

Anderson, K. L. and P. M. Dunman (2009). "Messenger RNA Turnover Processes in Escherichia coli, Bacillus subtilis, and Emerging Studies in Staphylococcus aureus." Int J Microbiol 2009: 525491.

Apprill, A. (2017). "Marine Animal Microbiomes: Toward Understanding Host–Microbiome Interactions in a Changing Ocean." Frontiers in Marine Science 4(222).

Araji, S., T. A. Grammer, R. Gertzen, S. D. Anderson, M. Mikulic-Petkovsek, R. Veberic, M. L. Phu, A. Solar, C. A. Leslie, A. M. Dandekar and M. A. Escobar (2014). "Novel Roles for the Polyphenol Oxidase Enzyme in Secondary Metabolism and the Regulation of Cell Death in Walnut." Plant Physiology 164(3): 1191-1203.

Asano, Y., T. Hiramoto, R. Nishino, Y. Aiba, T. Kimura, K. Yoshihara, Y. Koga and N. Sudo (2012). "Critical role of gut microbiota in the production of biologically active, free catecholamines in the gut lumen of mice." American Journal of Physiology-Gastrointestinal and Liver Physiology 303(11): G1288-G1295.

Ashburner, M., C. A. Ball, J. A. Blake, D. Botstein, H. Butler, J. M. Cherry, A. P. Davis, K. Dolinski, S. S. Dwight, J. T. Eppig, M. A. Harris, D. P. Hill, L. Issel-Tarver, A. Kasarskis, S. Lewis, J. C. Matese, J. E. Richardson, M. Ringwald, G. M. Rubin, G. Sherlock and G. O. Consortium (2000). "Gene Ontology: tool for the unification of biology." Nature Genetics 25(1): 25-29.

Augustin, R., K. Schröder, A. P. Murillo Rincón, S. Fraune, F. Anton-Erxleben, E. M. Herbst, J. Wittlieb, M. Schwentner, J. Grötzinger, T. M. Wassenaar and T. C. G. Bosch (2017). "A secreted antibacterial neuropeptide shapes the microbiome of Hydra." Nat Commun 8(1): 698.

Ayling, A. L. (1983). "Growth and regeneration rates in thinly encrusting demospongiae from temperate waters." The Biological Bulletin 165(2): 343-352.

Bacon, B., L.-M. Nisbett and E. Boon (2017). "Bacterial Haemoprotein Sensors of NO: H- NOX and NosP." Advances in microbial physiology 70: 1-36.

Baddal, B., A. Muzzi, S. Censini, R. A. Calogero, G. Torricelli, S. Guidotti, A. R. Taddei, A. Covacci, M. Pizza, R. Rappuoli, M. Soriani and A. Pezzicolia (2015). "Dual RNA-seq of Nontypeable Haemophilus influenzae and Host Cell Transcriptomes Reveals Novel Insights into Host-Pathogen Cross Talk." Mbio 6(6).

Bali, S., A. D. Lawrence, S. A. Lobo, L. M. Saraiva, B. T. Golding, D. J. Palmer, M. J. Howard, S. J. Ferguson and M. J. Warren (2011). "Molecular hijacking of siroheme for the synthesis of heme and d1 heme." Proc Natl Acad Sci U S A 108(45): 18260-18265.

128

Bates, D., M. Mächler, B. Bolker and S. Walker (2014). "Fitting linear mixed-effects models using lme4." arXiv preprint arXiv:1406.5823.

Baumgartner, M., S. Roffler, T. Wicker and J. Pernthaler (2017). "Letting go: bacterial genome reduction solves the dilemma of adapting to predation mortality in a substrate- restricted environment." ISME J 11(10): 2258-2266.

Bayer, K., M. T. Jahn, B. M. Slaby, L. Moitinho-Silva and U. Hentschel (2018). "Marine Sponges as Chloroflexi Hot Spots: Genomic Insights and High-Resolution Visualization of an Abundant and Diverse Symbiotic Clade." Msystems 3(6).

Bear, R., David Rintoul, Bruce Snyder, Martha Smith-Caldas, Christopher Herren, and Eva Horne (2019). Overview of Cellular Respiration. Principles of Biology.

Beaulieu, J.-M. and R. R. Gainetdinov (2011). "The Physiology, Signaling, and Pharmacology of Dopamine Receptors." Pharmacological Reviews 63(1): 182.

Belkaid, Y. and T. W. Hand (2014). "Role of the microbiota in immunity and inflammation." Cell 157(1): 121-141.

Bell, J. J. (2008). "The functional roles of marine sponges." Estuarine, Coastal and Shelf Science 79(3): 341-353.

Bender, D. A. and P. A. Mayes (2016). The Citric Acid Cycle: The Central Pathway of Carbohydrate, Lipid & Amino Acid Metabolism. Harper's Illustrated Biochemistry, 30e. V. W. Rodwell, D. A. Bender, K. M. Botham, P. J. Kennelly and P. A. Weil. New York, NY, McGraw-Hill Education.

Biscocho, D., J. G. Cook, J. Long, N. Shah and E. M. Leise (2018). "GABA is an inhibitory neurotransmitter in the neural circuit regulating metamorphosis in a marine snail." Developmental Neurobiology 78(7): 736-753.

Björk, J. R., C. Díez-Vives, C. Astudillo-García, E. A. Archie and J. M. Montoya (2019). "Vertical transmission of sponge microbiota is inconsistent and unfaithful." Nature ecology & evolution 3(8): 1172-1183.

Bolger, A. M., M. Lohse and B. Usadel (2014). "Trimmomatic: a flexible trimmer for Illumina sequence data." Bioinformatics 30(15): 2114-2120.

Bordenstein, S. R. and K. R. Theis (2015). "Host Biology in Light of the Microbiome: Ten Principles of Holobionts and Hologenomes." Plos Biology 13(8).

Boscaro, V., M. Kolisko, M. Felletti, C. Vannini, D. H. Lynn and P. J. Keeling (2017). "Parallel genome reduction in symbionts descended from closely related free-living bacteria." Nature Ecology & Evolution 1(8): 1160-1167.

Bravo, J. A., P. Forsythe, M. V. Chew, E. Escaravage, H. M. Savignac, T. G. Dinan, J. Bienenstock and J. F. Cryan (2011). "Ingestion of Lactobacillus strain regulates emotional behavior and central GABA receptor expression in a mouse via the vagus nerve." Proc Natl Acad Sci U S A 108(38): 16050-16055.

129

Britstein, M., G. Devescovi, K. M. Handley, A. Malik, M. Haber, K. Saurav, R. Teta, V. Costantino, I. Burgsdorf, J. A. Gilbert, N. Sher, V. Venturi and L. Steindler (2016). "A New N-Acyl Homoserine Lactone Synthase in an Uncultured Symbiont of the Red Sea Sponge Theonella swinhoei." Appl Environ Microbiol 82(4): 1274-1285.

Britstein, M., K. Saurav, R. Teta, G. D. Sala, R. Bar-Shalom, N. Stoppelli, L. Zoccarato, V. Costantino and L. Steindler (2017). "Identification and chemical characterization of N-acyl- homoserine lactone quorum sensing signals across sponge species and time." FEMS Microbiology Ecology 94(2).

Brodie, B. B. and P. A. Shore (1957). "A concept for a role of serotonin and norepinephrine as chemical mediators in the brain." Annals of the New York Academy of Sciences 66(3): 631-642.

Brown, A. J., S. M. Goldsworthy, A. A. Barnes, M. M. Eilert, L. Tcheang, D. Daniels, A. I. Muir, M. J. Wigglesworth, I. Kinghorn and N. J. Fraser (2003). "The Orphan G protein- coupled receptors GPR41 and GPR43 are activated by propionate and other short chain carboxylic acids." Journal of Biological Chemistry 278(13): 11312-11319.

Browne, H. P., S. C. Forster, B. O. Anonye, N. Kumar, B. A. Neville, M. D. Stares, D. Goulding and T. D. Lawley (2016). "Culturing of ‘unculturable’ human microbiota reveals novel taxa and extensive sporulation." Nature 533(7604): 543-546.

Brune, A. (2014). "Symbiotic digestion of lignocellulose in termite guts." Nature Reviews Microbiology 12(3): 168-180.

Butterfield, D. A., S. S. Hardas and M. L. B. Lange (2010). "Oxidatively modified glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and Alzheimer's disease: many pathways to neurodegeneration." Journal of Alzheimer's disease : JAD 20(2): 369-393.

Caldara, M., G. Dupont, F. Leroy, A. Goldbeter, L. De Vuyst and R. Cunin (2008). "Arginine biosynthesis in Escherichia coli: experimental perturbation and mathematical modeling." J Biol Chem 283(10): 6347-6358.

Camacho, C., G. Coulouris, V. Avagyan, N. Ma, J. Papadopoulos, K. Bealer and T. L. Madden (2009). "BLAST plus : architecture and applications." Bmc Bioinformatics 10.

Carabotti, M., A. Scirocco, M. A. Maselli and C. Severi (2015). "The gut-brain axis: interactions between enteric microbiota, central and enteric nervous systems." Annals of gastroenterology 28(2): 203-209.

Casagrande, F., M. Ratera, A. D. Schenk, M. Chami, E. Valencia, J. M. Lopez, D. Torrents, A. Engel, M. Palacin and D. Fotiadis (2008). "Projection structure of a member of the amino acid/polyamine/organocation transporter superfamily." J Biol Chem 283(48): 33240-33248.

César-Razquin, A., B. Snijder, T. Frappier-Brinton, R. Isserlin, G. Gyimesi, X. Bai, R. A. Reithmeier, D. Hepworth, M. A. Hediger and A. M. Edwards (2015). "A call for systematic research on solute carriers." Cell 162(3): 478-487.

130

Chalifoux, J. R. and A. G. Carter (2011). "GABAB receptor modulation of synaptic function." Curr Opin Neurobiol 21(2): 339-344.

Chen, H., P.-K. Nwe, Y. Yang, C. E. Rosen, A. A. Bielecka, M. Kuchroo, G. W. Cline, A. C. Kruse, A. M. Ring, J. M. Crawford and N. W. Palm (2019). "A Forward Chemical Genetic Screen Reveals Gut Microbiota Metabolites That Modulate Host Physiology." Cell 177(5): 1217-1231.e1218.

Chen, Y., A. T. Lun and G. K. Smyth (2016). "From reads to genes to pathways: differential expression analysis of RNA-Seq experiments using Rsubread and the edgeR quasi-likelihood pipeline." F1000Res 5: 1438.

Chevrot, R., R. Rosen, E. Haudecoeur, A. Cirou, B. J. Shelp, E. Ron and D. Faure (2006). "GABA controls the level of quorum-sensing signal in Agrobacterium tumefaciens." Proceedings of the National Academy of Sciences of the United States of America 103(19): 7460-7464.

Chu, H. and S. K. Mazmanian (2013). "Innate immune recognition of the microbiota promotes host-microbial symbiosis." Nat Immunol 14(7): 668-675.

Clarke, M. B., D. T. Hughes, C. Zhu, E. C. Boedeker and V. Sperandio (2006). "The QseC sensor kinase: A bacterial adrenergic receptor." Proceedings of the National Academy of Sciences 103(27): 10420.

Cohen, L. J., D. Esterhazy, S.-H. Kim, C. Lemetre, R. R. Aguilar, E. A. Gordon, A. J. Pickard, J. R. Cross, A. B. Emiliano, S. M. Han, J. Chu, X. Vila-Farres, J. Kaplitt, A. Rogoz, P. Y. Calle, C. Hunter, J. K. Bitok and S. F. Brady (2017). "Commensal bacteria make GPCR ligands that mimic human signalling molecules." Nature 549(7670): 48-53.

Colandene, J. D. and R. H. Garrett (1996). "Functional dissection and site-directed mutagenesis of the structural gene for NAD(P)H-nitrite reductase in Neurospora crassa." J Biol Chem 271(39): 24096-24104.

Collins, S. M., M. Surette and P. Bercik (2012). "The interplay between the intestinal microbiota and the brain." Nature Reviews Microbiology 10(11): 735-742.

Colosimo, D. A., J. A. Kohn, P. M. Luo, F. J. Piscotta, S. M. Han, A. J. Pickard, A. Rao, J. R. Cross, L. J. Cohen and S. F. Brady (2019). "Mapping Interactions of Microbial Metabolites with Human G-Protein-Coupled Receptors." Cell Host & Microbe 26(2): 273-282.e277.

Conesa, A. and S. Gotz (2008). "Blast2GO: A comprehensive suite for functional analysis in plant genomics." Int J Plant Genomics 2008: 619832.

Conesa, A., P. Madrigal, S. Tarazona, D. Gomez-Cabrero, A. Cervera, A. McPherson, M. W. Szczesniak, D. J. Gaffney, L. L. Elo, X. Zhang and A. Mortazavi (2016). "A survey of best practices for RNA-seq data analysis." Genome Biol 17: 13.

Coon, S., M. Walch, W. Fitt, R. Weiner and D. Bonar (1990). "Ammonia induces settlement behavior in oyster larvae." The Biological Bulletin 179(3): 297-303.

131

Dashty, M. (2013). "A quick look at biochemistry: Carbohydrate metabolism." Clinical Biochemistry 46(15): 1339-1352.

Dave, U. C. and R. K. Kadeppagari (2019). "Alanine dehydrogenase and its applications - A review." Crit Rev Biotechnol 39(5): 648-664.

Davidson, S. K., T. A. Koropatnick, R. Kossmehl, L. Sycuro and M. J. McFall-Ngai (2004). "NO means ‘yes’ in the squid-vibrio symbiosis: nitric oxide (NO) during the initial stages of a beneficial association." Cellular Microbiology 6(12): 1139-1151.

Davy, S. K., D. A. Trautman, M. A. Borowitzka and R. Hinde (2002). "Ammonium excretion by a symbiotic sponge supplies the nitrogen requirements of its rhodophyte partner." Journal of Experimental Biology 205(22): 3505. de Goeij, J. M., L. Moodley, M. Houtekamer, N. M. Carballeira and F. C. van Duyl (2008). "Tracing 13C-enriched dissolved and particulate organic carbon in the bacteria-containing coral reef sponge Halisarca caerulea: Evidence for DOM-feeding." Limnology and Oceanography 53(4): 1376-1386. de Goeij, J. M., D. van Oevelen, M. J. A. Vermeij, R. Osinga, J. J. Middelburg, A. F. P. M. de Goeij and W. Admiraal (2013). "Surviving in a Marine Desert: The Sponge Loop Retains Resources Within Coral Reefs." Science 342(6154): 108-110. de Mendoza, A., W. L. Hatleberg, K. Pang, S. Leininger, O. Bogdanovic, J. Pflueger, S. Buckberry, U. Technau, A. Hejnol, M. Adamska, B. M. Degnan, S. M. Degnan and R. Lister (2019). "Convergent evolution of a vertebrate-like methylome in a marine sponge." Nature Ecology & Evolution 3(10): 1464-1473.

Degnan, B. M., M. Adamska, A. Craigie, S. M. Degnan, B. Fahey, M. Gauthier, J. N. Hooper, C. Larroux, S. P. Leys, E. Lovas and G. S. Richards (2008). "The Demosponge Amphimedon queenslandica: Reconstructing the Ancestral Metazoan Genome and Deciphering the Origin of Animal Multicellularity." Cold Spring Harbor Protocol 2008: pdb.emo108.

Degnan, B. M., M. Adamska, G. S. Richards, C. Larroux, S. Leininger, B. Bergum, A. Calcino, K. Taylor, N. Nakanishi and S. M. Degnan (2015). Porifera. Evolutionary developmental biology of invertebrates 1, Springer: 65-106. den Besten, G., K. van Eunen, A. K. Groen, K. Venema, D.-J. Reijngoud and B. M. Bakker (2013). "The role of short-chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism." Journal of lipid research 54(9): 2325-2340.

Dendooven, T., B. F. Luisi and K. J. Bandyra (2020). "RNA lifetime control, from stereochemistry to gene expression." Curr Opin Struct Biol 61: 59-70.

Deutscher, J., C. Francke and P. W. Postma (2006). "How phosphotransferase system-related protein phosphorylation regulates carbohydrate metabolism in bacteria." Microbiology and Molecular Biology Reviews 70(4): 939-+.

132

Diaz Heijtz, R., S. Wang, F. Anuar, Y. Qian, B. Björkholm, A. Samuelsson, M. L. Hibberd, H. Forssberg and S. Pettersson (2011). "Normal gut microbiota modulates brain development and behavior." Proceedings of the National Academy of Sciences of the United States of America 108(7): 3047-3052.

Douglas, A. E. (2009). "The microbial dimension in insect nutritional ecology." Functional Ecology 23(1): 38-47.

Douglas, A. E. (2018). Fundamentals of microbiome science: how microbes shape animal biology, Princeton University Press.

Douglas, A. E. (2019). "Simple animal models for microbiome research." Nature Reviews Microbiology 17(12): 764-775.

Dreyfus, M. and P. Regnier (2002). "The poly(A) tail of mRNAs: Bodyguard in eukaryotes, scavenger in bacteria." Cell 111(5): 611-613.

Eitinger, T., D. A. Rodionov, M. Grote and E. Schneider (2011). "Canonical and ECF-type ATP-binding cassette importers in prokaryotes: diversity in modular organization and cellular functions." FEMS Microbiol Rev 35(1): 3-67.

Eliot, A. C. and J. F. Kirsch (2004). "Pyridoxal phosphate enzymes: Mechanistic, structural, and evolutionary considerations." Annual Review of Biochemistry 73: 383-415.

Elliott, G. R. D. and S. P. Leys (2010). "Evidence for glutamate, GABA and NO in coordinating behaviour in the sponge, Ephydatia muelleri (Demospongiae, Spongillidae)." The Journal of Experimental Biology 213(13): 2310.

Ellwanger, K. and M. Nickel (2006). "Neuroactive substances specifically modulate rhythmic body contractions in the nerveless metazoon Tethya wilhelma (Demospongiae, Porifera)." Frontiers in Zoology 3(1): 7.

Engelberts, J. P., S. J. Robbins, J. M. de Goeij, M. Aranda, S. C. Bell and N. S. Webster (2020). "Characterization of a sponge microbiome using an integrative genome-centric approach." Isme Journal.

Engelstadter, J. and G. D. D. Hurst (2009). "The Ecology and Evolution of Microbes that Manipulate Host Reproduction." Annual Review of Ecology Evolution and Systematics 40: 127-149.

Ereskovsky, A. V., D. J. Richter, D. V. Lavrov, K. J. Schippers and S. A. Nichols (2017). "Transcriptome sequencing and delimitation of sympatric Oscarella species (O. carmela and O. pearsei sp. nov) from California, USA." PLOS ONE 12(9): e0183002.

Esteves, A. I. S., N. Amer, M. Nguyen and T. Thomas (2016). "Sample Processing Impacts the Viability and Cultivability of the Sponge Microbiome." Frontiers in Microbiology 7(499): 499.

Ezenwa, V. O., N. M. Gerardo, D. W. Inouye, M. Medina and J. B. Xavier (2012). "Animal Behavior and the Microbiome." Science 338(6104): 198-199.

133

Falk-Krzesinski, H. J. and A. J. Wolfe (1998). "Genetic analysis of the nuo locus, which encodes the proton-translocating NADH dehydrogenase in Escherichia coli." J Bacteriol 180(5): 1174-1184.

Feehily, C. and K. A. G. Karatzas (2013). "Role of glutamate metabolism in bacterial responses towards acid and other stresses." Journal of Applied Microbiology 114(1): 11-24.

Feehily, C., C. P. O'Byrne and K. A. G. Karatzas (2013). "Functional gamma-Aminobutyrate Shunt in Listeria monocytogenes: Role in Acid Tolerance and Succinate Biosynthesis." Applied and Environmental Microbiology 79(1): 74-80.

Feng, H., N. Edwards, C. M. H. Anderson, M. Althaus, R. P. Duncan, Y.-C. Hsu, C. W. Luetje, D. R. G. Price, A. C. C. Wilson and D. T. Thwaites (2019). "Trading amino acids at the aphid–Buchnera symbiotic interface." Proceedings of the National Academy of Sciences 116(32): 16003.

Fernandez-Valverde, S. L., A. D. Calcino and B. M. Degnan (2015). "Deep developmental transcriptome sequencing uncovers numerous new genes and enhances gene annotation in the sponge Amphimedon queenslandica." BioMed Central Genomics 16: 387.

Fernandez-Valverde, S. L. and B. M. Degnan (2016). "Bilaterian-like promoters in the highly compact Amphimedon queenslandica genome." Scientific reports 6: 22496-22496.

Fettweis, J. M., M. G. Serrano, J. P. Brooks, D. J. Edwards, P. H. Girerd, H. I. Parikh, B. Huang, T. J. Arodz, L. Edupuganti, A. L. Glascock, J. Xu, N. R. Jimenez, S. C. Vivadelli, S. S. Fong, N. U. Sheth, S. Jean, V. Lee, Y. A. Bokhari, A. M. Lara, S. D. Mistry, R. A. Duckworth, S. P. Bradley, V. N. Koparde, X. V. Orenda, S. H. Milton, S. K. Rozycki, A. V. Matveyev, M. L. Wright, S. V. Huzurbazar, E. M. Jackson, E. Smirnova, J. Korlach, Y.-C. Tsai, M. R. Dickinson, J. L. Brooks, J. I. Drake, D. O. Chaffin, A. L. Sexton, M. G. Gravett, C. E. Rubens, N. R. Wijesooriya, K. D. Hendricks-Muñoz, K. K. Jefferson, J. F. Strauss and G. A. Buck (2019). "The vaginal microbiome and preterm birth." Nature Medicine 25(6): 1012-1021.

Fieth, R. A., M.-E. A. Gauthier, J. Bayes, K. M. Green and S. M. Degnan (2016). "Ontogenetic Changes in the Bacterial Symbiont Community of the Tropical Demosponge Amphimedon queenslandica: Metamorphosis Is a New Beginning." Frontiers in Marine Science 3: 228.

Fiore, C. L., D. M. Baker and M. P. Lesser (2013). "Nitrogen biogeochemistry in the Caribbean sponge, Xestospongia muta: a source or sink of dissolved inorganic nitrogen?" PLoS One 8(8): e72961.

Fiore, C. L., M. Labrie, J. K. Jarettt and M. P. Lesser (2015). "Transcriptional activity of the giant barrel sponge, Xestospongia muta Holobiont: molecular evidence for metabolic interchange." Frontiers in Microbiology 6: 364.

Flint, H. J., E. A. Bayer, M. T. Rincon, R. Lamed and B. A. White (2008). "Polysaccharide utilization by gut bacteria: potential for new insights from genomic analysis." Nature Reviews Microbiology 6(2): 121-131.

134

Folkers, M. and T. Rombouts (2020). Sponges Revealed: A Synthesis of Their Overlooked Ecological Functions Within Aquatic Ecosystems. YOUMARES 9 - The Oceans: Our Research, Our Future: Proceedings of the 2018 conference for YOUng MArine RESearcher in Oldenburg, Germany. S. Jungblut, V. Liebich and M. Bode-Dalby. Cham, Springer International Publishing: 181-193.

Fortunato, S. A. V., M. Adamski, O. M. Ramos, S. Leininger, J. Liu, D. E. K. Ferrier and M. Adamska (2014). "Calcisponges have a ParaHox gene and dynamic expression of dispersed NK homeobox genes." Nature 514(7524): 620-623.

Francis, W. R., M. Eitel, S. Vargas, M. Adamski, S. H. D. Haddock, S. Krebs, H. Blum, D. Erpenbeck and G. Wörheide (2017). "The genome of the contractile demosponge Tethya wilhelma and the evolution of metazoan neural signalling pathways." bioRxiv: 120998.

Freestone, P. P., P. H. Williams, R. D. Haigh, A. F. Maggs, C. P. Neal and M. Lyte (2002). "Growth stimulation of intestinal commensal Escherichia coli by catecholamines: a possible contributory factor in trauma-induced sepsis." Shock 18(5): 465-470.

Gaiti, F., S. L. Fernandez-Valverde, N. Nakanishi, A. D. Calcino, I. Yanai, M. Tanurdzic and B. M. Degnan (2015). "Dynamic and Widespread lncRNA Expression in a Sponge and the Origin of Animal Complexity." Mol Biol Evol 32(9): 2367-2382.

Gaiti, F., K. Jindrich, S. L. Fernandez-Valverde, K. E. Roper, B. M. Degnan and M. Tanurdzic (2017). "Landscape of histone modifications in a sponge reveals the origin of animal cis-regulatory complexity." Elife 6.

Galland, L. (2014). "The gut microbiome and the brain." Journal of medicinal food 17(12): 1261-1272.

Gantt, S. E., S. E. McMurray, A. D. Stubler, C. M. Finelli, J. R. Pawlik and P. M. Erwin (2019). "Testing the relationship between microbiome composition and flux of carbon and nutrients in Caribbean coral reef sponges." Microbiome 7(1): 124.

Gao, H., C. Cui, L. Wang, M. Jacobs-Lorena and S. Wang (2020). "Mosquito Microbiota and Implications for Disease Control." Trends in Parasitology 36(2): 98-111.

Garcia-Lavandeira, M., A. Silva, M. Abad, A. J. Pazos, J. L. Sanchez and M. L. Perez-Paralle (2005). "Effects of GABA and epinephrine on the settlement and metamorphosis of the larvae of four species of bivalve molluscs." Journal of Experimental Marine Biology and Ecology 316(2): 149-156.

Gardères, J., J. Henry, B. Bernay, A. Ritter, C. Zatylny-Gaudin, M. Wiens, W. E. Müller and G. Le Pennec (2014). "Cellular effects of bacterial N-3-Oxo-dodecanoyl-L-Homoserine lactone on the sponge Suberites domuncula (Olivi, 1792): insights into an intimate inter- kingdom dialogue." PLoS One 9(5): e97662.

Gauthier, M.-E. A., J. R. Watson and S. M. Degnan (2016). "Draft Genomes Shed Light on the Dual Bacterial Symbiosis that Dominates the Microbiome of the Coral Reef Sponge Amphimedon queenslandica." Frontiers in Marine Science 3(196).

135

Gensollen, T., S. S. Iyer, D. L. Kasper and R. S. Blumberg (2016). "How colonization by microbiota in early life shapes the immune system." Science 352(6285): 539.

Germer, J., N. Cerveau and D. J. Jackson (2017). "The Holo-Transcriptome of a Calcified Early Branching Metazoan." Frontiers in Marine Science 4(81): 81.

Ghosh, W. and B. Dam (2009). "Biochemistry and molecular biology of lithotrophic sulfur oxidation by taxonomically and ecologically diverse bacteria and archaea." FEMS Microbiol Rev 33(6): 999-1043.

Gibson, M. I., E. J. Brignole, E. Pierce, M. Can, S. W. Ragsdale and C. L. Drennan (2015). "The Structure of an Oxalate Oxidoreductase Provides Insight into Microbial 2-Oxoacid Metabolism." Biochemistry 54(26): 4112-4120.

Gilbert, J. A., M. J. Blaser, J. G. Caporaso, J. K. Jansson, S. V. Lynch and R. Knight (2018). "Current understanding of the human microbiome." Nature Medicine 24(4): 392-400.

Gilbert, S. F., J. Sapp and A. I. Tauber (2012). "A Symbiotic View of Life: We Have Never Been Individuals." Quarterly Review of Biology 87(4): 325-341.

Goh, E.-B., P. J. Bledsoe, L.-L. Chen, P. Gyaneshwar, V. Stewart and M. M. Igo (2005). "Hierarchical control of anaerobic gene expression in Escherichia coli K-12: the nitrate- responsive NarX-NarL regulatory system represses synthesis of the fumarate-responsive DcuS-DcuR regulatory system." Journal of bacteriology 187(14): 4890-4899.

Goldstein, D. S. (2010). Adrenaline and Noradrenaline. eLS.

Gopalakrishnan, V., B. A. Helmink, C. N. Spencer, A. Reuben and J. A. Wargo (2018). "The Influence of the Gut Microbiome on Cancer, Immunity, and Cancer Immunotherapy." Cancer Cell 33(4): 570-580.

Grandy, D. K., G. M. Miller and J. X. Li (2016). ""TAARgeting Addiction"-The Alamo Bears Witness to Another Revolution An Overview of the Plenary Symposium of the 2015 Behavior, Biology and Chemistry Conference." Drug and Alcohol Dependence 159: 9-16.

Grice, L. F., M. E. Gauthier, K. E. Roper, X. Fernandez-Busquets, S. M. Degnan and B. M. Degnan (2017). "Origin and evolution of the sponge aggregation factor gene family." Mol Biol Evol.

Griffiths, S. M., R. E. Antwis, L. Lenzi, A. Lucaci, D. C. Behringer, M. J. Butler IV and R. F. Preziosi (2019). "Host genetics and geography influence microbiome composition in the sponge Ircinia campana." Journal of Animal Ecology 88(11): 1684-1695.

Grote, A., D. Voronin, T. Ding, A. Twaddle, T. R. Unnasch, S. Lustigman and E. Ghedin (2017). "Defining Brugia malayi and Wolbachia symbiosis by stage-specific dual RNA-seq." Plos Neglected Tropical Diseases 11(3).

Group, N. H. W., J. Peterson, S. Garges, M. Giovanni, P. McInnes, L. Wang, J. A. Schloss, V. Bonazzi, J. E. McEwen, K. A. Wetterstrand, C. Deal, C. C. Baker, V. Di Francesco, T. K. Howcroft, R. W. Karp, R. D. Lunsford, C. R. Wellington, T. Belachew, M. Wright, C. Giblin,

136

H. David, M. Mills, R. Salomon, C. Mullins, B. Akolkar, L. Begg, C. Davis, L. Grandison, M. Humble, J. Khalsa, A. R. Little, H. Peavy, C. Pontzer, M. Portnoy, M. H. Sayre, P. Starke-Reed, S. Zakhari, J. Read, B. Watson and M. Guyer (2009). "The NIH Human Microbiome Project." Genome Res 19(12): 2317-2323.

Gruber, N. (2008). "The marine nitrogen cycle: overview and challenges." Nitrogen in the marine environment 2: 1-50.

Hadfield, M. G. and V. J. Paul (2001). "Natural chemical cues for settlement and metamorphosis of marine invertebrate larvae." Marine chemical ecology 13: 431-461.

Hadjiconstantinou, M. and N. H. Neff (2008). "Enhancing aromatic L-amino acid decarboxylase activity: implications for L-DOPA treatment in Parkinson's disease." CNS Neurosci Ther 14(4): 340-351.

Hajnsdorf, E. and V. R. Kaberdin (2018). "RNA polyadenylation and its consequences in prokaryotes." Philosophical Transactions of the Royal Society B-Biological Sciences 373(1762).

Hallam, S. J. (2006). "Genomic analysis of the uncultivated marine crenarchaeote Cenarchaeum symbiosum." Proc. Natl Acad. Sci. USA 103: 18296-18301.

Haugel-Nielsen, J., E. Hajnsdorf and P. Regnier (1996). "The rpsO mRNA of Escherichia coli is polyadenylated at multiple sites resulting from endonucleolytic processing and exonucleolytic degradation." The EMBO journal 15(12): 3144-3152.

Hazes, B., K. A. Magnus, C. Bonaventura, J. Bonaventura, Z. Dauter, K. H. Kalk and W. G. J. Hol (1993). "Crystal-Structure of Deoxygenated Limulus-Polyphemus Subunit-Ii Hemocyanin at 2.18-Angstrom Resolution - Clues for a Mechanism for Allosteric Regulation." Protein Science 2(4): 597-619.

Hedner, E., M. Sjögren, P.-A. Frändberg, T. Johansson, U. Göransson, M. Dahlström, P. Jonsson, F. Nyberg and L. Bohlin (2006). "Brominated Cyclodipeptides from the Marine Sponge Geodia barretti as Selective 5-HT Ligands." Journal of Natural Products 69(10): 1421-1424.

Heiden, M. G. V., L. C. Cantley and C. B. Thompson (2009). "Understanding the Warburg Effect: The Metabolic Requirements of Cell Proliferation." Science 324(5930): 1029-1033.

Heinken, A. and I. Thiele (2015). "Systems biology of host-microbe metabolomics." Wiley Interdiscip Rev Syst Biol Med 7(4): 195-219.

Henares, B. M., K. E. Higgins and E. M. Boon (2012). "Discovery of a nitric oxide responsive quorum sensing circuit in Vibrio harveyi." ACS Chem Biol 7(8): 1331-1336.

Hentschel, U., J. Piel, S. M. Degnan and M. W. Taylor (2012). "Genomic insights into the marine sponge microbiome." Nature Reviews Microbiology 10(9): 641-U675.

Herrmann, K. M. and L. M. Weaver (1999). "The shikimate pathway." Annual Review of Plant Physiology and Plant Molecular Biology 50: 473-503.

137

Hmelo, L. R. (2017). "Quorum Sensing in Marine Microbial Environments." Annual Review of Marine Science 9(1): 257-281.

Hoffmann, F. (2009). "Complex nitrogen cycling in the sponge Geodia barretti." Environ. Microbiol. 11: 2228-2243.

Holt, C. C., D. Bass, G. D. Stentiford and M. van der Giezen (2020). "Understanding the role of the shrimp gut microbiome in health and disease." Journal of Invertebrate Pathology: 107387.

Hooper J, V. S. R. (2006). "A new species of Amphimedon (Porifera, Demospongiae, Haplosclerida, Niphatidae) from the Capricorn-Bunker Group of Islands, Great Barrier Reef, Australia: target species for the ‘sponge genome project’." Zootaxa 1314: 31-39.

Hossain, S. and E. M. Boon (2017). "Discovery of a Novel Nitric Oxide Binding Protein and Nitric-Oxide-Responsive Signaling Pathway in Pseudomonas aeruginosa." ACS Infect Dis 3(6): 454-461.

Hsieh, P. C., T. H. Kowalczyk and N. F. Phillips (1996). "Kinetic mechanisms of polyphosphate glucokinase from Mycobacterium tuberculosis." Biochemistry 35(30): 9772- 9781.

Hu, J. F., J. A. Schetz, M. Kelly, J. N. Peng, K. K. Ang, H. Flotow, C. Y. Leong, S. B. Ng, A. D. Buss, S. P. Wilkins and M. T. Hamann (2002). "New antiinfective and human 5-HT2 receptor binding natural and semisynthetic compounds from the Jamaican sponge Smenospongia aurea." J Nat Prod 65(4): 476-480.

Hughes, D. T. and V. Sperandio (2008). "Inter-kingdom signalling: communication between bacteria and their hosts." Nature Reviews Microbiology 6(2): 111-120.

Hui, M. P., P. L. Foley and J. G. Belasco (2014). "Messenger RNA degradation in bacterial cells." Annu Rev Genet 48: 537-559.

Husted, A. S., M. Trauelsen, O. Rudenko, S. A. Hjorth and T. W. Schwartz (2017). "GPCR- Mediated Signaling of Metabolites." Cell Metabolism 25(4): 777-796.

Huttenhower, C., D. Gevers, R. Knight, S. Abubucker, J. H. Badger, A. T. Chinwalla, H. H. Creasy, A. M. Earl, M. G. FitzGerald, R. S. Fulton, M. G. Giglio, K. Hallsworth-Pepin, E. A. Lobos, R. Madupu, V. Magrini, J. C. Martin, M. Mitreva, D. M. Muzny, E. J. Sodergren, J. Versalovic, A. M. Wollam, K. C. Worley, J. R. Wortman, S. K. Young, Q. Zeng, K. M. Aagaard, O. O. Abolude, E. Allen-Vercoe, E. J. Alm, L. Alvarado, G. L. Andersen, S. Anderson, E. Appelbaum, H. M. Arachchi, G. Armitage, C. A. Arze, T. Ayvaz, C. C. Baker, L. Begg, T. Belachew, V. Bhonagiri, M. Bihan, M. J. Blaser, T. Bloom, V. Bonazzi, J. Paul Brooks, G. A. Buck, C. J. Buhay, D. A. Busam, J. L. Campbell, S. R. Canon, B. L. Cantarel, P. S. G. Chain, I. M. A. Chen, L. Chen, S. Chhibba, K. Chu, D. M. Ciulla, J. C. Clemente, S. W. Clifton, S. Conlan, J. Crabtree, M. A. Cutting, N. J. Davidovics, C. C. Davis, T. Z. DeSantis, C. Deal, K. D. Delehaunty, F. E. Dewhirst, E. Deych, Y. Ding, D. J. Dooling, S. P. Dugan, W. Michael Dunne, A. Scott Durkin, R. C. Edgar, R. L. Erlich, C. N. Farmer, R. M. Farrell, K. Faust, M. Feldgarden, V. M. Felix, S. Fisher, A. A. Fodor, L. J. Forney, L. Foster, V. Di Francesco, J. Friedman, D. C. Friedrich, C. C. Fronick, L. L. Fulton, H. Gao, N.

138

Garcia, G. Giannoukos, C. Giblin, M. Y. Giovanni, J. M. Goldberg, J. Goll, A. Gonzalez, A. Griggs, S. Gujja, S. Kinder Haake, B. J. Haas, H. A. Hamilton, E. L. Harris, T. A. Hepburn, B. Herter, D. E. Hoffmann, M. E. Holder, C. Howarth, K. H. Huang, S. M. Huse, J. Izard, J. K. Jansson, H. Jiang, C. Jordan, V. Joshi, J. A. Katancik, W. A. Keitel, S. T. Kelley, C. Kells, N. B. King, D. Knights, H. H. Kong, O. Koren, S. Koren, K. C. Kota, C. L. Kovar, N. C. Kyrpides, P. S. La Rosa, S. L. Lee, K. P. Lemon, N. Lennon, C. M. Lewis, L. Lewis, R. E. Ley, K. Li, K. Liolios, B. Liu, Y. Liu, C.-C. Lo, C. A. Lozupone, R. Dwayne Lunsford, T. Madden, A. A. Mahurkar, P. J. Mannon, E. R. Mardis, V. M. Markowitz, K. Mavromatis, J. M. McCorrison, D. McDonald, J. McEwen, A. L. McGuire, P. McInnes, T. Mehta, K. A. Mihindukulasuriya, J. R. Miller, P. J. Minx, I. Newsham, C. Nusbaum, M. O’Laughlin, J. Orvis, I. Pagani, K. Palaniappan, S. M. Patel, M. Pearson, J. Peterson, M. Podar, C. Pohl, K. S. Pollard, M. Pop, M. E. Priest, L. M. Proctor, X. Qin, J. Raes, J. Ravel, J. G. Reid, M. Rho, R. Rhodes, K. P. Riehle, M. C. Rivera, B. Rodriguez-Mueller, Y.-H. Rogers, M. C. Ross, C. Russ, R. K. Sanka, P. Sankar, J. Fah Sathirapongsasuti, J. A. Schloss, P. D. Schloss, T. M. Schmidt, M. Scholz, L. Schriml, A. M. Schubert, N. Segata, J. A. Segre, W. D. Shannon, R. R. Sharp, T. J. Sharpton, N. Shenoy, N. U. Sheth, G. A. Simone, I. Singh, C. S. Smillie, J. D. Sobel, D. D. Sommer, P. Spicer, G. G. Sutton, S. M. Sykes, D. G. Tabbaa, M. Thiagarajan, C. M. Tomlinson, M. Torralba, T. J. Treangen, R. M. Truty, T. A. Vishnivetskaya, J. Walker, L. Wang, Z. Wang, D. V. Ward, W. Warren, M. A. Watson, C. Wellington, K. A. Wetterstrand, J. R. White, K. Wilczek-Boney, Y. Wu, K. M. Wylie, T. Wylie, C. Yandava, L. Ye, Y. Ye, S. Yooseph, B. P. Youmans, L. Zhang, Y. Zhou, Y. Zhu, L. Zoloth, J. D. Zucker, B. W. Birren, R. A. Gibbs, S. K. Highlander, B. A. Methé, K. E. Nelson, J. F. Petrosino, G. M. Weinstock, R. K. Wilson, O. White and C. The Human Microbiome Project (2012). "Structure, function and diversity of the healthy human microbiome." Nature 486(7402): 207-214.

Illumina. "Ribo-Zero Gold rRNA Removal Kit (Epidemiology)." from https://sapac.illumina.com/products/by-type/molecular-biology-reagents/ribo-zero-gold-rrna- removal-epidemiology.html.

Ingle, R. A. (2011). "Histidine biosynthesis." Arabidopsis Book 9: e0141.

Ismail, A. S., J. S. Valastyan and B. L. Bassler (2016). "A Host-Produced Autoinducer-2 Mimic Activates Bacterial Quorum Sensing." Cell Host & Microbe 19(4): 470-480.

Jackson, M. R., S. L. Melideo and M. S. Jorns (2012). "Human Sulfide:Quinone Oxidoreductase Catalyzes the First Step in Hydrogen Sulfide Metabolism and Produces a Sulfane Sulfur Metabolite." Biochemistry 51(34): 6804-6815.

Jastrzebska, B. (2013). "GPCR: G protein complexes--the fundamental signaling assembly." Amino acids 45(6): 1303-1314.

Jiang, P., W. Du and M. Wu (2014). "Regulation of the pentose phosphate pathway in cancer." Protein & cell 5(8): 592-602.

Jiménez, E. and M. Ribes (2007). "Sponges as a source of dissolved inorganic nitrogen: Nitrification mediated by temperate sponges." Limnology and Oceanography 52(3): 948-958.

Johnson, K. V. A. and K. R. Foster (2018). "Why does the microbiome affect behaviour?" Nature Reviews Microbiology 16(10): 647-655.

139

Kahn, A. S., J. W. F. Chu and S. P. Leys (2018). "Trophic ecology of glass sponge reefs in the Strait of Georgia, British Columbia." Scientific Reports 8(1): 756.

Kahn, A. S., G. Yahel, J. W. F. Chu, V. Tunnicliffe and S. P. Leys (2015). "Benthic grazing and carbon sequestration by deep-water glass sponge reefs." Limnology and Oceanography 60(1): 78-88.

Kamke, J., A. Sczyrba, N. Ivanova, P. Schwientek, C. Rinke, K. Mavromatis, T. Woyke and U. Hentschel (2013). "Single-cell genomics reveals complex carbohydrate degradation patterns in poribacterial symbionts of marine sponges." The ISME journal 7(12): 2287-2300.

Kanehisa, M., M. Furumichi, M. Tanabe, Y. Sato and K. Morishima (2017). "KEGG: new perspectives on genomes, pathways, diseases and drugs." Nucleic Acids Research 45(D1): D353-D361.

Kanehisa, M. and Y. Sato (2020). "KEGG Mapper for inferring cellular functions from protein sequences." Protein Sci 29(1): 28-35.

Kanehisa, M., Y. Sato and K. Morishima (2016). "BlastKOALA and GhostKOALA: KEGG tools for functional characterization of genome and metagenome sequences." Journal of molecular biology 428(4): 726-731.

Karimi, E., B. M. Slaby, A. R. Soares, J. Blom, U. Hentschel and R. Costa (2018). "Metagenomic binning reveals versatile nutrient cycling and distinct adaptive features in alphaproteobacterial symbionts of marine sponges." FEMS Microbiol Ecol 94(6).

Kazanidis, G., D. van Oevelen, B. Veuger and U. F. M. Witte (2018). "Unravelling the versatile feeding and metabolic strategies of the cold-water ecosystem engineer Spongosorites coralliophaga (Stephens, 1915)." Deep Sea Research Part I: Oceanographic Research Papers 141: 71-82.

Kenny, N. J., W. R. Francis, R. E. Rivera-Vicéns, K. Juravel, A. de Mendoza, C. Díez-Vives, R. Lister, L. Bezares-Calderon, L. Grombacher, M. Roller, L. D. Barlow, S. Camilli, J. F. Ryan, G. Wörheide, A. L. Hill, A. Riesgo and S. P. Leys (2020). "The genomic basis of animal origins: a chromosomal perspective from the sponge Ephydatia muelleri." bioRxiv: 2020.2002.2018.954784.

Kim, I. V., E. J. Ross, S. Dietrich, K. Döring, A. Sánchez Alvarado and C.-D. Kuhn (2019). "Efficient depletion of ribosomal RNA for RNA sequencing in planarians." bioRxiv: 670604.

Klimovich, A. V. and T. C. G. Bosch (2018). "Rethinking the Role of the Nervous System: Lessons From the Hydra Holobiont." Bioessays 40(9): e1800060.

Kochanowska, A. J., K. V. Rao, S. Childress, A. El-Alfy, R. R. Matsumoto, M. Kelly, G. S. Stewart, K. J. Sufka and M. T. Hamann (2008). "Secondary metabolites from three Florida sponges with antidepressant activity." Journal of natural products 71(2): 186-189.

Koendjbiharie, J. G., S. Hon, M. Pabst, R. Hooftman, D. M. Stevenson, J. Cui, D. Amador- Noguez, L. R. Lynd, D. G. Olson and R. van Kranenburg (2020). "The pentose phosphate

140 pathway of cellulolytic clostridia relies on 6-phosphofructokinase instead of transaldolase." J Biol Chem 295(7): 1867-1878.

Komoda, T. and T. Matsunaga (2015). Chapter 4 - Metabolic Pathways in the Human Body. Biochemistry for Medical Professionals. T. Komoda and T. Matsunaga. Boston, Academic Press: 25-63.

Komori, H., Y. Nitta, H. Ueno and Y. Higuchi (2012). "Structural Study Reveals That Ser- 354 Determines Substrate Specificity on Human Histidine Decarboxylase." Journal of Biological Chemistry 287(34): 29175-29183.

Kopylova, E., L. Noe and H. Touzet (2012). "SortMeRNA: fast and accurate filtering of ribosomal RNAs in metatranscriptomic data." Bioinformatics 28(24): 3211-3217.

Koropatkin, N. M., E. A. Cameron and E. C. Martens (2012). "How glycan metabolism shapes the human gut microbiota." Nature Reviews Microbiology 10(5): 323-335.

Kowalik, M. A., A. Columbano and A. Perra (2017). "Emerging Role of the Pentose Phosphate Pathway in Hepatocellular Carcinoma." Front Oncol 7: 87.

Krishnan, A., R. Dnyansagar, M. S. Almen, M. J. Williams, R. Fredriksson, N. Manoj and H. B. Schioth (2014). "The GPCR repertoire in the demosponge Amphimedon queenslandica: insights into the GPCR system at the early divergence of animals." Bmc Evolutionary Biology 14.

Krishnan, A. and H. B. Schiöth (2015). "The role of G protein-coupled receptors in the early evolution of neurotransmission and the nervous system." The Journal of Experimental Biology 218(4): 562.

Kuypers, M. M. M., H. K. Marchant and B. Kartal (2018). "The microbial nitrogen-cycling network." Nature Reviews Microbiology 16(5): 263-276.

Lackner, G., E. E. Peters, E. J. N. Helfrich and J. Piel (2017). "Insights into the lifestyle of uncultured bacterial natural product factories associated with marine sponges." Proceedings of the National Academy of Sciences of the United States of America 114(3): E347-E356.

Lagier, J.-C., S. Khelaifia, M. T. Alou, S. Ndongo, N. Dione, P. Hugon, A. Caputo, F. Cadoret, S. I. Traore, E. H. Seck, G. Dubourg, G. Durand, G. Mourembou, E. Guilhot, A. Togo, S. Bellali, D. Bachar, N. Cassir, F. Bittar, J. Delerce, M. Mailhe, D. Ricaboni, M. Bilen, N. P. M. Dangui Nieko, N. M. Dia Badiane, C. Valles, D. Mouelhi, K. Diop, M. Million, D. Musso, J. Abrahão, E. I. Azhar, F. Bibi, M. Yasir, A. Diallo, C. Sokhna, F. Djossou, V. Vitton, C. Robert, J. M. Rolain, B. La Scola, P.-E. Fournier, A. Levasseur and D. Raoult (2016). "Culture of previously uncultured members of the human gut microbiota by culturomics." Nature Microbiology 1(12): 16203.

Lamarche, M. G., B. L. Wanner, S. Crépin and J. Harel (2008). "The phosphate regulon and bacterial virulence: a regulatory network connecting phosphate homeostasis and pathogenesis." FEMS Microbiology Reviews 32(3): 461-473.

141

Lambert, A., M. Østerås, K. Mandon, M. C. Poggi and D. Le Rudulier (2001). "Fructose uptake in Sinorhizobium meliloti is mediated by a high-affinity ATP-binding cassette transport system." Journal of bacteriology 183(16): 4709-4717.

Le Poul, E., C. Loison, S. Struyf, J. Y. Springael, V. Lannoy, M. E. Decobecq, S. Brezillon, V. Dupriez, G. Vassart, J. Van Damme, M. Parmentier and M. Detheux (2003). "Functional characterization of human receptors for short chain fatty acids and their role in polymorphonuclear cell activation." J Biol Chem 278(28): 25481-25489.

Lee, H.-J. (2019). "Microbe-host communication by small RNAs in extracellular vesicles: vehicles for transkingdom RNA transportation." International journal of molecular sciences 20(6): 1487.

Leonardi, R. and S. Jackowski (2007). "Biosynthesis of Pantothenic Acid and Coenzyme A." EcoSal Plus 2(2).

Lesnoff, M., R. Lancelot and M. R. Lancelot (2018). "Package ‘aods3’."

Leulier, F., L. T. MacNeil, W. J. Lee, J. F. Rawls, P. D. Cani, M. Schwarzer, L. P. Zhao and S. J. Simpson (2017). "Integrative Physiology: At the Crossroads of Nutrition, Microbiota, Animal Physiology, and Human Health." Cell Metabolism 25(3): 522-534.

Levin, M., L. Anavy, A. G. Cole, E. Winter, N. Mostov, S. Khair, N. Senderovich, E. Kovalev, D. H. Silver, M. Feder, S. L. Fernandez-Valverde, N. Nakanishi, D. Simmons, O. Simakov, T. Larsson, S. Y. Liu, A. Jerafi-Vider, K. Yaniv, J. F. Ryan, M. Q. Martindale, J. C. Rink, D. Arendt, S. M. Degnan, B. M. Degnan, T. Hashimshony and I. Yanai (2016). "The mid-developmental transition and the evolution of animal body plans." Nature 531(7596): 637-641.

Lewis, K., S. Epstein, A. D'Onofrio and L. L. Ling (2010). "Uncultured microorganisms as a source of secondary metabolites." The Journal of Antibiotics 63(8): 468-476.

Ley, R. E. (2010). "Obesity and the human microbiome." Curr Opin Gastroenterol 26(1): 5- 11.

Leys, S. P. (2015). "Elements of a 'nervous system' in sponges." Journal of Experimental Biology 218(4): 581-591.

Leys, S. P. and B. M. Degnan (2001). "Cytological basis of photoresponsive behavior in a sponge larva." Biological Bulletin 201(3): 323-338.

Leys, S. P., A. S. Kahn, J. K. H. Fang, T. Kutti and R. J. Bannister (2017). "Phagocytosis of microbial symbionts balances the carbon and nitrogen budget for the deep-water boreal sponge Geodia barretti." Limnology and Oceanography 63(1): 187-202.

Leys, S. P., C. Larroux, M. Gauthier, M. Adamska, B. Fahey, G. S. Richards, S. M. Degnan and B. M. Degnan (2008). "Isolation of amphimedon developmental material." CSH Protoc 2008: pdb prot5095.

142

Leys, S. P., J. L. Mah, P. R. McGill, L. Hamonic, F. C. De Leo and A. S. Kahn (2019). "Sponge Behavior and the Chemical Basis of Responses: A Post-Genomic View." Integrative and Comparative Biology 59(4): 751-764.

Leys, S. P., G. Yahel, M. A. Reidenbach, V. Tunnicliffe, U. Shavit and H. M. Reiswig (2011). "The sponge pump: the role of current induced flow in the design of the sponge body plan." PloS one 6(12): e27787-e27787.

Li, B., V. Ruotti, R. M. Stewart, J. A. Thomson and C. N. Dewey (2010). "RNA-Seq gene expression estimation with read mapping uncertainty." Bioinformatics 26(4): 493-500.

Li, H., B. Handsaker, A. Wysoker, T. Fennell, J. Ruan, N. Homer, G. Marth, G. Abecasis, R. Durbin and G. P. D. Proc (2009). "The Sequence Alignment/Map format and SAMtools." Bioinformatics 25(16): 2078-2079.

Li, Q., Y. Ren and X. Fu (2019). "Inter-kingdom signaling between gut microbiota and their host." Cellular and Molecular Life Sciences 76(12): 2383-2389.

Li, Z., J. Jiang, X. Yu, C. Wu, D. Shen and Y. Feng (2017). "Poly(A) polymerase I participates in the indole regulatory pathway of Pantoea agglomerans YS19." Microbiology 163(2): 197-206.

Li, Z. and S. K. Nair (2012). "Quorum sensing: how bacteria can coordinate activity and synchronize their response to external signals?" Protein science : a publication of the Protein Society 21(10): 1403-1417.

Li, Z., Y. Wang, J. Li, F. Liu, L. He, Y. He and S. Wang (2016). "Metagenomic analysis of genes encoding nutrient cycling pathways in the microbiota of deep-sea and shallow-water sponges." Marine Biotechnology 18(6): 659-671.

Li, Z. Y., Y. Z. Wang, L. M. He and H. J. Zheng (2015). "Metabolic profiles of prokaryotic and eukaryotic communities in deep-sea sponge Neamphius huxleyi indicated by metagenomics (vol 4, 3895, 2014)." Scientific Reports 4: 3895.

Liang, J., Q. Han, H. Z. Ding and J. Y. Li (2017). "Biochemical identification of residues that discriminate between 3,4-dihydroxyphenylalanine decarboxylase and 3,4- dihydroxyphenylacetaldehyde synthase-mediated reactions." Insect Biochemistry and Molecular Biology 91: 34-43.

Liang, J., Q. Han, Y. Tan, H. Ding and J. Li (2019). "Current Advances on Structure- Function Relationships of Pyridoxal 5'-Phosphate-Dependent Enzymes." Front Mol Biosci 6: 4.

Liberti, M. V. and J. W. Locasale (2016). "The Warburg Effect: How Does it Benefit Cancer Cells? (vol 41, pg 211, 2016)." Trends in Biochemical Sciences 41(3): 287-287.

Libiad, M., P. K. Yadav, V. Vitvitsky, M. Martinov and R. Banerjee (2014). "Organization of the human mitochondrial hydrogen sulfide oxidation pathway." The Journal of biological chemistry 289(45): 30901-30910.

143

Liebeskind, B. J., H. A. Hofmann, D. M. Hillis and H. H. Zakon (2017). "Evolution of Animal Neural Systems." Annual Review of Ecology, Evolution, and Systematics 48(1): 377- 398.

Liu, C., N. Zhou, M.-X. Du, Y.-T. Sun, K. Wang, Y.-J. Wang, D.-H. Li, H.-Y. Yu, Y. Song, B.-B. Bai, Y. Xin, L. Wu, C.-Y. Jiang, J. Feng, H. Xiang, Y. Zhou, J. Ma, J. Wang, H.-W. Liu and S.-J. Liu (2020). "The Mouse Gut Microbial Biobank expands the coverage of cultured bacteria." Nature Communications 11(1): 79.

Liu, H., Y. Mishima, T. Fujiwara, H. Nagai, A. Kitazawa, Y. Mine, H. Kobayashi, X. Yao, J. Yamada, T. Oda and M. Namikoshi (2004). "Isolation of Araguspongine M, a New Stereoisomer of an Araguspongine/Xestospongin alkaloid, and Dopamine from the Marine Sponge Neopetrosia exigua Collected in Palau." Marine Drugs 2(4): 154-163.

Lloyd-Price, J., C. Arze, A. N. Ananthakrishnan, M. Schirmer, J. Avila-Pacheco, T. W. Poon, E. Andrews, N. J. Ajami, K. S. Bonham, C. J. Brislawn, D. Casero, H. Courtney, A. Gonzalez, T. G. Graeber, A. B. Hall, K. Lake, C. J. Landers, H. Mallick, D. R. Plichta, M. Prasad, G. Rahnavard, J. Sauk, D. Shungin, Y. Vázquez-Baeza, R. A. White, J. Bishai, K. Bullock, A. Deik, C. Dennis, J. L. Kaplan, H. Khalili, L. J. McIver, C. J. Moran, L. Nguyen, K. A. Pierce, R. Schwager, A. Sirota-Madi, B. W. Stevens, W. Tan, J. J. ten Hoeve, G. Weingart, R. G. Wilson, V. Yajnik, J. Braun, L. A. Denson, J. K. Jansson, R. Knight, S. Kugathasan, D. P. B. McGovern, J. F. Petrosino, T. S. Stappenbeck, H. S. Winter, C. B. Clish, E. A. Franzosa, H. Vlamakis, R. J. Xavier, C. Huttenhower and I. Investigators (2019). "Multi-omics of the gut microbial ecosystem in inflammatory bowel diseases." Nature 569(7758): 655-662.

Lloyd-Price, J., A. Mahurkar, G. Rahnavard, J. Crabtree, J. Orvis, A. B. Hall, A. Brady, H. H. Creasy, C. McCracken, M. G. Giglio, D. McDonald, E. A. Franzosa, R. Knight, O. White and C. Huttenhower (2017). "Strains, functions and dynamics in the expanded Human Microbiome Project." Nature 550(7674): 61-66.

Lomelino, C. L., J. T. Andring, R. McKenna and M. S. Kilberg (2017). "Asparagine synthetase: Function, structure, and role in disease." Journal of Biological Chemistry 292(49): 19952-19958.

Lønborg, C., C. Carreira, T. Jickells and X. A. Álvarez-Salgado (2020). "Impacts of Global Change on Ocean Dissolved Organic Carbon (DOC) Cycling." Frontiers in Marine Science 7(466).

Love, M. I., W. Huber and S. Anders (2014). "Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2." Genome Biology 15(12): 550.

Lu, Y. J., Y. M. Zhang and C. O. Rock (2004). "Product diversity and regulation of type II fatty acid synthases." Biochemistry and Cell Biology 82(1): 145-155.

Luo, W. (2014). "Pathview: pathway based data integration and visualization."

Lymperopoulos, A., G. Rengo and W. J. Koch (2013). "Adrenergic nervous system in heart failure: pathophysiology and therapy." Circulation research 113(6): 739-753.

144

Lyte, M., L. Vulchanova and D. R. Brown (2011). "Stress at the intestinal surface: catecholamines and mucosa–bacteria interactions." Cell and Tissue Research 343(1): 23-32.

Macia, L., R. Nanan, E. Hosseini-Beheshti and G. E. Grau (2019). "Host- and Microbiota- Derived Extracellular Vesicles, Immune Function, and Disease Development." International journal of molecular sciences 21(1): 107.

Maes, A., C. Gracia, N. Innocenti, K. Zhang, E. Aurell and E. Hajnsdorf (2017). "Landscape of RNA polyadenylation in E. coli." Nucleic Acids Res 45(5): 2746-2756.

Maldonado, M. (2015). "Sponge waste that fuels marine oligotrophic food webs: a re- assessment of its origin and nature." Marine Ecology 37(3): 477-491.

Mangano, S., C. Caruso, L. Michaud and A. Lo Giudice (2018). "First evidence of quorum sensing activity in bacteria associated with Antarctic sponges." Polar Biology 41(7): 1435- 1445.

Marchler-Bauer, A., Y. Bo, L. Y. Han, J. E. He, C. J. Lanczycki, S. N. Lu, F. Chitsaz, M. K. Derbyshire, R. C. Geer, N. R. Gonzales, M. Gwadz, D. I. Hurwitz, F. Lu, G. H. Marchler, J. S. Song, N. Thanki, Z. X. Wang, R. A. Yamashita, D. C. Zhang, C. J. Zheng, L. Y. Geer and S. H. Bryant (2017). "CDD/SPARCLE: functional classification of proteins via subfamily domain architectures." Nucleic Acids Research 45(D1): D200-D203.

Martin, C. R., V. Osadchiy, A. Kalani and E. A. Mayer (2018). "The Brain-Gut-Microbiome Axis." Cell Mol Gastroenterol Hepatol 6(2): 133-148.

Matson, V., J. Fessler, R. Bao, T. Chongsuwat, Y. Zha, M.-L. Alegre, J. J. Luke and T. F. Gajewski (2018). "The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients." Science (New York, N.Y.) 359(6371): 104-108.

Matthies, A., M. Blaut and A. Braune (2009). "Isolation of a Human Intestinal Bacterium Capable of Daidzein and Genistein Conversion." Applied and Environmental Microbiology 75(6): 1740-1744.

Mazzoli, R. and E. Pessione (2016). "The Neuro-endocrinological Role of Microbial Glutamate and GABA Signaling." Frontiers in Microbiology 7.

McFall-Ngai, M., M. G. Hadfield, T. C. G. Bosch, H. V. Carey, T. Domazet-Loso, A. E. Douglas, N. Dubilier, G. Eberl, T. Fukami, S. F. Gilbert, U. Hentschel, N. King, S. Kjelleberg, A. H. Knoll, N. Kremer, S. K. Mazmanian, J. L. Metcalf, K. Nealson, N. E. Pierce, J. F. Rawls, A. Reid, E. G. Ruby, M. Rumpho, J. G. Sanders, D. Tautz and J. J. Wernegreen (2013). "Animals in a bacterial world, a new imperative for the life sciences." Proceedings of the National Academy of Sciences of the United States of America 110(9): 3229-3236.

McFall-Ngai, M., E. A. Heath-Heckman, A. A. Gillette, S. M. Peyer and E. A. Harvie (2012). "The secret languages of coevolved symbioses: insights from the Euprymna scolopes-Vibrio fischeri symbiosis." Semin Immunol 24(1): 3-8.

145

Miller, G. M. (2011). "The emerging role of trace amine-associated receptor 1 in the functional regulation of monoamine transporters and dopaminergic activity." Journal of Neurochemistry 116(2): 164-176.

Mohamed, N. M., A. S. Colman, Y. Tal and R. T. Hill (2008). "Diversity and expression of nitrogen fixation genes in bacterial symbionts of marine sponges." Environ Microbiol 10(11): 2910-2921.

Moitinho-Silva, L., C. Diez-Vives, G. Batani, A. I. Esteves, M. T. Jahn and T. Thomas (2017). "Integrated metabolism in sponge-microbe symbiosis revealed by genome-centered metatranscriptomics." Isme journal 11: 1651–1666.

Moitinho-Silva, L., S. Nielsen, A. Amir, A. Gonzalez, G. L. Ackermann, C. Cerrano, C. Astudillo-Garcia, C. Easson, D. Sipkema, F. Liu, G. Steinert, G. Kotoulas, G. P. McCormack, G. Feng, J. J. Bell, J. Vicente, J. R. Björk, J. M. Montoya, J. B. Olson, J. Reveillaud, L. Steindler, M.-C. Pineda, M. V. Marra, M. Ilan, M. W. Taylor, P. Polymenakou, P. M. Erwin, P. J. Schupp, R. L. Simister, R. Knight, R. W. Thacker, R. Costa, R. T. Hill, S. Lopez- Legentil, T. Dailianis, T. Ravasi, U. Hentschel, Z. Li, N. S. Webster and T. Thomas (2017). "The sponge microbiome project." GigaScience 6(10).

Morioka, C., Y. Tachi, S. Suzuki and S. Itoh (2006). "Significant enhancement of monooxygenase activity of oxygen carrier protein hemocyanin by urea." Journal of the American Chemical Society 128(21): 6788-6789.

Moroz, L. L., K. M. Kocot, M. R. Citarella, S. Dosung, T. P. Norekian, I. S. Povolotskaya, A. P. Grigorenko, C. Dailey, E. Berezikov, K. M. Buckley, A. Ptitsyn, D. Reshetov, K. Mukherjee, T. P. Moroz, Y. Bobkova, F. Yu, V. V. Kapitonov, J. Jurka, Y. V. Bobkov, J. J. Swore, D. O. Girardo, A. Fodor, F. Gusev, R. Sanford, R. Bruders, E. Kittler, C. E. Mills, J. P. Rast, R. Derelle, V. V. Solovyev, F. A. Kondrashov, B. J. Swalla, J. V. Sweedler, E. I. Rogaev, K. M. Halanych and A. B. Kohn (2014). "The ctenophore genome and the evolutionary origins of neural systems." Nature 510(7503): 109-114.

Morrison, D. J. and T. Preston (2016). "Formation of short chain fatty acids by the gut microbiota and their impact on human metabolism." Gut Microbes 7(3): 189-200.

Müller, W. E., X. Wang and H. C. Schröder (2009). "Paleoclimate and evolution: emergence of sponges during the neoproterozoic." Prog Mol Subcell Biol 47: 55-77.

Munroe, S., K. Sandoval, D. E. Martens, D. Sipkema and S. A. Pomponi (2019). "Genetic algorithm as an optimization tool for the development of sponge cell culture media." In Vitro Cellular & Developmental Biology-Animal 55(3): 149-158.

Murillo-Rincon, A. P., A. Klimovich, E. Pemöller, J. Taubenheim, B. Mortzfeld, R. Augustin and T. C. G. Bosch (2017). "Spontaneous body contractions are modulated by the microbiome of Hydra." Scientific reports 7(1): 15937-15937.

Nakanishi, N., S. Sogabe and B. M. Degnan (2014). "Evolutionary origin of gastrulation: insights from sponge development." BMC Biology 12(1): 26.

146

Nakayama, Y., M. Hayashi and T. Unemoto (1998). "Identification of six subunits constituting Na+-translocating NADH-quinone reductase from the marine Vibrio alginolyticus." Febs Letters 422(2): 240-242.

Nayfach, S., Z. J. Shi, R. Seshadri, K. S. Pollard and N. C. Kyrpides (2019). "New insights from uncultivated genomes of the global human gut microbiome." Nature 568(7753): 505- 510.

Neave, N. (2007). Hormones and the endocrine system. Hormones and Behaviour: A Psychological Approach. N. Neave. Cambridge, Cambridge University Press: 22-47.

Newman, M. A., T. Sundelin, J. T. Nielsen and G. Erbs (2013). "MAMP (microbe-associated molecular pattern) triggered immunity in plants." Front Plant Sci 4: 139.

Nichols, S. A., B. W. Roberts, D. J. Richter, S. R. Fairclough and N. King (2012). "Origin of metazoan cadherin diversity and the antiquity of the classical cadherin/β-catenin complex." Proceedings of the National Academy of Sciences 109(32): 13046.

Nicholson, J. K., E. Holmes, J. Kinross, R. Burcelin, G. Gibson, W. Jia and S. Pettersson (2012). "Host-gut microbiota metabolic interactions." Science 336(6086): 1262-1267.

Nilsson, N. E., K. Kotarsky, C. Owman and B. Olde (2003). "Identification of a free fatty acid receptor, FFA2R, expressed on leukocytes and activated by short-chain fatty acids." Biochemical and Biophysical Research Communications 303(4): 1047-1052.

Noriega, C. E., H.-Y. Lin, L.-L. Chen, S. B. Williams and V. Stewart (2010). "Asymmetric cross-regulation between the nitrate-responsive NarX-NarL and NarQ-NarP two-component regulatory systems from Escherichia coli K-12." Molecular microbiology 75(2): 394-412.

O’Brien, P. A., S. Tan, C. Yang, P. R. Frade, N. Andreakis, H. A. Smith, D. J. Miller, N. S. Webster, G. Zhang and D. G. Bourne (2020). "Diverse coral reef invertebrates exhibit patterns of phylosymbiosis." The ISME Journal.

O’Brien, P. A., N. S. Webster, D. J. Miller and D. G. Bourne (2019). "Host-Microbe Coevolution: Applying Evidence from Model Systems to Complex Marine Invertebrate Holobionts." mBio 10(1): e02241-02218.

Pandey, S., J. Maharana and A. K. Shukla (2019). "The Gut Feeling: GPCRs Enlighten the Way." Cell Host & Microbe 26(2): 160-162.

Parker, A., M. A. E. Lawson, L. Vaux and C. Pin (2018). "Host-microbe interaction in the gastrointestinal tract." Environmental Microbiology 20(7): 2337-2353.

Parks, D. H., C. Rinke, M. Chuvochina, P.-A. Chaumeil, B. J. Woodcroft, P. N. Evans, P. Hugenholtz and G. W. Tyson (2017). "Recovery of nearly 8,000 metagenome-assembled genomes substantially expands the tree of life." Nature microbiology 2(11): 1533-1542.

Pasolli, E., F. Asnicar, S. Manara, M. Zolfo, N. Karcher, F. Armanini, F. Beghini, P. Manghi, A. Tett, P. Ghensi, M. C. Collado, B. L. Rice, C. DuLong, X. C. Morgan, C. D. Golden, C. Quince, C. Huttenhower and N. Segata (2019). "Extensive Unexplored Human Microbiome

147

Diversity Revealed by Over 150,000 Genomes from Metagenomes Spanning Age, Geography, and Lifestyle." Cell 176(3): 649-662.e620.

Pawlik, J. R., D. E. Burkepile and R. V. Thurber (2016). "A Vicious Circle? Altered Carbon and Nutrient Cycling May Explain the Low Resilience of Caribbean Coral Reefs." BioScience 66(6): 470-476.

Peng, Y.-C., C. Lu, G. Li, Z. Eichenbaum and C.-D. Lu (2017). "Induction of the pho regulon and polyphosphate synthesis against spermine stress in Pseudomonas aeruginosa." Molecular Microbiology 104(6): 1037-1051.

Perez-Ortin, J. E., P. Alepuz, S. Chavez and M. Choder (2013). "Eukaryotic mRNA decay: methodologies, pathways, and links to other stages of gene expression." J Mol Biol 425(20): 3750-3775.

Pertea, M., D. Kim, G. M. Pertea, J. T. Leek and S. L. Salzberg (2016). "Transcript-level expression analysis of RNA-seq experiments with HISAT, StringTie and Ballgown." Nature Protocols 11(9): 1650-1667.

Petrova, O. E., F. Garcia-Alcalde, C. Zampaloni and K. Sauer (2017). "Comparative evaluation of rRNA depletion procedures for the improved analysis of bacterial biofilm and mixed pathogen culture transcriptomes." Scientific Reports 7: 41114.

Pfannkuchen, M., G. B. Fritz, S. Schlesinger, K. Bayer and F. Brümmer (2009). "In situ pumping activity of the sponge Aplysina aerophoba, Nardo 1886." Journal of Experimental Marine Biology and Ecology 369(1): 65-71.

Pietschke, C., C. Treitz, S. Foret, A. Schultze, S. Kunzel, A. Tholey, T. C. G. Bosch and S. Fraune (2017). "Host modification of a bacterial quorum-sensing signal induces a phenotypic switch in bacterial symbionts." Proceedings of the National Academy of Sciences of the United States of America 114(40): E8488-E8497.

Pita, L., L. Rix, B. M. Slaby, A. Franke and U. Hentschel (2018). "The sponge holobiont in a changing ocean: from microbes to ecosystems." Microbiome 6.

Pokusaeva, K., C. Johnson, B. Luk, G. Uribe, Y. Fu, N. Oezguen, R. K. Matsunami, M. Lugo, A. Major, Y. Mori-Akiyama, E. B. Hollister, S. M. Dann, X. Z. Shi, D. A. Engler, T. Savidge and J. Versalovic (2017). "GABA-producing Bifidobacterium dentium modulates visceral sensitivity in the intestine." Neurogastroenterology and motility : the official journal of the European Gastrointestinal Motility Society 29(1): e12904.

Ponnudurai, R., M. Kleiner, L. Sayavedra, J. M. Petersen, M. Moche, A. Otto, D. Becher, T. Takeuchi, N. Satoh, N. Dubilier, T. Schweder and S. Markert (2017). "Metabolic and physiological interdependencies in the Bathymodiolus azoricus symbiosis." ISME J 11(2): 463-477.

Poretsky, R., L. M. Rodriguez-R, C. W. Luo, D. Tsementzi and K. T. Konstantinidis (2014). "Strengths and Limitations of 16S rRNA Gene Amplicon Sequencing in Revealing Temporal Microbial Community Dynamics." Plos One 9(4).

148

Postma, P. W., J. W. Lengeler and G. R. Jacobson (1993). "Phosphoenolpyruvate:carbohydrate phosphotransferase systems of bacteria." Microbiological reviews 57(3): 543-594.

Pradeu, T. (2011). "A Mixed Self: The Role of Symbiosis in Development." Biological Theory 6(1): 80-88.

Priya, N., P. Ranjan, S. M. Sappal and A. L. Ramanathan (2017). 22 - Reactive Nitrogen Dynamics in the Mangroves of India. The Indian Nitrogen Assessment. Y. P. Abrol, T. K. Adhya, V. P. Aneja et al., Elsevier: 335-359.

Proctor, L. M., H. H. Creasy, J. M. Fettweis, J. Lloyd-Price, A. Mahurkar, W. Zhou, G. A. Buck, M. P. Snyder, J. F. Strauss, G. M. Weinstock, O. White, C. Huttenhower and H. M. P. R. N. C. The Integrative (2019). "The Integrative Human Microbiome Project." Nature 569(7758): 641-648.

Putnam, N. H., B. L. O'Connell, J. C. Stites, B. J. Rice, M. Blanchette, R. Calef, C. J. Troll, A. Fields, P. D. Hartley, C. W. Sugnet, D. Haussler, D. S. Rokhsar and R. E. Green (2016). "Chromosome-scale shotgun assembly using an in vitro method for long-range linkage." Genome Research 26(3): 342-350.

Quinlan, A. R. and I. M. Hall (2010). "BEDTools: a flexible suite of utilities for comparing genomic features." Bioinformatics 26(6): 841-842.

Ran, F. A., P. D. Hsu, J. Wright, V. Agarwala, D. A. Scott and F. Zhang (2013). "Genome engineering using the CRISPR-Cas9 system." Nature Protocols 8(11): 2281-2308.

Rea, K., T. G. Dinan and J. F. Cryan (2016). "The microbiome: A key regulator of stress and neuroinflammation." Neurobiology of Stress 4: 23-33.

Reeds, P. J. (2000). "Dispensable and indispensable amino acids for humans." Journal of Nutrition 130(7): 1835s-1840s.

Reen, F. J., J. A. Gutiérrez-Barranquero, R. R. McCarthy, D. F. Woods, S. Scarciglia, C. Adams, K. Fog Nielsen, L. Gram and F. O’Gara (2019). "Quorum Sensing Signaling Alters Virulence Potential and Population Dynamics in Complex Microbiome-Host Interactomes." Frontiers in Microbiology 10(2131).

Reiswig, H. M. (1974). "Water transport, respiration and energetics of three tropical marine sponges." Journal of experimental marine Biology and Ecology 14(3): 231-249.

Rémy, B., S. Mion, L. Plener, M. Elias, E. Chabrière and D. Daudé (2018). "Interference in Bacterial Quorum Sensing: A Biopharmaceutical Perspective." Frontiers in Pharmacology 9(203).

Repeta, D. J. (2015). Chapter 2 - Chemical Characterization and Cycling of Dissolved Organic Matter. Biogeochemistry of Marine Dissolved Organic Matter (Second Edition). D. A. Hansell and C. A. Carlson. Boston, Academic Press: 21-63.

149

Repeta, D. J. and R. M. Boiteau (2017). 7 Organic Nutrient Chemistry and the Marine Microbiome. The Chemistry of Microbiomes: Proceedings of a Seminar Series, National Academies Press (US).

Reshetnikov, A. S., O. N. Rozova, V. N. Khmelenina, Mustakhimov, II, A. P. Beschastny, J. C. Murrell and Y. A. Trotsenko (2008). "Characterization of the pyrophosphate-dependent 6- phosphofructokinase from Methylococcus capsulatus Bath." FEMS Microbiol Lett 288(2): 202-210.

Reveillaud, J., L. Maignien, A. M. Eren, J. A. Huber, A. Apprill, M. L. Sogin and A. Vanreusel (2014). "Host-specificity among abundant and rare taxa in the sponge microbiome." Isme Journal 8(6): 1198-1209.

Ribes, M., C. Dziallas, R. Coma and L. Riemann (2015). "Microbial Diversity and Putative Diazotrophy in High- and Low-Microbial-Abundance Mediterranean Sponges." Applied and environmental microbiology 81(17): 5683-5693.

Riesgo, A., N. Farrar, P. J. Windsor, G. Giribet and S. P. Leys (2014). "The Analysis of Eight Transcriptomes from All Poriferan Classes Reveals Surprising Genetic Complexity in Sponges." Molecular Biology and Evolution 31(5): 1102-1120.

Riesgo, A., K. Peterson, C. Richardson, T. Heist, B. Strehlow, M. McCauley, C. Cotman, M. Hill and A. Hill (2014). "Transcriptomic analysis of differential host gene expression upon uptake of symbionts: a case study with Symbiodinium and the major bioeroding sponge Cliona varians." Bmc Genomics 15.

Rivera, A. S., J. U. Hammel, K. M. Haen, E. S. Danka, B. Cieniewicz, I. P. Winters, D. Posfai, G. Wörheide, D. V. Lavrov and S. W. Knight (2011). "RNA interference in marine and freshwater sponges: actin knockdown in Tethya wilhelma and Ephydatia muelleriby ingested dsRNA expressing bacteria." BMC biotechnology 11(1): 67.

Rix, L., J. M. de Goeij, C. E. Mueller, U. Struck, J. J. Middelburg, F. C. van Duyl, F. A. Al- Horani, C. Wild, M. S. Naumann and D. van Oevelen (2016). "Coral mucus fuels the sponge loop in warm- and cold-water coral reef ecosystems." Sci Rep 6: 18715.

Rix, L., J. M. de Goeij, D. van Oevelen, U. Struck, F. A. Al-Horani, C. Wild and M. S. Naumann (2017). "Differential recycling of coral and algal dissolved organic matter via the sponge loop." Functional Ecology 31(3): 778-789.

Rix, L., M. Ribes, R. Coma, M. T. Jahn, J. M. de Goeij, D. van Oevelen, S. Escrig, A. Meibom and U. Hentschel (2020). "Heterotrophy in the earliest gut: a single-cell view of heterotrophic carbon and nitrogen assimilation in sponge-microbe symbioses." The ISME Journal.

Rooks, M. G., P. Veiga, A. Z. Reeves, S. Lavoie, K. Yasuda, Y. Asano, K. Yoshihara, M. Michaud, L. Wardwell-Scott, C. A. Gallini, J. N. Glickman, N. Sudo, C. Huttenhower, C. F. Lesser and W. S. Garrett (2017). "QseC inhibition as an antivirulence approach for colitis- associated bacteria." Proceedings of the National Academy of Sciences of the United States of America 114(1): 142-147.

150

Rosenberg, E., A. Kushmaro, E. Kramarsky-Winter, E. Banin and L. Yossi (2009). "The role of microorganisms in coral bleaching." The ISME Journal 3(2): 139-146.

Rosenberg, E. and I. Zilber-Rosenberg (2016). "Microbes Drive Evolution of Animals and Plants: the Hologenome Concept." Mbio 7(2).

Ross, A. A., K. M. Müller, J. S. Weese and J. D. Neufeld (2018). "Comprehensive skin microbiome analysis reveals the uniqueness of human skin and evidence for phylosymbiosis within the class Mammalia." Proceedings of the National Academy of Sciences 115(25): E5786-E5795.

Ross, A. A., A. Rodrigues Hoffmann and J. D. Neufeld (2019). "The skin microbiome of vertebrates." Microbiome 7(1): 79.

Routy, B., E. Le Chatelier, L. Derosa, C. P. M. Duong, M. T. Alou, R. Daillere, A. Fluckiger, M. Messaoudene, C. Rauber, M. P. Roberti, M. Fidelle, C. Flament, V. Poirier-Colame, P. Opolon, C. Klein, K. Iribarren, L. Mondragon, N. Jacquelot, B. Qu, G. Ferrere, C. Clemenson, L. Mezquita, J. R. Masip, C. Naltet, S. Brosseau, C. Kaderbhai, C. Richard, H. Rizvi, F. Levenez, N. Galleron, B. Quinquis, N. Pons, B. Ryffel, V. Minard-Colin, P. Gonin, J. C. Soria, E. Deutsch, Y. Loriot, F. Ghiringhelli, G. Zalcman, F. Goldwasser, B. Escudier, M. D. Hellmann, A. Eggermont, D. Raoult, L. Albiges, G. Kroemer and L. Zitvogel (2018). "Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors." Science 359(6371): 91-97.

Rowland, I., G. Gibson, A. Heinken, K. Scott, J. Swann, I. Thiele and K. Tuohy (2017). "Gut microbiota functions: metabolism of nutrients and other food components." European Journal of Nutrition 56: 1-24.

Ryu, T., L. Seridi, L. Moitinho-Silva, M. Oates, Y. J. Liew, C. Mavromatis, X. Wang, A. Haywood, F. F. Lafi, M. Kupresanin, R. Sougrat, M. A. Alzahrani, E. Giles, Y. Ghosheh, C. Schunter, S. Baumgarten, M. L. Berumen, X. Gao, M. Aranda, S. Foret, J. Gough, C. R. Voolstra, U. Hentschel and T. Ravasi (2016). "Hologenome analysis of two marine sponges with different microbiomes." Bmc Genomics 17.

Sacristán-Soriano, O., M. Winkler, P. Erwin, J. Weisz, O. Harriott, G. Heussler, E. Bauer, B. West Marsden, A. Hill and M. Hill (2019). "Ontogeny of symbiont community structure in two carotenoid-rich, viviparous marine sponges: comparison of microbiomes and analysis of culturable pigmented heterotrophic bacteria." Environmental Microbiology Reports 11(2): 249-261.

Saier, M. H. and W. T. Jenkins (1967). "Alanine aminotransferase I. Purification and properties." Journal of Biological Chemistry 242(1): 91-100.

Sanderson, S. J., S. S. KHAN, R. Graham McCARTNEY, C. MILLER and J. G. LINDSAY (1996). "Reconstitution of mammalian pyruvate dehydrogenase and 2-oxoglutarate dehydrogenase complexes: analysis of protein X involvement and interaction of homologous and heterologous dihydrolipoamide dehydrogenases." Biochemical Journal 319(1): 109-116.

151

Santos, A. A., S. S. Venceslau, F. Grein, W. D. Leavitt, C. Dahl, D. T. Johnston and I. A. C. Pereira (2015). "A protein trisulfide couples dissimilatory sulfate reduction to energy conservation." Science 350(6267): 1541.

Santos-Beneit, F. (2015). "The Pho regulon: a huge regulatory network in bacteria." Frontiers in Microbiology 6.

Saurav, K., N. Borbone, I. Burgsdorf, R. Teta, A. Caso, R. Bar-Shalom, G. Esposito, M. Britstein, L. Steindler and V. Costantino (2020). "Identification of Quorum Sensing Activators and Inhibitors in The Marine Sponge Sarcotragus spinosulus." Marine drugs 18(2): 127.

Say, T. E. and S. M. Degnan (2020). "Molecular and behavioural evidence that interdependent photo - and chemosensory systems regulate larval settlement in a marine sponge." Molecular Ecology 29(2): 247-261.

Schaechinger, T. J. and D. Oliver (2007). "Nonmammalian orthologs of prestin (SLC26A5) are electrogenic divalent/chloride anion exchangers." Proceedings of the National Academy of Sciences 104(18): 7693.

Schippers, K. J., D. E. Martens, S. A. Pomponi and R. H. Wijffels (2011). "Cell cycle analysis of primary sponge cell cultures." In vitro cellular & developmental biology. Animal 47(4): 302-311.

Schmieder, R., Y. W. Lim and R. Edwards (2012). "Identification and removal of ribosomal RNA sequences from metatranscriptomes." Bioinformatics 28(3): 433-435.

Schmitt, S., H. Angermeier, R. Schiller, N. Lindquist and U. Hentschel (2008). "Molecular microbial diversity survey of sponge reproductive stages and mechanistic insights into vertical transmission of microbial symbionts." Appl. Environ. Microbiol. 74: 7694-7708.

Schröder, I., C. D. Wolin, R. Cavicchioli and R. P. Gunsalus (1994). "Phosphorylation and dephosphorylation of the NarQ, NarX, and NarL proteins of the nitrate-dependent two- component regulatory system of Escherichia coli." Journal of Bacteriology 176(16): 4985.

Searcy-Bernal, R. and C. Anguiano-Beltran (1998). "Optimizing the concentration of gamma-aminobutyric acid (GABA) for inducing larval metamorphosis in the red abalone Haliotis rufescens (Mollusca : Gastropoda)." Journal of the World Aquaculture Society 29(4): 463-470.

Sebé-Pedrós, A., E. Chomsky, K. Pang, D. Lara-Astiaso, F. Gaiti, Z. Mukamel, I. Amit, A. Hejnol, B. M. Degnan and A. Tanay (2018). "Early metazoan cell type diversity and the evolution of multicellular gene regulation." Nature Ecology & Evolution 2(7): 1176-1188.

Seemann, T. (2014). "Prokka: rapid prokaryotic genome annotation." Bioinformatics 30(14): 2068-2069.

Selinger, D. W., R. M. Saxena, K. J. Cheung, G. M. Church and C. Rosenow (2003). "Global RNA half-life analysis in Escherichia coli reveals positional patterns of transcript degradation." Genome Research 13(2): 216-223.

152

Sender, R., S. Fuchs and R. Milo (2016). "Are We Really Vastly Outnumbered? Revisiting the Ratio of Bacterial to Host Cells in Humans." Cell 164(3): 337-340.

Seshadri, R., S. C. Leahy, G. T. Attwood, K. H. Teh, S. C. Lambie, A. L. Cookson, E. A. Eloe-Fadrosh, G. A. Pavlopoulos, M. Hadjithomas, N. J. Varghese, D. Paez-Espino, N. Palevich, P. H. Janssen, R. S. Ronimus, S. Noel, P. Soni, K. Reilly, T. Atherly, C. Ziemer, A.-D. Wright, S. Ishaq, M. Cotta, S. Thompson, K. Crosley, N. McKain, R. J. Wallace, H. J. Flint, J. C. Martin, R. J. Forster, R. J. Gruninger, T. McAllister, R. Gilbert, D. Ouwerkerk, A. Klieve, R. A. Jassim, S. Denman, C. McSweeney, C. Rosewarne, S. Koike, Y. Kobayashi, M. Mitsumori, T. Shinkai, S. Cravero, M. C. Cucchi, R. Perry, G. Henderson, C. J. Creevey, N. Terrapon, P. Lapebie, E. Drula, V. Lombard, E. Rubin, N. C. Kyrpides, B. Henrissat, T. Woyke, N. N. Ivanova, W. J. Kelly and c. Hungate project (2018). "Cultivation and sequencing of rumen microbiome members from the Hungate1000 Collection." Nature Biotechnology 36(4): 359-367.

Sharova, L. V., A. A. Sharov, T. Nedorezov, Y. Piao, N. Shaik and M. S. H. Ko (2009). "Database for mRNA half-life of 19 977 genes obtained by DNA microarray analysis of pluripotent and differentiating mouse embryonic stem cells." DNA research : an international journal for rapid publication of reports on genes and genomes 16(1): 45-58.

Shelp, B. J., A. W. Bown and M. D. McLean (1999). "Metabolism and functions of gamma- aminobutyric acid." Trends in Plant Science 4(11): 446-452.

Shigeta, Y., Y. Kanai, A. Chairoungdua, N. Ahmed, S. Sakamoto, H. Matsuo, D. K. Kim, M. Fujimura, N. Anzai, K. Mizoguchi, T. Ueda, K. Akakura, T. Ichikawa, H. Ito and H. Endou (2006). "A novel missense mutation of SLC7A9 frequent in Japanese cystinuria cases affecting the C-terminus of the transporter." Kidney Int 69(7): 1198-1206.

Shih, J. L., K. E. Selph, C. B. Wall, N. J. Wallsgrove, M. P. Lesser and B. N. Popp (2020). "Trophic Ecology of the Tropical Pacific Sponge Mycale grandis Inferred from Amino Acid Compound-Specific Isotopic Analyses." Microbial Ecology 79(2): 495-510.

Shin, J., G.-l. Ming and H. Song (2014). "Decoding neural transcriptomes and epigenomes via high-throughput sequencing." Nature Neuroscience 17(11): 1463-1475.

Siegl, A., J. Kamke, T. Hochmuth, J. Piel, M. Richter, C. G. Liang, T. Dandekar and U. Hentschel (2011). "Single-cell genomics reveals the lifestyle of Poribacteria, a candidate phylum symbiotically associated with marine sponges." Isme Journal 5(1): 61-70.

Sievers, F. and D. G. Higgins (2014). "Clustal Omega, Accurate Alignment of Very Large Numbers of Sequences." Multiple Sequence Alignment Methods 1079: 105-116.

Silva, Y. P., A. Bernardi and R. L. Frozza (2020). "The role of short-chain fatty acids from gut microbiota in gut-brain communication." Frontiers in Endocrinology 11: 25.

Simao, F. A., R. M. Waterhouse, P. Ioannidis, E. V. Kriventseva and E. M. Zdobnov (2015). "BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs." Bioinformatics 31(19): 3210-3212.

153

Simon, J. C., J. R. Marchesi, C. Mougel and M. A. Selosse (2019). "Host-microbiota interactions: from holobiont theory to analysis." Microbiome 7.

Sims, D., I. Sudbery, N. E. Ilott, A. Heger and C. P. Ponting (2014). "Sequencing depth and coverage: key considerations in genomic analyses." Nat Rev Genet 15(2): 121-132.

Sipkema, D., S. de Caralt, J. A. Morillo, W. A. Al-Soud, S. J. Sørensen, H. Smidt and M. J. Uriz (2015). "Similar sponge-associated bacteria can be acquired via both vertical and horizontal transmission." Environmental Microbiology 17(10): 3807-3821. siTOOLs. "riboPOOL Efficient, Affordable Ribosomal RNA Depletion for Any Species." from https://www.sitoolsbiotech.com/ribopools.php.

Sivan, A., L. Corrales, N. Hubert, J. B. Williams, K. Aquino-Michaels, Z. M. Earley, F. W. Benyamin, Y. Man Lei, B. Jabri, M.-L. Alegre, E. B. Chang and T. F. Gajewski (2015). "Commensal <em>Bifidobacterium</em> promotes antitumor immunity and facilitates anti–PD-L1 efficacy." Science 350(6264): 1084.

Slaby, B. M., A. Franke, L. Rix, L. Pita, K. Bayer, M. T. Jahn and U. Hentschel (2019). Marine Sponge Holobionts in Health and Disease. Symbiotic Microbiomes of Coral Reefs Sponges and Corals. Z. Li. Dordrecht, Springer Netherlands: 81-104.

Slaby, B. M., T. Hackl, H. Horn, K. Bayer and U. Hentschel (2017). "Metagenomic binning of a marine sponge microbiome reveals unity in defense but metabolic specialization." Isme Journal 11(11): 2465-2478.

Sobko, T., L. Huang, T. Midtvedt, E. Norin, L. E. Gustafsson, M. Norman, E. Å. Jansson and J. O. Lundberg (2006). "Generation of NO by probiotic bacteria in the gastrointestinal tract." Free Radical Biology and Medicine 41(6): 985-991.

Sogabe, S., W. L. Hatleberg, K. M. Kocot, T. E. Say, D. Stoupin, K. E. Roper, S. L. Fernandez-Valverde, S. M. Degnan and B. M. Degnan (2019). "Pluripotency and the origin of animal multicellularity." Nature 570(7762): 519-522.

Sogabe, S., N. Nakanishi and B. M. Degnan (2016). "The ontogeny of choanocyte chambers during metamorphosis in the demosponge Amphimedon queenslandica." Evodevo 7.

Song, H., O. H. Hewitt and S. M. Degnan (2020). "Bacterial symbionts in animal development: arginine biosynthesis complementation enables larval settlement in a marine sponge." bioRxiv: 2020.2008.2006.240770.

Song, S. J., J. G. Sanders, F. Delsuc, J. Metcalf, K. Amato, M. W. Taylor, F. Mazel, H. L. Lutz, K. Winker, G. R. Graves, G. Humphrey, J. A. Gilbert, S. J. Hackett, K. P. White, H. R. Skeen, S. M. Kurtis, J. Withrow, T. Braile, M. Miller, K. G. McCracken, J. M. Maley, V. O. Ezenwa, A. Williams, J. M. Blanton, V. J. McKenzie and R. Knight (2020). "Comparative Analyses of Vertebrate Gut Microbiomes Reveal Convergence between Birds and Bats." mBio 11(1): e02901-02919.

154

Soto, E. J. L., L. O. Gambino and E. R. Mustafa (2014). "Free fatty acid receptor 3 is a key target of short chain fatty acid What is the impact on the sympathetic nervous system?" Channels 8(3).

Spencer, J. B., N. J. Stolowich, C. A. Roessner and A. I. Scott (1993). "The Escherichia coli cysG gene encodes the multifunctional protein, siroheme synthase." FEBS Lett 335(1): 57- 60.

Sperandio, V., A. G. Torres, B. Jarvis, J. P. Nataro and J. B. Kaper (2003). "Bacteria-host communication: the language of hormones." Proc Natl Acad Sci U S A 100(15): 8951-8956.

Sperandio, V., A. G. Torres, B. Jarvis, J. P. Nataro and J. B. Kaper (2003). "Bacteria–host communication: The language of hormones." Proceedings of the National Academy of Sciences 100(15): 8951.

Srivastava, M., O. Simakov, J. Chapman, B. Fahey, M. E. Gauthier, T. Mitros, G. S. Richards, C. Conaco, M. Dacre, U. Hellsten, C. Larroux, N. H. Putnam, M. Stanke, M. Adamska, A. Darling, S. M. Degnan, T. H. Oakley, D. C. Plachetzki, Y. Zhai, M. Adamski, A. Calcino, S. F. Cummins, D. M. Goodstein, C. Harris, D. J. Jackson, S. P. Leys, S. Shu, B. J. Woodcroft, M. Vervoort, K. S. Kosik, G. Manning, B. M. Degnan and D. S. Rokhsar (2010). "The Amphimedon queenslandica genome and the evolution of animal complexity." Nature 466(7307): 720-726.

Steinert, G., K. Busch, K. Bayer, S. Kodami, P. M. Arbizu, M. Kelly, S. Mills, D. Erpenbeck, M. Dohrmann, G. Wörheide, U. Hentschel and P. J. Schupp (2020). "Compositional and Quantitative Insights Into Bacterial and Archaeal Communities of South Pacific Deep-Sea Sponges (Demospongiae and Hexactinellida)." Frontiers in Microbiology 11(716).

Stewart, R. D., M. D. Auffret, A. Warr, A. W. Walker, R. Roehe and M. Watson (2019). "Compendium of 4,941 rumen metagenome-assembled genomes for rumen microbiome biology and enzyme discovery." Nature biotechnology 37(8): 953-961.

Stewart, R. D., M. D. Auffret, A. Warr, A. H. Wiser, M. O. Press, K. W. Langford, I. Liachko, T. J. Snelling, R. J. Dewhurst, A. W. Walker, R. Roehe and M. Watson (2018). "Assembly of 913 microbial genomes from metagenomic sequencing of the cow rumen." Nature Communications 9(1): 870.

Strandwitz, P. (2018). "Neurotransmitter modulation by the gut microbiota." Brain Res 1693(Pt B): 128-133.

Strandwitz, P., K. H. Kim, D. Terekhova, J. K. Liu, A. Sharma, J. Levering, D. McDonald, D. Dietrich, T. R. Ramadhar, A. Lekbua, N. Mroue, C. Liston, E. J. Stewart, M. J. Dubin, K. Zengler, R. Knight, J. A. Gilbert, J. Clardy and K. Lewis (2019). "GABA-modulating bacteria of the human gut microbiota." Nature Microbiology 4(3): 396-403.

Sun, X. J., Q. Li, H. Yu and L. F. Kong (2014). "The effect of chemical cues on the settlement of sea cucumber (Apostichopus japonicus) larvae." Journal of Ocean University of China 13(2): 321-330.

155

Takami, H., T. Kawamura and Y. Yamashita (2002). "Effects of delayed metamorphosis on larval competence, and postlarval survival and growth of abalone Haliotis discus hannai." Aquaculture 213(1-4): 311-322.

Tannock, G. W., B. Lawley, K. Munro, I. M. Sims, J. Lee, C. A. Butts and N. Roy (2014). "RNA-Stable-Isotope Probing Shows Utilization of Carbon from Inulin by Specific Bacterial Populations in the Rat Large Bowel." Applied and Environmental Microbiology 80(7): 2240- 2247.

Theis, K. R., N. M. Dheilly, J. L. Klassen, R. M. Brucker, J. F. Baines, T. C. G. Bosch, J. F. Cryan, S. F. Gilbert, C. J. Goodnight, E. A. Lloyd, J. Sapp, P. Vandenkoornhuyse, I. Zilber- Rosenberg, E. Rosenberg and S. R. Bordenstein (2016). "Getting the Hologenome Concept Right: an Eco-Evolutionary Framework for Hosts and Their Microbiomes." mSystems 1(2): e00028-00016.

Thomas, T., L. Moitinho-Silva, M. Lurgi, J. R. Björk, C. Easson, C. Astudillo-García, J. B. Olson, P. M. Erwin, S. López-Legentil, H. Luter, A. Chaves-Fonnegra, R. Costa, P. J. Schupp, L. Steindler, D. Erpenbeck, J. Gilbert, R. Knight, G. Ackermann, J. Victor Lopez, M. W. Taylor, R. W. Thacker, J. M. Montoya, U. Hentschel and N. S. Webster (2016). "Diversity, structure and convergent evolution of the global sponge microbiome." Nature Communications 7: 11870.

Thomas, T., D. Rusch, M. Z. DeMaere, P. Y. Yung, M. Lewis, A. Halpern, K. B. Heidelberg, S. Egan, P. D. Steinberg and S. Kjelleberg (2010). "Functional genomic signatures of sponge bacteria reveal unique and shared features of symbiosis." Isme Journal 4(12): 1557-1567.

Tian, R. M., Y. Wang, S. Bougouffa, Z. M. Gao, L. Cai, V. Bajic and P. Y. Qian (2014). "Genomic analysis reveals versatile heterotrophic capacity of a potentially symbiotic sulfur- oxidizing bacterium in sponge." Environ Microbiol 16(11): 3548-3561.

Tierney, B. T., Z. Yang, J. M. Luber, M. Beaudin, M. C. Wibowo, C. Baek, E. Mehlenbacher, C. J. Patel and A. D. Kostic (2019). "The Landscape of Genetic Content in the Gut and Oral Human Microbiome." Cell Host & Microbe 26(2): 283-295.e288.

Tretter, L., A. Patocs and C. Chinopoulos (2016). "Succinate, an intermediate in metabolism, signal transduction, ROS, hypoxia, and tumorigenesis." Biochim Biophys Acta 1857(8): 1086-1101.

Tripathy, B. C., I. Sherameti and R. Oelmuller (2010). "Siroheme: an essential component for life on earth." Plant Signal Behav 5(1): 14-20.

Tully, B. J., E. D. Graham and J. F. Heidelberg (2018). "The reconstruction of 2,631 draft metagenome-assembled genomes from the global oceans." Scientific data 5: 170203.

Upadhyay, S. K., S. Sharma, H. Singh, S. Dixit, J. Kumar, P. C. Verma and K. Chandrashekar (2015). "Whitefly Genome Expression Reveals Host-Symbiont Interaction in Amino Acid Biosynthesis." Plos One 10(5).

156

Vacelet, J. and C. Donadey (1977). "Electron microscope study of the association between some sponges and bacteria." Journal of Experimental Marine Biology and Ecology 30(3): 301-314. van de Guchte, M., H. M. Blottière and J. Doré (2018). "Humans as holobionts: implications for prevention and therapy." Microbiome 6(1): 81. van de Water, J. A. J. M., D. Allemand and C. Ferrier-Pagès (2018). "Host-microbe interactions in octocoral holobionts - recent advances and perspectives." Microbiome 6(1): 64-64. van Duyl, F. C., J. Hegeman, A. Hoogstraten and C. Maier (2008). "Dissolved carbon fixation by sponge–microbe consortia of deep water coral mounds in the northeastern Atlantic Ocean." Marine Ecology Progress Series 358: 137-150. van Niel, G., G. D'Angelo and G. Raposo (2018). "Shedding light on the cell biology of extracellular vesicles." Nature Reviews Molecular Cell Biology 19(4): 213-228.

Vanwonterghem, I. and N. S. Webster (2020). "Coral Reef Microorganisms in a Changing Climate." iScience 23(4): 100972.

Vega, J. M. and R. H. Garrett (1975). "Siroheme: a prosthetic group of the Neurospora crassa assimilatory nitrite reductase." J Biol Chem 250(20): 7980-7989.

Venos, E. S., M. H. Knodel, C. L. Radford and B. J. Berger (2004). "Branched-chain amino acid aminotransferase and methionine formation in Mycobacterium tuberculosis." BMC Microbiology 4(1): 39.

Venturi, V. and S. Subramoni (2009). "Future research trends in the major chemical language of bacteria." HFSP journal 3(2): 105-116.

Verkhovsky, M. I. and A. V. Bogachev (2010). "Sodium-translocating NADH:quinone oxidoreductase as a redox-driven ion pump." Biochimica Et Biophysica Acta-Bioenergetics 1797(6-7): 738-746.

Walker, A. W., S. H. Duncan, P. Louis and H. J. Flint (2014). "Phylogeny, culturing, and metagenomics of the human gut microbiota." Trends in Microbiology 22(5): 267-274.

Walshaw, D. L., S. Lowthorpe, A. East and P. S. Poole (1997). "Distribution of a sub-class of bacterial ABC polar amino acid transporter and identification of an N-terminal region involved in solute specificity." FEBS letters 414(2): 397-401.

Wang, R., J.-Q. Lin, X.-M. Liu, X. Pang, C.-J. Zhang, C.-L. Yang, X.-Y. Gao, C.-M. Lin, Y.- Q. Li, Y. Li, J.-Q. Lin and L.-X. Chen (2019). "Sulfur Oxidation in the Acidophilic Autotrophic Acidithiobacillus spp." Frontiers in Microbiology 9(3290).

Watson, J. R., J. O. Krömer, B. M. Degnan and S. M. Degnan (2017). "Seasonal changes in environmental nutrient availability and biomass composition in a coral reef sponge." Marine Biology 164(6): 135.

157

Webster, N. S. and T. Thomas (2016). "The Sponge Hologenome." MBio 7(2): e00135- 00116.

Weigel, B. L. and P. M. Erwin (2017). "Effects of reciprocal transplantation on the microbiome and putative nitrogen cycling functions of the intertidal sponge, Hymeniacidon heliophila." Scientific Reports 7: 43247.

Weisz, J. B., U. Hentschel, N. Lindquist and C. S. Martens (2007). "Linking abundance and diversity of sponge-associated microbial communities to metabolic differences in host sponges." Marine Biology 152(2): 475-483.

Weisz, J. B., N. Lindquist and C. S. Martens (2008). "Do associated microbial abundances impact marine demosponge pumping rates and tissue densities?" Oecologia 155(2): 367-376.

Westermann, A. J., L. Barquist and J. Vogel (2017). "Resolving host-pathogen interactions by dual RNA-seq." Plos Pathogens 13(2).

Westermann, A. J., K. U. Forstner, F. Amman, L. Barquist, Y. J. Chao, L. N. Schulte, L. Muller, R. Reinhardt, P. F. Stadler and J. Vogel (2016). "Dual RNA-seq unveils noncoding RNA functions in host-pathogen interactions." Nature 529(7587): 496-+.

Westermann, A. J., S. A. Gorski and J. Vogel (2012). "Dual RNA-seq of pathogen and host." Nature Reviews Microbiology 10(9): 618-630.

White, J. P., J. Prell, V. K. Ramachandran and P. S. Poole (2009). "Characterization of a {gamma}-aminobutyric acid transport system of Rhizobium leguminosarum bv. viciae 3841." J Bacteriol 191(5): 1547-1555.

Wieland, O. H. (1982). "The Mammalian Pyruvate-Dehydrogenase Complex - Structure and Regulation." Reviews of Physiology Biochemistry and Pharmacology 96: 123-170.

Wilkinson, C. R. and P. Fay (1979). "Nitrogen fixation in coral reef sponges with symbiotic cyanobacteria." Nature 279(5713): 527-529.

Wilkinson, C. R., R. E. Summons and E. Evans (1999). "Nitrogen fixation in symbiotic marine sponges: ecological significance and difficulties in detection." Memoirs of the Queensland Museum 44(1-2): 667-673.

Wilson, J. E. (2003). "Isozymes of mammalian hexokinase: structure, subcellular localization and metabolic function." Journal of Experimental Biology 206(12): 2049-2057.

Wolf, T., P. Kammer, S. Brunke and J. Linde (2018). "Two's company: studying interspecies relationships with dual RNA-seq." Current Opinion in Microbiology 42: 7-12.

Wong, E. (2020). Early evolution of the nervous system: Insights from synaptic sub- machineries and sensory behaviour in the aneural sponge, Amphimedon queenslandica. Doctor of Philosophy, The University of Queensland.

158

Wong, E., J. Molter, V. Anggono, S. M. Degnan and B. M. Degnan (2019). "Co-expression of synaptic genes in the sponge Amphimedon queenslandica uncovers ancient neural submodules." Scientific Reports 9.

Wu, S., H. Ou, T. Liu, D. Wang and J. Zhao (2018). "Structure and dynamics of microbiomes associated with the marine sponge Tedania sp. during its life cycle." FEMS Microbiology Ecology 94(5).

Yang, E., E. van Nimwegen, M. Zavolan, N. Rajewsky, M. Schroeder, M. Magnasco and J. E. Darnell (2003). "Decay rates of human mRNAs: Correlation with functional characteristics and sequence attributes." Genome Research 13(8): 1863-1872.

Yang, Y., J. Sun, Y. Sun, Y. H. Kwan, W. C. Wong, Y. Zhang, T. Xu, D. Feng, Y. Zhang, J.- W. Qiu and P.-Y. Qian (2020). "Genomic, transcriptomic, and proteomic insights into the symbiosis of deep-sea tubeworm holobionts." The ISME Journal 14(1): 135-150.

Yen, J. H. and A. R. Barr (1971). "New hypothesis of the cause of cytoplasmic incompatibility in Culex pipiens L." Nature 232(5313): 657-658.

Youngblut, N. D., G. H. Reischer, W. Walters, N. Schuster, C. Walzer, G. Stalder, R. E. Ley and A. H. Farnleitner (2019). "Host diet and evolutionary history explain different aspects of gut microbiome diversity among vertebrate clades." Nature Communications 10(1): 2200.

Yuen, B., J. M. Bayes and S. M. Degnan (2014). "The Characterization of Sponge NLRs Provides Insight into the Origin and Evolution of This Innate Immune Gene Family in Animals." Molecular Biology and Evolution 31(1): 106-120.

Zan, J., C. Fuqua and R. T. Hill (2011). "Diversity and functional analysis of luxS genes in vibrios from marine sponges Mycale laxissima and Ircinia strobilina." ISME J. 5: 1505-1516.

Zeevi, D., T. Korem, A. Godneva, N. Bar, A. Kurilshikov, M. Lotan-Pompan, A. Weinberger, J. Fu, C. Wijmenga, A. Zhernakova and E. Segal (2019). "Structural variation in the gut microbiome associates with host health." Nature 568(7750): 43-48.

Zhang, C., H. Dang, F. Azam, R. Benner, L. Legendre, U. Passow, L. Polimene, C. Robinson, C. A. Suttle and N. Jiao (2018). "Evolving paradigms in biological carbon cycling in the ocean." National Science Review 5(4): 481-499.

Zhang, F., L. C. Blasiak, J. O. Karolin, R. J. Powell, C. D. Geddes and R. T. Hill (2015). "Phosphorus sequestration in the form of polyphosphate by microbial symbionts in marine sponges." Proceedings of the National Academy of Sciences of the United States of America 112(14): 4381-4386.

Zhang, F., L. Jonas, H. Lin and R. T. Hill (2019). "Microbially mediated nutrient cycles in marine sponges." FEMS Microbiol Ecol 95(11).

Zhang, F., J. Vicente and R. T. Hill (2014). "Temporal changes in the diazotrophic bacterial communities associated with Caribbean sponges Ircinia stroblina and Mycale laxissima." Frontiers in microbiology 5: 561-561.

159

Zhang, L., W. Jiang, J. Nan, J. Almqvist and Y. Huang (2014). "The Escherichia coli CysZ is a pH dependent sulfate transporter that can be inhibited by sulfite." Biochim Biophys Acta 1838(7): 1809-1816.

Zheng, D., T. Liwinski and E. Elinav (2020). "Interaction between microbiota and immunity in health and disease." Cell Research 30(6): 492-506.

Zheng, H., J. E. Powell, M. I. Steele, C. Dietrich and N. A. Moran (2017). "Honeybee gut microbiota promotes host weight gain via bacterial metabolism and hormonal signaling." Proceedings of the National Academy of Sciences 114(18): 4775.

Zhou, W., M. R. Sailani, K. Contrepois, Y. Zhou, S. Ahadi, S. R. Leopold, M. J. Zhang, V. Rao, M. Avina, T. Mishra, J. Johnson, B. Lee-McMullen, S. Chen, A. A. Metwally, T. D. B. Tran, H. Nguyen, X. Zhou, B. Albright, B.-Y. Hong, L. Petersen, E. Bautista, B. Hanson, L. Chen, D. Spakowicz, A. Bahmani, D. Salins, B. Leopold, M. Ashland, O. Dagan-Rosenfeld, S. Rego, P. Limcaoco, E. Colbert, C. Allister, D. Perelman, C. Craig, E. Wei, H. Chaib, D. Hornburg, J. Dunn, L. Liang, S. M. S.-F. Rose, K. Kukurba, B. Piening, H. Rost, D. Tse, T. McLaughlin, E. Sodergren, G. M. Weinstock and M. Snyder (2019). "Longitudinal multi- omics of host–microbe dynamics in prediabetes." Nature 569(7758): 663-671.

Zhou, Y., S. Holmseth, R. Hua, A. C. Lehre, A. M. Olofsson, I. Poblete-Naredo, S. A. Kempson and N. C. Danbolt (2012). "The betaine-GABA transporter (BGT1, slc6a12) is predominantly expressed in the liver and at lower levels in the kidneys and at the brain surface." American Journal of Physiology-Renal Physiology 302(3): F316-F328.

Zientz, E., T. Dandekar and R. Gross (2004). "Metabolic interdependence of obligate intracellular bacteria and their insect hosts." Microbiology and molecular biology reviews : MMBR 68(4): 745-770.

Zuniga, M., I. Comas, R. Linaje, V. Monedero, M. J. Yebra, C. D. Esteban, J. Deutscher, G. Perez-Martinez and F. Gonzalez-Candelas (2005). "Horizontal gene transfer in the molecular evolution of mannose PTS transporters." Mol Biol Evol 22(8): 1673-1685.

160

Appendices

Files marked with asterisk* are available online via CloudStor, a file sharing / cloud storage service offered by the Australia’s Academic and Research Network (AARNet). Link: https://cloudstor.aarnet.edu.au/plus/s/ofVx4c4vhtJeb3P Password: Sponge_Amphimedon2020

Supplementary files (Chapter2)

Supplementary File 2-1. Distribution Distribution of insert sizes in the Chicago libraries. The distance between forward and reverse reads is given on the X-axis in base pairs, and the probability of observing a read pair with a given insert size is shown on the Y-axis.

161

Supplementary File 2-2. Statistics of total raw data for the hologenome, from the Chicago long- read and Illumina short-read sequencing. The three Chicago libraries were paired-end sequenced and produced 277,225,058 read pairs (2 x 101 bp). A single paired-end Illumina library (insert size 350 bp) was sequenced across four lanes to yield 121,761,003 read pairs (2 x 74-75 bp). readNUM, the number of reads; baseNUM, the number of base pairs; aveReadLEN, the average length of the reads.

Total raw data Library readNUM baseNUM aveReadLEN ChicagoLibrary1_R1 111,419,501 11,253,369,601 101 ChicagoLibrary1_R2 111,419,501 11,253,369,601 101 ChicagoLibrary2_R1 51,348,871 5,186,235,971 101 ChicagoLibrary2_R2 51,348,871 5,186,235,971 101 ChicagoLibrary3_R1 114,456,686 11,560,125,286 101 ChicagoLibrary3_R2 114,456,686 11,560,125,286 101 Illumina Lane001_R1 30,365,914 2,261,219,761 74 Illumina Lane001_R2 30,365,914 2,260,735,113 74 Illumina Lane002_R1 30,111,093 2,242,259,535 74 Illumina Lane002_R2 30,111,093 2,242,192,132 74 Illumina Lane003_R1 31,796,023 2,367,531,897 74 Illumina Lane003_R2 31,796,023 2,368,225,851 74 Illumina Lane004_R1 29,487,973 2,195,837,617 74 Illumina Lane004_R2 29,487,973 2,195,571,650 74

162

Supplementary File 2-3. Statistics of raw data from the Chicago long-read and Illumina short-read sequencing, partitioned into the four genomes. For the Chicago library data, 178,327,335 read pairs are aligned to the sponge host Aq, and for the symbionts 262,408 read pairs are aligned to AqS1, 23,068 read pairs to AqS2, and 26,805 read pairs to AqS3. For the Illumina library data, 59,287,381 read pairs are aligned to the sponge host Aq, and for the symbionts 3,624,178 read pairs are aligned to AqS1, 928,606 read pairs to AqS2, and 606,539 read pairs to AqS3. Aq, the host sponge A. queenslandica; AqS1, AqS2 and AqS3, the three primary proteobacterial symbionts. readNUM, the number of reads; baseNUM, the number of base pairs; aveReadLEN, the average length of the reads.

Genome-partitioned raw data Aq AqS1 AqS2 AqS3 Library readNUM baseNUM aveReadLEN readNUM baseNUM aveReadLEN readNUM baseNUM aveReadLEN readNUM baseNUM aveReadLEN ChicagoLibrary1_R1 72,610,255 7,333,635,755 101 44,591 4,503,691 101 7,070 714,070 101 8,778 886,578 101 ChicagoLibrary1_R2 72,610,255 7,333,635,755 101 44,591 4,503,691 101 7,070 714,070 101 8,778 886,578 101 ChicagoLibrary2_R1 32,316,924 3,264,009,324 101 28,533 2,881,833 101 6,315 637,815 101 5,633 568,933 101 ChicagoLibrary2_R2 32,316,924 3,264,009,324 101 28,533 2,881,833 101 6,315 637,815 101 5,633 568,933 101 ChicagoLibrary3_R1 73,400,156 7,413,415,756 101 189,284 19,117,684 101 9,683 977,983 101 12,394 1,251,794 101 ChicagoLibrary3_R2 73,400,156 7,413,415,756 101 189,284 19,117,684 101 9,683 977,983 101 12,394 1,251,794 101 Illumina Lane001_R1 14,767,767 1,099,779,127 75 894,688 66,761,989 74 231,316 17,265,259 74 150,513 11,228,007 74 Illumina Lane001_R2 14,767,767 1,099,474,021 75 894,688 66,740,259 74 231,316 17,258,002 74 150,513 11,221,321 74 Illumina Lane002_R1 14,693,465 1,094,253,079 75 874,313 65,240,135 74 225,753 16,849,979 74 146,589 10,935,261 74 Illumina Lane002_R2 14,693,465 1,094,000,883 75 874,313 65,223,824 74 225,753 16,844,562 74 146,589 10,930,579 74 Illumina Lane003_R1 15,522,287 1,155,940,073 75 972,766 72,581,982 74 246,228 18,377,288 74 162,312 12,107,728 74 Illumina Lane003_R2 15,522,287 1,155,744,625 75 972,766 72,570,285 74 246,228 18,372,827 74 162,312 12,103,839 74 Illumina Lane004_R1 14,303,862 1,065,241,400 75 882,411 65,851,186 74 225,309 16,818,953 74 147,125 10,977,942 74 Illumina Lane004_R2 14,303,862 1,064,967,925 75 882,411 65,830,351 74 225,309 16,810,954 74 147,125 10,970,523 74

163

Supplementary File 2-4. The number of A. queenslandica, AqS1, AqS2 and AqS3 genes classified into the sub-categories of the six broad biological categories by KEGG Mapper.

Biological category Biological sub-category Aq AqS1 AqS2 AqS3 Cellular Processes Cell growth and death 718 18 15 14 Cellular Processes Cell motility 155 2 0 1 Cellular Processes Cellular community - eukaryotes 408 1 2 2 Cellular Processes Cellular community - prokaryotes 17 93 25 20 Cellular Processes Transport and catabolism 1030 12 3 9 Environmental Information Processing Membrane transport 98 153 72 27 Environmental Information Processing Signal transduction 1479 60 20 30 Environmental Information Processing Signaling molecules and interaction 193 0 0 1 Genetic Information Processing Folding, sorting and degradation 713 45 34 32 Genetic Information Processing Replication and repair 244 57 34 50 Genetic Information Processing Transcription 300 5 4 4 Genetic Information Processing Translation 572 90 72 80 Human Diseases Cancer: overview 1102 21 8 19 Human Diseases Cancer: specific types 521 5 2 6 Human Diseases Cardiovascular disease 238 4 3 10 Human Diseases Drug resistance: antimicrobial 0 27 16 12 Human Diseases Drug resistance: antineoplastic 281 11 6 10 Human Diseases Endocrine and metabolic disease 376 12 4 9 Human Diseases Immune disease 76 0 0 2

The table is continued on the next page.

164

Biological category Biological sub-category Aq AqS1 AqS2 AqS3 Human Diseases Infectious disease: bacterial 636 17 7 10 Human Diseases Infectious disease: parasitic 208 2 0 3 Human Diseases Infectious disease: viral 795 2 3 4 Human Diseases Neurodegenerative disease 414 10 2 6 Human Diseases Substance dependence 175 4 1 2 Metabolism Amino acid metabolism 377 233 113 141 Metabolism Biosynthesis of other secondary metabolites 64 36 22 27 Metabolism Carbohydrate metabolism 474 175 123 96 Metabolism Energy metabolism 213 162 81 74 Metabolism Glycan biosynthesis and metabolism 251 33 32 33 Metabolism Lipid metabolism 388 46 33 41 Metabolism Metabolism of cofactors and vitamins 220 114 79 93 Metabolism Metabolism of other amino acids 171 44 25 31 Metabolism Metabolism of terpenoids and polyketides 58 30 12 23 Metabolism Nucleotide metabolism 183 65 36 44 Metabolism Xenobiotics biodegradation and metabolism 123 42 7 27 Organismal Systems Aging 163 9 8 12 Organismal Systems Circulatory system 163 6 0 4 Organismal Systems Development and regeneration 517 2 1 1 Organismal Systems Digestive system 459 0 0 0 Organismal Systems Endocrine system 645 12 7 8 Organismal Systems Environmental adaptation 279 15 4 11 Organismal Systems Excretory system 209 0 0 0 Organismal Systems Immune system 772 13 1 3 Organismal Systems Nervous system 410 8 3 3 Organismal Systems Sensory system 201 0 0 0

165

Supplementary File 2-5.* Biological functional annotation results of A. queenslandica, AqS1, AqS2 and AqS3 genes. Spreadsheet Aqu3.1_gene_annotation present the Aqu3.1 gene annotation results, including the corresponding Aqu2.1 gene IDs, the KEGG annotation results (KO, KO_Enzyme, KO_Description) and Blast2GO annotation results (Blast2go_Description, Blast2go_Enzyme, Blast2go_Enzyme_Name). Spreadsheet Symbionts_gene_annotation present the gene annotation results of AqS1, AqS2, and AqS3, including, the KEGG annotation results (KO, KO_Enzyme, KO_Description), Blast2GO annotation results (Blast2go_Description, Blast2go_Enzyme, Blast2go_Enzyme_Name), and Prokka annotation

Supplementary files (Chapter3)

Supplementary File 3-1.* Summary statistic data of the gene depth of each sample.

Supplementary File 3-2.* The average proportion of expressed genes per GO term detected in A. queenslandica (Aq) and the primary symbionts (AqS1, AqS2 and AqS3) in the rRNA-depleted-, Poly(A)- and bacterial unenriched-Poly(A)-RNA data sets. The total gene column represents the total genes assigned to each GO term in A. queenslandica (Aq) and the primary symbionts (AqS1, AqS2 and AqS3).

Supplementary File 3-3.* The GO enrichment results (adjusted p-values) of the expressed genes in the rRNA-depleted-, Poly(A)- and bacterial unenriched-Poly(A)-RNA data sets for A. queenslandica (Aq) and the primary symbionts (AqS1, AqS2 and AqS3).

166

Supplementary files (Chapter4)

Quartile 0 1 2 3 4

Supplementary File 4-1. The expression levels of A. queenslandica holobiont genes involved in the histidine metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

167

Quartile 0 1 2 3 4

Supplementary File 4-2. The expression levels of A. queenslandica holobiont genes involved in valine, leucine and isoleucine biosynthesis pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub- block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

168

Quartile 0 1 2 3 4

Supplementary File 4-3. The expression levels of A. queenslandica holobiont genes involved in phenylalanine, tyrosine and tryptophan biosynthesis pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub- block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014)

169

Quartile 0 1 2 3 4

Supplementary File 4-4. The expression levels of A. queenslandica holobiont genes involved in glycine, serine and threonine metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub- block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

170

Quartile 0 1 2 3 4

Supplementary File 4-5. The expression levels of A. queenslandica holobiont genes involved in cysteine and methionine metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

171

Quartile 0 1 2 3 4

Supplementary File 4-6. The expression levels of A. queenslandica holobiont genes involved in arginine biosynthesis pathway in the adult holobiont. By manually check through NCBI blastp, the A. queenslandica ornithine carbamoyltransferase (EC 2.1.3.3) is contaminative sequence from AqS1. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

172

Quartile 0 1 2 3 4

Supplementary File 4-7. The expression levels of A. queenslandica holobiont genes involved in alanine, aspartate and glutamate metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

173

Quartile 0 1 2 3 4

Supplementary File 4-8 The expression levels of A. queenslandica holobiont genes involved in arginine and proline metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

174

Quartile 0 1 2 3 4

Supplementary File 4-9. The expression levels of A. queenslandica holobiont genes involved in fatty acid biosynthesis pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

175

Quartile 0 1 2 3 4

Supplementary File 4-10. The expression levels of A. queenslandica holobiont genes involved in fatty acid elongation pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

176

Quartile 0 1 2 3 4

Supplementary File 4-11. The expression levels of A. queenslandica holobiont genes involved in thiamine metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

177

Quartile 0 1 2 3 4

Supplementary File 4-12. The expression levels of A. queenslandica holobiont genes involved in folate biosynthesis pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

178

Quartile 0 1 2 3 4

Supplementary File 4-13. The expression levels of A. queenslandica holobiont genes involved in porphyrin and chlorophyll metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

179

Quartile 0 1 2 3 4

Supplementary File 4-14. The expression levels of A. queenslandica holobiont genes involved in nicotinate and nicotinamide metabolism pathway in the adult holobiont. The EC code is designated in each enzyme block. The four sub-blocks in each enzyme block from left to right represent adult holobiont expression levels in A. queenslandica, AqS1, AqS2 and AqS3. If the sub-block is blank, the enzyme is not present in that species. The colour legend shows the gene expression levels (0: not expressed, 1, 2, 3, and 4 indicate the gene expressed at quartile 1, 2, 3 and 4). This figure was created with R package pathview (Luo 2014).

180